Abell Braselton Modern Differential Equations Second Edition

ABELL
BRASELTON
MODERN DIFFERENTIAL
QU A T I O N S
SECOND EDITION
m ~
08 m
c;;o
z
»0
-I:;;
_ i"1
m
O~
z -~
Vl~
SECO N D
Harcourt ED ITI ON AB ELL & BRASELTON
College Publishers
-iorcourt Higher l earn ing Company
!tP://www.harcourtcollege.com l 9
•
Table of Integrals f cos au cos bu du =
sin(a - b)u
2(a - b)
+ f k u - a
du = Inlu +Y u
2
- a 21 + C
sin(a + b)u
f Un du = _I_U"+I
n+I
+C
'
n *--1 f cos 3
U du = }(2 + cos 2 u)sin u + C 2(a + b) + C
f~ du = Inlul +C f sm." u I sm
du = - ; . "-I u cos u + f sin au cos bu du = cos(a - b)u
2(a - b) f Ya2 - u 2 u
du = -2 Ya 2 - u2 +
2
a sin-I.!!.
2 a
+C
cos(a + b)u
fe"du=e"+C
n: I f sin"-2 U du 2(a + b) + C
2 2
f u Ya - u' du = f(2u 2 - a')Ya2 - u> +
au - I f cos" u du = ~ COS"-I u sin u + I sin" u cos m u du =
sinn - 1
u cos m + lU a 4 . -1
8sm
U
-;;-+ C
f ue Q
" du = --_ea"
a2
+C n+m +
-n n
- -I f COS"-2 u du
n - If s· "+1
- - sin"-2 u cos m u du = m
n+m
u cos
n+m
m-I
u + f sin-'u du = u sin-I u +~+C
a"
2
f tan u du = tan u -:-..'u +C m-If
n + m sinn u cosm- 2 U du f cos-1u du = U COS-I U - ~+C
f a"du = -- +C
+
In a
f sin u du = -cos u + C
3
f tan u du = + tan 2 u + Inlcos ul + C f ,r0--2
I
Va--u-
sm - +
d
u=
. -I U
a
C
f tan-1u du = u tan- t u - In(1 + u2 ) + C
f cos u du = sin u + C
3
f sec u du = + sec u tan u +
f +I
-2--2
a
1 tan - + C
u
dU = -
a
-I U
a
f e
a
" sin bu du =
I ea "
"2 Inlsec u + tan ul + C a' + b2 (a sin bu - b cos bu) + C
+C f I 1 sec - + C
,;-:;---0
uVu 2 -a 2
d
u= -
-I U
f tan u du = -Inlcos ul
f tan" u du = _1_
n- I
tan"-I u - f tan"-2 u du
.
a a
fe a
" cos bu du =
IU -- -al + C
f cot u du = Inlsin ul +C f I
-2--2
u-a
-2 I n
du = I
a u+a e "
a
f sec" u du =n~ I tan u sec"-2 u + a 2 + b2 (a cos bu + b sin bu) + C

f sec u du = Inlsec u + tan ul + C
: =if sec"-2 u du + C
f i =-2 I IUn+- -al + C
-2--2
a-u
du I
a U-a flnudu=ulnu-u+C
f esc u du = lnlcsc u - cot ul +C

f u sin u du = sin u - u cos u +C
f w-+;} =I Ya + + du
2
u
2
f u" In u du = _u_- [en

"+1
+ I) In u - I] +C
(n + I) 2
a2 , ; - :2; - - - 0
f sec u tan u du = sec u +C f u cos u du = cos u + u sin u + C

2" Inlu + Va
2
+u l +C
f udv=uv- f vdu
fsec 2 udu=tanu+C f u" sin u du = -un cos U +n f U"-I cos u du
f~ a +u
du
2
lnlu + Ya + u2 +
= 1 C
f sin2 u du = .!!.
2 4
- 1. sin 2u +C f u" cos u du = u" sin u - n f U"-I sin u du
f Yu' - a du I Yu a
2
= 2
-
2
-
f cos 2 I + 1- sin 2u + C .. sin(a - b)u

a 2
2" Inlu +
, ; - :2; - - - 0
V u - a'i +C
U du =
fsm au sm bu du = 2(a _ b) -
sin(a + b)u C
f sin3 u du = -}(2 + sin2 u)cos u + C 2(a + b) +
1 . Vkt 1 -Ids
ae ' 'erfc(a Vtt) -v:Jc
1 - a
v:;i sm(2 kt) s3/2 e
Table of Laplace Transforms Vs+a 'TTk
~ + aea"erf(aVt)
Vs ~ cosh(2Vkt) 1
Vs
e
Ids
k(s2 + 2k2) S - a2
F(s) = ~{f(t)}
f(t)
1
kt cosh kt S4 + 4k4 1 ~ sinh(2Vkt) 1
S3/2 e
Ids
lea"erf(aVt) 'TTk
S
k(s2 - 2k2) a Vs(s - a 2 ) s-a
n! kt sinh kt S4 + 4k4
l(e b' _ e a') In--
1 I s-b
t", n a positive integer sn+l a-b ea"erfc(aVt) + k2
Vs(Vs + a) 2 S2
1
eat _ e b1
(s - a)(s - b) t(l - cos kt) I n -- 2-
e a'
S
(a - b)s I k'
s-a
k ae at _ be bt (s - a)(s - b)
Jo(kt)
Ys2+k' t2 (1 - cosh kt)
s
S2 -
In-- 2-
sin kt + k2 n!
S2 _1_ - Ids 1 .
s t"e a" n a positive integer (s - a)"+1 ~ cos(2Vkt) Vs
e Ism kt tan-I!
s
cos kt + k' ( -1)"F(")(s)
S2 t"f(t), n a positive integer
k /,(t) sF(s) - f(O)
sinh kt S2 - k' s2F(s) - sf(O) - Special Formulas
/,,(t)
s f'(O)
cosh kt S2 - k', f(n)(t) s"F(s) - s"-lf(O) -
S"-2/, (0) - ... - Trigonometric Identities Hyperbolic Trigonometric Functions
k
e a, sin kt Sf"-2)(0) - f("-I)(O) cos 2 a + sin2 a = 1 1 _
(s - a)2 + k' cosh x = "2(e X + e X)
s-a ea'f(t) F(s - a) I + tan2 a = sec 2 a
e a, cos kt
(s - a)' + k' (f* g)(t) = F(s)G(s) 1 + cot 2 a = csc2 a
sinh x = t(e X - e- X )
k {f(t - v)g(v) dv
cos( a + (3) = cos a cos {3 - sin a sin (3
e a' sinh kt (s - a)' - k' 0 e- as
cos(a - (3) = cos a cos {3 + sin a sin (3 tanh x = sinh x = eX - e -x
OU(t - a), a ~ 0 sine a + (3) = sin a cos {3 + cos a sin (3 cosh x eX + e- x
s-a s
e a' cosh kt (s - a)' - k' e-asF(s) sin(a - (3) = sin a cos {3 - cos a sin (3 sech x = _1_ = __2__x
f(t - a)OU(t - a), a ~ 0 cosh x eX + e-
1 e-as~{f(t + a)} sin 2a = 2 sin a cos a
f(t)OU(t - a), a ~ 0
teat cos 2a = cos 2 a - sin2 a
t sin kt
(s - a)'
2ks
(S2 + k 2 )'
8(t - to), to ~ 0
f(t - T) = f(t)
e- roS
~
1- e
r0
e-s'f(t) dt cos2 a = 1 + cos 2a
2
. 2
sm a=
1 - cos 2a
2
cothx
1 2
cschx = sinh x = eX _ e- x
= _1_ = eX + e- x
x
C?S a
t cos kt
S2 - k'
(S2 + k')2
t- 1I2
Hy'7T
tana= sin a
cos a
1
cot a =
sm a
1
cosh2 x - sinh2
tanh x
X
eX - e-
= 1
2k 3 t l/2 seca=-- csc a = - . -
sin kt - kt cos kt (S2 + k 2 )2 2S 312 cos a sm a Maclaurin Series
1 . 3 . 5 ... (2n - 1)v7T cos( -a) = cos a sine - a) = -sin a
2ks 2 tn-liZ, n = 1,2, ... 2nsn+1/2
x2 x3 xn IX)
sin kt + kt cos kt eX=I+x+-+-+"'=)'-

(S2 + k')2 Logarithmic and Exponential 2! 3! ,:;:;'0 n!'
rea + I)
k2 ta, a>-I Function Properties -00 < x < 00
~
I-coskt
kt - sin kt
S(S2 + k 2 )
k3
S2(S2 + k')
+f(t) rF(s)
F(u) du
In aX = x In a, a > 0
In ab = In a + In b; a, b > 0
. x3
3!
-00 < x < co
x5
smx=x--+--'" = ),(-1) - ' - - -
5! "~o
n x 2n + 1
(2n + I)!'
00
8k 3s 2 {f(a) da b = In a - In b'' a , b > 0

In E. x2 X4 IXI x2n
(I + k 2t 2 )sin kt - kt cos kt (S2 + k')3 0
s cos X = I - 2! + 4! - ... = "~o ( - I)" (2n)! '
Inex=x
2k's ~ea'(l + 2at) <x<
(s - ai 12
-00 00
sin kt sinh kt e1nx=x,x>O
+ 4k4 1
S4
S3 _ _I_(e b ' _ e a') Vs=a-~
eX + Y = eXe Y 1 _ x = 1 + x + x 2 + ... = LX", - I < x < 1
00
cos kt cosh kl eX)' = (eXV n=O

S4 + 4e 2v:;;:(3
rr
Modern
DifferentiaL
Equations
r--~ is
Modern
[ It ~g"~~~~ublishers
A Harcourt Higher Learning Company DifferentiaL
Now you will find Saunders College
Publishing's distinguished innovation,
leadership, and support under a different
name ... a new brand that continues our
Equations
unsurpassed quality, service, and Second Edition
commitment to education.
We are combining the strengths of our

college imprints into one worldwide
brand: ~Harcourt
Our mission is to make learning

Martha L. Abell and James P. Braselton
accessible to anyone, anywhere, Georgia Southern University
anytime-reinforcing our commitment
to lifelong learning.
We are now Harcourt College Publishers.

Ask for us by name. F<_,
~! ~) VB ~~. ~ 11l0·' ~.~ ~ t1l ~.\ :}
AiWhe;~"ie~'r~'i~~"g
Comes to Life .. "
HARCOURT COLLEGE PUBLISHERS

Fort Worth Philadelphia San Diego New York Orlando Austin
San Antonio Toronto Montreal London Sydney Tokyo
Contents
Publisher: Emily Barrosse

Executive Editor: Angus McDonald
Introduction to Differential Equations
Acquisitions Editor: Liz Covello
Marketing Strategist: Julia Conover
Developmental Editor: James D. LaPointe
Project Editor: Frank Messina
1 1.1 Introduction 2
1.2 A Graphical Approach to Solutions: Slope Fields and
Production Manager: Charlene Catlett Squibb Direction Fields 13
Art Director and Text Designer: Paul Fry
Chapter 1 Summary 22
Cover Credit: Copyright 1996 Eric PearlelFPG International. Chapter 1 Review Exercises 23
MODERN DIFFERENTIAL EQUATIONS, Second Edition

First-Order Equations 25
ISBN: 0-03-028704-9
Library of Congress Catalog Card Number: 00-103663
2 2.1
2.2
Separable Equations 25
First-Order Linear Equations 36
Copyright © 2001. 1996 by Harcourt, Inc. 2.3 Substitution Methods and Special Equations 44
AU rights reserved. No part of this publication may be reproduced or transmitted in any form or by any 2.4 Exact Equations 55
means, electronic or mechanical. including photocopy. recording, or any information storage and retrieval 2.5 Theory of First-Order Equations 64
system, without permission in writing from the publisher.
2.6 Numerical Methods for First-Order Equations 70
Requests for pennission to make copies of any part of the work should be mailed to: Pennissions Depart- Chapter 2 Summary 82
ment, Harcourt, Inc., 6277 Sea Harbor Drive, Orlando, Florida 32887-6777. Chapter 2 Review Exercises 83
Chapter 2 Differential Equations at Work 86
Address for domestic orders:
A: Modeling the Spread of Disease 86
Harcourt College Publishers, 6277 Sea Harbor Drive, Orlando, FL 32887-6777
1-800-782-4479 B: Linear Population Model with Harvesting 88
e-mail collegesales@harcourt.com C: Logistic Model with Harvesting 89
D: Logistic Model with Predation 91
Address for international orders:
International Customer Service, Harcourt, Inc.
6277 Sea Harbor Drive, Orlando FL 32887-6777 Applications of First-Order Equations 93
(407) 345-3800
Fax (407) 345-4060
e-mail hbintl@harcourt.com
3 3.1 Population Growth and Decay 94
3.2 Newton's Law of Cooling and Related Problems 104
Address Jor editorial correspondence: 3.3 Free-Falling Bodies 111
Harcourt College Publishers, Public Ledger Building, Suite 1250, 150 S. Independence Mall West, Chapter 3 Summary 121
Philadelphia, PA 19106-3412
Chapter 3 Review Exercises 121
Web Site Address Chapter 3 Differential Equations at Work 125
http://www.harcourtcollege.com A: Mathematics of Finance 125
B: Algae Growth 128
Printed in the United States of America
C: Dialysis 129
0123456789 048 10987654321 D: Antibiotic Production 132
v
I
I
L
Contents vii
vi Contents
Higher Order Equations 134 6.5 First-Order Linear Nonhomogeneous Systems: Undetermined
4 4.1 Second-Order Equations: An Introduction 135

4.2 Solutions of Second-Order Linear Homogeneous Equations
Coefficients and Variation of Parameters 340
6.6 Phase Portraits 347
6.7 Nonlinear Systems 360
with Constant Coefficients 147 6.8 Numerical Methods 369
4.3 Higher Order Equations: An Introduction 153 Chapter 6 Summary 378
4.4 Solutions to Higher Order Linear Homogeneous Equations Chapter 6 Review Exercises 381
with Constant Coefficients 159 Chapter 6 Differential Equations at Work 383
4.5 Introduction to Solving Nonhomogeneous Equations with A: Modeling a Fox Population in Which Rabies Is Present 383
Constant Coefficients: Method of Undetermined B: Controlling the Spread of a Disease 384
Coefficients 168 C: FitzHugh-Nagumo Model 387
4.6 Nonhomogeneous Equations with Constant Coefficients:
Variation of Parameters 183
4.7 Cauchy-Euler Equations 196 Applications of First-Order Systems 389
4.8 Series Solutions of Ordinary Differential Equations 205
7 7.1 Mechanical and Electrical Problems with First-Order Linear
Systems 389
Chapter 4 Differential Equations at Work 230 7.2 Diffusion and Population Problems with First-Order Linear
A: Testing for Diabetes 230 Systems 399
B: Modeling the Motion of a Skier 232 7.3 Nonlinear Systems of Equations 408
C: The Schrodinger Equation 235 Chapter 7 Summary 417
Applications of Higher Order Equations 238
5 5.1
5.2
Simple Harmonic Motion 238
Damped Motion 247
A: Competing Species 419
B: Food Chains 420
C: Chemical Reactor 422
5.3 Forced Motion 257
5.4 Other Applications 267
5.5 The Pendulum Problem 274 Laplace Transforms 424
8 8.1 The Laplace Transform: Preliminary Definitions
and Notation 425
8.2 Solving Initial-Value Problems with the Laplace Transform 439
A: Rack-and-Gear Systems 285
B: Soft Springs 288 8.3 Laplace Transforms of Several Important Functions 445
C: Hard Springs 288 8.4 The Convolution Theorem 459
D: Aging Springs 289 8.5 Laplace Transform Methods for Solving Systems 465
E: Bode Plots 289 8.6 Applications Using Laplace Transforms 470
Systems of Differential Equations 291 Chapter 8 Review Exercises 489
6 6.1
6.2
Introduction 291
Review of Matrix Algebra and Calculus 299
A: The Tautochrone 492
B: Vibration Absorbers 493
6.3 Preliminary Definitions and Notation 314 C: Airplane Wing 494
6.4 First-Order Linear Homogeneous Systems with Constant D: Free Vibration of a Three-Story Building 495
Coefficients 323 E: Control Systems 497
L
viii Contents
9 Eigenvalue Problems and Fourier Series 499

9.1 Boundary-Value Problems, Eigenvalue Problems, Sturm-
Liouville Problems 500
Preface
9.2 Fourier Sine Series and Cosine Series 510
9.3 Fourier Series 520
9.4 Generalized Fourier Series 535
Chapter 9 Review &ercises 543
A: Signal Processing 545
B: Forced Damped Spring-Mass System 546
C: Approximations with Fourier Series 547
C omputer algebra system and sophisticated graphing calculators have changed
the ways in which we learn and teach ordinary differential equations. Instead of
focusing students' attention only on a sequence of solution methods, we want
them to use their minds to understand what solutions mean and how differential equa-
tions can be used to answer pertinent questions.
10 Partial Differential Equations

10.1
549
Introduction to Partial Differential Equations and Separation
of Variables 550
Interestingly, this metamorphosis in the teaching of differential equations oc-
curred relatively "overnight" and coincided with our professional careers at Georgia
Southern University. Our interest in the use of technology in the mathematics class-
room began in 1990 when we started to use computer laboratories and demonstrations
10.2 The One-Dimensional Heat Equation 557 in our calculus, differential equations, and applied mathematics courses. Over the past
10.3 The One-Dimensional Wave Equation 567 ten years we have learned some ways of how to and how not to use technology in the
10.4 Problems in Two Dimensions: Laplace's Equation 578 mathematics curriculum. In the early stages, we simply wanted to show students how
10.5 Two-Dimensional Problems in a Circular Region 588 they could solve more difficult problems by using a computer algebra system so that
Chapter 10 Summary 600 they could be exposed to the technology. However, we soon realized that we were miss-
Chapter 10 Review &ercises 600 ing the great opportunity of allowing students to discover aspects of the subject mat-
Chapter 10 Differential Equations at Work 602 ter on their own. We revised our materials to include experimental problems and
A: Laplace Transforms 602 thought-provoking questions in which students are asked to make conjectures and in-
B: Waves in a Steel Rod 604 vestigate supporting evidence. We also developed application projects called Differen-
C: Media Sterilization 604 tial Equations at Work, not only to emphasize technology, but also to improve the prob-
D: Numerical Methods for Solving Partial Differential lem-solving and communication skills of our students. To preserve the "wow" aspects
Equations 605 of technology, we continue to use it to observe solutions in classroom demonstrations
through such things as animating the motion of springs and pendulums. These demon-
Answers to Selected Exercises A-1 strations not only grab the attention of students, but also help them to make the con-
nection between a formula and what it represents.
In presenting our fmdings to colleagues around the country, we quickly found
Index 1-1 out that others were interested in our work. As a result, we decided to develop a dif-
ferential equations textbook to share this work with those who share our desire to im-
prove mathematics education. This book is a culmination of years of "trials and tribu-
lation" as the differential equations students at Georgia Southern can attest. Our hope
is that its use will inspire students to open their eyes to the exciting discoveries that
differential equations offer.
This book is designed to serve as a text for beginning courses in differential equa-
tions. Usually, introductory differential equations courses are taken by students who
have successfully completed a first-year calculus course, and this text is written at a
level readable for them.
ix
L
x Preface Preface xi
Technology Differential Equations is written in an easy-to-read, yet mathematically precise, style.

It contains all topics usually included in standard differential equations texts. Defini-
The benefits of incorporating technology into mathematics courses are well-known. tions, theorems, and proofs are concise but worded precisely for mathematical accu-
Some of the advantages include enhancing the ability to solve a variety of problems; racy. Generally, theorems are proved if the proof is instructive or has "teaching value";
helping students work examples; supporting varied, realistic, and illuminating applica- these proofs are optional. In other cases, proofs of theorems are developed in the ex-
tions; exploiting and improving geometric intuition; encouraging mathematical exper- ercises or omitted. Theorems and defmitions are boxed for easy reference; key terms
iments; teaching approximation; showing the mathematical significance of the com- are highlighted in boldface. Figures are used frequently to clarify material with a graph-
puter revolution; and making higher-level mathematics accessible to students. In ical interpretation.
addition, technology is implemented throughout this text to promote the following goals
in the learning of differential equations:
1. Solving problems: Using different methods to solve problems and generalize solu-
Second Edition Features
tions;
The second edition of Modern Differential Equations is an extensive revision of the
2. Reasoning: Exploiting computer graphics to develop spatial reasoning through vi-
first edition. Particular features include:
sualization;
3. Analyzing: Finding the most reasonable solution to real problems or observing Over 220 new exercises, for a total of 2567 exercises;
changes in the solution under changing conditions; Seven new Differential Equations at Work projects to provide a wider variety of proj-
4. Communicating mathematics: Developing written, verbal, and visual skills to com- ects to interest, motivate, and challenge classes and students;
municate mathematical ideas; and • Many more computer-generated graphs to help students understand the meaning of
5. Synthesizing: Making inferences and generalizations, evaluating outcomes, classi- the results, especially in the context of applications;
fying objects, and controlling variables. Greater emphasis on qualitative and graphical aspects of solving differential equa-
tions throughout the text (for example, see Section 2.1);
Students who develop these skills will succeed not only in differential equations, but
also in subsequent courses and in the workforce. Greater motivation of material through applications of differential equations through-
The (f]) icon is used throughout the text to indicate those examples in which out the text (for example, see Section 2.2);
technology is used in a nontrivial way to develop or visualize the solution or to indi- The inclusion of a chapter on Fourier series (Chapter 9) and a chapter on partial
cate the sections of the text, such as those discussing numerical methods, in which the differential equations (Chapter 10) to give those who desire to use the text for two
semesters or to use the text in an introductory applied mathematics or Fourier
use of appropriate technology is essential.
series/partial differential equations course the ability to do so.
Applications
Pedagogical Features
Applications in this text are taken from a variety of fields, especially biology, physics,
chemistry, engineering, and economics, and they are documented by references. These
Examples
applications can be found in many of the examples and exercises, in separate sections
and chapters of the text, and in the Differential Equations at Work subsections at the Throughout the text, numerous examples are given, with thorough explanations and a
end of each chapter. Many of these applications are well-suited to exploration with substantial amount of detail. Solutions to more difficult examples are constructed with
technology because they incorporate real data. In particular, obtaining closed form so- the help of graphing calculators or a computer algebra system and are indicated by an
lutions is not necessarily "easy" (or always possible). These applications, even if not icon.
formally discussed in class, show students that differential equations is an exciting and
interesting subject with extensive applications in many fields. "Think about it!"
Many examples are followed by a question indicated by a ~. Generally, basic knowl-
Style edge about the behavior of functions is sufficient to answer the question. Many of these
questions encourage students to use technology. Others focus on the graph of a solu-
To keep the text as flexible as possible, addressing the needs of both audiences with tion. Thus, "Think about it!" questions help students determine when to use technol-
different mathematical backgrounds and instructors with varying preferences. Modern ogy and make this text more interactive.
L
Preface
xiii
xii Preface
fer students extra practice on the topics in that chapter. These exercises are arranged
Technology by section so that students having difficulty can turn to the appropriate material for re-
Many students entering their first differential course have had substantial experience view.
with various sophisticated calculators and computer algebra systems.
Technology is used throughout the text to explore many of the applications and Figures
more difficult examples, especially those marked with ::l, and the problems in the sub- One of technology'S benefits is the ease of graphing and plotting. This text provides
sections Differential Equations at Work Solutions, partial solutions, or hints to those an abundance of figures and graphs, especially for solutions to examples. In addition,
exercises indicated by an asterisk are contained in the Student Resource Manual. students are encouraged to develop spatial visualization and reasoning skills, to inter-
Differential Equations at Work pret graphs, and to discover and explore concepts from a graphical point of view. To
ensure accuracy, the figures and graphs have been completely computer-generated. The
Differential Equations at Work subsections describe detailed economics, biology, Student Resource Manual shows typical graphical output using Mathematica and Maple
physics, chemistry, and engineering problems documented by references. These prob- for exercise solutions.
lems include real data when available and require students to provide answers based
on different conditions. Students must analyze the problem and make decisions about Historical Material
the best way to solve it, including the appropriate use of technology. Each Differential
Equations at Work project can be assigned as a project requiring a written report, for Nearly every topic is motivated by either an application or an appropriate historical
group work, or for discussion in class. note. We have also indicated photos of many famous mathematicians and descriptions
Differential Equations at Work also illustrate how. differential equations are used of the mathematics they discovered.
in the real world. Students are often reluctant to believe that the subject matter in cal-
culus, linear algebra, and differential equations classes relates to subsequent courses
Content
and to their careers. Each Differential Equations at Work subsection illustrates how the
material discussed in the course is used in real life. The highlights of each chapter are described briefly below.
The problems are not connected to a specific section of the text; they require stu-
dents to draw different mathematical skills and concepts together to solve a problem. Chapter 1 After introducing preliminary definitions, we discuss direction fields n~t
Because each Differential Equations at Work is cumulative in nature, students must only for first-order differential equations, but also for systems of eq~tlOns. In thIS
combine mathematical concepts, techniques, and experiences from previous chapters presentation, we establish a basic understanding of solutions and theIr graphs. We
give an overview of some of the applications covered later in the text to po~t out
and math courses. Detailed hints regarding the use of technology in solving the prob-
the usefulness of the topic and some of the reasons we have for studymg differ-
lems encountered in Differential Equations at Work are included in the Student Re-
ential equations in the exercises.
source Manual.
Chapter 2 In addition to discussing the standard techni~ues for solving several o/!>es
Exercises of fIrst-order differential equations (separable equations, homogeneous equatIOns,
exact equations, and linear equations), we introduce several numerical. methods
Numerous exercises, ranging in level from easy to difficult, are included in each sec- (Euler's Method, Improved Euler's Method, Runge-Kutta ~e~od) and dISCUSS the
tion of the text. In particular, the exercise sets for topics that students find most diffi- existence and uniqueness of solutions to first-order mItIal-value problems.
cult are rich and varied. The abundant "routine" exercises encourage students to mas- Throu"hout the chapter, we encourage students to build an intuitive approach to
ter basic techniques. Most sections also contain interesting mathematical and applied the solution process by matching a graph to a solution without actually solving
problems to show that mathematics and its applications are both interesting and rele- the equation.
vant. Instructors will find that they can assign a large number of problems, if desired, Chapter 3 Not only do we cover most standard applications of first-order equations
yet still have plenty for review in addition to those found in the review section at the in Chapter 3 (orthogonal trajectories, population growth and decay, Newton's la,:"
end of each chapter. Answers to most odd-numbered exercises are included at the end of cooling, free-falling bodies), but we also present many that are not (due to theIr
of the text; detailed solutions to those exercises marked with an asterisk are included computational difficulty) in Differential Equations at Work.
in the Student Resource Manual. Chapter 4 This chapter emphasizes the methods for solving homogeneous. an~ non-
homogeneous higher order differential equations. It also stresse~ the PnncIple of
Chapter Summary and Review Exercises Superposition and the differences between the properties of solutIons t~ lmear ~d
nonlinear equations. After discussing Cauchy-Euler equations and ~trodu~mg
Each chapter ends with a chapter summary highlighting important concepts, key terms
series methods of solution to differential equations, the chapter culmmates m a
and formulas, and theorems. The Review Exercises following the chapter summary of-
l
xiv Preface Preface xv
discussion of several special equations and the properties of their solutions-equa- problems. Suggestions regarding specific Differential Equations at Work are also
tions important in many areas of applied mathematics and physics. included.
Chapter 5 Several applications of higher order equations are presented. The distinc- The Student Resource Manual contains detailed solutions to selected exercises (those
tive presentation illustrates the motion of spring-mass systems and pendulums marked with an *) and problems and explanations of prerequisite techniques
graphically to help students understand what solutions represent and to make the needed for a standard differential equations course, but not necessarily mastered
applications more meaningful to them. by students enrolled in the course. The Student Resource Manual also contains so-
Chapter 6 The study of systems of differential equations is perhaps the most excit- lutions, partial solutions, or hints to some Differential Equations at Work prob-
ing of all the topics covered in the text. Although we direct most of our attention lems and substantial guidance for users of Mathematica and Maple software.
to solving systems oflinear first-order equations with constant coefficients, tech- A website (www.harcourtcollege.com!mathexpress) has specifically been created for
nology allows us to investigate systems of nonlinear equations and observe phase the second edition of Modern Differential Equations. This website offers additional re-
planes. We also show how to use eigenvalues and eigenvectors to understand the
sources to instructors and students in conjunction with the adoption of the text.
general behavior of systems of linear and nonlinear equations.
Chapter 7 Several applications discussed earlier in the text are extended to more than
one dimension and solved using systems of differential equations, in an effort to Acknowledgments
reinforce the understanding of these important problems. Numerous applications
involving nonlinear systems are discussed as well. The development of this text has involved a thorough program of review, for both ped-
Chapter 8 Laplace transforms are important in many areas of engineering and ex- agogical and topical content and for accuracy. We gratefully acknowledge the contri-
hibit intriguing mathematical properties as well. Throughout the chapter, we point butions of our colleagues that were involved in the revision of this text:
out the importance of initial conditions and forcing functions on initial-value prob- Elizabeth Bonawitz, Virginia Polytechnique Institute and State University
lems.
Steven Chapin, Ohio University
Chapter 9 Chapter 9 focuses on boundary value, eigenvalue, and Sturm-Liouville Charles MacCluer, Michigan State University
problems and Fourier series, including generalized Fourier series. Technology is John Spitler, University ofl'Vjloming
utilized to observe the convergence of these series and visualize the Gibbs phe- Craig Steenberg, Lewis-Clark State College
nomenon. Marie Vanisko, Carroll College of Montana
Chapter lOIn this chapter, we use the method of separation of variables to solve the
Thanks also go to those reviewers who provided feedback during the fIrst edition of
one-dimensional heat equation, one-dimensional wave equation, Laplace's equa-
tion (in a rectangular and circular region), and the wave equation in a circular re- this text:
gion. Exercises involving cylindrical and spherical coordinates are also included. Josefina Alvarez, New Mexico State University
Graphics are used so that students can see how the results obtained relate to the Sidney Birnbaum, California State Polytechnic University, Pomona
applications. Philip Crooke, Vanderbilt University
For a one-semester course introducing ordinary differential equations, many in- Steve R. Dunbar, University of Nebraska
Richard S. Falk, Rutgers University
structors will choose to cover topics from Chapters I to 7 or from Chapters I to 5 and
Ronald B. Guenther, Oregon State University
Chapter 8. For a two-semester course, the instructor will easily be able to cover the re-
Kenneth A. Heimes, Iowa State University
maining chapters of the text. In our introductory ordinary differential equations course, Dar-Veig Ho, Georgia Institute of Technology
we typically cover most of Chapters 1, 2, 4, and 6, and instructors choose a variety of Helmut Knaust, University of Texas, El Paso
applications from Chapters 3, 5, and 7. In our applied mathematics course, we cover David O. Lomen, University ofArizona
most of the material in Chapters 8 to 10. Marie Vanisko, Carroll College
We owe particular thanks for the careful comments and suggestions from those who
class-tested this project during its initial development:
Supplements
Lyle Cochran, Fresno Pacific College
The following aids prepared by the authors for instructors and students are available Stuart Davidson, Greensboro College
from the publisher: Marie Vanisko, Carroll College
The Instructor's Resource Manual contains detailed solutions to the exercises, plus The accuracy ofthe examples and answers in the back of the book was carefully checked
Mathematica and Maple code and output for the Differential Equations at Work in manuscript, galleys, and page proofs by Elizabeth Bonawitz (Virginia Polytechnic
h
,
I
xvi Preface
1
Institute and State University) and Stuart Davidson (Greensboro College). Their care-
ful and diligent work is greatly appreciated. Any remaining errors are our sole re-
sponsibility, and we would greatly appreciate hearing about them.
This text would not have been possible without the substantial efforts of many
at Harcourt College Publishers. In particular, we would like to express our gratitude to
Liz Covello, Senior Acquisitions Editor
James D. LaPointe, Developmental Editor
Amanda Loch, Editorial Assistant
Frank Messina, Senior Project Editor
Paul Fry, Senior Art Director
Charlene Squibb, Senior Production Manager
Introduction to
Joanne Cassetti, Director of Production
In addition, we would especially like to thank Dr. James Herod (Professor Emer-
DifferentiaL Equations
itus, Mathematics, The Georgia Institute of Technology) and Dr. Tim Howard (Assis-
tant Professor, Mathematics, Columbus State University) for their encouragement and
good ideas about ways to teach differential equations and incorporate applications (es-
pecially biological applications that students find interesting) into the differential equa-
tions course that Martha learned during her participation in the Faculty Development
Program at The Georgia Institute of Technology.
Finally, we must thank our families, particularly Imogene H. Abell, Lori Brasel-
ton, Ada Braselton, and Martha Braselton, for enduring with us the pressures of meet-
ing a deadline and for graciously accepting our demanding work schedules. We could
not have completed this task without their care and understanding.
Martha Abell and James Braselton

Statesboro, Georgia
March, 2000 he purpose of Modern Differential Equations: Theory, Applications, :ech-
IN HONOR OF
Imogene Holcomb Abell Joseph Abell
T nology is twofold. First, we introduce and discuss the tOPICS covered In an
undergraduate course in ordinary differential equations. Second, we in-
dicate how certain technologies such as computer algebra systems and graph-
ics calculators are used to enhance the study of differential equations, not only
Nelle Holcomb Karen Elizabeth Braselton by eliminating some of the computational difficulties that arise in. t~e study of
Margaret Holcomb Evelyn Albrecht Farnsworth differential equations but also by overcoming some of the Visual limitations as-
Emilie Abell Lorraine Maruth Braselton sociated with the solutions of differential equations.
Thomas Abell Ada Braselton The advantages of using technology such as graphics calculators and com-
puter algebra systems in the study" of differential equations are numerous~ but
Amy Abell Martha Braselton
perhaps the most useful is that of being able to produce the graphl~s ~ssoCiated
William Abell
with solutions of differential equations. This is particularly benefiCial In the diS-
cussion of applications because many physical situations are modeled with dif-
IN MEMORY OF ferential equations. For example, in Chapter 5, we see that the motion of a pen-
William J. Abell, Jr. W. Emmett Braselton dulum can be modeled by a differential equation. When we solve the problem
Louise Chambless Abell Doris Jacob Braselton of the motion of a pendulum, we use technology to watch the pendulum move.
The same is true for the motion of a mass attached to the end of a spring, as
Mildred Holcomb Ada Moyle Maruth
1
i
l
3
Chapter 1 Introduction to Differential Equations 1.1 Introduction
2
(C = 0), vCt) = -32t + 32 (C = 32), and v(t) = -32t - 8, (C = -8). This shows that
weil as many other problems. In having this ability to use technology, the study of dif-
ferential equations becomes much more meaningful as well as interesting. there are an infinite number of solutions.
We can verifY that vet) = - 32t + C is a general solution of dvldt = -32 through
Although Chapter I is short in length, the vocabulary introduced here is used
throughout the text. Although, to a large extent, this chapter may be read quickly, sub- substitution:
sequent chapters will take advantage of the terminology and techniques discussed here.
Any formal introduction to differential equations should begin with German sci- dv =!!... (-32t + C) = -32.
dt dt
entist Gottfried Wilhelm Leibniz (1646-1716) and British scientist Isaac Newton
(1642-1727), the inventors of calculus. Although we learn in integral calculus that the When we substitute our solution into the left side of the ODE, we obtain - 32, the value
Gottfried Wilhelm Leibniz area under the graph of a smooth positive function is given by a definite integral, both on the right side of the ODE. Therefore, we have verified that vet) = - 32t + C satis-
(1646-1716) Leibniz and Newton were more concerned with solving differential equations than fmd- fies the ODE for any choice of C.
(North Wind Picture Archives) ing areas. Many of the methods of solution we present in this textbook are from the Many times we are given a particular condition that the solution must satisfY. For
great Swiss mathematician Leonhard Euler (1707-1783). Subsequently, many prob- example, suppose that the object considered earlier has an initial v~locity ~f.-:64 ft/s.
lems, such as determining the motion of a string, not only lead to ordinary and partial In other words, the velocity at time t = 0 seconds is represented With the mItial co.n-
differential equations but to other areas of mathematics as well. Indeed, differential dition v(O) = -64. Therefore, we need to find a solution of the ODE that also satIs-
equations is full of rich and exciting applications; interesting applications are included fies the initial condition. We express this initial value problem (IVP) as
throughout the textbook to motivate discussions, make the study of differential equa-
tions more interesting and pertinent to the real world, and indicate how some people dv = -32, v(O) = -64,
dt
use differential equations beyond this course. However, mathematics is also interesting
in its own right; mathematical applications are also included throughout the textbook. where we solve the initial value problem by first finding a general solution to the ODE
and then by applying the initial condition to determine the arbitra,y' constant. When
Isaac Newton
(1642-1727) we substitute t = 0 into the general solution vet) = - 32t + C, we obtam v(O) = - 32(0)
+ C = C. Therefore, C = -64 so that v(O) = -64 is satisfied. This means that the so-
~
(North Wind PIcture Archil.:es)
Introduction lution to the IVP is vet) = -32t - 64. Notice that unlike the ODE, the IVP has only
one solution.
We begin our study of differential equations by explaining what a differential equation Another application of differential equations found in calculus is finding a func-
is. From our experience in calculus, we are familiar with some differential equations. tion when given (a) the slope of the line tangent to the graph of the functIOn at any
For example, suppose that the acceleration aCt) (measured in ft/s 2 ) of a falling object point (x, y), and (b) a point on the graph. For example, suppose that the slope of the
is aCt) = -32. Then, because aCt) = v'Ct), where vCt) is the velocity of the object (meas- tangent line at any point on the graph is given by
ured in ft/s), we have v'Ct) = -32 or
dy = 3x2 - 4x
dv = -32 dx '
dt .
Leonhard Euler
(1707-1783) and suppose that the graph passes through ~e point (1,4). In this case, the ordinary
An equation like this involving a function of a single variable is called an ordinary
(North Wind Picture Archives) differential equation is given by dyldx = 3x- - 4x, where we try to fmd the funct10n
differential equation (ODE). (If the equation involves partial derivatives, then it is
y(x) that satisfies the initial condition y(l) = 4. Therefore, we solve the IVP,
called a partial differential equation.) In this case, the function is v, which depends
on the variable t, representing time (measured in seconds). The goal in solving an ODE
is to find a function that satisfies the equation. We can solve this ODE through inte- dy = 3x2 - 4x, y(1) = 4.
dx
gration: 2
As in the previous example, we find a general solution to dyldx = 3x - 4x through
vCt) = f aCt)dt = f (-32)dt = -32t + C, integration. This yields
2
y(x) = x 3 - 2x + C.
where C is an arbitrary constant. This result indicates that vet) = - 32t + C is a solu-
tion of the ODE for any choice of C. (We call this a general solution because it in- Now, when we apply the initial condition, we find that
volves an arbitrary constant.) In fact, we have found every solution of the ODE be- y(I) = l' - 2(1)2 + C = -1 + C = 4
cause each is expressed as - 32t + C. Examples of solutions include vCt) = - 32t
l
4 Chapter 1 Introduction to Differential Equations
1.1 Introduction 5
so that C = 5. Therefore, the solution to the IVP is y 2

d y. .
y(x) = x 3 - 2x 2 + 5.
df2 + Y = -CI sm t - C2 cos t + CI sm t + C2 cos t = 0,
y 4
. Figure .1.1 shows the. graph of the solution together with a portion of the tangent so the function satisfies the ODE for any choice of CI and C2' We graph the solu-
hne at the pomt (1, 4). Notice that the slope of the tangent line is dy/dx evaluated when tion for several values of these constants in Figure 1.2. Because the solution
x = .1, dy/dx = 3(1)2 - 4(1) = -1. This observation will be useful in Section 1.2 in depends on at least one constant, we say that the functions shown in Figure 1.2
helpmg us bett~r understand the behavior of solutions of differential equations. are members of the family of solutions of the ODE.
-2
The prev.lOu~ two ~xamples are similar in that they each involve an ordinary dif- (b) Evaluating the function at t = 0 yields yeO) = CI sin 0 + C2 cos 0 = C2' Then,
ferential equatLOn m which the highest order derivative is the first derivative. We can -4 the initial condition, yeO) = 0, indicates that C2 = O. Similarly, y' (0) = CI cos 0 -
x equations of this type first-order ordinary differential equations because the order of C2 sin 0 = CJo so CI = 1 so that the second initial condition, y' (0) = I, is satisfied.
-6
~ differential equation is the order of the highest order derivative appearing in the equa- Therefore, y(l) = sin t is the solution to the IVP. We graph this solution in Figure
Figure 1.1 Solution to dyldx tion. 1.3. Notice that the ODE has an infinite number of solutions while the IVP has only
= 3x 2 - 4x, y(l) = 4 along Figure 1.2 Graph of y = CI
with the tangent line at (1, 4) one solution. Also observe that the number of initial conditions in the IVP matches
sin I + C2 cos I for various val·
the order of the ODE.
ues of CI and C2
'-
Detennine the order of each of the fonowing differential equations' The next level of classification is based on the fonowing dermition. An ordinary
2 differential equation (of order n) is caned linear if it is of the form
(a) dy/dx = x 1y2 cos y (b) Uxx + uyy = 0 (e) (dy/dx)4 = y + x .
d 2x dx
(d) df2 + 2 dt + 3x = sin t d~ ~-~ d~ ~
an(x) dxn + an-I(x) dxn-I + ... + a2(x) dx 2 + al(x) dx + ao(x)y = f(x),
: 1 So~ution (a) This eq~ation is first-order because it includes only one first-order where the functions aJ<x),j = 0, I, ... ,n, andf(x) are given and an(x) is not the zero
denvatlve, dy/dx. (b) ThiS equation is classified as second-order because the high- function.
est order derivatives, uxx, representing a2u/ax2, and uyy , representing a 2ulay2, are of If the equation under consideration is not of this form, the equation is said to be
Figure 1.3 Graph ofY(I) = sin I nonlinear. Therefore, some of the properties that lead to classifying an equation as
order :W? He~ce, Laplace's equation is a second-order partial differential equation.
(e) ThiS IS a frrst-order eq'fltion because the highest order derivative is the rrrst de- nonlinear are powers of the dependent variable (or one of its derivatives) and functions
rivative. Raising that derivative to the fourth power does not affect the order of the of the dependent variable. A similar classification is fonowed for partial differential
equation. The expressions equations. In this case, the coefficients in a linear partial differential equation are func-
tions of the independent variables.
(dyldx)4 and
do not represent the same quantities: (dy/dx)4 represents the derivative of y with re-
spect to x, dy/dx, raised to the fourth power; d 4yldx 4 represents the fourth deriva-
t~ve ?f y with respect to x. (d) The highest order derivative is d 2x/dt 2 , so the equa-
tion IS second order.
"1'''''••
Determine which of the fonowing differential equations are linear: (a) dy/dx = x 3
d 2u d 3y dy
(b) dx 2 + u = eX (e)(y - 1) dx + x cos(y) dy = 0 (d) dx3 + Y dx = x
'iE'"I,'., dy
(e) dx + x 2y = X
d 2x
(f) dl 2 + sin x = 0 (g) Uxx + YUy = 0 (h) U xx + uUy = 0
(a) Show that y = CI sin. t + C2 cos t satisfies the second-order ODE d 2y/d(2 + Y = 0
where CI and C2 are arbitrary constants. (b) Find the solution to the IVP d 2Yldt 2 +
Y = 0, yeO) = 0, y'(O) = 1.
:1 Solution (a) This equation is linear because the nonlinear term x 3 is the func-
tionf(x) of the independent variable in the general formula for a linear differential
:1 Solution (a) Differentiating, we obtain dy/dt = CI cos t - C2 sin t and d 2y/dt 2

equation. (b) This equation is also linear. Using u as the name of the dependent vari-
able does not affect the linearity. (e) Ify is the dependent variable, solving for dy/dx
= CI sm t - C2 cos t. Therefore,
gives us
l
1.1 Introduction 7
dy ~ (c) To integrate y = J16 ~ x 2 dx, we use a trigonometric substitution. Letting

dx x cosy
x = 4 tan 0 -~ < 0 < ~ so that dx = 4 sec2 0 dO gives us
Because the right side of this equation includes a nonlinear function of y, the equa- . ' 2 2
tion is nonlinear (in y). However, if x is the dependent variable, solving for dx/dy
yields
dx cos y Y--J 16 + 1
(4 tan 0)2
4sec2 0dO
-=--x
J__
dy l-y·
This equation is linear in the dependent variable x. (d) The coefficient of the term =J 1_2- 4 sec 0 dO
16 sec 0
2
dy/dx is y instead of an expression involving only the independent variable x. Hence,

this equation is nonlinear in the dependent variable y. (e) This equation is linear.
The term x 2 is the coefficient function. (f) This equation, known as the pendulum
= 1..
4
JdO = 1..4 0 + C =-4 = tan 8 :9"'= tan - =-4
t
equation because it models the motion of a pendulum, is nonlinear because it in-

volves a nonlinear function of x, sin x. Note that x is the dependent variable in this
case; t is the independent variable. (g) This partial differential equation is linear be-
cause the coefficient of u y is a function of one of the independent variables. (h) In
Graphy = ± tan-I (1) + C ifC = -1,0, and 1.
y
this case, there is a product of u and one of its derivatives, so the equation is non-
linear.
(d) To evaluate y = Jxe x dx by hand, we use the integration by parts formula,
J u dv = uv - J v du, with u = x, dv = eX dx, du = dx, and v = eX. This gives

If the ODE has the form dy/dx = I(x), we can use integration to determine y. The
following examples of differential equations of this type illustrate some of the meth-
ods of integration that may be encountered.
y = J xe x dx = xe x - Je~ x
= xe - eX + C.
vdu
"E'"I,I.' Find C ify(l) = 2.
(e) Use partial fractions to evaluate this integral. First, find the partial fraction de-
Figure 1.4 Graph of Solve the following differential equations.
y = sin x + C for various dy composition of - 4
1 2' which is determined by finding the constants A and B
dy dy x dy 1 -x
values of C (a) dx = cos x (b) dx = v'?"+l (c) dx = 16 + x2 (d) dx = xe x y
4 1 = _A_ + _B_. These values are
(e) dy =_1_
that satisfy the equation (2 _ x)(2 + x) 2- x 2 +x
dx 4 - x2 1
y
A = B = 4. (Why?) Thus,
:.1 Solution In each case, we integrate the indicated function and graph the solu-
tion for several values of the constant of integration. Each solution contains a con- y = 1..
4
J(_1_
2-x
+ _1_) dx
2+x
= 1.. [-In 12 - xl + In 12 + xl] + C
4
stant of integration so there are (infinitely) many solutions to each equation.
(a) y = J cos x dx = sin x + C. (See Figure 1.4.) = -

4
1 In \2- - \ + C.
+x
2-x
.
(See FIgure 1.6.)
I -4
(b) To evaluate y = J~+ x2 1

dx' we let u = x 2 + 1 so that du = 2x dx. Then,
Figure 1.6 Graph of
-2 1
y=-In 12-+- + C xl In Example 4, each solution is given as a function y = y(x~ of the in~epend~nt
Figure 1.5 Graph of

y - J1
-"2 v'?"+l
2x dx = 1..
2
J Vu
_1_ du = 1..
2
J
U- 112 du = u l/2 +C 4 2-x variable. In these cases, the solution is said to be explicit. In solVIng some dIfferentIal
equations, however, we can find only an equation involving x and y that the solution
y = \I'?+l + C for various
satisfies. In this case, we say that we have found an implicit solution.
values of C = v'?"+l + C. (See Figure 1.5.)
L
8 Chapter 1 Introduction to Differential Equations 1.1 Introduction 9
We will see that given an ar- Notice that the solutions of dy/dx = 1/(4 - x 2) are undefmed when x = -2 or x
Find an equation a/the line tangent to the graph 0/2x2 + y2 - 2xy + 5x = 0 at the
bitrary differential equation, points (-I, -3) and (-1, 1).
= 2. This is because lI(4 - x 2 ) has vertical asymptotes at these two values of x. Later,
constructing an explicit or im-
plicit solution is nearly always we will discuss in greater detail the relationship between the differential equation and In the same manner that we consider systems of equations in algebra, we can
its solutions. also consider systems of differential equations. For example, if x and y represent func-
impossible. Consequently, al-
though mathematicians were tions of t, we will learn in Chapter 6 to solve the system of linear equations
first concerned with rmding
analytic (explicit or implicit)
solutions to differential equa-
"ii,,','., {
X' = ax
y' = ex
+ by
+ dy'
2
tions, they have since (fre- VerifY that the equation 2x +/ - 2xy + 5x = 0 satisfies the differential equation
where a, b, e, and d represent constants and differentiation is with respect to t. We will
quently) turned their attention
dy 2y - 4x - 5 see that systems of differential equations arise naturally in many physical situations
to addressing properties of
the solution and finding dx 2y - 2x that are modeled with more than one equation and involve more than one dependent
algorithms to approximate variable. In addition, we will see that it is often useful to write a differential equation
solutions. : I Solution with order greater than one as a system of first-order equations.
. We use implicit differentiation to compute y' = dy/dx if 2x 2 + y2 - 2xy + 5x = 0:
2 Y dy dy
4x + 2y dx - 2x dx - 2y +5= 0
EXERCISES 1.1
-s -6 2X dy
dx (2y - 2x) = 2y - 4x - 5
dy 2y - 4x - 5
dx = 2y - 2x Determine if each of the following equations is an ordinary dif- d 2x . .
ferential equation or a partial differential equation. If the equa- 10. df2 + 2 smx = sm 21
. . f' d'ffi . I . dy 2y - 4x - 5 tion is an ordinary differential equation, then detennine (a) the
The equatIOn satls les the I erenlla equatIOn dx = 2y _ 2x order of the ordinary differential equation; and (b) if the equa- 11. u[ + UUx = UUxx • (J" constant
-s 2
Although we cannot solve 2x + y2 - 2xy + 5x = 0 for y as afimetion of x tion is linear or nonlinear. 12. (2x - I) dx - dy = 0
Figure 1.7 Graph of 2x 2 + y2 (see Figure 1.7), we can determine the corresponding y value(s) for a given value d y
2
dy 3 *13. (2x - y) dx - dy = 0
-2xy+5x=O of x. For example, if x = -1, then 1. dx 2 + dx - 2y = x
auau
14 . ax ay = u
2 + y2 + 2y - 5 = y2 + 2y - 3 = (y + 3)(y - 1) = O. 2. y : + y4 = sin x
15. (2x - y) dx - y dy = 0
Therefore, the points (-I, - 3) and (-1, 1) lie on the graph of 2x 2 + y2 - 2xy + 2 2
5x = O. (See Figure 1.8.) 3 a y = c2 a y 16. Write each of the following second-order equations as
• al2 ax 2 a system of f1rst-order equations.
L
4. ylll - 2y" + 5y + Y =
I eX d 2x dx
Figure 1.8 Notice that near
the points (-1, -3) and (-1,1)
the implicit solution looks like a
-2.S 1.2 *5. (:r + y= 0
(a) df2 -
d x2
*(b) 4 dl 2 +4
dt -
dx
dl
6x = 0
+ 37x = 0
-2.9
d 2y
2 dy
function. In fact, if we zoom in 1.1 6. x dx 2 + x dx + 2y = 0 2
near the points (-1, -3) and d x
(e) L dl 2 + g sin x = 0
(-1, I), we see what appears to -3 (-I, -3) I a2z a2 z a2z
be the graph of a function. 7. -;?aIi = ax 2 + ay2
d 2x dx
-3.1 0.9 *(d) df2 - J.L(1 - x 2 ) dt +x = 0
8. UU x + Ut = 0
o.s
x(~r + 2y = 2x
-3.2 2
d x dx
*9. (e) 1df2 + (b - I) dt - ax = 0
-1.2 -1.1 -1 -0.9 -O.S -1.2 -1.1 -1 -0.9 -O.S
L
Chapter 1 Introduction to Differential Equations 1.1 Introduction 11
10
2
d y dy (b) Determine lim,_~ V(I).
In Exercises 17-28, verify that each of the given functions is a 32. y cos x ax + (2y + sin x) dy = 0, y2 + Y sin x = 1; 48. ax 2 + 9 ax = O,y(O) = 2,y'(0) = -1,
solution to the corresponding differential equation. (A, B, and x=O 60. The number of cells P(t) in a bacteria colony after
C represent constants.) y(x) = A + Be- 9x I hours is determined by solving the initial-value
dy
33. (? + cos y) ax + (In x - x sin y) dy = 0,
49. J
d2
+ 9y = O,y(O) = O,y'(O) = 1,
problem
17. ax + 2y = O,y(x) = e- 2x,y(x) = Se- 2x y In x +x cos Y = 0; x = 1 dP=kP
y(x) = A cos 3x + B sin 3x dl
{
dy P(O) = Po
18. ax + xy = 0, y(x) = e-x '/2 In Exercises 34-43, use integration to fmd a solution to the dif-
50.
d 2y
ax' + 9y = 0, yeO) = 0, y (0) = 0,
,
ferential equation.
dy . _ 1 1 = A cos 3x + B sin 3x
(a) If a general solution of a,;: = kP is P(I) = Cela,
*19. ax + y = sm x, y(x) =e x - '2 cos x + '2 sin X dy
y(x)
34. ax = (x 2 - l)(x ' - 3X)' d'y d 2y use the initial condition to find C.
d y 2
dy *51. ax - ax = 0, yeO) = 0, y'(0) = 1,
2 2 (b) Find the value of k so that the popUlation doubles
20. ax 2 - ax - 12y = 0, y(x) = e 4x, y(x) = e -lx dy . '
35. ax = x sm(x 2) = 3, y(x) = A + Bx + Ce 2x
y"(O) in 8 hours.
d'y dy d'y dy *61. In 1840, the Belgian mathematician-biologist Pierre
21. ax 2 ax = 0, y(x) = A + Be- 9x
+9 36 dy = x 52. ax - 4 ax = O,y(O) = I,y'(O) = -1, F. Verhulst (1804-1849) developed the logistic equa-
'ax ~ ' tion dP/dl = rP - aP', where r and a are positive
22.
d 2x ax
'7ft2 + 3 dl - lOx = 0, X(I) = Ae 2t + Be-"
y"(O) = 0, y(x) = A + Be 2x + Ce- 2x
constants, to predict the population P(I) in certain
dy 1 d 2y dy
*37. ax = x In x 53. x 2 ax 2 - 12x ax + 42y = 0, y(1) = 1, countries.
d 2x
*23. dl 2 +x = I cos I - cos I, X(I) = A cos I + dy y'(1) = -I,y(x) =Ax 6 + Bx 7 (a) If a general solution to this equation is
38. ax =xlnx 2
r
B sin I + £. sin I - .!.. sin I +.!.. cos I 54. X
2 d y
ax dy
+ 3x ax + Sy = O,y(I) = O,y'(I) = 1, P(I) = a + ee-"
4 2 4 dy 2
39. ax = xe-
x
x -,(A cos(2 In x) + B sin(2 In x))
d 2y dy y(x) = fmd the solution that satisfies P(O) = Po.
24. ax 2 - 12 ax + 40y = 0, y(x) = e
6x
cos 2x,
dy -2(x + S)
(b) Determine limH~ P(I).
y(x) = e 6x
sin 2x 40. ax = (x + 2)(x - 4) In Exercises SS-S8, solve the initial-value problem. Graph the
25.
d'y
ax - 4
dy
ax = 0, y(x) = A + Be 2x + Ce- 2x *41 dy = x - x'
solution on an appropriate interval. 62. The differential equation ddS + __
I
3_- S =
1+ 100 '
where °
' . ax (x + 1)(x' + 1) dy S(I) is the number of pounds of salt in a particular
55. ax = 4X' - + 2, yeO) 1
42.:= ~
X =
d'y d 2y tank at time I, is used to approximate the amount
26. ax - 2 ax 2 = 0, y(x) = A + Bx + Ce 2x of salt in the tank containing a salt-water mixture in
' 56.
dy
ax .
= sm 2x - cos 2x, yeO) = 0
, d 2y dy which pure water is allowed to flow into the tank
27. X ax 2 - 12x ax + 42y = 0, y(x) = Ax 6 + Bx 7 43. : = (4 - 2
X )3/2 while the mixture is allowed to flow out of the tank.
2 d 2y dy dy 1
*57. : = xl, cos(~),y(~) = 1 If S(I) = IS,OOO,OOO
(I + 100)' ,
show that S satisfies
ax + 3x ax + Sy =
28. X 2
y(x) = x-I[A cos (2 In x)

0,
+B sin (2 In x)]
44. ax = x2 - 16
58.
dy
ax Inx
=--;-,y(I)=O
dS
Tt + 1+ 3100 S =. ° What IS
. th e 101ltal
... amount of
salt in the tank? As I ..... 00, what happens to the amount

In Exercises 4S-S4, use the indicated conditions with the indi- 59. The velocity of a falling object of mass m that is sub-
of salt in the tank?
cated solution to determine the solution to the given problem. jected to air resistance proportional to the instanta-
In Exercises 29-33, verify that the given equation satisfies the 63. The displacement (measured from x = 0) of a mass
neous velocity v of the object is found by solving the
differential equation. Use the equation to determine y for the dy attached to the end of a spring at time I is given by
45. ax + 2y = 0, yeO) = 2, y(x) = Ae- 2x initial-value problem
given value of x. Graph each equation using a graphing calcu- x(l) = 3 cos 41 + ~ sin 41. Show that x satisfies the or-
lator or computer algebra system.
46.
dy
ax .
+ y = smx,y(O) = -1,
mTt=mg-cv
dv
°
(Note that c > is the dinary differential equation x" + 16x = O. What is the
initial position of the mass? What is the initial veloc-
ax = y' x + y2 = 16; x = °
dy -x 2
{ proportionality constant.)
29. v(O) = Vo
ity of the mass?
y(x) = Ae -x - ~ cos x + ~ sin x (a) If a general solution to m ~~ = mg - cv is 64. Show that u(x,y) = In v'x 2 + y2 satisfies Laplace's
30. 3Y(X2 + y) ax + x(x 2 + 6y) dy = 0, x'y + 3xy2 = 8; 2 2
equation u= + Uyy = O.
x=2 d 2y dy v(t) = ~g + Ke -,,1m, find the solution of this
*47. ax
2 - ax - 12y = 0, yeO) = 0, y'(O) = 1, *65. The temperature in a thin rod oflength 217' after I min-
*31.
dy
ax 2y
= -~ - 3, x, + x'y = 100; x = 1 y(x) = Ae 4x + Be -lx initial-value problem. °
utes at a position x between and 217' is given by
l
r 1.2 A Graphical Approach to Solutions: Slope Fields and Direction Fields 13
74. The current I(t) in an L-R circuit, which contains a

I
u(x, I) = 3 - e -,61d cos 4x. Show that u satisfies u, = (b) Graph (x' + y')2 = 5xy on the rectangle [-2,2] X' = 4y
ku=. What is the initial temperature (t = 0) at x = 7r? resistor, an inductor, and a voltage source, satisfies the X [-2,2]. 82. A general solution of the system { y' = -4x is
What happens to the temperature at each point in the
wire as t ......,. oo?
differential equation RI + L 1ft = E(/), where R and (c) Approximate all points on the graph of
X = C, sin 4t - C 1 cos 41 h CdC
(x' + y2)' = 5xy with x-coordinate 1. y = C 1 sin 41 + C, cos 41' were 1 an 2 are
L are constants representing the resistance and the in- {
66. The displacement u of a string of length 1 at time 1 ductance and E(I) is the voltage source. Is this equa- (d) Approximate all points on the graph of
and position x where x is measured from x = 0 is given constants. Solve the initial-value problem
tion linear or nonlinear?'Determine the order of the (x' + y')' = 5xy withy-coordinate -0.319.
by u(x, t) = sin 7rX cos t. Show that u satisfies "rzu" equation. X' = 4y
= u"". What is the value of u at the endpoints x = 0 Often the calculus and algebra encountered in solving differ- y' = -4x
and x = I for all values of t? {
Throughout Modern Differen/ial Equations: Theory. Applica- ential equations can be tedious, if not completely overwhelm- x(O) = 4, y(O) = 0
*67. Find the value(s) ofm so thaty = xm is a solution of tions. Technology. we use graphs of solutions of differential ing or impossible. Today many sophisticated calculators and and then graph X(I), y(t), and the parametric equations
equations. In some cases, we are able to predict what the graph computer algebra systems are capable of performing the inte-
x 2y" - 2xy' + 2y = O. of a solution should look like. If the graph of our proposed so- X = X(/)
gration and algebraic simplification encountered when solving
lution does not appear as predicted, we know that either we
{ y=y(/),
many differential equations. Having access to these tools can
68. Find the value(s) of m so that y = e= is a solution of
y" - 3y' - ISy = O.
made a mistake in constructing our proposed solution or our
conjecture about the general shape of the graph of the solution
be a great advantage: a large number of problems can be solved . f
83. A general soluuon 0 the system
{x' = -5x+4y.
y' = 2x + 2y IS
relatively quickly, we make conj ectures as to the general form
X = C 1e - 4Cze-
3t 6t
is wrong. In other cases, we will find that it is easier to exam- of a solution to different forms of differential equations, and
d '" 2x dy y = 2C,e3, + C,e-6t, where C, and C, are con-
*69. Use the fact that dx (e y) = e dx + 2e 2xy and m-
• ine the graph of a solution than it is to examine the solution (if these tools allow us to check and verify our work. {
we are able to construct one in the first place).
dy x 80. Solve dy/dx = sin4 x, y(O) = 0 and graph the result- stants. Solve the initial-value problem
tegration to solve e
2x
dx + 2e 2xy = e .
ing solution on the interval [0, 47T].
In Exercises 75-77. (a) verify that the indicated function is a X' = -5x + 4y
solution of the given differential equation, and (h) graph the so- y' = 2x + 2y
d x xdy
70. Use the fact that dx (e y) = e dx
x
+ e y and mte-
•
lution on the indicated interval(s). 81. A general solution of /4) + ~ y" - 5y' + 629 y = {
x(O) = 4, y(O) = 0
o is given by 2 16
and graph X(/), y(t), and the parametric equations
gration to solve eX : + eXy = xe
x
• 75. xy' + Y = cos x, y = sin x; [-27r, 0) U (0, 27T]
x y = e- x12(C 1 cos 3x + C2 sin 3x) X =X(I)
71. The time-independent Schriidinger equation is 3x14
+ exl2 (C3 cos 2x + C4 sin 2x), { y=y(l)'
76. 16y" + 24y' + 153y = 0, y = e- cos 3x; [0, t7T]
given by where C lo C2 , C3 , and C4 are constants. Solve the
2
2 2 d 3y d y dy initial-value problem
- 2m
h d ljJ(x)
-;;;z + U(x)ljJ(x) = EljJ(x). *77. x
3
ax 3 +x 2
dx 2 + x dx - 40y = 0,
Y= X-I sin(3 lnx); (0, 7T] /4) + ~y" _ 5y' + 629 y = 0
2 16
If U(x) = 0, find conditions of E so that
78. Verify that {y(O) = 0, y'(O) = I, y"(O) = -1, y"'(O) = I
(n;:)
IjJ(x) = A sin
independent Schriidinger equation.

is a solution of the time-
(a)
X(I) = e-f 00 V3
3 sin v31+ 20 cos v3I]
and graph the resulting solution.
72. Use implicit differentiation to show that

-1. + ~ + 1.. - J, = C is an implicit solution of l y(l) = e- t [ -4~V3 sin v31 + 20 cos v3t]
x x y y A Graphical Approach to Solutions: Slope Fields and
. . . dy (x - 4)y3 is a solution of the system of differential
the differenual equatIon dx = x3(y _ 2)' Is y = 0 a Direction Fields
. {X'(I) = 4y(l)
solution of this differential equation? Is y = 0 a sin- equauons y'(t) = -x(l) - 2y(l)' C Systems of Ordinary Differential Equations and Direction Fields
gular solution? C Relationship Between Systems of First-Order and Higher Order Equations
(h) Graph X(I), y(I), and the parametric equations
2
73. Show that x + x = C is an implicit solution of the X-X(I)
y - for 0 OS t os 27T. Suppose that we are asked to solve the ordinary differential equation dyl dx = e -,? In
{ y=y(t)
. . dy . y2 + 2xy this Case, we do not attempt to solve this equation through integration as we did in the
dIfferential equatIOn dx = --x-2 - ' Is Y =0 a so- *79. (a) Show that (x 2 + y2)' = 5xy is an implicit solu-
previous section because of the presence of the function/(x) = e-'? on the right side
tion of
lution of this differential equation? Is y = 0 a singu- of the ODE. (See Exercise 21.) Instead, we Can gain insight into the behavior ofsolu-
lar solution? [4x(x' + 1) - 5y] dx + [4y(x' + y') - 5x] dy = O. tions of this ODE through a graphical approach by considering the slope of the tan-
L
14 Chapter 1 Introduction to Differential Equations 1.2 A Graphical Approach to Solutions: Slope Fields and Direction Fields 15
y y
gent line to solutions of the ODE. Recall from Section 1.1 that the differential equa-
tion gives the slope of the tangent line to solutions of the ODE at the given point (;c, y) 2
in the xy-plane. Therefore, if we wish to determine the slope of the tangent line to the
solution to the ODE that passes through (0, 1) at this point, we substitute (0, 1) into
the right side of dyldx = e-o? Because the right side only depends on x, we have the
slope
-2 -1 2 x -2 -I 2 x
dy = e-(O)' = 1
dx .
-1 -1
In fact, the slope of the line tangent to solutions at all points of the form (0, y) is 1.
At the point (vln2, 4), we find that the slope is -2 -2
dy = e-(v'hi2)' = e-1n2 = e1n2-' =..!..

dx 2' Figure 1.10 Slope field for Figure 1.11 Sketchi,
dy/dx = e-x' solution to dy/dx = e - ,
Again, we obtain the same slope at all points (vln2, y) = (0.832555, y) and y(-2) = -I
(-vln2,y) = (-0.832555,y).
In Figure 1.9, we draw several short line segments using points of the form (0, y)
for y = 0, :!: 1, :!:2, ±3. Notice that we use a triangle with base I and height I to as-
sist in sketching the tangent lines with slope I at these points. We use a triangle with
base 2 and height I to help us draw the lines of slope 112 that are tangent to solutions
y
'.\ \
': = ax + by
x
at the points (- vln2, y) for y = 0, :!: I, :!:2, :!:3. By drawing a set of short line seg-
ments representing the tangent lines to solutions of the ODE at numerous points in the
plane, we construct the slope field of the ODE. We show the slope field for dyldx =
0.5
......,",. -
6 x
1 dy
dt = ex + dy
,
'I:
-0.5 where a. b, c, and d are given constants. In the case of this system, we solve for x(t)
e- x ' on the square [-2,2] X [-2,2] in Figure 1.10. (Note that because constructing -I and yet). For example, if we consider the system
a slope field is time consuming, we usually let a computer algebra system do the work
for us.) Observe that at each point along the y-axis, the slope is 1 as we predicted ear- = y
lier. Notice also that the slope appears to be zero for larger values of IxI- This is Figure 1.12 Graphs of
xU) = sin t andy(t) = cos t dy ,
because for large values of x (in absolute value), dyldx = 0 because lim e-o? = O. - = -x
Figure 1.9 Several line over 0 ::; t:::; 2'1T.
Therefore, we expect solutions to "flatten ouf' as Ixl increases. x->±oo dt
segments in the slope field
for dy/dx = e -x' We can use the slope field to investigate the solution to an initial-value problem we can verify that the parametric equations
such as
x(t) = sin t
dy _, { y(t) = cos t
dx = eX, y( -2) = -I
satisfy the system because
by starting at the point ( - 2, - I) and tracing the solution by following the tangent line
dx d . dy d
slopes. This solution is sketched in Figure 1.11. Thus, although we did not determine dt = dt (sm t) = cos t = y and dt = dt (cos t) = -sin t = -x.
a formula for solutions to the ODE or the IVP, we were able to determine some prop-
erties of the solutions by using the slope field of the differential equation. We can graph each function separately as we do in Figure 1.12. [The graph of
x(t) is the solid curve; that of y(t) is dashed.] Another option is to graph them as a pair
of parametric equations as in Figure 1.13. As we recall, the graph of this pair of para-
Systems of Ordinary Differential Equations and metric equations is a circle of radius I centered at the origin because x 2 = sin2 t and
Figure 1.13 Graph of
Direction Fields
the parametric equations y2 = cos 2 t, so that x 2 + y2 = sin2 t + cos 2 t = I. However, we must indicate the ori-
We can also consider systems of differential equations. In Chapter 6, we willieam how X(t) = sin t entation of the curve (the direction of increasing parameter value t). For this pair of
{ yet) = cos t for 0 OS; t OS; 27T parametric equations, we find that at t = O. x(O) = sin 0 = 0 and yeO) = cos 0 = 1.
to solve systems of first-order ordinary differential equations of the form
,
l
16
::.:::.:::~:~-~o ,,~, O),~~oo~ f SffiUIM". (I.

to I = 'TT/2 because x( 'TT/2) = sine 'TT12) = I and y( 'TT/2) = cos( 'TT/2) = O. This means that
the solution moves from (0, I) to (1,0) as I increases, so the orientation is clockwise.
We place an arrow directed from (0, I) to (I, 0) to indicate this orientation. The para-
!
t·
!.i'
1.2 A Graphical Approach to Solutions: Slope Fields and Direction Fields
ponents from the system of differential equations. In the case of this system, we con-
sider (y, -x). At the point (l/V2, l/V2), we obtain the vector (l/V2, -l/V2). This
means that the solution through (1IV2, l/V2) has tangent vector (l/V2, -l/V2). The
17
direction field is made up of tangent vectors such as (112, -1!V2) to solutions at points
metric equations (x(l) = sin t, yet) = cos t} satisfy the initial-value problem in the plane, so it is similar to the slope field for dy/dx = -x/y shown in Figure 1.14,
except that vectors are used to indicate the orientation of solutions. In Figure 1.15, we
~ = y, x(O) = 0 show the direction field for this system. The vectors in the direction field indicate that
solutions to this system are circles in the xy-plane that are directed clockwise. We graph
{ dy several solutions along with the direction field in Figure 1.16. A collection of solutions
dt = -x,y(O) = I
in the xy-plane is called the phase portrait of the system. Notice that at points in the
first quadrant, where x> 0 andy> 0, dx/dt = y > 0 and dy/dt = -x < O. This means
because they satisfy the system of differential equations as well as the two initial con- that x(t) increases and yet) decreases along solutions in the first quadrant. In the sec-
ditions. Notice that in the case of an initial-value problem involving a system of dif- ond quadrant, where x < 0 and y > 0, dx/dt = y > 0 and dy/dt = -x > O. Therefore,
ferential equations, an initial condition is given for each of the variables x and y that
x(t) and yet) increase along solutions in the second quadrant. We can perform a simi-
depend on t. In the parametric plot, the solution to this initial-value problem passes lar analysis for points in the other two quadrants,
through the point (x(O), yeO»~ = (0, I).
Another way to view a system of ordinary differential equations is through the Relationship Between Systems of First-Order and
use of a direction field, which is similar to a slope field. First, we write the first-
Higher Order Equations
order system
dx Again, consider the system {ddx//ddt = y . If we differentiate the equation dx/ dt = y with
dt = ax + by y I=-x
respect to t, we obtain d 2 x/dt 2 = dy/dt. Therefore, if we equate dy/dt = -x and dy/dt
dy
{ dt=ex+dy = d x/dt , we have d x/dt = -x or d 2x/dl2 + x = O. We say that the system of two
2 2 2 2
first-order ODEs is equivalent to the second-order ODE d 2x/dt 2 + x = O. Often, how-

as a first-order equation with
dy dy/dt ex + dy
dx = dx/dt = ax + by'
y y y
Then, we can consider the slope field associated with this differential equation. For ex- ' : - -1(r 10
ample, if we refer back to the system
{
~ =y
dy , i I
dt= -x -10: '5 -10 -5 10 x -10 10 x
we obtain the first-order equation dy/dx = -x/y, so we can determine the slope of tan- 't.,., -5
gent lines to solutions at points in the xy-plane. At the point (l/V2, l/V2), the solu-
hon -1/V2 = - I . In a SlITU-
. h as sI ope ~
. to !his system that passes through!hi s pornt .. :'l '-'10
l/v2 I
lar manner, we can find the slope at other points in the plane. However, as we men-
Figure 1.14 Slope field for Figure 1.15 Direction field Figure 1.16 Direction field
tioned in our earlier discussion, we must indicate the orientation when we graph para- dyldx = -xly for {dx1
dt = y dxldt = y .
metric equations, so we consider the vector (dx/dt, dy/dt) or : i + d: j with com- dyldt= -x for { dyldt = -x and solutlOn
curves
r
[
I
l
ever, we begin with a second-order ODE of the form d 2x/dt 2 + b dx/dt + ex = J(t) and 3. dyldx = xly 6. dyldx = x 2 - Y
y
would like to write the equation as a system of first-order ODEs. We do this by letting
dx/dt = y and by differentiating this equation with respect to t to obtain d 2x/dt 2 = dy/dt.
Solving the second-order ODE for d 2x/dt 2 , we find d 2x/dt 2 = -b dx/dt - ex + J(t) so
that dy/dt = -b dx/dt - ex + J(t). Replacing dx/dt in this equation with y, we obtain
the following equivalent system of fIrst-order ODEs:
dx/dt = Y x x
{ dy/dt = -by - ex + J(t),
By writing the second-order ODE as a system of fIrst-order ODEs, we investi-

gate the behavior of the solution of the second-order ODE by observing the behavior
in the direction field and phase portrait of the corresponding system. A similar pro-
cedure is used to transform an ODE of degree greater than two into a system of fIrst-
Figure 1.19
order ODEs. Figure 1.22
4. dyldx = x 2 - Y
y 7. dyldx = x - y
y
2
1.5
0.5
-2 -l,5 -1-0.5 0.5 1 l,5 2

EXERCiSES '6.2 -0.5
0M!it -1
-1.5
-2
In Exercises 1-4, use the slope field to determine if the indi- 2. dyldx = x 2 +/ Figure 1.20
cated path is that of a solution to the differential equation. y Figure 1.23
1. dyldx = -ylx In Exercises 5-8, use the slope field to sketch the solutions that
pass through the given points.
y
5. dyldx = xly
y
Figure 1.18
Figure 1.17 ii Figure 1.21 Figure 1.24
l
9. Graph tbe slope field fory' = siny. Determine limx-oo (a) For given in.itial conditions, do you think that so- (a) For given initial conilltions, do you think that so-
y(x) if yeO) = Yo, where (a) Yo = -31'/2; (b) Yo = .15 -->.~' lutions of the system are sino.ilar or ilifferent? Why? lutions of the systems are similar or illfferent? Why?
- 1'/2; (c) Yo = 1'/2; (d) Yo = 31'/2. (b) Figure 1.27 sbows the illrection field associated (b) Figure 1.28 shows the illrection field associated
10. Graph the slope field fory' = sinx. Does limx_ooYex) with each system. Use the direction fields to graph the witb each system. Use the illrection fields to graph the
exist for any initial condition yeO) = Yo? Solve the solutions that satisfy these initial conilltions: solutions that satisfy these initial conilltions:
ODE and find limx_ooY(x). Does this match the graph- (i) x(O) = 0.5, yeO) = 0; (ti) x(O) = -0.25, yeO) = 0; (i) x(O) = 0.5, yeO) = 0.25; (ti) x(O) = -0.25, yeO) =
ical result? Hi IS x (iii) x(O) = 0, yeO) = 0.75; and (iv) xeD) = 0, yeO) = -0.5; (iii) x(O) = -0.5, yeO) = 0.75; and (iv) xeD) =
-IS -10 -5
*11. Graph the slope field for y' = e -x. Does Iimx _ oo y(x) -0.5. 0.75, yeO) = -0.5.
-5,
exist for any initial condition yeO) = Yo? Solve the (c) How do your graphs affect your conjecture in (a)? (c) How do your grapbs affect your conjecture in (a)?
ODE and find limx_ooY(x). Does this match the graph- -10 24. Consider the systems (d) For each system, use your graphs to calculate
ical result? Iim,_oo x(t) and Iim,_oo yet), as long as not both x(t)
, -IS
:=-t
l
1 x and yet) are the zero function.
12. Plot the slope field for y' = Does Iim x_ooY(x)
- 2 -- .
and
x + I dy 25. (Competing Species) The system of equations
exist for any initial condition yeO) = Yo? Solve the Figure 1.26 di= -y
ODE and find Iimx_ oo yex). Does this match the graph-
ical result? In Exercises 15-20, write the second-order equation as a sys-
tem of first-order equations. y
In Exercises 13-14, use the direction field of the given system d 2x

to sketch the graph of the solution that satisfies the .indicated 15. dt 2 +4x=0
initial conditions. Determine lim,_oox(t) and Iim,_ooy(t) in each 2 0.5 0.5
case (if they exist). 16. d x _ 5 dx = 0
dt 2 dt
13. System: {x; = y, (a) (x(O) = 0, yeO) = 2); 2
I
y =x *17. d x2 +4 dx + 13x = 0 -1 -0.5 0.5 1 x -I -0.5 0.5 1 x
(b) (x(O) = -2;y(0) = OJ; (c) (x(O) = -2,y(0) = 2) dt dt
d 2x dx -0.5 -D,S
18. dt' - 6 dt + 7x = 0
'4
y 19.
20. x"
2
d x
di2 + 16x = sin t
+ 4x' + 13x = e-'

I Figure 1.27
-1 ~I
2 *21. (a) Use a computer algebra system to solve dy/dx =

e-x'; (b) Graph the solution to the IVP dy/dx = e-x',
t y y
yeO) = a for a = -2, -1,0, 1,2. (c) Graph the slope
2 4 x field of the differential equation together with the so-
-4 -2
lutions in (b). Do the solutions appear to matcb the
results described at the beginning of the section? 0.5 0.5
22. (a) Use a computer algebra system to grapb the slope

field for dy/dx = sin(2x - y). (b) Use the slope field
-4 to graph the solutiony(x) that satisfies the in.itial con- "':1 -0.5 0.5 i x -I -0.5 0.5 I x
dition yeO) = 5. Does Iimx_oo y(x) appear to exist?

Figure 1.25 -0.5 -0.5
23. Consider the systems
: =-y : =-y -I -I
14. System.
.{x'y ' =43xx -- y2y.• (x(O) -_ 0, yeO) -_ 5}.
= 1 dy
-=x
dt
and
1
dy
-=-x
dt Figure 1.28
l
22 Chapter 1 Introduction to Differential Equations Chapter 1 Review Exercises 23
(b) Figure 1.30 shows the direction field for the sys- d"y d"-'y Trivial Solution
dx = x(a - b,,, - b-,y) + an_leX) + ...
dt tem if a = 1, h, = I, b, = 1, e = 0.67, d, = 0.75, and a"(x) dx" dx"-' y(x) = o.
d2 = 1. (i) Use the direction field to graph various so-
1 dy
dt = y(e - d,x - d2 y) lutions if both x(O) and yeO) are not zero. (li) Use the
direction field and your graphs to determine the fate
+ a2(x)
d 2y
dx 2 + aleX)
dy
dx + ao(x)y = f(x)
General Solution of an Nth Order Linear Equation
A solution that depends on n arbitrary constants and includes
where a, bb b2 , c, db and d2 represent positive con~ of the species with population yet). What happens to all solutions of the equation.
stants that can be used to model the population of two the species with population X(I)? Order of an Equation
species, represented by x(t) and yet), competing for The order of the highest derivative in the differential equation
26. (a) Use a computer algebra system to solve dy/dx =
common food supply.
e-x'. (b) Graph the solution to the IVP dy/dx = e-x',
is called the order of the equation.
ISection 1.21
(a) Figure 1.29 shows the direction field for the sys- yeO) = a for a = -2, -1,0,1,2. (c) Graph the slope Solution
tem if a = I, b, = 2, b2 = 1, e = 1, d, = 0.75, and d2 field of the differential equation together with the so- A solution of a differential equation on a given interval is a Slope Field
= 2. (i) Use the direction field to graph various solu- lutions in (b). Do the solutions appear to match the function that is continuous on the interval and has all the nec-
tions if both x(O) and yeO) are not zero. (ii) Use the di- results described at the beginning of the section? Phase Portrait
essary derivatives that are present in the differential equation
rection field and your graphs to approximate lim,_oo such that when substituted into the equation yields an identity A collection of graphs of solutions to the system in the xy-
x(t) and lim,_oo yet). for all values on the interval. plane.
Explicit Solution Direction Field
1 f 1.f A solution given as a function of the independent variable. A collection of line segments that indicate the slope of the
tangent line to the solution(s) of a differential equation.
Implicit Solution
o.s .. o.s·;· A solution given as a relationf(x, y) = o.
0.6-- 0.6;-
0.4-· CHAPTER 1 REVIEW EXERCISES
0.2--
L >-- L >-- In Exercises 1-5, determine (a) if the equation is an ordinary 8. y" + 4y' - 5y = 0, Y = e-'x, Y = eX
0.2 0.4 0.6 0.8 1 x 0.2 0.4 0.6 O.S 1 x differential equation or partial differential equation; (b) the or-
der of the ordinary differential equation; and (c) if the equation *9. y" - 6y' + 45y = 0, y = e 3x (cos 6x - sin 6x)
Figure 1.29 Figure 1.30 is linear or nonlinear. 10. x 2y" - xy' - 16y = 0, y = Ax' + Bx- 3
1. y' = Y 11. x 2y" + 3xy' + 2y = O,y = x-'(cos(lnx) - sin(\nx»
2. aux + U t = 0, a constant d 2y dy
CHAPTER 1 SUMMARY 2
12. dx 2 + 2 dx + 2y = x, y = (x - 1)/2
d d
3. ,;, +2 ~ +y=O *13. y" - 7y' + 12y = 2, y = Ae 3X + Be 4x + 1/6
Concepts & Formulas
4. mX + Icc = sin I, m and k constants
a¢ a ¢ a¢ 2 2
ISection 1.11 independent variable, the equation is called an ordinary dif- 5. ax ay;:
ax2 =
In Exercises 14 and 15, verify that the given implicit function
satisfies the differential equation.
ferential equation.
In Exercises 6-13, verify that the given function is a solution
14. (2x - 3y) dx + (2y - 3x) dy = 0, x 2 - 3xy + y2 = I
Differential Equation
15. (y cos(xy) + sin x) dx +x
!
Partial Differential Equation to the corresponding differential equation. (A and B denote con- cos(xy) dy = 0,
An equation that contains the derivative or differentials of
A differential equation that contains the partial derivatives or stants.) sin(xy) - cos x = 0
one or more dependent variables with respect to one or more
independent variables. differentials of one Or more independent variables with re-
dy .
spect to more than one independent variable. 6. dx + Y cos x = O,y = e-'"U
l
Ordinary Differential Equation
In Exercises 16-19, Tmd a solution of the differential equation.
If a differential equation contains only ordinary derivatives Linear Ordinary Differential Equation dy -x'
(of one or more dependent variables) with respect to a single A linear ordinary differential equation is of the form 7. : - y = sin x, y = (eX - cos x - sin x)/2 16. dx = xe
l
I
2
17. ! =xzsinx
perature (I = 0) on the surface of the hall? What hap-
pens to the temperature as t .... oo? !
18 dy = 2x - X + I
. dx
2
(x - 1)(x 2 + I)
26. The displacement (measured from x = 0) of a mass
attached to the end of a spring at time t is given
r
dy
19. dx = Vx 2 -
x 2
I
by x(t) = ± e- t ( cos V3st + ~ sin V3st)' Show
that x satisfies the ordinary differential equation x" +
2x' + 36x = O. What is the initial displacement of the
In Exercises 20-22, use the indicated initial or boundary con-
mass? What is the initial velocity of the mass?
ditions with the given general solution to determine the solu-
tion(s) to the given problem. 27. For a particular wire oflength I unit, the temperature
u at time t hours at a position of x feet from the end
First-Order Equations
dy 2
20. dx + 2y = x , yeO) = I, (x = 0) of the wire is estimated by u(x, t) = e-n'la sin
7TX - e- 4 n'la sin 27TX. Show that u satisfies the heat
y =.l _.lx + .lx 2 + Ae- Zz equation U t = ku=. What is the initial temperature
4 2 2 (t = 0) at x = I? What happens to the temperature at
21. y' + 4y = x, yeO) = I, y (1T/4) = 1T1l6, each point in the wire as t --+ oo? f
Y = x/4 + A cos 2x + B sin 2x
28. The height u of a long string at time t and position x
22. x 2y' + 5xy' + 4y = O.y(l) = l,y'(I) = 0, where x is measured from the middle of the string t
y = A,x-2 + Bx-zlnx (x = 0) is given by u(x, t) = sin x cos 2t. Show that u ~ ..
satisfies the wave equation Un = 4u xx What is the ini-
In Exercises 23 and 24. solve the initial-value problem. Graph tial height (t = 0) at x = O?
o
i
the solution on an appropriate interval. *29. Show that u(x, y) = tan -ley/X) satisfies Laplace's
dy 2.
equation Uxx + Uyy = O.
23. dx = cos x sm x, yeO) = 0
30. The slope field of y' = 2(y - yZ) is shown. Sketch
dy 4x - 9 the graph of the solution that satisfies the indicated
24. dx = 3(x _ 3)213' yeO) =0 initial condition. Also, determine if limx_~ y(x) ex-
ists. (a) yeO) = 0.5; (b) yeO) = 1.5; (e) yeO) = -0.5.
25. The temperature on the surface of a steel ball at time
e will devote a considerable amount of time in this course to develop-
t is given by u(t) = 70e -la + 30 (in OF) where k is a
positive constant. Show that u satisfies the first-order
equation du/dt = -k(u - 30). What is the initial tern-
31. The direction field for {x' = x, y' = 2y} is shown. Is
it possible to select initial conditions so that limt_~
x(t) = 0 and limt_~ yet) = O?
y
W ing explicit, implicit, numerical, and graphical solutions of differential
equations. In this chapter we discuss first-order ordinary differential
equations and some methods used to construct explicit, implicit, numerical, and
graphical solutions of them. Several of the equations and methods of solution
2 discussed here will be used in later chapters of the text.
1.5 0.5
'1 ~ Separable Equations
0.5 -1 -0.5 os 1 x Many interesting problems involving populations are solved through the use of first-
order ordinary differential equations. For example, let yet) be the fish population size
-3 -2 -1 2 3 x -0.5
-0.5
(in tons) at time t (years) and suppose that the population has birth rate by(t) and mor-
tality rate my(t) where we are assuming that these rates are proportional to yet). If there
-1 -1 are no other factors affecting the rate of change in the population, dy/ dt, then dy/ dt
equals the rate at which the population increases (the birth rate) minus the rate at which
Figure 1.31 Figure 1.32 it decreases (the mortality rate), or
25
l
26 Chapter 2 First-Order Equations 2.1 Separable Equations 27
y(t) = Ce kt .
: = (Birth Rate) - (Mortality Rate).
To solve the initial-value problem, we must choose C so that yeO) = Yo. Applying the
Mathematically, we represent this relationship with the differential equation initial condition, we find that yeO) = Ce k ' O = C, which indicates that C = Yo. There-
fore, the solution to the IVP is
dy dy
dt = by - my = (b - m)y or dt = ky, yet) = yoe
kl
•
where k = b - m> O. If yeO) = Yo is the initial population size, then we find yet) by Note: If we had not assumed that y(t) 2: 0 for all t, then we simplify IYI = Ce k ' with
solving the initial value problem Y = ±Cek '. Therefore, by letting C2 = ±C, we obtain y(t) = C,e k ', which is equivalent
to the result obtained above.
dy
dt = ky, yeO) = Yo· The equation dy/dt = ky is a member of a special category of separable equa-
tions because it has the form y' = fey). We find that, although we are able to solve
This is known as· the Malthus model and is due to the work of the English clergyman many equations of this form, we can learn much about the behavior of solutions with-
and economist Thomas R. Malthus. Notice that the ODE can be written as out actually solving the differential equation.
dy = kdt.
y
M
W
Equilibrium Solutions of y' = fey)
We say that this ODE is separable because we are able to collect all of the terms in- An equilibrium solution of y' = fey) is a (constant) solution of the ODE that satis-
volving y on one side of the equation and all the terms involving t on the other side of
the equation.
fies fey) = O. Consider the equation dy = 2y - y2 or ~= dt. (This is a non-
dt 2y - Y
linear ODE because it involves the termy2. We learn more about the differences in lin-
ear and nonlinear equations in Section 2.3.) Instead of solving by separating variables,
Definition 2.1 Separable Differential Equation
let's use a graphical approach to determining properties of solutions. First, we locate
A first-order differential equation that can be written in the form g (y)y' = f(x) the equilibrium solutions of the equation by solving 2y - y2 = 0, or y(2 - y) = O. The
y=2
or g(y) dy = f(x) dx is called a separable differential equation_ roots of this equation are y = 0 and y = 2, so the equilibrium solutions are yet) = 0
andy(t) = 2. Notice that each ofthese functions satisfies dy/dt = 2y - y2 because dy/dt
= 0 andy(2 - y) = O.
After we place the ODE in this form, we solve through integrating each side with re- Next, we investigate the behavior of solutions on the intervals y < 0, 0 < Y < 2,
spect to the indicated variable. This gives and y > 2 by sketching the phase line. After marking the two equilibrium solutions on
the vertical line in Figure 2.1, we determine the sign of dy/dt on each of the three in-
tervals listed above. This can be done by substituting a value of y on each interval into
Figure 2.1 Phase line for
fey) = 2y - y2. Forexample,f(-I) = -3, sody/dt< 0 ify< O. We use anarrowdi-
dy =2y- / rected downward to indicate that solutions decrease on this interval. In a similar man-
Integration yields dt
ner, we find thatf(l) = I andf(3) = -3, so dy/dt> 0 if 0 < Y < 2, and dy/dt < 0 if
lniyi = kt + CJ, y > 2. We include arrows on the phase line directed upward and downward, respec-
where C 1 is a constant of integration. When possible, we like to solve for yet). In this tively, to indicate this behavior. Based on the orientation, solutions that satisfy yeO) =
case, we use the exponential function and its properties: a, where either 0 < a < 2 or a> 2 approaches y = 2 as t --> +00, so we classify
elniyl = e kt + c , = ekte c ,. y = 2 as asymptotically stable. Solutions that satisfy yeO) = a where a < 0 or 0 <
a < 2 move away from y = 0 as t --> +00. We say that y = 0 is unstable because there
Sinlplifying this result gives us are solutions that begin near y = 0 and move away from y = 0 as t --> +00. In Figure
2.2, we graph solutions to this equation (in the ty-plane) for yeO) = a, a = -0.5, -0.4,
iyi = Ce kt ,
-0.3, -0.2, -0.1,0.0,0.25, 0.5, 0.75, 1.0, 1.25, 1.5, 1.75, 2.0, 2.25, 2.5, 2.75, 3.0,
C
where we replace the constant e ' with the constant C. Finally, because yet) represents Figure 2.2 Solutions to 3.25,3.5,3.75, and 4.0 to illustrate this behavior. Notice that solutions with initial value
population size, yet) 2: 0 for all t, and we eliminate the absolute value to obtain a gen- dy near y = 0 move away from this line while solutions that begin on 0 < Y < 2 or y > 2
eral solution of the ODE, di = 2y - y2
move toward y = 2 as t --> +00.
==~- ~.~-------------------------------------
Note: Sometimes, we refer to equilibrium solutions as steady-state solutions, be- Application of the natural logarithm yields
cause over time, the solution approaches a constant value that no longer depends
on time t. For example, if we interpret the solution y(t) of the differential equation 2.5 or
to represent the size of a population, then if y(O) = a where either 0 < a < 2 or a > 2,
we expect the population to approach 2 as t --> +00.
To find the value of C 1 so that the solution satisfies the condition yeO) = 0, we use
the equation e-Y = -e2x/2 - C1• Applying yeO) = 0 yields 1= -112 - C lo so
C 1 = -3/2 and thus the solution to the initial-value problem is
Show that the equation dy/dx = -x/y is separable and solve this ODE. -3-2.5-2-1.5-1-
-0.5
The domain of the natural logarithm is (0, +00), so this solution is valid only for
: :I Solution In this case, we see that the equation is separable by expressing it as Figure 2.4 the values of x such that 3 - e2x > 0, which is equivalent to e2x < 3 and has solu-
xdx = -ydy. tion x < In(3)/2 = 0.5493. Notice that the graph of the solution shown in Figure 2.4
passes through the point (0, 0). If it had not, we would check for mistakes in our
Integration yields
solution procedure.
x2 _ y2
T- -T+CI'
In some cases, a separable differential equation may be difficult to solve because of
Multiplying by 2 and simplifying gives us x 2 + y2 = 2C1 . If we let k = 2CI, then
the integration that is required.
the solutions satisfy the equation
x 2 + y2 = k,
Figure 2.3 Graph of
x 2 + y2 = k for various values
of k
so that an implicit solution is circles centered at the origin of radius
in Figure 2.3.
Yk, as shown
o "d""'·'Solve dy/dx = (x 4 + 1)(y4 + I), yeO) = o.
2
Is x + y2 = 0 an implicit solution olthe equation? Explain. : ]I Solution When we separate variables, we obtain
As we have seen, separable differential equations can be stated along with an ini-
~=(x4+ I)dx.
tial condition. If this is the case, we find a general solution to the differential equation y4 +I
using the separation of variables technique and we then apply the initial condition to
determine the unknown constant in the general solution. The right side of the equation is easily integrated. However, the left side presents a
challenge. Using a computer algebra system, we find that
= _1_[tan- (2Y- V2) + tan-I(2Y V2

"i'd','.' f~
+y4 2V21 V2
I + V2)]
Solve the initial-value problem dy/dx = e 2x + y , yeO) = o.

+ _1_ Inll + V2y + y21 + C
4V2 -I + V2y - y2 .
: 'I Solution If we write the equation as dy/dx = e 2x eY, we see that the variables are
separated as Therefore, we find that
e-Ydy = e 2xdx.
_ I[
2V2 tan
- v'2) + tan _1(2Y V2
_1(2Y V2 + V2)]
Integration gives us -e-Y = e 2x/2 + Clo and simplifying results in
+ _1_ Inll + V2y + y21 + C = ~ x5 + x.

4V2 -I + V2y - y2 5
30 Chapter 2 First·Order Equations
Substitution of x = 0 andy = 0 (from the initial condition) into this equation indi·
I EXERCISES 2.1
2.1 Separable Equations 31
cates that C= 0 because tan- 1(-Y2/Y2) = -'lT14, tan- 1(Y2/Y2) = 'lT14, and
Inl-II = O. This means that the solution to the IVP is
_ 1[ tan
,r.::
_1(2Y ,r.::
- Y2) +tan _1(2Y ,r.::
+ Y2)] In Exercises 1-30, solve each equation. (A computer algebra 2
25. x sin(x ) dx =
cos
vYvY dy
2v2 v2 v2 system or table of integrals may be useful in evaluating some
integrals.)
1 I n II + Y2y + y21 =-x
1 5 +x 26. x-2 dx=(I_.!.)2..!...dY
+--
dy 6x 2 x 2 _ 4x + 3 y y2
4Y2 -I + Y2y - y2 5 .
1. dx = 7y3
2. .!.x- l12 dx + y2 dy = 0
2
*27. cosy 2 dy = sin3 x cos x dx
We can gain insight into the graph of this solution by observing the slope field of dy 3y7 dy I (1 - Stny)
the ODE. For example, the slope of the tangent line to the solution passing through *3. dx =7 4. dx = x 2 (8 + 9y2) dy (S - 2 cos X)3 sin x cos' y
(0, 0) at (0, 0) is 28. dx = siny
5. (6 + 4x 3
) dx + (s + ;.) dy = 0
~
'~In 31y
= (0
4 4
+ 1)(0 + I) = 1. 29. _V_tn_XX dx=~dy
The slope at (0, I) is dyldx = (04 + 1)(14 + 1) = 2, and that at (0, 2) is dyldx =
6. (:9 -:3 +X7)dx+(9+y-2_4Y')dY=0
x
30. dy = 5-
y
x
4
(0 + 1)(24 + 1) = 17. As we see, slopes become steeper as the y·coordinate in· *7. 4 sinh 4y dy = 6 cosh 3x dx dx y2
creases [or decreases because we obtain the same values for dyldx at (0, -1) and
(0, -2)]. Along the x·axis, we obtain the same slopes at (I, 0) and (2, 0) as we did
8. tx -213 dx = (ys - 6y2 + 8) dy In Exercises 31-42, solve the initial·value problem. Graph the
at (0, 1) and (0, 2) because dyldx is symmetric in x and y. We generate the slope ,r -1 solution on an appropriate interval.
9. (x 2 + 2vx) dx = T y - Sl2 dy
field with a computer algebra system in Figure 2.5. To understand the solution of
31. t 3
= x ,y(0) = 4
the IVP, we begin at the point (0, 0) and follow the tangent line slopes shown in the
slope field for x > 0 and for x < O. Doing so, we obtain the curve shown in Figure 10. ;2 dx = ( ~ + vY) dy
2.6. We notice that y approaches +00 as x approaches a value near 0.9, and y *11. 3 sin x dx - 4 cos y dy = 0 32. !=COSX,y(-i)=-1
approaches -00 as x approaches a value near -0.9. We also note that given any
12. cosy dy = 8 sin 8x dx *33. dx = cos x dy, yeO) = 2
initial condition, the solution to the corresponding IVP becomes unbounded (i.e., y
approaches::!: (0) as the x-coordinate moves away from O. 13. (SX S - 4 cos x) dx + (2 cos 9y + 2 sin 7y) dy = 0 34. sin2 y dy = dx, yeO) = 0
14. 9 cosh 9y dy = 4 sinh 4x dx dy Vx
y y *15. (cosh 6x + S sinh 4x) dx + 20 sinh y dy = 0 35. dx = y' yeO) = 2
16 dy = e2YelOx
'dx 36. ! = A,Y(l) = 2
17. (10 + 7e -3x) dx - (0' - 8y 3) dy = 0
*37 dy = ~ yeO) = -2
18. sin2 x dx = cos 2 Y dy . dx y+ I'
*19. (3 sin x - sin 3x) dx = (cos 4y - 4 cos y) dy
20. dy = sec x
2 38. != e X - Y, yeO) =0
dx sec y tany
39. dy = L yeO) = e
21. (2 - :2) dy + 4 2 dx cos X = 0
dx Iny
22. dy = x
3 40. != x sin(x
2
), y(V:;;:) =0
+ 9)
Figure 2.5 Slope field for Figure 2.6 Solution to
dx yY(l - y2)(X'
*41 dy
. dx
=~
I +x '
yeO) = I
*23. tan y sec2 y dy + cos 3 2x sin 2x dx
dy = (x'
dx
+ 1)(y4 + I) ! = (x' + 1)(y4 + 1), yeO) = 0
24. dY =I+2o'
= 0
dy sin x
42. dx = cosy + I ,yeO) = 0
dx o'x Inx
43_ Find an equation of the curve that passes through the (c) How does the solution cbange for various initial dy dy Optical resonators are used to build up large
. y-2 conditions? (Hint: First determine how w and eo
50. dx = 3y 51. dx = -y
f"Ield intensities with moderate power inputs. A meas-
pomt (0, 0) and has slope - x _ 2 at each point (x, y)
affect the solution. Then, determine how ro affects dy Z ure of this property is the quality factor Q, where
on the curve. the solution. Try graphing the solution if w = 2, 52. dx = y - Y *53. : = 16y - 8y2
field energy stored by resonator
eo = 0, and ro = !, 1, and~. What happens if you
44. A differential equation of the form dy/dx = dy 2 Q = wX power dissipated by resonator
increase w? What happens if you increase eo?) 54. : = 12 + 4y - yZ 55. dx =y -5y+4
f(ax + by + k) is separable if b = 0. However, if
b * 0, the sUbstitution u(x) = ax + by + k as the new In Exercises 46-49, without actually solving the equations, 56. (Lasers) The variables that characterize a two-level
A technique called "Q-switching" is used to generate
short and intense bursts of oscillation from lasers. This
dependent variable yields a separable equation. Use match each differential equation in Group A with the graph of laser are listed in the follOWing table. is done by lowering the quality factor Q during the
this transformation to solve (3) dy/dx = (x + Y - 4)z its direction field and two solutions in Group B. pumping so that the inversion N2 - NI builds up to a
and (b) dy/dx = (3y + I)'. cf> Total number of photons in optical resonator high value without oscillation. When the inversion
I Length of resonator reaches its peak, the factor Q is suddenly restored to
*45. Let w> 0. (3) Show that the system Group A
N, Number of atoms per unit volume in Level 1 its ordinary value. At this point, the laser medium is
46. dy = y' +1 47. dy = x' + 1 N2 Number of atoms per unit volume in Level 2
well above its threshold, which causes a large buildup
dx = x(1 - r) - wy dx x' +1 dx y' + I V Volume of optical resonator
of the oscillation and a simultaneous exhaustion of the
17t
dt r=YxZ+yZ n Total inversion given by (N2 - N I ) V
d ' inversion by simulated transitions from Level 2 to
= y(1 - r) + wx dy dy nj Initial inversion
48. dx = (x' + I)(y' + 1) 49. dx n, Levell. This process converts most of the energy that
(;c4 + 1)(y4 + 1) Total inversion at threshold
was stored by atoms pumped into Level 2 into pho-
Ie Decay time constant for photons in passive
can be rewritten as the system
Group B resonator tons, which are now in the optical resonator. These
c Phase velocity oflight wave photons bounce back and forth between the reflectors
dr (a) (b)
- = r(1 - r) v light wave frequency with a fraction 1 - R escaping from the resonator with
dt
1
h magnetic f"Ield each bounce. This causes a decay of the pulse with a
de
- = lV photon lifetime
dt Suppose that a light wave of frequency v and inten-
sity Jv propagates through an atomic medium with Nz nl
by changing to polar coordinates {X = r cos e. atoms per unit volume in Level 2 and NI in Levell. Ie = c(l - R)"
y=rsme
If N z > N" then the medium is amplifying as is the The quantities of cf> and n are related by the differen-
(b) Show that the solution to r _2X x
case with lasers. On the other band, if N, > N2 , then tial equation
the medium is absorbing. These two situations are pic-
dr = r(1 - r)
dt tured in Figure 2.7.
de
-=w Amplifying medium (Nz > N I )
is
1 de
r(O) = ro, 8(0) = eo
(e) (d)
r e'
r(t)= 0 "
{ (1 - ro) + roe tal
e(t) = wi + eo Absorbing medium (N2 < N l )
and the solution to
dx = x(1 - r) - wy
dt r = Y XZ + y.2
is
1
dy
dt = y(l - r) + wx
,
Legend:
(b)
.;::. Atom in upper Atom in lower

In Exercises 50-55, find and classifY (as asymptotically stable
fx = r(t) cos 8(1) or unstable) the equilibrium solutions of the f"Irst-order equa-
State 2 State I
)y = r(l) sin e(I)' tions. Figure 2.7
.. _--------------------------------
(a) Solve this equation to detennine the total number 58. (pollution) Under nonnal atmospheric conditions the (c) For each pair of values in the previous table, dy JI ] KEo
density of soot particles N(t) satisfies the differe~tial
of photons in the optical resonator 1> as a func- dt = "LV(A o - y) Bo = y(Ao - y),y(O)
sketch the graph of N(t) if N(O) = No for No =
tion of n (where 1> depends on an arbitrary con- equation 0.01, 0.05, 0.1, 0.5, 0.75, I, \.5, and 2. Regard-
stant). = Yo·
less of the initial condition N(O) = No, what do
(b) Assuming that the initial values of 1> and n are 1>0 dN = -k"'+kN you notice in each case? Do pollution levels seem (In other words, the reaction stops when substance
dt "'. d,
and no at t = 0, show that 1> can be simplified to to be more sensitive to kc or kd? Does your result A is exhausted.) Solve this initial value problem.
obtain where kc' called the coagulation constant, is a con- make sense? Why? Calculate ,l!.r;;
yet).
stant that relates how well particles stick together; and
(d) Show that if kd > 0, lim,...,oo N(t) = kik,. Why is (d) Suppose that K = 0.2, V = I, Ao = I, Bo = 10,
1>=~ln-"--+.L(n
2n
+n)+-'-
2°'l'0
k d , called the dissociation constant, is a constant that
the assumption that kd > 0 reasonable? and Yo = 0.1. Compare the solution found in (b)
0 relates to how well particles fall apart. Both of these to that found in (c). Do the same for the parame-
constants depend on temperature, pressure, particle (e) For each pair in the table, calculate limt-+oo N(t)
(c) If the instantaneous power output of the laser is = kikc. Which situation results in the highest pol- ter values K = 0.2, V = I, Ao = I, Bo = 50, and
SlZe, and other external forces. t
given by lution levels? How could the situation be Yo = 0.1. How does the value of Bo affect the ac-
(a) Rewrite the equation dNldt = -kJV' + kdNin the changed? curacy of the approximation in (c)?
P=~ fonn
t, ' 59. (Kinetic Reactions) Suppose that substances A and B,
In many cases, computer algebra systems can be used to per-
show that the maximum power ontput occurs called the reactants, combine to fonn substance C. The
-kJV' + kdN dN = dt fonn the integration and algebraic simplification associated
when n = nt • (Hint: Solve aPlan = 0.) reaction depends on the concentration of the reactants
with a separable differential equation and can sometimes solve
and use partial fractions to show that because the rate of the reaction is proportional to the
*57. (Destrnction of Microorganisms) Microorganisms them. Similarly, initial-value problems can often be solved
product of their concentrations. When substances A
can be removed from fluids by mechanical methods k" quickly with a computer algebra system.
~ e:" + C
and B are placed in the test tube and begin to com-
such as filtration, centrifugation, or flotation. How- N(t) = bine to fonn C, the amounts of A and B diminish and
ever, they can also be destroyed by heat, chemical kd In Exercises 60-66, find a general solution of the given equa-
the reaction stops when either A or B is no longer pres-
agents, or electromagnetic waves. The fermentation in- tion and graph various solutions on the indicated interval.
where C represents an arbitrary constant. ent. Let yet) represent the amount of substance C in
dustry is interested in improving this process of ster- the test tube at any time. If the initial amounts of suh- dy
ilizing media. An important component in the desi"" (b) Find C so thatN(t) satisfies the condition N(to) = 60. dx = Y cos x; [0, 4'1T]
stances A and B are Ao and Bo and if the volume in
of a sterilizer is the kinetics of the death of micro~r No· the test tube is V, then the concentration of A at any
ganisms. The following table lists typical values of dy _ • r.---z .
time is.L. (Ao - y) and that of B is.L. (Bo - y). There- 61. dx - V I - y- sm x; [0, 4'1T]
The destruction of microorganisms by heat in- kc and kd ·
V V
dicates loss of viability as opposed to physical de- 1+ y2
fore, we detennine yet) by solving
struction. This destruction follows the rate of reaction 62. cos x dx = ~ dy; [0, 10]
dN 1£ = J.L. (Ao - y)][.L. (B - y)]
d[= -kN, kc kd dt "L V V 0
*63. eY cos y dy = V x2
2 dx; [-2, 2]
163 5 K 9 -x
where kis the reaction constant (min-') and is a func- = V2 (Ao - y)(Bo - y).
64. Solve the initial-value problem
tion of tem~erature, N is the number of viable organ- 125 26 2
Isms, and t IS time. * (a) Identify the equilibrium solutions of the equation. 1£ - x
Microbiologists use the tenn decimal rednc- 95 57 Assuming that Bo > A o, classify each as stable or dx - V97 eYcosy
tion time D to indicate the time of exposure to heat
during which the original number of viable microbes 49 85
unstable. Generate the direction field for the equa-
1 yeO) = 0
~
tion with Ao = I, Bo = 2, K= I, and V= I to
is reduced by one-tenth. If N(O) = No, rmd N(t). Use verify your results. What bappens when either yet) and graph the solution on an appropriate interval.
the fact that N(D) = t-oNo to find D in tenns of k. = Ao or yet) = Bo?
65. How does the graph of the solution to the initial-value
(b) Assume that yeO) = Yo and solve the initial value
problem
problem to determine yet). Calculate lim y(t). How
does this limiting value relate to th~physical sit- , CY
y=-
2
uation described by the problem? x
... S. Aiba, A. E. Humphrey, N. F. Millis, Biochemical Engineering, Second Edition, Academic Press (1973), pp. 240-242.
{
(c) If Bo is much larger than Ao, then we may ap- y(I) = I
t Chr. Feldennann, H. Jander and H Gg W e "S t p rti I C [. . proximate the solution to this problem by solving
of Research in Physical Chem;'stry and Chern;:Z ~hYs~~. V:l~: 1~~~:::~ ~f9~~;~;~~;;~1:~~/Air Flames at 10 bar," International Journal the simplified equation change as c changes from - 2 to 2?
2.2 First-Order linear Equations 37
36 Chapter 2 First-Order Equations
66. (a) Assume that y(l) > 0 and r I

f(t) dt exists for
is
yet) = 2e11J(u) du.
Multiplying the ODE by !L(t), we obtain
efp(l)dl d: + p(t)eJp(l)dly = q(t)eJp(l)dl.

I. Show that the solution to the initial-value
I ;;,:
(b) Find three functionsf(l) so that (i) y(1) is periodic;
problem (ii) limH~y(l) = 0; and (iii) limH~y(t) = "'.
dY Confirm your results by graphing each solution. However, notice that by the product rule,
di=yf(t)
(e) If c > 0 is given, is it possible to choosef(t) so
{ ~[eJp(l)dy] = ejp(l)dl dy + p(t)eJp(l)dly.
y(l) = 2 that limH~ f(t) = c? Explain. dt dt
This expression is the left side of the previous equation, so we can rewrite the equa-
tion as
12.21 First-Order linear Equations :t[eJP(I)dly] = q(t)eJp(l)dl.
In addition to the assumptions made in the fish population model at the beginning of We solve this equation by integrating each side. We illustrate the solution method in
Section 2.1, suppose that we harvest the population at the rate h. In other words, we
the following example.
remove fish from the population by fishing at the rate h. By modifying the Malthus
model developed in Section 2.1, we tmd that the rate at which the population size
changes satisfies
l'Ei,,','.'
Consider the IVP dy/dt = Y - 112, yeO) = a. In this case, the fish population grows
at the rate of I (ton per year) and fish are harvested (or fished) at the rate of 112
Therefore, we determine the population size by solving (ton per year). (a) Determine the behavior of solutions by drawing the phase line.
(b) Find the solution to the IVP. (e) Investigate the behavior of solutions of the IVP
dy
dt = ky - h, yeO) = Yo, using yeO) = 1/4, yeO) = I, and yeO) = 1/2.
where Yo represents the initial population size. In addition to being separable, this dif- : ') Solution (a) By solving y - 112 = 0, we find that the equilibrium solution is y
ferential equation is classified as a first-order linear differential equation. = 112. We draw the phase line in Figure 2.8. We include only nonnegative values
on the phase line because we are interested in y(t) 2: 0 because y(t) represents
population size. Note that if y > 112, then dy/dt> 0 while if y < 112, then
o dy/dt < O. Therefore, on the interval y > 112, the arrow is directed upward and on
Definition 2.2 First-Order Linear Differential Equation o oS y < 112, the arrow is directed downward. The phase line indicates that
A first-order linear differential equation can be written in the form Figure 2.8 Phase line for solutions to the IVP dy/dt = y - 112, yeO) = a with a > 112 increase as t -'> +00.
dy I Conversely, solutions to this IVP with 0 < a < 112 decrease as t -'> +00.
dy Yt-y=-Z
dt + p(t)y = q(t).
(b) We solve dy/dt = y - 112 as a first-order linear equation by rewriting the equa-
tion as dy/dt - y = -112. The integrating factor is
!L(t) = ejp(l)dl = eJ(-I)dl = e- I ,
Writing dy/dt = ky - h in this form, we have
dy
• .
so the equatlOn can be wntten as e
-I dy
dt - -I
e y = -2I e- I or
dt - ky = -h,
so p(t) = - k and q(t) = - h are constant functions.

:t [e-1y] = -te-'.
We solve a first-order linear ODE by using the integrating factor . Yle
. Ids e - IY = '2e
I -I + C, so
IntegratlOn
!L(t) = efp(l)dl.
38 Chapter 2 First·Order Equations 2.2 First·Order Linear Equations 39
0.25
Y
yet) = "21 + Cet. 0~" Figure 2.12 shows the graph of y = t+ Ce -x' for various values of C. Identify the
-0.25 0.5 1.5 x ~:;;'.: graph of the solution that satisfies the condition yeO) = -7.
-0.5
-0.75
(e) If yeO) = 114, then yeO) = 112 + C implies that C = -114. Therefore, In some cases, we must rewrite an equation to place it in the form of a linear
t - tel.
-1 first-order differential equation.
-1.25
y(t) =
1.5 2 x
Figure 2.9
dy = y _
Solution to
2. yeO) = 2.
We graph y(t) in Figure 2.9. Notice that the population becomes extinct when
t - tel = 0, or t = In 2 = 0.693147 year. [We ignore the portion of the solution
"E''''''OO
dt 2' 4
Solve dt =. I ,0 < t < 7T/2.
curve where y(t) < 0.] We should note that any solution to this IVP withy(O) < 112 dr smt-rtant
y becomes extinct at some time. If yeO) = 1, 1/2 + C = I indicates that C = 112.
4
3.5 Therefore, yet) = t + tel, so y(t) > 0 for all t, as we see in Figure 2.10. In fact,
Fig~re 2.12
J Solution Notice that if t is the dependent variable, the equation is nonlinear
(in t). (Why?) However, solving the equation for drldt yields
2.5
the population grows without bound. If yeO) = 112, then y(t) = 112 as expected be-
dr dr().
cause yet) = 112 is an equilibrium solution. If the population is initially 112 ton, then dt = sin t - r tan t or dt + tan t r = sm t,
1.5 it remains at that level. (See Figure 2.11.)
1 which is a linear equation in the dependent variable r. With pet) = tan t, the inte-
0.5
Figure 2.10
0.5 1.5
Solution to
x
"iI,,"'.1 grating factor is
JL
(t) = eJ<an I dl = e -Inlco, II = _1_, 0 < t <
cos t
7T/2.
dy 1 Solve the initial-value problem ~ + 5x y 4

= x4, yeO) = -7.
Then,
dt = y - 2' yeO) = 1
~ [JL(t)r] = JL(t) sin t
y
: ]Solution We begin by solving the linear equation ~ + 5x 4y = X4 using the
integrating factor eJ5x'dx = eX'. Then, the equation can be written as d[l] r sint
dt cost =cost
0.8
0.6
_I_r = -lnicos
cos t
ti + C
0.4 so that integration of both sides of the equation yields r = -(cos t)ln(cos t) + C cos t.
0.2
Solve the initial-value problem
0.5 1.5 x
~~
A general solution is
Figure 2.11 Solution to = sint-\tant
dy 1 1 {
dt = Y - 2' yeO) = 2 Y= 51. + Ce- x ' . t(l) = 7T/4 .
Graph the solution on an appropriate interval.
We find the unknown constant C by substituting the initial condition yeO) = -7.
This gives -7 = 115 + C, so C = -36/5. Therefore, the solution to the initial-value
problem is o 'ifi,,'·,M'
If a drug is introduced into the bloodstream in dosages of D(t) and is removed at a
rate proportional to the concentration, the concentration cet) at time t is given by
40 Chapter 2 First-Order Equations 2.2 First-Order Linear Equations 41
dC We graph cet) on the interval [0, 30] if k = 0.05, 0.10, 0.15, and 0.20 for 10 = 4, 8,
-=D(t) - kC 12, 16, and 20 in Figure 2.13. From the graphs, we see that as 10 is increased, the
dl
{ CCO) = 0 ' maximum concentration level decreases and occurs at later times, while increasing
k increases the rate at which the drug is removed from the bloodstream.
where k > 0 is the constant of proportionality. *
(a) Solve this initial-value problem. c C
30 30
(b) Suppose that over a 24-hour period, a drug is introduced into the bloodstream
at a rate of 24/10 for exactly 10 hours and then stopped so that to=4
20 to= 8 10::: 4
20
D (t) = { 24/10, i~ 0 :s I :s 10. to = 102= 16 10= 8
" //"
t, 0, If I > 10 :---':', to=20 10= 12
10= 16
What is the total dosage and average dosage over a 24-hour period? 10 10
. /<'~~
(e) Calculate and then graph CCI) on the interval [0, 30] if k = 0.05, 0.10, 0.15, and
0.20 for to = 4,8, 12, 16, and 20. How does increasing 10 affect the concentration . .
of the drug in the bloodstream? How does increasing k affect it? 10 20 30 1 0 10 20 30 1
(a) (b)
C C
30 30
1 Solution (a) After rewriting the equation as ~; + kC = D(I), we find the inte-
grating factor to be /L(t) = e JMr = e kt . Therefore, we obtain ~ [ektc] = ektD(t) so 20 to=4 20

to=4
that integration yields 10 =8
to= 12
ektC = fo
eksD(s) ds or cel) = e -kt fo eksD(s) ds.
10
/0 1
;\"
10= 16
to = 20
24
(b) In each case, the total dosage over a 24-hour period is
r, 24 i
0
Dr (t) dl=
'
10 o 10 20 30 1
fo t;dl = 24; the average dosage is

(c)
Figure 2.13 (a) k = 0.05 (b) k

(d)
= 0.10 (e) k = 0.15 (d) k = 0.20

1
24 - 0
i24
0
1
Dr,(I) dt = 24
f r
0
, 24
t; dl = L
EXERCISES 2.2
(e) To compute cel) = e- kt it eksDr (s) ds, we must keep in mind that Dr (I) is a
o ' ,
piecewise dermed function:
cel) = e-
kt
r
i eksDr,(s) ds =
je-
kt
f
eks 24 ds, if 0 :s 1:S 10
r, to 24
In Exercises 1-24, solve each equation.
9. : - ytanx = e- 2x
fo
0 dy I dy 1 .
e- kt 1. dx +~y=x 2. dx+~y=smx
o eks - ds, if I> 10
to
dy 1
10. dy = (a + x2 .=x6~ ! 10 y) dx
~: kt *3. dx + ~ Y = eX 4 dy +..!..y=xe-x
Ex.-~y=x"
= j ~:
(1 - e- ), if 0 :s 1:S 10
(e-k(t-r,) - e- kt ) if t > to'

'dx
6. Ex.-~y=x'
dx I + x"
x *11.
2 Ex.
1 . dx
dx x" - 4
_ lOx" - I
x(IOx 2 + 7x + 1) y
= 1
10x 2 + 7x + I
• 1. D. Murray. Mathematical Biology. Springet.Veriag (1990), pp. 645-649.

*7. dy = (a + ---:X-)
x- - I
dx
dy
8. dx + Y cot x = cos x 13. Ex._~y=X3
dx 4x' - 9
42 Chapter 2 First-Order Equations 2.2 First-Order linear Equations 43
14 ~_ 1& - 36. Ify = n(x) satisfies the homogeneous equation dyldx 41. Observe Figure 2.13. Describe how the value of k af-
*53. Solve y' - .!..2 y = 5 cos x + 2ex by assuming that
- dx 16x 2 + 25 Y - x + p(x)y = 0 and y = Y2(X) satisfies the nonhomoge- fects the concentration of the drug in the system. Do
neous equation dyldx + p(x)y = r(x), show that y = larger or smaller values cause the drug to remain in yp(x) = A cos x + B sin x + Ce x .
*15_ ~ - _ _9_x_ y =x y,(x) + Y2(X) satisfies the nonhomogeneous equation the system longer? How would the chemist working
dx 9x 2 + 49
dyldx + p(x)y = r(x). for the pharmaceutical company design the drug? In Exercises 54-59, solve each of the following equations us-
dy
16_ dx + (2 cot x)y = cos x *37. (a) Show that if y = n(x) is a solution of dyldx + dy ing the Method of Undetermined Coefficients.
p(x)y = r(x) and y = y,(x) is a solution of dy/dx
42. dx +Y = q(x), where
dy dy + p(x)y = q(x), then y = y,(x) + Y2(X) is a solu- 4,0 :5 x :5 2 54.y'+4y=8cos4x *55_ y' + lOy = 2ex
17. dx+>y=x' 18. dx - >y = x q(x) = { 0, x > 2 ,y(O) = O.
=
tion of dyldx + p(x)y r(x) + q(x). 56. y' - 3y = 27x 2 57_ y' - y = 2e x
*19. dy =_1_ dx
20. dy -x =y
(b) Use the result obtained in (a) to solve t dy
43. dx +Y = q(x), where
58. y' + y=4 + 3ex *59. y' +Y = 2 cos x +x
I
dx y2 + X dy _ 60. (First-Order Linear with Periodic Forcing Func-
dx + 2y = e '+ cos x. x,0:5X:51
22. dx = 3>y2 q (x) = { 0, x > I ' y(O) = 1. tion) Consider the differential equation
21. Y dx - (x + 3y2) dy = 0
dy 1- y' dy
dy
Without actually solving the problems in Exercises 38-40, 44. dx + p(x)y = 0, where dx + cy =f(x)
*23. dp = 13 + E.. 24'
dv
+v=e
-I
match each initial-value problem in Group A with the graph of I,0:5X:52
dl t ds ( where f(x) is a periodic function and c is a constant.
its solution in Group B. px)= { -I,x>2 ,y(O) =2.
The goal of this exercise is to determine if equations
In Exercises 25-34, solve the initial-value problem. Graph the Group A dy of this form have a periodic solution. (a) Solve the
solution on an appropriate interval. *45_ dx + p(x)y = 0, where
38. y' - >y = l,y(O) = I 2,O :5 x :5 I dy I
IVP dx + "2Y = sin x, y(O) = a. For what value of a
dy
25. dx - y = 4e',y(O) = 4 39. y' - sin(27JX)Y = l,y(O) = I p(x) = {4, x > I ' y(O) = 1.
does the IVP have a periodic solution? Graph the slope
dy _ 40. y' - X2~ I y= l,y(O) = I
46. (Method of Undetermined Coefficients) Consider field for this ODE. Describe the behavior of the other
26. dx + Y = e ',y(O) = -I the first-order nonhomogeneous equation solutions. Do they approach the periodic solution
Group B y' +Y = cos x. found in (a) as 1 --> +00 or as t --> -oo? (b) Solve the
*27. dy
dx + 3x'y = e-" ,y(O) = 2 (a) (b) dy I .
We can solve the problem in two parts. First, we can IVP dx - "2 y = sm x, y(O) = a. For what value of a
y solve the corresponding homogeneous equation
dy
28. dx + 2>y = 2x, y(O) = -I 10 10 y' + Y = O. (a) Show that Yh(X) = Ce -x is a general does the IVP have a periodic solution? Graph the slope
solution of this equation. (b) Next, we attempt to find field for this ODE. Describe the behavior of the other
dy Y cos x ( 7T) 4 a solution of the nonhomogeneous equations by guess- solutions. Do they approach the periodic solution
29. dx + -;:- = -x-' Y"2 = -:;;:, x >0 ing the form of the solution. Suppose that this par- found in (a) as x --> +00 or as x --> -oo? (c) Based on
dy
30_ dx
y
+ -;:- = 2e', y(l) = -I, x >0
6
4 4
ticular solution is yp(x) = A cos + B sin x. Substitute
yp into y' + y = cos x to show that the undetermined
your findings in (a) and (b), does the ODE !+ cy
coefficients are A = -112 and B = 112. Then, a gen- = f(x), where f(x) is a periodic function and c is a
*31. dy + ~y = _ x _ y(O) = I 2 eral solution of the nonhomogeneous equation is constant, have a periodic solution? How does the value
dx e'+1 e'+I'
o
y(x) = Yh(X) + yp(x) = Ce- x
- t cos x + t sin x.
of c affect the other solutions?
32_ ~dx + ----l=--

x +4
y = ----l=--,
x +4
y(O) = -4
0.5 1 1.5 2 2.5 3 x o 0.5 1 1.5 2 2.5 3 "
(What integral would need to be evaluated if we used
61. Suppose that a drug is added to the body at a rate r(t),
and lety(l) represent the concentration of the drug in
(c) an integrating factor to solve the ODE?) the bloodstream at time t hours. In addition, suppose
dx
33. dt = x + t + I, x(O) = 2 y
47. Solve y' - y = sin 2x. that the drug is used by the body at the rate /gI, where
10 k is a positive constant. Then, the net rate of change
48. Solve y' +Y = 5e 2x by assuming that yp(x) = A e 2x.
34. ~~ = e" + 2e, 8(0) = 0 iny(t) is given by the equation dy/dt = r(l) - /gI. Ifat
*49. Solvey' +y =
e- x by assuming thatyp(x) = Axe-X. t = 0, there is no drug in the body, we determine y(t)
35. The equation dyldx + p(x)y = 0 is called a homoge- Why don't we choose yp(x) = Ae- X? by solving the initial-value problem
neous first-order linear equation because q(x) = O. 50. Solve y' + Y = 2 - e2x by assuming that yp(x) = dy
(a) Show thaty = 0 (the trivial solution) is a solution. 4 A + Be 2x. dt = r(t) - /gI, y(O) = O.
(b) Show that ify = y,(x) is a solution and k is a con-
stant, y = /gI,{x) is also a solution. (c) Show that if 51. Solve y' - 5y = x by assuming that yp(x) = Ax + B.
(a) Suppose that r(t) = r, where r is a positive con-
y = y,(x) and y = y,(x) are solutions, y = n(x) + 52_ Solve y' + 3y = 27x 2 + 9 by assuming that yp(x) = stant. In this case, the drug is added at a constant
yz(x) is also a solution. 0.5 1 1.5 2 2.5 3 x Ax 2 + Bx + C. rate. Sketch the phase line for dyldt = r - /gI.
44 Chapter 2 First-Order Equations 2.3 Substitution Methods and Special Equations 45
Solve the IVP. Determine lim yet). How does this

limit correspond to the pliase line? 63. Compare the solutions of !+ y =I(x) subject to
dy
dt = (k- ay)y
(b) Suppose that rCt) = I + sin t and k = 1. In this yeO) = 0, where 1(;<) = x, sin x, cos x, and eX.
case, the drug is added at a periodic rate. Solve
the IVP. Determine ,lit;! yet) (if it exists). Describe 64_ Compare the solutions of !+ Icy = x, yeO) = I for
to show that it differs from dy/dt = ky in that the rate of growth (the coefficient ofy),
(k - ay), is not constant. Instead, this rate depends on y. Although this is a separable
what happens to the ct;;,g concentration over time. k= -2, -1,0, 1,2. equation, we can solve the ODE by another approach through the use of the substitu-
(e) Suppose that ret) = e-' and k = 1. In this case, tion
. dy
the rate at which the drug is added decreases over * 65. Compare the solutIOns of dx + Y = x, yeO) = k for
time. Solve the IVP. Determine lim yet) (if it ex- k= -2, -1,0, 1,2.
ists). Describe what happens to 'the drug concen- We would like to transform the logistic equation from an ODE that depends on
tration over time. 66. (a) Graph the direction field for the equation y' =
y - x. (b) Solve and graph the solution to the initial- y and I into one that depends on w and t, so we find dwldl with the chain rule:
One advantage of using technology is that often when a large value problems (i) ~O~ ~ ~ x; (ii) ~O~ ~ ~t; and dw _ dw dy _ _ -2 dy
number of problems are solved, their solutions are compared dt - dy dt - y dt .
and conjectures about general patterns can be discovered and (iii) e~o~ ~ ~;. (e) Comment on this statement: if
Therefore,
tested. Many computer algebra systems are capable of solving
a variety of linear equations, particularly those that are fre- we slightly change the initial conditions in a linear
quently encountered in an elementary differential equations initial-value problem, the solution also slightly changes.
course.
62. Find a general solution of the equation Substitution of this expression into dy/dt - ky = _ay2 gives us
x !+ Y = x cos x. Graph various solutions on the

or dw
dt
+ ky-l =
a.
rectangle [0, 27T] X [-10, 10].
Finally, with the substitution w = y -1, we obtain the first-order linear ODE
dw
dt + kw = a,
12.31 Substitution Methods and Special Equations which we solve with the integrating factor /1-(1) = e fkdt = e kt . Therefore,
d kt kt
dt [e w] = ae ,so
Substitutions are used often in integral calculus to transform integrals into forms that
can be computed easily. Similarly, with some differential equations, we can perform
substitutions that transform a given differential equation into an equation that is easier or w
a
=k+ C-
Ie
kt
.
to solve. For example, we have considered population problems modeled with the first-
order linear IVPs: Returning to the original variable, we have
dy dy
dt = ky, yeO) = Yo and dt = ky - h,y(O) = Yo· y
-1
= ka + C-
Ie
kt
.
However, we can consider a more complicated model that involves a nonlinear equa-
I a
tion with Applying the initial condition, yeO) = Yo, we find that Yo = k + C), so that
dy I a k - ayo -1 a k - ayo -kt •

C 1 = Yo - k' or C 1 = ~. Therefore, y k + ~ e . Solvmg for y,
2
dt = ky - ay , yeO) = Yo· =
we find that
The equation in this model, known as the logistic equation, is almost linear because
it can be written as dy/dt - ky = -a/. (The y2 term on the right side causes the equa-
tion to be nonlinear.) It can also be expressed as
46 Chapter 2 First·Order Equations 2.3 Substitution Methods and Special Equations 47
In Figure 2.14, we graph this function using the parameter values Yo = 114, a = dw 3w 3
I, and k = 3. Notice that unlike solutions to the Malthus model, solutions to the lo- ~+-;-=-;
2.5 gistic equation are bounded.
2 Equations such as the logistic equation are classified as Bernoulli equations. because w = y3. We solve this equation with the integrating factor
1.5
1 /-L(x) = e f (3/x)dx = e 3lnx = x 3, x > O.
0.5
i Definition 2.3 Bernoulli Equation Multiplying the ODE by /-L(x) and simplifying yields ~ [x 3w] = 3x 2, so that
2 4 10 x
A Bernoulli equation is a first-order equation of the form x 3w = x 3 + C. Solving for w, we find that w = 1 + Cx -3. Substituting w = y3 and
Figure 2.14 Solution to the solving for y gives us
dy n
logistic equation (yo = 1/4, dt + p(t)y = q(t)y .
a = 1, and k = 3) I
Of course, if n = 0 or n = I, this equation is linear, so we would not need to make N ow, consl'der thed'ffi . I equatIOn
l erenlia y = - x-+
, ddx -y, W h'IC h can be wntten
. as
a change of variable to solve the equation. However, for other values of n, such as y-x
n = 2 as we considered in the previous example, we solve these equations with the
substitution
(y - x) dy + (x + y) dx = O.
w = yl-n. Because this equation is not of the form fi (x)gl (y) dx + f2(x)gz(y) dy = 0, it is not
separable. However, if we let y = ux, then we use the product rule to obtain dy =
In doing so, we transform the nonlinear equation into the linear equation u dx + x duo Substituting these expressions into (y - x) dy + (x + y) dx = 0 and sim-
p lifying results in
dw
dt + (1 - n)p(t) w = (1 - n)q(t).
(ux - x)(u dx + x du) + (x + ux) dx = 0 or x(u z + I) dx + x 2 (u - I) du = O.
After solving this equation for w by using an integrating factor, we return to the orig- This equation is separable and can be written as
inal variable with the substitution w = y I-n and by solving for y. Of course, Bernoulli
equations may involve variables other than y or t, but we follow a similar solution dx=~du.
method. X u2 + 1
Because the right side of this equation is equivalent to (u2 ~1- u2 : I) du,
we find that
dy
Solve -
dx
+-
y
x
=
1
-2' X
-0/
> O. InJxJ = tan- l u - t In(u z + 1) + C.
: ] Solution In this case, the independent variable is x (instead of t as it was in the 1

Notice that the absolute value is not needed in the term '2ln(u2 + I) because
. previous problem), and the right side of the equation can be written as y -2/X , so
p(x) = lfx, q(x) = l/x, and n = -2. Therefore, we make the substitution w = u 2 + 1 > 0 for all u. Now, u = y/x because y = ux, so resubstitution gives us
yl-(-2) = y3, where dw/dx = 3y2 dy/dx so that dy/dx = (y-2/3) dw/dx. With this
substitution, we obtain InJxJ = tan-ley/X) - t InC(y/x)' + 1) + C
y-2 dw Y y-2
-3-~ + -; =-;-. as a general solution of (y - x) dy + (x + y) dx = O.
The equation (y - x) dy + (x + y) dx = 0 is called a homogeneous equation.
, dw 3y 3 3 We can always reduce a homogeneous equation to a separable equation by a suitable
Multiplication by 3y- gives us the first-order ODE ~ + --;- = -;, or
substitution.
Definition 2_4 Homogeneous Differential Equation rydy + y 2 dv - rydy - y~ dy = O.

A first-order differential equation that can be written in the form M(x, y) dx + Notice that by writing Y v 2y2 + y2 = Y /(v 2 + I) = Iyl~ as y~,
N(x, y) dy = 0, where M(tx, ty) = t"M(x, y) and N(tx, ty) = t"N(x, y) is called a we are making the assumption that y ~ O. The initial condition includes a positive
homogeneous differential equation (of degree n). value of y, so we can solve the IVP with this assumption. We can simplifY and sep-
arate the variables in this equation to obtain
y2dv=y~dy or dv = dy.
~ y
Using a table of integrals or a trigonometric substitution to evaluate

Show that the equation (x 2 + yx) dx - y2 dy = 0 is homogeneous.
dv .
.J Solution Let M(x, y) = x 2 + yx and N(x, y) = _y2. The equation

f , ~,weobtam
vv 2 + I
(x 2 + yx) dx - y2 dy = 0 Inlv + ~I = Inlyl + Ct.
is homogeneous of degree 2 because Returning to the original variables with v = x/y, we find that
= (txi + (ty)(tx) = t 2(X 2 + yx) = t 2M(x, y)

and
M(tx, ty)
Inl~ + J(~Y + 11 = Inlyl + C t
J
~ + (~r + 1 = C2y
Homogeneous equations are reduced to separable equations by either of the sub- 4 x+~
Y = C2y
stitutions
y = ux or x = ry. X + Y x 2 + y2 = C2y 2
Use the substitution y = ux if N(x, y) is less complicated than M(x, y), and use x = vy
if M(x, y) is less complicated than N(x, y). If a difficult integration problem is en-
y=±~~.
countered after a substitution is made, try the other substitution to see if it yields an
~2 VI + 2C2 • 0 =
-1
easier problem. As with the separation of variables technique, this technique was also Applying the initial conditions gives us I so C2 = I. Therefore,
discovered by Leibniz.
Figure 2.15 Graph of the solution to the IVP is y = ~. We graph this solution in Figure 2.15.
mI7+~J(~r + 11 = 0
Notice that the curve passes through the point (0, 1) as required by the initial
,'i'"I,I. condition.
dy
Solve the initial-value problem dx =
x +
k+ '
x2 y2
yeO) = I. In addition to verifYing that M(tx, ty) = t"M(x, y) and N(tx, ty) = t"N(x, y), there
are other ways to determine if an equation is homogeneous or not. For example, solv-
ing the differential equation 4xy2 dx + (x 3 + y3) dy = 0 for dy/dx, we obtain
I Solution When we write the ODE in differential form, y dx - (x +
1 4 ~)2
-4-
\1\2 + y2) dy = 0, we recognize that the ODE is homogeneous of order one. Then, dy -4xy2 7(- xy )
2
x (Y)
because M(x, y) = -y is less complicated than NC;<, y) = x + Yx 2 + y2, we let dx = x 3 + y3 = ..!.. (x 3 + y3) = I + (2::.)3 = F -; ,
x = ry so that dx = v dy + Y dv. Substitution yields x3 X
y(vdy + ydv) - (ry + Yv 2y2 + y2)dy = 0 where F(t) = -4t 2/(1 + t 3). Similarly, we find that
I
-(-4xy 2
) y y
y3 1.
0.5 : ;0.5.
where G(t) = -4tl(t3 + I). This indicates (although we have not shown it in general)
'1 x -I" . -::0.5. ·0.5 Ix -~ ,-0.5
that an equation is homogeneous if we can write it in either of the forms dyldx = F(ylx)
or dyldx = G(xly).
-0.5
.::[
'.fi,,'·'·'
~
--'I
£} (a)
y
(b)
Y
(e)
Find values of ao, a2, and a3, none of which are zero, so that the nontrivial
alo 1 1.
. dy aox + alY ..
solutlOns to dx = + are pen odIc. 0.5 0.5 05
a2x a3Y
: ] Solution This equation is homogeneous because -I -0.5 0.5 I x -} -0.5 0.5 IX -I -0,5 0.5 IX
-0.5 -0.5 ':'0.5

dy
dx
-I -I -I
(d) (e) (f)

ao + alt
where F(t) = a2 + a3 t ' Letting y = ux leads to Figure 2_16 (a) ao = -4, a, = 3, a2 = 7, a3 = 5 (b) ao = 4, a, = -5, a2 = -6, a3 = 3
(c) ao = 3, a, = -1, a2 = 1, a3 = -2 (d) ao = 3, a, = -1, az = 1, a3 = 2 (e) ao = -2.
-x(a3u2 + (a2 - al)u - ao) dx - x2(a3u + a2) du = 0 a, = -7, a2 = -7, a3 = 2 (f) ao = -5, a, = I, a2 = 2, a3 = -4
Integrating and substituting u = ylx yields a general solution of the equation

. ~ ~x+a~ . , .
In Vao + (al - a2)u - a3u2 - equatlOn dx = + . Several of the results we obtam are shown m FIgure 2.16.
a2x a3Y
(Use each direction field to graph several solutions to each equation.)
From these graphs, we see that the choices ao = 3, al = -I, a2 = I, and
a3 = -2 may lead to periodic solutions. A general solution of the equation : =
3
3x -2Y is -2x 2 - xy + y2 = C (why?), which is a family of ellipses as shown in
x - y
Fi~ure 2.17 Graph of Figure 2.17. We see that the choices ao = 3, al = -I, a2 = I, and a3 = -2 yield
~ - xy + y2 = C for various periodic solutions to the equation.
values of C
This solution is complicated and nearly impossible to analyze analytically, so we dy

proceed graphically. We begin by generating random values of ao, alo az' and a3,
Show that the nontrivial solutions to dx = aox : alY are periodic if a2 = -al and
a2x a3Y
2
none of which are zero, and then graphing the direction field associated with the a1 + aOa3 < O.
(a) (x - 2y + 1) dx + (4x - 3y - 6) dy = 0 46. Show that if the differential equation dyldx = f(x, y)
EXERCISES 2.3 (b) (5x + 2y + 1) dx + (2x + Y + 1) dy = 0 is homogeneous, the equation can be written as
dyldx = F(ylx). (Hint: Let t = 1Ix.)
(e) (3x - y + 1) dx - (6x - 2y - 3) dy = 0
(d) (2x + 3y + 1) dx + (4x + 6y + 1) dy = 0 *47. IfM(x,y) dx + N(x,y) dy = 0 is a homogeneous equa-
tion, show that the change of variables x = r cos aand
In Exercises 1-6, solve the Bernoulli equation. 24. Y dx + (xy + y) dy = 0 41. Find an equation of the Curve that passes through the y = r sin atransfonn the homogeneous equation into
point (Ve, Ve) and has slope ylx + xly at each point a separable equation.
,Ix 25. (2x 2 - txy + 5y2) dx + xy dy = 0
1• Y - '2 y = y 2. y
, 2
+Y = xy
(x, y) on the curve . 48. Equations of the fonn
26. (y + 2Yx 2 + y2) dx - x dy = 0
f(xy' - y) = g(y')
*3'Y'-~y=y3cOSX 4.y'-~y=y3Sinx *27. y2 dx = (xy - 4x 2 ) dy In Exercises 42-45, without actually solving the homogeneous
equations, match each equation in Group A with the graph of are called Clairaut equations after the French math-
28. Y dx - (3-.rxy + x) dy = 0 ematician Alexis Clairaut (1713-1765) who studied
its direction field in Group B.
5. y '-2y =C% 6.y'+3y=vYsinx 29. (x 2 - y2) dy + (y2 + xy) dx = 0 these equations in 1734. Solutions to this equation are
30. xy dy - (x 2e -yix + y2) dx = 0 detennined by differentiating each side of the equa-
Group A tion with respect to x.
In Exercises 7-16, detennine if the differential equation is ho- +.=.
*31. dx = 2y e-xly dy xeYlx dy yeYlx (a) Use the chain rule to show that the derivative of
mogeneous. If so, detennine its degree. dy x y 42. dx =y 43. dx = -x- f(xy' - y) is
7. (x + 3y) dx - 4x dy = 0 32. x(ln x - In y) dy = y dx xlY dy xe~/Y
44 dy = ye f'(xy' - y)(xy" + y' - y') = f'(xy' - y)(xy"),
8. (y2_X2)dx+xydy=0 . dx x 45. dx =y
where' denotes differentiation with respect to the
*9. Yx 2 +xydy-xydx=0 In Exercises 33-39, solve the initial-value problem. argument of the function, x.
Group B
dy 4/ _X2 (b) Show that the equation
10. dx + (x + y) dy = 0 33. dx =~,y(l)= 1 (a) (b)
11. cos (_x_) 2y

dx + e !r dy = 0 34. (x + y) dx - x dy = O,y(l) = 1
y
r(xy' - y)(xy") = g'(y')y",
x +Y which is equivalent to
*35. x dy - (y + Yx 2 + y2) dx = 0, y(l) = 0
12. y In (~)
Y
dx + L
x+y
dy = 0
36. (x 2 + y2Yx2 + y2) dx - xyYx 2 + y2 dy = 0,
2 [('(xy' - y)x - g'(y')]y" = 0,
I is obtained by differentiating both sides of the
*13. 2 In x dx - 1n(4y2) dy = 0 y(l) = 1
Clairaut equation with respect to x.
37. (y3 - x 3 ) dx - xy2 dy = 0, yell = 3
14. (~+~)dx+ ;2 dy=O
38. xy3 dx - (x 4 + y4) dy = 0, y(l) = 1
-3, -,2-)
-:-1
.) 3 x
(e) This result indicates that y" = 0 or
f'(xy' - y)x - g'(y') = O. If y" = 0, y' = c
sin 2x
15. cos 2y dx +
Y)
(In
~ dy = 0 *39. y4 dx + (x 4 - xy3) dy = O,y(l) = 2
~2 where c is a constant. Substitute y' = c into the
;'3 differential equation f(xy , - y) = g(y'), to find
40. First-order equations of the fonn that a general solution is f(xc - y) = g(c). If
16. v?"+"l dx + Y dy = 0
f' (xy' - y)x - g' (y ') = 0, this equation can be
(a,x + b,y + C1) dx + (a2x + b2y + C2) dy = 0 (e) (d) used along withf(xy' - y) = g (y ') to detennine
In Exercises 17-32, solve each equation. y y
can be transfonned into a homogeneous equation with another solution by eliminating y'. This is called
-3 -3 I :
17. 2x dx + (y - 3x) dy = 0 a transfonnation. If a2101 '" b21blo the transfonnation the singular solution of the Clairaut equation.
X=X+h . 2 -.2- I
18. (2y - 3x) dx + x dy = 0 49. Use the following steps to solve the Clairaut equation
{ Y = Y + k' where (h, k) satIsfies the linear system
. 1-. ,i ; :r xy' _ (y')3 = y .
*19. (xy - y2) dx + x(x - 3y) dy = 0 o,h + b,k + c, = 0 ..
20. (x 2 + xy + y2) dx - xy dy = 0 { 02h + b k + C2 = 0' reduces this equatIon to a (a) Place the equation in the appropriate fonn to find
2 '::'~C-2 -1 'I '.2.' 3 x thatf(x) = x andg(x) = x 3.
21. (x 3 + y3) dx - xy2 dy = 0 homogeneous equation in the variables X and Y. If '-1,
021a, = b2lb, = k, the transfonnation z = o,x + b,y (b) Use the form of a general solution to find that
2 dy_x+4y -2 xc - y = c3 and solve this equation for y.
2· dx - 4x + y reduces this equation to a homogeneous equation in
the variables x and z. Use this transfonnation to solve ~3 :'3 (e) Find the singular solution by differentiating
*23. (x - y) dx + x dy = 0 the following equations. xy' - (y')3 = Y with respect to x to obtain
54 Chapter 2 First-Order Equations 2.4 Exact Equations 55
xy" + y' - 3(y'iy" = y', which can be simpli- (a) Differentiate the equation with respect to x to 60. Consider the solution to the logistic equation y = *63. (a) The sine integral function, Si(x), is defined by
fied to [x - 3(y'i]y" = O. Sincey" = owas used
to find the general solution, solve x - 3(y')2 = 0
for y' to obtainy' = (X/3)112. Substitute tbis ex-
obtainy' = -y' - xy" + (Y'ty".
(h) Substitute y' = p to obtain p =

dp
-p - x dx
ayo
kkyo
+( - ayo)e
kt . Find lim y(I). Sketch the phase
,...~
line for dy/dl = (k - ay)y and compare it to the value

Si(x) = r si~ I dl.
pression for y' into xy' - y = (y ')' to obtain a (i) Evaluate Si'(x) and limx_o Si(x). (ii) Graph
+ (p)4 : . SimplifY this equation to obtain of,~~y(t). Si(x) on the interval [0, 67T]. (iii) Approximate the
relationship between x and y.
dp 2p maximum value of Si(x). (iv) Can you predict
In the same manner that computer algebra systems and graph- limx_~ Si(x)?
In Exercises 50-53, solve the Clairaut equation. dx = pO_x'
ing lltilities can be used to help Tmd solutions of separable equa-
(h) Graph the direction field associated with the equa-
50. xy' - y- 2(xy' -y)' =y' +I . th' . dx p4 - x d tions, they can also be used to help solve and graph solutions
c Rewnte IS equatlOn as dp =
() an solve---;;:p- of homogeneous equations. tion y sin(x/y) dx - (x + x sin (x/y» dy = O.
51. xy' - y- I = (y')2 -y'
this first-order linear equation for x to obtain (c) Solve the initial-value problem
52. I + Y - xy' = In(y') 61. Solve the equation
*53. I - 2(xy' - y) = (y,)-2 x = tp4 + Cp-ll2. (x lI3y 213 + x) dx + (X2/3y 113 + y) dy = O. Graph sev- y sin (x/y) dx - (x + x sin(x/y» dy = 0
{ y(l) = 2
eral solutions.
54. Equations of the fonn y = xf(y ') + g (y ') are called
Lagrange equations. These equations are solved by (d) Substitute y' = p and x = t p4 + Cp -1/2 into the y'-
y2 _ x2 and graph the solution.
making the substitution 62. Solve the initial-value problem { y(4) = 0xy and
differential equation y = -xy' + t(y')5 to ob-
p =y'(x). graph the solution for 0 < x :5 4.
tain a fonnula for y.
(a) Differentiate y = xf(y') + g (y ') with respect to (e) Graph the splution curves for various values of C.
x to obtain
y' = xf'(y')y" + f(y') + g'(y')y". In Exercises 56-58, solve the Lagrange equation.
~ Exact Equations
(b) Substitute p into the equation to obtain
dp dp
56. y =x(:r + 3(:r -2(:)' Unlike homogeneous equations, many first-order differential equations carmot be re-
p = xf'(p) dx + f(p) + g'(p) dx
duced to separable differential equations by a suitable substitution. For example, the
dp
57. y = x(: + 1) + (2 : + 1) first-order differential equation
= f(p) + dx [xf'(p) + g'(p)].
(sin y + Y cos x) dx + (sin x + x cos y) dy = 0
(c) Solve this equation for dx/dp to obtain the linear 58. y=X(2 - : ) + (2(:r + I)
is neither separable nor can be reduced to a separable equation by an appropriate sub-
equation stitution. Nevertheless, a general solution of
59. Euler was the first mathematician to take advantage
dx xf'(p) + g'(p) of integrating factors to solve linear differential equa-
tions. * Euler used the following steps to solve the
(sin y +y cos x) dx + (sin x + x cos y) dy = 0
dp p - f(p)
which is equivalent to equation can be calculated, as we will see in this section.
dx z 1
dx + f'(p) x=~ --2z+-=-.
dv v v
dp f(p) - p p - f(p)"
Definition 2.5 Exact Differential Equation
This linear first-order equation can be solved (a) Multiply the equation by the integrating factor
for x in tenns of p. Then, x(p) can be used with e- 2",v.
A first-order differential equation that can be written in the form
y = xf(p) + g(p) to obtain an equation for y.
(b) Show that ; (e- 2,vz) = e- 2vv : - 2e- 2,,,", M(x, y) dx + N(x, y) dy = 0,
55. Solve the equation y = -xy' + t (y')5 using the + e- 2vz. where
following steps. In this case, f(y ') = -y' and
(c) Express the equation as :v (e- 2v ,,",) = e- 2v and af af
g(y')=t(y')'. M(x, y) dx + N(x, y) dy = ax (x, y) dx + ay (x, y) dy
solve this equation for z.
for some functionf(x, y) is called an exact differential equation.
II< Victor J. Katz, A History of Mathematics: An Introduction. HarperCollins (1993), p. 503 .
. _-------------------------------------------------------~~- ..~---
56 Chapter 2 First·Order Equations 2.4 Exact Equations 57
In calculus, we learn that the total differential of the functionf(x, y) is M(x, y) dx + N(x, y) dy = 0
af af df= O.
df = ax (x, y) dx + ay (x, y) dy.
A general solution of the equation is
Therefore, the equation M(x, y) dx + N(x, y) dy = 0 is exact if there exists a function f(x,y) = C,
f(x, y) such that M(x, y) dx + N(x, y) dy is the total differential of f(x, y).
where C is a constant.
In multivariable calculus, we learned that if f, af af a'i and a'i are
, ax' ay' axay' ayax
. . a'i a'i
contmuous on an open regIOn R, then - - = - - on R. Hence if
tial off(x, y),

axay ayax '
M(x, y) dx + N(x, y) dy = 0 is exact and M(x, y) dx + N(x, y) dy is the total differen· ,"',,','M'
Find a general solution of (sin y +y cos x) dx + (sin x + x cos y) dy = O.
~~ = :x (:~) = :y (:~) = aa~' ~ .1 Solution The equation is exact because
In fact, we can prove the following. (See Exercise 55 in this section.) a~ (sin y + Y cos x) = cos y + cos x = :x (sin x + x cos y).
f
Theorem 2.1 Test for Exactness Letf(x, y) be a function with aa = sin y + Y cos x and af = sin x +x cos y. In·
x ay
The flrst·order differential equation M(x, y) dx + N(x, y) dy = 0 is exact if and . af . ·th I .
t egratmg ax = sm y + Y cos X WI respect to x resu ts m
only if aM = aN.
ay ax
f (siny + y cos x) dx = x siny + y sinx + g(y),
where g (y) denotes an arbitrary function of y. We must include this arbitrary func·
lif'"I,'.' tion g (y) because the derivative of a function of y with respect to x is zero. That is,
the general form of the functionf(x, y) whose partial derivative with respect to x is
Show that the equation 2xy3 dx + (I + 3x 2y2) dy = 0 is exact and that the equa·
iJf/ ax = sin y + Y cos x is given by
tion x 2y dx + 5xy2 dy = 0 is not exact.
f(x,y) =x siny +y sin x + g(y).
~ ] Solution The equation 2xy3 dx + (1 + 3x 2y2) dy = 0 is an exact equation
Because we are looking for a functionf(x, y) that satisfies
because
af .
:y (2xy3) = 6xy2 = a~ (1 + 3x 2y2). ay = sm x + x cos y,
Conversely, the equation x 2y dx + 5xy2 dy = 0 is not exact because and differentiating f(x, y) = x sin y + y sin x + g (y) with respect to y results in
af .
..i.. (x 2y)
ay
= x2 *' 5y2 = ..i..ax (5xy2). ay = smx +x cosy + g'(y),
it must be true that
If the equation M(x, y) dx + N(x, y) dy = 0 is exact, we can find a function sin x + x cosy + g'(y) = sin x + x cosy,
af af which indicates that g'(y) = O. This means that g (y) = k for some constant k. Thus,
f(x, y) such that M(x, y) = ax (x, y) and N(x, y) = ay (x, y). Then the differential equa·
tion becomes f(X, y) = x sin y +y sin x + k.
2
y and g(y) = -y + C I so that substitution into f(x, y) = x sin y + g(y) yields
Therefore, the implicit function x sin y + y sin x + k = C I or x sin y + y sin x = C,
where C = C 1 - k represents an arbitrary constant, is a general solution of 2
f(x,y) = x siny - y + CI .
(sin y + y cos x) dx + (sin x + x cos y) dy = O. [Note that there is no need to in-
clude an arbitrary constant inf(x, y), because it is included inf(x, y) = C.] Several
2
A general solution of the exact equation is then x sin y - y + CI = C. Simplify-
ing, we have
members of the family of solutions are graphed in Figure 2.18 by graphing several
level curves of the functionf(x, y) = x siny + y sin x.
-5
_--
........
lOX x 2 sin y - y = k,
where k = C - C I is a constant. Our solution must satisfy yeO) = t so we must find
y
~IO the solution that passes through the point (0, !). Substituting x = 0 and y = ! into
the general solution, we obtain 0 2 sine!) - ! = k so that k = -! and the solution is
x 2 sin y - y = -!. The solution is graphed in Figure 2.19. We see that the graph
Figure 2.19 passes through the point (0, !), as required by the initial condition.
Figure 2.18 We graph various solutions to the Solving the Exact Differential Equation
equation by graphing several level curves of the func-
tion/ex, y) = x sin y + y sin x. M(x, y) dx + N(x, y) dy = 0
af af
CD
'if',,',".1 @
Assume that M(x, y) = ax (x, y) and N(x, y) = ay (x, y).
Integrate M(x, y) with respect to x. (Add an arbitrary function of y, g(y).)

Solve 2x sin y dx + (x 2 cos Y - 1) dy = 0 subject to yeO) = 112. G) Differentiate the result in step 2 with respect to y and set the result equal to
N(x, y). Solve for g'(y).
: 1 Solution The equation is exact because o Integrate g' (y) with respect to y to obtain an expression for g (y). (There is no
need to include an arbitrary constant.)
a~ (2x sin y) = 2xcos y = :x (x 2 cos y - 1). ® Substitute g(y) into the result obtained in step 2 for f(x, y).
® A general solution isf(x, y) = C, where C is a constant.
Letf(x, y) be a function with aflax = 2x sin y and aflay =x 2
cos Y - 1. Integrat-
(j) Apply the initial condition if given.
ing aflax with respect to x yields
A similar algorithm can be stated so that in step 2, N(x, y) is integrated with respect
f(x, y) = J2x siny dx = x 2 siny + g(y).
to y as we show in Example 4.
Notice that the arbitrary function g(y) serves as a "constant" of integration with re-
spect to x. From the differential equation, we have
af = x2 cos Y - 1
ay ,
"ii,,',I.'
and differentiatingf(x, y) = x 2 sin y + g(y) with respect to y gives us Solve (e Y1X -
x 1+x
1_) dx +
~eYlx + __ 2 e y1x dy = O.
af 2
ay = x cos Y + g'(y).
Thus,
] Solution This equation is exact because ~ (eY1X - ~ ey1x + __
ay x 1+x
1_) 2 =
2
x cosy - 1 = x cosy 2
+ g'(y) _L2 e y1x = ~ (eY1x ). Let f(x, y) be a function such that af = e
y1x -
~ e y1x +
g'(y) = -1 x ax ax x
----------------------------------------------------~----------------------------------~
60 Chapter 2 First·Order Equations 2.4 Exact Equations 61
f y1x
y y
-+11
x
d
2 and aa = e • Integrating af with respect to y because it is a less compli-
.
y
af
cate expreSSiOn than ax gives us
ay
21t
1t o 0
21t
1t
()()
f(x, y) = Je y1x
dy = I;x e
y1x
+ g(x) = xe y1x + g(x)
-1t -"
where g (x) is an arbitrary function of x. Differentiating f(x, y) with respect to x ;--..
-21t -21t
leads to
Figure 2.20 Graph of (a) (b) (e)
x cosy + 4x - siny = C
for various values of C :~ = e
y1x
+x (- :2 e
Y1x
) + g'(x) = e
y1x
- ; eY 1x
+ g'(x), 2" 2" '-../ \ 21t
1t
" "
so g'(x) = I ~ x 2 ' This implies thatg(x) = 1
tan- x, sof(x,y) = xe y1x + tan-I x.
Therefore, a general solution of the exact equation is xe y1x + tan-I X = c.
~ c::
-"
Figure 2.20 shows the graph of x cos y + 4x - sin y = C for various values of C.
IdentifY the curve(s) corresponding to C = 5. -21t
Up to now, ali initial-value problems we have considered have had a unique so- (d) (e) (I)
y
lution. However, this need not be true. (This topic is considered again in Section 2.5.) 2" 2"
o l'fi"",Wj 3"
<:::
4" x
"
"
~
21t 3"
\:
4" x
Find a value of Yo so that there is not a unique solution to the initial-value problem
{
(cos x cosy - sin x) dx + (cosy - sin x siny) dy = O.
yeO) = Yo
C0 <::: ~ \: Figure 2.21 (a) C = -j(b) C = -1
(e) C = -4(d) C = 0 (e) C = l (f) C = 1
(g) (h) (g) C = l (b) C = !
: ISolution The equation (cos x cos y - sin x) dx + (cosy - sin x siny) dy = 0 is
exact because
4.9
a (cos x cosy - smx

ay . )= . = axa (cosy - smxsmy).
-coSx smy '. 4.8
4.7
A general solution is found to be cos x + sin x cos y + sin y = C, which we graph
for various values of C in Figure 2.21. 4.6
From the graphs, we see that if C = 0, it is possible to find Yo such that there Figure 2.22 The solution is not unique near this point
4.5
is not a unique solution to the initial-value problem. Indeed an exact value of Yo is because when we zoom in, we see the graph of more than
371'/2. (Why?) (See Figure 2.22.) -0.2 -0.1 0.1 0.2 one function passing through the point.
(~
46. (lntegrating Factors) If the differential equation
40. + y) dx + (eY + x) dy = 0, yeO) = 0 M(x, y) dx + N(x, y) dy = 0 is not exact, multiplying
EXERCISES 2.4 l+x
it by an appropriate function J.L(x, y) yields an exact
equation. To find J.L(x, y), we use the fact that if the
In Exercises 41-44, without actually solving the equations, equation J.L(x, y)M(x, y) dx + J.L(x, y)N(x, y) dy = 0 is
match each equation in Group A with the graph of some of its exact, [J-LMl y = lJ.LNlx-
In Exercises 1-10, determine if the equation is exact.
*25. (1 + y2 cos(xy» dx + (xy cos(xy) + sin(xy» dy =0 solutions in Group B. (a) Use the product rule to show that J.L must satisfy
1- (y2 - 2~) dx + (2xy - \IX + 1) dy = 0 26. yeX)' (cos(xy) + sin(xy» dx

sin(xy» dy = 0
+ xeX)' (cos(xy) + Group A
the differential equation
MJ.Ly - NJ.Lx + (My - Nx)J.L = O.

dy eY-X(cosx - sin x)
27. «3 + x) cos (x + y) + sin (x + y» dx + 41- dx = cos y + siny
2_ x
YX 2 +y2
dx+ Y
YX 2 +y2
dy=O
(3 + x) cos (x + y) dy = 0 (b) Use this equation to show that if J.L = J.L(x), J.L sat-
dy-ex-y(cos x + sin x) isfies the differential equation
*3_ y cos(xy) dx + x cos(xy) dy = 0 2
42. dx = cos y - sin y
2x y cos(x 2 ) - Y sin(x 2 ) dJ.L _ My - Nx h My - Nx
4_ (y sec 2 x + 2x) dx + (tan x) dy = 0 28. x2 dx + dy -eYcosx _eX cosy (f;- N J.L,were N
5. 3xy2 dy + y3 dx = 0 43. dx = eY sin x - eX sin y
2xy + sin(x2) is a function of x only.
6. (x - y sin x) dx + Cy6 + cos x) dy = 0 x dy = 0 44 dy = eX sin x + eY siny
(e) Show that if J.L = J.L(x),
*7. Cy sin 2x) dx - (vY + cos 2x) dy = 0 eY1x( y)
- dx eYcosy-eXcosx
8.
2r
(e + y) dx - (e Y - x) dy = 0
*29. x - dx + eY1X dy = 0 J.L(x) =
x
(J-
Group B
l _ [aM(x, y) _ ON(x, y)] dx).
9. In(xy) dx + ~ dy = 0 (al (b) exp N(x, y) ay ax
y X1Y
e (x2 cos(?) + y2 sin(?)) (d) If J.L = J.L(Y), find a differential equation that J.L
10. eX)' dx + ~ eX)' dy = 0 30. In
dx-
y x 2y must satisfy and a restriction on
N-M
----=--r.
Show
X1Y 2n·-f----~
In Exercises 11-30, solve each equation. e (x2 cos(?) + y' sin(?)) that
11. 3x 2 dx - dy = 0 xy2
dy= 0 J.L(Y) =
12. -dx + 3y2 dy = 0 - 1 _[ ON(x, y) _ aM(x, y)] d ).
exp (J M(x, y) ax ay Y
*13. y2 dx + 2xy dy = 0
In Exercises 31-40. solve the initial-value problem. Graph the
3x 2 x3 solution on an appropriate region.
14. - dx - 2 dy = 0 7t 27t 37t X
In Exercises 47-54, use an integrating factor to solve each dif-
y Y
15. (2x + y3) dx + (3xy2 + 4) dy = 0
31. 2xy2 dx + 2x 2y dy = 0, y(l) = I (e) (d) ferential equation. (See Exercise 46.)
47. x 2y dx + x 3 dy = 0
16. -~ dx + (;2 + 3y2 ) dy = 0 32. (I + :2) dx - ~ dy = 0, y(2) = 1 37t
48. y(2e X + 4x) dx + 3(e X + x 2 ) dy = 0
*17. 2xy dx + (x2 + y2) dy = 0 *33. (2xy + 3x 2) dx + (x 2 - 1) dy = 0, yeO) = 1 49. y dx + (2x - yeY) dy = 0
18. 2xy3 dx + (1 + 3x 2y2) dy = 0
34. (l + 5x - y) dx - (x + 2y) dy = 0, yeO) = 0 27t SO. (2xy + y2) dx - x 2 dy = 0
19.
2
sin ydx + x sin 2y dy = 0 35. (eY - 2xy) dx + (xeY - x 2) dy = 0, yeO) = 0 51. (y + 2x 2 ) dx + (x 2y - x) dy = 0
20. (3x 2 + 3y2) dx + 6xy dy = 0 36. (2xex' + 2xe-Y ) dx + (ex' - x 2e-Y + 1) dy = 0, 52. (5xy + 4y2 + 1) dx + (x 2 + 2xy) dy = 0
yeO) = 0 7tC
53. (2xy2 + y) dx + (2x 3 - x) dy = 0
y+y2 X+X2
*21. (y _ X)2 dx - (x _ yj' dy = 0 *37. Cy2 - 2 sin 2x) dx + (1 + 2xy) dy = 0, yeO) = 1 54. 2
(2x + tany) dx + (x - x tan y) dy = 0
2 2
2 38. (cos x - sin x + y) dx + (sec y tan y + x) dy = 0, 27t *55. Suppose that M(x, y) dx + N(x, y) dy = 0 is an equa-
22. (3x y + 3y2 - 1) dx + (x 3 + 6xy) dy = 0
yeO) = 0 " tion for which aMlay = aNlax.
23. _2xy2 sin(x 2 ) dx + 2y COS(X2) dy = 0
24. (2x - y2 sin(xy» dx + (cos(xy) - xy sin(xy» dy = 0 39. C~ x2 - y2) dx - 2xy dy = 0, yeO) = 0
45. Show that a separable equation of the form g(y) dy -
hex) dx = 0 is exact.
(a) LetgCy) = J(N(X,y) - ~J M(x,y) dx) dy.
64 Chapter 2 First-Order Equations 2.5 Theory of First-Order Equations 65
Show that g is a function of y. Hint: Show that Graph several solutions. would see that no other solution curves intersect. This tells us that if another initial
N(x, y) - ~ f M(x, y) dx is a function of y by 58. (a) Find a general solution for each of the following
differential equations:
condition were used instead of yeO) = 0, then the IVP would have only one solution.
Hence, we say that it has a unique solution. For example, if we consider
showing that
(i) (2x + 2y) dx + (2x + 2y) dy = 0;
dy x
:x (N(X, y) - :y f M(x, y) dx) = O.
(ii) (1.8x + 2y) dx + (2x + 2y) dy = 0; and
(iii) (2x + 1.9y) dx + (1.9x + 2y) dy = O.
dx = Y"y(1) = 0,
(b) Letf(x, y) = g(y) + f M(x, y) dx. Show that (b) Graph the direction field for each equation along
with several solutions.
we have the same general solution, y2 - x 2 = K, as before. However, when we apply
the initial condition y(1) = 0, we find that (0)' - (1)2 = K, so K = -1. Therefore, the
M(x, y) dx + N(x, y) dy = af dx + af dy. (c) Comment on this statement: "If we slightly change IVP has the unique solution y2 - x 2 = -lor x 2 - y2 = I. As we see from this ex-
ax ay
a differential equation, the solutions also slightly ample, an initial-value problem may have one or more solutions depending on how the
As with other types of equations, technology is useful in solv- change." problem is stated. The following theorem helps us understand the types of initial-value
ing exact equations or in performing the steps necessary to solve *59. How does the graph of the solution to the initial-value Figure 2.24 Graph of problems that have a unique solution.
an exact equation. problem x 2 = K for several values
y2 _
ofK<O
56. Find a general solution of the equation (sin(cy) - yc sin(cx» dx +
(2x - y2 sin(xy» dx + (cos(xy) - xy sin(xy» dy = O. {
(xc cos(cy) + cos(cx» dy = 0 Theorem 2_2 Existence and Uniqueness
Graph various solutions on the rectangle [0, 37Tj X yeO) = I
[O,37T]. Consider the initial-value problem
change as c takes on values from - 2 to 2?
57. Find a general solution of the equation 60. Find restrictions on p and q so that the first-order lin- y' = I(x, y), y(xo) = yo·
(-1 + e""y + y cos(xy» dx + .dy
ear equatwn dx + p(x)y = q(x) is exact. If/and af/ay are continuous fimctions on the rectangular region R: a < x < b,
(1 + e""x + x cos(xy» dy = O. c < Y < d containing the point (xo, Yo), then there exists an interval Ix - xol < h
centered at Xo on which there exists one and only one solution to the differential
equation that satisfies the initial condition.
~ Theory of First-Order Equations
(' Some Differences Between Linear Equations and Nonlinear Equations
To better understand the solution or solutions to an initial-value problem consider the "f''''''.'
following: ' Does the fact that the IVP dy/dx = x/y, yeO) = ° has two solutions contradict the
Existence and Uniqueness Theorem?
dy x
dx = Y" yeO) = o.
y : .I Solution In this case,/(x, y) = x/y and (xo, Yo) = (0, 0). The hypothesis of the
Thi~ equation is solved by separating the variables to obtain y dy = x dx and by inte- Existence and Uniqueness Theorem is not satisfied because I is not continuous at
gratmg each SIde of the equation to obtain y2/2 = x 2 /2 + Cor (0, 0). Therefore, the fact that the IVP has two solutions does not contradict the Ex-
istence and Uniqueness Theorem.
y2 -x 2 = K,
where K = 2C. if K *
0, this solution is represented graphically by a family of hyper-
bolas. For K > 0, we have hyperbolas that intersect the y-axis (Figure 2.23), and for
K < 0, the hyperbolas intersect the x-axis (Figure 2.24). In the case of this initial-value
"i',,',••'
problem, we are interested in finding the solution or solutions that satisfY yeO) = O. In VerifY that the initial-value problem dy/dx = y, yeO) = 1 has a unique solution.
other words, we require that the solution curve(s) pass through (0, 0). When we sub-
2 °
stitute (0, 0) into y2 - x = K, we find that K = 0, so y2 - x 2 = satisfies the IVP.
Factoring indicates that (y - x)(y + x) = 0, so the IVP has two solutions, y = x and
: ] Solution In this case, I(x, y) = y, Xo = 0, and Yo = 1. Hence, both I and aj/ ay
are continuous on all rectangular regions containing the point (xo, Yo) = (0, 1). By
Figure 2_23 Graph of
y2 _ x 2 = K for several values y = -x. A questIOn we may ask at this point is whether the IVP has more than one so- the Existence and Uniqueness Theorem, there exists a unique solution to the dif-
ofK>O lution if we use another initial condition. By observing the family of hyperbolas, we ferential equation that satisfies the initial conditiony(O) = 1. We verifY this by solv-
2.5 Theory of First·Order Equations 67
ing the initial-value problem. This equation is separable and equivalent to dyly =
Some Differences Between Linear Equations and
dx. A general solution is given by y = Ce x , and the solution that satisfies the initial
condition yeO) = 1 is y = eX. Nonlinear Equations
While solving both linear and nonlinear first-order equations throughout Chapters I
and 2, you may have noticed some of the differences between the two types of. equa-
The Existence and Uniqueness Theorem gives sufficient, but not necessary, con-
tions. For example, in Exercises 1.1, we considered the ODE dyldx = _2xy2 WIth so-
ditions for the existence of a unique solution of an initial-value problem. If an initial-
lution y = l/(x 2 + k) where k is constant. We noted that although y = 0 also satisfies
value problem does not satisfy the hypotheses of the theorem, we cannot conclude that
the ODE it cannot be obtained from y = l/(x 2 + k) for any choice of k. (We call y =
a unique solution does not eJcist. In fact, the problem may have a unique solution, no
solution, or many solutions. o a sin~ar solution.) Therefore, the solution technique used to solve this sepa:able
nonlinear ODE does not generate all solutions. This is a difference between nonlmear
and linear equations. When we find a general solution of a linear equation, we obtain
all possible solutions to the equation. . . .
1'£·,,"'- dy x
We can also state an EJcistence and Uniqueness Theorem for IVPs mvolvmg flrst-
order linear equations.
Find a solution to the initial-value problems (a)
dx = y213
y { and
6 yeO) = 0
2
(b) {2(xy3 - xSy) dx + (x - x y2) dy = O'f 'bl
yeO) = 0 ' 1 POSSI e.
Theorem 2.3 Existence and Uniqueness: First-Order Linear Equations l
l
CJ Solution The EJcistence and Uniqueness Theorem does not guarantee the eJcis- I Consider the initial-value problem i
tence of a solution to either problem because both +
y- 3
and 2(xy3 - xSy) are
x6 _ x 2y2
y'+p(x)y=q(x),y(xo)=yo.
discontinuous at the point (0, 0) specified by the initial condition. I If p and q are continuous on an open interval I that contains x = xo, then there
eJcists one and only one function that satisfies the differential equation for each
-4
()Th
a
. dy x.
e equatIOn dx = y213 IS separable and has solution II x in I, and this also satisfies the initial condition.
Figure 2.25
1y513 = lx2 + C1
52
y=(%X2 + cr. o '.fi,,'lUEFind an interval on which the IVP xy' + 2y = x cos x, y(2) = 4 has a unique solu-
Application of the initial condition leads to the unique solution y = (516)315 X615, tion.
which is graphed in Figure 2.25.
5
(b) The equation 2(xy3 - x y) dx + (x 6 - x 2y2) dy = 0 is more difficult to solve. : .1 Solution First, we write the ODE in the form y' + (2/x)y = cos x to identify
Dividing the equation by (X4 + y2)2 leads to an exact equation that has general so- p(x) = 21x and q(x) = cos x. The function p is continuous for x < 0 or for x > 0; q
2 4 y
lution x y = C(x + y2). Applying the initial conditiony(O) = 0 results in the iden- 400 is continuous for all values of x. Therefore, we must determine the interval con-
2
tity 0 = 0, which means thatx y = C(x4 + y2) is a solution to the initial-value prob- 350 taining Xo = 2 on which both functions are continuous. This interval is 0 < x < +co.
lem for any value of C. There are infinitely many solutions to the initial-value 300 (Of course, if the initial condition were y( -2) = 4, then we would have selected
250
problem, so the solution is nonunique. Several solutions are graphed in Figure 2.26. 200 -co < x < 0.) If we solve this IVP, we fmd that
Figure 2.26
150
100
50 Y = -2 cos x - "'2
2 sm. 2.
x -"'2 (sm 2 +2cos2-8 ) ' x.
+ sm
Are the hypotheses of the Existence and Uniqueness Theorem satisfied for the initial- x x x
value problem ~ = ~, yeO) = I? If J
y -1
h = lnly + ~I+ c,
+-"-::::;~==F-'-~
0.5 1 1.5 2 2.5 3 x
Figure 2.27 Solution of

We graph this function in Figure 2.27 to observe that the solution is valid only on
0< x < +co. The solution becomes unbounded as x -> 0+. It approaches zero as
does a unique solution to this problem exist? Does more than one solution exist? xy' + 2y = x cos x, y(2) = 4 x -> +co.
69
68 Chapter 2 First-Order Equations 2.5 Theory of First-Order Equations
As we see, the interval on which the solution to a first-order IVP is unique de- tradict the Existence and Uniqueness Theorem? 19. y' + _I_y = _1_,y(_I) = 0
x-3 x-I
pends on the functions p and q in the general form of the ODE. This is not necessar- 5. Show that y = 0 and y = xlxl both sa~fy the initial-
I
ily true with an IVP involving a nonlinear ODE. In this case, the interval of definition value problem dyldx = 2V1Yi, yeO) - O. Does this 20. (x - 2)y' + (x 2 - 4)y = x + 2'Y(0) = 3
of the solution may depend on the initial condition, as we illustrate in the following contradict the Existence and Uniqueness Theorem?
example. 2
6. Does the initial-value problem dy/dx = 4x - xy2, *21. y' + Yy , =x,y(O) = 0
y(2) = 1 have a unique solution on an interval con- 4 - x-
y
"ii,,','., taining x = 2?
*7. Does the initial-value problem dy/dx = yVx, y(l) =
22.y'+ Y y
4-x
2 =x,y(3)=-1
Determine the interval of definition of the solution to y' = y3, yeO) = a, a > O. I bave a unique solution on an interval containing
10 23. (a) Over what interval is the solution to the IVP y' +
x = I? Verify your result by solving the problem.
(1Ix)y = sinx,Y(1T) = I certain to exist? (b) ~olvethis
: 1 Solution This equation is separable. Integrating f ~ = f dx yields -± y -2 = S. Does the initial-value problem dy/dx = 6y 213, y(l) =
o have a unique solution on an interval containing
IVP and graph the solution. (e) Is the solution valId
on a larger interval than what is guaranteed by the Ex-
X + C, such that simplification gives us y -2 = -2x + K where K = -2C. There- x = I? Solve the problem using separation of vari- istence and Uniqueness Theorem for Linear IVPs?
= lIYK - = a indicates that K = ables. Is y = 0 obtained through this method? Is
fore, y 2x. Applying the initial condition yeO) 24. (a) Over what interval is the solution to the IVP y' +
y = 0 a solution? (tan x)y = sin x, yeO) = 0 certain to exist? (b) So~ve
a- 2 , so y = lIYa- 2 - 2x. This solution is defined only if a- 2 - 2x > 0 or x <
-1 -0.5 0.5 1.5 x 9. Does the initial-value problem y' = sin y - cos x, this IVP and graph the solution. (e) Is the solution
(a- 2 )/2. In Figure 2.28, we graph the solution for K = a- 2 = 1,2,3. In this graph,
y( 1T) = 0 have a unique solution on an interval con- valid on a larger interval than what is guaranteed by
Figure 2_28 Graph of notice that when a- 2 = I the solution is defined only for x < 112. The other two
taining x = 7T'? the Existence and Uniqueness Theorem for Linear
solutions are defined on x < I and x < 3/2, respectively. Notice also that each be-
Y= _ _I_-forK=1 23
YK-2x ' , 10. Does the initial-value problemxy' = y,y(O) = I have IVPs?
comes unbounded as x approaches (a -2)/2 from the left.
a Wlique solution on an interval containing x = O? Ver-
ify your response by solving the initial-value problem. In Exercises 25-28, determine the interval of definition of the
Another difference between linear equations and nonlinear equations that you 11. Show thaty = sec x satisfies the initial-value problem solution to each IVP. Graph the solution for several values of a
may have noticed while working problems in this chapter is that a linear equation has y' = Y tan x, yeO) = 1. What is the largest open inter- to verify your result. Assume that a is positive.
an explicit solution because of the way we solve linear ODEs. Conversely, a nonlinear val containing x = 0 over which y = sec x IS a solu-
25. y' = y>, yeO) = a
equation often has an implicit solution that cannot be solved analytically to find an ex- tion? Explain.
26. y' = xy2, yeO) = a
plicit form of the solution. 12. Show thaty = tan x satisfies the initial-value problem
y' = I + y2, yeO) = o. What is the largest open inter- *27. y' = -x/y, yeO) = a
val containing x = 0 over which y = tan x is a solu- 2S. y' = _y3, yeO) = a
tion? Explain. 29. Find all points (xo, 0) in the rectangule
EXERCISES 2.5 13. Using the Existence and Uniqueness Theorem, de- R = {(x, y)IO :5 x :5 21T, -1T:5 Y :5 1T) such that the
termine if dy/dx = ~ has a unique solution initial-value problem
passing through the point (a) (0, 2); (b) (4, -1);
dY cosxsinx+sinxsiny
(e) (0, 1/2); (d) (2, I) is guaranteed. dx = cosxcosy-cosysiny
L Does the Existence and Uniqueness Theorem guaran- Separate the initial-value problem into two parts. First 14. Using the Existence and Uniqueness Theorem, deter- {
solve the problem for y 2: 0 and then solve it for y(xo) = 0
tee a unique solution to the following initial-value mine if dy/dx = Y25 - y2 has a unique solution
problems on some interval? Explain. y < O. passing through the point (a) (-4, 3); (b) (0, 5); has more than one solution. Confirm your results
dy *3. The Existence and Uniqueness Theorem implies that (e) (3, -6); (d) (4, -5). graphically.
(a) dx + x 2 = y2, yeO) = 0
at least one solution to the initial-value problem dyldx 30. Find all points (xo, Yo) in the rectangle
dy = yllS, yeO) = 0 exists. (a) Must the solution be R = {(x, y)IO :5 x :5 41T, 0 :5 Y :5 41T) such that the
(b) dx + x 2 = y-2, yeO) = 0 unique according to the theorem? (b) Solve the prob-
In Exercises 15-22 use the Existence and Uniqueness Theo-
initial-value problem
rem (for linear IVP;) to determine the largest interval on which
lem by hand. Is the solution unique? Note that y = 0
dy I the solution is guaranteed to exist. dY = cosy-ycosx
(e) dx = Y + I _ x' y(1) = 0 satisfies the differential equation as well as the initial
2 dx xsiny+smx I
condition. 15. y' + (I/x)y = x , y(l) =0 {
2. According to the Existence and Uniqueness Theorem, 4
+ x4y = 2x', yeO) = 0 y(xo) =Yo
the initial-value problem dy/dx = Iyl. y(l) = 0 has a 4. Show that y = 0 and y = ~6 both satisfy the initial- 16. x'y'
*17. 2y' + xy = In x, y(e) = 0 has no solutions. Hint: Graph the direction field as-
solution. (a) Must the solution be unique? (b) Solve
the problem by hand. Is the solution unique? Hinl:
value problem xI dydx = _vry, yeO) = O. Does th,S. con- 18. y' + (sec x)y = x, yeO) = 0
sociated with the equation.
2.6 Numerical Methods for First-Order Equations 71
o~ Numerical Methods for First-Order Equations

, Euler's Method (' Improved Euler's Method (' Errors
Y
=
Tangent Line
Y I(x", Yo)(x - xo) + Yo
Therefore, the approximate value of the solution at x = Xl is
(" Runge-Kutta Method (X).YI) Yl = f(xo, YO)(XI - xo) + Yo = hf(xo, Yo) + Yo·
I
Notice that we refer to the approximate value of y at x = Xi as Yi while w~ refer
In many cases, we cannot obtain a formula for the solution to an initial-value problem to the exact value of the solution at x = Xi as Y(xi)' This means that Yl approxlmat~s
of the form Y(XI)' (See Figure 2.29.) We remember from calculus that ~is gives a .good approxI-
mation if h is a small number. Next, we assume that the pomt (x" Yl) IS on the solu-
dy
dx = f(x, y), y(xo) = Yo· tion curve. If we determine the equation of the line with slope f(x" Yl) that passes
through (X" Yl), we obtain
x
For example, we cannot solve
Y - Yl = f(x" Yl)(x - Xl),
dy . Figure 2.29 so we find the approximate value of y at X = X2 is
dx = sm(xy), yeO) = I
Y2 = f(xo, YO)(X2 - XI) + Yl = hf(xo, Yo) + YJ.
using any of the methods we have discussed to this point. Of course, we can determine
certain properties of the solution by looking at the slope field. However, we sometimes We continue this process until we reach YN, the approximate value of Y(XN) = y(X). In
would like to find numerical values of y at particular values of x. For that reason, we doing so, we obtain a sequence of points of the form (x", Yn)~ n =:' I, 2, ... , N. We
now discuss numerical methods for approximating solutions to initial-value problems. Exact solution (X•• y(x.)) show several points of this type along with actual values of Y m Figure 2.30.
We begin with a method based on tangent line approximations.
Euler's Method
Euler's Method
The solution of the initial-value problem
Suppose that we wish to approximate the solution to
y' = f(x, Y), y(xo) = Yo
dy is approximated at the sequence of points (xn, Yn) (n = I, 2, ... ), where Yn is
dx = f(x, y), y(xo) = Yo
Figure 2.30 the approximate value of y(xn ) by computing
over the interval Xo :5 x :5 X, where X is a specified value of x. For example, we may Yn = hf(xn-" Yn-I) + Yn-I (n = 1, 2, ... ),
wish to approximate the solution on 0 :5 X :5 1. This indicates that the initial condi-
tion is specified at Xo = 0 and that X = I is the end of the interval. Now, we partition i where Xn = xo. + nh and h is the selected stepsize.
Xo :5 x :5 X into N subintervals of equal length h, where
h = X-xo
N .
Therefore, the nth value of x in the partition is
o "fi""'.' Use Euler's method with h = 0.1 and with h = 0.05 to approximate the solution of
y' = xy, yeO) = 1 on 0 :5 X :5 1. Determine the exact solution and compare the
Xn = Xo + nh, n = 1,2, ... ,N.
results.
This means that Xl = Xo + h,X2 = Xo + 2h, ... ,xN= Xo +Nh = Xo +N(X~XO) = 1 Solution First, we note thatf(x, y) = xy, Xo = 0, and Yo = 1. With h = 0.1, we
X If we are solving the IVP dy/dx = f(x, y), y(xo) = Yo, then the point (xo, Yo) is on have the formula
the solution curve. Therefore, as we discussed in relation to slope fields, we can fol-
low the line tangent to the solution curve at (xo, Yo) to approximate the value of the so- Yn = hf(xn-" Yn-l) + Yn-l = O.lxn-IYn-1 + Yn-l·
lution at the next value of x, x = Xl' Because the slope of this tangent line isf(xo,yo), Then, for x, = Xo +h = 0.1, we have
we find that the equation for the line is y - Yo = f(xo, Yo)(x - xo) or
Y, = O. lxoYo + Yo = 0.1(0)(1) + 1 = 1.
y = f(xo, Yo)(x - xo) + Yo. Similarly, for X2 = Xo + 2h = 0.2,
72 Chapter 2 First-Order Equations 2.6 Numerical Methods for First-Order Equations 73
Y2 = O. IX ,y, + YI = 0.1(0.1)(1) + 1 = 1.01. y

1.6
y
1.6
In Table 2.1, we show the results of this sequence of approximations. From this, we 1.5 • 1.5
see that y( 1) is approximately 1. 54711.
1.4 • 1.4
TABLE 2.1 Euler's method with h = 0.1

1.3 • 1.3
1.2 • 1.2
Xn Xn •
• 1.1
Yn I Yn 1.1
0.0 1.0 0.6 1.15873 0.2 0.4 0.6 O.S 1 x 0.2 0.4 0.6 O.S
Ca) (b)
0.1 1.0 0.7 1.22825
0.2 1.01 Figure 2.31 (a) h = 0.1 (b) h = 0.05

0.8 1.31423
0.3 1.0302 0.9 1.41937
0.4 1.061 II 1.0 1.54711 Improved Euler's Method
0.5 1.10355 Euler's method is improved by using an average slope over each interval. Using the
tangent line approximation of the curve through (xo, Yo), Y = f(xo, Yo)(x - xo) + Yo, we
find the approximate value of Y at x = x" which we now call yj. Then,
For h = 0.05, we use yj = hf(xo, Yo) + Yo,
Yn = hf(xn-lo Yn-l) + Yn-l = 0.05Xn-IYn-1 + Yn-l and with the differential equation Y' = f(x, y), we find that the approximate slope of
Tangent Line Approximation the tangent line at x = XI is f(x" yf). The average of the two slopes, f(xo, Yo) and
to obtain the values given in Table 2.2. With this stepsize, the approximate value of
y(1) is 1.59594. Average .
ApproXImauon
fe *).
Xh Yl , IS
f(xo, Yo) +
2'
f(x" yi) and an equation of the line throu"h (x
/:l 0,
y) with
0
The exact solution !o the initial-value problem, which is found with separa-
tion of variables, isy = eX-l2, so the exact value ofy(l) is e '12 = 1.64872. The smaller
value of h, therefore, yields a better approximation. We graph the approximations
obtained with h = 0.1 and h = 0.05 as well as the graph ofY = er12 in Figure 2.31.
Notice from these graphs that the approximation is more accurate when h is f(xo, Yo) + f(x" y!)
Y= 2 (x - xo) + Yo·
decreased. Xo XI
X
We illustrate the determination of this average slope and the position of these lines in
TABLE 2.2 Euler's method with h = 0.05 Figure 2.32 Figure 2.32.
At x = x" the approximate value of Y is given by
Xn Yn Xn Yn Xn Yn
0.0 1.0 0.35 1.05361 0.70 1.2523 f(xo, Yo) + f(x" yi) ( ) + _ f(xo, Yo) + f(XI, yi) h + Yo.
YI = 2 x, - Xo Yo - 2
0.05 1.0 0.40 1.07204 0.75 1.29613
Continuing in this manner, the approximation in each step in the improved Euler's
0.10 1.0025 0.45 1.09348 0.80 1.34474 method depends on the following two calculations:
0.15 1.00751 0.50 1.11809 0.85 1.39853
y~ = hf(xn-" Yn-I) + Yn-l
0.20 1.01507 0.55 1.14604 0.90 1.45796
i
0.25 1.02522 0.60 1.17756 0.95 1.52357
0.30 1.03803 0.65 1.21288 1.00 1.59594

where Xn = Xo + nh.
2.6 Numerical Methods for First·Order Equations 75
TABLE 2.3 Improved Euler'S method with h = 0.1

Improved Euler's Method
Yn (I EM) Yn (EM) Actual Value
0.0 1.0 1.0 1.0
y' = f(x, y), y(xo) = Yo 1.00501
0.1 1.005 1.0
is approximated at the sequence of points (X., Yn) (n = 1, 2, ... ), where Yn is
0.2 1.0201755 1.01 1.0202
the approximate value of y(xn ) by computing at each step the two calculations:
OJ 1.0459859 1.0302 1.04603
y~ = hf(xn-I> Yn-l) + Yn-l
(n = 1, 2, ... ), 1.08329
1.083223 1.06111
_ f(xn-I> Yn-l) + f(x., y~) 0.4
Yn - 2 h + Yn-l 1.10355 1.13315
0.5 1.1330513
where Xn = Xo + nh and h is the selected stepsize. 1.1970687 1.15873 1.19722
0.7 1.277392 1.22825 1.27762
""".,'.' 0.8 1.3767731 1.31423 1.37713
0.9 1.4987552 1.41937 1.49930

Use the improved Euler's method to approximate the solution of y' = xy, yeO) =1 1.0 , 1.6478813 ,1.547111 1.64872
on 0 :=; x :=; I for h = 0.1. Compare the results to the exact solution. ~~~-.r-~====."=='~~c,,"
: ] Solution In this case, f(x, y) = xy, Xo = 0, and Yo = I. Therefore, we use the Errors
equations
When approximating the solution of an initial-value problem, there are several sources
y~ = hxn-1Yn-l + Yn-l of error. One of these sources is round-off error because computers and calculators
Xn-IYn-1 + xnY~ use only a fmite number of digits in all calculations. Of course, the error is compounded
Yn= 2 h+Yn-l as rounded values are used in subsequent calculations. Therefore, one way to minimize
round-off error is to reduce the number of calculations.
for n = 1,2, ... , 10. For example, if n = I, we have Another source of error is truncation error, which results from using an ap-
proximate formula. We begin our discussion of error by considering the error associ-
Y1' = hxoYo + Yo = (0.1)(0)(1) + I = I ated with Euler's method, which uses only the first two terms of the Taylor series ex-
pansion.
YI
= XoYo +2 XIY1' h + Yo = (0)(1) +2 (0.1)(1) (01)
.
+ I = 1.005. Recall from calculus the Taylor formula with remainder:
dy 1 dZy Z
y Then, y(x) = y(xo) + dx (xo)(x - xo) + 2!dxz (xo)(x - xo) + ...
1.6
y~ = hx1Yl + Yl = (0.1)(0.1)(1.005) + 1.005 = 1.01505 I dn 1 dn+1y
1.5 + ~ : . (xo)(x - xor + (n + 1)1 dxn+1 (e)(x - xor+\
1.4
XIYl + xzyi (0.1)(1.005) + (0.2)(1.01505)
1.3 yz = 2 h + Yl = 2 (0.1) + 1.005 where e is a number between x and Xo. The remainder term is
1.2 I dn+1y n+l
1.1 = 1.0201755. Rn(x) = (n + 1)1 dxn+1 (e)(x - xo) ,
In Table 2.3, we list the approximations obtained with this improved method and
0.2 0.4 0.6 0.8 1 x so the accuracy of the approximation obtained by using the first n terms of the Taylor
compare them to those obtained with Euler's method in Example 1 (see Figure 2.33).
Figure 2.33 Which method yields a better approximation? series depends on the size of Rn(x). For Euler's method, the remainder is
where
and
so at x = xI = Xo + h, the remainder is We also use the Taylor series expansion ofy to obtain another representation ofYn+1 =
y(xn + h):
"( )
Therefore, a bound on the error is given by
y(xn + h) = y(xn) + hy'(xn) + hZ Y ;n + ... = Yn + hy'(xn) + hZ Ty"(x ) + ....
h2 Now, because
IRI(xI)! oS -2 max
Xo s xS x! Yn+1 = Yn + Akl + Bkz = Yn + Ahf(xm Yn) + Bhf(xn + ah, Yn + bhf(xn.Yn»),
we wish to determine values of A, B, a, and b such that these two representations of
"ii,,1,".' Yn+ I agree. Notice that if we let A = I and B = 0, then the relationships match up to
order h. However, we can choose these parameters more wisely so that agreement oc-
Find a bound for the local truncation error when Euler's method is used to approx- curs up through terms of order hZ. This is accomplished by considering the Taylor ex-
imate the solution of the initial-value problem y' = xy, yeO) = 1 at XI = 0.1. pansion of a function F of two variables about (xo, Yo), which is given by
: ] Solution We found in Example 1 that the exact solution of this problem is F(x, y) = F(xo, Yo) + ax
aF
(xo, Yo)(x - xo) +
aF
a; (xo, Yo)(Y - Yo) + ....
Y = eX'12. Therefore, a bound on the error is
In our case, we have
f(xn + ah, Yn + bjh(x., Yn») =
dy d y, 2
, af af Z
'10
where - = xeX" - and - - = ern + x 2eX"12 f(X., Yn) + ah ax (X., Yn) + bhf(x., Yn) ax (X., Yn) + O(h ).
dx dx 2 .
Then, The power series is then substituted into the following expression and simplified to
yield
Yn+1 = Yn + Ahf(xm Yn) + Bhf(xn + ah, Yn + bhf(x., Yn»)
oS (0.005)[ max leX'12 1 + max Ixz/lzll Z af Z af 3
O:SxsO.l OsxSO.I = Yn + (A + B)hf(x., Yn) + aBh ax (X., Yn) + bBh I(x., Yn) ax (X., Yn) + O(h ).
oS (0.005)[e(O.I)'12 + (O.lie(o.I)'12l = 0.005075.
Comparing the above expression to the following power series obtained directly
Notice that the value of Y = eX'12 at X = 0.1 is 1.005012521 and the approximate from the Taylor series of Y given by
value obtained with Euler's method is YI = 1. Therefore, the error is 1.005012521 hZ af hZ af 3
- 1.0 = 0.005012521, which is less than the bound IRI(xI)! oS 0.005075. Y(Xn + h) = Y(Xn) + hf(x., Yn) + '2 ax (X., Yn) + '2 ff:x., Yn) ay f:x., Yn) + O(h )
or
Runge-Kutta Method
hZ af hZ af 3
In an allempt to improve on the approximation obtained with Euler's method and to Yn+1 = Yn + hf(X., Yn) + '2 ax (X., Yn) + '2 f(X., Yn) ay (X., Yn) + O(h ),
avoid the analytic differentiation of the functionf(x, y) to obtain y", y"', ... , we in-
we see that A, B, a, and b must satisfy the following system of nonlinear equations:
troduce the Runge-Kulla method. We begin with the Runge-Kulla method of order two.
Suppose that we know the value of Y at Xn. We now use the point (xm Yn) to ap- I I
proximate the value of Y at a nearby value x = Xn + h by assuming that
A +B = I, aA =2' and bB=2·
Yn+l = Yn + Akl + Bk2 , Choosing a = b = I, the Runge-Kulla method of order two uses the equations
1 1 Therefore, the Runge-Kutta method of order two approximates that the value of Y
Yn+1 = Y(xn + h) = Yn + "2hf(x., Yn) + "2hf(xn + h, Yn + hf(x., Yn)) at x = 0.1 is 1.005. Similarly, if n = I, then
= Yn
1
+ "2 (k l + k2 ) kl = O.lX1Yl = 0.1(0.1)(1.005) = 0.01005
k2 = O.I(XI + 0.1)(y1 + k 1) = 0.1(0.2)(1.01505) = 0.020301
where kl = hf(x., Yn) and k2 = hf(xn + h, Yn + k l ). Notice that this method is equiv- 1 I
alent to the improved Euler method. Y2 = Yl + "2 (k1 + k2) = 1.005 + "2 (0.01005 + 0.020301) = 1.0201755.
In Table 2.4, we display the results obtained for the other values on 0 S x S 1 us-
Runge-Kutta Method of Order Two ing the Runge-Kutta method of order two.
y' = f(x, y), y(xo) = Yo TABLE 2_4 Runge-Kutta Method of

is approximated at the sequence of points (X., Yn) (n = 1, 2, ... ), where Yn is Order Two with h = 0.1
the approximate value of y(xn ), by computing at each step Xn i Yn (RK) Actual Value
I
0.0 1.0 1.0
Yn+1 = y(xn + h) = Yn + thf(x., Yn) + thf(xn + h, Yn + hf(x., Yn)) I
0.1 1.005 1.00501

(n = 0, I, ... )
I 0.2
I
I 1.0201755 T 1.0202
= Yn + "2(k1 + k2), iI
0.3 1.0459859 1.04603
where kl = hf(x", Yn), k2 = hf(xn + h, Yn + kl)' Xn = Xo + nh, and h is the se- 0.4 1.083223 1.08329
lected stepsize.
1.1330513 1.13315
0.6 1.1970687 1.19722
'ifi,,"'·' 0.7 1.277392 1.27762
Use the Runge-Kutta method of order two with h = 0.1 to approximate the solution 0.8 1.3767731 1.37713
of the initial-value problem y' = xy, yeO) = I on 0 S x S 1.
0.9 1.4987552 1.4993
] Solution In this case,f(x, y) = xy, Xo = 0, and Yo = 1. Therefore, on each step 1.0 1.6478813 1.64874
we use the three equations
kl = hf(xn, Yn) = O.lxnYn

k2 = hf(xn + h, Yn + k 1) = O.I(xn + O.I)(yn + k 1) The terms of the power series expansions used in the derivation of the Runge-
I Kutta method of order two can be made to match up to order four. These computations
Yn+l = Yn + "2 (k1 + k2)' are rather complicated, so they are not discussed here. However, after much work, the
For example, if n = 0, then approximation at each step is found to be made with
kl = O.lxoYo = 0.1(0)(1) = 0
k2 = O.I(xo + O.I)(yo + k 1) = 0.1(0.1)(1) = 0.01
1 I and
Yl = Yo + "2 (k1 + k2 ) = I + "2 (0.01) = 1.005.
Runge-Kutta Method of Order Four Y1 = Yo + °f/ [k + 2k2 + 2k3 + k41

1

=1+ 0 1 [0
6 + 2 . 0.05 + 2 . 0.050125 + 0.100501251 = 1.005012521.
y' = [(x, Y), y(xo) = Yo
is approximated at the sequence of points (X., Yn) (n = I, 2, ... ), where Yn is In Table 2.5, we list the results for the Runge-Kutta method of order four to five
the approximate value of y(xn), by computing at each step decimal places. Notice that this method yields the most accurate approximation of
the methods used to this point.
h
Yn+l = Yn + "6 [k1 + 2k2 + 2k3 + k41 (n = 0, I, ... ), TABLE 2.5 Fourth-Order Runge-Kutta Method with h = 0.1
I Yn Yn (IEM) Actual Value
where kl = [(X., Yn), k2 = [ ( Xnh Yn + 2
+ 2' hkl) ' k3 = [ ( Xn + 2'
h Yn + hk ) -f ' (RK Order 4)
----~--------------~----
0.0 1.0 1.0 1.0
k4 =[(Xn+1, Yn + hk3), Xn = Xo + nh, and h is the selected stepsize. ------~--------
0.1 1.00501 1.005 1.00501
0.2 1.0202 1.0201755 1.0202
o ,id,,"'., ----~--------------~----
0.3
----~--------------~----
0.4
1.04603
1.08329
1.0459859
1.083223
1.04603
1.08329
Use the fourth-order Runge-Kutta method with h = 0.1 to approximate the solution 0.5 1.13315 1.1330513 1.13315
of y' = xy, yeO) = I on 0 :0=; x :0=; 1. ----~--------------~----
0.6 1.19722 1.1970687 1.19722
Detailed discussions of error : ] Solution With[(x, y) = xy, Xo = 0, and Yo = I, the formulas are
----+-------------r---
0.7 1.27762 1.277392 1.27762
analysis as well as other nu-
----~--------------~----
merical methods can be found kl = [(X., Yn) = XnY., 0.8 1.37713 1.3767731 1.37713
in most numerical analysis ----~--------------~----
texts.* h Yn
k2 = [ (Xn + 2' + 2hk1) =
( Xn + 01)(Yn + -T
T 0Ik ) ' 0.9 1.4993 1.4987552 1.4993
1.0 1.64872 1.6478813 1.64874
k3 =[(Xn +~, Yn + h~2) = (Xn + O~I )(Yn + 0.~k2),

k4 = [(Xn+b Yn + hk3) = Xn+l(Yn + 0.1k3)
EXERCISES 2.6
Yn+! = Yn + i[k1 + 2k2 + 2k3 + k41 = Yn + 0(/ [k1 + 2k2 + 2k3 + k4].
For n = 0, we have k1 = XoYo = (0)(1) = 0, k2 = (xo + Oil )(YO + 0.~k1) =

In Exercises 1-8, use Euler's method with h = 0.1 andh = 0.05 *7. y' = smy,y(O) = 1, x = 1
(0.05)(1) = 0.05, k3 = (xo + Oil )(YO + 0.~k2) = (0.05)(1 + 0.0025) = 0.050125, to approximate the solution at the given value of x. 8. y' = sm(y - x), yeO) = 0, x =1
and k4 = X1(Yo + 0.lk3) = (0.1)(1 + 0.0050125) = 0.10050125. Therefore, 1. y' = 4y + 3x + 2, yeO) = 1, x = 1
In Exercises 9-16, use the improved Euler's method with
2. y' = 4x - Y + 1, yeO) = 0, x = 1 h = 0.1 and h = 0.05 to approximate the solution of the corre-
*3. y' - x = y2 - 1, yeO) = 1, x = 1 sponding exercise above at the given value of x. Compare these
4. y' + x = 5yll2, yeO) = 1, x = 1 results with those obtained in Exercises 1-8.
.. See, for example, Richard L. Burden and 1. Douglas Faires. Numen"cal Analysis. Third Edition, PWS
5. y' = ~ + 5y, y(l) = 1, x = 2 9. Exercise 1 10. Exercise 2
Publishers (1985). 6. y' =xyll3 - y,y(1) = l,x = 2 *11. Exercise 3 12. Exercise 4
82 Chapter 2 First-Order Equations Chapter 2 Review Exercises 83
13. Exercise S 14. Exercise 6 As indicated in our examples, computers and computer algebra Bernoulli Equation ! Section 2.6!
*15. Exercise 7 16. Exercise 8 systems are of great use in implementing numerical techniques. A differential equation of the form
In addition to being able to implement the algorithm illustrated Euler Method
In Exercises 17-24, use the Runge-Kutta method with h = 0.1 earlier, many computer algebra systems contain built-in
y' + p(x)y = q(x)yn.
The approximate value of y at x = Xn is
and h = O.OS to approximate the solution of the corresponding commands that you can use to implement various numerical
~ection 2.4! + Yn-l
exercise above. Compare these results with those obtained in methods. Yn = hf(Xn-lo Yn-l)
Exercises 1-16.
28. Graph the solution to the initial-value problem where Xn = Xo + nh.
17. Exercise I 18. Exercise 2 Exact Differential Equation
dy = sin(2x _ y) Improved Enler's Method
*19. Exercise 3 20. Exercise 4 dx . A differential equation that can be written in the form
{ yeO) = 0.5 The approximate value of y at x = Xn is computed with
21. Exercise 5 22. Exercise 6 on the mterval [0, IS].
M(x,y) dx + N(x,y) dy = 0
*23. Exercise 7 24. Exercise 8 *29. Graph the solution of y' = sin(xy) subject to the ini-
y~ = hf(Xn-lo Yn-l) + Yn-l
where
25. Consider the IVP dy/dx = -y, yeO) = 1. (a) Find the tial condition yeO) = i on the interval [0, 7] for i = _ J(Xn-lo Yn-l) + J(xn, y~) h +
0.5, 1.0, I.S, 2.0, and 2.S. In each case, approximate aJ aJ Yn - 2 Yn-l'
exact solution to the problem. What is y(1)? (b) Ap-
the value of the solution if x = 0.5. M(x, y) dx + N(x, y) dy = ax (x, y) dx + ay (x, y) dy
proximate y(1) with Euler's method using h = 0.1,
Runge-Kurta Method of Order Two
h = 0.05, and h = 0.025. (c) Do the values in (b) ap- 30. (a) Graph the direction field associated with for some functionJ(x, y) is called an exact differential The approximate value of y at x = Xn is Yn+ 1 =
proach the exact value of y(1) as h approaches zero?
26. Repeat Exercise 25 using the improved Euler's
dy
dx = x 2 + y2 for -1 :SO x :SO 1 and -1 :SO Y :SO 1.
equation.
Yn + 21 (k, + k2) where k, = hJ(x., Yn) and k2 =
method. Do the approximate values of y(l) obtained hJ(xn + h, Yn + k,).
with the improved Euler's method converge more (b) Graph the solution to the initial-value problem !Section 2.5!
Runge-Kulta Method of Order Four
{t
quickly than those obtained with Euler's method?
27. Repeat Exercise 25 using the Runge-Kutta method of =x 2 + y2 Existence and Uniqueness Theorem The approximate value of y at x = Xn is
order four. Do the approximate values of y(1) obtained
with this method converge more quickly than those
yeO) = 0 IfJ and Z are continuous functions on the rectangular region Yn+' = Yn
h
+ 6"[k, + 2k2 + 2k3 + Ie.], n = 0, 1,2, .
obtained with the improved Euler's method and Euler's on the interval [- I, 1]. R: a < x < b, c < Y < d containing the point (xo, Yo), then
method? (c) Approximate y(I). there exists an interval Ix - xol
< h centered at Xo on which h Yn + --:2
hk') '
there exists one and only one solution to the initial-value ( + 2'
where k, = J(X., Yn), k2 = J Xn
problem
h hk,)
k3 = J (Xn + 2' Yn + 2 ' and Ie. = J(Xn+1- Yn + hk3)'
y' = J(x, y), y(xo) = Yo·
CHAPTER 2 SUMMARY
" ... _,
~ Concepts & Formulas
CHAPTER 2 REVIEW EXERCISES
! Section 2.1 ! Integrating Factor
An integrating factor for the first-order linear equation,
Separable Differential Equation : + p(x)y = q(x) is efp(x)dx. 6. (-x-' + x- 3) dx - (2y4 - 6y 9) dy = 0
In Exercises 1-23, solve each equation.
A differential equation that can be written in the form
g(y)y' = J(x) or g(y) dy = J(~) dx is called a separable dif- 1 dy _ 2x' *7. dy = __ y_
dx e2x Iny
ferential eqnation. • dx - S/
! Section 2.3! 2. cos 4x dx - 8 siny dy = 0 8 dy _ (4 - 7x)(2y - 3)
• dx - (x - 1)(2x - S)
! Section 2_2! Homogeneous Differential Equation (of degree n) *3 dy = sinhx
· dx 2 coshy 9. dy = cos 7x sin(3x)
A differential equation that can be written in the form
First-Order Linear Differential Equation dx e Y cos 6y
M(x, y) dx + N(x, y) dy = 0, where dy e
8y
A differential equation that can be written in the form =""7
t + p(x)y = q(x) is called a first-order linear differential M(tx, ty) = tnM(x, y) and N(tx, ty) = tnN(x,y)
4. dx
5 dy
• dx
= e'x
y4
10. 3x dx + (x - 4y) dy = 0
*11. (y - x) dx + (x + y) dy = 0
12. (y2 - x 2 ) dx + (x + y) dy = 0
equation. is called a homogeneons differential equation (of degree n).
13_ (2y - x) dx + (4y2 - 2xy) dy = ° 37. y' = Yx - y, yell = 1, [1, 2] Hence, the initial point (xo, Yo) is used to determine (b) Use the four-term method with h 0.1 to ap-=
dy y2 + x 2
14. dx=~ 38. y' = x + yl/3, yell = 1, [1, 2] y" A IlIst-order approximation is obtained from this proximate the solution of y' = 1 + Y + x 2 ,
2
series by disregarding the terms of order h and higher. yeO) = 0 on 0 S x S 1.
*39. y' = sin(xy), yeO) = 1, [0, 1]
*15 dy =~ In other words, we determine y I from
• dx y2 + x 2 40. Does the Existence and Uniqueness Theorem guaran- 42. (Running Shoes) The rate of change of energy ab-
y, = Yo + l(Xo, yo)h.
2
16. (x - y) dx + (y - x) dy = ° tee a unique solution to the following initial-value
problems? We next use the point (X"YI) to approximate the value
sorption in a running shoe is given by
2
17. (x y + sin x) dx + (fX' - cos y) dy = ° dy
(a) dx = xy', yeO) = 0
of y at X2 = x, + h. Calling this value Yo, we have
+ Y'(XI)(X +
dE =Fdu
dl dl'
18. (tany -x) dx + (xsec 2 y+ 1) dy= ° y(x) = y(X,) - x,)
y"(XI) 0
where
*19. (x Iny) dx + (~; + 1) dy °

=
*(b) : = xy-', yeO) = 0 2 i ( x -x,) + ... F(I) =
Fo
2" [1 - COS(wl)]
20. y' +Y = 5 *21. y' + xy = x

dy y
(e) dx = - x _ 2,y(2) = ° so that an approximate value of y(x2) is represents the force magnitude and pulse duration ex-
Y2 = y(X,) + I(x" YI)h = y, + I(x" y.)h. erted on the shoe and
22_ dx +~= 2 23. y' + 1. y = cos x 41. (Higher-Order Methods with Taylor Series Expan-
dy y Y x Continuing this procedure, we have that the approxi-
sion) Consider the initial-value problem du = UoW sine wI - 0)
mate value of y at x = X" = X"-I + h is given by dl 2
Solve the following Bernoulli equations. (See Section 2.3.) y' = I(x,y),y(xo) = Yo.
y" = Y(X"-I) + I(x"-"Y"-I)h represents the rate of change of the vertical displace-
Recall that the Taylor series expansion of y about ment of the midsole.· The constant Fo represents the
dy dy x = Xo is given by
24. dx - Y =xy3 *25. dx +Y = e Xy-2 = Y"-I + I(x"-" Y"-I)h. maximum maguitude of the input force in Newtons
(N), w represents the frequency of the input profile in
Solve the following Clairaut equations. (See Exercise 48, Sec-

y(x) = y(xo) + y' (xo)(x - xo) + zf- (x -
y"(x)
xo)2
If we use the first three terms of the Taylor series ex-
pansion instead, then we obtain the approximation radians per second (rad!s), Uo represents the maximum
rate of change of the vertical displacement of the mid-
tion 2.3.) y'''(xo) 3
h
2
+-3-!-(x-xo) + .... y" = Y(X"-I) + J(X"-" Y"-I)h +
dJ
dx (x"_" Yo-I) 2
sole in meters (m), 8 represents the phase angle be-
26. Y = xy' + 3y,4 *27. Y - xy' = 2 In(y') tween F(I) and u(t) in radians (rad), and I represents
We know the value of y at the initial value of x = Xo, time in seconds (s). Thus, the rate of change of energy
dl h'
Solve the following Lagrange equations. (See Exercise 54, Sec- so we use this value to approximate y at x, = Xo + h, = Y"-I + J(X"-lo y"_,)h + dx (x"_" Yo-I) 2' absorbed by the shoe is given by
tion 2.3.) which is near xo, in the following manner. We IlIst
evaluate the Taylor series at x, = Xo + h to yield We call the approximation which uses the first three de Fouow .
dt = - 4 - [1 - cos(wl)] sm(wl - 0).
28_ y - 2xy' = -2(y')' *29. y - 2xy' = -4(y'j' terms of the expansion the three-term Taylor
"(x)
21 h2
0 method.
y(xo + h) = y(xo) + y'(xo)h + Y (a) Find the energy absorbed by the shoe from I = I,
In Exercises 30-35, solve each initial-value problem. to 1=12,
(a) Use the three-term Taylor method with h = 0.1 to
+ y'''(xo) h' ...
30. (2x - y - 2) dx + (2y - x) dy = 0, yeO) = 1 3! . approximate the solution ofthe initial-value prob- (b) The maximum energy absorbed by the shoe,
lem y' = xy, yeO) = 1 on 0 S X S 1.
31. cos(x - y) dx
32. (ye'" -
+ (1 - cos(x - y» dy = 0, y( 1T) =
2x) dx + xe'" dy = 0, yeO) = °
1T
Substituting y' = I(x, y), y =
2
f (x, y), and y''' = A four-term approximation is derived in a
similar manner to be
ME, is found by calculating the energy absorbed
if I, = 8/wand 12 = (1T + O)lw. Show that
*33. (sin y - y cos x) dx + (x cos Y - sin x) dy = 0, d1 Fouo ( 1T. )

ME=-2- 1 +4 sm8 .
Y(1T) = ° dx2 (x, y) into this expansion, using y(xo) = Yo and
calling this new value y" we have
y" = y(Xn-I) + J(xn-" Yn-I)h
2 2
+
34. y2 dx + (2xy - 2 cos y sin y) dy = 0, yeO) = 1T dJ h d1 h' 43. (Fermentation) In the fermentation industry, one of
+ h) = + I(xo, yo)h + dx (xn-"y"-I)2 + d7(Xn-"Yn-I)6
(~+ lOY) dx +(~+ lnx) dy = O,y(l) =
YI = y(xo y(xo) the goals of the molecular biologist is to control the
35. 1 environment and regulate the fermentation. To achieve
1 dl , 1 d 21
2T dx (xo, yo)h- + 3! dx2 (xo, yo)h' + ... meaningful environmental control, fermentation re-
For each of the following initial-value problems, use (a) Euler's search must be carried out on fully monitored envi-
method, (b) the improved Euler's method, and (e) the Runge- ronmental systems, the environmental observations
Kutta method with h = 0.05 to approximate the solution to the must be correlated with existing knowledge of cellu-
initial-value problem 00 the indicated interval.
'" John F. Swigart, Arthur G. Erdman, and Patrick 1. Cain, "An Energy-Based Method for Testing Cushioning Durability of Running Shoes," Jour-
36. y' = y2 - x, yeO) = 0, [0, 1]
nalofApplied Biomechanics, Volume 9 (1993), pp. 47-65.
86 Chapter 2 First-Order Equations Differential Equations at Work 87
lar control mechanisms, and environmental control bulk liquid, kLa is the volumetric oxygen transfer co-
conditions must be reproduced through continuous
1. Susceptible and infected individuals die at a rate proportion to the number of sus-
efficient, Qo, is the specific rate of oxygen uptake ceptible and infected individuals with proportionality constant J.L called the daily
computer monitoring, analysis, and feedback control (microbial respiration), and X is the cell mass con-
of the fermentation environment. death removal rate; the number 11J.L is the average lifetime or life expectancy_
centration. If there are no cells present, the equation _
One component of envirorunental control is the becomes 2. The constant A represents the daily contact rate_ On average, an infected person
measurement of dissolved oxygen with a steam- will spread the disease to A people per day.
sterilizable dissolved oxygen sensor made up of a dC - 3. Individuals recover from the disease at a rate proportional to the number infected
dt = kLa(C* - C)*
polymer membrane-covered electrode. Suppose that with the disease with proportionality constant 1'. The constant l' is called the daily
the electrode is immersed in a liquid medium and that recovery removal rate; the average period of infectivity is 111'.
(a) If C(O) = 0, determine C(t) as a function of C'
the oxygen is reduced according to the chemical and kLa. 4. The contact number a = AI(l' + J.L) represents the average number of contacts an
reaction
infected person has with both susceptible and infected persons.
(b) Suppose that Cp is the concentration of dissolved
00 + 2HoO + 4e --+ 40H- oxygen that corresponds to the sensor reading and If a person becomes susceptible to a disease after recovering from it (such as
The rate of change dC/dt in dissolved oxygen con- kp is the sensor constant that depends on the con- gonorrhea, meningitis, or streptococcal sore throat), then the percentage of people sus-
centration at a particular point in a fermentor vessel ductance of the membrane and the liquid film ceptible to becoming infected with the disease, Set), and the percentage of people in
is given by o~ide the ~ns~ If Cp ~tisfies the equation the population infected with the disease, let), can be modeled by the system
dC - C£CrJdt = kp(C - Cp) and Cp(O) = 0, determine
dt = kLa(C' - C) - Qo,x, Cit) as a function of kp, kLa, and C'. By know- S'(t) = -AlS + 1'1 + J.L - J.LS
ing kp in advance and by observing Cp experi- l' (t) = AlS - 1'1 - J.Ll
where C' is the concentration of dissolved oxygen that {
mentally, the value of kLa can be estimated with S(O) = So, 1(0) = 10, S(t) + let) = I
is in e<J!:liIibrium with partial pressure p in bulk gas the solution oftbis initial-value problem. Thus, the
phase, C is the concentration of dissolved oxygen in amount of dissolved oxygen can be determined. This model is called an SIS (susceptible-infected-susceptible) model because once an
individual has recovered from the disease, the individual again becomes susceptible to
the disease.'
Since Set) = 1 - let), we can write l' (t) = AlS - 1'1 + J.Ll as
l' (t) = Al(1 - 1) - 1'1 - J.Ll
and thus we need to solve the initial-value problem
': ' " Differential Equations at Work:
1'(t) = [A - (1'+ J.L)]1- Al2
A. Modeling the Spread of a Disease { 1(0) = 10
I. Convert the Bernoulli equation l' (t) = Al( 1 - I) - 1'1 - J.Ll to a linear equation and
I
Suppose that a disease is spreading among a population of size N. In some diseases,
solve the result with the substitution y = 11.
like chicken pox, once an individual has had the disease, the individual becomes im-
mune to the disease. In other diseases, like most venereal diseases, once an individual 2. (a) Show that the solution to the initial-value problem
has had the disease and recovers from it, the individual does not become immune to 1'(t) = [A - (1' + J.L)]1 - 1\A 2
the disease. Subsequent encounters can lead to recurrence of the infection. { 1(0) = 10
Let Set) denote the percentage of the popUlation susceptible to a disease at time is
t, l(t) the percentage of the popUlation infected with the disease, and R(t) the percent- e(A+J.')(u-l)t
age of the population unable to contract the disease. For example, R(t) could represent if a oF 1
the percentage of persons who have had a particular disease, recovered, and have sub- a[e(HJ.')(u-l)t - 1]/(a - 1) + 1110 '
let) =
sequently become immune to the disease.
To model the spread of various diseases, we begin by making several assump-
tions and introducing some notation.
1 At ; lII ' if a = I
o
and graph various solutions if (b) A = 3.6, l' = 2, and J.L = 1; (c) A = 3.6, l' = 2,
>I< S. Aiba, A E. Humphrey, N. F Millis, Biochemical Engineen"ng. Second Edition, Academic Press
* Herbert W Hethcote, "Three Basic Epidemiological Models." Applied Mathematical Ecology, edited by
I
(1973), pp. 317-336.
Simon A. Levin. Thomas G. Hallan, and Louis J. Gross. Springer-Verlag (1989). pp. 119-143.
I
and /-L = 2. In each case, find the contact number. How does the contact number af- second year, instead of harvesting, you decide to add fish at a rate of r = 112. If
fect I(t) for large values of t? yeO) = 0.5, fmdy(2). How does this value compare to the original size of the pop-
3. Evaluate liml->~I(t). ulation? What (approximate) value of r should be used so that y(2) = yeO)?
6. Suppose that in problem 5 above, you do not fish or add to the populatio~ after ~he
The incidence of some diseases, such as measles, rubella, and gonorrhea, oscillates first year. If yeO) = 0.5, find y(2). How does this value compare to the ongmal Size
seasonally. To model these diseases, we may wish to replace the constant contact rate of the population? How much time T is required so that yeT) = yeO)?
A by a periodic function A(t). 7. Suppose that fishing occurs on a seasonal basis so that the situation is modeled with
4. Graph various solutions if (a) A(t) = 5 - 2 sin(6t), 'Y = I, and /-L = 4; and (b) A(t) y' = ay - h( 1 - sin 7), yeO) = Yo· What is the maximum value, minimum value,
= 5 - 2 sin(6t), 'Y = I, and /-L = 2. In each case, calculate the average contact num-
ber. How does the average contact number affect I(t) for large values of t? and period of the harvesting function h( 1 - sin 7)? Using the indicated parame-
5. Explain why diseases such as gonorrhea, meningitis, and streptococcal sore throat
ter values, describe what happens to the size of the population over one year
continue to persist in the population. Do you think there is any way to eliminate
(a) a = 1110, h = 1, yeO) = 20
these diseases completely from the population? Why or why not?
(b) a = 1110, h = 2, yeO) = 20
(e) a = 1/10, h = 4, yeO) = 20
(d) a = 1110, h = 10, yeO) = 20
B. Linear Population Model with Harvesting 8. Suppose that a situation is modeled with y' = ay - h cos 7, yeO) = Yo. That is, at
some times, fishing occurs while at others fish are added. During what time inter-
I. Consider: y' = ay - h, yeO) = Yo, where a and h are positive constants. The value vas does fishing occur? During what time intervals are fish added? What IS the
of a represents the growth rate, while that of h represents the fishing (or harvest-
ing) rate. The solution yet) represents the size of a population at time t. maximum value, minimum value, and period of the function, h cos 7? Using the
(a) Determine the equilibrium solution.
indicated parameter values, describe what happens to the size of the population over
(b) Sketch the phase line.
(e) Solve the initial-value problem. one year.
(d) If Yo > hla, determine lim y(t). (a) a = 1/10, h = 1, yeO) = 50
(b) a = 1/10, h = 4, yeO) = 50
(e) If Yo = hla, determine 1;;
t ...~
yet). (e) a = 1110, h = 10, yeO) = 50
(f) If Yo < hla, determine the value of t such that y(t) = O. What does this mean in (d) a = 1110, h = 20, yeO) = 50
terms of the popUlation?
2. Consider y' = IY - I, yeO) = Yo·

(a) Determine the equilibrium solution.
C. Logistic Model with Harvesting
(b) Sketch the phase line.
(e) Solve the initial-value problem. I. Consider the logistic model with harvesting, y' = ay - cy2 - h, yeO) = Yo, where
(d) If Yo = 3, determine lim yet). a represents the growth rate, c the inhibitive factor, and h the harvesting rate. Each
(e) If Yo = 2, determine 1;; yet). of these constants is positive.
(a) Find the equilibrium solutions by solving ay - cy2 -:- h = O. (What .condition
(f) If Yo = 1.5, determin'.;-'the value of t such that y(t) = O. What does this mean in
must hold so that there are two real solutions to this equatIOn? What conditIOn must
terms of the popUlation?
hold so that there is only one equilibrium solution? What condition must hold so
3. Consider the modely' = y - l,y(O) = 0.5, and the modely' = h - 1,y(0) = 0.5. that there is no equilibrium solution?)
In which model do you predict that the popUlation will become extinct more quickly? (b) Sketch the phase line assuming that there are two equilib?um solution~.
VerifY your response. (e) Sketch the phase line assuming that there is only one eqUlhbnum soluhon.
4. Consider the model y' = h - ty(O) = 0.5, and the model y' = h - 1, yeO) = (d) What happens to the population if there is no equilibrium solutIOn?
0.5. In which model do you predict that the population will become extinct more 7
quickly? VerifY your response. 2. Consider the problem, y' = 10 Y - /0 y2 - 1, yeO) = Yo·
5. Suppose that you own a "fish farm" and that during the first year, the growth rate (a) What are the two equilibrium solutions?
of the population is a = 112 and the harvesting rate is h = 1/2. However, during the (b) Sketch the phase line.
(c) If yeO) = 3, then detennine lim y(t). ing occurs at different rates. Investigate the solution to this IVP over 0 :S t:S 10 for
t-->~
(d) If yeO) = 6, then detennine lim yet). a = 2, 4, 6, 8, 10, 12, 14. What eventually happens to the size of the population in
(e) If yeO) = I, what do you ex~-;ct to happen to the population? each case?
3. Consider the problem, y' = ?o y - l~ y2 - h, yeO) = Yo.

(a) What value of h, h = h o, causes this model to have only one equilibrium solu- D. Logistic Model with Predation
tion? (This value is called a bifurcation point because the dynamics of the model
change for values slightly larger and smaller than this value.) One of the more predominant pests in Canadian forests is the spruce budwonn. These
(b) Describe solutions to the model with h = ho. insects, which are moths in the adult stage, lay eggs in the needles of the trees. After
(c) Describe solutions to the model with h < ho. hatching, the larvae eventually tunnel into the old foliage of the tree. Later, they spin
(d) Describe solutions to the model with h >- ho. a webbing in the needles and devour new growth until they are interrupted. Although
4. Suppose that over the first year, the rate of change in the size of a fish population they do not destroy the tree, budwonns weaken the trees and make them susceptible to
. . b ,7 1 0 1 disease and forest fires. We can model the rate of change in the size of the budworm
IS gIVen y y = 10 y - 10 y- - 2' where yeO) = 0.5. After the first year, however, popUlation by building on the logistic model. Let Wet) represent the size of the bud-
the growth rate remains the same while no fishing is allowed. What is the size worm population at time t. One factor that helps control the budworm population is
of the population after five years? Describe what happens to the population size predation by birds (that is, birds eat budworms). Consider the model,
on the interval 0 :S t :S 5. Compare this result to population that follows y' =
7 1 0 I
10 y - 10 y- - 2' where yeO) = 0.5 over the entire interval 0 :S t :S 5. (That is,
dW =
dt rW ( I - 1)
kW - peW),
fishing continues after the first year.) where peW) is a function of W describing the rate of predation. Notice that for low
5. Suppose that over the first year, the rate of change in the size of a fish population population levels, the model demonstrates exponential growth rate r. Based on the lo-
. . b ,13 I 0 I gistic model, k represents the carrying capacity of the forest. In the absence of preda-
IS gIven y y = 10 y - 10 y- - 2' where yeO) = 0.5. After the first year, however,
tion, this gives a maximum size of the budworm population.
the growth rate remains the same while no fishing is allowed. What is the size Properties of the Predation Function To develop a formula for P( W), we make
of the population after five years? Describe what happens to the population size two assumptions. First, if W= 0, then peW) = 0 (that is, if there are no budworms,
on the interval 0 :S t :S 5. Compare this result to population that follows y' = the birds have none to eat). Second, peW) has a limiting value (that is, the birds have
13 I 0 1
10 y -10 y- - 2' where yeO) = 0.5 over the entire interval 0 :S t:S 5. (That is, a limited appetite. Even if the budworm population is large, the birds eat only what
they need). A function that satisfies both of these properties is peW) = (aW2)!
fishing continues after the first year.)
(b 2 + W2). Therefore, our model becomes:
6. CanSI'der the rnadeIy' = 10
13 y - 1 y2 - 2
1SIn m yeO) =
. 6' a, . .. .
10 In WhICh fIShIng
occurs at certain times while at other times, fish are added. Investigate the solution
-dW = rW ( I
dt
1)
- -W - -2aW2
k
- - Wi(O)
b + W2 '
= W;o·
to this IVP over 0 :S t:S 12 for d = 2,4, 6, 8, 10, 12, 14. What eventually happens
to the size of the population in each case? 1. What is the limiting value of peW) = (aW2)/(b 2 + W2) as W ...... oo?
7..CanSI'der themo deIy ' 13 -lOy
= lOy 1 2- 2
1 ( 1 - SIn
. 6 m),yeO) = a, in which fish- 2. Using the indicated parameter values, approximate the equilibrium solutions. Sketch
a phase line in each case. (a) r = I, k = IS, a = 5, b = 2. (b) r = I, k = 20,
Ing occurs at dIfferent rates. Investigate the solution to this IVP over 0 :S t :S 12 for a = 5, b = 2.
a = 2, 4, 6, 8, 10, 12, 14. What eventually happens to the size of the population in
each case? 3. Using the values r = I, a = 5, b = 2, approximate the value of k that leads to three
8. Consider the model y' = 1 Y -0 (0

7
y2 - + 7,
sin yeO) = a, in which fishing
equilibrium solutions. (Do this experimentally by selecting values for k and then
plotting. This value is a bifurcation point.) Sketch the phase line.
4. Using various initial conditions, generate numerical solutions to the two situations
occurs at certain times while at other times, fish are added. Investigate the solution
to this IVP over 0 :S t:S 12 for a = 2,4, 6, 8, 10, 12, 14. What eventually happens in problem 2 above and to that in problem 3. Compare these results to the corre-
to the size of the population in each case? sponding phase line.
I
7 3
9. Consider the model y' = 10 Y - 10 y2 - +(I - sin 7),
yeO) = a, in which fish-
5. Suppose that your job is to determine the carrying capacity of the forest. How im-
portant is it that you do a good job? (Use the three cases above for guidance.) I
i
I
I
3
6. Discuss the differences in the following three models:
(a) y' = 0.48Y(1 - +Y) -~;

5 4 +Y
(b) y' = 0.48Y(1 - -.L Y) -~.
17 4 + y2'
(c)y' = 0.48Y(1 _ _ 1_y) _ ~
15.5 4 + y2'
References:
Laurentian Forestry Centre Success Stories. July 30, 1997: http://www.cfl.forestry.cal74a.htm.
Forest Heal~h Notes, by Do~a I?ekker-Robertson, Peter Griessmann, Dave Bawngartner, and Don Hanley
of the Washmgton State UruverSIty Cooperative Extension, july 7, 1997:
http://ext.nrs.wsu.edulinfo/fhn/wesprbdw.html.
Spruce Budworm, Manitoba Natural Resources Forestry Branch: http://www.gov.mb.ca/natres/forestry(
Applications of First-Order
budwonnldocwnenthtml.
An Introduction to ~he Mathematics of Biology with Computer Algebra Models, by Edward K. Yeargers,
DifferentiaL Equations
Ronald W. ShonkwIler, and James V. Herod, Birkhauser, Boston, 1996.
Spruce Budwonn Management Plan, Manitoba Model Forest Inc.:
http://www.winnipeg.rreenetmb.calmanmodf/mmfar/mmfar26.html.
Sproce Budworm, by J. A. Drouin, Northern Forest Research Centre, Edmonton:
http://www.ag.usask.calcofaldepartmentslhortlhortinfo/pestslbudworm.html.
Eating Worms. by Tim Howard, Columbus State University.
http://www.colstate.edul-thowardlwelcome.html.
hen a space shuttle is launched from the Kennedy Space Center, the
W minimum initial velocity needed for the shuttle to escape the Earth's at-
mosphere is determined by solving a first-order differential equation.
The same can be said for finding the flow of electromagnetic forces, the tem-
perature of a cup of coffee, the population of a species, and numerous other
applications. In this chapter, we show how these problems can be expressed as
first-order equations. We focus our attention on setting up problems and ex-
plaining the meaning of the subsequent solutions because techniques for solv-
ing the first-order equations were discussed in Chapter 2.
93
95
3.1 Population Growth and Decay
94 Chapter 3 Applications of First-Order Differential Equations
") Solution Let Yo represent the original amount of 209pO that is present. The
[I!J Population Growth and Decay amount present after t years is yet) = yoe • Using y(100) = ~yo and y(100) =
kt
k
100k
(' Logistic Equation (' Population Model with a Threshold yoe 100k, we solve yoe = ~yO for e :
e IOOk =!2
As we discussed in Section 2.1, the Malthus model,
dY=ky
dt ,
{
yeO) = Yo
can be used. to investigate population growth if k > O. It can also be applied to prob-
lems mvolvmg the decay of radioactive material if k < O. Hence,
Forms of a given element with different nwnbers of neutrons are called nuclides. 1 )tI100
Some nuclides are not stable. For example, potassiwn-40 (4oK) naturally decays to reach yet) = yoekt = yo(ek)t = Yo ("2 .
argon-40 (40 Ar). This decay that occurs in some nuclides was first observed, but not
under.stood,. by He~ Becquerel (1852-1908) in 1896. Marie Curie, however, began To determine the percentage of Yo that remains, we evaluate
studymg thiS decay ill 1898, named it radioactivity, and discovered the radioactive sub- 1)501100
stances poloniwn and radiwn. Marie Curie (1867-1934), along with her husband Pierre y(50) = Yo ( "2 = 0.7071yo·
Curie (1859-1906) and Henri Becquerel, received the Nobel Prize in Physics in 1903
for their work on radioactivity. Marie Curie subsequently received the Nobel Prize in Therefore, 70.71% of the original amount of209po remains after 50 years.
Chemistry in 1910 for discovering poloniwn and radiwn.
Marie Curie (1867-1934) and Given a sample of 40K of sufficient size, after 1.2 X 109 years, approximately
Pierre Curie (1859-1906) in In Example 1, we determined the percentage of the original amount of 209pO that
half of the sample will have decayed to 40Ar. The half-life of a nuclide is the time for
their Paris laboratory (1896). remains even though we do not know the value of Yo, the initial amount of 209PO. In-
half the nuclei in a given sample to decay (see Table 3.1). We see that the rate of de-
(AlP Emilio Segre VISual Archives) stead of letting yet) represent the amount of the substance present after time t, we can
cay ~f a. nuclide is prop~rtional to the amount present because the half-life of a given
let it represent the fraction (or percentage) of yo that remains after time t. In doing so,
nucILde IS constant and llldependent of the sample size.
we use the initial condition yeO) = 1 = 1.00 to indicate that 100% of Yo is present at
t= O.
TABLE 3.1 Half-lives of Various Nuclides
Element
Aluminum
Beryllium
Nuclide
26Al
IDse
7.4
l.51
Half-Life
X
X
10' years
6
10 years
Element
Polonium
Polonium
Nuclide
209pO
21OpO
Half-Life
100 years
138 days
,"'Ui,'.'
The wood of an Egyptian sarcophagus (burial case) is found to contain 63% of the
carbon-14, 14C, found in a present-day sample. What is the age of the sarcophagus?
Carbon I4C
36Cl
5730 years Radon 222Ro 3.82 days
:J Solution From Table 3.1, we see that the half-life of 14C is 5730 years. Let yet)
kt
Chlorine 3.01 X 10' years Radium 226Ra 1700 years . be the percent of 14C in the sample after t years. Theny(O) = 1. Now, y(t) = yoe ,
"II 230Th so y(5730) = e5730k = 0.5. Solving for k yields:
Iodine 8.05 days Thorium 75,000 years
In(e5730k) = In(.5)
Potassium .0K l.2 X
9
10 years Uranium 23·U 4.51 X 10
9
years
5730k = In(.5)
Henri Becquerel (1852-1908)
discovered radioactivity in 1896. 1n(.5) -In 2
Marie Curie, Pierre Curie, and k = 5730 = 5730'
Henri Becquerel shared the No- "ii""'N' Sarcophagus A Roman sar-
cophagus in the Antayla Mu-
Thus, yet) = e kt = e ~i~g t = Z-tI5730. (An alternate approach to obtain the solution
bel Prize in Physics in 1903 for seum, Perge, Turkey.
their work on radioactivity. If the half-life of poloniwn, 209pO, is 100 years, determine the percentage of the is to solve e5730k = 0.5 for e k instead of for k, as we did in Example 1. This
(Borys MalkiniAnthro-Phoro)
(North Wind Picture Archives) original amount of 209pO that remains after 50 years.
96
Chapter 3 Applications of First-Order Differential Equations
3.1 Population Growth and Decay 97
yields e k = (.5)115730 = (!)1I5730 = 2-115730 S b ftuti

_ kt _ k 1 . . . u S I on 0
f thi .
s expreSSIOn into
y(t) - yoe . - yo(e ) gives the same solutIOn as found previously.)
. In this problem, we must find the value of t for which y(t) = 0.63. Solvin
120
y (in millions)
•
•
"f'"I,'.'
this equation results in: g • The population of the United States was recorded as 5.3 million in 1800. Use the
so Malthus model to approximate the population for years after 1800 if k = 0.03. Com-
2-115730 = 0.63 = flo • pare these results to the actual population. Is this a good approximation for years
•
In(T'15730) = In flo • after 1800?
40
-t : I Solution In this example, k = 0.03, Yo = 5.3, and our model for the popula-
5730 In 2 = In flo t (in years) tion of the United States at time t (where t is the number of years from 1800) is
o 40 SO 120 yet) = 5.3e o.031 . To compare this model with the actual population of the United
-5730 In flo _ 5730(1n 100 - In 63) States, census figures for the population of the United States for various years are
t= In 2 - In 2 = 3819.48. Figure 3.1 listed in Table 3.2 along with the corresponding value ofy(l). A graph ofy(t) with
We conclude that the sarcophagus is approximately 3819 years old. the corresponding points is shown in Figure 3.l.
L
TABLE 3.2 U.S. Population and Values of yet)

. We can use the Malthus model to predict the size of a population at a given time
If the rate of growth of the population is proportional to the present population. Actual
Population Value of
Actual
Population
I Value of
Year (I) (in millions) y(f) = S.3eo.ol l Year (f) (i millions)
n yet) = S.3eo.o ll
1800 (0) 5.30 5.30 1870 (70) 38.56 43.28
1810 (10) 7.24 7.15 1880 (80) 50.19 58.42

Suppose that the number of cells in a bacteria culture doubles after three days. De-
termme the number of days reqUired for the initial population to triple. 1820 (20) 9.64 9.66 1890 (90) 62.98 78.86
1830 (30) 12.68 13.04 1900 (l00) 76.21 106.45

. ) Solution In this case, yeO) = Yo, so the population is given by yet) = yoe kt and
: ~(3) = 2yo bec!use the popUlation doubles after three days. Substituting this value 1840 (40) 17.06 17.60 1910 (110) 92.23 143.70
, mto y(t) = yoe , we have
1850 (50) 23.19 23.75 1920 (120) 106.02 193.97
y(3) = yoe 3k = 2yo.
1860 (60) 31.44 32.06 1930 (130) 123.20 261.83
~olving for ek,;::e find.that e = 2 or (e k )3 = 2. Therefore, e k = 2113. Substitution
3k I
mto yet) = yoe then Yields
yet) = yo(e k )' = Y0203. Although the model appears to approximate the data for several years after
1800, the accuracy of the approximation diminishes over time because the popula-
We find when the popUlation triples by solving y(t) = Y02'13 = 3yo for t. This yields
tion of the United States does not exclusively increase at a rate proportional to the
2
03
= 3 or fin
2 = In 3. Therefore, the population triples in t = 31n~3 = 4.755 population. Another model that better approximates the population would take other
factors into account.
days.
L
L
0~ Determine the number of days required for the culture of bacteria considered in Ex- Logistic Equation
~~r, ample 3 to reach nine times its initial size.
Because the approximation obtained with the Malthus model is less than desirable in
, To obsen:e son:e of the limitations of the Malthus model, we consider a popu- the previous example, we see that another model is needed. The logistic equation (or
lahon problem m which the rate of growth of the population does not exclusively de- Verhulst equation), which was mentioned in Section 2.3, is the equation
pend on the present popUlation.
y'(t) = (r - ay(t»)y(t),
98
Chapter 3 Applications of First-Order Differential Equations
3.1 Population Growth and Decay 99
:vhere r and a are constants, subject to the condition yeO) = Yo. This equation was first Use a compuler algebra system to solve the logistic equation. If the result you obtain
mtrodu~e~ by the.Bel~an mathematician Pierre Verhulst to study population growth. is not in Ihe same form as that given above. show (by hand) that the two are the
The loglslic equation differs from the Malthus model in that the term (r - ay(t)) is not
same.
constant. This equation can be written as dyldt = (r - ay)y = ry - ay2 where the term _
(_y2) represents an inhibitive factor. Under these assumptions the population is nei-
ther allowed to grow out of control nor grow or decay constantly as it was with the
Malthus model.
Here, we solve the logistic equation as a separable equation. For convenience we
write the equation as '
o 'if'"I,I.,
Use the logistic equation to approximate the population of the United States using
r = 0.03, a = 0.0001, and Yo = 5.3. Compare this result with the actual census val-
dy ues given in Table 3.2. Use the model obtained to predict the popUlation of the United
y' = (r - ay)y or dt = (r - ay)y. States in the year 2010.
Separating variables and using partial fractions to integrate with respect to y, we have : :1 Solution We substitute the indicated values of r, a, and Yo into
dy = dt = ryo rt to obtain the approximation of the population of the
(r - ay)y y ayo + (r aYo)e
United States at time I, where t represents the number of years since 1800,
(~
r -ay
+ l!!:.)
y
dy = dl 0.03 . 5.3 0.159
yet) = 0.0001' 5.3 + (0.03 - 0.0001 . 5. 3)e-· 031 .03/ .
0.00053 + 0.02947e
( _a_
r - ay
+ 1.)
y
dy = rdl In Table 3.3, we compare the approximati on of the popUlation of the United States
given by the approximation y (t) with the actual population obtained from census
-Inir - ayi + Iniyi = rt + c. figures. Note that this model appears to approximate the population more closely
We solve this expression for y using the properties of logarithms: over a longer period of time than the MaIthus model did (Example 4), as we can
In/-y-/
r -ay = rl +c TABLE 3.3 U.S. Population and Value 5 of yet)
--y- = ±ert + c = Ke r , Actual Actual

r- ay Population Value Population Value
Year (t) (in millions) of yet) Year (I) (in millions) of yet)
I e- rl
y = r( if +a )-1 . 1800 (0)
I
5.30
I
5.30 1900 (100) 76.21 79.61
Applying the initial condition yeO) = Yo and solving for K, we find that 1810 (10) 7.24 7.11 1910 (110) 92.23 98.33
1820 (20) 9.64 9.52 1920 (120) 106.02 119.08

~=K.
r - ayo 1830 (30) 12.68 12.71 1930 (130) 123.20 141.14
After .substituting this value into the general solution and simplifying, the solution can 1840 (40) 17.06 16.90 1940 (140) 132.16 163.59
be wntten as
1850 (50) 23.19 22.38 1950 (150) 151.33 185.45
y = ryo
ayo + (r - aYo)e rI·
1860 (60) 31.44 29.44 1960 (160) 179.32 205.82
1870 (70) 38.56 38.42 1970 (170) 203.30 224.05

Notice that liml->ro y(t) = ria because lim/->oo e- = 0 if r > O. This makes the solu-
rl
tion to the logistic equation different from that of the Malthus model in that the solu- 1880 (80) 50.19 49.63 1980 (180) 226.54 239.78
tion to the logistic equation approaches a finite nonzero limit as I -> 00 while that of
the Malthus model approaches either infinity or zero as 1-> 00. 1890 (90) 62.98 63.33 1990 (190) 248.71 252.94
...
100
Chapter 3 Applications of First·Order Differential Equations 3.1 Population Growth and Decay 101
250 Y (in millions)

se.e in ~igure 3.2. To predict the population of the United States in the year 2010 EXERCISES 3.1
With thiS model, we evaluate y(210) to obtain
y(210) =
0.00053
0.159
+ 0.02947e
- 300
0.03,210 - .
-
Solve the following problems. Unless otherwise stated, use the 10. If 90% of the initial amount of a radioactive element
Thus, we predict that the population will be approximately 300 million in the year Malthus method. remains after one day, what is the half-life of the ele-
2010. Note that projections of the popUlation of the United States in the year 2010 ment?
1. Suppose that a culture of bacteria has initial popula-
made by the Bureau of the Census range from 298.71 million to 300.59 million so tion of n = 100. If the population doubles every three *11. If y(l) represents the percent of a radioactive element
the approximation obtained with the logistic equation seems reasonable. ' days, determine the number of bacteria present after that is present at time I and the values of y(ll) and
L
30 days. How much time is required for the popula- y(t2) are known, show that the half-life H is given by
Population Model with a Threshold tion to reach 4250 in number?
Figure 3.2 (12 - t l ) In 2
2. Suppose that the popUlation in a yeast culture triples H = In[y(lll] - In[y(12l] .
Suppose that when the population of a species falls below a certain level (or thresh- every seven days. What is the population after 35 days?
old), the species cannot sustain itself, but otherwise, the population follows logistic How much time is required for the popUlation to be 12. The half-life of 14C is 5730 years. If the original
~rowth. To describe this situation mathematically, we introduce the differential equa- 10 times the initial population? amount of l4C in a particular living organism is 20 g
non
*3. Suppose that two-thirds of the cells in a culture re- and that found in a fossil of that organism is 0.0 I g,
main after one day. Use this information to determine determine the approximate age of the fossil.
the number of days until only one-third of the initial 13. After 10 days, 800 g of a radioactive element remain,
population remains. and after 15 days, 560 g remain. What is the half-life
where 0 < A < B. In this case, the equilibrium solutions are y = 0, y = A, andy = B. 4. Consider a radioactive substance with half-life 10 of this element?
To understand the dynamics of the equation, we generate the corresponding phase line. days. Ifthere are initialiy 5000 g of the substance: how
14. After one week, 10% of the initial amount of a ra-
If 0 <y < A, theny' < 0; if A <y < B, theny' > 0; and ify > B, theny' < O. From much remains after 365 days?
dioactive element decays. How much decays after two
the phase line in Figure 3.3, we see that y = 0 is asymptotically stable, y = A is un- 5. Suppose that the half-life of an element is 1000 hours. weeks? When does half of the original amount decay?
stable, and y = B is asymptotically stable. Notice that we expect the population to die If there are initially 100 g, how much remains after
1 hour? How much remains after 500 hours? *15. Determine the percentage of the original amount of
out when y < A, .so we call y = A the threshold of the population. As with the logistic 226Ra that remains after 100 years.
equatIOn, y = B IS the carrying capacity. In Figure 3.4, we graph several solutions to 6. Suppose that the popUlation of a small town is initially
5000. Due to the construction of an interstate high- 16_ If an artifact contains 40% of the amount of 230Th as
y' = -0?5(l - y)(l - y(3)y using different values for the initial population yo. No-
way, the population doubles over the next year. If the a present-day sample, what is the age of the artifact?
tice that If 0 < Yo < 1, then liml->ro yet) = 0, and if 1 < Yo < 3 or Yo > 3, then lim,....oo
yet) = 3. In this case, the threshold is y = 1 and the carrying capacity is y = 3. rate of growth is proportional to the current popula- 17. On an archeological dig, scientists find an ancient tool
tion, when will the popUlation reach 25,000? What is near a fossilized human bone. If the tool and fossil
the popUlation after five years? contain 65% and 60% of the amount of 14C as that in
*7. Suppose that mold grows at a rate proportional to the present-day samples, respectively, determine if the tool
amount present. If there are initially 500 g of mold could have been used by the human.
and 6 hours later there are 600 g, determine the amount 18. A certain group of people with initial popUlation
of mold present after one day. When is the amount of 10,000 grows at a rate proportional to the number pres-
4 mold !OOO g? ent. The population doubles in five years. In how many
8. Suppose that the rabbit population on a small island years will the popUlation triple?
grows at a rate proportional to the number of rabbits
*19. Solve the logistic equation by viewing it as a Bernoulli
present. If this population doubles after 100 days,
equation.
when does the population triple?
20. What is the limiting population, limHoo y(ll, of the
9. In a chemical reaction, chemical A is converted to
U.S. population using the result obtained in Example
15 20 chemical B at a rate proportional to the amount of
5?
chemical A present. If half of chemical A remains af-
Figure 3.3 Phase line for ter five hours, when does 1/6 of the initial amount of 21. Solve the logistic equation if r = 1/100 and a = l! I O'
Figure 3.4 Several solutions
Y' = -r(1 - ~)(I - f)y to y' = -0.25(1 - Yl( 1- f)y chemical A remain? How much of the initial amount given that y(Ol = 100,000. Find y(25l. What is the
remains after 15 hours? limiting population?
102 Chapter 3 Applications of First-Order Differential Equations 3.1 Population Growth and Decay 103
22. Five college students with the flu return to an isolated 28. (Dating Works of Art) We can determine if a work equilibrium solutions (as asymptotically stable, semi- to model the popUlation under consideration. Assume
campus of2500 students. If the rate at which this virus of art is more than 100 years old by determining if the stable, or unstable) of the following differential that h 2: r2/4a.
spreads is proportional to the number of infected stu- lead-bearing materials contained in the work Were equations. (a) Show that if h 2: r2/(4a), a general solution of
dents y and to the number not infected 2500 - y, solve manufactured within the last 100 years. The half-life. (a) y' = y(y - 2)' (b)y' =y2(y-l) dP/dl = rP - aP'- - h is
the initial-value problem dy/dl = ky(2500 - y), yeO) oflead-210 ('loPb) is 22 years, while the half-life of
= 5 to find the number of infected students after I days
if 25 students have the virus after one day. How many
radium-226 e26
Ra) is 1700 years. Let SF denote the
(e) y' =y-Vy (d)y' = y2(9 _ y2)
32. Consider the Malthus population model with k = 0.01,
P(t) =
ratio of 210Pb to 226Ra per unit mass of lead. The
students have the flu after five days?
*23. One student in a college organization of 200 members
approximate value of SF for works of art created
in the last 80 years is 100. Then, the quantity oflead
0.05, 0.1, 0.5, and 1.0 using Yo = I. Solve the model,
plot the solution with these values, and compare the
L [r + Y4ah
1
- r2 tan(2 a (C - al)Y4ah - 1'2)].
proceeds to spread a rumor. If the rate at which this (i - Ra)/(Po) at time I is given by results. How does the value of k affect the solution? (b) Suppose that for a certain species it is found
rumor spreads is proportional to the number of stu- 33. Considerthelogisticequationwithr = 0.01,0.05,0.1, that r = 0.03, a = 0.0001, h = 2.26, and
dents y that know about the rumor as well as the num- _I_-_R_a = (SF - I)e- Ar 0.5, and 1.0 using Yo = I and a = I. Solve the model, C = -1501.85. At what time will the species be-
ber that do not know, then solve the initial-value prob- Po (SF-I)e Ar+l' plot the solution with these values. and compare the come extinct?
lem to find the number of students informed of the results. How does the value of r affect the solution?
rumor after I days if 50 students are informed after where A is the disintegration constant for 21OPb.* (e) If r = 0.03, a = 0.0001, and P(O) = 5.3, graph
34. We may use graphing utilities to investigate the be- P(I) if h = 0, h = 0.5, h = 1.0, h = 1.5, h = 2.0,
one day. How many students know the rumor after two On the other hand, for very old paintings (i - Ra)!
Po = O. havior of solutions of the logistic equation under dif- h = 2.25 and h = 2.5.
days? Will all of the students eventually be informed ferent conditions. (a) Use a graphing utility to inves-
of the rumor? (See Exercise 22.) (a) Determine the disintegration A for 210Pb where the tigate the behavior of the solution if values ofr, a, and (d) What is the maxunum allowable harvest rate to
24. Suppose that glucose enters the bloodstream at the amount of 210Pb at time tis y = yoeAt. Yo are varied. Under what conditions does the limit of assure that the species survives?
constant rate of r grams per minute while it is removed (b) Graph (i - Ra)/(Po) for 0,;; I ,;; 250 using tbe solution approach (b) a nonzero number as (e) Generalize your result from (d). For arbitrary a
at a rate proportional to the amount y present at any SF= 100. 1 - 00; and (e) zero as I~ oo? Hinl: Consider and r, what is the maximum allowable harvest rate
time. Solve the initial-value problem dy/dl = r - ky, 0< 1'';; 1, r> I; 0 < a < I, a > I. that ensures survival of the species?
The table on page 103 shows the ratio of
yeO) = Yo to find y(I). What is the eventual concen- 35. (Harvesting) If we wish to model a population of size
(I - Ra)/(po) for various famous paintings. 36. Shortly after the Chernobyl accident in the Soviet
tration of glucose in the bloodstream according to this P(I) at time I and consider a constant harvest rate h
The last two paintings, Lace Maker and Laugh- Union in 1986, several nations reported that the level
model? (like hunting, fishing, or disease), then we might mod-
ing Girl were painted by the Dutch painter Jan Ver- of 131 1 in milk was five times that considered safe for
25. What is the concentration of glucose in the blood- meer who lived from 1632 to 1675. ify the logistic equation and use the equation human consumption. Make a table of the level of l311
stream after 10 min if r = 5 g per min, k = 5, and the in milk as a multiple of that considered safe for hu-
initial concentration is yeO) = 500? After 20 min? (e) Determine if it is likely that the first six paintings dP = rP - ap2 - h
were also painted by Vermeer (which would make dl man consumption for the first three weeks following
Does the concentration appear to reach its limiting
value quickly or slowly? (See Exercise 24.) them very valuable!). If not, approximate when
they were painted.
26. Suppose that we deposit a sum of money in a money
market fund that pays interest at an annual rate k, and *29. Consider the differential equation dy/dl =
let S(I) represent the value of the investment at time I. -r(i - y/A)y, where r and A are positive constants. 22. Ra concentration
210po concentration 1 - Ra
If the compounding takes place continuously, then the (a) Find the equilibrium solutions, sketch the phase
Painting (dpm/g of Pb) (dpm/g of Pb) Po
rate at which the value of the investment changes line, and classify the equilibrium solutions. (b) De-
is the interest rate times the value of the investment, scribe how this equation differs from the logistic equa- Washing of Feel 12.6 0.26 0.98
dS/dl = kS. Use this equation to find S(I) if S(O) = So. tion. (e) If A = 2 and y(O) = 1, what is limr-.oo y(I)?
(d) If A = 2 and y(O) = 3, what is limr-.oo y(I)? Woman Reading Music 10.3 0.30 0.97
27. Banks use different methods to compound interest. If
m times per year, then S(I) = So( I + 1;r'.

the interest rate is k, and if interest is compounded
When
30. Find and classify the equilibrium solutions of the
Gompertz equation, dy/dl = y(r - a Iny), where r and
a are positive constants.
Woman Playing Mandolin
Woman Drinking
8.2
8.3
0.17
0.1
0.98
0.99
Compare limr-.oo So( I

in Exercise 26.
+ r'
interest is compounded continuously, then m ~
1;
00.
to the formula obtained

*31. The equilibrium solution y = c is classified as semi-
stable if solutions on one side approach y = c as
I _ 00, but on the other side they move away from
y = c as I _ 00. Use this definition to classify the
Disciples of Emmaus
Boy Smoking
8.5
4.8
0.8
0.31
0.91
0.94
Lace Maker 1.5 1.4 0.07
... Bernard Keisch, "Dating Works of Art through Their Natural Radioactivity: Improvements and Applications," Science, Volume 160 (April 26, Laughing Girl 5.2 6.0 -0.15
1968), pp. 413-415. ',::.,..~-'t="C!"'~_=z:--=.::.~..=~_~~~":::""'-~=='-~-:""
....
3.2 Newton's Law of Cooling and Related Problems 105
the accident. After how long did the milk become safe quires a large number of cohorts to achieve success_ is separable, and separating variables gives us
for human consumption? ful reproduction. Having fallen below this level, the dT
37. Consider a solution to the logistic equation with ini- popUlation size continued to decrease and the bird is T- Ts = kdt,
tial population Yo where 0 < Yo < ria. Show that this now extinct. Which popUlation model should be used -
solution is concave up for 0 < Y < rl(Za) and concave to describe this situation? so lnlT - Tsl = kt + C. Using the properties of natural logarithms and simplifYing
down for r/(Za) < y < ria. Hint: Differentiate the 39. (a) Find the solution to the IVP dyldt = -r(1 - yIA)y, yields
right side of the logistic equation and set the result yeO) = Yo, where r and A are positive constants. (b) If
equal to O. Describe the behavior of dyldt based on T = C 1e kt + Ts,
Yo < A, then determine limH~ y(t). (e) If Yo > A,
this result. show that limH~y(t) = +00. (d) Show that if Yo > A, where C1= eC. Applying the initial condition implies that To = C1+ Ts, so C1=
38. From the early 1800s to the mid-1800s, the passenger Yo ) . To - Ts. Therefore, the solution of the equation is
then Y(I) has a vertical asymptote at I = -1 In( --A-
pigeon population was thriving. However, due to hunt- r Yo-
T = (To - Ts)e kt + Ts·
= r( 1 -~)( 1 - ~)y,
ing, the population size was reduced dramatically by
the late 1800s. Unfortunately, the passenger pigeon re- 40. Solve the IVP y' yeO) = Yo.
Notice that if k < 0, lim,...,oo e kt = O. Therefore, lim,...,~ T(t) = Tso so the temperature
of the body approaches that of its surroundings.
To better understand the model, suppose that k < O. If T> Tso then dT/dt < 0,
so T(t) is a decreasing function. Similarly, if T < Tso then dT/dt > 0, so T(t) is an in-
§ Newton's Law of Cooling and Related Problems
creasing function. Therefore, the object cools off if the surrounding temperature is
greater than the object's temperature, whereas it warms up if the temperature of the
surroundings is warmer than the object's temperature.
First-order linear differential equations can be used to solve a variety of problems that Notice that instead of solving the ODE as a separable equation, we could have
involve temperature. For example, a medical examiner can find the time of death in a
solved it as a first-order linear equation.
homicide case, a chemist can determine the time required for a plastic mixture to cool
to a hardening temperature, and an engineer can design the cooling and heating sys-
tem of a manufacturing facility. Although .distinct, each of these problems depends on
a basic principle, Newton's law of cooling, which is used to develop the differential
equation associated with each problem.
"f·"I,I.'
A pie is removed from a 350°F oven and placed to cool in a room with tempera-
ture 75°F. In 15 min, the pie has a temperature of l50°F. Determine the time re-
quired to cool the pie to a temperature of 80°F.
Newton's Law of Cooling
: 1 Solution In this example, To = 350 and Ts = 75. Substituting thesektvalues into
The rate at which the temperature T(t) changes in a cooling body is proportional T = (To - Ts)e kt + Tso we obtain T(t) = (350 - 75)e kt + 75 = 275e + 75. To
to the difference between the temperature of the body and the constant temper- solve the problem we must find k or e k. We know that T(15) = 150, so T(15) =
ature Ts of the surrounding medium. 275e 15k + 75 = 150. Solving this equation for e k gives us
275e 15k = 75
This situation is represented as the first-order initial-value problem

dT
dt = k(T - Ts), T(O) = To,
where To is the initial temperature of the body and k is the constant of proportionality.
The equation e
k = (.2....)1115
11 .
dT = k(T- T) Thus, T(t) = 275(e k )' + 75 = 275(fr),/15 + 75.

dt s
106 Chapter 3 Applications of First-Order Differential Equations 3.2 Newton's law of Cooling and Related Problems 107
To find the value of I for which T(/) = 80, we solve the equation 275(nyils This result means that the time of death occurred approximately 1.53 hours
+ 75 = 80 for I: before the body was discovered, as we observe in Figure 3.5. Therefore, the time of
death was approximately 10:30 A.M. because the body was discovered at noon.
3 )'/1S
275 (
11 = 5 T(t)
60 In each of the previous cases, the temperature of the surroundings was assumed
2-.)'/1S =...!...
( 11 55 40
to be constant. However, this does not have to be the case. For example, determining
the temperature inside a building over the span of a 24-hour day is more complicated
3
In( 11 t s
In( 515) = -In 55
=
20
t (in hours)
because the outside temperature varies. If we assume that a building has no heating or
air conditioning system, the differential equation that needs to be solved to find the
temperature u(t) at time I inside the building is
I 5 In( {I) -In 55
t -2 -1 2 4
= -1.53
Figure 3-5 Graph of ~~ = k( C(t) - u(t))

-15ln55 -15ln 55 T(t) = 17(f,)t12 + 65
t= 46.264.
3 In3-ln11
where cet) is a function that describes the outside temperature and k > 0 is a constant
In( 11 )
that depends on the insulation of the building. According to this equation, if C(t) >
u(/), then duldt > 0, which implies that u increases. However, if cet) < u(t), then
Thus, the pie will reach a temperature of 80°F after about 46 minutes.
dul dt < 0, which means that u decreases.
An interesting problem associated with this example is to determine if the pie
ever reaches room temperature. From the formula T(t) = 275(n)'lls + 75, we note
'-
that 275(n)'/1S > 0, so T(/) = 275(n),lls + 75 > 75. Therefore, the pie never actu-
ally reaches room temperature according to our model. However, we see its tem-
perature approaches 75° as t increases because lim ....= [275(n)'lls + 75] = 75. o "ii,,',',.. Suppose that during the month of April in Atlanta, Georgia, the outside temperature
in OF is given by cet) = 70 - 10 cos( 711/12), 0 :5 t :5 24. (This implies that the
average value of cet) is 70 0 E) Determine the temperature in a building that has an
l'i'"I,I., initial temperature of 60°F if k = 114.
In the investigation of a homicide, the time of death is important. Newton's law of
cooling can be used to approximate this time. For example, the normal body tem- : ] Solution The initial-value problem that we must solve is
perature of most healthy people is 98.6°E Suppose that when a body is discovered
at noon, its temperature is 82°E Two hours later, it is 72°E If the temperature of the du = k (70 - 10 cos 711 - u)
dt 12
surroundings is 65°F, what was the approximate time of death? { u(O) = 60 .
:I Solution This problem is solved like the previous example. Let T(t) denote the The differential equation can be solved if we write it as
temperature of the body at time t, where T(O) represents the temperature of the body ~~ + ku = k (70 - lOcos ~) and use an integrating factor. This gives us
when it is discovered and T(2) represents the temperature of the body two hours af-
ter it is discovered. In this case we have To = 82 and Ts = 65, and substituting these
values into T = (To - Ts)e kt + Ts yields T(t) = (82 - 65)e kt + 65 = 17e kt + 65. u(t) = 9 !O,,? (63 + 7"? - 9 cos ~ - 37Tsin ~) + C e- 1
1/4
.
Using T(2) = 72, we solve the equation T(2) = 17e 2k + 65 = 72 for e k to find
e k = (il)1/2, so T(t) = 17(e k )' + 65 = 17(ily/2 + 65. To find the value of t for which We then apply the initial condition u(O) = 60 to determine the arbitrary constant C1
T(t) = 98.6, we solve the equation 17(il)'/2 + 65 = 98.6 for t and obtain and obtain the solution
t = 2 In(1.97647)
-1.53569.
_ 10 (63
t - 9 + ,,?
u() + 771'" -
0 9 711
cos 12 -
3 . 711)
7T sm 12 - 9
10"? -1/4
+ ,,? e ,
ln7-lnl7
108 Chapter 3 Applications of First-Order Differential Equations 3.2 Newton's law of Cooling and Related Problems 109
where k = 114, which is graphed in Figure 3.6. From the graph, we see that the tem- 1 Solution We notice that the volume (of liquid) does not remain constant in the
perature reaches its maximum (approximately 72°F) near t = 15.5 hours, which cor- '*
tank because R, R 2 . Therefore, we determine Vet) by solving the IVP dV/dt =
responds to 3:30 P.M. 4 - 3 = 1, V(O) = 500 with the general solution V(t) = t + C,. Applying the initial
L condition yields V(t) = t + 500, so that the IVP used to find y(t) is
At what time during the day is the temperature in the building increasing (decreas- dy 3y
ing) at the fastest rate? dt = 4- t + 500' yeO) = 250.
In many situations, a heating or cooling system is installed to control the tem- Rewriting the ODE as dy/dt + (3y)/(t + 500) = 4, we use the integrating factor
u (in oF)
perature in a building. Another factor in determining the temperature, which we ig- /L(t) = eJ(3dt)I(t+500) = e 3'n(t+500) = (t + 500)3, t 2: 0
nored in prior calculations, is the generation of heat from the occupants of the build-
75
ing, including people and machinery. We investigate the inclusion of these factors in
70 the exercises at the end of this section.
to obtain ;t [(t + 500iy] = 4(t + 500)3 so that (t + 50Wy = (t + 500)4 + C2 •
65
An application that involves' a differential equation similar to the one encoun- Therefore, yet) = t + 500 + C2 /(t + 50W, so that yeO) = 250 indicates that
t (in hours)
tered with Newton's law of cooling is that associated with mixture problems. In this C2 = -250.500 3 = -31,250,000,000. After graphing yet) = t + 500 -
case, suppose that a tank contains Vo gallons of a brine solution, a mixture of salt and 31,250,000,000/(t + 500)3, we can use a numerical solver to find that the approxi-
10 15 20
water. A brine solution of concentration S, pounds per gallon is allowed to flow into mate solution to t + 500 - 31,250,000,000/(t + 500)3 = 1000 is t = 528.7 min.
Figure 3.6 the tank at a rate R, gallons per minute while a well-stirred mixture flows out of the However, if t = 528.7, then V(528.7) = 528.7 + 500 = 1028.7 gal, so the tank
tank at the rate of R2 gallons per minute. If yet) represents the amount (in pounds) of would overflow before 1000 lb of salt could be present in the tank.
salt in the tank at time t minutes, then the equation that describes the net rate of change
in the amount of salt in the tank is
dy
dt = (rate salt enters the tank) - (rate salt leaves the tank)
EXERCISES 3.2
= (S, ~)(R
gal
~) - (y(t)
' min
~)(R
V( t) gal 2
~)
min '
where Vet) is the volume of solution (liquid) in the tank at any time t and y(t)/V(t) rep- 1. A hot cup of tea is initially 1000e when poured. How that the temperature of the surroundings is 60°F. (Nor-
resents the salt concentration of the well-stirred mixture. To determine V(t), we solve long does it take for the tea to reach a temperature of mal body temperature is 9S.6°F.)
the equation that gives the net rate of change in the amount of liquid in the tank, 50 0e if it is at sooe after 15 min and the room tem- 6. At the request of his children, a father makes home-
perature is 30 0e? made popsieles. At 2:00 P.M., one of the children asks
~= (Rate liquid enters the tank) - (Rate liquid leaves the tank) 2. Suppose that the tea in Exercise 1 is allowed to cool if the popsicles are frozen (O°C), at which time the fa-
at room temperature for 20 min. It is then placed in a ther tests the temperature of a popsiele and finds it to
cooler with temperature 15°e. What is the tempera- be 5°e. If the father placed the popsieles with a tem-
=(R ~)-(R~)
, min 2 min ture of the tea after 60 min if it is 60 0e after 10 min
in the cooler?
perature of 15°e in the freezer at 12:00 P.M. and the
temperature of the freezer is - 2°e, when will the pop-
using the initial condition V(O) = Vo. We then use the solution Vet) to solve the previ- *3. A can of orange soda is removed from a refrigerator sieles be frozen?
ous ODE for y(t) where yeO) = Yo is the initial amount of salt in the tank. having temperature 40°F. If the can is at 50°F after *7. A thermometer that reads 90°F is placed in a room
5 min, how long does it take for the can to reach a with temperature 70°F. After 3 min, the thermometer
temperature of 60°F if the surrounding temperature reads SO°F. What does the thermometer read after
I'i',,','.' is 75°F?
4. Suppose that a container of tea is placed in a refrig-
5 min?
8. A thermometer is placed outdoors at SO°F. After 2 min,
Suppose that S, = 1, R, = 4, and R2 = 3. In addition, suppose that the tank initially erator at 35°F to cool. Ifthe tea is initially at 75°F and the thermometer reads 6soF, and after 5 min, it reads
holds 500 gal of liquid and 250 Ib of salt. Find the amount of salt contained in the it has a temperature of 70°F after 1 hour, then when n°F. What was the initial temperature reading of the
tank at any time. Determine how much time is required for 1000 lb of salt to accu- does the tea reach 55°F? thermometer?
mulate. If the tank holds a maximum of 800 gal, can 1000 Ib of salt be present in 5. Determine the time of death if a corpse is at 79°F when 9. A casserole is placed in a microwave oven to defrost.
the tank at any time before the tank reaches its maximum capacity? discovered at 3:00 P.M. and 6soF 3 hours later. Assume It is then placed in a conventional oven at 300°F and
-=~--------------------------------------------------~-----------------------------------
110 Chapter 3 Applications of First-Order Differential Equations 3.3 Free-Falling Bodies 111
= 4 gal/min, R2 = 4 galImin, SI = 2Ib/gal, a thermostat. Suppose that the desired temperature is Ud' Then
bakes for 30 min, at which time its temperature is (a) RI ~~ = k(C(t) - u(t)) + A(t) + B(t).
150°F. If after baking an additional 30 min its tem- Vo = 400, Yo = 0 B(t) = k,AUd - u(t)), where kd is a constant approximately equal
perature is 200°F, what was the temperature of the to two for most systems.
(b) RI = 4 gal/min, R2 = 2 gal/min, SI = lib/gal, where the constant k > 0 depends on the insulation of
casserole when it was removed from the microwave? Vo = 500, Yo = 20 the building. 22. Determine the temperature (and the maximum tem-
10. A bottle of wine at room temperature (70°F) is placed (e) R, = 4 gal/min, R2 = 6 gal/min, S, = lIb/gal, 21. Find the temperature (and the maximum temperature) perature) in a building with k = 114 and initial
in ice to chill at 32°F. After 20 min, the temperature Vo = 600, Yo = \00 in the building with k = 114 if the initial temperature temperature 70° if (a) A(t) = 0.25, B(t) =
of the wine is 58°F. When will its temperature be is 70°F, and (a) A(t) = 0.25, C(t) = 75. and B(t) = 0; \.75(68 - u(t)), and C(t) = 70 - 10 cos (711112);
Describe how the values of R, and R2 affect the
50°F? (b) A(t) = 0.25, C(t) = 70 - 10 cos (711/12), and (b) A(t) = 1, B(t) = \.75(68 - u(t)), and C(t) =
volume of liquid in the tank.
*11. When a cup of coffee is poured its temperature is B(t) = 0; (e) A(t) = I, C(t) = 70 - 10 cos (711112), 70 - 10 cos (711112); (e) A(t) = 0.25, B(t) =
200°F. Two minutes later, its temperature is 170°F. If 18. A tank contains \00 gal of a brine solution in which 1.75(68 - u(t)), and C(I) = 80 - 10 cos (711/12).
andB(t) = O.
20 Ib of salt is initially dissolved. (a) Water (contain-
the temperature of the room is 68°F, when is the tem- *23. If A(t) = 0.25, B(t) = 1.75(Ud - u(I)), and C(t) = 70
perature of the coffee 140°F? ing no salt) is then allowed to flow into the tank at a
If a heating or cooling system is considered, we must model the - lOcos (711112); determine the value of Ud needed so
rate of 4 gal/min and the well-stirred mixture flows
12. After dinner, a couple orders two cups of coffee. Upon system with an appropriate function. Of course, the system that the average temperature in the building over a 24-
out of the tank at an equal rate of 4 gal/min. Deter-
being served, the gentleman immediately pours one could run constantly, but we know that most are controlled by hour period is 70°F.
mine the amount of salt yet) at any time t. What is the
container of cream into his cup of coffee. His com- eventual concentration of the brine solution in the
panion waits 4 min before adding the same amount of
tank? (b) If instead of water a brine solution with con·
cream to her cup. Which person has the hotter cup of
coffee when they both take a sip of coffee after she
centration 2 Ib/gal flows into the tank at a rate of
4 gal/min, what is the eventual concentration of the
§ Free·Falling Bodies
adds the cream to her cup? (Assume that the cream's
brine solution in the tank?
temperature is less than that of the coffee.) Explain. The motion of some objects can be determined through the solution of a first-order
*19. A tank contains 200 gal of a brine solution in which equation. We begin by explaining some of the theory that is needed to set up the dif-
13. Suppose that during the month of February in Wash-
10 Ib of salt is initially dissolved. A brine solution with ferential equation that models the situation.
ington, DC., the outside temperature in OF is given by
concentration 2 Ib/gal is then allowed to flow into the
C(t) = 40 - 5 cos (711112).0 :5 t :5 24. Determine the
tank at a rate of 4 gal/min and the well-stirred mix-
temperature in a building that has an initial tempera-
ture flows out of the tank at a rate of 3 gal/min. De-
ture of 50° if k = 1/4. (Assume that the building has Newton's Second law of Motion
termine the amount of salt yet) at any time t. If the
no heating or air conditioning system.)
tank can hold a maximum of 400 gal, what is the con-
14. Suppose that during the month of August in Savan- The rate at which the momentum of a body changes with respect to time is equal
centration of the brine solution in the tank when the
nah, Georgia, the outside temperature in OF is given volume reaches this maximum? to the resultant force acting on the body.
by C(t) = 85 - 10 cos (711112), 0 :5 t :5 24. Deter-
20. A tank contains 300 gal of a brine solution in which
mine the temperature in a building that has an initial
30lb of salt is initially dissolved. A brine solution with
temperature of 60°F if k = 114. (Assume that the
concentration 4 Ib/gal is then allowed to flow into the The body's momentum is defined as the product of its mass and velocity, so this
building has no heating or air conditioning system.)
tank at a rate of 3 gal/min, and the well-stirred mix- statement is modeled as
*15. Suppose that during the month of October in Los An- ture flows out of the tank at a rate of 4 gal/min. De-
geles, the outside temperature in OF is given by C(t) termine the amount of salt yet) at any time t. What is d
= 70 - 5 cos (711/12), 0 :5 t:5 24. Find the tempera-
the concentration of the brine solution after 10 min?
dt (mv) = F,
ture in a building that has an initial temperature of What is the eventual concentration of the brine solu-
65°F if k = 1/4. (Assume that the building has no heat tion in the tank? For what values of t is the solution where m and v represent the body's mass and velocity, respectively, and F is the sum
or air conditioning system.) defined? Why? of the forces acting on the body. The mass m of the body is constant, so differentia-
16. Suppose that during the month of January in Cincin- tion leads to the differential equation
nati, Ohio, the outside temperature in OF is given by The temperature u(t) inside a building can be based on three
C(t) = 20 - 5 cos (711112), 0 :5 t :5 24. Find the tem- factors: (I) the heat produced by people or machinery inside
perature in a building that has an initial temperature the building, (2) the heating (or cooling) produced by the fur·
of 40°F if k = 1/4. (Assume that the building has no nace (or air conditioning system), and (3) the temperature out- If the body is subjected to the force due to gravity, then its velocity is determined by
heat or air conditioning system.) side the building based on Newton's law of cooling. If the rate solving the differential equation
17. Find the amount of salt yet) in a tank with initial vol- at which these factors affect (increase or decrease) the temper-
ume Vo gallons of liquid and Yo pounds of salt using ature is given by A(t), B(t), and C(t), respectively, the differen- dv dv
the given conditions. tial equation that models this situation is
n!di= mg or 7t=g,
where g = 32 ftlsec 2 (English system) or g = 9.8 mlsec2 (international system). See - Solution (a) First, we set up the initial-value problem to determine the velocity
the summary of units in Table 3.4. of the object. The air resistance is equivalent to the instantaneous velocity, so
FR = v. The formula m ~~ = mg - FR then gives us

TABLE 3.4 Units Useful in Solving Problems
Associated with Newton's Second Law of Motion
dv = 32 - v
dt '
English International
and imposing the initial velocity yeO) = 2 yields the initial-value problem
Mass I Cb-sec2)
sug -ft- kilogram (kg)
dv = 32 - v
dt
Force pound (lb) Newton(~)
sec
{
yeO) = 2 '
which can be solved through several methods. We choose to solve it as a linear first-
Distance foot (ft) meter (m)
order equation and use an integrating factor. (It also can be solved by separating
Time second (sec) second (sec)
variables.) With the integrating factor e', we have ! (e'v) = 32e'. Integrating both
sides gives us e'v = 32e' + C, so
This differential equation is applicable only when the resistive force due to the v = 32 + Ce-',
medium (such as air resistance) is ignored. If this offsetting resistance is considered,
we must discuss all of the forces acting on the object. Mathematically, we write the
and applying the initial velocity gives us yeO) = 32 + Ceo = 32 + C = 2, so
C = - 30. Therefore, the velocity of the object is
equation as
v = 32 - 30e-'.
m ~~ = I (forces acting on the object),
Notice that the velocity of the object cannot exceed 32 ftlsec, called the limiting
negative direction velocity, which is found by evaluating limt->~ v(t).
FR
where the direction of motion is taken to be the positive direction.
To determine the distance traveled at time t, s(t), we solve the first-order
We use a force diagram in Figure 3.7 to set up the differential equation that mod-
too~
equation
els the situation. Air resistance acts against the object as it falls and g acts in the same
direction of the motion. We state the differential equation in the form:
ds = v = 32 - 30e-'
dt
dv
or mat = mg-FR ,
with initial condition s(O) = O. This differential equation is solved by integrating
s (in feet)
where FR represents this resistive force. Note that down is assumed to be the positive both sides of the equation to obtain s = 32t + 30e -, + C2 • Application of s(O) = 0
mg
direction. The resistive force is typically proportional to the body's velocity (v) or a 2000 ,nm / then gives us s(O) = 32(0) + 30e o + C2 = 30 + C2 = 0, so C2 = -30, and the dis-
positive direction tance traveled by the object is given by
power of its velocity. Hence, the differential equation is linear or nonlinear based on 1500 (32.1875\0/
Figure 3.7 Force diagram the resistance of the medium taken into account. s = 32t + 30e-' - 30.
1:::, / (b) The object strikes the ground when s(t) = 1000. Therefore, we must solve the
"ii,,','.' V. . t (in seconds)
equation s = 32t + 30e-' - 30 = 1000 for t. The roots of this equation can be
approximated with numerical methods like Newton's method. From the graph of
(a) Determine the velocity and the distance traveled by an object with mass m = I o 20 _I. 40 60 this function, shown in Figure 3.8, we see that set) = 1000 near t = 35. Numerical
32.1875
slug that is thrown downward with an initial velocity of 2 ftlsec from a height of methods show that s = 32t + 30e-' - 30 = 1000 when t = 32.1875.
1000 ft. Assume that the object is subjected to air resistance that is equivalent to the Figure 3.8 Graph of The velocity at the point of impact is found to be 32.0 ftlsec by evaluating the
instantaneous velocity of the object. (b) Determine the time at which the object s = 32t + 30e -, - 30 derivative at the time at which the object strikes the ground, given by s' (32.1875).
strikes the ground and its velocity when it strikes the ground.
"F''''''.'
Suppose that the object in Example 1 of mass I slug is thrown downward with an
30
25
v (in ftis)
----_._._._.__._-_.:g
v (I) from Examp~.!- _ - - -
,."
initial velocity of 2 ftlsec and that the object is attached to a parachute, increasing- ,,/
20 ..-
this resistance so that it is given by v 2 • Find the velocity at any time t and deter-
IS ..- ..- ..-
mine the limiting velocity of the object. /
10 /
/ Figure 3.9 The velocity functions from Ex-
4--125 /- ample I and Example 2. Notice how the dif-
1 Solution This situation is modeled by the initial-value problem v (t) from Example 2 I (in s)
ferent forces due to air resistance affect the
dv = 32 _ v 2 0.5 1.5 2 2.5 velocity of the object.
dt
{
yeO) = 2
We solve the differential equation by separating the variables and using partial
fractions: o "ii,,'.,.,
Determine a solution (for the velocity and the height) of the differential equation
dv that models the motion of an object of mass m when directed upward with an ini-
positive direction
32 _ v2 = dt
motion tial velocity of va from an initial position So, assuming that the air resistance equals
dv _ dt ev where e is a positive constant.
(4v2 + v)(4v2 - v) -
: ) Solution The motion of the object is upward, so g and FR act against the upward
_1_ [ 1 ___1__] dv = dt motion of the object as shown in Figure 3.10.
8v2 v + 4v2 v - 4v2 Therefore, the differential equation that must be solved in this case is the lin-
ear equation dvldt = -g - (clm)v. We solve the initial-value problem
lnlv + 4v21 - Inlv - 4v21 = 8v2t +C
g
Inl v-4v2
v + 4v21 = 8v2t + C negative direction ~~=-g-;;v
{
Figure 3.10 By drawing a yeO) = Vo
force diagram, we see that g and
v + 4v21 = Ce 8Vh .. the equation
· rewntmg dv = -g - -;;;
e v as dv + -;;;
e v = -g and t h en ca-
Iv-4v2
FR are in the negative direction. bY f Irst .
at I at
v + 4v2 = Ke 8VZt (K = :tC). culating the integrating factor eI~d' = e
ctlm . Multiplying each side of the equation
v-4v2
by e ctlm gives us e ctlm dv + .£... ve ctlm = _geC'lm so that..E... (ecdmv) = _ge ctlm . Inte-
8VZt )v = dt m dt
Solvi~ for v, we find that v + 4v2 = Ke 8VZ,(v - 4v2) or (l - Ke
-4Y2(Ke 8VZt + 1), so grating we obtain ectlmv = - gm e ctlm + C and, consequently, the general solution is
e
-4v2(Ke 8Vzt + 1)
v= vet) = -gm + Ce- ctlm .
1 - Ke 8Vzt e
1+ 2v2 Applying the initial condition yeO) = va and solving for C yields C =
Application of the initial condition yields K = , (;:;. The limiting velocity of
1- 2v2 (cvo + gm)/e so that the solution to llie initial-value problem is
.. . . I . ( -4v2K , (;:;2 ft/
the object IS found wIth I:Hopltal's ru e to be hmt-->oo v t) = ~ = 4 v L. sec. gm eva + gm -cdm
v(t) = --;::- + e e .
In Example 1, the limiting velocity is 32 ft/sec, so the parachute causes the
velocity of the object to be reduced (see Figure 3.9). This shows that the object does For example, llie velocity function for llie case willi m = 11128 slugs, e =
not have to endure as great an impact as it would without the help of the parachute. 11160, g = 32 ft/sec 2 and va = 48 ft/sec is v(t) = 8Se- 4t15 - 40. This function is
v (in ft/s) graphed in Figure 3.11. Notice where v(t) = O. This value of t represents the time This problem must be broken into two parts: an irlitial-value problem for the
40 at which the object reaches its maximum height and begins to fall toward the ground. object above the pond and an irlltial-value problem for the object below the surface
Similarly, this function can be used to investigate numerous situations with- of the pond. Using techniques discussed in previous examples, the initial-value prob-
30
out solving the differential equation each time. lem above the pond's surface is found to be
20
The height function set), which represents the distance above the ground at
10 time t, is determined by integrating the velocity function: dv = 32 - v
dt
{ yeO) = 0 .
0.5
-10 s(t) = Jvet) dt = J( -~ + cVo ; gm e- ct1m ) dt
However, to detennine the irlltial-value problem that yields the velocity of the ob-
-20
gm cmvo + gm2 ject beneath the pond's surface, the velocity of the object when it reaches the sur-
= -- t- e- ct1m
+ C. face must be known. Hence, the velocity of the object above the surface must be
Figure 3.11 c c2
s (in ft)
detennined first.
If the initial height is given by s(O) = So, solving for C results in 120
The equation dvldt = 32 - v is separable and rewriting it yields dv/(32 - v) = dt.
v (I) C = (gm2 + c2so + cmvo)/c2, so that 100
Integrating and applying the initial condition results in v(t) = 32 - 32e -to To find
80
the velocity when the object hits the pond's surface, we must know the time at which
40
_ gm cmvo + gm 2 -crlm gm 2 + ?so + cmvo 60
the object has fallen 50 ft. Thus, we find the distance traveled by the object by solv-
30 s (t) - ---;;- t - c2 e + c2 ing dsldt = vet), s(O) = 0, obtaining s(t) = 32e- t + 32t - 32. From the graph of set)
40
20
20
shown in Figure 3.13, we see that the value of t at which the object has traveled 50
10 The height and velocity functions are shown in Figure 3.12 using the parameters ft appears to be about 2.5 sec.
m = 11128 slugs, c = 11160, g = 32 ftlsec 2 , and Vo = 48 ftlsec as well as So = O. A more accurate value of the time at which the object hits the surface is t =
The time at which the object reaches its maximum height occurs when the 2.47864. The velocity at this time is then detennined by substitution into the ve-
-10 derivative of the position is equal to zero. From Figure 3.12, we see that s' (t) = locity function, resulting in v(2.47864) = 29.3166. Note that this value is the initial
Figure 3.13 Graph of
-20 vet) = 0 when t = 1. Solving s' (t) = 0 for t yields the better approximation set) = 32e - t + 32t - 32 velocity of the object when it hits the surface of the pond.
t = 0.985572. Thus, the initial-value problem that determines the velocity of the object
Figure 3.12 Graph of vet) = beneath the surface of the pond is given by dvldt = 32 - 6v, yeO) = 29.3166. The
SSe -4t15 - 40 (in blue) and solution of this initial-value problem is v(t) = 16/3 + 23.9833e -6t, and solving
set) = 110 - 40t - IIOe- 4tI5 Weight and Mass Notice that in the 'English system, pounds describe force. There· ds/dt = 16/3 + 23.9833e -6t, s(O) = 0 we obtain
(in black) fore, when the weight W of an object is given, we must calculate its mass with the
relationship W = mg or m = Wig. Conversely, in the International system, the mass s (in ft)
set) = 3.99722 - 3.99722e -61 + 136 t,
of the object (in kilograms) is typically given. 30 (3.93802,25)
25+-------..,..
We now combine several of the topics discussed in this section to solve the fol- which gives the depth of the object at time t. From the graph of this function, shown
20
lowing problem. in Figure 3.14, we see that the object is 25 ft beneath the surface of the pond after
15
10 approximately 4 sec.
A more accurate approximation of the time at which the object is 25 ft be-
"f·,,·,'.'
5
t (in seconds) neath the pond's surface is t = 3.93802.
3 4 Finally, the time required for the object to reach the pond's surface is added
3.93802
A 32-lb object is dropped from a height of 50 ft above the surface of a small pond. to the time needed for it to travel 25 ft beneath the surface to see that approximately
While the object is in the air, the force due to air resistance is v. However, when the Figure 3.14 Graph of s(t) = 6.41667 sec are required for the object to travel from a height of 50 ft above the
object is in the pond it is subjected to a buoyancy force equivalent to 6v. Detennine 3.99722 - 3.99722e- 6t
+ lft pond to a depth of 25 ft below the surface.
how much time is required for the object to reach a depth of 25 ft in the pond.
Note The model presented in this section in which the force due to air resistance
: I Solution The mass of this object is found using the relationship W = mg, where
W is the weight of the object. With this, we find that 32 Ib = m(32 ftlsec 2 ), so
is proportional to the velocity m ~~ = mg - cv is a simplified model used to illus-
m = 1 (lb-sec 2 )/ft (slug). trate how differential equations are used to model physical situations. In most cases,
the equation in which the force due to air resistance is assumed to be proportional any time I. (b) What is the maximum height attained (b) What is the parachutist's velocity vet) after the
parachute is opened? (c) What is the parachutist's lim-
to the square of the velocity m ~~ = mg - cV' does a much better job in predicting
by the projectile?
iting velocity? How does this compare to the limiting
12. In a camival game, a participant uses a mallet to pro-
the velocity of the falling object. In fact, the first model typically works only when- ject an object up a 20-m pole. If the mass of the ob- velocity if the parachute does not open?
an object is dropped in a highly viscous medium or when the object has negligible ject is I kg and the object is subjected to resistance 21. (Escape Velocity) Suppose that a rocket is launched
mass.'" equivalent to 1/10 times the instantaneous velocity, de- from the Earth's surface. At a great (radial) distance r
termine if an initial velocity of 20 mlsec causes the from the center of the Earth, the rocket's acceleration
object to reach the top of the pole. is not the constant g. Instead, according to Newton's
13. Assuming that air resistance is ignored, find the ve- law of gravitation, a = Idr 2 , where k is the constant of
locity and height functions if an object with mass m proportionality (k > 0 if the rocket is falling toward
EXERCISES 3.3 is thrown vertically up into the air with an initial ve- the Earth; k < 0 if the rocket is moving away from the
locity Vo from an initial height So. Earth). (a) If a = -g at the Earth's surface (when
r = R), find k and show that the rocket's velocity
14. Use the results of Exercise 13 to find the velocity and
is found by solving the initial-value problem dvldt =
height functions if m = 11128, g = 32 ftlsec 2 , Vo =
_gR2Ir2, yeO) = Yo. (b) Show that dvldt = v dvldr so
1. A rock that weighs 32 lb is dropped from rest from is equivalent to half of the instantaneous velocity. 48 ftlsec, and So = 0 ft. What is the shape of the height
that the solution to the initial-value problem v dvldr =
the edge ofa cliff. (a) Find the velocity of the rock at (b) When does the ball reach its maximum height? function? What is the maximum height attained by the
_gR2Ir2. vCR) = Vo is v 2 = (2gR2Ir) + v~ - 2gR.
time I if the air resistance is equivalent to the instan- object? When does the object reach its maximum
*7. A rock of weight 0.5 Ib is dropped (with zero initial (c) Compute lim,~oo. Ifv > 0 (so that the rocket does
taneous velocity v. (b) What is the velocity ofthe rock height? When does the object hit the ground? How do
velocity) from a height of 300 ft. If the air resistance not fall to the ground), show that the minimum value
at I = 2 sec? these results compare to those of Example 3?
is equivalent to 1164 times the instantaneous velocity, of Vo for which this is true (even for very large values
2. An object that weighs 4 Ib is dropped from the top of find the velocity and distance traveled by the object *15. Consider the situation described in Exercise 13. What of r) is Vo = "I/2iR. This value is called the escape
a tall building. (a) Find the velocity of the object at at time t. Does the rock hit the ground before 4 sec is the velocity of the object when it hits the ground, velocity and signifies the minimum velocity required
time I if the air resistance is equivalent to the instan- elapse? assuming that the object is thrown from ground level? so that the rocket does not return to the Earth.
taneous velocity v. (b) What is the velocity of the ob- 16. Consider the situation described in Exercise 13. If the
8. An object of weight 0.5 lb is thrown downward with 22. If R = 3960 miles andg = 32 ftlsec 2 = 0.006 mi/sec2 ,
ject at I = 2 sec? How does this compare to the result an initial velocity of 16 ftlsec from a height of300 ft. object reaches its maximum height after T sec, when
use the results of Exercise 21 to compute the escape
in Exercise I? If the air resistance is equivalent to 1164 times the in- does the object hit the ground, assuming that the ob-
velocity of the Earth.
*3. An object weighing 1 Ib is thrown downward with an stantaneous velocity. find the velocity of and distance ject is thrown from ground level?
*23. (See Exercise 21) Determine the minimum initial ve-
initial velocity of 8 ftlsec. (a) Find the velocity of the traveled by the object at time t. Compare these results 17. Suppose that an object of mass 10 kg is thrown verti-
object at time I if the air resistance is equivalent to locity needed to launch the lunar module (used on
to those in Exercise 7. cally into the air with an initial velocity vomlsec. If
early space missions) from the surface of the Earth's
twice the instantaneous velocity. (b) What is the ve- 9. An object of mass 10 kg is dropped from a great the limiting velocity is -19.6 mlsec, what can be said
locity of the object at I = I sec? moon given that the moon's radius is R = 1080 miles
height. (a) If the object is subjected to air resistance about c in the force due to air resistance FR = cv act-
and the acceleration of gravity of the moon is 16.5%
4. An object weighing 16 Ib is dropped from a tall build- equivalent to 10 times the instantaneous velocity, find ing on the object?
of that of the Earth.
ing. (a) Find the velocity of the object at time I if the the velocity. (b) What is the limiting velocity of the 18. If the limiting velocity of an object of IDass m which
air resistance is equivalent to twice the instantaneous object? is thrown vertically into the air with an initial veloc- 24. (See Exercise 21) Compare the Earth's escape veloc-
velocity. (b) What is the velocity of the object at I = ity Vo mlsec is -9.8 mlsec, what can be said about the ity to those of Venus and Mars if for Venus R = 3800
10. Suppose that an object of mass I kg is thrown with a miles and acceleration of gravity is 85% of that of the
I sec? How does this compare to the result in Exer- downward initial velocity of 5 mlsec and is subjected relationship between m and c in the force due to air
cise 3? resistance FR = cv acting on the object? Earth; and for Mars R = 2100 and acceleration of
to an air resistance equivalent to the instantaneous ve- gravity is 38% of that of the Earth. Which planet has
5. A ball of weight 4 oz is tossed into the air with an ini- locity. (a) Find the velocity of the object and the dis- *19. A parachutist weighing 192 lb falls from a plane (that
the largest escape velocity? Which has the smallest?
tial velocity of 64 ftlsec. (a) Find the velocity of the tance fallen at time t. (b) How far does the object drop is, Vo = 0). When the parachutist's speed is 60 ftlsec,
object at time I if the air resistance is equivalent to after 5 sec? the parachute is opened and the parachutist is then sub- 25. In an electric circuit with one loop that contains
1116 the instantaneous velocity. (b) When does the ball jected to air resistance equivalent to FR = 3v 2 • Find a resistor R, a capacitor C, and a vo1tage source
*11. A projectile of mass 100 kg is launched vertically from
reach its maximum height? the velocity vet) of the parachutist. What is the limit- E(t), the charge Q on the capacitor is found by solv-
ground level with an initial velocity of 100 mlsec.
6. A tennis ball weighing 8 oz is hit vertically into the (a) Ifthe projectile is subjected to air resistance equiv- ing velocity of the parachutist? ing the initial-value problem R ~~ + ~ Q = E(t),
air with an initial velocity of 128 ftlsec. (a) Find alent to 1/10 times the instantaneous velocity, deter· 20. A parachutist weighing 60 kg falls from a plane (that
Q(O) = Qo. Solve this problem to find Q(I) if
the velocity of the object at time I if the air resistance mine the velocity of and the height of the projectile at is. Vo = 0) and is subjected to air resistance equiva-
E(t) = Eo where Eo is constant. Find 1(1) = Q' (t)
lent to FR = lOy. After one minute, the parachute is
where l(t) is the current at any time t.
opened so that the parachutist is subj ected to air re-
'" Lyle N. Long and Howard Weiss, "The Velocity Dependence of Aerodynamic Drag: A Primer for Mathematicians," The American Mathematics sistance equivalent to FR = 100v. (a) What is thepara- 26. If in the R-C circuit described in Exercise 25 C =
Monthly. Volume 106, Number 2, pp. 127-135. chutis!'s velocity when the parachute is opened? 10- 6 farads, R = 4000 ohms, and E(t) = 200 volts,
120 Chapter 3 Applications of First-Order Differential Equations Chapter 3 Review Exercises 121
find Q(t) and let) if Q(O) = O. What eventually hap- v =

8
-4V2(Ke rv'2 + 1) ? CHAPTER 3 SUMMARY
pens to the charge and the current as t ..... oo? 1 - K e 8 rv'2 . ~ ::;;:ug;:;,~.:. . ~ _~z::nE:3:i~~~,l.~-""':~~:::-~~-:::::'~~=:C:i;:.-':':::':'... ~. -"~~Gn:J:"*-=:::::r_"=:",,:,,.-,-~~-=,::.~,,,",,,,,,.~ ..•;. ..,~;:::::;::,,,,,:.:J
*27. An object that weighs 48 Ib is released from rest at ~ Concepts & formulas
the top of a plane metal slide that is inclined 30° to *31. Find the equilibriwn solution to dvldt = -g - (c/m)v._
the horizontal. Air resistance (lb) is nwnerically equal What is the limiting velocity?
32. Find the equilibrium solution to dvldt = -g_
~ection 3_1
to one-half the velocity (ftlsec), and the coefficient of
friction is j.L = 114. Using Newton's second law ofmo- (c/m)v2 What is the limiting velocity? 1 1Section 3_31
tion by summing the forces along the surface of the 33. Suppose that a falling body is subjected to air resist_
slide, we find the following forces: Malthus model Newton's second law of motion
ance asswned to be F R = cv. Use the values of
(a) the component of the weight parallel to the slide: The initial-value problem dyldt = ky, yeO) = Yo has the The rate at which the momentum of a body changes with
c = 0.5, I, and 2; and plot the velocity function with
Fl = 48 sin 30° = 24; solution respect to time is equal to the resultant force acting on the
m = I. g = 32, and va = O. How does the value of c
d
(b) the component of the weight perpendicular to the affect the velocity? y = yoe"'. body: d/Cmv) = F.
slide: N = 48 cos 30° = 24V3; 34. Compare the effects that air resistance has on the ve- Logistic equation (or Verhulst equation) The velocity of the falling body is found by solving the dif-
(c) the frictional force (against the motion of the locity of a falling object of mass m = 0.5 that is re- ferential equation determined with
The initial-value problem y'(t) = (r - ay(t»y(t), yeO) = Yo
object): F2 - j.LN = -l(24V3) = -6V3; and leased with an initial velocity of va = 16. Consider
bas the solution
(d) the force due to air resistance (against the motion FR = 16v 2 andFR = 16v. m dv =
dt
I (forces acting on the object).
of the object): F, = -lv. *35. Compare the effects that air resistance has on the ve-
Because the mass of the object is m = 48/32 = locity of a falling object of mass m = 0.5 that is re- y = ayo + (r - aYo)e
312, we tmd that the velocity of the object leased with an initial velocity of va = O. Consider
satisfies the initial-value problem m dvldt = FR = 16v' andFR = 16Vv.
Fl + F2 + F, or [section 3_21
36. Consider the velocity and height functions found in
1.2 dv = 24 - 6V3 - 1..2 v' yeO) = 0. Example 3 with m = 11128, c = 11160, and g =
~
32 ftlsec 2 • Newton's law of coo1ing
Solve this problem for vet). Determine the distance (a) Suppose that on the first toss, the object is thrown The initial-value problem dTldt = k(T - T,), reO) = To has
traveled by the object at time t, x(t), if x(O) = O. the solution
with va = 48 ftlsec from an initial height of
28. A boat weighing 150 Ib with a single rider weighing So = 0 ft and on the second toss with va = T = (To - T,)e/a + T,.
170 Ib is being towed in a particular direction at a rate 36 ftlsec and So = 6. On which toss does the
of 20 mph. At t = 0, the tow rope is cut and the rider object reach the greater maximwn height?
begins to row in the same direction, exerting a con-
(b) If So = 0 ft, compare the effect that the initial
stant force of 12 Ib in the direction that the boat is
velocities va = 48 ftlsec, va = 64 ftlsec, and va =
moving. The resistance is equivalent to twice the in-
80 ftlsec have on the height function.
stantaneous velocity (ftlsec). The forces acting on CHAPTER 3 REVIEW EXERCISES
the boat are F 1 = 12 in the direction of motion and (c) If va = 48 ftlsec, compare the effect that the ini-
the force due to resistance in the opposite direction, tial heights So = 0 ft, So = 10ft, and So = 20 ft
F2 = - 2v. Because the total weight (boat and rider) have on the height function.
is 320 Ib, m = 320/32 = 10. Therefore, the velocity 37. A woman weighing 125 Ib falls from an airplane at
In Exercises 1-4, classifY the equilibrium solutions. 6. Suppose that a culture contains 200 cells. After one
satisfies the differential equation m dvldt = F, + F2 an altitude of 4000 ft and opens her parachute after
or 10 dvldt = 12 - 2v with initial velocity 20 mph, day the culture contains 600 cells. How many cells
5 sec. If the force due to air resistance is FR = v be- 1. y' =y(I-2y)
does the culture contain after two days?
' ·IS eqmvaent
w hlch . I to v(0) = 2 OhfX
mi 5280 ft
~ x fore she opens her parachute and F R = 10v afterward, 2. y' = -y
how long does it take for the woman to reach the *7. What percentage of the original amount of the element
I hr 88 . ground? I 226Ra remains after 50 years?
3600 sec =""3 ftlsec. Fmd vet) and the distance *3. y' = -"4Y (Y - 4)
38. Consider the problem discussed in Example 4. Instead 8. If an artifact. contains 10% of the amount of 14C as
traveled by the boat, x(t), if x(O) = O.
~)(1 - *)
of a buoyancy force equivalent to 6v, suppose that that of a present-day sample, how old is the artifact?
29. What is the equilibriwn solution of dvldt = 32 - v? 4. y' = -y(l -
when the object is in the pond, it is subjected to a 9. Suppose that in an isolated population of 1000 peo-
How does this relate to the solution v = 32 + Ce-I? buoyancy force equivalent to 6v 2 Determine how 5_ The initial population in a bacteria culture is Yo. Sup- ple, 250 initially have a virus. If after one day, 500
30. What is the equilibriwn solution of dvldt = 32 - v 2 ? much time is required for the object to reach a depth pose that after four days the popUlation is 3yo. When have the virus, how many days are required for three-
How does this relate to the solution of 25 ft in the pond. is the population 5yo? fourths of the population to acquire the virus?
dx
10. The Gompertz equation given by dyldt = the velocity of the object at any time t. When does the *(b) ~~ - 2 sec otan 0 = 0, 1'(0) = 4 dt = (r - ax(t) )x(t) - qEx(t).
y(r - a In y) is used by actuaries to predict certain object reach its maximum height? What is its maxi_
populations. If yeO) = Yo, then find yet). Find mum height if it is thrown from an initial height of This equation can be rewritten in the form
limH~y(t) if a> O. 5 ft? (c) dO
dr - 6 sin! cos! = 0 reO) = 0
2 2 '
*11. A bottle that contains water with a temperature of 40°F *19. A parachutist weighing 128 lb falls from a plane 24. A cylindrical tank 1.50 m high stands on its circular
«r - qE) - ax(t))x(t).
'::; =
is placed on a tennis court with temperature 90°F. Af- (vo = 0). When the parachutist's speed is 30 filsec, the base of radius r = 0.50 m and is initially filled with (b) Solve the equation dxldt = «r - qE) - ax(t))x(t)
ter 20 min, the water is 65°F. What is the water's tem- parachute is opened and the parachutist is then sub. water. At the bottom of the tank, there is a hole of ra· and find the solution that satisfies the initial con-
perature after 30 min? jected to air resistance equivalent to FR = 2v'. Find dius r = 0.50 cm that is opened at some instant so that dition x(O) = Xo. What is h(t)?
the velocity vet) of the parachutist. What is the limit. draining starts due to gravity. According to Torricelli's
12. A can of diet cola at room temperature of 70°F is
ing velocity of the parachutist? law, v = 0.600v2gh, where g = 980 cmlsec' and h Suppose that Xo = I, r = I, and a = 112.
placed in a cooler with temperature 40°F. After 30 min
20. A relief package weighing 256 lb is dropped from a is the height of the water. By determining the rate
the can is at 60°F. When is the can at 45°F? (c) Graph x(t) if there is no harvesting for 0 :s t :s
plane (vo = 0) over a war-ravaged area and is sub· at which the volume changes. we fmd that
10. Hint: If there is no harvesting, qE = O.
13. A frozen turkey breast is placed in a microwave oven jected to air resistance equivalent to FR = 16v. After dh 0.600~V2g Vh, where A is the cross- (d) Graph x(t) and h(t) for 0 :s t :s 20 using qE = 0,
to defrost. It is then placed in a conventional oven at 2 sec the parachute opens and the package is then sub- dt
325'F and bakes for 1 hour, at which time its tem- sectional area of the outlet and B is the cross-sectional O. I, 0.2, ... , 2. What is the maximum sustain-
jected to air resistance equivalent to FR = 4v'. Find able harvest rate? In other words, what is the high-
perature is 100°F. If after baking an additional 45 min area of the tank. In this case, A = 0.500'.". cm' and
the velocity vet) of the package. What is the limiting est rate at wruch the fish can be harvested with·
its temperature is 150°F, what was the temperature of B = 50.0'.".cm', so dhldt = -0.00266 Vh. Find h(t)
velocity of the package? Compare this to the limiting out becoming extinct? At what rate should the fish
the turkey when it was removed from the microwave? velocity if the parachute does not open. if h(O) = 150.
be harvested to produce the largest overall har-
14. Suppose that during the month of July in Statesboro, 21. Atomic waste is placed in sealed canisters and dumped 25. (Fishing) Consider a popUlation of fish with size at vest? How does this result compare to (a)?
Georgia, the outside temperature in OF is given by C( t) in the ocean. It has been determined that the seal will time t given by x(t). Suppose the fish are harvested at
= 85 - 10 cos (7Tl/12). O:s t:S 24. Find the temper· not break and leak the waste when the canister hits the a rate of h(t). If the fish are sold at price p and 8 is In 1965, the values of 1', a,p, c, and q for the Antarc-
alure in a parked car that has an initial temperature of bottom of the ocean as long as the velocity of the can· the interest rate, the present value P of the harvest is tic whaling industry, were determined to be r = 0.05,
70°F if k = 1/4. (Assume that the car has no heat or ister is less than 12 mlsec when it hits the bottom. Us· given by the improper integral. a = 1.25 X 10- 7 , P = 7000, c = 5000, and q = J.3
air conditioning system.) ing Newton's second law, the velocity satisfies the X 10- 5 • Assume that t = 0 corresponds to the year
*15. A rock weighing 4 Ib is dropped from rest from a large equation m(dvldt) = W - B - kv, where v(O) = 0, W
p= J(~ e- [p - _c_]h(t)
8t
qx(t)
dt,* 1965 and that x(O) = 78,000. Assume that a typical
o
is the weight of the canister, B is the buoyancy force, firm expects a return of 10% on their investment, so
height and is subjected to air resistance equivalent to
and the drag is given by - kv. Solve this first-order where c and q are constants related to the cost of the that 8 = 0.10.
F R = v. Find the velocity v(t) of the rock at any time
linear equation for vet) and then integrate to find the effort of catching fish (q is called the catchability).
(e) Approximate
r
t. What is the velocity of the rock after 3 sec? How
far has the rock fallen after 3 sec? position yet). If W = 2254 Newtons, B = 2090 New· (a) Evaluate P if 8 = 0.05, x(t) = I, and h(t) = ~.
16. A container of waste weighing 6 Ib is accidentally reo
tons, and k = 0.637 kg/sec, determine the time at
wruch the velocity is 12 mls. Determine the depth H
Ifwe assume that the harvesting rate h(t) is
proportional to the population of the fish, then
P= e-St[p - q:Ct) }(t) dt
leased from an airplane at an altitude of 1000 ft with of the ocean so that the seal will not break when the if (i) E = 5000 and (ii) E = 7000. What happens
an initial velocity of 6 filsec. If the container is sub· canister hits the bottom. h(l) = qEx(t),
to the whale population in each case? What ad-
jected to air resistance equivalent to F R = 2v/3, find where E represents the effort in catcrung the fish,
22. According to the law of mass action, if the tempera· vice would you give to the whaling industry?
the velocity vet) of the container at any time t. What and, under certain assumptions, the size of the
ture is constant, then the velocity of a chemical reac· (f) Approximate
r e-&~
is the velocity of the container after 5 sec? How far population of the fish x(t) satisfies the differen-
tion is proportional to the product of the concentra·
has the container fallen after 5 sec? Approximately
when does the container hit the ground? tions of the substances that are reacting. The reaction tial equation P = - q:Ct) }(t) dt
A + B .... M combines a moles per liter of substance
17. An object of mass 5 kg is thrown vertically in the air A and b moles per liter of substance B. If y(t) is the
dx = (I' - ax(t) )x(t) - h(l) using the values in the following table.
dt
from ground level with an initial velocity of 40 mlsec. number of moles per liter that have reacted after time
If the object is subjected to air resistance equivalent t, the reaction rate is dyldt = k(a - y)(b - y). Find
to FR = 5v, find the velocity of the object at any time y(t) if yeO) = O. Find limH~ y(l) if a> b and if
t. When does the object reach its maximum height? b > a.
'" 1. N. Kapur, "Some Problems in Biomathematics," International Journal of Mathematical Education in Science and Technology, Volume 9,
What is its maximum height?
23. Solve the following equations for 1'( 0). Graph the po' Number 3 (August 1978), pp. 287-306; and Colin W. Clark, "Bioeconomic Modeling and Resource Management," Applied Machemacical Ecol-
18. A ball weighing 0.75 Ib is thrown vertically in the air lar equation that results. ogy, edited by Simon A. Levin, Thomas G. Hallam, and Louis J. Goss, Springer-Verlag, New York (1980), pp. 11-:7. F~r m~re ~fo~atl.on, see
with an initial velocity of 20 filsec. If the ball is sub- dr Robert M. May, John R. Beddington, Colin W. Clark, Sidney 1. Holt, and Richard M. Laws, "Management of MultlspeCles Fishenes, SClence,
(a) I' dO + 4 sin 20 = O. 1'(0) = 2 Volume 205, Ntunber 4403 (July 20, 1979), pp. 267-277.
jected to air resistance equivalent to FR = vl64, find
124 Chapter 3 Applications of First-Order Differential Equations Differential Equations at Work 125
trajectory is dy/d:!. = -1/fCx, y). Therefore, the fam- m2, respectively; and let 0 denote the angle between Hint: Studying Figure 3.16 might help.
E p= r e-B'[p - q~t) ]h(t)

Approximation of
dt
ily of orthogonal trajectories is fouod by solving
dy/d:!. = - I/fCx, y). Find the family of orthogonal tra-
jectories of the set of curves y = ex 2 by carrying out -
them as shown in Figure 3.15. Then,
tan () = m2 - ml
1 + m2m,'
y
1000 the following steps. Ca) Show that the slope of the
tangent line at any point on the parabola y = ex 2 is
1500 dy/d:!. = 2ex. Cb) Solve y = ex 2 for c and substitute
this result into dy/d:!. = 2ex to obtain dy/d:!. = 2ylx.
2000
Cc) Solve the equation dy/d:!. =
-l/f(x, y) = -x/(2y)
2500 to find that the family of orthogonal trajectories is e;
2
x /2 + y2 = k, a family of ellipses.
3000
28. Let T(x, y) represent the temperature at the point
x
3500 (X, y). The curves given by T(x, y) = c, where c is a
constant, are called isotherms. The orthogonal trajec. Figure 3.16
4000 tories are curves along which heat flows. Show that
the isotherms are Cx 2/2) + xy - (y2/2) = c if the
4500 (c) Find a family of curves that intersects the family
curves of heat flow are given by y 2 + 2xy - x 2 = k.
of curves x 2 - y2 = C at an angle of -rr/4. Graph
Hint: On the heat flow curves, dy/d:!. = (x - y)/(x + y), Figure 3.15
several members of both families to confirm your
What value of E produces the maximum profit? so we must solve dy/d:!. = Cx + y)/(x - y), which is an
result.
What happens to the whale population in this exact differential equation. m, + tan 0
case? (al Show that m2 = 1 - m, tan O' (d) Find a family of Curves that intersects the family
In Exercises 29-32, determine the orthogonal trajectories of the of curves x 2 + y2 = c' at an angle of -rr/6. Graph
(g) Some reports have indicated that the optimal stock
(b l Suppose we are given a family of curves that sat- several members of both families to confirm your
level of whales should be about 227,500. How given family of curves. (Graph the orthogonal trajectories and
curves simultaneously.) isfies the differential equation dy/d:!. = f(x, y). result.
does this number compare to the maximum num-
Use (al to show that if we want to find a family
ber of whales that the environment can sustain? 38. (a) Determine the orthogonal trajectories of the fam-
29. Y + 2x = c of curves that intersects this family at a constant
ily of curves y2 = 2cx + 2c2 (h) Graph several mem-
Hint: Evaluate lim,~~ x(t) if there is no harvest-
30. y = e= angle 0, we must solve the differential equation
ing. How can the whaling industry make a profit bers of both families of curves on the same set of axes.
and maintain this number of whales in the ocean? *31. y2 =x 2 + c dy f(x, y) ~ tan 0 (c) What is your reaction to the graphs?
26. CFamily of Orthogonal Trajectories) Two curves C, 32. y2 = x 2 + ex d:!. 1 + fCx,y) tan o·
and C2 are orthogonal at a point of intersection if their 33. A family of curves is self-orthogonal if the family of
respective tangent lines to the curves at that point are orthogonal trajectories is the Same as the original fam·
orthogonal. For example, consider the curves y = x ily of curves. Is y2 - 2ex = c 2 a self-orthogonal fam-
and y = v'1="? (a) Show that the derivatives of ily of curves (parabolas)?
these functions are y' = 1 and y' = -x/v'1="?, re-
spectively. (b) Show that the curves are orthogonal at
34. Find a value of c so that the two families of curves ,:~ Differential Equations at Work:
y = k,x 2 + c andx2 + 2y2 = k2 + y, where k, andk2
the point CV2/2, V2/2) because the derivatives eval-
uated at this point are 1 and - 1, respectively. There-
are constants, are orthogonal. A. Mathematics of Finance
fore, the tangent lines are perpendicular at (V2/2, 35. Suppose that an electrical current is flowing in a wire
V212). along the z-axis. Then, the equipotential lines in the Suppose that P dollars are invested in an account at an annual rate of 1""10 compounded
xy-plane are concentric circles centered at the origin. continuously. To find the balance of the account x(t) at time t, we must solve the
27. Given a family of curves F, we would like to find a
If the electric lines of force are the orthogonal trajec· initial-value problem
set of curves that are orthogonal to each curve in F.
tories of these circles, find the electric lines of force.
We refer to this set of orthogonal curves as the fam- dx
ily of orthogonal trajectories. Suppose that a family 36. The path along which a fluid particle flows is called -=rx
a streamline, and the orthogonal trajectories are called dt
of curves is defined as FCx, y) = c and that the slope {
equipotential lines. If the streamlines are y =
k/x, x(O) = P
of the tangent line at any point on these curves is
dy/d:!. = fCx, y) when it is obtained by differentiating find the equipotential lines.
for x.
F(x, y) = c with respect to x and solving for dy/d:!.. e, e
*37. (Obliqne Trajectories) Let and 2 denote two lines,
Then the slope of the tangent line of the orthogonal not perpendicular to each other, with slopes and m, 1. Show that x(t) = Pe".
I
I
I
Differential Equations at Work 127
2. If $1000 is deposited into an account with an annual interest rate of 8% com-

pounded continuously, what is the balance of the account at the end of 5, 10, 15,
and 20 years?
If we allow additions or subtractions of sums of money from the account, the
problem becomes more complicated. Suppose that an account like a savings account,
8. Suppose an investor begins by investing $250 per month in an account paying an
annual interest rate of 10%, compounded monthly, and in addition increases the
amount invested by 6% each year for an increase of Wo each month. How much
will the investor have accumulated at the end of 10, 20, and 30 years?
9. Suppose that a 25-year-old investor begins investing $250 per month in an account
paying an annual interest rate of 10%, compounded monthly. If at the age of 35
I.
home mortgage loan, student loan or car loan, has an initial balance of P dollars and
r denotes the interest rate per compounding period. Let pet) denote the money flow per
unit time and !5 = In(l + r). Then the balance of the account at time t, xU), satisfies
the investor stops making monthly payments, what will be the account balance
when the investor reaches 45, 55, and 65 years of age? Suppose that a 35-year-old
friend begins investing $250 per month in an account paying an annual interest I
II
the initial-value problem rate of 10%, compounded monthly, at the same time the first investor stops. Who
has a larger investment at the age of 65?
dx _ Ox = pU) 10. If you are given a choice between saving $150 a month beginning when you first
{
dt * start working and continuing until you retire, or saving $300 per month beginning
x(O) = P . 10 years after you first start working and continuing until you retire, which should
3. Show that the balance of the account at time to is given by
you do to help ensure a financially secure retirement?
From Exercises 8 and 9, we see that consistent savings beginning at an early age
I
4. Suppose that the initial balance of a student loan is $12,000 and that monthly pay-
ments are made in the amount of $130. If the annual interest rate, compounded
monthly, is 9%, then P = 12,000, pet) = -130, r = 0.09/12 = 0.0075, and !5 =
can help assure that a large sum of money will accumulate by the age of retirement.
How much money does a person need to have accumulated to help ensure a financially
secure retirement? For example, corporate pension plans and social security generally
provide a relatively small portion of living expenses, especially for those with above-
average incomes. In addition, inflation erodes the buying power of the dollar; large
I
II
In(l + r) = In(1.0075).
(a) Show that the balance of the loan at time t, in months, is given by
x(t) = 17398.3 - 5398.25 . 1.0075'.
(b) Graph xU) on the interval [0, 180], corresponding to the loan balance for the
sums of money today will not necessarily be large sums of money tomorrow.
As an illustration, we see that if inflation were to average a modest 3% per year
for the next 20 years and a person has living expenses of $20,000 annually, then after
20 years the person would have living expenses of
$20,000' 1.03 20 = $36,122.
IE
first 15 years. (e) How long will it take to pay off the loan?
5. Suppose that the initial balance of a home mortgage loan is $80,000 and that Let t denote years and suppose that a person's after-tax income as a function of years I
monthly payments are made in the amount of $599. If the annual interest rate, com- is given by I(t) and E(t) represents living expenses. Here t = 0 might represent the year
pounded monthly, is 8%, how long will it take to pay off the mortgage loan? a person enters the work force. Generally, during working years, I(t) > EU); during re-
6. Suppose that the initial balance of a home mortgage loan is $80,000 and that tirement years, when I(t) represents income from sources such as corporate pension
monthly payments are made in the amount of $599. If the annual interest rate, com- plans and social security, IU) < EU)·
pounded monthly, is 8%, how long will it take to pay off the mortgage loan if the Suppose that an account has an initial balance of So and the after-tax return on
monthly payment is increased at an annual rate of 3%, which corresponds to a the account is r%. We assume that the amount deposited into the account each year is
monthly increase of ~%? I(t) - E(t). What is the balance of the account at year t, Set)? S must satisfy the initial-
7. If an investor invests $250 per month in an account paying an annual interest rate value problem
of 10%, compounded monthly, how much will the investor have accumulated at
the end of 10, 20, and 30 years? ~~ = rS(t) + I(t) - EU).
{
S(O) = So
II. Show that the balance of the account at time t = to is

'" Thoddi C. T. Kotiah, "Difference and Differential Equations in the Mathematics of Finance," Interna-
tional Journal of Mathematics in Education, Science, and Technology, Volume 22, Number 5 (1991),
pp.783-789.
SUo) = erro{so + f;o [IU) - EU)]e- rr dt}.
I
Edward W Herold ·'Inflation Mathematics for the Professional," Mathematical Modeling: Classroom
No'es in Applied Mathematics. Edited by Murray S. Klamkin, SIAM (1987), pp. 206-209. Assume that inflation averages an annual rate of i. Then, in terms of E(O), Im
~
R
t!i
~
<:1
128 Chapter 3 Applications of First-Order Differential Equations Differential Equations at Work 129
E(t) = E(O)e it • provides a continuous supply of nutrients. The organisms, however, must spread over
the surface without being washed away. If conditions become unfavorable, they must
Similarly, during working years we assume that annual raises are received at an be able to free themselves from the surface and recolonize on a new surface.
annual rate of}. Then, in terms of 1(0), The rate at which cells accumulate on a surface is proportional to the rate of
let) = l(o)ef r. growth of the cells and the rate at which the cells attach to the surface. An equation
describing this situation is given by
However, during retirement years, we assume that let) is given by a fixed sum F,
such as a corporate pension or annuity, and a portion indexed to the inflation rate dN
V, such as social security. Thus, during retirement years
dt = /-L(N+ A),
let) = F + Vei(r-T), where N represents the cell density, /-L the growth rate, A the attachment rate, and t
time. *
where T denotes the number of working years. Therefore,
I. If the attachment rate A is constant, solve the initial-value problem
f,( ) _ {1(0).!1, 0 s; t s; T .
t - F+ Vei(l-T), t> T dN
- = /-L(N+A)
dt
12. Suppose that a person has an initial income of 1(0) = 20,000 and receives annual { N(O) = 0
average raises of5%, so that} = 0.05 and initial living expenses areE(O) = 18,000.
Further, we assume that inflation averages 3%, so that i = 0.03, while the after- for N and then solve the result for /-L.
tax return on the investment is 6%, so that r = 0.06. Upon retirement after 2. In a colony of cells it was observed that A = 3. The number of cells N at the end
T years of work, we assume that the person receives a fixed pension equal to 20% of t hours is shown in the following table. Estimate the growth rate at the end of
of his living expenses at that time, so that
each hour.
F = 0.2 . 18,000 . e°.Q3T,
t N J1-
while social security provides 30% of his living expenses at that time, so that
1 3
V = 0.3 . 18,000 . e°.Q3T.
2 9
(a) Find the smallest value of T so that the balance in the account is zero after
30 years of retirement. Sketch a graph of S for this value of T. 3 21
(b) Find the smallest value of T so that the balance in the account is never zero.
4 45
Sketch a graph of S for this value of T.
13. What is the relationship between the results you obtained in Exercises 9 and 10
and that obtained in Exercise 12? 3. Using the growth rate obtained in Problem 2, estimate the number of cells at the
end of 24 hours and 36 hours.
14. (a) How would you advise a person 22 years of age first entering the work force
to prepare for a financially secure retirement?
(b) How would you advise a person 50 years of age with no savings who hopes
to retire at 65 years of age? C. Dialysis
(c) When should you start saving for retirement?
The primary purpose of the kidney is to remove waste products like urea, creatinine,
and excess fluid from blood. When the kidneys are not working properly, wastes ac-
B. Algae Growth cumulate in the blood; when toxic levels are reached, death is certain. The leading
causes of chronic kidney failure in the United States are hypertension (high blood
When wading in a river or stream, you may notice that microorganisms like algae are pressure) and diabetes mellitus. In fact, one-fourth of all patients requiring kidney
frequently found on rocks. Similarly, if you have a swimming pool, you may notice
that without maintaining appropriate levels of chlorine and algaecides, small patches ... Douglas E. Caldwell, "Microbial Colonization of Solid-Liquid Interfaces," Biochem.ical Engineering V,
of algae take over the pool surface, sometimes overnight. Underwater surfaces are at- Annals of the New York Academy of Sciences, Volume 56, New York Academy of SCiences (1987), pp.
tractive environments for microorganisms because water movement removes wastes and 274-280.
130 Chapter 3 Applications of First-Order Differential Equations 131
Differential Equations at Work
where k is the proportionality contant. *

We let L denote the length of the dialyser, and the initial concentration of wastes
in the blood is u(O) = uo, while the initial concentration of wastes in the dialysate is
veL) = O. Then, we must solve the initial-value problem
Qsu' = -k(u - v)
-QDV' = k(u - v) .
{
u(O) = uo, veL) = 0
I. Show that the solution to

Qsu' = -k(u - v)
-QDV' = k(u - v)
{
u(O) = uo, veL) = 0
is
aL
Qse"" - QDe
u(x) = Uo e""(Qs _ QDeaL)
eaL _ ecxx
vex) = Uo QD '
e""-(e aL -I)
Qs
k k . {QSU' = -/.:(u - v) .
where (l' = - - - . First add the equatlOns Q ' _ k( _ ) to obtam the
Qs QD - DV - U v
linear equation - v) = -~ (u - v) + ~ (u - v), then let z = u - v.
.!L (u
dx Qs QD
Compressed air Next solve for z and subsequently solve for u and v.
dialyzing temperature.
solution bath In healthy adults, typical urea nitrogen levels are 11-23 mg/dL (1 dL = 100
mL), while serum creatinine levels range from 0.6-1.2 mg/dL, and the total volume
Diagram of a kidney dialysis machine of blood is 4 to 5 L (1 L = 1000 mL).
2. Suppose that hemodialysis is performed on a patient with urea nitrogen level of 34
mg/dL and serum creatinine level of 1.8 using a dialyser with k = 2.2~ and L = I.
dialysis have diabetes. Fortunately, kidney dialysis removes waste products from the If the flow rate of blood, Qs, is 2 dL/min while the flow rate of the dialysate, QD,
blood of patients with improperly working kidneys. During the hemodialysis process, is 4 dL/min, will the level of wastes in the patient's blood reach normal levels after
the patient's blood is pumped through a dialyser, usually at a rate of 1-3 deciliters/min. dialysis is performed? For what waste levels would dialysis have to be performed
The patient's blood is separated from the "cleaning fluid" by a semipermeable mem- twice?
brane, which permits wastes (but not blood cells) to diffuse to the cleaning fluid. The 3. The amount of waste removed is given by
o
L
i
k[u(x) - vex)] dx. Show that
cleaning fluid contains some substances beneficial to the body which diffuse to the
blood. The cleaning fluid, called the dialysate, is flowing in the opposite direction as
the blood, usually at a rate of 2-6 dL/min. Waste products from the blood diffuse to
to
k[u(x) - vex)] dx = Qs[uo - u(L)].
the dialysate through the membrane at a rate proportional to the difference in concen-
4. The clearance of a dialyser, CL, is given by CL = ::[uo - u(L)]. Use the solu-
tration of the waste products in the blood and dialysate. If we let u(x) represent the
concentration of wastes in blood and vex) represent the concentration of wastes in the tion obtained in (a) to show that
dialysate, where x is the distance along the dialyser, QD is the flow rate of the dialysate
through the machine, and Qs is the flow rate of the blood through the machine, then '" D. N. Burghess and M. S. Borrie, Modeling with Differential Equations. Ellis Horwood Limited,
pp. 41-45. Joyce M. Black and Esther Matassarm-Jacobs, Luckman and Sorensen's Medical-Surgical
Qsu' = -k(u - v) Nursing: A Psychophysi%gicApproach. Fourth Edition, W B. Saunders Company (1993). pp. 1509-1519,
{ -QDV' = k(u - v) ,
1775-1808.
132 Chapter 3 Applications of First-Order Differential Equations 133
1 - e- aL 2. Substitute these values into the result obtained in Problem I. (a) Graph X(!) on the
CL=QB-:"'--~--
1 - QB e- aL ' interval [0, 24]. (b) Find the mass concentration at the end of 4, 8, 12, 16,20, and
QD 24 hours.
The rate of accumulation of cycloheximide is the difference between the rate
Typically, hemodialysis is perfonned 3-4 hours at a time three or four times
of synthesis and the rate of degradation:
per week. In some cases, a kidney transplant can free patients from the restrictions
of dialysis. Of course, transplants have other risks not necessarily faced by those on dP
dialysis; the number of available kidneys also affects the number of transplants per-
dt = Rs - Rd,
fonned. For example, in 1991 over 130,000 patients were on dialysis while only It is known that Rd = KdP, where kd = 5 X 10- 3 h- I , so dPldt = Rs - Rd is equiv-
I
7000 kidney transplants had been perfonned.
alent to dPldl = Rs - KaP. Furthennore,
)-1 ,
o D. Antibiotic Production
P
Rs = QpaEX ( 1 + K[
When you are injured or sick, your doctor may prescribe antibiotics to prevent or cure
infections. In the journal article "Changes in the Protein Profile of Streptomyces griseus
where Q represents the specific enzyme activity with value Qpo = 0.6 g CH/g pro-
tein . hr ~d K[ represents the inhibition constant. E represents the intracellular con-
centration of an enzyme, which we will assume is constant. For large values of K[
I
I
during a Cycloheximide Fennentation," we see that production of the antibiotic cyclo-
and I, X(I) = 10 and (I + (PIK[))-I = I. Thus, Rs = IOQpaE, so
heximide by Streptomyces is typical of antibiotic production. During the production of
cycloheximide, the mass of Streptomyces grows relatively quickly and produces little dP
cycloheximide. After approximately 24 hours, the mass of Streptomyces remains rela-
dt = 10QpaE - KdP.
tively constant and cycloheximide accumulates. However, once the level of cyclohex- 3. Solve the initial-value problem
imide reaches a certain level, extracellular cycloheximide is degraded (feedback in-
hibited). One approach to alleviating this problem and to maximize cycloheximide dP
dt = 10QpaE - KdP
production is to remove extracellular cycloheximide continuously. The rate of growth
of Streptomyces can be described by the equation { p(24) = 0
~= /J-max(1 - XX
max
)x, 4. Graph tp(l) on the interval [24, 1000].
5. What happens to the net accumulation of the antibiotic as time increases?
where X represents the mass concentration in gIL, /J-m"" is the maJdmum specific growth
6. If instead the antibiotic is removed from the solution so that no ?egradation occurs,
I
rate, and Xm "" represents the maximum mass concentration.'
what happens to the net accumulation of the antibiotic as time Increases?
I. Find the solution to the initial-value problem
~= X~ax) Xto a linear equation with
I
by first converting the equation /J-max( 1 -
the substitution y = X-I.
Experimental results have shown that /J-max = 0.3 hr- I and Xmax = 10 gIL.
>I< Kevin H. Dykstra and Henry Y. Wang. "Changes in the Protein Profile of Streptomyces griseus during a
I
CycloheXimide Fermentation," Biochemical Engineering V, Annals of the New York Academy of Sciences,
Volume 56, New York Academy of Sciences (1987), pp. 511-522. I
Ii
4.1 Second-Order Equations: An Introduction 135
4
~ Second-Order Equations: An Introduction
r: The Second-Order Linear Homogeneous Equation with Constant
Coefficients r: The General Case r: Reduction of Order
In this chapter, we focus our attention on solving linear ordinary differential equations
of order two or higher. There are many instances in which we encounter equations of
this type. For example, we see them in the study of vibrations and sound, and as we
mentioned in Chapter I, there is a relationship between a system of two first-order
ODEs and a second-order ODE.
Higher Order Let's consider the second-order equation y" + Y = 0 or ;/' = -Yo To find a solu-
tion, we must determine a function with the property that the second derivative of the
Equations function is the negative of the function itself. When we consider functions familiar to
us, such as polynomials, trigonometric functions, exponential functions, and natural
logarithms, we can conclude that YI(t) = sin I is a solution because
yj = cos I and y7 = -sin I;
Similarly, y2Ct) = cos I is a solution because

Y2 = -sin I andy~ = -cos I;
Of course, we are not always able to determine solutions of a differential equation by

inspection as we have in this case. Therefore, we now investigate a more general ap-
proach to solving second-order equations. We begin this discussion by considering a
problem studied in physics.
The Second-Order Linear Homogeneous Equation

with Constant Coefficients
Suppose that we attach an object to the end of a spring that is mounted to a horizon-
I n Chapters 2 and 3, we saw that first-order differential equations can be

used to model a variety of physical situations. However there are many
~hysical situations that need to be modeled by higher ~rder differential
equations.
tal rod (see Figure 4.1). The object comes to rest. We call this the equilibrium position.
Let x(t) represent the displacement of the object from equilibrium. If x(t) > 0, then the
object is below equilibrium; if X(I) < 0, it is above equilibrium. If we assume there is
a damping force acting on the object that impedes motion, then we can find x(t) by
For example, in 1735, Daniel Bernoulli's (1700-1782) study of the vibration solving a second-order linear ODE such as
of an elastic beam led to the fourth-order differential equation
d 2x dx
k4
4
d y dl 2 +3 at + 2x = O.
dx 4 = y,
This is a linear second-order ODE because it has the form
whic.h de~cribes the displacement of the simple modes. This equation can be
d 2x dx
rewritten In the form
a2(1) dl2 + al(l) at + ao(l)x = 1(1),
4 Figure 4.1 A spring-mass
4 d y
Y- k dx' = O. system and the highest order derivative in the equation is of order two. Of course in this equa-
tion, a2(t) = 1, a I (t) = 3, and ao(t) = 2, so the coefficient functions are constant. In
Both Bernoulli ~nd Euler realized that a solution to the equation is y = e x1k but addition,f(t) = 0 for all t, so this equation is homogeneous. (We give more details on
that other solutions to the equation must also exist. this spring-mass problem in Chapter 5.) For now, we attempt to find a solution of this
134
136 Chapter 4 Higher Order Equations 137
4.1 Second-Order Equations: An Introduction
~DE by assuming tha~ solutions have the form xCt) = e rt because err and its deriva_ After finding a general solution of the ODE, we apply the two initial conditions. In
tlV?S are cons~t ~ultlples of one another. We find the value(s) of r that lead to a so- this case, we have x(O) = Ct + Cz and dxldt (0) = -Ct - 2cz, so we solve the system
lution by subshtutmg x(t) = e into the ODE. Notice that dxldt = re rt and d 2xldt 2 '"
rt
CI + Cz = 0
r2e rt. Substitution into d 2xldt 2 + 3 (dxldt) + 2x = 0 then gives us r2e rt + 3re rt + 2en- { -C! - 2cz = -1
= 0 or
for C! and Cz. Adding these equations, we obtain -Cz = -I, so Cz = l. Substituting
e rt (r2 + 3r + 2) = O.
this value into C! + Cz = 0 then indicates that Ct = -l. Therefore, the unique solution
This algebraic equation tells us that either e rt = 0 or r2 + 3r + 2 = O. We know that to this IVP is
rt
there are no values of t so that e = O. Therefore, we solve the characteristic equation x(t) = e- zi - e- t,
r2 + 3r + 2 = 0
x which we graph in Figure 4.3. Notice that the curve passes through (0, 0) because of
V
23 5
to find the ~lues of r that ~atisfY the equation. In the case of a second-order equation, -0.05 the initial condition x(O) = O. The curve then moves in the downward direction because
the charactenshc equatIOn IS quadratic, so we can either factor or use the quadratic for- -0.1 dxldt (0) = -l. Notice also that limt->oo x(t) = 0 (as with any general solution shown
mul~, Factoring yie:ds (r + 1)(~ + 2) = 0, so r = -lor r = -2. This mean; that X(I) -0.15 in Figure 4.2). This means that the object eventually comes to rest.
Zt
= e . and. x(t) = e 21 each satIsfY the ODE. We verifY this by substituting these two -0.2 In the previous problem, a general solution is x(/) = Cte-t + cze- . Notice that
-0.25
functIOns mto the ODE. IfxCt) = e-', then dxldt =_e- r and d 2xldt 2 = e-'. Therefore, if we select Ct = Cz = 0, we obtain the trivial solution xCt) = O. This shows us that the
Y trivial solution always satisfies a linear homogeneous ODE. Therefore, when solving
d 2x dx
d(2 + 3 It + 2x = e- r + 3(-e- t ) + 2e- r = O. figure 4.3 Graph of solution equations of this type, we seek nontrivial solutions.
XCI) = e- 2t - e- r to the IVP
Because simplification of the left side of the ODE yields zero (the riaht side of The General Case
th:2;quation~, x(t) = e -t satisfie~ the ODE. In a similar manner, we verifY iliat xCt) '" Now that we have discussed a representative problem considered in this chapter, we
e also satIsfIes the ODE. NotIce that these two functions are not constant multiples can state more precisely several theorems and definitions that were mentioned earlier.
of one another, so w.e say that x(t) = e -t and x(t) = e -2t are linearly independent. No- Let's begin with the definition of linear dependence and linear independence.
nce also that for arbItrary constants Ct and C2, the function x(t) = Ct e -t + C2 e -z', called
the linear combination of X(/) = e -t and X(/) = e -Zt, is also a solution of the ODE. We
verifY this through substitution as well. If x(t) = c!e- t + C2e-zl, then dxldt = -cle- I
- 2cze- Zt and d Zxldt 2 = c!e- t + 4c2e-zr. Therefore, Definition 4.1 Linearly Dependent
dZx dx Let S = {f! Ct), f2(t)} be a set of functions. The set S is linearly dependent on an
d(2 + 3 It + 2x = (cle- t + 4c2e-zl) + 3( -Cte-t - 2cze- 2r ) + 2(cle- t + C2e-Zt) interval I if there are constants C! and Cz not both zero, so that
r
= cl(e- - 3e- r + 2e- t ) + cz(4e- Zt - 6e- zr + 2e- zl ) Ct/! (t) + czh(t) = 0
=0 for every value of t in the interval!. The set S is linearly independent if S is

y not linearly dependent (that is, if C! = C2 = 0).
This property, called the Principle of Superposition, tells that if we have two linearly
mdependent solutions of a linear homogeneous ODE, then any linear combination of
the ~o solutions is also a solution. We call xCt) = cle- t + cze- Zt a general solution
of thIS second-order ODE because all solutions are obtained from it. In Figure 4.2, we Note: We can rewrite ct/t(t) + c2h(t) = 0 ashCt) = -(Ct/C2)fi(t), so we say that the
3 3.5 x graph this solution for several different choices for CI and Cz. two functionsfiCt) andf2(t) are linearly dependent if they are constant multiples. They
. .. Ifwe are given additional information about the spring-mass system such as the are linearly independent otherwise.
mltIal dIsplacement and the initial velocity of the object, then we have a second-order
initial-value problem (IVP). For example, suppose that the object is released from equi-
lIbnum (x = 0) with initial velocity dx/dt (0) = -l. Therefore. we need to solve
Figure 4.2 Graph of general
solution for various values of dx Determine if the following sets of functions are linearly dependent or linearly in-
Cl and C2 x(O) = 0, It (0) = -l.
: dependent. (a) {2t, 4t}; (b) {e ', e- t }; (e) {sin 2/, 5 sin t cos t}; (d) {t, OJ.
138 Chapter 4 Higher Order Equations 4.1 Second·Order Equations: An Introduction 139
:. Solution (a) Notice that 4t = 2(2t), so the functions are linearly dependent. Another way to express Theorem 4.1 is to say that any linear combination of two
(b) These functions are not constant multiples of one another, so they are linearly or more solutions of y" + p(t)y' + q(t)y = 0 is also a solution of this ODE.
independent. (c) We use the identity sin 2t = 2 sin t cos t to show that ~ sin 2t '"
~(2 sin t cos t) = 5 sin t cos t, so the functions are linearly dependent. (d) This se(
contains the zero function. Because we can obtain the zero function by mUltiplying
the second function by 0, 0 . t = 0, the functions are linearly dependent. Therefore,
"i'u"'.'
if 0 is a member of a set of two functions, the set is linearly dependent. Show that Yt (t) = cos 2t and Y2(t) = sin 2t are solutions of the second-~rder linear
homogeneous ODE y" + 4y = O. Use the Principle of SuperpoSItIOn to fmd another
solution of this ODE.
Consider the second-order linear ODE
a2(t)y" + al(t)y' + ao(t)y = F(t), : I Solution For YI(t) = cos 2t, we have yi(t) = -2 sin 2t and y~(t) = -4 cos 2:.
. Then, y~ + 4YI = (-4 cos 2t) + 4(cos 2t) = 0, so YI satisfies the OD.E. In a SImI-
where the coefficient functions a2(t), at(t), ao(t), and F(t) are continuous on the open
lar manner we can show thatY2 is a solution of the ODE. (We leave thIS task to you
as an exer~ise.) Therefore, by the Principle of Superposition, any linear combina-
interval 1. Also, assume that a2(t) '" 0 for each value of t in I so that we may divide
the equation by a2(t) to write the equation in normal form given by
tion of these two functions, yet) = Cl cos 2t + C2 sin 2t, is also a solution of the
y" + p(t)y' + q(t)y = f(t), ODE. Notice that the two functions YI (t) = cos 2t and y,(t) = sin 2t are linearly
independent because they are not constant multiples .of each other. Then, because
where p(t) = at(t) la 2(t), q(t) = ao(t)lazCt), andf(t) = F(t)la2(t). Iff(t) == 0, that is, if
the ODE is second-order and we know two linearly mdependent solutIOns, a gen-
f(t) is identically the zero function, then the equation is homogeneous and is repre.
eral solution of the ODE is the linear combination of these two functions,
sented by
yet) = CI cos 2t + C2 sin 2t.
y" + p(t)y' + q(t)y = O.
If f(t) is not identically the zero function, the equation, y" + p(t)y' + q(t)y = f(t), is
nonhomogeneous. As with first-order initial value problems, we can state an existence and unique-
ness theorem for an initial-value problem involving a second-order linear ordinary dif·
ferential equation.
Theorem 4.1 Principle of Superposition
Iffl(t) andh(t) are solutions of the linear homogeneous ODE y" + p(t)y' + q(t)y
= 0 on the interval I, and if Ct and C2 are arbitrary constants, then y = cdt (t) + Theorem 4.2 Existence and Uniqueness
c2h(t) is also a solution of y" + p(t)y' + q(t)y = 0 on I.
Suppose that p(t), q(t), and f(t) are continuous functions on an open interval I
that contains t = to. Then, the initial-value problem
PROOF OF THEOREM 4.1
y" + p(t)y' + q(t)y = f(l), y(to) = a, y' (to) = b
Suppose that fi(t) and h(t) are solutions of the linear homogeneous ODE y" + has a unique solution on I.
p(t)y' + q(t)y = 0 on the interval I. Then, f~(t) + p(t)fi(t) + q(t)fi(t) = 0 and
f2(t) + p(t)f',(t) + q(t)h(t) = O. Let y = Cdl(t) + c2h(t). Then, y' = cdi(t) +
Co!2(t) and y" = cd~(t) + c2!'2(t). Substitution into the left side ofthe ODE yields
y" + p(t)y' + q(t)y =
=
(cdrct) + c2!'2(t)) + p(t)(cdi(t) + C2!2(t»)
+ q(t)(clfi(t) + c2h(t»)
CI(f~(t) + p(t)fi(t) + q(t)fi(t»)
"ii"',,,..
Find the solution to the IVP y" + 4y = 0, yeO) = 4, y' (0) = -4.
+ c2(fM + P(t)f2(t) + q(t)h(t»)
: 1 Solution We found a general solution of the ODE, yet) = CI cos 2t + C2 sin 21,
= CI(O) + C2(0) = 0 . in Example 2. Application of the initial condition yeO) = 4 gives us yeO) =
Therefore, y = clfi (t) + c2h(t) is also a solution of y" + p(t)y' + q(t)y = O. CI cos 0 + C2 sin 0 = c[, so Ct = 4. To apply the condition y' (0) = -4, we com-
pute y'(t) = -2CI sin 2t + 2C2 cos 2t, so that y'(O) = -2CI sin 0 + 2C2 cos 0 =
140 Chapter 4 Higher Order Equations
4.1 Second-Order Equations: An Introduction 141
2C2. Therefore, 2C2 = -4, so C2 = -2. The unique solution to the IVP is yet) '"
4 cos 2t - 2 sin 2t. This solution is unique over the interval (-00, 00) because the Theorem 4.3 Wronkians of Solution _ _ _ _ _ _ _ _ _ _ _ _ _ _,
coefficient functions pet) = 0, q(t) = 4, and !(t) = 0 are continuous on (-00, 00),
Consider the second-order linear homogeneous equation y" + p(t)y' + q(t)y =
and the Existence and Uniqueness Theorem guarantees that the solution is as well.'
0, where pet) and q(t) are continuous on the open interval 1. Suppose that YI (t)
and yz(t) are solutions of this ODE on 1. If YI (t) and Y2(t) are linearly depende,nt,
then W(YI(t), Y2(t)) = 0 on 1. If YI(t) and Y2(t) are linearly independent, then
Now we discuss why we can write every solution of the ODE, y" + p(t)y' + q(t)y
= 0 with linearly independent solutions YI (t) and Y2(t), as the linear combination
W(YI (t), yzCt))"* 0 for all values on 1.
y(t) = CIYI(t) + C2Y2(t). Generally, we call the formy(t) = CIYI(t) + C2Y2(t) a general
solution of this ODE. We begin by stating the following definition.
We omit the proof of this theorem here. We leave it as part of the section exer-
cises (see Abel's formula).
Definition 4.2 Wronskian of a Set of Two Functions
Let S = ifl (t), h(t)} be a set of differentiable functions. The Wronskian of S,
denoted by W(S) = W(jj(t),h(t)), is the determinant
Show that YI (t) = t V3 and yz(t) = t are solutions of 3t 2y" - 2ty' + 2y = O. Com-
W(S) = IAi:j j~~ZI = fi(t)!,(t) - !!(t)h(t)· pute the Wronskian of S = {t 2/3 , t}. Does your result contradict Theorem 4.3?
;] Solution First, we find YIlt) = 2t- 1I3 /3 and y~(t) = -2t- 41319. Then,
.. 3t>y" - 2ty' + 2y = 3t 2( - 2t - 4/319) - 2t(2t - 1/3 /3) + 2t 2/3 = (-~ - j + 2)t 2/3 =
"f·,,·,'.' 0, so n(t) = t 2/3 satisfies the ODE. In a similar manner, we can show that
Y2(t) = t is also a solution. (We leave this to you as an exercise.) Now,
12~::/3 I =
Compute the Wronskian of each of the following sets of functions.
V3 2/3 2l3
(a) S = {sin t, cos t}; (b) S = {e', tet}; (e) S = {2t, 4t}. W(S) = : t - tt = +t •
jJ Solution We compute a 2 X 2 determinant with I~ ~I = ad - bc. "*

Notice that W(S) 0 for all values of t except t = O. However, this does not con-
tradict Theorem 4.3. After writing the ODE in normal form,y" - 2/(3t)y' + 2/(3t 2 )y
= 0, we see that the coefficient functions p(t) = -2/(3t) and q(t) = 2/(3t 2) are not
(a) W(S) = Iat
d s(in. t)
sm t
d c(os t
dt cos t)
I= I sin t
cos t
cos t
-sin t
I continuous at t = O. The theorem holds only on an open intervall where these func-
tions are continuous.
= -sin2 t - cos 2 t = -(sin2 t + cos 2 t) = -1.
(b) W(S) = I~ ~t;:e')1

e;e') = I:: t/~ e,l Theorem 4.4 General Solution
Consider the second-order linear homogeneous equation y" + p(t)y' + q(t)y =
= e'(te t + e') - e'(te t ) = te 2' + e 2t _ te 2t = e 2'. 0, where pet) and q(t) are continuous on the open interval 1. Suppose that YI (t)
and Y2(t) are linearly independent solutions of this ODE on 1. Then, if Y is any
(e) W(S) = 2t
2 4tl
4 = 8t - 8t = 0 solution of this ODE, there are constants CI and C2 so that yet) = CIYI(t) + CzY2(t)
1 for all t on 1.
In part (e) of Example 4, we found that W(S) = O. Recall also that we concluded
in Example 1 that the functions in S = {2t, 4t} are linearly dependent. This theorem states that if we have two linearly independent solutions of a
second-order linear homogeneous equation, then we have all solutions, called a gen-
...
142 Chapter 4 Higher Order Equations 4.1 Second-Order Equations: An Introduction 143
2t
eral solution, given by the linear combination of the two solutions. In addition, we call so Yl (t) = e -51 is a solution. In a similar manner, we can show that Y2(t) = e is
{Yl (t), Y2(t)} a fundamental set of solutions, and we note that a fundamental set of also a solution. Next, we verify that these functions are linearly independent. We
solutions for y' + p(t)y' + q(t)y = 0 must contain two linearly independent solutions. can establish this by noting that the two are not constant multiples, or we can com-
The number of functions in the set equals the order of the ODE, and these functions- pute
must be linearly independent. We will generalize this result to higher-order equations
later in this chapter. W(Yl> Y2) =
e-St
_5e-51
e
2t
I
2e 21 = 2e
-31
- (-5e
-3t -
) - 7e
-3t
.
I
PROOF OF THEOREM 4.4 The Wronskian is not identically zero, so the functions are linearly independent.
Therefore, because we have two linearly independent solutions to the second-
*"
Let t = to be a value on I where WCYI(t), Y2(t)) O. Then, Y(to) = CIYl(tO) + order ODE, a general solution is the linear combination of the solutions,
C2Y2(tO) and Y' (to) = Clyi(tO) + C2y;(tO)' We can write this system of two equa- Y = cle- 51 + c2e 21 . In addition, we say that {e- 51 , e 21 } is a fundamental set of
tions as solutions of y" + 3y' - lOy = O.
Reduction of Order
where the determinant of the matrix of coefficients at t = to, W(Y1 (to), Y2(to)) = In the next section, we learn how to find solutions of homogeneous equations with con-
Yl (to)y;(to) - yi(tO)Y2(tO), is not zero because Yl (t) and Yit) are linearly independ- stant coefficients. In doing so, we will find it necessary to determine a second linearly
ent. This means that the system has a nontrivial solution. That is, (Cl> C2) (0, 0),*" independent solution from a known solution. We illustrate this procedure, called re-
or equivalently, at least one of the values Cl or C2 is not zero. These values can be duction of order, by considering a second-order equation.
found through elimination to be Suppose we have the equation
- Y(to)y[(to) + Y'(tO)YI(tO) y" + p(t)y' + q(t)y = 0,
and
Yl (to)y2(to) - y;(to)Y2(to) .
and that y = f(t) is a solution. We know from our previous discussion that to find a
*"
Notice that these values make sense only if YI(tO)y;(to) - y;(to)Y2(to) 0, as we general solution of this second-order equation, we must have two linearly independent
mentioned earlier in connection to the linear independence of Yl(t) and Y2(t). solutions. We must determine a second linearly independent solution. We accomplish
Using these values of Cl and C20 let Z(t) = CIYI (t) + C2Y2(t) be a solution of y" + this by attempting to find a solution of the form
p(t)y' + q(t)y = O. Then, in addition to satisfying the ODE, Z(to) = CIYI (to) +
y = v(t)f(t)
c2y2Cto) = Y(t o) and Z' (to) = C,y; (to) + C2y;(tO) = Y' (to). By the Existence and
Uniqueness Theorem, the IVP and solving for vet). Differentiating with the product rule, we obtain
y" + p(t)y' + q(t)y = 0, y(to) = Y(to), y'(to) = Y'(to) y' = f'v + v'f and y" = f"v + 2v'f' + fv".
has a unique solution. Because Z(t) and yet) each satisfy the ODE and have the For convenience, we have omitted the argument of these functions. We now substitute
same initial conditions, they must be equal (that is, Z(t) '" yet) on I). There is only y,y', andy" into the equationy" + p(t)y' + q(t)y = O. This gives us
one solution to the IVP
y" + p(t)y' + q(t)y =!"v + 2v'f' + fv" + p(t)(f'v + fv') + q(t)vf
= If'' + p(t)f' + q(t)flv + fv" + 2v'f' + p(t)fv'
=0
l'i'"I,I.,1 = fv" + (2f' + p(t)f)v'.
Show thaty = cle- 51 + C2e21 is a general solution ofy" + 3y' - lOy = O.
Therefore, we have the equation
: ] Solution First, we must verify that Yl (t) = e - 51 and Y2(t) = e 21 satisfy the ODE. fv" + (2f' + p(t)f)v' = 0,
Withy;(t) = -5e- 51 andyj'(t) = 25e- 51 , we have which can be written as a first-order equation by letting w = v'. Making this substitu-
yj' + 3y; - IOYI = 25e- 51 + 3(-5e- 51 ) - lOe- 51 = 0, tion gives us the linear first-order equation
144 Chapter 4 Higher Order Equations 4.1 Second-Order Equations: An Introduction 145
I Solution In this case, we must divide by 4t 2 to obtain an equation of the fonn

fw' + (2f' + p(t)f)w = 0 or f~~ + (2f' + p(t)f)w = 0 . . .
2
d y 2 dy 1 _
y" + p(t)y' + q(t)y = O. ThIS gIves us the equatIon """if2 + Idt + 4t 2 y - O.
which is separable, so we obtain the separated equation
Therefore, pet) = 21t and f(t) = t -112. Using the fonnula for v, we obtain
-fp(t)dt f e-f2ltdt f e- 21n(t) f-t 1 dt = In t, t
We solve this equation by integrating both sides of the equation to yield
vet) =
f _e_ _ dt =
[f(t)f
- _ - - dt =
[t 1/2f
----1-
t
dt = > O.
A second solution is y = f(t)v(t) = t -1/2 In t.
Inlwi = In ()2) - f p(t) dt.
This means that w = J...e-fp(t)dt so we have the fonnula dv = J...e-fp(t)dt or

f2' dtf2
e-fP(t)dt
v(t) = f [f(t)1' dt.
EXERCISES 4_1
If Y = f(t) is a known solution of the differential equation y" + p(t)y' + q(t)y = 0, we
can obtain a second linearly independent solution of the form y = f(t)v(t), where
e-Jp(t)dt
vet) = J [f(t)f dt.

In Exercises 1-6, calculate the Wronskian of the indicated set y" + 4y = 0
e-fp(t)dt of functions. Classify each set of functions as linearly inde- 14. {yeO) = -1, y'(O) = 2'
We leave the proof that Y1 (t) = f(t) and Y2(t) = f(t)v(t) = f(t)
linearly independent as an exercise. f ---2
[f(t)]
dt are pendent or linearly dependent. y = CI cos 2t + C2 sin 2t,
y" - 8y' + l6y = 0
-<Xl < t< <Xl
1. S = {t, 4t - I} 2. S= {t, e'}

*15. {y(l) = 0, y'(l) = _e 4 ,
*3. S = {e- 6 ', e- 4t } 4. S = {cos 2t, sin 2t}
y = cle 41 + C2te4r , -00 < t < 00
*5. S = {e -3t cos 3t, e -3t sin 3t}
'if',,',·.· 6. S = {eSt cos 4t, eSt sin 4t}
2
16 [t y" + 7ty' -7y= 0 y=c l t- 7
. )y(I) = 2, y'(l) = -22'
+ C2t , t>O
Determine a second linearly independent solution to the differential equation [y" + Y = 2 cos t
y" + 6y' + 9y = 0 given thaty = e- 3t is a solution.
In Exercises 7-12, show thaty = c,y,(t) + C2y,(t) satisfies the
given differential equation and that {n(t), Y2(t)} is a linearly 17. )YeO) = 1, y'(O) = l'
independent set. Y= Cl cos t + C2 sin t + t sin t, -00 <t< 00
: 1 Solution First we identify the functions pet) = 6 andf(t) = e- 3t. Then we de-
18. Use the Wronskian to show that ifn andY2 are solu-
tennine the function vet) such that y(t) = f(t)v(t) with the fonnula 7. Y = cle t
+ c2e-r,y" - Y= 0
tions of the first-order differential equation
vet) = f e-fp(t)dt
[f(t)1' dt =
f e- f6dt
[e-3tf dt =
f e- 6t
e- 6t dt =
f dt = t.
8. Y =
*9. Y =
clef+ C2e- 3t, y" + 2y' - 3y = 0
cle- r + czte- l , y" + 2y' + Y = 0
y' + p(t)y = 0,
10. Y = c,e 2t + C2te2', y" - 4y' + 4y = 0 then y I and Yz are linearly dependent.
A second linearly independent solution is y = f(t)v(t) = te -3t. 11. y = c, cos 2t + C2 sin 2t, y" + 4y = 0 19. Consider the hyperbolic trigonometric functions
cosh t = (e t + e -')/2 and sinh t = (e t - e-')/2. Show
12. y = e"(cl cos 4t + C2 sin 4t), y" - 6y' + 25y = 0
that
In Exercises 13-17, use the given solution on the indicated in- (a) dldt (cosh t) = sinh t.
terval to find the solution to the initial-value problem. (b) dldt (sinh t) = cosh t.
Determine a second linearly independent solution to the differential equation Y" - y' - 2y 0 = (e) cosh2 t - sinh2 t = l.
2
2d d
4t --? + 8t ...L
dt dt
+ y = 0, t> 0, ify = t- 1/2 is a solution. 13. {yeO) = -1,y'(O) = -5'
Y = Cle- r + e2e21, -co < t < 00
(d) cosh t and sinh t are linearly independent func-
tions.
...
4.2 Solutions of Second·Order linear Homogeneous Equations
with Constant Coefficients 147
In Exercises 20-23, show that S is a fundamental set of solu·
tions for the given equation. (See Exercises 1-6 to verify lin.
.
twnf(t)
f e-Ip(t)dt
U(t)f dt obtained by reduction of order 14.21 Solutions of Second-Order Linear Homogeneous
ear independence.) are linearly independent. Hint: Use the WronSkian. Equations with Constant Coefficients
20. S = {e- 6', e- 4'}; y" + lOy'+ 24y = 0 38. Show that a general solution of y" - k 2y = 0 is Y "'- (' Two Distinct Real Roots r: Repeated Roots (' Complex Conjugate Roots
21. S = {cos 2t, sin 2t}; y" + 4y = 0 c, cosh kt + C2 sinh kt, where cosh kt = (e kt + e -kt)/2
and SInh kt = (e'" - e- kt )/2.
22. S = {e-" cos 3t, e-" sin 31}; y" + 6y' + l8y = 0 In Section 4.1, we started to investigate the method by which we solve second-order
39. (Abel's Formula) Suppose that y, and Y2 are two so. linear homogeneous equations. Here, we restrict our attention to equations with con-
*23. S = {e s , cos 4t, e S' sin 4t}; y" - lOy' + 41y = 0
lutions ofy" + p(t)y' + q(l)y = 0 on an open interval stant coefficients,
24. Let ay" + by' + cy = 0 be a homogeneous second. I, where pet) and q(t) are continuous. Then, the Wron.
order equation with constant coefficients and let m I skian of y, and y, is ay" + by' + cy = 0,
and m2 be the solutions of the equation am2 + bm +
c = O. W(y" y,) = Ce - Jp(>)d'. and we solve the equation by assuming that Y = e" is a solution for some value(s) of
Prove Abel's formula using the following steps. Be. r. We find r by substituting this solution into the ODE. With y' = re" and y" = r 2e",
(a) Ifml '" m2 and both ml and m2 are real, show that
{em", em,,} is a fundamental set of solutions of cause YI and y, are solutions, begin with the system we obtain ar 2e" + bre" + ce" = 0, or
ay" + by' + cy= O. YI + p(t)y! + q(t)YI = 0 e"(ar 2 + br + c) = O.
(h) If m, = m2, show that {em", xe m,,} is a funda- { yz + p(tM + q(t)y, = 0.
mental set of solutions of ay" + by' + cy = O. Therefore, we solve the characteristic equation
(a) Multiply the tITst equation by (-y,) and the
(c) If m, = " + i{3, {3'" 0, and m2 = m; = 01 - i{3, second by YI' Add the reSUlting equations to obtain ar2 + br + c = 0
show that {e a , cos {3t, e a , sin {3t} is a fundamen-
tal set of solutions of ay" + by' + cy = O.
(YIY2 - Yly,) + P(Y'Y2 - YI Y2) = O.
that (dldt) W(YJ, y,) = y'Y2 - Ylh (c) Use
(b) Show
to determine r (because e" * 0 for all values of t). Of course, the roots are
the results in (a) and (b) to obtain the first-order -b+~
In Exercises 25-34, use reduction of order with the solution ODE dWldt + pW= O. Solve this equation for w: r=
2a
n (I) to find a second linearly independent solution of the given 40. Can the Wronskian be zero at only one value of t on
differential equation. so the roots of this quadratic equation depend on the values of a, b, and c. There are
l? Hint: Use Abel's formula.
25. YI(t) = e3 ', y" - 5y' + 6y = 0 *41. (a) Find conditions on the constants c" C2, c" and e,
three possibilities: two real distinct roots when b2 - 4ac > 0, one real repeated root
when b 2 - 4ac = 0, and two complex conjugate roots when b 2 - 4ac < O. We now
26. YI(t) = e- 2',y" + 6y' + 8y = 0 so thaty(l) = c, + C2 tan(c, + C4 In t) is a solution of
the nonlinear second·order equation ty" - 2yy' = O. consider each case.
*27. YI(t) = e2', y" - 4y' + 4y = 0
(b) Is the Principle of Superposition valid for this
28. YI(t) = e-", y" + lOy' + 25y = 0 equation? Explain.
Two Distinct Real Roots
29. YI(t) = cos 2t, y" + 4y = 0 42. Find a general solution of 4t 2y" + 4ty' +
30. n(t) = sin 7t, y" + 49y = 0 2 Suppose that the characteristic equation of an ODE has two roots rl and r2, where
*31. YI(t) = t- 4 , t 2y" + 4ty' - 4y = 0
(36t - l)y = 0 given that y = (cos 3t)I"Vt is one
solution. rl *
r2, so two solutions to the ODE are YI (I) = e'I' and y,(t) = e'z'. We show that these
*43. Given that y = (sin 4t)lt is a solution of ty" + 2y' + solutions are linearly independent by computing the Wronskian
32. YI(t) = t- 2 , t 2y" + 6ty' + 6y = 0
16y = 0, find and graph the solution of the equation
33. YI(t) = t, t y" + 3ty' + Y = 0
2
that satisfies y (7T/8) and y' (7T/8) = - 321 7T.
34. YI(t) = t-" t'y" - ty' +Y = 0
44. Find a general solution of y" + b(t)y' + c(t)y = 0
35. Find a and b so that S = {e -, cos 2t, e -, sin 2t} is a given that y = (sin t)lt 2 is one solution. These functions are linearly indepe'ldent because r2 - rl *
0, so that Wen, Y2) * O.
fundamental set of solutions for y" + ay' + by = O. Therefore, a general solution is y(t) = CIYI(t) + C2Y2(t) = cle'I' + C2e'z'.
*45. Given that y = (cos t)lt' is a solution of y" +
36. Is the Principle of Superposition ever valid for linear b(l)y' + c(l)y = 0, tmd and graph the solution(s), if
nonhomogeneous equations? Explain. any, that satisfy the boundary conditions y( 7T) =
*37. Suppose that f(t) is a solution to the equation y(27T) = O.
Theorem 4.5 Distinct Real Roots
y" + p(t)y' + q(t)y = O. Show thatf(t) and the solu- --------------------------~
Let rl and r2 be the real solutions of ar2 + br + c = O. If rl * r2, then a gen-
eral solution of ay" + by' + cy = 0 is yet) = cle'I' + C2e"'.
148 Chapter 4 Higher Order Equations 4.2 Solutions of Second-Order Linear Homogeneous Equations
I'ii,,','.' Complex Conjugate Roots

Solve y" + 3y' - 4y = O. Suppose that the characteristic equation of an ODE has the complex conjugat: ~s
rl = '" + f3i and r2 = '" - f3i where", and f3 (f3 > 0) are real numbers and I = Y-1.
Note: Sometimes, we denote the roots as rl.2 = '" :!: f3i.
: .1 Solution The characteristic equation for this ODE is r2 + 3r - 4 '"
To construct a real-valued general solution, we use Euler's formula,
(r + 4)(r - 1) = 0, with roots rl = -4 and r2 = I. Therefore, a general solution
is yet) = cle- 4t + C2e t. ei8 = cos IJ + i sin IJ,
l.
which can be obtained through the use of the Maclaurin series:
Repeated Root tk t2 t3 t4 t S
Suppose that the characteristic equation of an ODE has a repeated root r = rl = rz. In
et ="
.6'0 -kl =I + t + -2! + -3! + -41 + -51. + ....
00
2 (_I)kt 2k t2 t4
other words, b - 4ac = 0 in the quadratic formula so that the root is r = -bl(2a) or 00
2
(r + b/(2a)) is a factor in the characteristic equation. Because r = -b/(2a) is repeated, cos t = ~o (2k)! = I - 2T + 4T - .'.,
we only have one solution of the ODE, YI (t) = erl = e -btl(2a). We use this solution to
obtain a second linearly independent solution through reduction of order with the for- and
mula • 00 (_I/t 2k+l t3 t5
e-fp(t)dt SID t =~ e2k + 1)1 = t - 3T + 51 - ....
yz(t) = YI(t) f LYI(t)f dt,
We derive Euler's formula using these Maclaurin series and substitution:
where p(t) appears in the general equation y" + p(t)y' + q(t)y = O. In this case, the 00 (iIJ/ i2~ hP i4rj is!)'
e,8 = " - - = 1 + iIJ + - - + - - + - - + - - + ...
ODE is ay" + by' + cy = o. Dividing by a yields y" + (bla)y' + (da)y = 0, so pet) .6'0 k! 2! 3! 4! 5!
= bla. Returning to the reduction of order formula, we find that
~ i~ rj i!)'
-feb/aId, f _ee-btla
-btla f = I + iIJ - 2T - 3! + 4T + 5! + ...
Y-o(t) = e- bt/(2a)
f e
[e bt/(2a)f
dt = e- bt/(2a) __ dt = -bt/(2a) dt = te-btl(2a)
e .
= (I - ~2! + !t....4! + ... ) + i(IJ - !...3! + !..5! + ...) = cos IJ + i sin IJ.
Therefore, a general solution is y(t) = cle-bt/(Za) + czte -bt/(20) = cIeri + C2 te ".
Using the properties of sine and cosine, Euler's formula also implies that e -i8 =
cos IJ - i sin IJ. The roots of the characteristic equation are m 1.2 = '" :!: f3i, so YI =
e(,,+/3i)t = e"t(cos f3t + i sin f3t) and Y2 = e("-/3i)t = eot(cos f3t - i sin f3t) are both
Theorem 4.6 Repeated Root solutions to the differential equation. By the Principle of Superposition, any linear
Let r = rl = r2 be the repeated (real) solution of ar2 + br + c = O. A general combination of Yl and Y2 is also a solution. For example, Zl = t(Yl + Yz) = e"t cos f3t
solution of ay" + by' + cy = 0 is yet) = cIeri + C2te". and Z2 = -1(n - Y2) = e'" sin f3t are both solutions. You should verify that
{e'" cos f3t, e"t sin f3t} is linearly independent, so a general solution is y =
eot(cl cos f3t + C2 sin f3t).
"",,','M'
Solve y" + 2y' + Y = O. Theorem 4.7 Complex Conjugate Roots
Let r = '" + f3i and r2 = '" - f3i, where", and f3 (f3 > 0) are real numbers, be
:.J Solution The characteristic equation for this ODE is r2 + 2r + I = (r + 1f = 0, the s~lutions of ar2 + br + c = O. A general solution of ay" + by' + cy = 0 is
with roots rl = r2 = -1. Therefore, a general solution is y(t) = cle- t + C2 te - t. y(t) = CleO' cos f3t + cze'" sin f3t = e"'(cl cos f3t + C2 sin f3t).
====:;;;",J
150 Chapter 4 Higher Order Equations 4.2 Solutions of Second-Order linear Homogeneous Equations
"ii,,',".1
Solve y" + 4y' + 20y
! ] Solution
= 0, yeO) = 3, y'(O) = -1.
The characteristic equation is r2 + 4r + 20 = 0 with roots

- EXERCISES 4.2
In Exercises 1-15, find a general solution of each equation. (b) Substitute y and y' into the Ricatti equation to ob-
tain the second-order equation
-4 + Y4
2
- 4(20) -4 ± V-64 -4 ± 8i . 1. y" + 8y' + 12y = 0 2. y" - 4y' - 12y = 0
1".2 = 2 = 2 = --2-- = -2 ± 4,.
*3. y" + 3y' - 4y = 0 4. y" + y' - 72y = 0 w"(t) (W'(t»2 a'(t)w'(t)
a(l)w(t) - a(t)(w(t»)' - (a(t»)'w(t) +
5. y" + 16y = 0 6. 4y" + 9y = 0
Therefore, in our earlier notation, '" = - 2 and j3 = 4, so that a general solution of
the ODE is *7. y" + 7y = 0 8. y" + 8y = 0 (W,(t»2 + b(t)w'(t) + c(t) = O.
4 a(t)(w(t»2 a(t)w(t)
yet) = e- 2t(c, cos 4t + C2 sin 4t), 9. y" - 6y' + 25y = 0
10. y" - 8y' + 20y = 0 Multiply this equation by a(t)w(t) to obtain
because the solutions of the characteristic equation are complex conjugates. To find + b(t)w'(t) + a(t)c(t)w(t)
*11. y" + 6y' + 18y = 0 w"(t) - a'(t)w'(t) = 0
the solution for whichy(O) = 3 andy'(O) = -I, we first calculate aCt)
12. 4y" + 21y' + 5y = 0 and simplify the result to obtain the second-order
2t
y' = 2e- (-c, cos 4t + 2C2 cos 4t - 2c, sin 4t - C2 sin 4t) 13. 7y" + 4y' - 3y = 0 equation
with the product rule and then evaluate both 14. y" + 4y' + 4y = 0
w" - (a'(t) - b(I»)W' + a(l)c(I)W = O.
-1 2 4 x *15. y" - 6y' + 9y = 0 aCt)
yeO) = c, and y'(O) = 2(2c2 - c,)
In Exercises 16-27, solve the initial-value problem. Graph the 29. Convert the Ricatti equation y' + (t 4 + t 2 + l)y2 +
obtaining the system of equations {2(2 C)I : 3 l' Substituting c, = 3 into the 2 3 4
Figure 4.4 Graph of C2 - Cl - -
solution on an appropriate interval. 2(1 - t + t - 21 + t ) + 1 = 0 to a
I + t2 + t4
2
yet) = e- '(3 cos41+ i sin 4t) second equation results in C2 = 5/4. Thus, the solution of y" + 4y' + 20y = 0 for
whichy(O) = 3 andy'(O) = -1 isy(t) = e- 2t(3 cos 4t + ~ sin 4t).
16. y" - y' = O,y(O) = 3,y'(0) = 2
17. 3y" - y' = 0, yeO) = 0, y'(O) = 7
y t 4 + 12 + I
second-order equation by using the following steps.
18. y" + y' - 12y = 0, yeO) = 0, y'(O) = 7 (a) Note that aCt) = t 4 + t 2 + 1,
2(1 - t + t 2 - 21 3 + (4)
In Example 3, evaluate limH~y(t). Does this limit depend on the initial conditions? *19. y" - 7y' + 12y = 0, yeO) = 3, y'(O) = -2 bet) = 1 + 12 + t4 , and
Does this result agree with the graph in Figure 4.4? 20. 2y" - 7y' - 4y = 0, yeO) = O,y'(O) = I 1 w'(t) 1
+ lOy = 0, yeO) = I, y'(O) = 5 c(t) = . Let yet) = - - - - so that
Theorems 4.5 through 4.7 can be summarized as follows: 21. y" - 7y' t4 + t 2 + I wet) aCt)
22. y" + 36y = 0, yeO) = 2,y'(0) = -6 the second-order equation is
Solving Second-Order Equations with Constant Coefficients

*23. y" + 100y = 0, yeO) = 1, y'(O) = 10 4t3 + 2t 2(1 - t + t - 2t
2 3
+ t4 ) ,
24. y" - 2y' +Y = O,y(O) = 4,y'(0) = 0 W" - ( (4 + (2 + I - 1 + P + t4 w +
Let ay" + by' + cy = 0 be a homogeneous second-order equation with constant 25. y" + 4y' + 4y = O,y(O) = 1, y'(O) = 3
real coefficients and let m 1 and m2 be the solutions of the equation am 2 + bm +
26. y" + 2y' + 5y = 0, yeO) = I, y'(O) = 0 (t
4
+ t2 + I) t 4 + ~2 + 1 w = 0,
c = O.
*27. y" + 4y' + 20y = 0, yeO) = 2, y'(O) = 0
which simplifies to w' - 2w' + w = O.
CD If m, =fo m2 and both m, and m2 are real, a general solution of ay" + by' + cy 28. A Ricatti equation, named for the Italian mathe-
= 0 is matician Jacopo Francesco Ricatti (1676-1754), is a (b) Show that wet) = C,e' + C 2te t is a general solu-
nonlinear first-order equation of the form tion ofw" - 2w' + w = o.
y' + a(t)y2 + b(t)y + c(t) = O. w'(t) 1

(c) Use y(l) = - - - - to show that
@ If m, = m2, a general solution of ay" + by' + cy = 0 is
. . w'(t) 1 wet) a(l)
We make the substitutIOn yet) = ~ aCt) to solve ......,_C:::!.;,e~'...:.+_C:::2"'(:-et-;+,.....:le-;'):---;-:: satisfies the
Ricatt, equanons. yet) = (t4 + t 2 + 1)(C,e' + C 21e')
m, is the complex conjugate
ofm,:
® If m 1 = a + ij3, j3
by' + cy = 0 is
* 0, and m2 = m, = a - ij3, a general solution of ay" + (a) Show that
Ricatti equation.
, w"(t) (w'(t»)' _ a'(t)w'(t) In Exercises 30-35, solve the given Ricatti equation. (See Ex-
Tn; + if3 = a
= '" a - if3. y = Cle ' cos j3t + C2eat sin j3t = e'''(ci cos j3t + C2 sin j3t). y (t) = a(t)w(t) - a(t)(w(t»2 (a(t»)'w(t)' ercises 28 and 29.)
4.3 Higher Order Equations: An Introduction 153
2
30' 1 2 _ 2(-6 + t - 6t ) 41. Show that the boundary-value problem
'Y+t 2 +1 Y t 2 +1 y+ \4.3\ Higher Order Equations: An Introduction
J y"+2y'+5y=0 h inr'
45(t 2 + 1) = O. )YeO) = 0,y(7T/2) = 0 as Imtely many solutions,
The ideas presented for second-order linear homogeneous equations in Sections 4.1
31. y' + sin ty2 - (2 - cot t)y + csc t = 0
32. y' + t 2 cos ty2 _ t tan t + 2t - 2 y + 2 s;c t = 0

the boundary-value problem
L
y "+2y '+5y -0 -
(0) = 0, y( 7T/4) = 0
has no nontrivial solutions, and that the boundary.
and 4.2 can be extended to those of order three or higher. We begin with the general
form of an nth-order linear equation.
t t J. "+2 '+5 =0
value problem L/c0) = ;'Y(7T~) = 0 has one (non.
33. y' + sin t / _ I - 2t - t cot t Y + 26t csc t = 0
t t trivial) solution.
Definition 4.3 nth-order Ordinary Linear Differential Equation
34.
y
'+ _,_t_
t-+ 4y
2 _ t: -+
t(t-
4
4) y
+ 4(/2 + 4) = 0
/
*42. Use factoring to solve each of the following nonlin.
eat equations. For which equations, if any, is the Prin.
An ordinary differential equation of the form
ciple of Superposition valid? Explain an(/)y(nlCt) + an_ICt)/n-ll(/) + ... + al(/)y'Ct) + ao(/)y(/) = g(/),
35. y' +ttan4t+ 1+ lOt + 4t csc4/sec 4/ y+
t (a) (y")2 - 5y"y + 4y2 = 0 where an(/) ¢ 0, is called an nth-order ordinary linear differential equation.
41 cot 4/ = 0 (b) (y'')2 - 2y"y + / = 0 If g (I) is identically the zero function, the equation is said to be homogeneous;
/
43. Find conditions on a and b, if possible, so that the so. if g(/) is not the zero function, the equation is said to be nonhomogeneous; and
36. Consider the homogeneous equation y" + p(t)y' + lution to the initial-value problem if the functions all), i = 0, 1,2, ... , n are constants, the equation is said to have
q(t)y = O. (a) If net) satisfies the equation and c is a constant coefficients. An nth-order equation accompanied by the conditions
constant, show that yet) = CYI (t) satisfies the equation. Jy" + 4y' + 3y = 0
(b) If h(t) and Y2(t) ate both solutions of the equa- )YeO) = a,y'(O) = b YCto) = Yo, y'Cto) = Yo, ... ,/n-I)Cto) = y~-l)
tion, show thaty(/) = Yl(t) + h(t) is a solution.
has (a) neither local maxima nor local minima; (b) ex. where Yo, Yo, ... , y~-ll are constants is called an nth-order initial-value
37. Use the substitution t = eX to solve (a) 3t 2y" - 2ty' actly one local maximum; and (c) exactly one local problem.
+ 2y = 0, t> 0 and (b) t 2y" - ty' + y = 0, x> O. minimum on the interval [0, "').
2
Hint: Show that if t = eX, dy = ~ dy and d y = 44. (a) Show that the roots of the characteristic equation
dt t dx dt 2 On the basis of this definition, the equation y'" - 8y" + 1Oy' - 3y = cos t is a
2y d 2x dx
.!.(d _ dY) of dt 2 + 01 dt + aox = 0 ate third-order nonhomogeneous equation with constant coefficients, while /4) - 12y = 0
t 2 dx 2 dx'
is a fourth-order homogeneous equation. We assume that the coefficient functions
-at±~
~ kI
k
38. (a) Use the Maclaurin series e t = L t
and the ml,2 = 2 anCt), ... , 00(1) and g (I) in the equation
k=Q •
Maclaurin series for sin t and cos t to prove e -;8 = (b) Graph ao = kay and al = 0, using the horizontal
an(/)/nl(/) + an_I(/)/n-ll(/) + ... + aoCt)y(/) = g(/)
cos 8 - i sin 8. (b) Use (a) and trigonometric identi- axis to represent ao and the vertical axis to repre. are continuous on an open interval I. Therefore, if we also assume that anCt) 0/= 0 for
ties to prove that e i • = cos 8 + i sin 8. sent al' Randomly generate three points on the all I on J, we can divide by an(/) to place the ODE in normal fonn,
39. Show that a general solution of the differential equa- graph of ao = ~ay, three points in the region ao <
tion ay" + 2by' + cy = 0 where b 2 - ac> 0 can be laI, and three points in the region ao > ~ai. /nl(/) + PnCt)y(n-Il(/) + '" + PI(/)y(t) = J(/),
written as (e) Write the second-order homogeneous equation where PnCt) = an-I(/)lan(t), ... , PICt) = ao(t)lai/), andJCt) = gCt)lan(/).
d 2x dx
Y - e- -btla[ Cl cosh t~ + C2 81'nh t~] . dt' + a 1 dt + aox = 0 as a system by letting In Sections 4.1 and 4.2, we discussed how we needed two linearly independent
a a solutions to solve a second-order linear homogeneous ODE. Therefore, we gave a def-
y=x,=dx inition for the linear dependence or linear independence of a set of two functions. We
40. Express the solution to each differential equation in dt
now define this concept for a set of n functions because we need to understand this
terms of the hyperbolic trigonometric function (see
Exercise 39).
(d) For each pair of points obtained in (b), graph property to solve the higher order equations.
the phase plane associated with the system for
(a) y" + 6y' + 2y = 0 -I :$ x :$ I and -I :$ Y :$ 1. Compate your
(b) y" - 5y' + 6y=0 results with your classmates. Definition 4.4 Linearly Dependent and Linearly Independent
(e) y" - 6y' - l6y = 0 (e) How does the phase plane associated with the sys· Let S = ifl Ct),f2(t),/J(/), ... ,fn-l Ct),fnCt)} be a set of n functions. S is linearly
tern change as the roots of the chatacteristic equa·
(d) y" - 16y = 0 dependent on an interval J if there are constants c" C2, ••• , Cn , not all zero, so that
tion change?
154 Chapter 4 Higher Order Equations 4.3 Higher Order Equations: An Introduction 155
Another way to express Theorem 4.8 is to say that any linear combination of so-
cdi(t) + cz!2(t) + ... + C.-lJ,,-I(t) + c.J.(t) = 0
lutions of/')(I) + P.(t)/·-l)(t) + ... + PI(t)y(t) = 0 is also a solution of this ODE.
for every value of t in the interval I.
S is linearly independent if S is not linearly dependent.
"i'd,,'M'
Thus, S = {fi(I),!2(t),h(I), ... ,J,,-I(t),J,,(t)} is linearly independent if Show that YI (t) = e - I , Y2(t) = cos I, and Y3(t) = sin t are solutions of the third_
order linear homogeneous ODEy''' + y" + y' + Y = o. Use the Principle of Super-
cJ,.(t) + cz!2(t) + ... + C.-tin-l(t) + cnJ,,(t) = 0 position to find another solution to this ODE.
implies that Cl = Cz = ... = c. = o. I
: .1 Solution For YI(t) = e- I, we havey;(t) = -e-I,Yl(t) = e- , andy~'(t) = -e- •
I
Then, Y'{' + Yl + Y; + YI = -e- I + e- I - e- I + e- = 0, so YI satisfies the ODE.

I
In a similar manner, we can show that yz and Y3 are solutions ofthe ODE. (We leave
this to the reader.) Therefore by the Principle of Superposition, the linear combina-
Classify the sets of functions as linearly independent or linearly dependent.
tion of these three functions, yet) = cle - I + Cz cos t + C3 sin t, is also a solution of
(a) S = {t, 3t - 6, I} (b) S = {cos 2t, sin 21, sin I cos t} (e) S = {I, I, IZ}
the ODE.
: ] Solution (a) Here, we must find constants CJ, Cz, and C3 such that
As with first-order and second-order initial-value problems, we can state an ex-
Cit + cz(3t - 6) + C3(l) = (Cl + 3cz)t + (C3 - 6cz) = O.
istence and uniqueness theorem for an initial-value problem involving a higher order
Equating each of the coefficients to zero leads to the system of equations linear ordinary differential equation.
{CI + 3cz = 0, C3 - 6cz = O}. This system has infinitely many solutions of the fonn
{CI = -3cz, C3 = 6cz, Cz arbitrary}. With the choice CI = -3 Cz = 1 and C = 6
the equati?n is satisfie~. Therefore, the set S = {I, 31 - 6, I} i~ linearl~ depe~dent: Theorem 4.9 Existence and Uniqueness
(b) For thiS set of functIOns, we apply the identity sin 2t = 2 sin I cos t. Doing this,
we conSider the equation Suppose that P.(t), ... , PI (t), and J(t) are continuous functions on an open in-
terval I that contains t = to. Then, the initial value problem
CI cos 2t + Cz sin 21 + C3 sin I cos I =
CI cos 2t + 2cz sin 1 cos t + C3 sin t cos t = O. y(')(I) + Pn(t)y(n-Il(t) + ... + PI (t)y(t) = J(I), y(to) = b o, y' (to) = bJ, ... ,
y(n-l)(to) = bn - I
The choices of Cl = 0, Cz = 1, and C3 = -2 lead to a solution. At least one of
the constants is not zero, so the set S = {cos 2t, sin 2t, sin t cos t} is linearly has a unique solution on I.
dependent.
(c) For the set S = {I, t, tZ}, we consider the equation CI(l) + czt + C31z = O. The
?nly constants that satisfy the equation are CI = Cz = C3 = 0, so the set is linearly The proof of this theorem is well beyond the scope of this text but can be found
mdependent. in advanced differential equations textbooks. *
At this point, we need to discuss why we can write every solution of the
. We also extend the Principle of Superposition to include more than two ODE y(n)(t) + Pn(t)y(n-Il(t) + ... + Pl(t)y(t) = 0 as a linear combination of solu-
solutIOns. ' tions if we have n linearly independent solutions of the nth-order equation. (We
call this a general solution of the ODE.) Before moving on, we must define the
Wronskian of a set of n functions. This definition is particularly useful because
Theorem 4.8 Principle of Superposition determining whether a set of functions is linearly independent or dependent be-
rf!I(t)~!A(t), ... ,J,,(I) are solutions of the linear homogeneous ODE y(')(t) + comes more difficult as the number of functions in the set under consideration
P?(t)/ Jet) + ... + PI (t)y(t) = 0 on the interval I, and if CJ, Cz, ... , c. are ar- increases.
bl~ary constants, then y = CI!I(I) + cz!2(I) + ... + c.J,,(t) is also a solution of
this ODE on I. ... For example. see Chapter 2 of C. Corduneanu, Principles oj Differential and Integral Equations.
CheLsea PubLishing Company (1971).
43 Higher Order Equations: An Introduction
Definition 4.5 Wronskian In part (a) of Example 3, we found that W(S) = O. Recall that we concluded that
the functions in S = {t, 3t - 6, I} are linearly dependent because 3t - 6 = 3(t) - 6(1),
Let S.= f!i(t),h:(t),h('), ... ,J,,-I(I),J,,(t)} be a set ofn functions for which which means that one of the functions in the set can be written as a linear combina-
each IS differentiable at least n - I times. The Wronskian of S, denoted by tion of other members of the set. This is no coincidence. The Wronskian of a set of lin-
early dependent fUnctions is identically zero, whereas the Wronskian of a set of linearly
W(S) = W(!J(t),fz(I),h(t), ... ,!n-l(t),J,,(t)),
independent functions is not equal to zero for at least one value. In part (b), we found
is the determinant '*
that W(S) = 2 O. Therefore, the functions are linearly independent as we discussed
Nt) fz(t)
earlier.
J,,(t)
fi(t) f2(t) f~(t)
W(S) =
Theorem 4.10 Wronskian.::.o~f.::.S.::.o~lu~t::.io:..:n::s:"-_ _ _ _ _ _ _ _ _ _ _ _- i
Consider the nth-orderlinearhomogeneous equationy<nJet) + Pn(t)y<n-I)(t) + ...

+ PI(t)y(t) = 0, where Pn(t), ... , PI(t) are continuous functions on an open in-
terval I. Suppose that YI(t), Y2(t), ... , Yn(t) are solutions of this ODE on T. If
. Although most computer algebra systems can quickly find and simplify the Wron- YI(t), Y2(t), ... , Yn(t) are linearly dependent, then W(YI(t), Y2(t), ... , Yn(I)) ==
o on I. If YI (t), Y2(t), ... , Yn(t) are linearly independent, then W(YI (I), Y2(t), ... ,
skian of a set of functions, the following example illustrates how to compute the Wron-
skian for a set of three functions by hand. Yn(t)) '*
0 for all values of t on I.
================~
,'i'"I,iM! We omit the proof of this theorem here. We leave it as part of the section exer-
cises (see Abel's formula).
Compute the Wronskian of each of the following sets of functions.
(a) S = {t, 3t - 6, I}; (b) S = {I, t, t 2}.
:1 Solution We compute a 3 X 3 determinant with

I"i,,','.'
The functions YI(t)= t- 4, Y2(t) = t- 1, Y3(t) = t, andY4(t) = tint are solutions of
t 4y<4) + 9t 3y'" +
Ilt 2y" - 4ty' + 4y = O. Show that these solutions are linearly in-
dependent on the interval t > O.
, I Solution Notice that we have assumed that I > 0, so we can divide each term by
. t 4 to place the ODE in normal form,
(a) W(S) = I:
3t; 6 ~ I. We can carry out the step given above for computing a Y
(4) + 2- y '" + _1_1
I 2
t y"
__ ~y' + ~Y
3
1 4
t
= 0
'
000
3 X 3 determinant. However, the row of zeros indicates that W(S) = O. (You should for use with Theorem 4.10. (Note that we assumed t > 0 so that the coefficient func-
verify this.) tions in the ODE are continuous.) Then, if S = {t-4, t- 1, t, t In I}, we find with the
I: :fH ~H: ~I "I: :1-'

help of a computer algebra system that
(b) "\S) ~ + In .dIDti", ~ W(S) =

t- 4
-4/1 5
I-I
-llt 2
I In t
1 + In t
=(9"
300
20lt 6 21t 3 0 lit
could have. observed that all nonzero entries lie on the diagonal or above, so the de-
-1201t 7 -6114 0 -1/t2
tenrunant IS the product of the numbers on the diagonal. Note: This approach also
works ~f the nonzero entries lie on the diagonal or below. Matrices having these '*
so W(S) 0 for t > O. Therefore, the functions are linearly independent on the in-
:..- propertJes are called upper triangular and lower triangular, respectively. terval t > O.
158 Chapter 4 Higher Order Equations 4.4 Solutions to Higher Order Linear Homogeneous Equations
8. S = {e -, cos 2t, e -, sio 2t, e" cos 5t, e" sin 5t} (e) Solve this ODE to fiod that W(n, Y2. ... ,Yn) =
Theorem 4.11 General Solution Ce
-Jp,(I)d,
.
Consider the nth-order linear homogeneous equationy(n\t) + Pn(t)y(n-I\t) + '" .

In ExerCIses 9-14, show. that S is a fundamental set of solu-
tions for the given equatIOn. Using traditional pencil and paper to compute. the Wronskian
+ PI (t)y(t) = 0, where Pn(t), ... ,PI(t) are continuous functions on an open in- of a set of functions quickly becomes ted,ous and tIme-
terval I. Suppose that YI (t), Y2(t), ... ,Yn(t) are linearly independent solutions of 9. S = {e-', e", te"}; y'" - 5y" + 3y' + 9y = 0 consuming. Fommately, computer algebra systems ar~ capable
this ODE on 1. Then, ifY is any solution of this ODE, there are constants CI, C2, 10. S = {e-', e- 2', e- 4'}; y'" + 7y" + 14y' + 8y = 0 of computing derivatives and evaluating the determmant en-
... , Cn so that yet) = CIYI(t) + C2Y2(t) + ... + cny.(t) for all ton 1. countered when calculating the Wronskian of a set of functIOns.
*11. S = {e', e- 51 sio t, e- 5' cos t};
y'" + 9y"+ 16y' - 26y = 0 Moreover, the graphing capabilities can be used to help deter-
mine whether the resulting expresslon IS or IS not zero. Co~-
12. S = {e 3t, e- Zt • eSt, teSt}; ute the Wronskian of each of the following sets S to claSSIfy
Theorem 4.11 tells us that if we have n linearly independent solutions of an nth-
order linear homogeneous equation, then we have all solutions, called a general solu-
/4) _ lIy'" + 29y" + 35y' - 150y = 0 ~ach set as lioearly independent or lioearty dependent. If ~ec
13. S = {e-", e-', e- 4 ' cos 3t, e- 4 ' sin 3t}; essary, graph the Wronskian to see if it is the zero functIOn.
tion, given by the linear combination of the solutions. In addition, we call {Y1 (t), Y2(t),
/4) + 12y'" + 60y" + 124y' + 75y = 0
... , Yn(t)} a fundamental set of solutions, and we note that a fundamental set of so- 16. S = {1 - 2 sin2 t, cos 2t}
lutions for y(nl(t) + Pn(t)y(n-Il(t) + ... + PI (t)y(t) = 0 must contain n linearly inde- 14. S = {e -, cos 2t, e -, sin 2t, e 21 cos 5t, e" sin 5t}; 2
17. S = (I, d(3t 2 - I), 4(5t' - 3t), k(35t
4
y(4) _ 2y'" + 26y" + 38y' + 145y = 0 - 30t + 3)}
pendent solutions. (In other words, the number of functions in the set equals the order
15. (Abel's Formula for Higber Order Equations) Con- 18. S = (sio t, sio 2t, sin 3t, sin 4t)
of the nth-order linear homogeneous ODE, and these functions must be linearly inde-
pendent.) . sider the nth-order linear homogeneous equatIOn 19. S = {e t , te t , t 2 e t , t 3 e l , (4e t }
inlet) + Pn(t)/n-I)(t) + .,. + PI(t)y(t) = 0 WIth so- 20. Is it possible to choose values of c" ... , not all
We omit the proof of this theorem because it follows the same steps as those used C7,
lutions YI(l), 12(t), ... ,Yn(t). We can show that the zero, so that
in the proof of the case for second-order equations discussed in Section 4.1. Wronskian of these n solutions satisfies the same Iden-
tity as that presented io Exercises 4.1. We do thIS for f(x) = c, + C2t- 5/2 + c,t -3/2 + C4t -1/2+ c, tll2+
Ii""",." the third-order ODE y(') + p,(t)y" + P2(t)y' +
PI(t)y(t) = 0, with solutions y" Y2, andy,.
C6 + C7 tS12
t3/2
is the zero function?

Determine a general solution of t 4 4) + 9t 3y'" l + Ilt 2y" - 4ty' + 4y = 0 using the (a) Show that 21. Use the symbolic manipulation capabilities of a com-
solutions given in Example 4. puter algebra system to help you prove the followmg
: .1 Solution
.
We showed in Example 4 that the four solutions YI(t) = t- 4 , Y2(t) =
t -1, Y3(t) = t, and Y4(t) = tint are linearly independent. Therefore, because we have
ft W(y" Y2, ... ,Yn) = \ ~i) ~~)
y, Y2
i \.
y,
theorem. Suppose that f(t) is a differentiable function
*
on an ioterval I and that f(t) 0 for all x m 1. (a)
Prove thatf(t) and tf(t) are lin<;arly iodependent on I.
four linearly independent solutions of a fourth-order linear homogeneous ODE, we (b) Use the ODE to solve lor ,
y,(') ,Y2(3) ,and y,(3) . Sub-
_ (b) Prove that f(t), tf(t), and t f(t) are lmearly mde-
can write a general solution as the linear combination stitute these values to obtam dWldt + p,(t)W - O. pendent on 1. (e) Generalize your result.
y(t) = CIYI(t) + C2Y2(t) + C'Y3(t) + C4Y4(t) = C1t-4 + C2t-I + C3t + C4t In t.

4 I
The set {t- , t- , t, tin t} is a fundamental set of solutions for the ODE.
Solutions to Higher Order linear Homogeneous

Equations with Constant Coefficients
EXERCISES 4.3
r: Distinct Real Roots r: Repeated Real Roots r: Complex Conjugate Roots
Suppose that we wish to solve

In Exercises 1-8, calculate the Wronskian of the indicated set 3. S = (e -', e", te"). anY (nl(t) + a n-l y(n-Il(t) + .,. + a,y'(t) + aoy(t) = 0,
of functions. Classify each set of functions as linearly inde-
4. S = (e', e- z" e- 4') a are real constants with an #- O. Following the same procedure
pendent or linearly dependent.
*5. S = {e t , e- St, sin t, e- 51 cos t} where an> a n - 1> ••• , 0 th t (t) = e" is a solutIOn, and we
we used to solve ay" + by' + cy = 0, we assume a Y I
*1. S = (3t 2, t, 2t - 2t 2)
substitute this function into t1;e ODE t~ determ!:'e ~(e) ~l~:~) ::vt~~t; l:a~3:~t.s;0~~
6. S = {e 3t, e -2/, eSt, teSt}
2. S = {cos 2t, sin t, I)
7. S = {e- 3r, e- t , e- 4l cos 3t, e- 4r sin 3t} tions. Taking derivatives, we fmd that Y (t) = re ,Y t - r , _
lowing this pattern, we see that the kth derivative is y<k)(t) = rke rt • Therefore, substi. -, and y"(t) = C2e' + C3e -'. Then, yeO) = c, + C2 + C3, y' (0) = C2 - C3, and
tution yields ~~{O) = C2 + C3, so we obtain the system of three equations and three unknowns:
rt
Qnr"e + an_Ir(n-l)e rt + ... + Qlre rt + aoe rt = 0 or Cl + C2 + C3 = 0
ert(anrn + an_l rn - l + ... + a,r + ao) = 0, {
C2 - C3 = -2
so the characteristic equation is Cz + C3 = 2
anY" + an_lyn-l + ... + aIr + ao = 0

Adding the last two equa~ons i~dicates that Cz = O. SU?st~tution of~is :;~e :t~
either of the last two equations gIves C3 = 2. Then, substitut.lOn of Cz . 0 .3
because e
r
, *- 0 for all values of t. into the first equation gives us c, = -2. Therefore, the umque solutlOn to th,S IVP
is yet) = -2 + (O)e' + (2)e-' = -2 + 2e-'.
Distinct Real Roots
Suppose that the characteristic equation has the roots 1'], 1'2, ... , rn and they are dis.
Repeated Real Roots
tinct (that is, no two are equal). Then we have the n solutions h(t) = e r", Y2(t) = er",
r
... , Yn(t) = e "'. Later in Exercise 39, we show that these solutions are linearly inde. When solving second-order equations in Section 4.2, we encountered equations in whi~h
pendent. Therefore, we obtain a general solution by taking the linear combination of the root of the characteristic equation was repeated. For e~ample, the. charactenstic
y + Y = 0 is 1'2 - 21' + 1 = 0 or (I' - 1) = O. In this
the n solutions. .
equatlOn ,or y " - 2'
c
. case,
b we say
(t)-
that the root I' = I has multiplicity two, and we found a general solutlOn to e?,. .-
c,e' + C2te'. As we increase the order of the ODE, we can have roots of multiphclty
Theorem 4.12 Distinct Real Roots greater than two, so we present the following theorem.
If the roots 1'], 1'2, ... , rn of anrn + an_,r n-' + ... + a,r + ao = 0 are real and
distinct, then a general solution of any(n)(t) + an_,y(n-I)(t) + '" + aly'(t) + Theorem 4.13 Repeated Real Roots
aoy(t) = 0 is yet) = c,e r" + C2er" + .... + Cne r.,!.
n 1
Su pose that the root I' = R of the characteristic equation anrn ": an_lr - + ...
+ ~lr i~,
+ ao = 0 has mUltiplicity k (that (I' - Rl,S a fa~tor
m the Ch~ct:
. ti'
Iscequa. tl'on) Then, the k linearly mdependent .. -Rof anY (t)
solutions
,'f',,·,'.' an_ly<n-I)(t) + ... + aly'(t) + aoy(t) = 0 assocIated With I' -
eRt, [eRe, t2eRt, . . . , tk-le
Rt
.
are
Solve y'" - y' = O.

~======.J
: J Solution In this case, the characteristic equation is 1'3 - I' = O. Factoring, we
For example, applying this theorem to a second-order equation w~th k = 2 and
have 1'(1'2 - 1) = 1'(1' - 1)(1' + 1) = O. The three distinct roots are I' = 0, I' = I, and R = -1, we obtain yet) = Cle-' + czle-' = (Cl + c2t)e-' as we found m Example 2
t
I' = -1 with corresponding solutions eO = I, e , and e-', respectively. Therefore, a
in Section 4.2.
general solution is the linear combination of these functions, yet) = Cl + C2 e' + C3 e -to
"i'"',,., Solve 2y(6) - 7y(5) - 4y<4) = o.

Solve y'" - y' = 0, yeO) = 0, y'(O) = -2, y"(O) = 2.
: I Solution The characteristic equation is 21'6 - 71'5 - 41'4 = 0, or
, I Solution (Notice that an initial-value problem involving a third-order ODE reo
1'4(21'2 - 71' - 4) = 1'4(21' + 1)(1' - 4) = O.
quires three initial conditions.) Typically, we solve an initial-value problem by first
finding a general solution. However, we found this in the previous example to be Therefore, the roots are I' = 0 (of multiplicity k = 4), I' = -l/~ (of multiplicity
yet) = Cl + C2 e ' + C3 e -'. Next, we apply the initial conditions where y' (I) = C2 e ' _ k = 1), and I' = 4 (of multiplicity k = I). Then, the four lmearly mdependent solu-
tions corresponding to r = 0 are e O.t, te o.t, t 2e O·t, and t 3e O.t, which are simplified to
I, t, t 2, and t 3 •
ondin
:~.t cos 7= cos I and eo· t
to the complex conjugate pair r = :'::.i (where a = 0 and f3 = 1) ~re
sin I = sin t (see Table 4.2). Therefore, ~ gene:al solutlOn
Notice that the nwnber of solutions associated with r = 0 equals the multi_ is the linear combination of these four linearly independent functlOns given by
plicity of r = O. The solution associated with r = -112 is e -';2 and that associated
4t yet) = c,e t + C2e-' + C3 cos t + C4 sin t.
with r = 4 is e (Table 4.1). We find a general solution of2y(6) - 7/ S ) - 4/4 ) = 0
by
3 taking
tI2 the linear combination of the six (linearly independent) solutions: I, t, t2,
t , e- , e 4t . This solution is TABLE 4.2 Linearly Independent Solutions of
y(4) _ Y = 0
y(t) = c, + c2t + C3t2 + c4 3 + cse -t12 + C6e4t.
Roots Multiplicity Corresponding Solution(s)
TABLE 4.1 Linearly Independent Solutions of r = 1 k= 1 y= e l
2y(6) - 7y(S) - 4y(4) = 0
r= -1 k=I Y = e- r
Root Multiplicity Corresponding Solution(s)
r= ±i k= 1, k= 1 Y = cos t, Y = sin t
r=O
I k=4 Y = 1, y = I, Y = 12, Y = 13
r= -1/2 k= 1 y=e- tI2
r=4 Fi re 4.5 shows Ihe graph of y = c,e t + C2e -, + C3 ~ os 1+ c4 _sin I _for various
_. val-
k= 1 Y = e 4t
OJ CJ, Cz, C3, and c4· Idenlzify those graphs for whzch (a) c, - C3 - C4 - 0, (b) c,
gu.r
ues
(- = C2 = 0; and (c) C3 = C4 = O.
tI2 4tthat in Example 3, a fundamental set of solutions for the ODE is {l, t, t ,
Note 2
3
1 e- , e }. The nwnber of solutions in the set equals the order of the ODE. There-
,
fore, 10 form a general solulion of an nth-order linear homogeneous ODE, we lake the 20
linear combination of Ihe n jUnctions in Ihe fundamental sel.
15
Complex Conjugate Roots
10
Theorem 4.14 Complex Conjugate Roots

0.5
Suppose that 1', = a + f3i and r2 = a - f3i (a and f3 > 0 are real) are roots of
the characteristic equation anr" + an_,r n-' + ... + aIr + aD = O. Then, two lin- 3x
-1 4 x -3 -2 -1
early independent solutions associated with this pair of roots are
(b) (e)
(a)
e'" cos f3t y
and eat sin f31. 20
-1 4 x
10
I'E'"I,'.'
Solve y(4) - Y = O.
-3 -2 3x
-10
: ) Solution In this case, the characteristic equation is
-20
4
r - I = (r2 - 1)(1'2 + I) = (I' - 1)(1' + 1)(r2 + I) = 0,
(e) (f)
so the roots are r = I, r = -I, r = i, and r = -i. The solutions corresponding to Cd)
the roots r = I and r =-I t
are e and e -t, respectively, while the solutions corre- Figure 4.5 (a)-(f)
We state the following rules for finding a general solution of an nth-order lin y(t) = e 2t(cI cos 3t + Cz sin 3t + C3t cos 3t + c4t sin 3t)
h '. ~
omogeneous equatJ.on WIth constant coefficients because of the many situations tha + cse-St + e 2t (c6 + C7t + C8t2).
are encountered. t
c
Find an eighth-order homogeneous linear ODE with constant coefficients that has
Determining a General Solution of a Higher Order Equation general solution
y(t) = e 2t(cl cos 3t + C2 sin 3t + C3t cos 3t + C4t sin 3t) +
CD Let r be a real root of the characteristic equation cse-St + e 2t(c6 + C7t + C8t2).
anrn + an_Ir n- 1 + ... + aIr + ao = 0
of an nth-order homogeneous linear differential equation with real constant co-

efficients. Then, ert is the solution associated with the root r.
"ii"I,'.,
(a) If a is a positive constant, find conditions on the constant b so that y(t)
If r is a root of mUltiplicity k where k ;;,: 2 of the characteristic equation,
then the k solutions associated with r are satisfies
ert, ter(, (2e rt , •.• , tk-1e rt• 5.69889y '" + 5.21742y" - 3.39914y' - 5.6932y = 0
{ y(O) = 0, y'(O) = a, y"(O) = b
@ Supp.ose that r .and r represent the' complex conjugate pair a :!: {3i. Then the two y
solutIOns assocIated with these two roots are and limH~ y(t) = O. (b) For this function, find and classify the first critical point
0.8 on the interval [0, +co).
e'" cos {3t and eat sin {3t.
If the values a :!: {3i are each a root of multiplicity k of the characteristic : 1 Solution A general solution of the equation
equation, then the other solutions associated with this pair are
5.69889y'" + 5.21742y" - 3.39914y' - 5.69232y = 0
te'" cos {3t, te'" sin {3t, t 2e'" cos {3t, tZe at sin {3t, ... ,
k
t - 1e at COS {3t, t k- 1e at sin (3t. is
5 x
. . A general solution to the nth-order differential equation is the linear com- y(x) = e-O.91693t(CI cos 0.496889t + C2 sin 0.496889t) + c3eO.91834St,
bmatlOn of the solutions obtained for all values of r.
Note that if r." r2, ... , rj are the roots of the equation of multiplicity k"
Figure 4.6 Graphofly'(C) = while the solution to the initial-value problem is
a
kz, ... , '9, respecllvely, then kl + k2 + ... + k· = n, where n is the order of the e-O.91693t(cos 0.496889t -
y(t) = e-O.9I693t[(-0.507273a - 0.276615b) cos 0.496889t - (-0.138893a +
differential equation. 1 1.84534 sin 0.496889t)
1.02168b) sin 0.496889t] + (0.507273a + 0.276615b)eO.91834St.
,'ii".".,
Determine a general solution of the eighth-order homogeneous linear ODE with
We see that limt->~ y(t) = 0 if 0.507273a + 0.276615b = 0, which leads to
b = - 1.83386a.
To find and classify the first critical point of y(t), we compute
?O?stant c?efficients if the roots and corresponding multiplicities of the character- y'(t) = ae-O.91693t(cos 0.496889{- 1.84534 sin 0.496889t)
ISltC equatIOn are r = 2 - 3i k = 2' r = 2 + 3i k = 2' r = -5 k - I' and r - 2
k = 3. " ' " -, - ,
and graph e -0.91693t(cos 0.496889t - 1.84534 sin 0.496889t) in Figure 4.6 to locate
the fITst zero of y' .
: .) Solution The solutions that correspond to the complex conjugate pair 2 :!: 3i are From the graph, we see that y' (t) = 0 near I and, numerically, we obtain the
2t 2t
y = e cos 3t and y = e sin 3t. These roots are repeated so the two other solutions critical number t = 0.999432. At this critical number y(0.999432) = 0.383497a, so
8 x
corresponding to the pair are y = te 2' cos 3t and y = te 2t sin 3t. For the single root by the first derivative test, (0.999432, 0.383497a) is a local maximum: To see ~t
r = - 5: the corresponding solution isy = e- 5t • Finally, the root of multiplicity three Figure 4_7 Graph ofy(t) for
(0.999432, 0.383497a) is the absolute maximum, we graph y(t) for vanous chOIces
2t
r = ~ Yle.lds the three solutions y = e , y = te 2t, and y = tZe Zt . Therefore, a general various choices of a of a in Figure 4.7.
solutIOn IS L
166 Chapter 4 Higher Order Equations 4.4 solutions to Higher Order Linear Homogeneous Equations
167
with Constant Coefficients
two distinct roots, rl of multiplicity one andr2 ofmul-

EXERCISES 4.4
-
40. (Operator Notation) The nth-order derivative of a
tiplicity k = n - I. Then, the characteristic equation
function y is given in operator notation by D"y = can be written as (r - rl)(r - r2j" = O. Similarly,
d"yldt". Then, the left side of the nth-order linear
the ODE can be written in differential operator nota-
homogeneous ODE with constant coefficients,
tion as (D - rl )(D - rdy = o. Therefore, every
In Exercises I - I 0, given the list of roots and mUltiplicities of 26. y(4) - 5y'" - lOy" + 80y' - 96y = 0 a,,/")(t) + a"_I/,,-I)(t) + .. , solution of (D - r,j"y = 0 is also a solution of
the characteristic equation, fonn a general solution. What is the
27. /4) + 7y'" + 6y" - 32y' - 32y = 0
+ aly'(t) + aDy(t) = 0, =
(D - rl)(D - r,j"y 0 (see Exercise 42). To find
order of the corresponding ODE? solutions of (D - r2j"y = 0, assume that y = v(l)e"".
28. y(4) + y'" + lOy" - 52y' + 40y = 0 can be expressed as
(a) Use the product rule to show that (D - r,)(v(l)e'2~
1. r = -2, k = 3; r = 2, k = I
*29. y(4) + 2y'" - 2y" + 8y = 0 + a,,_,D"-'y + ... + alDy + aDY
a,.D"y or = (Dv)e'" so that, by induction, (D - r2)k(v(t)e'2') =
2. r= l,k=2;r= -1O,k=2 (a,.D" + a,,_ID,,-1 + ... + aID + aD)y, (Dkv)e'" for any function vet). (b) Therefore, a solu-
30. y(4) + 32y" + 256y = 0
*3. r = 0, k = 2; r = 3i, k = I; r = - 3i, k = I tion of (D - r2j"y = 0 must satisfy (Dkv)e'" = 0
wherep(D) = a"D" + a,,_ID"-1 + .. , + aID + aD is
4. r = 4, k = 3; r = I + i, k = I; r = I - i, k = I In E~ercises 31-38, solve the initial-value problem. Graph the or Dkv = O. Solve D"v = 0 (or v(k) = 0) to find
called an nth-order linear differential operator with
solution on an appropriate interval. vet) and to conclude that the portion of the solu-
5. r= -3 +4i, k=2; r= -3 - 4i k= 2' r=-5 constant coefficients. Therefore, the ODE can be writ-
k=l;r=-J/3,k=1 " , 31. y'" - 2y" = O,y(O) = l,y'(O) = 2,y"(0) = 0 ten as (a"D" + an_ID"-1 + .. , + aID + aD)y = O.
tion corresponding to r = r2 is y = v(t)e'" =
(CI + C21 + ... + ci/k-I )e"'.
6. r = i/2, k= 3; r = -i/2, k= 3; r = -I, k= 2 32. y'" + 49y' = 0, yeO) = -I, y'(O) = 0, y"(O) = -2 For example, we can write y" + 2y' - 8y = 0 as
(d - 2D - 8)y = 0 or in either of the forms
7. Can {I, t, t 2, 2t - 8) be a fundamental set of solu· *33. y'" - y = 0, yeO) = 0, y'(O) = 0, y"(O) = 3
(D + 2)(D - 4)y = 0 or (D - 4)(D + 2)y = O. Write Computer algebra systems have built-in functions that can be
tions for a fourth·order linear ODE? 34. y(4) - 16y = 0, yeO) = 0, y'(O) = -8, y"(O) = 0 each of the following equations in differential opera- used to find exact solutions of nth-order linear homogeneous
8. Can (e-', e', cos t, sin t, cos 2t) be a fundamental set y"'(O) = 0 ' differential equations with constant coefficients as long as n is
tor notation.
of solutions for a fifth-order linear ODE? 35. y(4) - 8y" + 16y = 0, yeO) = 0, y'(O) = 0, smaller than 5. In cases when the roots of the characteristic
(a) y" + y' + 7y = 0
equation are symbolically complicated, when approximations
*9. Can (e", e'l4, e-' cos I, e-' sin t) be a fundamental y"(O) = 8, y"'(O) = 0
(h) y" + 16y = 0 are desired, or when n is greater than 4, the roots of the char-
set of solutions for a fifth·order linear ODE? 36. /5) - 4y'" = O,y(O) = O,y'(O) = O,y"(O) = 0,
(e) y'" + 8y' - y = 0 acteristic equation can be approximated by taking advantage of
fl4 y'"(O) = 0, /4)(0) = 32
10. Can (e 2t , e , e-', le-', I'e-') be a fundamental set the numerical approximation capabilities of the system being
of solutions for a fourth-order linear ODE? *37. /5) + 8y(4) = 0, yeO) = 8, y'(O) = 4, y"(O) = 0,
(d) 16/4 ) +y = 0
41. Consider the linear differential operator p(D) = used.
y"'(O) = 48, y(4)(0) = 0
(2D - I). Wben we apply p(D) to the functionf(l) = 44. Find a general solution of
In Exercises II-30, find a general solution of each equation. 38. /6) -3y(4) + 3y" - y = 0, yeO) = 16, y'(O) = 0, 12, we obtain (2D - I)(x') = 2D(x') - x =
2
(A computer algebra system may be useful in solving some of y"(O) = 0, y"'(O) = 0, y(4)(0) = 0, /5)(0) = 0
2(2x) - x 2 = 4x - x 2. Apply the given differential
(a) y'" + 2y" + 5y' - 26y = 0
the equations.)
39. Suppose that the roots rIo r" and r3 of the character· (b) 0.9y'" + 18.78y' - 0.2987y = 0
+ 25y' =
11. y'" - lOy" ° istic equation of a third-order linear homogeneous
operator to the given function.
= 3D,f(t) = I 8.9/4 ) - 2.5y" + 32.0y' + 0.773y =
+ y" =
12. 8y'" ° ODE with constant coefficients are real and distinc~
(a) p(D) (e) O.
*13. y'" + 7y" + 17y' + 15y = ° and,. ~onsider a general solution y = Cl e r1t + c2e"lt +
r1t
(h) p(D) = (I - D),f(I) = t Graph the solution for various initial conditions.
14. y'" + 7y" + 24y' - 32y = ° C3e 3. Show that the functions e , el).t, and ef)t
are linearly independent using the following steps:

(c) p(D) = D2,f(I) = 12
(d) p(D) = (D' + 1),J(t) = cos I
45. Solve each of the following initial-value problems.
Verify that yoU! result satisfies the initial conditions
15. y'" - 8y" + 5y' + 14y = 0 (a) Assume that the functions are linearly dependent. by graphing it on an appropriate interval.
42. The linear differential operator p(D) is said to anni-
16. y'" - 6y" - 7y' + 60y = 0 Then, there are constants c" C2, and C3 (not all zero)
such that. cle'" + c,e'" + C3e'" = O. (h) Multiply
hilate a functionf(l) ifp(D)[f(t)] = 0 for a1l t. In fact, f y'" + 3y" + 2y' + 6y = 0
*17. y'" + y" - 16y' + 20y = 0 this ~quation bye-Tit to obtain CI + cze('"2- r L)t + if PI = PI(D) annihilatesf(x) and if P2 = P2(D) is an- (a) "\YeO) = O,y'(O) = l,y"(O) =-1
18. y'" + 12y" + 36y' + 32y = 0 other linear differential operator, then P2P I = PIP, y(4) _ 8y'" + 30y" - 56y' + 49y = 0
C3e(" ")' = o. (c) Differentiate this equation with
annihilates f(t). For example, (D - I) annihilates e' = =
19. y'" + 11y" + 24y' - 36y = 0 respect to t to obtain (r2 - rl)c,e("-")' + (h) { yeO) = l,y'(O) 2,y"(0) -1,y"'(O) = -I
because (D - I)(e') = e' - e' = O. Therefore, if we
C3(r3 - rl)e("-")' = O. (d) Multiply this equation by
20. y'" + 2y" - 15y' - 36y = 0 apply another linear differential operator such as D, O.3IY '" + 11.2y" - 9.8y' + 5.3y = 0
e-v:-,,), and differentiate the resulting equation to
*21. /4) - 9y" = 0 obtam c3(r3 - rl)(r, - rl)e("-")' = O. This indicates
we have D(D - I)(e') = D(e' - e') = D(O) = O. (e) { yeO) = -J,y'(O) = -1,y"(O) = 0
22. 8y(4) + y' = 0 Show that (D - I)D(e') = O. In terms of ODEs, this 46. Use a computer algebra system to fmd a general so-
that C3 = O. (e) Follow similar steps to show that CI =
means that a solution to (D - I)y = y' - y = 0 is also lution of
23. y(4) - 16y = 0 C2 = 0, which contradicts the assumption. (f) How
a solution to D(D - I)y = y" - y' = O.
could you apply a similar argument to show that the
24. 9y(4) + 4y'" = 0
43. (Repeated Roots) Suppose that the characteristic
~ctions e"/', er'].t, . . . , er"t are linearly independent
*25. /4) - 6y'" - y" + 54y' - 72y = 0 If the n roots rb r2, ... , rn are real and distinct? equation anrn + an_lr n- 1 + '" + air + ao = 0 has
4.5 Introduction to Solving Nonhomogeneous Equations with Constant
Coefficients: Method of Undetermined Coefficients
Show that the result you obtain is equivalent to *47. Show that if Yo '" 0 the problem liii,,','.'
Y = Ae-
xJk
+ B~k + C sin::£k + D cos::£ 4.02063Y '" - 0.224975y" + 4.486y' - 2.48493y = 0
k' { . Verify that yp(t) = ~ sin t is a particular solution of y" - 4y = - 3 sin t.
y(O) = O,y'(O) =yo,y(4) = 0
has (a) neither local maxima nor local minima; (b) ex-
actly one local maximum; and (c) exactly one local has a unique solution. : 1 Solution First we compute y;,ct) = ~ cos t and y;(t) = -~ sin t. Substituting into
minimum on the interval [0, co). 48. Complete the following table. y" - 4y results in
Roots of I y; - 4yp = -53 sm

.
t-
3. 3 .
4 . "5 sm t = - sm t.
Differential Characteristic Characteristic
Equation Equation Equation I General Solution We conclude that y (t) = ~ sin t is a particular solution of y" - 4y = - 3 sin t be-
yen) = 0 cause yp satisfies the equation y" - 4y = - 3 sin t and contains no arbitrary con-
stants.
(r - k)"
<-
rI,2 = a ± i{3, Show that yp(t) = !e -21(15 - 10e4' + 3e 21 sin t) is also a particular solution of
y" - 4y = - 3 sin t.
13*0
In this example, we see that yp(t) = ~ sin t is a particular solution of y" ~ 4y ~
y = ea'[(c,., + c,.,1 + ... C"n_,l n,-') cos f31
+ (C2.' + C'.2t + ···C2.n_,t n- ) sin f31] _ 3 sin t. The corresponding homogeneous equation of y" - 4y = - 3 sm t IS
y" _ 4y = 0 with general solution Yh(t) = cle- 2' + C2 e2'. Let y(t) = Yh(t) + Yit).
Then,
y" - 4y = (Yh(t) + yp(t»)" - 4(Yh(t) + Yit))
Introduction to Solving Nonhomogeneous Equations = y'j, + y; - 4Yh - 4yp
with Constant Coefficients: Method of Undetermined
Coefficients = y'j, - 4Yh + Y; - 4yp = -3 sin t.
~~
o -3sint
C" General Solution of a Nonhomogeneous Equation ("" Method of
Undetermined Coefficients We see thaty(t) = Yh(t) + yp(t) is a solution of the n0n?0mogeneo~s equationy" ~ 4y
= _ 3 sin t. In fact, it is not hard to show that if y(t) IS any solutJOn to the equatiOn,
General Solution of a Nonhomogeneous Equation then there are constants Cl and C2 SO that y(t) = Yh(t) + yp(t) (see Exercise 65). .
More generally, if yP(t) is a particular solution of the nonhomogeneous equatJOn
Suppose that we encounter the homogeneous equation y" + Y = O. In previous sec-
tions, we learned that this equation has the general solution yet) = Cl cos t + C2 sin t. any<n)(t) + an_ly<n-l)(t) + .,. + aly'(t) + aoy(t) = g(t)
Now we investigate ways to solve nonhomogeneous equations. For example, consider
the equation y" + Y = 2e -I. Notice that if we substitute the function y(t) = e - I with and
derivativesy'(t) = -e- I andy"(t) = e- I into this equation, we obtain Yh(t) = cd/t) + C2I2(t) + ... + cnfn(t)
y" +Y = e- I + e- I = 2e-'. is a general solution of the corresponding homogeneous equation
Therefore, yet) = e - I satisfies the nonhomogeneous equation y" + Y = 2e -I. In fact, any<n)(t) + an_ly(n-l)(t) + ... + aly'(t) + aoy(t) = 0,
this solution is called a particular solution of the nonhomogeneoUs equation.
every solution, y (t), of the nonhomogeneous equation can be written in the form
Definition 4.6 Particular Solution y(t) = Yh(t) + yp(t),

A particular solution, yp(t), of the nonhomogeneous differential equation for some choice of c" C2, ... , Cn (see Exercise 65).
any(n)(t) + an_ly(n-l)(t) + ... + aly'(t) + aoy(t) = get) is a specific function To prove this theorem for second-order equations, consider
that contains no arbitrary constants and satisfies the differential equation.
a2(t)y"(t) + al (t)y' (t) + ao(t)y(t) = g (t)
170 Chapter 4 Higher Order Equations 4.5 Introduction to Solving Nonhomogeneous Equations with Constant
Coefficients: Method of Undetermined Coefficients 171
and asswne that {fi(t),fi(t)} is a fundamental set of solutions for the corresponding
:I Solution We first show that yp(l) = ~t3 - ~t2 + ~t is a particular solution of
homogeneous equation y" + 2y' + 2y = t 3. Calculating y~ = ~t2 - 3t + ~, y; = 3t - 3, and y; + 2y~ +
2yp gives US
az(t)y"(t) + a, (I)y' (I) + ao(t)y(t) = O.
If both yet) and yit) are solutions to y; + 2y~ + 2yp = (3t - 3) + 2(lt2 - 3t + l) + 2( 3
t t -ltZ + It)
az(t)y"(t) + a, (t)y' (I) + ao(t)y(t) = g (t), = 3t - 3 + + 3 + t 3 - 3t 2 + 31 = f3.
3t 2 - 6t
3
let Yh(t) = yet) - yit) so that substitution into the nonhomogeneous equation yields The corresponding homogeneous equation of y" + 2y' + 2y = t is y" + 2y' +
2y = 0 with characteristic equation r2 + 2r + 2 = 0 and roots r = - I ± i. Thus, a
az(t)y;;(t) + a, (t)yl,(t) + aO(t)Yh(t)
general solution of y" + 2y' + 2y = 0 is Yh(t) = e -t(c, cos t + C2 sin f), so that
= az(t)[y"(t) - y;;Ct)] + a,(t)[y'(t) - YAt)] + ao(t)[y(t) - y~(t)]
y(t) = Yh(t) + yp(t)
= [a2(t)y"(t) + a,(t)y'(t) + ao(t)y(t)] - [az(t)y;(t) + a,(t)y~(t) + ao(t)y~(t)J
= get) - get) = O. = e-t(c, cos t + C2 sin t) + t t 3 -ltZ + It
Thus, Yh(t) = yet) - yp(t) is a solution to the homogeneous equation az(t)y"(t) + is a general solution ofy" + 2y' + 2y = t
3
a, (t)y' (t) + ao(t)y(t) = O. There are constants c, and C2 so thaty,,(t) = c,fi(t) + cz!i(t) '~
which indicates that ' 3
Ify,(t) and Y2(t) are nontrivial solutions ofy" + 2y' + 2y = t , is y,(t) + Y2(t) also
yet) - yP(t) = c,fi (t) + czfi(t) or y(t) = ,c,fi (I) + C2!i(l), + yp(t). a solution?
'----.,------'
Yh(t) y,,(t)
Method of Undetermined Coefficients
This means that every solution to the nonhomogeneous equation can be written as the
swn of a general solution to the corresponding homogeneous equation and a particular Now that we understand how to form a general solution to a nonhomogeneous equa-
equation of the nonhomogeneous equation. This leads us to the following definition. tion, we present a method, called the Method of Undetemlined Coefficients, that can
be used to find a particular solution in some cases. Consider the nonhomogeneous
second-order linear differential equation
Definition 4.7 General Solution of a Nonhomogeneous Equation y" + 4y' + 3y = e t.
A general solution to the linear nonhomogeneous nth-order differential equa- A general solution to this nonhomogeneous equation is y(t) = Yh(t) + yp(t), where
tion Yh(t) is a solution of the corresponding homogeneous equation y" + 4y' + 3y = O.
This equation has characteristic equation r2 + 4r + 3 = (r + 1)(r + 3) = 0 so Yh(t) =
cle-
t
+ C2e - 3t•
is Now we must select the proper form of a particular solution yp(t). Because g(t)
= e', we are safe to assume that a particular solution is a multiple of e'. Notice that
yet) = Yh(t) + Yp(t),
the function we choose for yp(t) must satisfy y" + 4y' + 3y = e', so we need to select
where Yh(t) is a general solution of the corresponding homogeneous equation a function that includes e' in the function and its derivatives. We let yp(t) = Ae ' and at-
an(t)/n)(t) + an_,(t)y(n-'\t) + ... + a,(t)y'(t) + ao(t)y(t) = 0, tempt to find A. Substituting this function into y" + 4y' + 3y = e ' yields
and yp(t) is a particular solution to the nonhomogeneous equation. y; + 4y~ + 3yp = Ae' + 4Ae' + 3Ae' = 8Ae' = e' .
Equating the coefficients of e ' gives us 8A = 1 or A = i. Therefore, yp(t) = ie', and a
,'ii,,','.,
general solution is
_, -3, 1 ,
Y ( f) = Yh ( t) + yp (t) = c,e + C2 e + "8 e.
Let yp(t) = tt 3 - ~t2 + ~t and Yh(t) = e-t(c, cos 1 + C2 sin t). Show that y(t) ~
y,,(f) + yp(f) is a general solution of y" + 2y' + 2y = t 3 Find a particular solution of y" + 4y = sin t of the form yp(t) = A cos t +B sin t.
-
Now consider the equation and yP(t) is a particular solution involving no arbitrary constants of the nonhomoge-
y" + 4y' + 3y = e- 3t . neous equation
We just saw that the solution of the corresponding homogeneous equation any(n)(t) + an_Iy(n-I)(t) + ... + alY '(t) + aoy(t) = get).
y" + 4y' + 3y = 0 We learned how to solve h~mogeneous equations in Sections 4.2 and 4.4, so we
t must learn how to find the form of a particular solution to solve nonhomogeneous
iSYh(t) = cle- Ae- 3t as we did in the
+ C2e-3t. Therefore, if we assume thatyp(t) = equations with the method of undetermined coefficients.
previous example, we have the derivativesy~(t) = -3Ae- 3t andy;(t) = 9Ae- 3t. Sub-
stitution into y" + 4y' + 3y = e- 3t then yields
y" + 4y' + 3y = 9Ae- 3t - 12Ae- 3t + 3Ae- 3t = 0 "* e- 3t Outline of the Method of Undetermined Coefficients to Solve
which does not lead to determining the value of A. Of course, this should not surprise an/nJet) + an_l/n-I)(t) + '" + aly'(I) + aoy(t) = g(l),
us because e -3t is a solution of the corresponding homogeneous equation. Therefore,
we cannot include this function in the particular solution. However, we can obtain a
function that resembles e -3t and includes e -3t in yp(t), Y ~(t), and y;(I), by multiplying where g (t) is a linear combination of the functions
3t
e- by I. We let yp(t) = Ate- 3t . The derivatives of this function are
1, t, t 2 , . . . , e'd, te kt , t 2 e kt, ... ,eat cos {3t, teat cos {3t, t 2 e at cos {3t, ... ,
y~(t) = Ae- 3t - 3Ate- 3t eO" sin {31, teat sin {31, t 2e at sin (3t, ....
Substitution into y" + 4y' + 3y = e -3t gives us ---------------------------
(j) Solve the corresponding homogeneous equation for Yh(t).
y; + 4y~ + 3yp = -6Ae- 3t + 9Ale- 3t + 4(Ae- 3t - 3Ate- 3t ) + 3Ate- 3t @ Determine the form of a particular solution yp(t) (see Determining the Form of
= -2Ae- 3r = e- 3t . Yp(t)).
o Determine the unknown coefficients in yAI) by substituting yp(t) into the nonho-
Equating the coefficients of e -3t then yields -2A = I or A = -1. Hence, yp(t) = mogeneous equation and equating the coefficients of like terms.
-1te -3t and a general solution is o Form a general solution withy(t) = Yh(t) + yp(t).
yet) = Yh(t) + yp(t) = cie- t + C2e - 3t - t te- 3t .
Show that it is not possible to find a particular solution of y" + 4y = sin 2t of the
Determining the Form of Yp(t) (step 2):
form yp(t) = A cos 2t + B sin 2t, but that it is possible to find a particular solution Suppose that g(t) = blgl(t) + bzg2 (t) + ... + bJgJCt) where bl> b 2, ... , bj are con-
of the form yp(t) = t(A cos 2t + B sin 21}. stants and each g;(I), i = 1,2, ... ,j is a function of the form 1m, Ime kt , Ime at cos {3t,
or 1mear sin {31.
These two examples illustrate the guesswork involved in choosing the form of a
(A) If g, (I) = 1m, the associated set of functions is
particular solution. To eliminate some of the guessing requITed to apply this method,
we introduce an algorithm that can be used to determine the form of yp(t). S = {1m, 1m-I, ... , 12, I, I}.
Consider the linear nonhomogeneous nth-order differential equation with con-
stant coefficients (B) If g,(t) = Ime kt , the associated set of functions is
S = {tme kt, Im- Iekt, ... , 12 e kt, lekt, e kt }.
an/n)(t) + an_Iy(n-I)(t) + ... + alY' (t) + aoy(t) = g (t)
(C) If gi(t) = Ime a, cos (3t, or gi(t) = tme a' sin {31, the associated set of functions is
where an "*,0 and g(t) is a linear combination of the functions I, t, t 2, ... , e kt, te kt,
2 kt
t e , ... , eat cos {3t, teat cos {3t, t 2 e at cos {3t, ... , eat sin {3t, teat sin (3t, t 2 e at sin f3t, S = {Ime a' cos {31, 1m-Ie"" cos {31, ... , 12 e a , cos {31, Ie"" cos {31, e",r cos {31,
. . . . A general solution of this differential equation is Ime a' sin {31, Im- Ie ",t sin {31, ... , 12 e"" sin {3t, lear sin (3t, e"" sin {3t} .
y(t) = Yh(t) + Yp(t), For each function in g (I), determine the associated set of functions S. If any of the
where Yh(t) is a solution of the corresponding homogeneous equation functions in S appears in the homogeneous solution Yh(t), multiply each function in S
by I' to obtain a new set S', where r is the smallest positive integer so that each func-
any(n)(t) + an_Iy(n-I)(t) + ... + aly'(t) + aoy(t) = 0, tion in S' is not a function in Yh(I).
. The correct form of a particular solution is obtained by taking the linear combi. Substitution of the derivatives y ~(t) = 2AI + B and y;(t) = 2A into
natIon of all functions in the associated sets where repeated functions should appear y" + 5y' + 6y = 212 + 3t yields
only once m the form of the particular solution. y; + 5y~ + 6yp = 6At 2 + (lOA + 6B)1 + (2A + 5B + 6C) = 212 + 3t
Therefore, 6A = 2, lOA + 6B = 3, and 2A + 5B + 6C = O. This system of equa-
tions has the solution A = L B = -t.,
and C = -Ji" so
I? I 7
For each of the following functions, determine the associated set of functions: (a) yp(t) = 3"1" -18 t - 108'
f(t) = 14, (b)f(t) = 13e-2t, and (c)f(t) = 12 e- ' cos 4t.
~ .. Solution -2' -31 1 2 1 7
(a) Using (A), we have S = {1 4 , 13 , 12, t, I}. yet) = Yh(l) + yp(l) = cle + C2e + 3"t -18 1 - 108'
(b) In this case we use (B): 2
(c) In this case, we see that g(l) = 3e- 21 . Then the associated set is S = {e- '}.
S = {t 3e- 2t, 12e- 2t, le- 2t, e- 2t }. However, because e -2, is a solution to the corresponding homogeneous equation,
we must multiply this function by I' so that it is no longer a solution. We mUltiply
(c) According to (C),
the element of S by II = t to obtain S' = {te -2,} because te -2' is not a solution of
S = {t2e-' cos 4t, le- 1 cos 41, e- 1 cos 41, t 2e- ' sin 4t, le- 1 sin 41, e- 1 sin 41}. y" + 5y' + 6y = O. Hence, yp(l) = Ale -2'. Differentiating yit) twice and substitut-
ing into the equation yields:
2
."E''''''.'
Solve the nonhomogeneous equations (a) y" + 5y' + 6y = 2e' , (b) y" + 5y' +
y; + 5y~ + 6yp = -4Ae- 2' + 4Ale- 2 ' + 5(Ae- 21 - 2Ale- 2') + 6Ate- ,
= Ae- 2' = 3e- 2 '.
2t
6y = 212 + 31, (c) y" + 5y' + 6y = 3e- 2', and (d) y" + 5y' + 6y = 4 cos I. Thus, A = 3 so yp(t) = 3Ie- 21 , and a general solution ofy" + 5y' + 6y = 3e- is
3
y(l) = Yh(t) + yP(t) = c,e- 2t + c ze- ' + 31- '.
2
1 Solution (a) The corresponding homogeneous equation y" + 5y' + 6y = 0 has

general solution Yh(l) = cje- 21 + C2e -3,. (Why?) Next, we determine the form of (d) In this case, we see that get) = 4 cos t, so the associated set is S = {cos I, sin I}.
yp(I). Because g (t) = 2e', we choose S = {e'}. Because neither of these functions is a solution of the homogeneous equation, we
take the particular solution to be the linear combination
Notice that e' is not a solution to the homogeneous equation, so we take y (I) to be
the linear combination of the functions in S. Therefore, p yp(t) = A cos I +B sin t.
Yit) = Ae'. We determine A and B by differentiating yP(t) twice and substituting into the non-
homogeneous equation
Substituting this solution into y" + 5y' + 6y = 2e', we have
y; + 5y~ + 6yp = -A cos 1- B sin I - 5A sin t + 5B cos 1+ 6A cos t + 6B sin I
Ae' + 5Ae' + 6Ae' = l2Ae' = 2e t.
= (5A + 5B) cos 1+ (-5A + 5B) sin I
Equating the coefficients of et then gives us A = k. A partiCUlar solution is
yp(t) = ke', so a general solution of y" + 5y' + 6y = 2e' is = 4 cos t.
yet) = Yh(t) + yp(t) = cle- 21 + C2e-3' + ie'. Hence,5A + 5B = 4 and -5A + 5B = 0, so A = B =~. Therefore,
(b) Here, we see that g(l) = blgl(t) + bzg2(t) = 2t 2 + 31. The set of functions yp(t) =
2 2 .
"5 cos I + "5 sm I,
associated with gl(t) = 212 is SI = {t2, I, I} and that associated with g2(t) = 3t is
S2 = {t, I}. Notice that none of these functions are solutions of the corresponding and a general solution is
homogeneous equation. Notice also that SI and S2 have two elements in cornmon.
If we take the linear combination of the functions t 2 , I, and 1, then
-?l
y(t) = Yh(t) + YP (t) = cle - + C2 e
-31
+ "52 cos t + "52 sm
. I
.
yil) = AI2 + Bt + c.
"f'"I,'.) y y'(t) = Ae- 2t cos 3t

p
+ (3B - 2A)te- 21 cos 3t + Be- 21 sin 3t+
(-3A - 2B)te- 2t sin 3t
Solve the following nonhomogeneous equations: (a)y" + 5y' + 6y = 4te- 2t + e-2t,
and (b) y" + 4y' + 13y = 4e- 2t sin 3t. and
y;(t) = (-4A + 6B)e- 21 cos 3t + (-5A - 12B)te- 21 cos 3t +

2 x
: I Solution (a) As we saw in Example 4, Yh(t) = c]e- 21 + C2e-31 is a genera! -1
(-6A - 4B)e- 21 sin 3t + (12A - 5B)te- 21 sin 3t,
solution of the corresponding homogeneous equation y" + 5y' + 6y = O. Nex~
we notice that get) = b]g](t) + b~2(t) = 4te- 2t + e- 2t . The set associated
with g](t) = te- 21 is S, = {te- 2t, e- 2t } and that associated with git) = e- 21 is ( which when substituted into the equation y" + 4y' + 13y = 4e -21 sin 3t yield
82 = {e -21}. Notice that both sets contain a function that appears in the corres-
y; + 4y~ + l3yP = 6Be- 21 cos 3t - 6Ae- 21 sin 3t = 4e- 21 sin 3t.
ponding homogeneous solutionYh(t) = c]e- 21 + C2e-31. We mUltiply both sets by t Figure 4.8 Graph of y(l) =
to obtain 8i = {t 2e- 2t , te- 21 } and 8; = {te- 21 } so that neither set contains a solu- Yh(l) + Yp(t) for various values
tion to the corresponding homogeneous set. However, because Si and 8; contain the Equating the coefficients oflike terms, we have -6A = 4 and 6B = O. Hence, A =
of c] and C2
common element te -21, we let -j and B = 0, so
yit) =At 2e- 2t + Bte- 21 . yp(t) = -t te -2t cos 3t.
Notice that if we include the repeated function twice, we arrive at the same result: Therefore, a general solution is
yp(t) = c]t 2e- 2t + C2te-21 + C3te-21 = c]t 2e- 21 + (C2 + c3)te- 21 . y(t) = Yh(t) + Yp(t) = c]e-
21 cos 3t + C2e-21 sin 3t - tte- 21 cos 3t.
Differentiating yp(t) twice and substituting into y" + 5y' + 6y = 4te -21 + e- 21
yields
What is the limit as t -> DO of every solution? Why doesn't the choice of c] and C2
y; + 5y~ + 6yp = (2A +B)e- 21 + 2Ate- 2t = 4te- 21 + e- 2t matter? Hint: First, identify and use the graph ofYp(t) in Figure 4.8 (or generate
your own) and then apply L'Hopital's rule, if necessary.
Therefore, 2A + B = I and 2A = 4 soA = 2 and B = - 3. This results in the par-
ticular solution To solve an initial-value problem, first determine a general solution and then use
the initial conditions to solve for the unknown constants in the general solution.
and the general solution
y(t) = Yh(t) + Yp(t) = c]e- 2t + C2e-3t + 2t 2e- 21 - 3te- 21 .

Solve the initial-value problem y'" + y" = 12t 2, yeO) = 0, y'(O) = I, y"(O) = 25.
(b) As in (a), first find a general solution of the corresponding homogeneous equa-
tion y" + 4y' + l3y = O. Verify that a general solution of the corresponding ho- : 1 Solution First, we solve the corresponding homogeneous equation y'" + y" = 0,
mogeneous equation is . which has characteristic equation r3 + r2 = r2(r + I) = 0 and thus, general solu-
Yh(t) = c]e- 2t cos 3t + C2e-2t sin 3t. tion
Because g(t) = 4e- 2t sin 3t, the associated set of functions is 8 = {e- 21 cos 31,
e -21 sin 3t}, but both of the functions in 8 are solutions to the homogeneous
Next, we set up the set of functions associated with get) = 12t 2, which is
equation, so we multiply by t so that they are no longer solutions. Hence, 8' :
8 = {t 2 , t, I}. However, both of the functions t and I are solutions ofy'" + y" = 0,
{te -21 cos 3t, te -21 sin 3t} and we let
so we multiply the functions in 8 by t 2 so that the functions that result do not ap-
yp(t) = Ate- 21 cos 3t + Bte- 2t sin 3t. pear inYh(t). Hence, 8' = {t 4, t 3, t 2}, so
This function has first and second derivatives Yit) = At4 + Bt 3 + Ct 2.

The derivatives of ypCt) are y ~(t) = 4At 3 + 3Bt2 + 2Ct, y;(t) = 12At 2 + 6Bt + 2C, y y y
and y;;(t) = 24At + 6B. Substitution into y'" + y" = 12t 2 then gives us
20 20
y;; + y; = 12At 2 + (24A + 6B)t + (2C + 6B) = 12t ,
2
so 12A = 12, 24A + 6B = 0, and 2C + 6B = O. This system of equations has the

solution A = 1, B = -4, and C = 12, so
yp(t) = t 4 - 4t 3 + 12t 2, -20
and a general solution is
Figure 4.9 yCt) = YhCt) + yp(t) = CI + C2t + C3e-c + t 4 - 4t 3 + 12t 2.
Figure 4.9 shows the graph ofy(t) = CI + C2t + C3e-c + t 4 - 4t 3 + 12t 2 for vari.
ous values of CI, C2, and C3' Identify the graph of the solution that satisfies the
initial conditions yeO) = 0, y' (0) = 1 and y"(O) = 25.
To determine the unknown c()efficients Ch C2, and C3, we apply the initial
conditions. First, we compute y'(t) = C2 - C3e-c + 4t 3 - 12t 2 + 24t and
y"(t) = C3e-c + 12t2 - 24t + 24. Applying the initial conditions, we have yeO) =
CI + C3 = 0, y'(O) = C2 - C3 = 1, andy"(O) = C3 + 24 = 25, so CI = -1, C2 = 2,
and C3 = 1. Therefore, the solution of the initial-value problem is
yet) = Yh(t) + yp(t) = -1 + 2t + e -c + t 4 - 4t 3 + 12t2 (see Figure 4.1 0). (d) (e)
y
(I)
y
y
o 'ifi""'.If w 2:
. .. { y" + Y = cos wt
0, solve the IIDtral-va1ue problem yeO) = 0, y' (0) = l'
20 20 20
-20
: 1 Solution A general solution of the corresponding homogeneous equation y" + -20 -20
y y = 0 is YhCt) = CI cos t + C2 sin t. We see that if w*,I we can find a particular
solution of the nonhomogeneous equation of the formyAt) = A cos wt + B sin wI,
while if w = 1 we can find a particular solution of the form ypCt) = teA cos t + (g) (h) (i)
B sin t). Solving for a particular solution, forming a general solution, and applying
y" + Y = cos wt
the initial conditions yields the solution Figure 4.11 Graph of the solution of { yeO) = 0, y'(O) = I for various values of w
(a) w = 3/4 (b) w = 23/28 (e) w = 25/28 (d) w = 27/28 (e) w = I (f) w = 29/28
-2--
1
w - 1
(
cos t - cos wt) .
+ sm .
t, If w *' 1 (g) w = 31/28 (h) w = 33/28 (i) w = 514
yet) = 1
-0.5
-0.5
0.5 x
1 '2 Ct + 2) sin t, if w = 1
Figure 4.10 Graph of y(t) = Notice that the behavior of the solution changes dramatically when w = 1: if
-I + 2t + e -, + t 4 - 4t 3 + o os: w < 1 or w > 1, the solution is periodic and bounded (why?), but if w = 1, tbe
12t 2 solution is unbounded (see Figure 4.11).
'-
, l~-
;
EXERCISES 4.5
For each function in Exercises 1-12, determine the associated + 2y' + 26y = -33St
29. y"
- -
59. { x"+4x=cost
x(O) = 0, x'(O) = 0
x" + 4x = cos 3t
Group A
x"+4x=cos2t
60. { x(O) = 0, x'(O) = 0
d 2y
.
dy
adj2+bdi+cy=k.
fInd limt_ yet).

. d~ •
(d) Determme the solutlOn of a dj2 + b di = k and
61. { x(O) = 0, x'(O) = 0

set of functions. + 3y' - 4y = -32t 2
30. y" (e) Determine the solution of a d 2y/dt 2 = k and find
= 2t + 5
1. f(t) *31. Sy" + 6y' + Y = 5t 2 Group B limHooy(t).
= 3 - St 2
2. f(t) 32. y" - 6y' + 8y = -256t 3 Ca) (b) 64. Let Yit) = ~ sin t. Show that if yet) is any solution of
x
3. f(t) = 7e 2t + 2 y" - 4y = -3 sin t then yet) - yp(t) is a solution of
= 52 sin 3t
33. y" - 2y' ) 0.) y" - 4y = O. Explain why there are constants C, and
4. f(t) = -2te-' 34. y" - 6y' + 13y = 25 sin 2t 2t
C2 so that y(t) - yp(t) = c,e- + C2e2t and, thus,
2 0.2
*5. f(t) = 4e- t - 2t4 + t *35. y" - 9y = 54t sin 3t yet) = cle- 2t + C2 e2t + yit).
6. f(t) = 4t - St 2e- 3t
1
!\I 0.1 f~
36. y" - 5y' + 6y = -7S cos 3t .lJ, 65. Show that if Yit) is a particular solution of the non-
7. f(t) = 6 sin 2t + 3e- 4' 37. y" + 4y' + 4y = -32t 2 cos 2t 1 12.5 homogeneous equation
-I -0.1
S. f(t) = 2t sin t 3S. y" - y' - 20y = -2e t an/n)(t) + an_l/n-l\t) + ...
-2 -0.2
9. f(t) = 2 sin 3t + 4 sin 3t - S cos 2t *39. y'~ - 4y' - 5y = -64St 2e" + a,y'(t) + aoy(t) = get)
, -) -0.3
10. f(t) = 2te' sin t - 7t 3 + 8 40. y" - 7y' + 12y = _2t 3e 4t and
*11. f(t) = e-' cos 2t + I 41. y'" + 6y" + lly' + 6y = 2e- 3t - te- t (e) Yh(t) = cd,(t) + Cz/2(t) + ", + cnln(t)
12. f(t) = te 2' sin t + cos 4t + 3t 2 - Se -4t 42. y'" + lOy" + 34y' + 40y = te- 4 ' + 2t- 3, cos t x
is a general solution of the corresponding homoge-
In Exercises 13-24, determine the form of a particular solution

*43. y'" + 6y" - 14y' - 104y = -llie' 0.4r~ neous equation
44. /4) - 6y'" + l3y" - 24y' + 36y = lOSt 0.2
any(n)(t) + an_ly(n-I\t) + ...
for the given nonhomogeneous equation. (Do not solve for the
unknown coefficients.) 45. /4) - lay'" + 3Sy" - 64y' + 40y = l53e-' + a,y'(t) + aoy(t) = 0
.5
+ Y = Se '
13. y" 2 -0.2 (assume that {f,(t),li(t), ... ,f,,(t)} is a fundamental
In Exercises 46-5S, solve the initial-value problem. set of solutions for the corresponding homogeneous
+ 3y = -e- 9 '
14. y" - 4y' -0.4
equation), then every solution of the nonhomogeneous
46. y" + 3y' = IS,y(O) = O,y'(O) = 3
*15. y" - 4y' + 3y = 2e 3'
16. y" - y = 2t - 4
47. y" - Y = 4,y(0) = O,y'(O) = a -0.6 V equation, yet), can be written in the form
4S. y" - 4y = 32t, yeO) = 0, y'(O) = 6 yet) = Yh(t) = yp(t),

17. y" - 2y' +y= t 2
*49. y" + 2y' - 3y = -2, yeO) =~, y'(O) = S 62. T he method of undetermined coefficients can be used for some choice of Clo C2, . . . , en'
IS. y"+ 2y' = 3 -4t
50. y" + y' - 6y = 3t, yeO) = ¥" y'(O) = -~ to solve fIrst-order nonhomogeneous equations. Use 66. (a) Find conditions on w so that Yit) = A cos wt +
*19. y" + Y = cos 2t this method to solve the following problems. B sin wt, where A and B are constants to be deter-
y" + Sy' + 16y = 4,y(0) = b'(o) = a
-*,
51.
20. y" + 4y = 4 cos t - sin r (a) y' - 4y = t 2 mined, is a particular solution of y" + Y = cos wt.
52. y" + 7y' + lOy = te-t, yeO) = y'(O) = to (b) For the value(s) of w obtained in (a), show that
21. y" + 4y = cos 2t + t (b)y'+y=cos2t
*53. y" + 6y' + 25y = -I, yCO) = -,'"y'(O) = 7 every solution of the nonhomogeneous equation
22. y" + 4y = 3te- t (c ) y' - y = e 4t
54. y" - 3y' = _e 3t - 2t, yeO) = 0, y'(O) = ~ y" + Y = cos wt is bounded (and periodic).
*23. y" = 3t 4 - 2t
55. y" - y' = -3t - 4t 2e 2t, yeO) = -1, y'(O) = 0 63. ( a) Suppose that Yl(t) and Y2(t) are solutions of 67. (a) Find conditions on w so tilat Yit) = t(A cos wt +
24. y" - 4y' + 13y = 2te- 2J sin 3t d 2y dy B sin wt), where A and B are constants to be deter-
56. y" - 2y' = 2t2,y(0) = 3,y'(0) =! a dj2 + b dt + cy = f(t), where a, b, and c are
mined, is a particular solution of y" + 4y = sin wt.
1n Exercises 25-45, find a general solution of each equation. *57. y'" + 5y" = 125t, yeO) = 0, y'(O) = 0, y"(O) = 0 Positive constants. Show that (b) For the value(s) of w obtained in (a), show that
5S. y'" + 25y' = 325e- t, yeO) = 0, y'(O) = 0, y"(O) =0 every solution of the nonhomogeneous equation
25. y" + y' - 2y = -I limHoo [y,(t) - YI(t)) = o.
y" + 4y = sin wt is unbounded.
26. 5y" + y' - 4y = -3 (b) Is the result of (a) true if b = O?
1n Exercises 59-61, without actually solving the equations,
*27. y" - 2y' - 8y = 32t match each initial-value problem in Group A with the graph of (c SupjJOse that f(t) = k where k is a constant. Show As when calculating solutions of homogeneous equations, tech-
28. 16y" - Sy' - 15y = 75t its solution in Group B. that limHoo y(t) = Idc for every solution yet) of nology is useful in helping us calculate and visualize solutions
j Ioi.
182 Chapter 4 Higher Order Equations 4.6 Nonhomogeneous Equations with Constant Coefficients: Variation of Parameters 183
of nonhomogeneous equations, particularly when the calcula- 72. Show that the solution to the initial-value problem Nonhomogeneous Equations with Constant
tions encountered are symbolically complicated.
y" + y' - 2y = f(l) Coefficients: Variation of Parameters
68. Find a general solution of each equation. { yeO) =0, y'(O) a =
(a) y'" - y" - 7y' + 15y = 12e- 3' + e 2t cos 31 (" Second-Order Equations (" Higher Order Nonhomogeneous Equations
is
(" Green's Function
(b) y(4) - 9y'" + 24y" - 36y' + 80y = cos 21 +
4 sin 21 Y(I) =.1 ae' _.1 ae- 2t +.1 e' It f(x)e-' dx-
3 3 3 0 Second-Order Equations
69. Solve the following initial-value problems. Graph the
solution. .1 e -2t It f(x)e 2t dx
3 0
For the nonhomogeneous equation y" + h = sec (tI2) + esc (tI2), 0 < t < 7T, the non-
2Y '" - 6y" + 18y' - 56y = e- 2' + 4e 3t homogeneous function is g (t) = sec (tI2) + esc (t12), which is not a function of terms
(a) { y"(O) = -1,y'(O) = O,y(O) = I (as long as the integrals can be evaluated). of the form tm, tme"', tme"tI cos f3t or tme"tI sin f3t. Because the method of undetermined
[ y(4) _ h'" - 21y" - ~y' + 5y = I (a) Use the Fundamental Theorem of Calculus to vet. , coefficients or the annihilator method are limited to equations involving these func-
ify this result. . tions, we need to introduce another method, called variation of parameters, discov-
(b) [y"'(0) = 0, y"(O) = 0, y'(O) = 0, yeO) = I
(b) If possible, find a function f(l) and a value of a ered by Lagrange, that can be used to find a particular solution of other equations such
70. Consider the initial-value problem
so that (i) the solution is periodic; (ii) the solution as this one.
y'" + y" + 4y' + 4y = e- t cos 21 approaches 0 as x approaches infinity; and (iii) A general solution to the corresponding homogeneous equation y" + ~y = 0 is
{ yeO) = a, y'(O) = b, y"(O) = C . the solution has no limit as x approaches infinity. Yh(t) = CI cos (1/2) + C2 sin (t12).
Confirm your results graphically. In the method of variation of parameters, we try to find a particular solution of
(a) Find conditions on a, b, and c, if any, so that
Y(I) ~ 0 as I~ 00. Confirm your results by graph- 73. How does the solution to the initial-value problem the form
ing a (the) solution that satisfies the conditions. Xu + 9x = sin (CI)
(b) Is it possible to choose a, b, and c so that the so- { x(O) = 0, x'(O) = 0
lution behaves like cos 21 for positive large values
change as c assumes val ue from 0 to 6?
of I sin 21? where we replace the constants CI and C2 in Yh(t) with the unknown functions UI (t) and
74. Ifwe consider a differential equation over a certain in.
71. Determine if it is possible to find values of a3, a2, a" U2(t). We arrive at the name of the method from this replacement because we vary the
terval and are given conditions that must be satisfied
and ao so that a general solution of parameters CI and C2 by allowing them to be jUnctions of t instead of constants. (No-
at the endpoints of the interval, then we call this a
y(4) + Q3yIfI + Q2Y" + Q1Y' + GoY = cos t bolindary-value problem (BVP). tice that there should be many choices for U1(t) and U2(t) because we have two unknown
functions and only one equation, the nonhomogeneous differential equation, to use in
has the indicated form. In each case, either find such (a) Show that the boundary-value problem
values (and confirm your results) or explain why no finding them.) We find possible choices for UI (t) and U2(t) by substitution of Yp(t) into
such numbers exist. 4Y " + 4y' + 37y = cos 31 the nonhomogeneous equationy" + h = sec (t12) + esc (t12). Differentiating Yit), we
{ yeO) = y(7T)
(a) y(l) = cle -It + C2e -t + C3e' + C4e2t + -10 cos I find that
(b) y(l) = cle -t + C2e' + C3 cos

-icost-ttsint
I + C4 sin I
has infinitely many solutions. Confirm your re-
sults graphically.
(b) If possible, fmd conditions ony(O) andY(7T) so
y~(t) = -t UI(t) sin f + u;(t) f + t U2(t) f + uS(t) f·
cos cos sin
(e) y(l) = CIcos I + C2 sin I + C31 cos I

that the solution to the boundary-value problem ~ . To simplifY the process of finding UI (t) and U2(t), we assume that
+ c41 sin t + i cos t + ~ sin t - it 2 cos t unique.
(d) y(l) = CI cos I + C2 sin I + C31 cos I
+ C41 sin I + 13 sin I u;(t) cos f + uS(t) f
sin = 0,
which is our first restriction on UI(t) and U2(t). (In other words, of all functions uI(I)
and U2(t) that lead to a particular solution to the nonhomogeneous differential equa-
tion, we look for functions that satisfY this condition.) Eliminating this expression from
y~(t) gives us
so that the second derivative is

4.6 Nonhomogeneous Equations with Constant Coefficients: Variation of Parameters 185
Yp"( t) = -4I Ul (t) cos 2t - 2I Ul'()t sm

. 2t - 4I U2 ().
t sm 2t + 2IU2 t
, (t) cos 2' so one possibility for Ul(t) is
Next, we substitute Yp(t) into the nonhomogeneous equation . 2

-sm I sec (t2 + csc 2t) sm-
y" + ~y = sec (tI2) + csc (tI2) to obtain
I
dt = -2 J( -42
. t
cos
+I
)
dt =
Yp"+ I
4 Yp = -4I () t
Ul t cos 2 - 2I 'C) . t
Ul t sm 2 - t sm 2t
4I U2 (). 2
+ t uz(t) cos t + t[ Ul(t) cos t + u,(t) sin t]

-2t + 4 Inlcos tl.
Again, we have selected the anti derivative with constant of integration equal to zero.
-_ -2I Ul'C)t sm
. 2t + 2I U2'C) t
t cos 2' Then by variation ofparameters,
Then, because Yp(t) satisfies the nonhomogeneous equation, we have a second restric_ Yit) = ul(I)Yl(t) + U2(t)Yz(t)
tion on Ul (t) and U2(t),
-2I Ul'C)t sm
. 2t + 2I U2'C)t cos 2I = sec 2t + csc
I
2'
= cos t[ -21 + 4 Inlcos tl] + sin t[2t + 4 InlSin tlJ
is a particular solution of y" + ±Y = sec (t/2) + csc (t12) and

This gives us the system of two equations
Y = Yh(t) + Yp(t)
u;(t) cos t + uz(t) sin t = 0
= Cl cos t + C2 sin t + cos t[-21 + 4lnlcos tl + sin t[2t + 41nlsin tl]
1 . 2t + 2I U2'C)
-2I Ul'C)t sm t
t cos 2 t l
= sec 2 + csc 2
that we solve for the two functions u;(t) and uz(t). Multiplying the first equation byl
is a general solution. Note that cos (tIZ) > 0 and sin (1/2) > 0 on 0 < x < 7T, so the
absolute value signs can be eliminated from the arguments of the natural logarithm
sin (t12), the second by cos (t12), and adding the resulting equations yields functions.
In general, to solve the second-order linear nonhomogeneous differential
cos 2I [t
sec 2 + csc 2t equation
Uz(t) = a2(I)y" + al(t)y' + ao(t)y = g(I),
I
2 where Yh(t) = C1Yl(t) + C2Y2(t) is a general solution of the corresponding homogeneous
equation a2(t)y" + al(t)y' + ao(t)y = 0, we first divide the equation by a2(t) to rewrite
so that one choice for U2(t) is obtained through integration to be
it in the form
U2(t) = J
(t2 +csc 2t)
cos t sec
2
t =2J
( cos-
I + sin
I )
f dt = 21 + 4 Inlsin tl·
y" + p(t)y' + q(t)y = f(t),
assume that a particular solution has a form similar to the general solution by varying
the parameters Cl and C2, and let
(Notice that any anti derivative of uz(l) is a possible choice for U2(t). Here, we assume ~' Yp(t) = Ul (t)Yl (I) + U2(t)Y2(t).
that the constant of integration is zero.) Similarly, if we multiply the first equation in .• We need two equations to determine the two unknown functions Ul(t) and U2(t), and
the system by -i cos (1/2), the second by sin (tIZ), and add the equations that resu1~. we obtain them by substituting yp(t) = Ul(t)Yl(t) + U2(t)Y2(t) into the nonhomogeneous
-~ .
differential equation y" + p(t)y' + q(t)y = f(I). Differentiating Yp(t), we obtain
. I(I
-sm"2 sec 2 + csc 2I) y~(t) = ul(l)y;(I) + Uj(t)Yl(t) + U2(t)YZ(t) + u2(t)Y2(t),
ul(t) =
I which can be simplified to
2
with the assumption that u;(t)Y1(t) + uz(t)yz(t) = O. Notice that this assumption is made Summary of Variation of Parameters for Second-Order Equations
to eliminate the derivatives of UI (t) and uz(t) from this expression for y ~(t), which shu_
plifies the process for finding Ul (t) and uz(t). Therefore, the first equation satisfied by Given the second-order equation y" + p(t)y' + q(t)y = f(t):
UI(t) and uz(t) is
CD Find a general solution Yh(t) = ClYI (t) + czyz(t) and fundamental set of solu-
u; (t)YI (t) + uz(t)yzCt) = o. tions S = {Yl (I), Y2(t)} of the corresponding homogeneous equation y" + p(t)y'
+ q(t)y = o.
With our simplified expression for y ~(t), the second derivative is
, _ -Yz(t)f(l) , _ Y1 (t)f(t)
y;(t) = Ul(t)y]'(t) + u;(t)y;(t) + u2(tM(t) + uz(t)yz(t). @ Let Ut(t) - W(S) and U2(t) - W(S) .
Substitution into y" + p(t)y' + q(t)y = f(t) then yields ® Integrate to obtain UI(t) and uz(t).
@ A particular solution of y" + p(t)y' + q(t)y = f(t) is given by
y; + p(t)y~ + q(t)yp = uI(t)[y]'(t) + p(t)y;(t) + q(t)YI(t)] yit) = UI(t)YI(t) +U2(t)Y2(1).
+ U2(t)[y~(t) + p(t)y z(t) + q(t)Y2(t)] ® A general solution of y" + p(t)y' + q(t)y = f(l) is given by y = Yh(t) + yp(t).
+ u;(t)y;(t) + uz(t)yz(t)
= u;Ct)y; (t) + uz(t)y2(t) = f(t)
because YI(t) andYz(t) are solutions of the corresponding homogeneous equation. There-
"E',,','.'
Solve y" - 2y' +Y = e' In I, I> O.
fore, our second equation for determining UI(t) and uz(t) is u;(t)y;(t) + uz(t)yz(t) =
f(t), so we have the system
:1 Solution The corresponding homogeneous equation has the characteristic
u; (t)YI (t) + uz(t)yzCt) = 0 equation r2 - 2r + I = (r - 1)2 = 0 so Yh(t) = CI e' + C2 te' and
~
{ uj(t)y;(t) + uz(t)yz(t) = f(t)'
h· h· wntten
W IC IS
. . matrix. ~,orm as
In
(YI(t)
y; (t)
Yz(t)) (ul(t))
y;(t) u;"(t) =
(0) In Imear
f(t).
. algebra, S = {e', te'}. Therefore, we compute
Wi(S) = Ie'
e'
te'
e' + te'
I = eZ' + teZ' - le z , = eZ'.
we learn that this system has a unique solution if and only if Iy~((t)
Y I t)
y;(t) I
yz(t)
*o. " Through the use of integration by parts, a table of integrals, or a computer algebra
system,
J- J
Notice that this determinant is the Wronskian of the set S = {YI(t), Y2(t)}, W(S). We .""
*
stated in Section 4.1 that W(S) 0 if the functions Yt (t) and Y2(t) in the set S are lin- ." UI(t) =
te'(e' In t)
e Z' dt =- tin t dl
1 z
="4 t -"2 t
1 Z
In t
early independent. Because S = {Yl (t), Yz(t)} represents a fundamental set of solutions
"* and
of the corresponding homogeneous equation, W(S)
solution that can be found by Cramer's rule to be
O. Hence, this system has a unique
U2(t) =
J e'(e' In t)
e Z' dt =
J In I dl = t In I - t.
I0 yz(t) I Then a particular solution, which is assumed to have the form
f(t) A(t) -yz(t}f(t)
u;(t) =
W(S) W(S) yp(t) = UI(t)Y1(t) + UZ(t)Y2(t),
and is
IYI(t)
y;(t)
0
f(t)
I
YI(t}f(I)
yp(t) = (+ + 2
t -
Z
I In I) e' + (I In I - t)le' = + IZe' In t - i 2
t e'.
uz(t) =
W(S) W(S) . Therefore, a general solution is
Note: If you are not familiar with Cramer's rule, these formulas can be found with 'Ini - "4
elimination as we did in solving y" + iy = sec (1/2) + csc (t/2) at the beginning of the
yet) = Yh(t) + yit) = cle , + czle,+"2
1 l2 e 3 t 2 e,, t> 0 .
section. The functions UI (t) and U2(t) are then found through integration.
Of course, the problems that were solved in the preceding sections by the method We then evaluate yeO) = CI = 0 and y'(O) = 2C2 - 114 = 1. Hence, c, = 0 and
of undetermined coefficients can be solved by variation of parameters as well. C2 = 5/8, so the solution is
"f'"I,'., yet) = i sin 2t - i t cos 2t.
Note: If we had used the method of undetermined coefficients, we should not

Solve the initial-value problem y" + 4y = sin 2t, yeO) = 0, y'(O) = 1 using varia. expect a particular solution of the same form as that obtained through variation of
tion of parameters.
parameters unless we simplifY the original functions u I(I) and U2(t) obtained through
integration. (Why?)
:.J Solution The characteristic equation of the corresponding homogeneous equa...
tion has the complex conjugate roots r = :':2i, so Yh(t) = CI cos 2t + C2 sin 21
'---.-' '---.--" In the following example, a condition is given at each of the endpoints of the in-
y,(I) h(l) terval 0 :=;: t :=;: 7T. Solutions to problems of this type, called boundary value problems,
and S = {cos 2t, sin 21}. Thus, W(S) = Ic0 s 2: sin 21 I differ from those of initial-value problems in that there may be no solution or more
- 2 sm 2t 2 cos 21 = 2. Through the use
of trigonometric identities, we obtain than one solution.
() J
and
UI I = - -1sm-
2
: ' 2t d t = - -
1I
4
+ - 1sm
16
. 41
o 'ifi""'·'
For what value(s) of A> 0, if any, does the boundary-value problem
U2(t) = J± cos 21 sin 21 dt = -t cos z 2t. {Y "+ A2y = sin 2t have (a) one solution, (b) no solutions, and (e) m
yeO) = O,y(7T) = 0
. 1y
. f'lrute
Simplifying and using the identity sin 41 = 2 sin 2t cos 2t leads to the particular so. many solutions?
lution
y : 1 Solution A fundamental set of solutions of the corresponding homogeneous
. equation is S = {cos At, sin At} with W(s) = A. By variation of parameters,
=
1 1 . 4) 1
( -"4 t + 16 sm t cos 21 - "8 cos 2t· sm 2t
2 .
UI(t) = J -sin A~ sin 2t dt
= -i t cos 21 + ;6 sin 2t cos 2t . cos 21 - t cos 2 2t . sin 2t

=
(2 + A) sin((2 - ,1.)1) - (2 - A) sin((2
2,1.(,1.2 - 4)
+ A)t) .
, If A", 2
= _.1 1 cos 2t
4
and, thus, a general solution is
1 sin 4t - 4t 'f' = 2
16 ,1"
and
Figure 4.12
yet) = Yh(t) + yp(t) "" CI cos 2t + C2 sin 2t - i t cos 2t. uz(t) = f cos A~sin 2t dl
*
Figure 4.12 shows the graph y (I) = CI cos 2t + C2 sin 2t - t cos 2t for various . (2 + A) cos((2 - A)t) + (2 - A) cos((2 + A)t) .
values of CI and Cz· IdentifY the graph of the solution that satisfies the initial con. 2A(A2 _ 4) , If A '" 2,
ditions (a) yeO) = 0 and y'(O) = 1; and (b) yeO) = 0 and y'(O) = -1.
To solve the initial-value problem, we compute
y'(t) = -2cI sin 2t + 2cz cos 2t - i cos 2t + ± t sin 2t.

1 2
_ cos 2t if A = 2
8 '
so a general solution of the nonhomogeneous equation is

y(t) = Cl cos At + C2 sin At + Ul(t)Yl(t) + U2(t)Y2(t) so f(t) = 2. Therefore, using integration by parts, we obtain
2
Ul(t)=J -ln t(2) dt=-2Jtlntdt=-t 2 Int+tt
I
Cl cos At + C2 sin At + _2_1- sin 2t, if A
A - 4
"* 2 l
= Cl cos 2t + C2 sin 2t - t t cos 2t, if A = 2. and
Applying the boundary conditions yields the system of equations
{Cl cos A1T+ C2

. Cl = 0, if A
Sin A1T= 0
"* 2
and so Yit) = Ul(t)Yl(t) + U2(t)Y2(t) 2
= (I)(-t In t + t 2
t ) + (In
2
t)(t ) = t 2
t , and a
general solution to the nonhomogeneous equation is
; = 0, if A = 2.
{ Cl -"4= 0 y(t) = Yh(t) + Yit) = Cl + C2 In t + "2I t 2, t> 0
.
y
'-
Cl = 0 .'
The system { Cl _ 1T14 = 0 does nothave a solution (why?), so the boundary-value
Higher Order Nonhomogeneous Equations
problem does not have a solution if A = 2. However, the solution to the system
Higher order nonhomogeneous linear equations can be solved through v~riation of pa-
{ Cl cos '+ C2 SID
A'TT
Cl =
" 1l7T = 00 IS
. Cl = 0 and C2 = 0, proVl'de d th at " IS not an mte-
1\
. rameters as well. In general, if we are given the nonhomogeneous equatIOn
"*
ger, because sin A1T = 0 if A is an integer. But, if A 2 is an integer, then any value inlet) + a n_l(t)/n-l)(t) + ... + al(t)y'(t) + ao(t)y(t) = get)
of C2 is a solution to this system, so there are infinitely many solutions to the
-3 "*
boundary-value problem if A 2 is an integer. and a fundamental set of solutions {YI (t), Y2(t), ... , Yn(t)} of the associated homoge-
Thus, the boundary-value problem has no solution if A = 2, no solution if ,\ neous equation
Figure 4.13
"*
is not an integer, and infinitely many solutions if A 2 is an integer (see Figure /n)(t) + an_I(t)/n-I)(t) + ... + al(t)y'(t) + ao(t)y(t) = 0,
4.13).
we generalize the method for second-order equations to find Ul (t), U2(t), ... , unCt)
such that
An advantage that the method of variation of parameters has over the method of yp(t) = UI(t)Yl(t) + U2(t)Y2(t) + ... + un(t)Yn(t)
undetermined coefficients is that nonhomogeneous equations with nonconstant coeffi-
cients can be considered. is a particular solution of the nonhomogeneous equation.
If
I'i·,,·,'.' 2
then
A general solution of t y" + ty' = 0, t> 0 is Yh(t) = Cl + C2 In t. Solve the non- y~m)(t) = UI (t)y\m)(t) + u2(t)Am)(I) + ... + un(t)y~m)(t)
homogeneous equation t 2y" + ty' = 2t2, t> O.
for m = 0, I, 2, ... , n - I, and if
I Solution In this case, S = {Yl (t), Y2(t)} = {I, In t}, so W(S) = I~ I~/:I = III. uj(t)y\m-l)(t) + u;(t)Am-l)(t) + ... + u~(t)y~m-I)(t) = 0
Recall that we divided the second-order equation a2(t)y" + al(t)y' + ao(t)y = get) for m = I, 2, ... , n - 1, then
by the coefficient function of y" to place it in the form y" + p(t)y' + q(t)y = f(t)
before deriving the formulas for Ul(t) and U2(t). Doing this, we have y" + +
y' = 2,
y~n)(t) = (Ul(t)y~)(t) + u2(t)An)(t) + ... + Un(t)y~)(t)!
+ (u;(t)y\n-I)(t) + u;(t)yr-I)(t) + ... + un(t)y~
n-I)
(t»).
192 Chapter 4 Higher Order Equations 4.& Nonhomogeneous Equations with Constant Coefficients: Variation of Parameters 193
zl
Following a method similar to that of a second-order differential equation, we obtain
the system of n equations e-
-Ze- Zl )(Uj(t))
u;(I) = (0 )
0 ,
Uj(t)Yl(t) + u;(t)yz(t) + ... + u~(t)Yn(t) = 0 4e -21 u3(t) COS t
uj(t)yj(t) + u;(t)y;(t) + ... + u~(t)y~(t) = 0 ,
1
where
y\n-l)(t)uj(t) + y~-I)(t)u2(t/+ ... + y~-l)(t)u~(t) = get) -,
l~
e
-e- r
which can be solved for uj (t), u;(t), ... , u~(t) using Cramer's rule. WCs) =
Let Wm(yl(t), yz(t), ... , yit)) denote the determinant of the matrix obtained by e- I
replacing the mth column of
Using this with Cramer's rule, we have
Yl (t) Yz(t)
yit) )
(
yj(t)
y\n~l)(t)
y;(t)
yr-1)(t)
y~(t)
y~~l)Ct) IL, ~-' ~:~,:ul

uj (I) = .!....---_-Z-e--.37t----'-
_e- 3t cos t 1
-Ze- 3 , = 2" cos t,
by the column ( 1). Then, by Cramer's rule, 1 0

o 0
e -2t
_Ze- Zt
I
get)
o cos t 4e- 2 ,
Ze- 2t cos I
u;Ct) = -Ze 31
=-~3:=-'~ = -e cos
t
I,
-Ze
'( ) _ .o;.g-,-Ct;....) and

u· t - - W,--'C.::...Y.:..:1(t-,-,),. ::..Y::..:z(t.:. :.,_.
) _.....:.'.::..:yn::.o,(t=))
'W(yl(t), YZ(t), ... , Yn(t)) e-'
1~
1 0
for i = I, Z, ... , n, and -e- t 0
J g(t)W,CYt(t), h(I), ... , Yn(t))

t
u;(t) = e- cos I -e-' cos I 1 21
U3(1) = -'------Z-e--.3-.,-,---'- = -Ze 31 = 2"e cos t.
W(yl(t), yz(t), ... , Yn(t)) <ix,
so Integration then gives us u, (t) =~ sin t, U2(t) = -~ etc cos t + sin I), and U3(1) =
Yo e 2t (Z cos I + sin t).
Yit) = UI Ct)YI Ct) + uz(t)yz(t) + '" + un(t)yil) Because
is a particular solution of the nonhomogeneous equation. A general solution ofthe non-
Yp(l) = ul(t)y,(I) + U2(t)Y2(t) + U3(t)Y3(1),
homogeneous equation is given by y(t) = Yh(t) + yp(t), where Yh(t) is a general solu·
tion of the corresponding homogeneous equation. I-j, " we find through substitution into this equation and simplification that yit) =
-\ / -l'o cos I + Yo sin t is a particular solution of the nonhomogeneous equation. A gen-
-------._--
l'f'"I,'.,
Solve y'" + 3y" + Zy' = cos t.
-2 ___ .._
-3
------- eral solution is
y(l) = YhCt) + Yit) = Cl

3
+ C2 e -, + C3 e -21 -To cos
t+ 1 . t
To sm .
Figure 4.14
: ) Solution You should verify that a general solution of the corresponding homo·
geneous equation iSYh(t) = CI + cze- I + C3e-zl and a fundamental set of solutions Figure 4.14 shows the graph of y(l) = YhCt) + yit) for various values of c" Cz, and
is S = {l, e-', e- zl }. Therefore, we must solve the system C3' IdentifY the graph of yp(t).
EXERCISES 4.6
.. d' + ~,(t)u' ~ aoCt)u ,= 0
{ u(lo) - Yo, u (to) Yo
and v is the solution of

=
lution obtained with the Green's function satisfies the
IVP.
Suppose that we wish to solve y" + 4y = f(I), yeO) = 0,
y' (0) = O. In this case, y.(t) = c, cos 2t + C2 sin 2t, so
I~ Exercises 1-38, solve each differential equation using vari-
atIon of parameters. 30. y" + 8y' + 16y = e-
4t v" + a,(t)v' + ao(t)v = f(t)
{ veto) = 0, v' (10) = 0 .
y (t) = I'
sin 21 cos 2x - cos 2t sin 2x f(x) dx =
1+12
, o
sin 2(t - x)
2
sin 2(1 - x)
1. y" - 7y' + lOy = e't
2. y" + 16y = 2 cos 41
*31 "+ I t t
' Y " 4 y =sec"2+ csc "2
48. Let y, (t) and yit) be two linearly independent solu·
tions of the second-order homogeneous linear differ-
Io 2 fex) dx. Therefore, G(I, x) = 2 . For
the IVP, y" + 4y = sin I, yeO) = 0, y'(O) = 0, the solution is
*3. y" + 4y' + 4y = e- 2t 32. y"'+y' = -2sinl- sint ential equation t sin 2(1 - x) . I. I.
4. y" - 4y' + I3y = 8e 2t cos 31
33. y'" + 4y' = sec 2t
cos 2 t
a2(I)y" + a,(I)y' + ao(t)y = O.
yet) = 0
I 2 sm x dx = 3" sm t - 3" sm 21.
5. y" + 4y' + 20y = 21e- 2t 50. Use the Green's function obtained above to solve
What is a general solution of a2(I)y" + a,(I)y' +
6. y" + 4y = sec 2t 34. y'" - 2y" = - I + 2t I> 0 y" + 4y = sin wt, yeO) = 0, y'(O) = 0 for (a) w = 3;
t2 , ao(l)y = g (t)? (Hint: In order to apply the formulas of
*7. y" + 16y = csc 4t (b) w = 2.5; (c) w = 2.1; (d) w = 2. What happens to
variation of parameters, the lead coefficient must be
8. y" + 16y = cot 41 *35.Y"'-3Y"+3y'-y=£ t>O I; divide by a2(t) and apply the variation of parame- the solution as w -> 2?
t' + k 2y =
9. y" + 2y' + 50y = e- t csc 7t ters formula.) 51. Use a Green's function to solve y" f(t),
36. y':' - 4y" - Ily' + 30y = e 4t yeO) = O,y'(O) = O.
10. y" + 6y' + 25y = e-3t(sec 4t + esc 41)
37. y"; + 3y" - lOy' - 24y = e- 3t 52. If we use the solution in Exercise 51 to solve y" +
*11. y" - 2y' + 26y = et(sec 51 + esc 5t) Green's Functions
-6r •
38. y'" - 13y' + 12y = cos t k 2y = sin wi, yeO) = 0, y'(O) = 0, what value of w
Suppose that we are trying to solve the initial-value problem
12. y" + 12y' + 37y = ~ I> 0 y" + p(t)y' + q(t)y = f(I), y(lo) = 0, y'(to) = O. Based on leads to the behavior seen in Exercise 50 (d)?
t ' 2
In Exercises 39-45, solve the initial-value problem. the Method of Variation of Parameters, we write the solution *53. Use a Green's function to solve y" - k y = f(t),
13. y" - 6y' + 34y = e 't tan 5t
of this equation as y(l) = ",(t)y,(t) + u2(I)Y2(1), where yeO) = 0, y'(O) = O.
39. y" + Y = tan t, yeO) = 2, y'(O) = I,
-St .
14. y" - lOy' + 34y = e sm 3t t> 0 '
-Y2(x)f(x) J.t y,(x)f(x) 54. Use the solution obtained in Exercise 53 to solve
t2 , -71"/2 < t < 71"12
J.
",(I) = t. W(S) dx and "2(1) = t, ~ dx; y, and
the IVP y" - 4y = f(I), yeO) = 0, y'(O) = 0 for
*15. y" - 12y' + 37y = e 6t sec t 40. y" + 9y = t csc t, y("./4) = \12, Y'(71"/4) = 0 + p(I)Y' + q(t)y = O. Here W(S) represents the
Y2 satisfy y" (a) f(l) = 1; (b) f(t) = e'; (c) f(l) = e2t; (d) f(l) =
16. y" - 8y' + 17 = e 4t sec t *41. y" + 5y' + 6y = -3 - sin 4t, yeO) = -23150 Wronskian of y, and Yz as a function of x. Therefore, cos t; (e)f(l) = e- 2t cos I; (fif(t) = e2t cos I. What
y'(O) = 27/25 ' type offunctionf(t) (if any) yields a periodic solution
17. y" - 9y = __I _ y(l) = ",(I)y,(t) + U2(t)Y2(1)
I + e3t 42 . y " - 16y 161 to this IVP?
= 7'Y(0) = O,y'(O) = 0
= (t) J.t -Y2(x)f(x) dx + (t) J.t y,(x)f(x) d; 55. Given that y = c,t-' + C2t-' In I is a general solu·
18. y" - 25y = _ I _ y,
I - "W(S) Y2" W(S) X tion to t 2y" + 3ty' + Y = 0, I > 0, use variation of
eSt
43. y" - 2y' +Y = _e- 2t cos 31 yeO) =
,
-.!!
13' parameters to solve t 2y" + 3ty' + Y = In I, I> O.
*19. y" - y = 2 sinh 1
= J.t Yz(t)y,(x) - y,(I)Y2ex) f( )dx
y'(O) = _l.Q. t, W(S) x. *56. Given that y = c, sin (2 In t) + C2 cos (2 In t) is a
20. y" - 2y' +Y = £t' I > 0 13 general solution to t 2y" + ty' + 4y = 0, I> 0, use
Thus, we can solve the nonhomogeneous problem using an in- variation of parameters to solve 12y" + ty' + 4y = t,
2t
+ 4y =!!.....-
44. y" - 6y' + 18y = _e't sin 3t, yeO) = _1 tegral involving the functionf(I). The function
21. y" - 4y' t 2 , I> 0 y'(O)=o 6' t> O.
-4t G( ) = Yz(t)y,(x) - y,(t)Y2ex) 57. Given that y = c,t 6 + c21-' is a general solution to
22. y"+ 8y' + 16y=_e_
t4 '
1>0 *45. y'" - y" = 31 2, yeO) = 0, y'(O) = 0, y"(O) = 0 t,x W(S) t 2y" - 4ty' - 6y = 0, t> 0, use variation of param-
46. Solve y" + 4y' + 3y = 65 cos 2t by (a) the method eters to solve t 2y" - 4ty' - 6y = 2 In I, I> O.
*23. y" + 6y' + 9y =.J... I> 0 of undetermined coefficients; (b) the method of is called the Green's function.
58. Consider the initial-value problem
3
le "
variation of parameters. Which method is more easily ' -Y2(x)f(x)
24. y" + 6y' + 9y = e-" In 1,1> 0 applied?
49. Show that u,(I) = t, J. W(S) dx and u2(1) = 4y" + 4y' + Y = e-'12
25. y" + 3y' + 2y = cos(e t ) 47. Show that the solution of the initial-value problem ' y,(x)f(x) dx indicate that u,(lo) = 0 and
{ yeO) = a,y'(O) = b .
26. y" + 4y' + 4y = e - 2tv'l""=7, -1 :S t:S I J.t, W(S) (a) Show that regardless of the choices of a and b,
fy" + ~,(t)y' ~ ao(l)y ,= f(l) U2(tO) = O. Recall that the Method of Variation of
*27. y" - 2y' + Y = e tv'l""=7, -I :S I:S I LY(lo) - Yo, y (to) = Yo
the limit as I -> IX) of every solution is O.
Parameters is based on the two equations
28. y" - lOy' + 25y = e 5t In 21 Y(I) = u,(I)y,(t) + ",(t)y,(t) and y'(I) = u;(t)y,(t) +
(b) Find conditions on a and b, ifpossible, so thaty(t)
can be written as Y(I) = U(I) + V(I) where U is the SD-
29. y" - 4y' + 4y = e 2t tan-' t lutlOn of u2(l)Y2(1). Show that y (to) = y' (to) = 0, so that the so·
has (i) no local minima or maxima; (ii) exactly
4.7 Cauchy-Euler Equations
one local mInlmwn or local max' . and anxny<n) + an_IXn-ly<n-l) + ... + aIX)" + aoy = g(x),
(iii) two distinct local minima and m=:: y(l) = sin (2t) + .!. sin(2t) ]' f(x) cos 2x dx
2 0
where ao, aj, a2, ... , an are constants, called Cauchy-Euler equations. (Notice that
(e) If possible, find a and b so that y(l) has local
- "21 cos(2t) ]' cos(21) ]' f(x) sin 2x dx in Cauchy-Euler equations, the order of each derivative in the equation equals the power
extrema If I = 1 and I = 3. Graph y(l) on the o 0
mterval [0, 5]. of x in the corresponding coefficient.) Euler observed that equations of this type are
is the solution to reduced to linear equations with constant coefficients with the substitution x = e'. *
(d) If possible, find a and b so that yet) has exactly
one local extremwn at t = 3. Graph yet) on the f y" + 4y = f(t)
mterval [0, 6]. [YeO) = 0, y'(O) = 2'
(b) Show.th~t iff(t) is constant, the resulting solution Definition 4.8 Cauchy-Euler Equation
(e) Is it possible to determine a and b so that y(l) has
local extrema
b ? if I = to and I = t I (I0 *- I 1 b0 th ar- IS pen odIc and find its period. Conf'mn your re- A Cauchy-Euler differential equation is an equation of the fonn
O
Itrary). Support your conclusion with several sult WIth a graph.

randomly generated values of 10 and 11' (e) If f(l) is periodic, is the resulting solution .
anxny<n) + an_IXn-ly<n-l) + ... + aIX)" + aoy = g(x),
odic?. A s m pen-
. (b) , confirm your result with a graph.
59. Solve the equation where ao, aj, a2, ... , an are constants.
*61. ~ne solution of 12y" - 4ty' + (12 + 6)y = 0 is y =
2
e -2'[y d dtY)2J -
dl y2 - (d 21(1 + t)y2 = 0
t co~ I. (a) Solve 12y" - 4ty' + (t 2 + 6)y = 13 + 2t
(b) ~I~d the solutions, if any, that satisfY the initi~ Second-Order Cauchy-Euler Equations
condltlOnsy(O) = 0 :mdy'(O) = I. (e) Does the result
~y making the substitution y = e"U). Graph the solu- found ID (b) contradIct the Existence and Uniqueness Consider the second-order homogeneous Cauchy-Euler equation
tIOn for various values of the constants. theorem? ExplalD.
3x 2y" - 2xy' + 2y = 0, x > O.
60. Show that the solution to the initial-value problem
62. One solution of ty" + 2y' + ty = 0 is y = ~
I . Here, the solution y is a function of x. We assume that x > 0 because we are guaran-
f y" + 4y = f(t)
[YeO) = O,y'(O) = 2
(a) Solve ty"
~at
+ 2y' + ty =-to (b) Find the soluti
satisfies the initial conditions y( 1T) = -I ;~ teed solutions where the coefficient functions in y" - ~
JX
y' + ~ y =
3x
0 are contin-
is y (7T) = .-li1T. (e) Find the solutions, if any, that sat- uouS. (Similarly, we can solve the equation for x < 0.)
yet) = sin(2t) + t sin(2l) r o
f(t) cos 21 dt
ISfY the mltlal conditions y (0) = y' (0) = O. (d) Does
the re~ult found m (e) contradict the Existence and
Suppose that a solution of this differential equation is of the fonny = x' for some
constant(s) r. Substitution of y = x' with derivatives y' = rx,-l and y" = r(r - 1)X,-2
Umqueness theorem? Explain.
into this ODE yields
- "21 cos (2t) ]'0 f(t) sin 2t dt 63. O~e s.olution of 4t 2y" + 4ty' + (161 2 - l)y = 0 is
y - (sm21)/Vt. (a) Solve412y" + 4ty' + (161 2 - I) 3x2[r(r - 1)x,-2] - 2x[rx,-l] + 2x' = 0
(as long as the integrals can be evaluated). : 16~312 (b) Find the solutions, if any, that satiSfY;' ".1 3r(r - I)x' - 2rx' + 2x' = 0
oun ary conditi~ns y(1T) = y(37T/2) = O. (e) Under
(a) VerifY this result by using the Fundamental The- what condItions, If any, do infinitely many solutions [3r(r - 1) - 2r + 2]x' = 0
orem of Calculus to show that satIsfY the boundary conditions y (7T) = y' (21T) = O?
(3r2 - 5r + 2)x' = O.
2
We are seeking a nontrivial solution, so we solve the quadratic equation 3r - 5r + 2
= O. Factoring gives us (3r - 2)(r - I) = 0, so the roots are rl = 2/3 and r2 = I. There-
I 4.71 Cauchy-Euler Equations fore, two solutions of this equation are X2l3 and X. These two functions are linearly in-
dependent because they are not constant multiples of one another, so a general solu-
(' Second-Order Cauchy-Euler Equations (' N h
Equations (' Higher Order Cauchy-Euler Equa~:n~mogeneous Cauchy-Euler
2/3
tion is y = CIX + C2 X •
This approach works nicely if the roots of the quadratic equation are real and
distinct. However, if we obtain a repeated root or complex conjugate roots, additional
In previous sections we solved r d ' ff< .
Generally, solving ::n
arbitrary :~ar ~. ~renllal equations with constant coefficients. work is required to detennine a general solution (see Exercises 43-45). Therefore, we
task, particularly when the coeff' I' erten la equallon is a fonnidable if not impossible refer back to Euler's idea by using the substitution
IC en s are not constants Howe bl
certain equations with variable c ff" . . ver, we are a e to solve
oe IClents usmg techniques sim'l t th di
previously. We begin by conside' d'ff<' . I ar 0 ose scussed * Carl B. Boyer, A History of Malhematics, Princeton University Press (1985), p. 496.
nng I erenllal equatIOns of the fonn
4.7 Cauchy-Euler Equations 199
t
x = e .
In. doing t.his,. the c~ain rule must be used to transform the derivatives with respect to
l'f'"I,'.,
x Into denvatIves WIth respect to t. If x = e t , t = In x, so Solve xZy" - 5xy' + lOy = 0, y(1) = I, y'(I) = O.
: ] Solution After substituting the appropriate derivatives, we obtain the equation
Z[-xI2 (d- d )] - 5 [I
Y 2y
and x -- Y ] + lOy = 0
x -d- or
dt 2 dt x dt
The characteristic equation is r2 - 6r + 10 = 0 with complex conjugate roots

r = 4(6 ± v'36 - 40) = 3 ± i. A general solution of the transformed differential
equation is yet) = e 3'(cI cos t + C2 sin t). We can transform the initial conditions as
well. If x = I as indicated in the two initial conditions, then t = In I = O. There-
fore, in terms of t, the initial conditions are yeO) = I and y' (0) = 0, and the trans-
Substitution o~ these .derivat!ves into a second-order Cauchy-Euler equation yields a formed initial-value problem is
second-order lInear dIfferential equation with constant coefficients. y
2
7d y - 6 d(
dy
+ lOy = O,y(O) = l,y'(O) = O.
,'e·",,'.' 2
Solve x y" - xy' +y = 0, x > O.
0.5
-0.5
2 x The characteristic equation is r2 - 6r + 10 = 0 with complex conjugate roots
r = 4(6 ± \1'36 - 40) = 3 ± i, so the general solution of the transformed differen-
-1 tial equation is yet) = e 3'(cI cos t + C2 sin t). Then, yeO) = CI = I. With
3t
! Solution Substituting the derivatives above, we have -1.5 y'(t) = 3e (ct cos t + C2 sin t) + e 3t( -Cj sin t + C2 cos t),
X2[:2 (~;~ - ~)J - [~ ~J

x +y = 0
-2
Figure 4.16
we find that y'(O) = 3cI + C2 = 0, so C2 = -3ct = -3. Therefore, y(t) =
e 3t(cos t - 3 sin t). In terms of x, this solution isy(x) = x 3(cos(lnx) - 3 sin(lnx)),
which is graphed in Figure 4.16.
2
d y dy dy
dt 2 -d(-d(+y=O
~§:~ What is limx_o+ y(x)?
2
d y dy
dt2 - 2d( + Y = O. C)J~ -
Theorem 4.14 Solving Second-Order Homogeneous Cauchy-Euler Equations

This equation has ~o~stant c~efficients, so solutions are of the form y = e't. In this
case, the charactenstrc equatIon is r2 - 2r + I = (r - Il
= 0 with repeated root Consider the second-order Cauchy-Euler equation ax 2y" + bxy' + cy = 0,
rl =. r2 = 1. There.fore, a general solution of this equation is yet) = clet + c2te t. Re. x > O. With the substitution x = e t , this equation is transformed into the second-
turnl.ng to the ongInal variable, x = e' (or t = In x), we have y(x) = CtX + C2 X Inx. order differential equation with constant coefficients, .a(d2y/dt 2 ) + (b - a)
In FIgure ~.15, we graph the solution for several values of CI and C2' Notice that (dy/dt) + cy = O. Let rl and r2 be solutions of the equation ar z + (b - a)r +
Figure 4.15
each functIOn approaches zero as x ---> 0 + . C = O.
(a) If rl
r
*' rz are real, a general solution of the Cauchy-Euler equation is y(x) =
CIX ! + C2 xr2 .
Graphs of the solutions corresponding to various values of CI and C2 are shown in
(b) If rio r2 are real and rl = rz, a general solution is y(x) = CIX" + C2X" In x.
FIgure 4.15. Regardless of the choice of CI and Cz, calculate lirnx_o+ (CjX + C2 X In x).
Why lS a rzght-hand limit necessary? Are the solutions in Figure 4.15 bounded as
*'
(c) If rl = r; = a + f3i, f3 0, a general solution is
X~ oo? y(x) = Xa(CI cos(f3ln x) + C2 sin(f3 In x)).
200 Chapter 4 Higher Order Equations 4.7 Cauchy-Euler Equations 201
Nonhomogeneous Cauchy-Euler Equations ditions is given by y(x) = x 2 (Wo

cos(3 In x) - sin(3 In x» + + m and then fr !oX
approximate the first value ofx greater than 10for whzch (a) y (x) - 0 and
Just as we encountered in nonhomogeneous linear differential equations with constant
(b) y(x) = O.
coefficients, the methods of variation of parameters and undetermined coefficients can
be used to solve nonhomogeneous Cauchy-Euler equations. The following two exam_ When solVing nonhomogeneozts Cauchy-Euler equations, the method of undetermined
ples illustrate two different approaches to solving nonhomogeneous Cauchy-Euler equa- coefficients should be used only when the equation is transformed to a constant co-
tions: (I) solving the transformed equation with the method of undetermined coeffi_ efficient equation. On the other hand, the method of variation ofparameters ca~ be
cients, and (2) solving the equation in its original form using variation of parameters. used with the original equation in standard form or with the transformed equatzon.
l'i'd'.".1 "ii"I,'.'
Solve xV' + 4.xy' + 2y = In x, y(l) = 2, y'(1) = O.
2
Solve x y" - 3.xy' + l3y = 4 + 3x, x > O.
:j Solution We begin by determining a solution of the
2
co~espondingo
homogeneous
. f 2y" + 4.xy' + 2y = 0, which is transformed mto d yldt- + 3(dyldt) +
: J Solution After substitution of the appropriate derivatives and functions of t, we
obtain the differential equation
equa
2y = IOn . r + 2) -.:..
x characteristic equation r 2 + 3r + = ( r + 1)(
0 with 3 - 0 Therefore '
Yh(t) = c,e- t + C2e-2t,SOYh(X) = c,x- I + C2X .2.Ass~.mgthatyp(x) - uI(x)y,(x)
2 I (d
2y
dY ) I dy t + U2(x)YzCX) = UI(X)X- I + U2(X)X -2, we fi~d usmg vanatlOn of parameters With the
X :? dt2 - dt - 3x ~ dt + 13y = 4 + 3e , equationy" + (4Ix)y' + (2Ix2)y = (lnx)lx- that
2
which in simplified form is d yldt 2 - 4(dyldt) + 13y = 4 + 3e t. The characteristic 4 X -3 In x
equation of the corresponding homogeneous equation is r2 - 4r + 13 = 0, which has u\(x) =
-x- Inx=lnx
_x- 4 and uz(x) = - - - - 4 - = -x In x.
-x
complex conjugate roots r[,2 = 2 :t 3i and general solution Yh(t) = e 2t (c[ cos 3t +
C2 sin 3t). Using the method of undetermined coefficients to solve the equation Then, u[ (x) = f In x dx = x In x - x and U2(X) = - f 2 2
x In x dx = 114 x - 1/2 x In x,
2 so
d y dy
-2 - 4 - + 13y = 4 + 3e t,
dt dt
yp(x) = uI(x)y[(x) + U2(X)Y2(X) = (x In x - x ) x -I + (1.4 x 2 _ 1.2 x 2Inx)x-2
we assume that the particular solution is Yit) = A + Bet. The derivatives of
this function are (dyJdt)(t) = Be' and (d 2Y/dt 2 )(t) = Bet; thus, substitution into
2 2
d yldx - 4(dyldt) + 13y = 4 + 3e' yields
= 1. Inx _1..
2 4
t
Be' - 4Be + I3A + I3Be t = IOBe t + I3A = 4 + 3e t.
Therefore, B = fo and A = 1" so Yit) = 1, + foe'. A general solution in the variable
I In _1.
t is given by + yix ) =
-[
+ C2 X
-2
+ 2"
0.5\0;'1.5 1.5 y(x) = Yh(X) c[x x 4'
t·
2 x x
5
-10
-1 Y (t) = e 2'(c, cos 3t + C2 sm
. 3) 4 + 10
t + 13 3 et.
-20 h
were Y '(x) = -CIX -2 - 2c2 X -3 + 1I(2x) • Applying the initial
. .
conditions,
th we find
t
-30 that (1) = c + C2 - 3/4 and y'(I) = -c, - 2C2 + 112. Solvmg e sys em
Returning to the original variable x = et (or t = In x), we have
-40 {c[ 1C2 - 3/~ = 2, -c[ - 2C2 + 112 = O} indicates that c[ = 5 and C2 = -9/4, so
-50 the solution to the IVP is
Figure 4_17
y(x) = X\CI cos(3 In x) + C2 sin(3 In x» + I~ 3
+ 10 x.
y(x) = Yh(X) + Yp(x) = 5x
-, 9 _2+11
- 4"x 2" nx
_1.4·
Figure 4_18 Graph of y(x) =
This general solution is plottedfor various values of the arbitrary constants CI and 5x-,-2.x-2+.!.lnx-~ We graph this function in Figure 4.18. The solution becomes unbounded as x --> 0+.
C2 in Figure 4.17. IdentifY the graph of the solution that satisfies the initial condi- 424
tions y(I) = 2 and y'(1) = O. Show that the solution that satisfies these initial con-
4.7 Cauchy-Euler Equations
-
Higher Order Cauchy-Euler Equations EXERCISES 4.7
Solutions of higher order homogeneous Cauchy-Euler equations are determined in the
same manner as solutions of higher order homogeneous differential equations with COn.
stant coefficients.
In Exercises 1-20, find a general solution of the Cauchy·Euler In Exercises 31-42, solve the initial-value problem.
"",,"'"
equation. 3X'y" - 4xy' + 2y = 0
1. 4x 2y" - &y' + 5y = 0 31. { y(l) = 2, y'(1) = I
Solve 2x 3y'" - 4x'y" - 20xy' = 0, x > O. 2. 3x 2y" - 4xy' + 2y = 0 2x'y" - 7xy' + 7y = 0
32. {y(l) = -1,y'(1) = 1
*3. 2x 2y" - 8xy' + 8y = 0
: J Solution In this case, if we assume that y = x' for x > 0, we have the deriva. x 2y " +xy' + 4y = 0
4. 2x 2y" - 7xy' + 7y = 0
. tives y' = rx,-l, y" = r(r - I)x'-', and y'" = r(r - I)(r - 2)X,-3. Substitution *33. {y(l) = 1, y'(I) = 0
into the differential equation and simplification yields (2r 3 - lOr' - 12r)x' = O. 5. 4x 2y" + 17y = 0
x 2y " + xy' + 2y = 0
Because x' '" 0, we must solve 6. 9x 2y" - 9xy' + lOy = 0
34. {y(l) = O,y'(I) = Z
7. 2x 2y" - 2xy' + 20y = 0
(2r 3 - lOr' - I2r) = r(2r + 2)(r - 6) = 0 x 3y '" + 10x2y" - 20xy' + 20y = 0
8. x 2y" - 5xy' + lOy = 0
for r. Hence, rl = 0, r2 = -1,_ and r3 = 6, and a general solution is 35. { yell = O,y'(1) = -1,y"(I) = 1
y = CI + C2X-1 + C3x6. *9. x y" 2
+ 2xy' - 6y = 0
x3y m + 15x 2y" + 54xy' + 42y = 0
10. x 2y" + xy' - 16y = 0 36. {yell = 5. Y '(1) = 0, y"(1) = 0
11. x 2y" + xy' + 4y = 0
Find and classifY all relative extrema ofy(x) in the interval (1, 4) ify satisfies the x3y m - 2x 2y" + 5xy' - 5y = 0
12. x 2y" + xy' + 36y = 0
boundary conditions y(I) = y(4) = 0 (see Figure 4. 19). *37. {yell = 0, y'(1) = -1, y"(l) = 0
13. x'y" + 5xy' + 13y = 0
x3y m - 6x 2y" + 17xy' - 17y = 0
If a root r of the auxiliary equation is repeated m times, the m linearly inde.
14. x 2y" + 2xy' + Y = 0 38. { y(l) = -2, y'(I) = 0, y"(1) = 0
pendent solutions that correspond to r are x', x' In x, x'(ln x)', ... , x'(ln x)m-l be.
*15. x 3y'" + 22x2y" + 124xy' + 140y = 0 2x'y" + 3xy' - y = x-,
cause in the transformed variable I, the solutions are e", Ie", I'e", ... , I'-le".
16. x 3y'" - 4x 2y" - 46xy' + 100y = 0 39. {y(1) = 0, y'(I) = 2
!'ii,,','. 17.
18.
x 3y'" + 2x'y" - 4xy' + 4y = 0
x 3y'" + 4x 2y" + 6xy' + 4y = 0
x 2y " + 4xy' + 2y = In x
40. {y(l) = 2,y'(I) = 0
Solve x 3y'" + xy' - Y = 0, x > O. *19. x 3y'" + 2xy' - 2y = 0 4X"y" + Y = x'
20. x 4y(4) + 9x 3y''' + llx 2y" - 4xy' + 4y = 0 *41. {y(l) = l,y'(I) =-1
: ] Solution As in Example 5, we substitute y = x' into the equation. This gives us
3 2 9X 2y " + 27xy' + lOy =
x'(r - 3r + 3r - 1) = 0, where r3 - 3r' + 3r - 1 = (r - Ii, so the roots are X-I
In Exercises 21-30, solve the nonhomogeneous Cauchy-Euler 42. {y(1) = O,y'(l) = -1
rl = r2 = r3 = 1. One solution to this equation is x, a second is x In x, and a third equation.
is x(ln x)'. Therefore, a general solution is y = CIX + C,X In x + C3X(1n xf. 43. Suppose that a second-order Cauchy-Euler equation
21. x 2y" + 5xy' + 4y = x- 5 has solutions x'" and xr'2, where rl and r2 are real and
22. x 2y" - 5xy' + 9y = x 3 unequal. Calculate the Wronskian of these two solu-
tions to show that they are linearly independent. There-
*23. x 2y" + xy' + Y = x- 2 fore, y = C1X r1 + C2xr2 is a general solution.
24. x 2y" + xy' + 4y = x- 2 44. (a) Show that the transformed equation for ~e
25. x 2y" + 2xy' - 6y = 2x Cauchy-Euler equation ax 2y" + bxy' + cy = 0 IS
Figure 4.19 Several solutions a(d 2yldt') + (b - a)(dyldt) + cy = O. (b) Show that
to the 2x 3y 'll - 4x 2y" - 20xy' = 26. x 2y" + xy' - 16y = In x
if the characteristic equation for a transformed
o equation that satisfy the *27. x 2y" + xy' + 4y = 8
Cauchy-Euler equation has a related root, then this
boundary conditions yell = 28. x'y" + xy' + 36y = x'
y(4) = 0 root is r = (-(b - a)/Za). (e) Use reduction of order
*29. x 3y'" + 3x 2y" - llxy' + 16y = x- 3 to show that a second solution to this equation is x'
30. x 3y'" + 16x'y" + 70xy' + 80y = X-
13 In x. (d) Calculate the Wronskian of x' and x' In x to
l~···
204 Chapter 4 Higher Order Equations 4.8 Series Solutions of Ordinary Differential Equations 205
show that these two solutions are linearly independ- 55. x 2y"+ 4xy' + 2y = 0, x < 0 (d) If both a and b are not zero, is it possible to I(x, y, y')y" + g(x, y,y') = 0
ent.
2
56. x y" + xy' + Y = x 2 , X < 0 find a and b so that both limx-+o+ y(x) = 0 and
45. Suppose the roots ofthe characteristic equation for the and that a general solution of
2 limx_ y(x) = O? Explain.
transformed Cauchy-Euler equation are r = '" ± f3i, 57. x y" + xy' + 4y = O,y(-I) = O,y'(-I) = 2 X(y')2 + yy' + W" = 0
65. An equation of the form/(x,y,y')y" + g(x,y,y') =
so that two solutions are x a + f3; andx a -{3;. Use Euler's 2
58. x y" - xy' + Y = O,y(-l) = O,y'(-l) = 1
formula and the Principle of Superposition to show
o that satisfies the system is y2/2 = C1 + C2 In x.
a 59. Consider the second·order Cauchy-Euler equation
that x cos(f3lnx) and x a sin(f3lnx) are solutions /= + 2plxy + p2/::: = gxP + pgyp - gy, 66. (a) Find a general solution of the nonlinear equation
of the original equation. Hint: x{3; = (e 'nx )"; = x 2y " + Bxy' + Y = 0, x > 0, where B is a constanl { /xP + p/yp + 2/y - gPP ~(y' _ xy',), = 2yy" _ (y')2
cos(f31n x) + i sin(f31n x). (a) Find limx-+ooY(x) wherey(x) is a general solution
where p = y' is called an exact second-order differ- by differentiating both sides of the equation, set·
46. The solution in Example 4 isy(x) = 5x- 1 - £x- 2 + using the given restriction on B. ential equation. (a) Show that the equation
! In x - ~ = (-9 + 20x - 3x 2 + 2x 2 In x)/4x 2 • Use ting the result equal to zero, and factoring. (b) Is
(b) Determine if the solution is bounded as x -> 00 in
I:Hopital's Rule to show that limx-+oo y(x) = +co. X(y')2 + yy' + W" = 0 the Principle of Superposition valid for this non-
each case: (i) B = 1; (ii) B > I; (iii) B < I. linear equation? Explain.
47. Consider the equation x y" + Axy' + By = 0, x > 0,
2 is an exact second-order equation. (b) Show that
where A and B are constants. Solve the equation
for each of the following. Investigate limx...o+ y(x) and
60. Consider the second-order Cauchy-Euler equation
stated in Exercise 59. Determine limx-+o+ y(x) as ",(x, y, p) = hex, y) + f I(x, y, p) dp is a solution
well as if y(x) is bounded as x -> 0+ in each case: of the exact equation
limx-+ooY(x).
(a)B= I; (b)B> 1; (c)B< I.
(a) A = -1,B=2
61. Use variation of parameter to solve the nonhomoge.
(b) A = 4, B = 2
(e) A = I, B = I
neous differential equation that was solved in Exam.
pIe 3 with undetermined coefficients. Which method
~ Series Solutions of Ordinary Differential Equations
48. Use the results of Exercise 47 to show that solutions is easier? , Series Solutions About Ordinary Points 'Legendre's Equation
2
to x y" + Axy' + By = 0, x > 0 (a) approach zero as 62. Use a computer algebra system to assist in finding a (' Regular and Irregular Singular Points and the Method of Frobenius
x -> 0+ if A < 1 and B> 0; (b) approach zero as general solution of the following Cauchy-Euler equa. , Gamma Function (' Bessel's Equation
x -> co if A > 1 and B> 0; (c) are bounded as tions.
x-> 0+ andx-> co if A = I andB> O.
(a) x 3y'" + 16x 2y" + 79xy' + 125y = 0 In calculus, we learned that Maclaurin and Taylor polynomials can be used to approx-
49. Show that if x = e' and y = y(x), then ~{ = 4
(b) x y(4) + 5x 3y'" - 12x 2y" - 12xy' + 48y = 0
imate functions. This idea can be extended to finding or approximating the solution of
a differential equation.
1 (d
3y 2
d y dY ) (c) x 4y(4) + 14x 3y'" + 55x 2y" + 65xy' + 15y = 0
7' dt 3 - 3 dt2 + 2 dt . 4 4
(d) x / ) + 8x 3y'" + 27x 2y" + 35xy' + 45y = 0 Series Solutions About Ordinary Points
4
(e) x y(4) + IOx 3y'" + 27x 2y" + 21xy' + 4y = 0
In Exercises 50-53, use the substitution in Exercise 49 to solve Consider the equation a2(x)y" + al(x)y' + ao(x)y = 0 and let p(x) = al(x)la2(X) and
the following equations. 63. (a) Solve the initial-value problem q(;:) = aO(x)la2(x). Then, a,(x)y" + al(x)y' + ao(x)y = 0 is equivalent to y" + p(x)y'
SO. x y'"
3 + 3x 2y" + 37xy' = 0, x > 0 . x 3y '" + 9x"y" + 44xy' + 58y = 0
+ q(x)y = 0, which is called the standard form of the equation. A number Xo is an
3 ordinary point of the differential equation if both p(x) and q(x) are analytic at Xo. If
51. x y'" + 3x 2y" - 3xy' = 0, x > 0 {y(1) = 2,y'(1) = lO,y"(I) = -2'
Xo is not an ordinary point, Xo is called a singular point. Note: A function is analytic
52. x 3y'" + 3x 2y" + 7xy' + Y = 0, x > 0 (b) Graph the solution on the interval [0.2, 1.8] and at Xo if its power series centered at x = Xo has a positive radius of convergence.
53. x 3y'" + 3x 2y" - 3xy' = -8, x > 0 approximate all local minima and maxima of the If Xo is an ordinary point of the differential equation y" + p(x)y' + q(x)y = 0,
solution on this interval. p(n)(xo) n
L L
00 00
54. Letx = -e'. (a) Show that dy = dy dt =..!:.. dYand we can write p(x) = bn(x - xo)", where b n = - - , - , and q(x) = cn(x - xo) ,
. dx dt dx x dt 64. (a) Solve the initial-value problem n=O n. n=O
2 2y
d y _ I (d dY ) '. 6X 2y " + 5xy' - Y = 0
(nl( )
dx 2 - x2 dt2 - dt . (b) Show that the different131 where Cn = ~,xo. Substitution into the equation y" + p(x)y' + q(x)y = 0 results in
{y(1) = a, y'(1) = b . n.
2
equation ax y" + bxy' + cy = I(x) is transformed
. d 2y dy (b) Find conditions on a and b so that limx-+o+ y(x) y" + y' L bn(x - xo)" +Y L cn(x - xo)" = O.
mto a dt 2 + (b - a) dt + cy = I( -e'). = O. Graph several solutions to confirm your reo n=O n=O
suIts.
In Exercises 55-58, use the substitution described in Exercise (c) Find conditions on a and b so that limx-+ooY(x) =
If we assume that y is analytic at xo, we can write y(x) = L an(x -
n=O
xo)". Because a
54 to solve the indicated equation or initial-value problem. O. Graph several solutions to confirm your results. power series can be differentiated term by term, we can compute the tITst and second
derivatives of y and substitute back into the equation to calculate the coefficients a indices in the series, n 2: 2. We use this formula to determine the values of an for
Thus, we obtain a power series solution of the equation. ,. n = 2, 3, ... , 10, and give these values in Table 4.3.
",i,,','MI TABLE 4.3

n n
(a) Find a general solution of (4 - x 2 ) dyldx +Y = 0 and (b) solve the initial-value
(4 - x 2 ) dyldx + Y = 0 o 6
pro bl em { yeO) = 1 .
187
: ] Solution (a) Because x = 0 is an ordinary point of the equation, we assume that 7 - 262144 ao
y= ~;;'~o a,x'. Substitution of this function and its derivatives into the equation
gives us 1 1843
2 3zao 8388608 Qo
4859
9 - 33554432 ao
= f
n=1
4n~'nxn-l - i
n=1
nanx n+ l + f
n=O
anx" = O. 4 10
12767
268435456 ao
Note that the fIT~t terms in these three series involve xc, x 2 , and xc, respectively. If 31
5 - 8192 Qo
we pull off the first two terms in the first and third series, all three series will be-
gin with x 2 • Doing so, we have
(4a l + ao) + (8a2 + al)x + L 4nanxn~1 - L nanx n+ + L anx n = O.

1
Therefore,
n=3 n=1 n=2
1 2 3 3 11 4 31 5 69 6
Now the indices of these three series do not match, so we change two of the three Y = ao - 4aox + 32aox -128aox + 2048 aox - 8192 aox + 65536 aox
to match the third. Substitution of (n + 1) for n in ~;;'~3 4nanxn~1 yields
187 1843 4859 12767 +
i 7 8 9 10
n+l=3
4(n + l)an+IXn+I~1 = f
n~2
4(n + l)a n+IXn. - 262144 aox + 8388608 aox 33554432 aox + 268435456 aox ...
-Polynomial Approximation (b) When we apply the initial conditiony(O) = 1, we substitute x = 0 into the so-
Similarly, substitution of (n - I) for n in ~;;'~ I nanx n+I yields
_y ~IX~21/4 lution obtained in (a)_ Hence, ao = 1, so the series solution of the initial-value prob-
i
n-I=1
(n - l)an~lxn~I+1 = i
n=2
(n - l)an~IXn.
x+2
y
lem is
1 1 2 3 3 11 4_..2l.. 5 ~ 6
After combining the three series, we have the equation 4 Y= 1- 4x+ 32 x -128 x + 2048 x 8192 x + 65536 x
+
+ ao) + (8az + al)x + L
187 7 1843 8 4859 9 12767 10
(4a l (an + 4(n + l)an+l - (n - I)an~l)xn = O. - 262144 x + 8388608 x - 33554432 x 268435456 x + ...
n~2
Equating the coefficients of Xo and x to zero yields The equation (4 - x 2 ) dyldx +Y = 0 is a separable first-order equation. Separat-
- 2[1/4
Figure 4.20 Comparison of ing, integrating, and applying the initial condition yields y = [~ + 2 . We can ap-
and
exact and approximate solutions proximate the solution of the problem by taking a finite number of terms of the se-
n to the initial-value problem
When the coefficient ofx , an + 4(n + l)a n+1 - (n - l)an~b n 2: 2, is set to zero, (4 ~ x 2 ) dy/dx + Y = 0 ries solution. The graph of the polynomial of degree ten is shown in Figure 4.20
we obtain the recurrence relation an+l = ((n - l)an~1 - a n)/4(n + 1) for the { yeO) = 1 along with the solution obtained through separation of variables. Notice that the ac-
curacy of the approximation decreases near x = 2 and x = - 2, which are sin~

points of the differential equation. (Why?) The reason for this is discussed in the
"fi"i".,
following theorem. Find a general solution of Legendre's equation (1 - x 2 )y" - 2xy' + k(k + I)y = O.
The following theorem explains where a series solution of a differntial equation

.
:. Solution
. ." 2x
In standard form, the equation IS y - 1 _ x 2 Y
, +~
k(k + 1) - 0
Y- .
is valid.
Because x = 0 is an ordinary point, there is a solution of the form y = L:~O anxn.
This solution will converge at least on the interval ( -I, I), because the closest sin-
gular points to x = 0 are x = ± I. Substitution of this function and its derivatives
Theorem 4.15 Convergence of a Power Series Solution into the differential equation yields
Let x = Xo be an ordinary point of the differential equation

a2(x)y"(x) + al(x)y'(x) + ao(x)Y(x) = 0,
i
n=2
n(n - l)a nx n- 2 - i
n=2
n(n - I)anx n - i
n= 1
2nanx
n
+ t
n-O
k(k + I)anxn = 0
[2a2 + k(k + I)ao]x o + [-2al + k(k + I)al + 6a3]X

and suppose that R is the distance from x = Xo to the closest singular point of
the equation. Then the power series solution y = L:~O a.(x - xo)" converges at
least on the interval (xo - R, Xo +"R). *
+ i
n=4
n(n - l)a nx·- 2 - i
n=2
n(n - I)anx' - i
n=2
2na.x n + i
n=2
k(k + l)a.xn = O.
After substituting (n + 2) for each occurrence of n in the first series and simplify-
ing, we have
Series solutions have been
This theorem indicates that a polynomial approximation may not be accurate near
used since the days of New-
singular points of the equation. Now we understand why the approximation in Exam- [2az + f..:(k + I)ao]x o + [-2al + k(k + I)al + 6a3]X
ton. At that time, however.
mathematicians did not con-
cern themselves with such is-
ple I breaks down near x = 2 and x = - 2; these are the closest singular points to the
ordinary point x = O. + i
11=2
{en + 2)(n + l)a n+2 + [-n(n - I) - 2n + k(k + n
I)]a.}x = O.
sues as convergence, which is Of course, x = 0 is not an ordinary point for every differential equation. How.
essential to successful applica- ever, because the series y = L:~o an(x -·xo)" is easier to work with if Xo = 0, we can Equating the coefficients to zero, we find that
tion of the method. always make a transformation so that we can use y = L:~O anx n to solve any linear
k(k + I) k(k+I)-2 (k - I)(k + 2)
equation. For example, if x = Xo is an ordinary point of a linear equation and we make a2 = ---2--ao, a3 = 6 al = 6 a),
the change of variable t = x - Xo, then t = 0 corresponds to x = Xo. Hence, t = 0 is an
ordinary point of the transformed equation. and
_ n(n - I) + 2n - k(k + I) = (n - k)(n + k + I) an"'" 2.
Legendre's Equation a n+2- (n +2)(n + I) an (n+2)(n+l) m
Legendre's equation is
Using this formula, we find the following coefficients:
(I - X2)y" - 2xy' + k(k + l)y = 0,
(2 - k)(3 + k) _ (2 - k)(3 + k)k(k + I)
where k is a constant, named after the French mathematician Adrien Marie Legendre a4 = 4 .3 a2 - - 4.3.2 ao
(1752-1833). The Legendre polynomials, solutions of Legendre's equation, were in·
(3 - k)(4 + k) (3 - k)(4 + k)(k - I)(k + 2)
troduced by Legendre in his three-volume work Traite des fonctions elliptiques et des as = 5.4 a3 = - 5.4.3. 2 al
integrales euleriennes (1825-1832). Legendre encountered these polynomials while
trying to determine the gravitational potential associated with a point mass. (4 - k)(5 + k) (4 - k)(5 + k)(2
- k)(3 + k)k(k + 1)
a6 = 6.5 a4 = - 6.5.4.3.2 ao
* A proof of this theorem can be found in more advanced texts, such as Introduction to Ordinary Differ- (5 - k)(6 + k) (5 - k)(6 + k)(3 - k)(4 + k)(k - I)(k + 2)
ential Equations by Albert L. Rabenstein, Academic Press (1966), pp. 103-107. a7 = 7.6 as =- 7 . 6 . 5 ·4· 3 . 2 al'
TABLE 4.4 Pn(x) for n = 0, We have the two linearly independent solutions
1, ... , 5 Theorem 4.16 (Method of Frobenius)
_ ( k(k + 1) 2 (2 - k)(3 + k)k(k + 1) X4
YI (X) - ao 1 - --2-1- x - 4!
Po(X) = 1 Let Xo be a regular singular point of y" + p(x)y' + q(x)y = O. Then this differ-
(4 - k)(5 + k)(2 ~lk)(3 + k)k(k + 1) x 6 _ ... )
ential equation bas at least one solution of the form
~
1
P2(x) = z(3x 2 - 1) y = I an(x - xo)"+r,
and n=O
_ ( (k - 1)(k + 2) 3 (3 - k)(4 + k)(k - 1)(k + 2) 5 where r is a constant that must be determined. This solution is convergent at least
Y2 (x ) - al x - 6 x - 5! x on some interval Ix - xol < R, R > O.
1
P4 (x) = S(35x 4 - 30x 2 + 3) _ (5 - k)(6 + k)(3 - k)(4 + k)(k - 1)(k + 2) x 7 _ ... )
7! '
Suppose that x = 0 is a regular singular point of the differential equation y" +
1
P,(x) = S(63x' - 70X' + 15x) so a general solution is p(x)y' + q(x)y = O. Then the functions x p(x) and x 2q(x) are analytic, which means that
y=ao (l- k(k+ 1) x 2
2'
-
(2 - k)(3 + k)k(k +
4!
1) 4
x -
both of these functions have a power series in x with a positive radius of convergence.
Hence,
(4 - k)(5 + k)(2 ~,k)(3 + k)k(k + 1) x 6 _ ... j xp(x) = Po + PIX + P2 X2 + ... and

. J Therefore,
+ ( (k - 1)(k + 2) 3 (3 - k)( 4 + k)(k - 1)(k + 2) 5
al x - 6 x- x
51
(5 - k)(6 + k)(3 - k)(4 + k)(k - 1)(k + 2) x7 _ ... ) and

7! .
An interesting observation from the general solution to Legendre's equation given Substitution of these series into the differential equationy" + p(x)y' + q(x)y = 0 and
here is that the series solutions terminate for integer values of k. If k is an even inte. multiplying through by the first term in the power series for p(x) and q(x), we see that
ger, then the first series terminates, while if k is an odd integer the second series ter- the lowest term in the series involves x n + r - 2 :
minates. Therefore, polynomial solutions are found for integer values of k. We list sev.
eral of these polynomials for suitable choices of ao and al in Table 4.4 and graph them
in Figure 4.21. Because these polynomials are useful and are encountered in numer.
ous applications, we have a special notation for them: Pn(x) is called the Legendre
Figure 4.21 Po(x), PI (x), polynomial of degree n and represents the nth-degree polynomial solution to + (PI + P2X + P3 X2 + P4 X3 + "')C~o anCn + r)x n+r- I) + (~o anqOxn+r-2)
P 2(x), P,(X), P 4 (x), and P,(x) Legendre's equation.
Match each Legendre polynomial to the appropriate graph in Figure 4.21. + (~l + q2 + q3X + q4x2 + q5x3 + .. ·)C~o anxn+r) = O.
Regular and Irregular Singular Points and the Then, with n = 0, we find that the coefficient of x r - 2
is
Method of Frobenius
-rao + r2ao + raoPo + aoqo = ao(r2 + (Po - l)r + qo) = ao(r(r - I) + por + qo).
Let Xo be a singular point of y" + p(x)y' + q(x)y = O. Then Xo is a regular singular
point of the equation if both (x - xo)p(x) and (x - xo)2q(X) are analytic at x = Xo. If Thus, for any equation of the formy" + p(x)y' + q(x)y = 0 with regular singular point
Xo is not a regular singular point, then Xo is called an irregular singular point of the x = 0, we have the indicial equation
equation.
r(r - 1) + por + qo = O.
The values of r that satisfy this equation are called the exponents or indicial roots and Solution In standard form, this equation is y" + (3/x)y' - (J/x)y = o. Hence,
are xp(x) = x(3/x) = 3 and x 2q(x) = x 2( -lIx) = -x, so Po = 3 and qo = O. Thus, the
rl = +(1 - Po +V I- 2po + P6 - 4q o)
indicial equation is r(r - 1) + 3r = r2 + 2r = r(r + 2) = 0 with roots rl = 0 and
r2 = -2. (Notice that we always use rl to denote the larger root.) Therefore, we
attempt to find a solution of the form Y = ~;;'~o an x n - 2 with derivatives y' =
and ~;;'~o (n - 2)a nx n- 3 andy" = ~;;'~o (n - 2)(n - 3)a nx n- 4 . Substitution into the dif-
r2 = +(1 - Po - V I - 2po + P5 - 4q o).

ferential equation yields
f (n - 2)(n - 3)a nx n- 3 + f 3(n - 2)a nx n- 3 - f a nx n-

2
= 0
Note that rl 2: r2 and rl - r2 = VI - 2po + P5 - 4qo. n=O n=O n=O
Several situations can arise when rmding the roots of the indicial equation. (6ao - 6ao)x-3
I. If rl *" r2 and rl - r2 = VI - 2po + pi; - 4qo is not an integer, then there are two
+ f (n - 2)(n - 3)a nx n- 3 + f 3(n - 2)anx n- 3 - f n 2
anx - = 0
linearly independent solutions of the equation of the form Yl (x) = x" I anx n and n=l n=l n=O
00 n=O
hex) = x" I bnxn.
L
"
00
(n - 2)(n - 3)anxn- 3
00
+ L" 3(n - 2)a nx n --3Lan-Ix

"
00
n-3 = 0
n~O r---~..--~---- n=i n=1 '1=1
2. If rl *" r2 and rl - r2 = V I - 2po + P5 - 4qo is an integer, then th;re are two
linearly independent solutions of the equation of the form Yl (x) = x" I anx" and I {[en - 2)(n - 3) + 3(n n
- 2)]a n - a n_dx - 3 = O.
00 n=O n=1
heX) = CY1(X) In x + x" )' b"x n. Equating the coefficients to zero, we have
n"':;"'O
3. If rl - r2 = VI - 2po + P5 - 4qo = 0, then there are two linearly independent a = an-I =~,n::::l,n*2.
n (n - 2)(n - 3 + 3) n(n - 2)
solutions of the problem of the form Yl(X) = x"
00
I
n=O
anx n and Y2(X) = Yl(x) In x +
Notice that from this formula, al = -ao. When n = 2, we refer to the recurrence
xr! I bnxlZ. relation n(n - 2)an - an-I = 0 obtained from the coefficient in the series solution.
n=O When n = 2, 2(0)a2 - al = 0, which indicates that al = O. Because al = -ao, ao
In any case, if Yl (x) is a solution of the equation, a second linearly independent solu- = o. However, ao *"
0 by assumption, so there is no solution of this form.
tion is given by Because there is no series solution corresponding to r2 = - 2, we assume there
is a solution of the form Y = ~;;'~o anx n corresponding to rl = 0 with derivatives
y' = ~;;'~I nanx n- 1 and y" = ~;;'~2 n(n - l)a nx n- 2. Substitution into the differen-
e-Jp(x)dx
Y2(X) = Yl(X) f [Yl(x)f dx,

tial equation yields
00 00
which can be obtained through reduction of order.
When solving a differential equation in Case 2, first attempt to find a general
I n(n - l)anx n- 1 + I 3na nx n- 1 - I n
anx = 0
n=2 n=1 n=O
solution using Y = x" ~;;'~o anx n = ~;;'~o anx n+", where r2 is the smaller of the two
roots. A general solution can sometimes be found with this procedure. However, if the
contradiction ao = 0 is reached, find solutions of the formYl(x) = x" ~;;'~o anx n and
(3al - ao)x° + f
n=2
n(n - l)a nx n-' + f
n=2
n 1
3nanx - - f
n=1
n
anx = 0
Y2(X) = cY1(x) In x + x" ~;;'~o bnxn. 00
(3a, - ao)x° +I ([n(n - 1) + 3n]an - n

an_dx -' = o.
n=2
"fi"I,'. Equating the coefficients to zero, we have a, = 1ao and

Find a general solution of xy" + 3y' - Y = 0 using a series expansion about the a = an-l -~ n>2
regular singular point x = o. n n(n - 1) + 3n - n 2 + 2n'
- .
TABLE 4.5 We use this formula to calculate several coefficients in Table 4.5 and use them to Now we choose 00 = IIc, so
2 3
+ x24
cI (I"3 + 12 + 120
form YI(X) = 00
X
I + "3 x
+ 360 )
+ ....
2
( and yi(x) = Xx + .... )
o ao To determine a second linearly independent solution, we assume that

Substitution into the previous equation then yields
+I bn x n - 2
Y2(X) = CYI(X) In x
n=O
TABLE 4.6 l [I + ~ + .£.
x 3 24
+ L360 + ... J+ 2 [l3
+ ...E... +
12 120
L
+ ... J
2 and substitute this function into the differential equation to fmd the coefficients b
bn
Because the derivatives of yix) are n· n +I [en - 2)nb n - bn _,]x n - 3 = 0
0 -2 11=1
CYI(X) ~
y;(x) = - x - + cyj(x) In x + L... (n - 2)b n x n - 3
n=O 1 2 l
(x
+ ± + ~ +..£.. +
3 4 45
?
... ) + I00
n~1
[en - 2)nb n - bn _ l ]x n - 3
= 0
and 2 bz
2
2 4 x x )
+(- b , - b) -2 b
,x - I
( ~+"3+4+45+'"
" -CYI(X) 2cyi(x) ~ 3b z - 4 OX -
Y2(X) = - - 2 - + - - - + cy~(x) In x + L... (n - 2)(n - 3)b n x n - 4,
3 -9-
x x n=O
2
12bz - 25
+ (3b 3 - b2)x O + (8b 4 - b3)x + (15b s - b4 )X = 0,
substitution into the differential equation yields
4
288 so we have the sequence of equations -b , - bo = 0, -b , + 2 = 0, 3b 3 - b2 + 1= 0,
x
-CYI(X)
--2-
2cyi(x)
+ - - - + cy~(x) In x + I
00
(n - 2)(n - 3)b n x n - 4
J 60b z - 157
*
8b4 - b3 + = 0, 15b s - b4 + ~ = 0, .... Solving these equations, we see that b ,
[ x x n=O 5 = 2 and b o = -2. However, the other coefficients depend on the value of b 2 • We
21600
give these values in Table 4.6. Hence, a second linearly independent solution is given
by
Y2(X) = CYI(X) In x + x
-2( -2 + 2x + b x 2+ -3b- 9 -4 x 3+
2
2 - 12b2 - 25
288 X
4
60b 2 - 157 s )
3cYI(x) ~ + 21600 x + ...
+ - - - + 3cyi(x) In x + L... 3(n - 2)b n x"-3 - CYI(X) In x - ') bn
00
xn- 2= 0
2cy (x)
x n~ ~
= (1 + f + ~: + ;;0 + .. }n x
, - + 2cyi(x) + I I
00 00
__ (n - 2)(n - 3)b n x n
- 3+ 3(n - 2)b n xn- 3
3b - 4 3 12b2 - 25 4
+ x- 2 -2 + 2x + b2x- + - -2 9 - x +
?
x 11=0 11=0
(
288 x
- n=O
I bx n
n
-
2 + c[xy~(x) + 3y i (x) - YI (x)] In x = O. 60b 2 - 157 s + )
+ 21600 X"',
= 0 because Yt is a soLution
of the differential equation where b 2 is arbitrary. In particular, two linearly independent solutions of the equa-
Simplifying this expression gives us tion are (c = 1100 = I)
X x2 x3
2cYI(x) YI(X) = 1 + "3 + 24 + 360 + ...
- - x - + 2cyi(x) + 6b ox- 3 - 6box- 3 + I00
(n - 2)(n - 3)bn x n
-
3+
11=1
and (b 2 = 0)
00 00
I 3(11 - 2)bn x n - 3
- ') bn x n - 2
= 0
n=1 n=O
2cy, (x) ~
- - - + 2cyi(x) + / [en - 2)nb n - bn _,]x n - 3
= o.
X n~l
216 Chapter 4 Higher Order Equations 4.8 Series Solutions of Ordinary Differential Equations
A gene.ral solution is therefore given by y(x) = ClYI (x) + C2Y2(X), where CI and c A useful property associated with the Gamma function is
are arbItrary. 2
l..
rex + I) = xr(x).
If x is an integer, using this property we have r(2) = r(1 + I) = I . r(1) = reI),
&.'
", Explain why the choice ao = lie does not affect the general solution obtained in Ex-
.W' ample 3. HInt: If c = 0, b o = 0, which is impossible (why?), so c cannot be zero. f(3) = r(2 + I) = 2r(2) = 2, r(4) = r(3 + I) = 3r(3) = 3 . 2, reS) =
r(4 + I) = 4r(4) = 4 . 3 . 2, ....
Gamma Function and for the integer n,
~ne ~f the more useful fu_nctio~s, which we will use shortly to solve Bessel's equa_ r(n + I) = n!
tIOn, IS the Gamma function, fIrst mtroduced by Euler in 1768, which is defmed as
follows. This property is used in solving the following equation.
Notice that because integration Bessel's Equation

is with respect to u, the result Definition 4.9 Gamma Function
Another important equation is Bessel's equation (of order IL), named after the Ger-
is a function of x.
The Gamma function, denoted rex), is given by man astronomer Friedrich Wilhelm Bessel (1784- I 846), who was a friend of Gauss.
rex) = r
0
e-uu x - I ~u, x > O. (See Figure 4.22.)
Bessel's equation is
x 2y" + xy' + (x 2 - /l,z)y = 0,
I where J.t 2: 0 is a constant. Bessel determined several representations of J".(x), the
~ V LJ
I I Bessel function of order /L, which is a solution of Bessel's equation, and noticed some
I I of the important properties associated with the Bessel functions. The equation received
I I
I I I I its name due to Bessel's extensive work with J".(x), even though Euler solved the equa-
I I I I
I I I I tion before Bessel.
- I 2 4 x To use a series method to solve Bessel's equation, first write the equation in stan-
I dard form as
n
I -2
: I I
I x2 - J.t2
Figure 4.22 Although we have only defined rex) y" + -y' + - - 2 - y = 0,
iA -4
for x > 0, rex) can be defined for all real numbers
except x = 0, -1, -2, .... *
X X
so x = 0 is a regular singular point. Using the Method of Frobenius, we assume that

there is a solution of the form y = ~;:'~o anxn+r. We determine the value(s) of r with
I 2 2 (2 :2 2)
"E''''''.'
Evaluate r(1).
the indicial equation. Becausexp(x) = x-; = I andx2q(x) = x x
Po = I and qo = -/L2 • Hence, the indicial equation is

J.t = x - J.t2,
2
r(r - I) + por + qo = r(r - I) + r - /L = r2 - J.t2 = 0
: 1 Solution n
with roots rl = /L and r2 = - J.t. We assume that y = ~;:'=o anx +". with derivatives
r(l) = r
o
e-uu l - I du = foo0 e- u du = lim [-e-"]g
b-)r:¥) tion into Bessel's equation yields
n
y' = ~;:'=o (n + J.t)anxn+".-I and y" = ~;:'=o (n + J.t)(n + J.t - I)a nx +".-2. Substitu-
= 10;, (-e- b + I) = 1. x2 f (n + /L)(n + J.t - l)a nx n+".-2 +x f (n + J.t)anxn+".-I

n~O n=O
... This top~c. is discussed in most complex analysis books such as Functions of One Complex Variable, + (x 2 - /L 2) I anx n+". = 0
Second Edmon, by John B, Conway, Springer-Verlag (1978), pp. 176-185. n=O
Using the Gamma function, rex), we write this solution as

~ (-1)" (X)2n+,u 1
Y= ~o n!r(I + fL + n) 2' ' where ao = -,,-.
2~fL!
This function, denoted J,u(};:), is called the Bessel function of the first kind of order
/L. For the other root rz = - fL, a similar derivation yields a second solution
fL(fL - I)aox,u + (1 + fL)fLalX,u+l + I (n + fL)(n + fL - I)anxn+,u
n=2 ~ (-I)" (X)2n-,u
y=) -
~ ~ ,;70 n!f(1 - fL + n) 2 '
+ fLaoX,u + (1 + fL)alx,u+l + I (n + fL)anXn+,u + I anxn+,u+Z
n=2 n=O which is the Bessel function of the first kind of order - /L and is denoted L,u(x).
Now, we must determine if the functions J,u(x) and J_,u(x) are linearly inde-
- fLZaoX,u - fLZalX,u+l - I fL2anXn+,u = 0 pendent.
n~Z Notice that if fL = 0, these two functions are the same. If fL > 0, rl - rz =
fL - (- fL) = 2fL· If 2fL is not an integer, by the Method of Frobenius the two solutiolfs
J,u(x) and J_,u(x) are linearly independent. Also, we can show that if 2fL is an odd in-
teger, then J,u(x) and J_,u(x) are linearly independent. In both of these cases, a general
+I ([en + fL)(n +'fL - 1) + (n + fL) - fL2]a n + an_2}X n+,u = o. solution is given by y = c1Jp.(x) + C2L,u(X). The graphs of the functions J,u(x), fL = 0,
n=2
Figure 4.23 1, 2, 3 are shown in Figure 4.23. Notice that these functions have numerous zeros.
Notice that the coefficient of aox,u is zero because rl = fL is a root of the indicial
equation. After simplifying the other coefficients and equating them to zero, we have What happens to the maximum value of Jm(x) as 111 increases?
(1 + 2fL)al = 0 and
If fL is not an integer, we define the Bessel function of the second kind of or-
der /L as the linear combination of the functions J,.{x) and J_,u(x). This function, de-
an =
(n + fL)(n + fL - 1) + (n + fL) - fL2 n(n + 2fL)' n;O: 2. noted by Y,u(x), is given by
From the Erst equation, al = O... Therefore, from an = -(an-2In(n + 2fL», n;O: 2, _c..::o.:.s.!.fL_7T...:::J.t!,u",(x-,-)_-.....:..J.=-"",uo:.(x-,-)
Y,u(x) = sin fL7T
an = 0 for all odd n. The coefficients ,that correspond to even indices are given by
We can show that J,u(x) and Y,u(x) are linearly independent solutions of Bessel's equa-
ao ao a2 ao
a2 = + 2fL) 22(1 + fL)' a4 = tion of order fL, so a general solution ofthe equation is y = c1J,u(x) + C2Y,u(X). We show
2(2 4(4 + 2fL) 24 . 2(2 + fL)(1 + fL) ,
the graphs of the functions Y,u(x), fL = 0, 1,2 in Figure 4.24. Notice that limx->o+ Y,.{x)
a4 ao 0.5 = -00. We can show that if m is an integer and if Ym(x) = lim,u->m Y,u(x), then Jm(x)
a6 = and Ym(x) are linearly independent. Therefore, y = c1J,u(x) + C2Y,u(X) is a general so-
6(6 + 2fL) 26 . 3 . 2(3 + fL)(2 + fL)(1 + fL) , 0.25
lution to x 2y" + xy" + (X2 - fL2)y = 0 for any value of fL.
a6 ao A more general form of Bessel's equation is expressed in the form
a8 =
8(8 + 2fL) 2 8 • 4 . 3 . 2(4 + fL)(3 + fL)(2 + fL)(I + fL) . -0.25
x2y" + xy' + (A2X2 - fL2)y = O.
-0.5
A general formula for these coefficients is given by Through a change of variables, we can show that a general solution of this parametric
-0,75
_ (-I)"ao Bessel equation is
aZn - 22nn!(1 + fL)(2 + fL)(3 + fL) ... (n + fL)' n;O: 2. Our solution can then be -1
"i',,·,I.,
written as
ao(-I)nx 2n+,u Figure 4.24
y = I
n~O
az nx 2n +,u = I 2n
n~O 2 n!(1 + fL)(2 + fL)(3 + fL)" '(n + fL)
Find a general solution of each of the following equations:
= f ao( -1)"2,u (!..)2n+,u (a) x 2y" + xy' + (X2 - l6)y = 0; (b) xV' + .l.y' + (X2 - is)y = 0;
n";;;'O n!(1 + fL)(2 + fL)(3 + fL) .. ·(n + p,) 2 . (c) x2y" + xy' + (9x 2 - 4)y = O.
:11 Solution (a) In this case, JL = 4. Hence, y = C1J 4(X) + C2Y4(X). We graph this 13. y" + xy' = sin x. yeO) = I. y'(O) = 0 27. The differential equation d 2yldx 2 + p(x)(dyldx) +
solution on [I, 20] for various choices of the arbitrary constants in Figure 4.25(a). q(x)y = 0 has a singular point at infinity if after sub-
14. y" + y' + xy = cos x. yeO) = 0, y'(O) = I
stitution of w = I/x the resulting equation has a sin-
Notice that we must avoid graphing near x = 0 because of the behavior of Y4(x). 15. (a) If y = ao + a,x + a,x 2 + a3x3 + .... what are gular point at w = O. Similarly, the equation has an
(b) Because J.L = t, y = c1JlIS(X) + C2YlIS(X), This solution is graphed for several the first three nonzero terms of the power series for ordinary point at infinity if the transformed equa-
values of the arbitrary constants in Figure 4.25(b). (e) Using the parametric Bessel's y2? (b) Use this series to find the first three nonzero tion has an ordinary point at w = O. Use the chain rule
equation with A = 3 and JL = 2, we have y = C1J 2(3x) + C2Y2(3x). We graph this so- terms in the power series solution about x = 0 to Van- and the substitution w = I/x to show that the differ-
lution for several choices of the arbitrary constants in Figure 4.25(e). der-Pol's equation, ential equation d 2yldx 2 + p(x)(dyldx) + q(x)y = 0 is
y" + (y2 _ I)y' +Y = 0 equivalent to
4
y
4
y
4
y
ify(O) = 0 andy'(O) = 1.
2
d y +
dw'w
(1. + P(I/;'») dy
w dw
+ q(l/,w) y = O.
w
*16. Use a method siroilar to that in Exercise 15 to find the
2 2 first three nonzero terms of the power series solution 28. Use the definition in Exercise 42 to determine if in-
about x = 0 to Rayleigh's equation, finity is an ordinary point or a singular point of the
10 x given differential equation.
-2 -2 y" + (t(y')' - I )Y' + Y = 0 (a) d 2yldx' + xy = 0
-4 -4 -4 if yeO) = 1 and y'(O) = O. an equation that arises in
(b) x 2 d 2yldx 2 + x dyldx + (x 2 - n')y = 0
the study of the motion of a violin string. (e) (I - x') d 2yldx 2 - 2x dyldx + n(n + l)y = 0
(a) (b) (c) 29. Hermite's equation is given by
In Exercises 17 and 18. determine the singular point of each
y" - 2xy' + 2ky = 0, k 2: O.
Figure 4.25 (a) y = c,J,(x) + coY,(x) (b) y = C,J1I5 (X) + C2Y1I5(X) equation. In each case. classify the point as regular or irregu-
(e) y = c,J2(3x) + C2Y2(3x) lar.
Using a power series expansion about the ordinary
point x = O. obtain a general solution of this equation
*17. x(x + l)y" -..;. y' + 5y = 0 for (a) k = I and (b) k = 3. Show that if k is a non-
x- negative integer. then one of the solutions is a poly-
18. (x 2 - 25}'y" - (x + 5)y' + lOy = 0 nomial of degree k.
EXERCISES 4.8 30. Chebyshev's equation is given by
In Exercises 19-24. use the Method of Frobenius to obtain two
-linearty independent solutions about the regular singular point (1 - x 2 )y" - xy' + k'y = O. n 2: O.
x= O. Using a power series expansion about the ordinary
point x = 0, obtain a general solution of this equation
In Exercises I and 2. determine the singular points of the equa- *19. 2xy" - 5y' - 3y = 0 for (a) k = 1 and (b) k = 3. Show that if k is a non-
In Exercises 9 and 10. determine at least the first five nonzero
tions. Use these points to find an upper bound on the radius of terms in a power series expansion about x = 0 for the solution 20. 9xy" + 14y'+ (x - l)y = 0 negative integer. then one of the solutions is a poly-
convergence of a series solution about Xo. of each initial-value problem. 21. x 2y" +xy' + (x - I)y= 0 nomial of degree k.
31. The hypergeometric equation is given by
1. x'y" - 2xy'
2. (x-2)y"+y'-y=0,xo=-2
+ 7y = O. Xo = I 9. y" - 4x 2y = O. yeO)
10. (-I + 2x')y" + 2xy'
= I. y'(O) = 0
- 3y = 0, yeO) = -2.
22. y"+ 3>' - (3~2 -I)y=o
x(1 - x)y" + [c - (a +b+ Ilx]Y' - aby = O.
y'(O) = 2 where a. b. and c are constants. (a) Show that x = 0
In Exercises 3-8. solve the differential equation with a power
*23. y" + ( 2x
I - 2 )y' - 35 y = 0
"j6x2 and x = I are regular singular points. (b) Show
that the roots of the indicial equation for the series
(~+ 2)Y' +(:2 +x)y= 0
series expansion about x = O. Write out at least the first five In Exercises II and 12, solve the equation with a power series
nonzero terms of each series. expansion about t = 0 by making the indicated change of vari-
able.
24. y"-
i: anx n+, are r = 0 and r = I-c. (e) Show that
+ 30y = 0
3. y" - 11y'
In Exercises 25 and 26. solve the differential equation with a nf~~r = O. the solution obtained with the Method of
4. y" + y= 0 11. 4xy" + y' = O. x = t + I
series expansion about x = O. Compare these results with the Frobenius is
*5. y" - y' - 2y = e- x 12. 4x'y" + (x + I)y' = O. x = t + 2 solution obtained by solving the problem as a Cauchy-Euler ab ala + l)b(b + I) x'
6. (I + 3x)y" - 3y' - 2y = 0 equation. y,(x) = I + llix + 2!c(c + I)
In Exercises 13 and 14. determine at least the first five nonzero
7. (2 + 3x)y" + 3xy' = 0 2S. x 2y" + 7xy' - 7y = 0 ala + I)(a + 2)b(b + I)(b + 2) x3 + ...
terms in a power series expansion about x = 0 of the solution
8. (2 - x 2 )y" + 2(x - I)y' + 4y = 0 to each nonhomogeneous initial-value problem. 26. x'y" + 3xy' + Y = 0 + 3!c(c + 1)(c + 2)
4.8 Series Solutions of Ordinary Differential Equations 223
*
where c 0, -1, - 2, .... This series is called the
36. Show that the equation xy" + y' + k 2xy = 0 is 49. Laguerre's equation is xy" + (1 - x)y' + ny = O.
hypergeometric series. Its sum denoted F(a b c' x) sin x if x '" 0
is called the hypergeometric f:nction. (d) Sho~ ilia; equivalent to ~ (xy ') + k xy =2
O. Show that
x '
1(X) = { 1, if x = 0 (a) Show that x = 0 is an
(a) Show that x = 0 is a regular singular point of
Laguerre's equation. (b) Use the Method ofFrobenius
F(1, b, b; x) = 1/(1 - x). (e) Find the solution that cor.
responds to r = I-c. f xJo(kmx)Jo(knx) dx = 0, m * n. (This shows that ordinary point of the equation. (b) Find a power se-
ries solution of the equation and graph an approxi-
to show that one solution of Laguerre's equation is
n (-I)m n! m .
32. Laguerre's equation is given by Jo(kmx) and Jo(knx) are orthogonal on the interval mation of the solution on an interval. (c) Generate a
Ln(x) = 2:
m=O
I I( _ )1 x , where Ln(x) IS
m. m. n m.
xy" + (1 - x)y' + ky = O. [0, I].) Hint: Follow a procedure like that deScribed numerical solution of the equation. Explain any un- called the Laguerre polynomial of order n.
in Exercise 33. Assume that Jo(km) = Jo(kn) = O. expected results. (c) Calculate the first eight Laguerre polynomials
(a) Show that x = 0 is a regular singular point of La. with this formula (or find them using a built-in
37. Find a general solution of each equation. . fy" - (cos x)y = sin x
guerre's equation. (b) Use the Method of Frobenius to 2 46. (a) UsepowersenestosolveLY(O) = l,y'(O) = 0 . computer algebra system command). (d) Show that
determine one solution of Laguerre's equation. (a) x y" + xy' + (x 2 - i)y = 0
2
the Laguerre polynomials satisfy the formula
(c) Show that if k is a positive integer, then the solu. (b) x y" + xy' + (16x 2 - 25)y = 0 (b) Compare the polynomial approximations of de- eX dn(xne -X)
tion is a polynomial. This polynomial, denoted L,/..x), gree 4, 7, 10, and 13 to the numerical solution ob- Ln(x) = -;;! ~ for n = I, 2, 3, . . . , 8.
is called the Laguerre polynomial of order k. In Exercises 38 and 39, solve each hypergeometric equation. tained with a computer algebra system.
(e) Show that f'oe-rLn(x)Lm(x) dx = 0 for n '" m, n,
*33. Show that Legendre's equation can be written as ~xpress all solu:ions in terms of the function F(a, b, c; x). (No- 47. The differential equation y" - xy = 0 is called Airy's m = 1,2,3, ... , 8. (This indicates that the Laguerre
d tIce that when eIther a or b is a negative integer, the solution is equation and arises in electromagnetic theory and polynomials are orthogonaI.) (f) Determine the value
dx [(1 - X2)y'] + k(k + I)y = O. Use this equation a polynomial.) (See Exercise 31.) quantum mechanics. Two linearly independent solu- of foe-X[Ln(x)f dx by experimenting with n = 1, 2,
to show that P m(x) and P,,(x) are orthogonal on the tions to Airy's equation, denoted by Ai(x) and Bi(x), 3, ... , 8. Note: All of the properties mentioned in
38. x(i - x)y" + y' + 2y = 0
interval [-1, 1] by showing that are called the Airy functions. The function Ai(x) ---> this problem hold for all n. We simply chose the val-
*39. x(1 - x)y" + (1 - 2x)y' = 0 o as x ---> 00 while Bi(x) ---> +00 as x ---> 00. (Most com- ues n = I, 2, 3, ... , 8 so that the calculations can be
( Pm(x)P,,(x) dx = 0, m * n. 40. Use the power series of expansion of the Bessel func-
tion of the first kind of order n (n an integer),
puter algebra systems contain built-in definitions of
the Airy functions.)
carried out with technology instead of by hand.
Hint: P m(x) and P,,(x) satisfy the differential equa. 50. The values of x that satisfy the equation Jo(x) = 0 are
tions ~ [(1 - x2)p~(x)] + n(n + I)P,,(x) = 0 and

Jm(x) = f I (-I)" (!..)2n+m
,,~o n.(n + m)! 2
(a) If your computer algebra system contains built-in
definitions of the Airy functions, graph each on the useful in many applications in applied mathematics.
(a) Approximate the first ten zeros (or roots) of the
interval [-15, 5]. (h) Find a series solution of Airy's
d 2 to verify thatJo(x) = -JI(x). equation and obtain formulas for both Ai(x) and Bi(x). Bessel function of the first kind of order zero, Jo(x),
dx [(1 - x )p;,,(x)] + m(m + l)Pm(x) = 0, respec- which is graphed in Figure 4.26. (b) Approximate the
(c) Graph the polynomial approximation of degree
41. Use the change of variables y = vex) to transform first nine zeros of J,,{x) for JL = 1,2, ... , 8.
tively. Multiply the first equation by P,,(x) and the sec- n of Ai(x) for n = 6, 15, 30, and 45 on the interval
Yx [-15, 5]. Compare your results to (aJ, if applicable.
ond by Pm(x), and subtract the results. Then, integrate Bessel's equation x 2y" + xy' + (x 2 - n 2 )y = 0 into 51. (a) Verify that the Legendre polynomials given in
from -I to 1.
the equation v" + [1 + i ]v :2
n2
= O. By substitut-
I (d) Graph the polynomial approximation of degree n
of Bi(x) for n = 6, 15, 30, and 45 on the interval
Table 4.4 satisfy the relationship
34. (Relationships among Bessel functions) (a) Using
[-15, 5]. Compare your results to (aJ, if applicable. ( Pm(x)Pn(x) dx = 0, m '" n
J,.(x) = ,,~o n!r(1(:I~ + n) (ft+" and ing n = ~ into the transformed equation, derive the so-
lution to Bessel's equation with n = 1/2.
48. (a) Use a series solution to approximate the solution
(called an orthogonality condition).
d to the initial-value problem
f(x + 1) = xf(x), show that dx [x"J,,(x)] =
x"J,.-I(X).
42. Show that a solution of x(d 2yldx 2) + dyldx
o is y = J o(2 Wv) where v is a constant.
+ ylv =
xy" + (2 - x)y' + (1/4x)y = 0,
y(l) = l,y'(J) =-1.
(b) Evaluate f, [Pn(x)f dx for n = 0, I, ... , 5.
*43. Use integration by parts to show that f(p + I) = How do these results compare to the value of
(b) U' J ().:(', (-1)" (X)2n+,. pf(p), p > O. Note: p is any real number. (b) Use a computer algebra system to generate a nu- 2/(2n + I), for n = 0, 1, ... , 5?
smg ,. x = ~o n!f(1 + JL + n)"2 '
merical solution of the initial-value problem and
d _
show that dx [s "J,,(x)] = -X-"J,.+I(X). 44. (a) Show that fG) = 2 fe- X2 dx. (Hint: Let compare these results with those obtained in (a)
2 by graphing the two approximations simultane-
u= x .) (b) Use polar coordinates to evaluate
ously.
(c) Using the results of parts (a) and (b), show that
J,,_I(X) - J,.+I(X) = 2f,.(x). (r e-x'dx)(r e-y'dY ) = rfe-(x'+y')dxdy.
(d) Evaluate f x"J,._I(X) dx. (c) Use the results of (a) and (b) to evaluate f(4) and
then fG).
35. Show that y = Jo(kx) where k is a constant is a solu-
tion of the parametric Bessel equation of order zero.
xy" + y' + exy = o.
45. Consider the initial-value problemy" +I(x)y' +Y = 0,
yeO) = l.y'(O) = -I, where
224 Chapter 4 Higher Order Equations Chapter 4 Summary
CHAPTER 4 SUMMARY
0.8 0.8 0.8 ~~~~:::J"~~,rr~~-=
0.6 0.6 0.6 ~ Concepts & Formulas
0.4 0.4
0.2 0.2
a general solution of the equation is

-0.2 -0.2 §tion4.11
-0.4 -0.4 f(t) = cd,(t) + c2h(t)
-0.6 -0.6 Second-order ordinary linear differential equation
where {Clo C2} is a set of two arbitrary constants.
(a) (b) (e) a2(t)/2)(t) + al(t)y'(t) + ao(t)y(t) = F(t)
Reduction of Order
If y = f(t) is one solution of y" + p(t)y' + q(t)y = 0, a
, Homogeneous
second linearly independent solution is y = f(t) vet), where
0.8 0.8 azCt)/2l(t) + al(t)y'(t) + ao(t)y(t) = 0
0.6 0.6
0.8
0.6
Constant coefficients
vet) = J_1_ e - Jp(') d, dt
If(t)f .
a2/2l(t) + aly'(t) + aoy(t) = F(t)
-0.2 -0.2 Linearly dependent and independent set of functions 1Section 4.21
-0.4 -0.4 If S = {II (t), h(t)}, S is linearly dependent if there are con-
-0.6 -0.6 stants c, and C2 not both zero, so that Characteristic equation
(d) (e) (f) cdi(t) + czh(t) = o. The equation
S is linearly independent if S is not linearly dependent. ar2 + br + c = 0

1
is the characteristic equation of the second-order linear
0.8 0.8 0.8 Wronskian
homogeneous differential equation with constant coefficients
0.6 0.6 0.6 The Wronskian of
0.4 0.4 0.4
S = {l1(t),h(t)},
ay" + by' + cy = o.
0.2 0.2 0.2
is the determinant Solving second-order homogeneous linear equations with
-0.2 -0.2 -0.2 constant coefficients
-0.4 -0.4 -0.4 W(S) = INt) h(t) I. If ay" + by' + cy = 0 and r, and r2 are the solutions of the
-0.6 -0.6 -0.6 fi(t) f2(t) characteristic equation ar2 + by + c = 0, then a general solu-
(g) (h) If W(S) "* 0 for at least one value of t in the interval I, S is tion is
(i)
linearly independent. (a) y = cle'" + C2e"', if r, "* '2and both rl and r2 are real;
Figure 4.26 (a) Jo(x) (b) J,(x) (c) J 2 (x) (d) J,(x) (e) J 4 (x) (I) Js(x) (g) J 6 (x) (b) J 7 (x) (i) Jg(x) (b) y :;::: Y = Cle rlt + c2terlf, if rl = r2; and
Principle of Superposition (c) y = e"'(c, cos {3t + C2 sin {3t), if'l = a + i{3, {3"*0 and
Any linear combination of a set of solutions of the second- r2 = " = a - i{3.
order linear homogeneous equation is also a solution.
Fundamental set of solutions

1Section 4.31
A set S = {fi (t), h(t)} of two linearly independent solutions of
the second·order linear homogeneous equation. Every second-
nth-order ordinary linear differential equation
order linear homogeneous equation has a fundamental set of
an(t)/n)(t) + ... + al(t)y'(t) + ao(t)y(t) = g(t)
solutions from which a general solution can be obtained.
General solution of a homogeneous equation Homogeneous

an(t)/n)(t) + ... + al(t)/(t) + ao(t)y(t) = 0
If S = {I, (t), f2(t)} is a fundamental set of solutions of the
second-order linear homogeneous equation Constant coefficients
a2(t)/2l(t) + a,(t)y'(t) + ao(t)y(t) = 0, any(n)(t) + ... + aly'(t) + aoy(t) = get)
Chapter 4 Summary
Linearly dependent and independent set of functions

is the characteristic equation of the nth-order linear homo- where Yh(t) is a general solution of the corresponding homo- where ao, ah a2, ... , an are constants.
If S = {f,(t),h(I),h(t), ... ,f.-1(t),f.(I)), S is linearly geneous differential equation with constant coefficients geneOUS equation General solution of a second-order Cauchy-Euler equation
dependent if there are constants c" C2 , ••• , Cm not all zero,
so that any(n) + an_1y(n-1) + ... + alY' + aoy = O. an/n)(t) + ... + aly'(I) + aoy(t) = 0, Y = c1x ml + C2 xm2 , if ml =I=- m2 are real;
C,h(t) + c-Ji(t) + ... + cJ,,(t) = o. and ypCI) is a particular solution of the nonhomogeneous Y = CIX ffll + C2 Xml In x, if ml = m2; and _ .
Solving higher-order homogeneous linear equations with
y = Xa[C1 cos({3 In x) + C2 sin({3 In x)] if ml = m2 = a + 1{3,
constant coefficients equation.
S is linearly independent if S is not linearly dependent.
1. Let r be a real root of the characteristic equation ,"ethod of undetermined coefficients
{3 * o.
Wronskian Nonhomogeneous Cauchy-Euler Equations
n 1
anrn + an_1r - + '" + aIr + ao = 0 Method used to find a particular solution:. .
The Wronskian of
I. If gll) = t m , the associated set of functlOns 1S Higher-order Cauchy-Euler equations
of an nth-order homogeneous linear differential equation
S = (h(t),h(t),h(t) + '" + fn-,(t),f.(t»), S = {1 m , Im - 1
, ••. , t 2 , t - I}.
with real constant coefficients. Then, ert is the solution as..
is the determinant sociated with the root r.
Z. If g,(t) = t m e"', the associated set of functions is
If r is a real root of multiplicy k where k 2: 2 of the \ Section 4.8\
Nt) h(t) f.(t)
characteristic equation, then the k solutions associated with S = {tme kt , tm-1e kt , •..• t 2e kt , tela, e kr }.
W(S) =
Jl(t) f2(t) nt) rare = tme a • cos {31, or g,(t) = tme
3. If g,(t) a
• sin {3t, the associ- Ordinary and Singular points
ated set of functions is Xo is an ordinary point ofy" + p(x)y' + q(x)y =.0 ifboth
p(x) and q(x) are analytic at Xo. If Xo is not an ordmary pomt,
r s::= {tmeat cos f3t, ... , teat cos f3t, eat cos {3t,
If W(S) * 0 for at least one value of t in the interval I, S is
2. Suppose·that rand represent the complex conjugate pair
a ± {3i. Then, the two solutions associated with these two tme at sin {3t, ... , teat sin /3t, ear sin ,Bt].
Xo is called a singular point.
linearly independent Power Series Solution method about an ordinary point
roots are For each function in S, determine the associated set of
Principle of Superposition functions. If any of the functions in S appear in the homo- 1. Assume that Y = 2:~~o a.(x - xoJ"·
eat cos f3t and eat sin f3t. 2. After taking the appropriate derivatives, substitute
Any linear combination of a set of solutions of the nth-order geneous solution Yh(I), multiply each function .in S by I' to
If the values a ± {3i are each a root of multiplicity k of the y = 2:~~o an(x - xoJ" into the differential equation.
linear homogeneous equation is also a solution. obtain a new set S' (r is the smallest pOS1llve mteger so
characteristic equation, then the other solutions associated 3. Find the unknown series coefficients an through an equa-
that each function in S' is not a function in Yh(I). A partie·
Fundamental set of solutions wi th this !lair are ular solution is obtained by taking the linear combination tion relating the coefficients.
A set S = (h(t),f2(t),h(I), . .. ,f.-I(t),J.(t)) ofn linearly 4. Apply any given initial conditions, if applicable.
te
Uf
cos {3t, terti sin ,Bt, t 2 e rtr cos {3t, t 2e al sin /3t, ... , of all functions in the associated sets where repeated func-
independent solutions of the nth-order linear homogeneous tions should appear only once in the particular solution.
tk-t e
rtt
cos /3t, t k - 1ear sin f3t. Convergence of the power series solution
equation. Every nth-order linear homogeneous equation has a
fundamental set of solutions from which a general solution A general solution to the nth-order differential equation is Legendre's equation
can be obtained. the linear combination of the solutions obtained for all
values of r. Note that if rl, r" ... , r, are the roots of the
[Section 4.6\ (1 - x 2 )y" - 2xy' + k(k + l)y = 0
General solution of a homogeneous equation equation of multiplicity k .. k2' ... , ~, respectively, then
Variation of parameters Regular and irregular singular points
If S = (fI(t),h(t), ... ,f.-I(I),f.(t)} is a fundamental set of kl + k2 + '" + ~ = n, where n is the order of the differ. Let Xo be a singular point of y" + p(x)y' + q(x)y = O. Xo is a
solutions of the nth-order linear homogeneous equation ential equation. A particular solution ofy" + p(t)y' + q(l)y = f(t) is
regular singular point of the equation if both (x - xo)p(x)
an(I)/n>Ct) + ,.. + al(t)y'(t) + ao(t)y(t) = 0, yp(t) = ul(I)YI(t) + U,(t)Y2(t), and (x - xo?q(x) are analytic at Xo· If Xo is not a regular sm-
a general solution of the equation is
\ Section 4.51 where UI(I) =
-12(t)f(l)
J----w(S) dl and u2(1) =
J y,(t)f(t) .
W(S) dt,
gular point, Xo is called an irregular singular point of the
equation.
f(t) = c,h(t) + c-Ji(t) + ... + Cn-If._I(t) + cnf.(t), S = {Y1 (I), 12(I)} is a fundamental set of solutions of the Method of Frobenius
where {eb C2, ... ,en-I. en} is a set of n arbitrary constants. Particular solution corresponding homogeneous equation.
A particular solution, Yp(t), of the differential equation Indicial equation
any(n\t) + ... + a,y'(t) + aoY(I) = get) is a specific func- r(r - 1) + por + qo = 0
ISection 4.41 tion, containing no arbitrary constants, that satisfies the
equation.
\ Section 4.7\
Gamma function
Characteristic equation
General solution of a nonhomogeneous equation
Cauchy-Euler Equation
A Cauchy-Euler differential equation is an equation of the
rex) = L
OO
e-"u x - I
du, x> 0
The equation
A general solution to the nonhomogeneous equation is form Bessel's equation
yet) = y.(t) + yp(I), x'y" + xy' + (x' - /.L2 )y = 0
228 Chapter 4 Higher Order Equations Chapter 4 Review Exercises 229
CHAPTER 4 REVIEW EXERCISES 53. Use the substitution in the previous problem to solve Y" + 2y' - (A - I)y = 0
~~.~~~::a:::...;"..-.",.,z=:::;~\i..!..~:;;:;;,;:m,,,; p;.~'"TfJ----"--'-'~~~~~ y" - 2ty' + tZy = O. Hint: You should obtain the dif- (d) { yeO)= 0, y(2) 0 =
ZUE&oiio
ferential equation duldt = - (u - If Make the sub- Hinl: Consider three cases: A = 0, A < 0, and
" stitution v = u - t to solve this equation. A>O.
In Exercises 1-6, determine if the given set is linearly inde- *25. y" + Sy' = Si z *54. (Abel's formula) Suppose that Yl and yz are two so-
pendent or linearly dependent. lutions of y" + p(t)y' + q(t)y = 0 on an interval I In Exercises S9-6S, solve the Cauchy-Euler equation.
26. y" - 4y' = -3 sin I
1. S = {e s" I} *27. y" + 2y' + Sy = 3 sin 21
where p(l) and q(t) are continuous functions. Then, the 59. x 2y" - 4xy' + 6y = 0
Wronskian of Yl and yz is
2. S = {cos z t, sin2 t} 2S. y" - 9y = cos 3t *60. xZy" + 7xy' + 8y = 0
W(YhY,)(X) = Ce-Jp(,)d,.
*3. S= {I, tlnl} 29. y" - 2y' = 2 cos 41 J2x Zy" + Sxy' + Y = 0
4. S = {I, 1- 1, 3t} 30. y" - 3y' + 2y = -4e- z, Prove Abel's formula by computing fr W(y" Y2)(t) 61. '\yCI) = 1, y'(I) = 0
5. S = {t, cos I, sin I} *31. y" - 6y' + 13y = 3e- 2 , and using the relationship y" = -p(l)y' - q(l)y for yr *62. xZy" + xy' + Y = 0
6. S = ie', e- z" e-'} 32. y" + 9y' + 20y = -2Ie' to obtain a first-order ordinary differential equation 63. x 2y" + 7xy' + 2Sy = 0
33. y" + 7y' + 12y = 31 ze -4, for W. *64. Sx 2y" - xy' + 2y = 0
In Exercises 7-9, verify thaty is a general solution of the given
differential equation. 34. y'" + 3y" - 9y' + Sy = e' 55. Can the Wronskian be zero at only one value of x on *65. x 2y" - 7xy' + ISy = Sx
*35. y!" - 12y ' - 16y = e 4t - e- 2t
n Hinl: Use Abel's formula.
7. y" - tyt + lOy = 0; Y = cleSt + cze 2t 56. Use Abel's formula to find the Wronskian (within a
36. /4) + 6y'" + ISy" + 30y' + 2Sy = e-' cos 2t + In Exercises 66-68, solve the differential equation with a se-
8. y" - y' - 2y = O;y = cre2l + cze- r constant multiple) associated with the following dif- ries expansion about x = O. Write out at least the first five
e- 2t sint
9. y" - 2y' + 2y = 0; Y = e'(c, sin I + C2 cos I) ferential equations. Also, obtain a fundamental set of nonzero terms of each series.
37. /4) + 4y'" + 14y" + 20y' + 2Sy = 12 solutions and compute the Wronskian directly. Com-
In Exercises 10 and 11, show that the function Yl (t) sati~fies pare the results.
66. y" - 4y' + 4y = 0
the differential equation and find a second linearly independ- In Exercises 3S-4S, solve the initial-value problem.
(a) y" + 3y' - 4y = 0 *67. y" + 2y' - 3y = xe'
ent solution. 68. (-1 + 2xz)y" + 2xy' - 3y = 0
38. y" + Sy' + 6y = O,y(O) = 2,y'(0) = 0 *(b) y" + 4y' + l3y = 0
10. y,(x) = I + I,y" _ _2_y' + __2_zy = 0 39. y" + lOy' + 16y = 0, yeO) = O,y'(O) = 4
(e) y"+4y'+4y=0 In Exercises 69-71, USe the Method of Frobenius to obtain two
t+ 1 (1+ I)
40. y" +'16y = 0, yeO) = O,y'(O) = -8 linearly independent solutions about the regular singular point
(d) y" + 9y = 0
11. y,(x) = sin t, y" + 1.y ' + Y = 0 *41. y" + 2Sy = 0, yeO) = 1, y'(O) = 0 x= O.
I I 57. Use Abel's formula to find the Wronskian (within a
42. y" - 4y = x, yeO) = 2, y'(O) = 0
constant multiple) associated with the following dif- 69. 3xy" + lly' - Y = 0
In Exercises 12-37, f'md a general solution for each equation. 43. y" + 3y' - 4y = e', yeO) = O,y'(O) = 4 ferential equations. *70. 2x 2y" + 5xy' - 2y = 0
12. y" + 7y' + lOy = 0 44. y" + 9y = sin 31, yeO) = 6,y'(0) = 0 (a) ty" + y' + ty = 0
*13. 6y" + Sy' - 4y = 0 45. y" + Y = cos I, yeO) = 0, y'(O) = 0
(b) IZy" - Sty' + Sy = 0 71. y" -~y' + (:Z - 2)y= 0
14. y" +2y' + y= 0 (e) IZy" - ty' + Sy = 0

In Exercises 46-SI, use variation of parameters to solve the in- In Exercises 72 and 73, find a solution to each hypergeomet-
*15. y" + 3y' + 2y = 0
dicated differential equations and initial-value problems. 58. Solve each of the following boundary-value problems. ric equation. Express the solution in terms of the function
16. y" - lOy' + 34y = 0
*17. 2y" - Sy' + 2y = 0
46. y" + 4y = tan 21 J y" - y = 0 F(a, b, c; x). (See Exercise 31 in Section 4.8).
47. y" + Y = csc I (a) V(O) + 3y(0) = 0, y'(I) + y(l) = 1

72. x(I - x)y" + (1 + 2x)y' + lOy = 0
IS. ISy" - lly' + 2y = 0
J y" + Ay = 0
*19. 20y" + y' - Y = 0
4S. y" - Sy' + 16y = t- 3 e 4 '
*49. y" - Sy' + 16y = 1- 3 e 4',y(l) = O,y'(l) = 0
*(b) '\yeO) = 0, yep) = 0' p >0 73. x(I + x)y" + (t - 12x)Y' - lOy = 0
20.12y"+Sy'+y=0 Hint: Consider three cases: A = 0, A < 0, and
*21. 2y'" + 3y" + y' = 0 50. y" - 2y' + Y = e' In I A> O. 74. (Simple Modes of a Vibrating Cbain) The equation
that describes the simple modes of a vibrating chain
22. 9y" + 36y' + 40y = 0
51. y" - 2y' + Y = e' In I,y(l) = O,y'(l) = 0 y" + Ay = 0 >0 of length C is
52. Show that the substitution u = y' Iy converts the equa· (e) { y'(O) = O,y'(p) = O,p
*23. 9y" + 12y' + 13y = 0
tion Pa(t)y" + p,(t)y' + pz(I)Y = 0 to Pa(t)(U' + u') Hinl: Consider three cases: A = 0, A < 0, and
24. y" - 2y' - Sy = -I + p,(t)u + pz(t) = o. A> O.
where y is the displacement and x is the distance from (b) Convince yourself that for any value of e,the Let G denote the cumulative level of glucose in the blood, g = G - Go, H the cumu-
the bottom of the chain. This equation was studied ex- equation J o(2Vliv) = 0 has infinitely many so. lative level of hormones that affect insulin production (such as glucagon, epinephrine,
tensively by Daniell Bernoulli around 1727. lutions. cortisone, and thyroxin), and h = H - Ho. Notice that g and h represent the fluctua-
(a) If the chain is fixed at the top so that y(e) = 0 (e) If C = I, graph J o(2Vliv) and approximate the tion of the cumulative levels of glucose and hormones from their equilibrium values.
and yeO) = 1, show that a solution to this equa- last ten solutions of Jo(2Vliv) = o. The relationship between the rate of change of glucose in the blood and the rate of
tion is J o(2Wv). change of the cumulative levels of the hormones in the blood which affect insulin pro-
duction is
g' = 11(g, h) + J(t)
{ h' = h(g, h) ,
where J(t) represents the external rate at which the blood glucose concentration is be-
_: J - _ Differential Equations at Work: ing increased. * If we assume that II and h are linear functions, then this system of
equations becomes
A. Testing for Diabetes
g' = -ag - bh + J(t)
{ h' = -eh + dg ,
Diabetes mellitus affects approximately 12 million Americans; approximately one-half
of these people are unaware that they have diabetes. Diabetes is a serious disease: it is where a, b, e, and d represent positive numbers.
the leading cause of blindness in adultS and the leading cause of renal failure, and it (a) Show that if g' = -ag - bh + J(t) then
is responsible for about one-half of all nontraumatic amputations in the United States.
In addition, people with diabetes have an increased rate of coronary artery disease and h= 1.. (_g' - ag +J) and h' = i (-g" - ag' + J').
II
strokes. People at risk for developing diabetes include those who are obese; those suf- b
fering from excessive thirst, hunger, urination, and weight loss; women who have given (b) Substitute h = (lIb)(-g' - ag + J) and h' = (lIb)(-g' - ag' + J') into h' =
birth to a baby of weight greater than 9 lb; those with a family history of diabetes; and - elI + dg to obtain the second-order equation
those who are over 40 years of age.
1 . e
People with diabetes cannot metabolize glucose because their pancreas produces b(-g' - ag' + J') = -b(-g' - ag +J) + dg
an inadequate or ineffective supply of insulin. Subsequently, glucose levels rise. The
body attempts to remove the excess glucose through the kidneys; the glucose acts as a g' + (a + e)g' + (ae + bd)g = J' + cJ.
diuretic, resulting in increased water" consumption. Since some cells require energy, Since the glucose solution is consumed at t = 0, for t> 0 we have that
I
which is not being provided by glucose, fat, and protein are broken down and ketone
levels rise. Although there is no cure for diabetes at this time, many cases can be ef- g' + (a + e)g' + (ae + bd)g = O.
fectively managed by a balanced diet and insulin therapy in addition to maintaining an (c) Show that the solutions of the characteristic equation of g' + (a + e)g' +
optimal weight. (ae + bd)g = 0 are
Diabetes can be diagnosed by several tests. In the fasting blood sugar test, a pa-
tient fasts for at least four hours and the glucose level is measured. In a fasting state, 1.. (-a - e - V(a - e)2 - 4bd) and 1.. (-a - e + V(a - e)2 - 4bd).
2 2
the glucose level in normal adults ranges from 70-110 mg/mL. An adult in a fasting
state with consistent readings of over 150 mg probably has diabetes. However, indi- (d) Explain why it is reasonable to assume that glucose levels are periodic and subse-
viduals vary greatly, so people with mild cases of diabetes might have fasting state glu- quently that (a - ef - 4bd < O.
cose levels within the normal range. A highly accurate test that is frequently used to (e) If (a - e)2 - 4bd < 0 and t > 0, show that a general solution of
diagnose mild diabetes is the glucose tolerance test (GTT), which was developed by
Rosevear and Molnar of the Mayo Clinic and Ackerman and Gatewood of the Univer- g' + (a + e)g' + (ae + bd)g = 0
sity of Minnesota. During the GTT, a blood and urine sample are taken from a patient is
in a fasting state to measure the glucose (Go), hormone (Ho), and glycosuria levels, re-
spectively. We assume that these values are equilibrium values. The patient is then given D. N. Burghess and M. S. Borrie, Modeling with Differential Equations, Ellis Horwood Limited, pp.
oj<
100 g of glucose. Blood and urine samples are then taken at 1-. 2-, 3-, and 4-hour in- 113-116. loyce M. Black and Esther Matassarin-Jacobs, Luckman and Sorensen s Medical-Surgical
tervals. In a person without diabetes, glucose levels return to normal after two hours; NurSing: A Psychophysiologic Approach. Fourth Edition, W B. Saunders Co., Philadelphia (1993),
in diabetics the blood sugar levels either take longer or never return to normal levels. pp. 1775-1808.
If the slope has constant angle a, we can rewrite the equation as

g(t) = e-Ca+cltI2[A cos (fV 4bd - (a - C)2) + B sin (fV4bd - (a - c i )] 2
d s CoAp (tis)2 ,
and that - 2 = g[sin a - fL cos a] - - - -
dl 2m dl
G(t) =
where fL is the coefficient of friction, CD the drag coefficient, A the projected area of
Go + e- ca +cl'/2[A cos( fV4bd - (a - C)2) + B sin (fV4bd - (a - C)2)} the skier, p the air density, and g "" 9.81 m' S-2 the gravitational constant.* Because
g, a, fL, m, CD, A, and p are constants, g [sin a - fL cos a] and CoAp!2m are constants.
Let a = !(a + c) and w = !V4bd - (a - c)'. Then we can rewrite the gen- Thus, if we let k 2 = g [sin a - fL cos a] (assuming that g [sin a - fL cos a] is non-
eral solution obtained here as negative) and h 2 = CoAp!2m, we can rewrite this equation in the simpler form
G(t) = Go + e-a'[A cos wt + B sin wt].

Research has shown that lab results of 27r! w > 4 indicate a mild case of diabetes.
(f) Suppose that you have given the GTT to four patients you suspect of having a mild
Remember that the relationship between displacement, s, and velocity, v, is
case of diabetes. The results for each patient are shown in the following table.
Which patients, if any, have a mild case of diabetes? ds
v =di'
Thus, we can find displacement by integrating velocity, if the velocity is known.
Patient 1 Patient 2 Patient 3 Patient 4
I
1. Use the substitution v = ds! dl to rewrite d 2s!dt = k' - h (tis! dl)2 as a first-
I 2 2
Go 80.00 90.00 100.00 110.00 order equation and find the solution that satisfies yeO) = Yo. Hint: Use the method
t= I 85.32 91.77 103.35 114.64 of partial fractions.
If
~
2. Find s(t) if s(O) = O.
t= 2 82.54 '85.69 98.26 ,9
- Thus, in this case, we see that we are able to express both s and v as functions of t.
t= 3 78.25 92.39 96.59 8.14
i --
Typical values ofthe constants m, p, fL, and CoA are shown in the following table.
t= 4 76.61 91.13 99.47
- 113.76
Constant Typical Value
m 75 kg
B. Modeling the Motion of a Skier 1.29 kg· m- 3
P
During a sporting event, an athlete loses strength because the at!- ~ has to perform }.t 0.06
work against many physical forces. Some of the forces acting' an athlete include 0.16 m 2
CoA
gravity (the athlete must usually work against gravity when m llg body parts; fric-
tion is created between the ground and the athlete) and aerod .unic drag (a pressure
gradient exists between the front and the back of the athlete; • athlete must overcome 3. Use the values given in the previous table to complete the entries in the following
the force of air friction). In downhill skiiing, friction and dr affect the skier most be- table. For each value of a, graph v(t) and s(t) if yeO) = 0, 5, and 10 on the inter-
cause the skier is not working against gravity (why?). Thf ,stance traveled by a skier val [0, 40]. In each case, calculate the maximum velocity achieved by the skier.
moving down a slope is given by What is the limit of the velocity achieved by the skier? How does changing the
d 2s initial velocity affect the maximum velocity?
m dt 2 = Fg - F". - L~
where m is the mass of the skier, s is the distance traveled by the skier at time I, Fg is
>to Sauli Savolainen and Reijo Visuri. "A Review of Athletic Energy Expenditure, Using Skiing as a Practi~
the gravitational force, F". is the friction force between ski and snow, and D is the aero-
cal Example;' Journal ofApplied Biomechanics. Volume 10. Number 3 (August 1994), pp. 253-269.
dynamic drag.
234 Chapter 4 Higher Order Equations Differential Equations at Work 235
a k 2 = g[sin a - JL cos a] h2 = (oAp

2m
- skier. How does changing the initial velocity affect the maximum velocity? Com-
pare your results to (3). Are your results consistent?
8. How do these results change if p., is increased or decreased? if CrA is increased
a = 30° = ",16 rad or decreased? How much does a 10% increase or decrease in friction decrease or
increase the skier's velocity?
a = 40° = 2",19 rad
9. The values of the constants considered here for three skiers are listed in the fol-
a = 45° = ",/4 rad lowing table.
a = 50° = 5",118 rad
Constant Skier 1 Skier 2 Skier 3
4. For each value of a in the previous table, graph v(t) and s(t) if v(O) = 80 and 100 m 75 kg 75 kg 75 kg
on the interval [0, 40], if possible. Interpret your results. 1.29 kg· rn- 3 1.29 kg· rn- 3 1.29 kg· rn- 3
p
Sometimes, it is more useful to express v (velocity) as a function of a differ-
ent variable, like s (displacement). To do so, we write dtlds = lIv because ds/dt = J.L 0.05 0.06 0.Q7
v. Multiplying dvldt = k 2 - h 2v 2 by lIv and simplifying leads to
evA 0.16 rn 2
0.14 rn z 0.12 rn2
ldv = l(~ - hV)
v dt v Suppose that these three skiers are racing down a slope with constant angle
a = 30° = ",/6 rad. Who wins the race if the length of the slope if 400 m? Who
loses? Who wins the race if the length of the slope is 800 or 1200 m? Who loses?
Does the length of the slope matter as to who wins or loses? What if the angle of
the slope is increased or decreased? Explain.
10. Which of the variables considered here affects the velocity of the skier the most?
Which variable do you think is the easiest to change? How would you advise a
group of skiers who want to increase their maximum attainable velocity by 10%?
...!..
dw + h2v = k 2v- 1
2v dx c. The Schrodinger Equation
Multiplying this equation by 2v and applying the substitution w = v 2 leads us to
The time independent Shriidinger equation, proposed by the Austrian physicist Erwin
the first-order linear differential equation
Schriidinger (1887-1961) in 1926, describes the relationship between particles and
dw + 2h2w = 2k2 waves. The Shriidinger equation in spherical coordinates is given by
ds . 2
- a ( r 2 -aI/J) + -1- - -
1 - a (. aI/J) +----+-(E-V)I/J=O*
smli- 1 a I/J 2p.,
5. Show that a solution of this differential equation is given by r2 ar ar r2 sin Ii ali ali r2 sin Ii acf/
2
h2 .
w(s) = (k/h)2 + Ce- 2h\ If we assume that a solution to this partial differential equation of the form
where C is an arbitrary constant. Resubstitute w = v2 to obtain ",(r, Ii, cjJ) = R(r)6(Ii)g(cjJ)
[v(s)f = (k/h)2 + Ce -2h 2s. exists, then by a technique called separation of variables, the Schriidinger equation
can be rewritten as three second-order ordinary differential equations:
6. Show that the solution that satisfies the initial condition v(O) = Vo is
d 2g
[V(S)]2 =
-k 2
+ ~e2h2s +
h2 e 2/'-'
h 2 2
v
0
(k)2 (1 -
= h e-
2h
's) + v6e- 2h2s .
Azimuthal Equation
,
dcjJ2 = -meg
7. For each value of a in the previous table, graph v(s) ifv(O) = 0, 5, and 10 on the >10 Stephen T. Thornton and Andrew Rex, Modern Physics for Scientists and Engineers, SaWlders College
interval (0, 1500]. In each case, calculate the maximum velocity achieved by the Publishing (1993). pp. 248-253.
236 Chapter 4 Higher Order Equations Differential Equations at Work 237
4. Show that the solutions of the indicial equation of

Radial Equation
2
2
d u + (~- p) du +
dr 2
[2 (2 _E..) _ e(er2+ 1)] u = 0
Angular Equation I
-'--d
sm
sin d( (J-de) + [ece + 1) - smm; () ]e = O.
-'-2-
r dr r
(J (J d(J
are e and -(e + I). ~
The orbital angular momentum quantum number e must be a nonnegative integer: f 5. Find a power series solution to the equation of the form u(r) = I anr n+{.
= 0, ::': 1, ::':2, ... and the magnetic quantum number me depends on me = -f e: -e, 6. Find a series solution of n~O
+ 1, ... , -2, -1,0, 1,2, ... , e, e
+ I.
. ..
If e= 0, the radial equatIOn IS ?1 d ( dR)
dr r2 dr
2/L
+ l1" (E - V)R = O. Using
VCr) = -e2/471"Eor (the constants /L, h, e, and Eo are described later), the equation
2 2
d R 2 dR 2/L ( e )
becomes - d 2 + - -d + -h2 E + - - R = O.
r r r 471"Eor
I. Show that if R = Ae -rlao is a solution of this equation, ao = 471"Eoh2/ /Le2 and
E = - h 2 /2p.a6.
The constants in the expression 471"Eoh2/ /Le2 are described as follows:
34
h = 1.054572 X 10- J . s is the Planck constant (6.6260755 X 10- 34 J . s divided
by 271"), EO = 8.854187817 X 10- 12 F/m is the permitivity of vacuum, and
19
e = 1.60217733 X 10- C is the elementary charge. For the hydrogen atom,
/L = 9.104434 X 10- 31 kg is the reduced mass of the proton and electron.
2. Calculate ao, 2ao, and E for the hydrogen atom.
3. The diameter ofa hydrogen atom is approximately 10- 10 m while the energy, E, is
known to be approximately 13.6 eY. 'How do the results you obtained in problem' .. '
compare to these?
For the hydrogen atom, the radial equation can be written in the form
2
d R + ~ dR + [2E + 22 _ e(e + I)] R = 0
dr2 r dr r r2 ,
where E is the energy of the hydrogen atom and 2 is a constant. *

When E < 0 and we let p = V - 2E, a solution of
+~
2
d R dR + [2E + 22 _ e(e + I)] R = 0
dr 2 r dr r r2
has the form
R(r) = e -pru(r).
When this solution is substituted into the equation, we obtain the equation
2
----,-
dr r
(1 )
d u + 2 - - p -du + [2(2 - -p) - - -
dr r r2
e(e
+ - u = O. 1)]
... Yi-Hsin Liu and Wai-Ning Mei, "Solution of the radial equation for hydrogen atom: series solution or
Laplace transform?" International Journal of Mathematical Education in Science and Technology, Volume
21, Number 6, (1990), pp. 913-918.
5.1 Simple Harmonic Motion 239
5
stretched s units past its natural length to the equilibrium position x = O. When the sys-
tem is in motion, the displacement from x = 0 at time t is given by x(t).
By Newton's second law of motion,
where m represents mass of the object and a represents acceleration. If we assume that
there are no forces other than the force as a result of gravity acting on the mass, we
determine the differential equation that models this situation by summing the forces
Applications of Higher acting on the spring-mass system with
d 2x
Order Differential Equations m -0
dr
= S (forces acting on the system)
= -k(s + x) + mg
= - ks - kx + mg.
At equilibrium ks = mg, so after simplification we obtain the differential equation
or
The two initial conditions that are used with this problem are the initial position
x(O) = '" and the initial velocity x' (0) = (3. The function xU), which describes the dis-
placement of the object with respect to the equilibrium position at time t, is found by
M'--'" - .' :alue problem
I
n Chapter 4, we discussed ,several techniques for solving higher G - '
differential equations. In this chapter, we illustrate how those methods - ': ,,-
d2x + kx = 0
m dt 2
can be used to solve some initial-value problems that model physical
situations. 1x(O) = "', x' (0) = (3.
Based on the assumption made in deriving the differential equation, positive values of
x(t) indicate that the mass is below the equilibrium position and negative values ofxU)
~ Simple Harmonic Motion indicate that the mass is above the equilibrium position. The units that are encountered
in these problems are summarized in Table 5.1.
Suppose that an object of mass m is attached to an elastic spring that is suspended from
a rigid support such as a ceiling or a horizontal rod. The object causes the spring to
stretch a distance s from its natural length. The position at which it comes to rest is
called the equilibrium position. According to Hooke's law, the spring exerts a restor· TABLE 5.1 Units Encountered when Solving Spring-Mass Systems
ing force in the upward (opposite) direction that is proportional to the distance s that
the spring is stretched. Mathematically, this is stated as System Force Mass Length Time
F= ks, English pounds (lb) 2
slugs (lb-s /ft) feet (ft) lb/ft seconds (s)
Figure 5.1 A spring-mass where k> 0 is the constant of proportionality or spring constant. In Figure 5.1 we Metric Newton (N) kilograms (kg) meters (m) N/m seconds (s)
system. see that the spring has natural length b. When the object is attached to the spring, it is
238
~..
240 Chapter 5 Applications of Higher Order Differential Equations 5.1 Simple Harmonic Motion 241
1'1''''''.' (c) Because the mass is released from the equilibrium position, x(O) = O. Also, the
initial velocity is S fils in the downward (positive) direction, so x'(O) = S.
Determine the spring constant of the spring with natural length lOin. that is stretched '
o "d".....
to a distance of 13 in. by an object weighing 5 lb.
I Solution Because the mass weighs 5 lb, F = 5 lb, and because displacement from
the equilibrium position is 13 - 10 = 3 in., s = 3 in. X I filI2 in. = 114 ft. There. An object weighing 60 lb stretches a spring 6 in. Determine the function x(t) that
fore, describes the displacement of the object if it is released from rest 12 in. below the
F=ks equilibrium position.
5= 1..k :I Solution First, the spring constant k is determined from the supplied informa-
4 tion. By Hooke's law, F = ks, so we have 60 = k· 0.5. Therefore, k = 120 lb/ft.
k = 20. Next, the mass m of the object is determined using F = mg. In this case, 60 =
m . 32, so m = 15/S slugs. Because kim = 64, and 12 in. is equivalent to I ft, the
Notice that the spring constant is given in the units lblft because F = 5 Ib and initial-value problem that models the situation is
s = 114 ft.
~:~ + 64x = 0
"ii"I,INM I x(O) = I, x'(O) = O.
The characteristic equation that corresponds to the differential equation is r2 +
An object of weight 161b stretches a spring 3 in. Determine the initial-value prob. 64 = O. It has solutions r = ±Si, so a general solution of the equation is xU) =
lem that models this situation if (a) the object is released from a point 4 in. below CI cos St + C2 sin St.
the equilibrium position with an upward initial velocity of 2 fils; (b) the object is To find the values of CI and C2 that satisfY the initial conditions, we calculate
released from rest 6 in. above the equilibrium position; (c) the object is released x'(t) = -SCI sin St + SC2 cos St. Then x(O) = Cl = I and x'(O) = SC2 = O. So CI = I,
from the equilibrium position with a downward initial velocity of S fils. C2 = 0, and x(t) = cos St.
Nnti<"p '\.._L , . , indicates that the spring-mass system never comes
: .1 Solution We first determine the differential equation that models the spring- _.ow " 1S set mto motion. The solution is periodic, so the mass moves verti-
mass system (we use the same equation for all three parts of the problem). The in· cally, retracing its motion, as shown in Figure 5.2. Motion of this type is called sim-
formation is given in English units, so we use g = 32 filS2. We must convert all ple harmonic motion.
measurements given in inches to feet. The object stretches the spring 3 in., so L
s = 3 in. X I filI2 in. = 114 ft. Also, because the mass weighs 16 Ib, F = 16. Ac-
What is the maximum displacement of the object in Example 3 from the equilibrium
cording to Hooke's law, 16 = k· 114, so k = 64 lb/ft. We then fmd the mass m of
position?
the object with F = mg to find that 16 = m . 32 or m = 1/2 slug. The differentia]
equation used to find the displacement of the object at time t is
2
1..2 ddt2x + 64x = 0
.
"",,1,'.'
An objective weighing 2 lb stretches a spring 1.5 in. (a) Determine the function x(t)
The initial conditions for parts (a), (b), and (c) are then found. that describes the displacement of the object if it is released with a downward ini-
(a) Because 4 in. X I filI2 in. = 113 ft and down is the positive direction, x(O) : tial velocity of 32 fils from 12 in. above the equilibrium position. (b) At what value
113. Notice, however, that the initial velocity is 2 fils in the upward (negative) di· of t does the object first pass through the equilibrium position?
rection. Hence, x'(O) = -2.
(b) A position 6 in. (or 112 ft) above the equilibrium position (in the negative di· : :I Solution (a) We begin by determining the spring constant. Because the force
rection) corresponds to initial position x(O) = -112. Being released from rest indio F = 2 stretches the spring 3/2 in. X I filI2 in. = liS ft, k is found by solving 2 =
cates that dxldt(O) = O. k· liS. Hence k = 16 lb/ft. With F = mg, we have m = 2/32 = 1116 slug, The dif-
x (in tt) x (in it) x (in it) x(in it) with derivative x'(t) = -16cl sin 16t + 16c2 cos l6t. Application of the initial con-
ditions then yields x(O) = Cl cos 0 + C2 sin 0 = Cl = -I and
x'(O) = -16cl sin 0 + l6c2 cos 0 = 16c2 = 32, C2 = 2.
tt--H---t+-+--t-+-Jt-- t
(in s) The position function is given by x(t) = -cos l6t + 2 sin 16t. (See Figure 5.3.)
-1 (b) To determine when the object first passes through its equilibrium position, we
solve the equation x(t) = -cos 16t + 2 sin 16t = 0 or tan l6t = 112. Therefore,
t = 1116 tan- I (.5) = 0.03 s, which appears to be a reasonable approximation based
(a) (b) (e)
-2 on the graph of x(t).
x x
'-
Figure 5.3 Approximate the second time the mass considered in Example 4 passes through the
equilibrium position.
A general formula for the solution of the initial-value problem

d2X
m--;ji2+kx=O
(d) (e)
1
(I)
x x x(O) = a, dx (0) = {3
dt
is
x (t) = a cos wt
{3 .
+ -;;; Sin wt where w= [f;.
-1 -1
Through the use of the trigonometric identity cos (a + b) = cos a cos b - sin a sin b,
we can write x(t) in terms of a cosine function with a phase shift. First, let
(g) (h) (i)
x(t) = A cos (wt - </».
Figure 5.2
__ " 'I' T A Sin wt sin </>. Comparing the functions
x(t) = a cos wt + Ji sin wt and x(t) = A cos wt cos </> + A sin wt sin </>
w
ferential equation that models this situation is 1116 d 2xldt 2 + 16x = 0 or d 2xldt 2 + indicates that
256x = O. Because 12 in. is equivalent to I ft, the initial position above the equi·
librium (in the negative direction) is x(O) = -1. The downward initial velocity A cos</>= a and
(in the positive direction) is x' (0) = 32. Therefore, we must solve the initial-value
problem Thus
d2X + 256x = 0
cos </> = ~ and sin A-. =.Ji..
dt 2 A 'I' Aw'
(
x(O) = -I, x'(O) = 32.
Because cos 2 </> + sin </> = 1, (alA)' + [{3I(Aw)f =
2
1. Therefore, the amplitude of the
Because the characteristic equation is r2 + 256 = 0, the roots are r = ± 16i. A gen· solution is
eral solution is
x(t) = Cl cos 16t + C2 sin 16t,
-
and (b) From our knowledge of trigonometric functions, we know that the maximum
value ofx(t) = Vs cos (5t - q,) is x = Vs. Therefore, the maximum displacement
x(t) = Jcl- + ~ cos (wt - q,), of the mass from its equilibrium position is Vs = 2.236 meters.
(e) The period of this trigonometric function is T = 2'TTlw = 2'TT15. The mass returns
1 to its initial position every 2'TT15 = 1.257 s.
where q, = cos- ( V"
,,2 + {i-lw2
) and w = "I/ki;;. Note that the period ofx(t) is
In many cases, questions about the displacement function are more easily answered if EXERCISES 5.1
the solution is written in this form.
How does an increase in the magnitude (absolute value) of the initial position and
-
initial velocity affect the amplitude of the resulting motion of the spring-mass In Exercises 1-4, determine the mass m and the spring constant d 2x 1
system? From experience, does this agree with the actual physical situation? Hor the given spring-mass system.
10. 10 d(2 + lox = 0, x(O) = -5, x'(O) = 1
d2X 11. A 16-lb object stretches a spring 6 in. If the object is

m dt 2 + /0: = 0
"ii,,',I.,
lowered 1 ft below the equilibrium position and re-
1
leased, determine the displacement of the object. What
x(O) = a, : (0) = {3. is the maximum displacement of the object? When
does it occur?
A 4-kg mass stretches a spring 0.392 m. (a) Determine the displacement function
Interpret the initial conditions. (Assume the English system.) 12. A 4-lb weight stretches a spring 1 ft. A 16-lb weight
if the mass is released from I m below the equilibrium position with a downward
is then attached to the spring, and it comes to rest in
initial velocity of 10 mls. (b) What is the maximum displacement of the mass? d 2x
1. 4 dt' + 9x = 0, x(O) = -1, x'(O) = 0 its equilibrium position. If it is then put into motion
(e) What is the approximate period of the displacement function? with a downward velocity of 2 ftls, determine the dis-
d 2x , placement of the mass. What is the maximum dis-
2. 2 dt' + 128x = 0, x(O) = -0.5, x (0) = I
: J Solution (a) Because the mass of the object (in metric units) is m = 4 kg, we placement of the object? When does it occur?
use this with F = mg to determine the force. We first compute 2
*3. "4I ddt2x + 16x = 0, x(O) = 0.75, x'(O) = -2
*13. A 6-lb object stretches a spring 6 in. If the object is
lifted 3 in. above the equilibrium position and released,
F = (4)(9.8) = 39.2 N.
2 c1D'-- _~quired for the mass to return to
We then find the spring constant with 39.2 = k· 0.392, so k = 100 N/m. The
4 ..!... d x2
. 25 dt
+ 4x = 0 x(O)
'
= -..!..4' x'(O) = 1 _~ c<julllbrium position. What is the displacement of
differential equation that models this spring-mass system is 4 d 2xldt 2 + 100x = 0 the object at t = 5 seconds? If the object is released
2 2
or d xldt + 25x = O. The initial position is x(O) = 1, and the initial velocity is InExercises 5-10, express the ~ulutJ(in of the initial-value prob-
from its equilibrium position with a downward initial
velocity of I ftls, determine the time required for the
x (in meters) x'(O) = 10. We must solve the initial-value problem lem in the formx(t) = Va2 + ~/W2 cos (wt -</». What is the
object to return to its equilibrium position.
III ~ period and amplitude of the solution?
!
2 ~ {\ 1\
d2x + 25 = 0 2
14. A 16-lb weight stretches a spring 8 in. If the weight
dt 2 x d x
5. dt' +x = 0, x(O) = 3, x'(O) = -4 is lowered 4 in. below the equilibrium position and re-
t (ip s)
x(O) = 1, x'(O) = 10 leased, find the time required for the weigbt to return
d 2x to the equilibrium position. What is the displacement
+ 4x =
/n"1 rr
l fr either directly or with the general formula obtained previously. Using the general 6. dt 2 0, x(O) = I, x'(O) = 1 of the weight at t = 4 s? If the weight is released from
R2 g02
-1 formula with" = I, (3 = 10, and w = VI00/4 = 5, we have . 2 its equilibrium position with an upward initial veloc-
1 d x
*7. 16 d(2 + x = 0, x(O) = -2, x'(O) = 1 ity of 2 ftls, determine the time required for the weight
-2 V V V V x(t) = ,,2 + 2 cos (wt - q,) = 12 + - , cos (5t - </J) 2
to return to the equilibrium position.
w 5- d x
15. Solve the initial-value problem d 2xldt 2 + /0: = 0,
8. dt' + 256x = 0, x(O) = 2, x'(O) = 4
= Vs cos (5t - q,), 2
x(O) = -I. x'(O) = 0 for values of k = 1, 4, and 9.
Figure 5.4 Simple harmonic d x I Comment on the effect that k has on the resulting
motion. where q, = 1
cos- (l/Vs) = 1.11 rad, which we graph in Figure 5.4. 9. dt' + 9x = 0, x(O) = 3' x'(O) = -1
motion.
246 Chapter 5 Applications of Higher Order Differential Equations 5.2 Damped Motion 247
16. Solve the initial-value problem d'xldt' + .i. x = 0, Positive y-direction of the cylinder ifit is released with 1.22 in. of its height displacement from the equilibrium position; (b) the
m above the water with no initial velocity. (Hint: At time at which the object first passes through its equi-
x(O) = -1, x'(O) = 0 for values of m = 1,4, and 9. Equilibrium =
t 0, h = 9.78 in., so 0.22 in. is above the water. librium position; (e) the period of the motion.
Comment on the effect that m has on the resulting position Therefore, the initial position is yeO) = 1.) 29. If the spring in Problem 28 has the spring constant
motion.
*27. An object with mass m = 1 slug is attached to a spring k = 161b/ft, what is tbe maximum displacement from
*17. Suppose that a I-lb object stretches a spring 118 ft. the equilibrium position? How does this compare to
with spring constant k = 4 lb/ft. (a) Determine the dis-
The object is pulled downward and released from a "-"'-"--r"-----""h--+-Y'l0..----':!I-If--Water line the result in the previous exercise? Determine the max-
placement function of the object ifx(O) = "andx'(O)
position b ft beneath its equilibrium with an upward
= O. Graph the solution for" = 1,4, -2. How does =
imum displacement if k 24 lb/ft. Do these results
initial velocity of 1 ftls. Determine the value of b so agree with those that would be obtained with the gen-
varying the value of " affect the solution? Does it
that the maximum displacement is 2 ft. eral formula A = V d + ~/w2?
1
change the values of t at which the mass passes
18. Suppose that a 70-gram mass stretches a spring 5 cm through the equilibrium position? (b) Determine the 30. An object of mass m = 3 slugs is attached to a spring
and that the mass is pulled downward and released displacement function of the object if x(O) = 0 and with spring constant k = 15 Ib/ft. If the object is re-
from a position b units beneath its equilibrium with x' (0) = {3. Graph the solution for {3 = 1, 4, - 2. How leased from 9 in. below its equilibrium with a down-
an upward initial velocity of 10 cmls. Determine the does varying the value of {3 affect the solution? Does ward initial velocity of 1 ftls, find (a) the maximum
value of b so that the mass first returns to its equilib- Figure 5.5 it change the values of t at which the mass passes displacement from the equilibrium position; (b) the
rium at t = I s. (g = 980 cmls') through the equilibrium position? time at which the object first passes through its equi-
19. The period of the motion of an undamped spring-mass 28. An object of mass m = 4 slugs is attached to a spring librium position; (e) the period of the motion.
system is 7T/2 seconds. Find the mass m if the spring Of the displaced liquid equals the force of gravity on
the object. (See Figure 5.5.) Consider the cylinder of
=
with spring constant k 20 lb/ft. If the object is re- 31. If the mass in Problem 30 is m = 4 slugs, find the
constant is k = 32 lb/ft. leased from 7 in. above its equilibrium with a down- maximum displacement from the equilibrium position.
20. Find the period ofthe motion of an undamped spring- radius r and height H of which h units of the height Compare this result to that obtained with m = 3 slugs.
ward initial velocity of2.5 ftls, find (a) the maximum
is submerged at equilibrium.
mass system if the mass of the object is 4 kg and the
spring constant is k = 0.25 N/m. (a) Show that the weight of liquid displaced at equi.
21. If the motion of an object satisfies the initial-value librium is 7Tr'hp. Therefore, at equilibrium 7Tr'hp
= mg.
problem
d'x
(b) Lety(t) represent the vertical displacement of the ~ Damped Motion
m-+kx=O cylinder from equilibrium. Show that when the
2
dt cylinder is raised out of the liquid, the downward Because the differential equation derived in Section 5.1 disregarded all retarding forces
1 x(O) = ", : (0)
find the maximum velocity of the object.

= (3,
force is 77T'[h - y(t)]p.
(e) Use Newton's second law of motion to show that
2
m d yldt' = 7Tr'[h - y(t)]p - mg. Simplify this
acting on the motion of the mass, a more realistic model is needed. Studies in me-
chanics reveal that the resistive force due to damping is a function of ~he velocity of
the motion. For c > 0, functions such as
equation to obtain a second-order equation that
22. Show that the solution to the initial-value problem models this situation. 1, x' >0
~ I dt) , and FR = c Sgn(~), where Sgn(~) =
l
m~:~+kx=O 24. Determine if the cylinder can float in a deep pool of
water (p = 62.4 lb/ft') using the given radius r, height \
0, x'
\ -1, x'
=0
<0
H, and weight W: (a) r = 3 in., H = 12 in., W = Sib;
x(O) = ", dx (0) = (3 (b) r = 4 in., H= 8 in., W= 20 lb; (c) r = 6 in., can be used to represent the damping force. We follow procedures similar to those
dt used in Section 5.1 to model simple harmonic motion and to determine a differential
H= 9 in., W= 50 lb.
can be written as x(t) = u(t) + vet), where u and v sat- *25. Determine the motion of the cylinder of weight 512 equation that models the spring-mass system, which includes damping. Assuming that
isfy the same differential equation as x, u satisfies the Ib, radius r = 1 ft, and height H = 4 ft if it is released FR = c dxldt, after summing the forces acting on the spring-mass system, we have
initial conditions u(O) = ", duldt (0) = 0, and v satis- with 3 ft of its height above the water (p = 62.51b/ft J ) d 2x dx d 2x
fies the initial conditions v(O) = 0, dvldt (0) = {3. with a downward initial velocity of 3 ftls. What is the m -- = -c - - 10: or m- 2 + c -dx + 10: = O.
maximum displacement of the cylinder from its equi·
dt 2 dt dt dt
23. (Archimedes' principle) Suppose that an object of
mass m is submerged (either partially or totally) in a librium? The displacement function is found by solving the initial-value problem
liquid of density p. Archimedes' principle states that 26. Consider the cylinder of radius r = 3 in., heigbt 2
a body in liquid experiences a buoyant upward force H = 12 in., and weight 10 lb. Show that the portion m d x2 + c dx + 10: = 0
equal to the weight of the liquid displaced by the body. of the cylinder submerged in water of density (P) dt dt
The object is in equilibrium when the buoyant force J
=
62.5 Ib/ft is h 0.815 ft =
9.78 in. Find the motion \ x(O) = a, x' (0) = {3.
From our experience with second-order ordinary differential equations with con tan Differentiating yields
ff" . s t
coe IClents ill Chapter 4, the solutions to initial-value problems of this type depend 0
the values of m, k, and c. Suppose we assume (as we did in Section 5.1) that solut' n x' (t) = (-4cI + c2)e -41 - 4c2te -4'.
'ffi . 1 . IOns
o f th e dI erentm equation have the form xU) = e rl • Then the characteristic eq ti (a) The initial conditions in this case are x(O) = 0 and x'(O) = 1, so CI = 0 and
is mr2 + cr + k = 0 with solutions ua on
-4c, + C2 = 1. Thus, C2 = 1 and the solution is x(t) = te -41, which is shown in
r=
-c±~ Figure 5.6. Notice that x(t) is always positive, so the object is always below the equi-
2m . librium position and approaches zero (the equilibrium position) as t approaches in-
finity. Because of the resistive force due to damping, the object is not allowed to
The solution depends on the value of the quantity c 2 - 4mk. In fact problems of !hi pass through its equilibrium position.
type are classified by the value of c 2 - 4mk as follows. ' s
What is the maximum displacement from the equilibrium position? Jfyou were
Case 1: c? - 4mk > 0 looking at this spring, would you perceive its motion?
This situation is said to be overdamped, because the damping coefficient c is lar . x (in ft) (b) In this case, x(O) = -112 and x' (0) = 5. When we apply these initial conditions,
comparison with the spring constant k. ge m we find that x(O) = CI = -112 and x'(O) = -4cI + C2 = -4(-112) + C2 = 5.
Hence C2 = 3, and the solution is
Case 2: CO - 4mk = 0
This situati~n is described as criticaUy damped, because the resulting motion is os-
Cillatory, With a slight decrease in the damping coefficient c.
This function, which is graphed in Figure 5.7, indicates the importance of the ini-
Case 3: CO - 4mk < 0 tial conditions on the resulting motion. In this case, the displacement is negative
I (in s) (above the equilibrium position) initially, but the positive initial velocity causes the
This situation .is called underdamped, because the damping coefficient c is small in
companson with the spnng constant k. function to become positive (below the equilibrium position) before approaching
,i,.".".,
zero.
Figure 5.6 Critically damped
motion.
Jf the object in Example 1 is released from any point below its equilibrium position
with an upward initial velocity of 1 fils, can it possibly pass through its equilibrium
An 8-lb object is atta~hed to a. spring of length 4 ft. At equilibrium, the spring has position? Jf the object is released from below its equilibrium position with any up-
x (in ft)
le~gth of 6 ft. Determme the displacement function xU) if FR = 2 dxldt and (a) the ward initial velocity, can it possibly pass through the eqUilibrium position?
object IS release~ fro~ the equilibrium position with a downward initial velocity of 0.1
1 fils; (b) the object IS released 6 in. above the equilibrium position with an initial
0.5 1.5 t (in s)
velocity of 5 fils in the downward direction.
I Soluti~n Notice that s = 6 - 4 = 2 ft and that F = 8 lb. We find the spring con-
stant With 8 = k . 2, so k = 4 lb/ft. Also, the mass of the object is m = 8132 = 1/4
-0.1
-0.2
-0.3
"""'4-A 32-lb object stretches a spring 8 ft. If the resistive force due to damping is FR =
slug. The differential equation that models this spring-mass system is -0.4 5 dxldt, determine the displacement function if the object is released from 1 ft be-
low the equilibrium position with (a) an upward velocity of 1 fils; (b) an upward
-0.5
1 d 2x dx d 2x dx velocity of 6 fils.
"4 dt 2 + 2 dt + 4x = 0 or --;;;z + 8 dt + 16x = O.
Figure 5.7 Critically damped
motion. :.1 Solution (a) Because F = 32 lb, the spring constant is found with 32 = k(8), so
The corresponding characteristic equation is
k = 4 lb/ft. Also, m = 32/32 = I slug. The differential equation that models this sit-
r2 + 8r + 16 = (r + 4)2 = 0, uation is d 2xldt 2 + 5 dxldt + 4x = O. The initial position is x(O) = 1, and the ini-
tial velocity in (a) is x' (0) = -1. The characteristic equation of the differential equa-
so r = -4 is a root of mUltiplicity two. A general solution is tion is
xU) = c,e- 4, + C2te-4'. 1'2 + 51' + 4 = (1' + 1)(1' + 4) = 0,
x (in ft) with roots rl = -1 and r2 = -4. A general solution is d2X + dx + 16x = 0
dt 2 dt
0.8 {
and : (t) = -cle-' - 4c2e-4'. x(O) = 0, x' (0) = 1.
0.6 2
2 . 0 f ddt x dx + 16x = 0 IS
.
(Because c - 4mk = 52 - 4(1)(4) = 9 > 0, the system is overdamped.) Applica. A general solutIOn 2 + dt
0.4
tion of the initial conditions yields the system of equations {Cl + C2 = I with
x(t) = e _,/2 (Cl cos -
3V7t . -3V7t)
0.2 -Cl - 4C2 = -I
solution {Cl = I, C2 = O}, so the solution to the initial-value problem is x(t) = e-I
2- + C2 sm 2- .
4 The graph of x(t) is shown. in Figure 5.8(a). Notice that this solution is always pos~
Itlve and, due to the dampmg, approaches zero as t approaches infinity. Therefore x (in ft) Notice that c2 - 4mk = (1/2)2 - 4(1/2)(S) = -63/4 < 0, so the spring-mass sys-
(a)
the object is always below its equilibrium position. ' tem is underdamped. Because
(b) Using the initial velocity x'(O) = -6, we solve the system {Cl + C2 = I
x'(t) = -2I e -II"- ( c] 3V7t
cos - 2- . -3V7t)
+ C2 sm 2-
x (in ft) -Cl - 4C2 = -6'
which has solution {CI = -2/3, C2 = 5/3}. The solution of the initial-value problem is
3V7 ( . 3V7t 3V7t)
0.8 + e -,/2 -2- -C] sm - 2 - + C2 cos - 2 - ,
0.6
0.4
-0.2 application of the initial conditions yields {Cl = 0, C2 = 2/(3 V7)}. Therefore, the
The graph of this function is shown in Figure 5.8(b). As in Example I, these results
solution is
0.2 indicate the importance of the initial conditions on the resulting motion. In this case,
the displacement is positive (below its equilibrium) initially, but the larger negative ' (3V7t)
(a) () = -2- e - , / 2sm
xt --.
-0.2
initial velocity causes the function to become negative (above its equilibrium) be- 3V7 2
fore approaching zero. Therefore, we see that the initial velocity in part (b) causes x (in ft)
the mass to pass through its equilibrium position. Solutions of this type have several interesting properties. First, the trigono-
(b)
L metric component of the solution causes the motion to oscillate. Also, the expo-
•o
Figure 5.8 nential portion forces the solution to approach zero as t approaches infinity. These
Can you find a minimum value for the upward initial velocity Vo in Example 2 so qualities are illustrated in the graph of x(t) in Figure 5.9(a). Physically, the position
that the object passes through its eqUilibrium position for all values greater than vo? of the mass in this case oscillates about the equilibrium position and eventually
comes to rest in the equilibrium position. Of course, with our model the displace-
ment function x(t) --'> 0 as t --'> 00, but there are no values t = T such that x(t) = 0
-0.2
for t > T as we might expect from the physical situation. Our model only approxi-
"fi,,','·' mates the behavior of the mass. Notice also that the solution is bounrl~rl "hr.",! and
below by the ell _______._ .• "~V~5.1 the
A 16-lb object stretches a spring 2 ft. Determine the displacement x(t) if the resis- (b)
horizontal axis, -e '''". This is illustrated with the simultaneous display of these
tive force due to damping is FR = (1/2) dxldt and the object is released from the Figure 5.9 functions in Figure 5.9(b); the motion of the spring is illustrated in Figure 5.10.
equilibrium position with a downward velocity of I fils.
'-
:11 Solution Because F = 16 lb, the spring constant is determined with 16 = k· 2. Notice that when the system is underdamped as in Example 3, the amplitude (or
Hence, k = 8 lb/ft. Also, m = 16/32 = 112 slug. Therefore, the differential equation damped amplitude) of the solution decreases as t --'> 00. The time interval between two
is successive local maxima (or minima) of x(t) is called the quasiperiod of the solution.
I d 2x I dx 2
d x Approximate the quasiperiod of the solution in Example 3.
--+--+Sx=O
2 dt 2 2 dt
or - + -dx
dt 2
+
dt
16x=0
.
In Section 5.1 we developed a general formula for the displacement function. We
The initial position is x(O) = 0 and the initial velocity is x' (0) = I. We must solve can do the same for systems that involve damping. Assuming that the spring-mass
the initial-value problem system is underdamped, the differential equation m d 2xldt 2 + c dxldt + Icc = 0 has the
- x (inft) x
5.2 Damped Motion 253
t
(in s)
2" t
characteristic equation mr2 + cr + k = 0 with roots r = (-c ±
2
2
4mk - c )/(2m). tv
Ifwe let p = c/(2m) and f.L = v'4mk - c /(2m), then r = -p ± f.Li and a general so-
lution is x(l) = e-P'(cl cos f.Lt + C2 sin f.LI). Applying the initial conditions x(O) = a
-0.2 -0.2 and x'(O) = f3 yields the solution X(I) = e- p ,( a cos f.LI + f3: ap sin f.LI), which can
x x
be written as
x(i) = Ae- P' cos (f.LI - cp).
Then, x(l) = e -P'(A cos f.LI cos cp + A sin f.LI sin cp). Comparing the functions
20T t
2" , x(l) = e -Pt( a cos f.LI + --f.L-
f3 + ap.
SID f.LI )
-0.2
and
-0.2
x(l) = e - PI(A cos f.LI cos cp + A sin f.LI sin cp),
x x
we have
. f3+ap
Acoscp=a and A SIDCP=---,
f.L
which indicates that
27r t
2" I sin .I. = f3 + ap
and 'f' Af.L·
-0.2 -0.2 Now, cos 2 cp + sin2 cp = 1, so a/A2 + [(f3 + ap)/(Af.L)f = 1. Therefore, the decreas-
ing amplitude of the solution is
x x
A= J a [f3-+-a-
2 +
f.L
p
]2
and
20T t 21T t x(l) = e- Pi J+ ~ [f3 + ap

--f.L- ]2 cos (f.LI - cp),
a \
-0.2 -0.2 where cp = cos- 1 (
\i -,.,/'t""J f
x The quantity
47rm
27r'
is called the quasiperiod of the function. (Note that functions of this type are not pe-
riodic.) We can also determine the times at which the mass passes through the equi-
librium position from the general formula given here. We do this by setting the argu-
-0.2 -0.2 ment equal to odd multiples of 7r/2, because the cosine function is zero at these values.
The mass passes through the equilibrium position at
Figure 5.10
25"4 5.2 Damped Motion 255
Chapter 5 Applications of Higher Order Differential Equations
[ (2n + 1)
t= ~_ _,,_ _-2. -
f+ cf>]
_ [m(2n
•
+ 1)'1T + 2mcf>l ,
~ n = 0, ±I, ±2,
.. EXERCISES 5.2
r V4mk-c·
In Exercises 1-4, determine the mass m (slugs), spring constant d 2x dx
Use this formula to find the quasiperiod for the solution in Example 3. 9. dt 2 + 8 dt + 15x = 0, x(O) = 0, x'(O) = I
kO b/ft ), and damping coefficient c inFR = c dxldl for the given
spring-mass system d 2x dx
10. dJl + 7 dt + 12x = 0, x(O) = -1, x'(O) = 4
d2X dx
m-, +c-+kx=O
dr dt 2
An ~bject .of mass I slu~ is attached to a spring with spring constant k = 13 Ib/ft { d " 3dx
*11. dJl + '2 I x = 0, x(O) = -I, x'(O) = 2
dt + '2
x(O) = a, x' (0) = /3.
and. I.S s~bJected to a reslstiv.e .force of ~R = 4 dxldt due to damping. If the initial
posltlon IS x(O) = I and the mltlal velOCity is x' (0) = I, determine the quasiperiod Describe the initial conditions. (Assume the English system.) d 2x dx
12. dJl + 5 dt + 4x = 0, x(O) = 0, x'(O) = 5
of the solutlOn and fmd the values of I at which the object passes through th . 2
librium position. e equi' d x dx
1. dJl + 4 dt + 3x = 0, x(O) = 0, x'(O) = -4 2
d " + 8 dt
13. dJl dx + 16x = 0, x(O) = 4, x'(O) = -2
J Solution The initial-value problem that models this situation is I d x 2
dx
2. 32 dt2 + 2 dt + x = O. x(O) = I, x'(O) = 0 d 2x dx
14. dJl + 6 dt + 9x = 0, x(O) = 3, x'(O) = -3
1
~;~~'4:+13x=0 *3.
I d x
4'
dJl
2
+ dx
2 dt +x = 0, x(O) = -0.5, x'(O) = I 2
d x dx
*15. dt 2 + 10 dt + 25x = 0, x(O) = -5. x'(O) = I
x(O) = I, x'(O) = 1.
2
d x dx
4.4 dl 2 +2 +8x=0,x(0)=0,x'(0)=2 d 2x dx I
The characteristic equation is r2 + 4r + 13 = 0 with roots r = - 2 + 3' A dt 16. dJl+dt+4'x=O,x(O)= -1,x'(0)=2
eral solution is x(t) = e- 21(cl cos 31 + C2 sin 3t) with deriva:;~e - l. gen-
In Exercises 5-8. express the solution of the initial-value prob- 17. Suppose that an object with m = I is attached to the
dx . end of a spring with spring constant k = I. After
21
dt (t) = -2e- (cl cos 31 + C2 sin 3t) + 3e- 21
(-cl sin 31 + C2 cos 31). lem in the form
reaching its equilibrium position, the object is pulled
Ap~lication of th: initial conditions yields the solution x(t) = e -21(cos 31 + sin 31), X(I) = e -J'
pI a- + [/3
- -+
j .a
p
t -]2 cos (j.tt - cp).
one unit above the equilibrium and released with an
initial velocity Vo. If the spring-mass system is criti-
which can be wntten as cally damped. what is the value of c?
-'1
x (I) = e· J a2 + [({3 + a )]2
--j.L-- cos (j.LI -
p
cf» =
In each case, find the quasiperiod and the time at which the
mass first passes through its equilibrium position.
18. A weight having mass m = I is attached to the end of
a spring with k = 5/4 andFR = 2 dxldt. Determine the
displacement of the mass if the object is released from
1.5
x (inft)
e -21 J2 [I
I +-
+ -3 2]2 cos (31 - cf» = .V;:::
2e- 21
cos (31 - cp).
d 2x
5. dl 2
d x 2
dx
dx
+ 4dt + 13x = O,x(O) = l,x'(O) =-1
the equilibrium position with an initial velocity of
3 units/s in the downward direction.
*19. A 32-lb weight is attached to the end of a spring with
6. dJl + 4 dt + 20x = 0, x(O) = 1, x'(O) = 2 spring constant k = 24 Ib/ft. If the resistive force is FR
The quasiperiod is
0.5 = 10 dxldt, determine the displacement of the mass if
d 2x dx it i~ rp.leased with no initial velocity from a position
t (in s) 471711 4'1T 2'1T *7. dJl + 2 dt + 26x = 0, x(O) = I, x'(O) = I
2 3
-0.5 Y4(13) - 4 d 2x dx
mass passes through its equilibrium position and, if
8. dJl + 10 dt + 41x = 0, x(O) = 3, x'(O) = -2 so, when. Determine the maximum displacement of
-1 The values of I at which the mass passes through the equilibrium are the object from the equilibrium position.
-1.5
1= m(2n + 1)'1T + 2mcf> _ (2n + 1)'1T + 2cf> In Exercises 9-16, solve the initial-value problem. Classify each 20. An object weighing 8 Ib stretches a spring 6 in. be-
Y 4mk _ c2 - 6 ' n = 0, 1,2, as overdamped or critically damped. Determine if the mass yond its natural length. If the resistive force is FR =
Figure 5.11 X(I) = V2e- 2' passes through its equilibrium position and. if so, when. Deter- 4 dxldt, find the displacement of the mass if it is set
cos (3t - cp), cp = Tr14. where cf> = cos- 1 (IIYz) = '1T14. (See Figure 5.11.) mine the maximum displacement of the object from the equi- into motion from its equilibrium position with an ini-
c... librium position. tial velocity of 1 fils in the downward direction.
256 Chapter 5 Applications of Higher Order Differential Equations 257
5.3 Forced Motion
21. An object of mass m = 70 kg is attached to the end 38. Solve the initial-value problem 4x" + ex' + 5x = 0,
*29. Suppose that the spring-mass system 36. In Problem 35, consider the solution that results us-
of a spring and stretches the spring 0.25 m beyond its ing the damping coefficient that produces critical x(O) = 3, x'(O) = 0 using c = I, -4, and O. Plot the
{ m~:~+c~+kx=O
naturallength. Ifthe restrictive force isFR = 280 dx/dl damping with x'(O) = -1,0, I, and 2. In which case solutions and compare them.
find the displacement of the object if it is released does the object pass through its equilibrium position? 39. Solve the initial-value problem x" + ex' + x = 0,
from a position 3 m above its equilibrium position with x(O) = a, x'(O) = {3 When does it pass through the equilibrium position? x(O) = -1,x'(O) = 3 for values ofc = 2 andc = vB.
no ~iti~1 veloci~. Does the object pass through its (Notice the effect of the coefficient of damping on the
37. Using the values of x'(O) and c in Problem 36, in ad·
eqUlhbnum posltlOn at any time? ~ critically dampe~. If (3 = 0, show that limH~ X(I) solution to the corresponding critically damped and
dition suppose that the equilibrium position is I unit
22. Suppose that an ohject of mass m = I slug is attached - 0, but that there IS no value I = 10 such that x(lo) ~ overdamped equation.)
above the floor. Does the obj ect come into contact with
to a spring with spring constant k = 25 Ib/ft. If the re- O. (!he mass approaches but never reaches its equi.
hbnum position.) the floor in any of the cases? If so, when?
sistive force is FR = 6 dx/dl, determine the displace-
ment of the object ifit is set into motion from its equi- 30. Suppose that the spring-mass system
hbnum position with an upward velocity of 2 fils.
What is the quasiperiod of the motion?
m ~;~ + c ~ + kx = 0
*23. ~ obje~t of mass m = 4 slugs is attached to a spring { ~ Forced Motion
With spnng constant k = 64 Ib/ft. If the resistive force x(O) = a, x' (0) = {3
IS FR = c dx/dl, find the value of c so that the motion In some cases, the motion of a spring is influenced by an external driving force, f(I).
is critically damped. If {3 > 0, find a condition on Il
IS critically damped. For what values of c is the mo- Mathematically, this force is included in the differential equation that models the situ-
s~ that the mass passes through its equilibrimn posi.
tion underdamped?
lion after it is released. ation by
24. An object of mass m = 2 slugs is attached to a spring
31. In the case of underdamped motion, show that the 2
with spnng constant k Ib/ft. If the resistive force is
FR = 8 dx/dl, find the value of k so that the motion is
amount of time between two successive times at which mddlx2 = -kx -
dx
cdt + f(l)·
the mass passes through its equilibrium is one-half of
critically damped. For what values of k is the motion
the quasiperiod.
underdamped? For what values of k is the motion over- The resulting initial-value problem is
damped? 32, In the case of underdamped motion, show that the
d2X dx
25. If the quasiperiod of the underdamped motion is
amount of time between two success;V positive max.
Ima of the position function is 471711 4km - c2 •
m dl 2 + C dt + kx = f(l)
71'/6 seconds when a 1/13 slug mass is attached to a {
spring with spring constant k = 13 Ib/ft, find the 33. In the case of underdamped motion, show that the x(O) = a, x' (0) = {3.
damping constant c. ratio between two consecutive maxima (or minima) is
e2c1r/V 4mk-c 2• Therefore, differential equations modeling forced motion are nonhomogeneous and re-
26. If a mass of 0.2 kg is attached to a spring with spring quire the method of undetermined coefficients or variation of parameters for solution.
constant k = 5 N/m that undergoes damping equiva. 34. The natural logarithm of the ratio in Exercise 33
We first consider forced motion that is undamped.
lentto ~ dx/dl, find the quasiperiod of the resulting called the logarithmic decrement, is '
motIOn.
; """.".'
d= 2c71'
lSO~U~:; :(I~ th~ :~I~ValUe problem
27. Y4mk-c 2
Show that the
and indicates the rate at which the motion dies out be.
An object of mass m = I slug is attached to a spring with spring constant k =
cause of damping. Notice that because m, k, and dare
4 Ib/ft. Assuming there is no damping and that the object begins from rest in the
all quantities that can be measured in the spring-mass
x(O) = a, dx (0) = (3 equilibrium position, determine the position function of the object if it is subjected
dl ~ystem, the value of d is useful in measuring the damp·
mg constant c. Compute the logarithmic decrement of to an external force of (a).," ~ 0: (b)f(t) = 1; (e)f(l) = cos I.
Can be written as X(I) = u(l) + V(I), where u and v sat-
the system in (a) Example 3 and (b) Example 4.
isfy the same differential equation as x u satisfies the '~'Q nroblem
initial conditions u(O) = a, u' (0) = 0 'and v satisfies 35. Determine how the value of c affects the solution of : J Solution First, we note that we must solve u,~ -
the initial conditions v(O) = 0, v' (0) ~ {3. the initial-value problem
28. If the spring-mass system m d 2x/dl 2 + c dx/dl + kx
~2~ + 4x = f(l)
X(~) = 0, x'(O) = 0
d2X dx
= 0 is either critically damped or overdamped, show dt' +c'dt+ 6x =0 {
that the mass can pass through its equilibrium posi- {

x(O) = 0, x'(O) = I,
tion at most one time (independent of the initial con- for each of the forcing functions in (a), (b), and (e). A general solution of the ho-
ditions).
where c = 2\16, 4\16, and \16. mogeneous problem d 2xldl 2 + 4x = 0 is Xh(l) = CI cos 21 + C2 sin 21.
~ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _--JlI.-._ _ _ _ _ _ _ ~~~_
5.3 Forced Motion
(a) Witb/(t) = 0, tbe equation is homogeneous, so we apply tbe initial conditions homogeneous equation. "'e
VV'
consider this type of situation with the initial-value
to x(t) = c] cos 2t + C2 sin 2t witb derivative x'(t) = -2cI sin 2t + 2cz cos 2t. Be. problem
cause x(O) = CI = 0 and x' (0) = 2C2 = 0, C] = Cz = 0, tbe solution is x(t) = o. This '
solution indicates that tbe object does not move from tbe equilibrium position be. dZx2 + u/x = FI cos UJt + F z sin wt + G(t)
cause tbere is no forcing function, no initial displacement from tbe equilibrium po- dt
sition, and no initial velocity. \ x(O) = a, x' (0) = {3,
(b) Using the metbod of undetermined coefficients with a particular solution oftbe nstants and G is any function of t. (Note that one of the co~
form xpCt) = A, substitution into the differential equation d 2xldtZ + 4x = 1 yields where Fl and F2 are co I d G can be identically the zero function.) In thiS
4A = 1, so A = i. Hence, stants F] and:~ c~ ~~u:.a:~~1 a;;equency of the system because a general solution
case, we say a w . . () _ wt + C sm wt. In tbe
x (in ft)
x(t) = Xh(t) + xpCt) = CI cos 2t + C2 sin 2t + t, of the corresponding homogeneous equatIOn 1S Xh t - CI cos
e of tbis initial-value problem, tbe forced frequency, tbe frequency 0
2 f th tri ono
e g -
~~tric functions in F] cos wt + F2 sin wt + GCt), equals tbe natural frequency.
0.75 with derivative x' Ct) = - 2cI sin 2t + 2C2 cos 2t. Witb x(O) = C] + *= 0 and
0.5 x'(O) = 2cz = 0, we have C] = -i and C2 = 0, so
0.25
-0.25
x(t) = _.l4 cos 21 + .l
4· 'f",,"'*)
. Investi"ate the effect that tbe forcing functionfCt) = cos 2t has on the solution of
-0.5 Notice from tbe graph of this function in Figure S.12(a) tbat tbe object never moves
the initial-value problem
-0.75 above tbe equilibrium position. (Positive values of x indicate that tbe mass is below
-J tbe equilibrium position.) d2x
-+
2 4x =f(t)
(e) In this case, we assume tbat xit) = A cos t + B sin t. Substitution into d 2 xld? + dt
(a) 4x = cos t yields 3A cos t + 3B sin t = cos t, so A = t and B = O. Therefore, \ x(O) = 0, x'(O) = O.
x (inft)
xCt) = xhCt) + xit) = c] cos 2t + C2 sin 2t + t cos t, 'I Solution
. I 1 ()-
As we saw m Ex~mp e , Xh t - c]
cos 2t + C sin 2t. Because f(t) =
_ 2 2t + Bt sin 2t with
0.75 cos 2t is contained in tbis solutIOn, we assume that xit) - At cos ,
with derivative
0.5 first and second derivatives
0.25
x'(t) = -2cI sin 2t + 2cz cos 2t - t sin t. x~(t) = A cos 2t - 2At sin 2t +B sin 2t + 2Bt cos 2t
-0.25 and
-0.5
Applying tbe initial conditions then gives us x(O) = c] + t = 0 and x' (0) = 2C2 = 0,
so C] = -t and C2 = O. Thus, x;(t) = -4A sin 21 - 4At cos 2t + 4B cos 2t - 4Bt sin 2t.
-0.75
-J . . Z + 4x = cos 2t yields -4A sin 2t
. t dZxldtO + 4B cos 2t = cos 2t.
xCt) = -3I cos 2t + 3I cos t. SubslltutIOn m t .
Thus A = 0 and B = 1/4, so xp(t) = .t sm 2t, and
(b)
The graph of one period of this solution is shown in Figure S.12(b). In this case, 1 . 2
x(t) = Xh(t) + xit) = c] cos 2t + C2 sin 2t + "4 t sm t.
Figure 5.12 Ca) Graph of the mass passes through tbe equilibrium position twice (near t = 2 and t = 4) over
x(t) = -±
cos 21 + ±. (b) Graph tbe period, and it returns to tbe equilibrium position witbout passing through it at
t
of xCI) = -t cos 21 + cos I.
t = 27T. (Can you predict other values oft where this occurs?) This function has derivative
L
x'(t) = -2cI sin 2t + ~ .. - 2t + -1.
4
1
2
2
sm 2t + - t cos t,
How does changing the initial position to x(O) = I affect the solution in (c)? How
does changing the initial velocity to x' (0) = I affict the solution? Is the maximum so application of tbe initial conditions yields X\'. -. = 0 and x'(O) = 2C2 = o.
displacement affected in either case?
*
Therefore, c] = C2 = 0, so
When we studied nonhomogeneous equations, we considered equations in which
the nonhomogeneous (right-hand side) function was a solution of tbe corresponding x(t) = t sin 2t.
5.3 Forced Motion
x
The graph of this solution is shown in Figure 5.13. Notice that the amplitude in. Notice that the solution can be represented as
creases as t increases. This indicates that the spring-mass system will encounter a .<
serious problem; either the spring will break or the mass will eventually hit its sUp- . (w + f3)t
x(t) = A(t) SIn - - 2 - '
port (like a ceiling or beam) or a lower boundary (like the ground or floor).
where
The phenomenon illustrated in Example 2 is called resonance and can be ex. 2F . (w - f3)t
A(t) = w2 _ 132 SIll --2--'
-4
tended to other situations such as vibrations in an aircraft wing, a skyscraper, a glass,
or a bridge. Some of the sources of excitation that lead to the vibration of these struc-
tures include unbalanced rotating devices, vortex shedding, strong winds, rough sUr. Therefore, when the quantity (w - 13) is small, (~ + 13) is relatively larg~ in com~
. Th function sin (w + f3)tl2 oscillates qUIte frequently because It has pe
faces, and moving vehicles. Many engineers must overcome problems caused When
Figure 5.13 Resonance. p.anson. e 13) M hil the function sin (w - f3)tl2 oscillates relatively slowly
structures and machines are subjected to forced vibrations. nod 47T/( w + . eanw e, th funcf
because it has period 47T/( w - 13). When we graph x(t), we see that e IOns
Over a suffiCient amount of time, do changes in the initial conditions affect the mo-
2F . (w - f3)t form an envelope for the solution.
tion of a spring-mass system? Experiment by changing the initial conditions in the ± w2 _ ~ SIn 2
initial-value problem in Example 2.
Let us investigate in detail initiar~value problems of the form
1
d2x
dt
2
- 2 + w x = F cos f3t, w '" 13
x(O) = 0, x' (0) = o.

o 'ifi""'·'
Solve the initial-value problem
A general solution of the corresponding homogeneous equation is Xh(t) = 2
CI cos wt + C2 sin wt. Using the methodof undetermined coefficients, a particular so- -d X
+ 4x=f(t)
dt 2
lution is given by xAt) = A cos f3t + B sin f3t. The corresponding derivatives of this {
x(O) = 0, x' (0) = 0,
solution are
x~(t) = -Af3 sin f3t + Bf3 cos f3t and x;(t) = -Af32 cos f3t - Bf32 sin f3t. with (a) f(t) = cos 3t and (b) f(t) = cos St.
Substituting into the nonhomogeneous equation d 2xldP + w2x = F cos f3t and equat- :)1Solution (a) A general solution of the corresponding ~omogeneo.u~ equation is
ing the corresponding coefficients yields : . 2t By the method of undetermIned coeffICIents, we as-
, Xh(t) = CI cos 2t + C2 SIn . "4 3t 'elds
: that x (t) = A cos 3t + B sin 3t. Substitution into x + x = cos ,)'1 3
and B= O. (a) .' sume P
-5Acos3t-5Bsin3t=cos3t,soA=-1 an IS d B -
-.0 Thus xP
(t) = --5
cos t
and
Therefore, a general solution is
. I
x(t) = Xh(t) + xAt) = CI cos 2t + C2 SIll 2t - '5 cos 3t.
x(t) = CI cos wt + C2 sin wt + w2 ~ ~ cos f3t.
Because the derivative of this function is
Application of the initial conditions yields the solution
3 . 3
F x'(t) = -2cI sin 2t + 2C2 cos 2t + '5 SIn t,
x(t) = - 2 - - ' (cos f3t - cos wt).
w - 13-
application of the initial conditions yields x(O) = CI - 115 = 0, and x' (0) = 2C2 =
Using the trigonometric identity Hcos (A - B) - cos (A + B)] = sin A sin B, we have O. Therefore, CI = 1/5 and C2 = 0, so
(b)
x(t) = ~ sin (w + f3)t sin (w - f3)t. x(t) = '51 cos 2t - '51 cos 3t.
w 2 _f32 2 2 Figure 5.14
263
262 Chapter 5 Applications of Higher Order Differential Equations 5.3 Forced Motion
. I -I 4 -I' 2
The graph of this function is shown in Figure 5.14(a) along with the envelo
functions pe
,. x(t) = Xh(t) + xit) = C1 cos 2t + C2 sm 2t + 17 e cos 2t - 17 e sm t.
2 . t 2 . t Applying the initial conditions with x(t) and

5 sm 2 and -5 sm 2' . 9 -I 2 + 2 -I' 2
x'(t) = -2C1 sm 2t + 2C2 cos 2t - 17 e cos t 17 e sm t,
(b) In a similar manner, the solution of the initial-value problem
we have
d2X
d2 + 4x = cos 5t I
+ 17 = and x'(O) = 2C2 - ;7 = o.
X(~) = 0, x'(O) = 0
x(O) = C1 0
(
Therefore, C1 = -1/17 and C2 = 9/34, so

is
Figure 5.15 1 9. I -I 2 4 -I' 2
I I x(t) = -17 cos 2t + 34 sm 2t + 17 e cos t - 17 e sm t,
x(t) = 21 cos 2t - 21 cos 5t.
which is graphed in Figure 5.15. Notice that the effect of terms involving the ex-
~he graph of the solution is shown in Figure 5.14(b) along with the envelope func. ponential function diminishes as t increases. In this case, the forcing function f(t)
tlons .. , = e - I cos 2t approaches zero as t increases. Over time, the solution of the nonho-
mogeneous problem approaches that of the corresponding homogeneous problem,
2 . 3t 2 . 3t
21 sm 2 and -21 sm 2' so we observe simple harmonic motion as t -> 00.
Some computer algebra sys-
tems contain commands that We now consider spring problems that involve forces due to damping as well as
allow us to "play" functions. Oscillations like those illustrated in Example 3 are called beats because of the
If this is possible with your periodic variation of amplitude. This phenomenon is commonly encountered when two external forces. In particular, consider the initial-value problem
technology, see Exercises 27 musicians try to simultaneously tune their instruments or when two tuning forks with d2X dx
and 28. almost equivalent frequencies are played at the same time. m --, +C- + kx = p cos At
dt- dt
Consider the problem (
x(O) = a, x' (0) = f3,
~2~ + 4x = cos f3t which has a solution of the form

(
X(~) = 0, x'(O) = 0 x(t) = h(t) + s(t),
for f3 = 6, 8, and 10. What happens to the amplitude of the beats as f3 increases? where limt->oo h(t) = 0 and set) = C1 cos At + C2 sin At.
The function h(t) is called the transient solution, and s(t) is known as the steady-
state solution. Therefore, as t approaches infinity, the solution x(t) approaches the
steady-state solution. (Why?) Note that the steady-state solution co~r:sponds to a par-
£3 ,'i',,"'" ticular solution obtained through the method of undetermined coeff,c,ents or vanatlOn
Investigate the effect that the forcing functionf(t) = e- r cos 2t has on the initial· of parameters.
value problem
d2X
dt 2 + 4x = f(t) 'if''''''·'
( Solve the initial-value problem
x(O) = 0, x' (0) = o.
- -2 + 4dx- +
d2X
13x = cos t
dt dt
: ] Solution Using the method of undetermined coefficients, a general solution of (
the equation is x(O) = 0, x'(O) = I
264 Chapter 5 Applications of Higher Order Differential Equations 5.3 Forced Motion 265
tha~ models the motion of an object of mass m = I slug attached to a spring wi x x x
spnng constant k = 13 Ib/ft that is subjected to a resistive force of FR = 4 dxl:

0.15 0.15
a.nd an external force of f(t) = cos t. Identify the transient and steady-state sol.
tIOns. U
0.1
0.1
l\
'I Solu6on A general solution o f h - + 4 -dx + \3x = 0 is xh(t) = i \
dt 2 dt
e- (c] cos 3t + C2 sin 3t). We assume that a particular solution has the ~
2/
10/ \5 t -0.001
xi t) = A cos 1.+ B sin t, with. de~vatives x~(t) = -A sin I + B cos I and x;(t~r::: \j \
-A cos t - B sm t. After substItutIOn mto d 2xldt 2 + 4 dxldl + \3x = cos t we
12A + 4B = I ' see (b) (e)
(a)
that { -4A + 12B = 0 must be satisfied. Hence A = 3/40 and B = 1140, so
Figure 5.16
3 I .
xit) = 40 cos I + 40 sm t.
Therefore,
EXERCISES 5.3
X(I) = x;,(t) + xiI) = e -2/(C] cos 3t + C2 sin 3t) + .l.- cos t + J.... sin t
40 40' 1. An 8·lb weight stretches a spring 1 ft. Ifa 16·lbweight 6. An object of mass m = 2 kg is attached to a spring
is attached to the spring, it comes to rest in its equi· with spring constant k = I kg/m. If the resistive force
with derivative is FR = 3 dxldl and the external force is f(l) =
Iibrium position. If it is then put into motion with a
downward initial velocity of 2 ftls, determine the dis· 2 cos wi, find the displacement of the object if
x'(t) = -2e- 2/(c] cos 3t + C2 sin 3t) placement of the mass if there is no damping and an x(O) = 0 and x'(O) = O. Will resonance occur for any
external force f(t) = cos 31. What is the natural fre- values of w?
" 3t + C2 cos 3t) - - 3 sm
+ 3e -2/( -c] sm . t + - I cos t quency of the spring-mass system?
" 40 40' *7, An object of mass m = I slug is attached to a spring
2. A 16·lb weight stretches a spring 6 in. If the mass is with spring constant k = 25 lb/ft. If the resistive force
lowered I ft below its equilibrium position and re- is FR = 8 dxldl and the external force is f(l) =
Application of the irntial conditions yields x(O) = c] + (3/40) = 0 and x'(O) = leased, determine the displacement of the mass if there cos I - sin I, find the displacement of the object if
-2c] + 3C2 + (1/40) = I. Therefore, c] = -3/40 and C2 = 11/40, so is no damping and an external force off(l) = 2 cos I. x(O) = 0 and x'(O) = O.
What is the natural frequency of the spring-mass
x(t) -- e -2/( - 403 cos 3t + 11

40 sm .3t) + 40
3 cos t + 40
I sin t. system? 8. An object of mass m = 2 slug is attached to a
spring with spring constant k = 6 lb/ft. If the resistive
*3. A 16-lb weight stretches a spring 8 in. If the mass is
lowered 4 in. below its equilibrium position and re- force is FR = 6 dxldl and the external force isf(t) =
2 sin 21 + cos I, find the displacement of the object
This indicates that the transient solution is e -2'( -i<; cos 3t + k\ sin 3t) and the steady. leased, determine the displacement of the mass if there
is no damping and an external force off(t) = 2 cos I. if x(O) = 0 and x' (0) = O.
sta~e solution is i<; cos t + .'0 sin t. We graph this solution in Figure 5.16(a). The so·
lutIOn and the steady-state solution are graphed together in Figure 5.16(b). Notice 4, A 6-lb weight stretches a spring 6 in. The mass is 9. Suppose that an object of mass 1 slug is attached to a
that the two curves appear identical for t > 2.5. The reason for this is shown in the raised 3 in. above its equilibrium position and released. spring with spring constant k = 4 lb/ft. If the motion
subsequent plot of the transient solution (in Figure 5.16(c)), which becomes quite Determine the displacement of the mass if there is no of the object is undamped and subjected to an exter-
small near t = 2.5. damping and an external force off(l) = 2 cos 51. nal force of f(l) = cos I, determine the displacement
5, An object of mass m = I kg is attached to a spring of the object if x(O) = 0 and x' (0) = O. What func-
with spring constant k = 9 kg/m. If there is no damp- tions envelope this displacement function? What is the
Over a sufficient amount of time, do changes in the initial conditions in Example 5 ing and the external force is f(l) = 4 cos wi, find the maximum displacement of the object? If the external
affect the motion of the spring-mass syslem? Experiment by changing the initial displacement of the object if x(O) = 0 and x'(O) = O. force is changed to f(l) = cos (tI2), does the maxi-
conditions in the initial-value problem in Example 5. What must the value of w be for resonance to occur? mum displacement increase or decrease?
266 Chapter 5 Applications of Higher Order Differential Equations 5.4 Other Applications 267
10. Anobject of mass I slug is attached to a spring with I, 0 OS I OS I 22. Solve the initial-value problem x" + x = cos wI, 26. Consider the function g (y) =
spnng constant k = 25 Ib/ft. If the motion of the ob-
J(I) = 2 - I, I < I OS 2 x(O) = 0, x'(O) = 1 using values w = 0.9 and FN (wz - -I l' + 4A defined in Exercis\, zq.
2-1
ject is undamped and subjected to an external force of { w = 0.7. Graph the solutions simultaneously to deter- (a) Graph this function for k = 4, m = I, F = 2, and
*
J(I) = cos {31. (3 5. determine the displacement of 0,1>2.
mine the effect that the value of w has on each the damping coefficient: c = 2A = 2, I, 0.75, 0.50,
the object ifx(O) = 0 andx'(O) = O. What is the value (Hinl: Solve the initial-value problem over each' solution. 0.25. (b) Graph the function for k = 49, m = 10,
of (3 if the maximum displacement of the object is t~rval. Choos~ constants appropriately so that the fun~: *23. Investigate the effect that the forcing functions
F = 20, and the damping coefficient: c = 2A = 2, 1,
21l1V2 ft? tlOns x and x are continuous.) 0.75,0.50,0.25. In each case, describe what happens
(a) J(t) = cos 1.9t and (b) f(t) = cos 2.lt have on
*11. Anobject of mass 4 slugs is attached to a spring with 18. Find the solution to the initial-value problem to the maximum magnitude of g (y) as c -> O. Also,
the initial-value problem
s~r~ng constant k = 26 Ib/ft. It is subjected to a re- as c -> 0, how does the resonance frequency relate to
~:~ + 4x = J(t)
d 2x dx the natural frequency of the corresponding undamped
sIstIve force of FR = 4 dxldl and an external force dl2 +4 dl + 13x = J(I), x(O) = 0, x' (0) ~ 0
J(I) = 250 sin I. Determine the displacement of the system?
object ifx(O) = 0 andx'(O) = O. What is the transient
solution? What is the steady-state solution? whereJ(I) =
1,0 OS I OS '1T'
I - I, '1T'< los 2'1T' \
x(O) = 0, ! (0) = O.
*27. To hear beats, solve the initial-value problem
12. Anobject of mass I slug is attached to a spring with {

0,1>2'1T'. How do these results differ from those of Example 3?
dZ~
dl-
+ w2x = F cos {31, w * (3
spnn~ constant k = 4000 Ib/ft. It is subjected to a Are there more or fewer beats with these two func- {
resIstIve force of FR = 40 dxldl and an external force 19. Show that a general solution of x(O) = 0, x'(O) = 0
tions?
J(I) = 600 sin I. Determine the displacement of the d 2x dx using W Z = 6000, f3 = 5991.62, and F = 2. In each
dl 2 + 2Adi + w2x = Fsin-yt
object if x(O) = 0 and x' (0) = O. What is the transient case, plot and, ifpossible. play the solution. (Note: The
solution? What is the steady-state solution? x" + O.lx' + x = 3 cos 2t, x(O) = O. x'(O) = 0 purpose of the high frequencies is to assist in hearing
is
13. Find the solution of the differential equation using the method of undetermined coefficients and the solutions when they are played.)
d 2x
X(I) = Ae- Ar sin cV w2 - A21 + <1» compare the result with the forcing function 3 cos 2t. 28. To hear resonance, solve the initial-value problem
m dl 2 + kx = F sin wi, w * -...;7J;;, F Determine the phase difference between these two
+ -=-yr(:=w"'2=_=-1~)2=+=4A=:02=='-1 sin (-yt + 0), functions. dZ~ + w2x = F cos {3t. w 1'(3
~at satisfies the initial conditions: (a) x(O) = a, dl
25. Solve the following initial-value problem involving a {
x (0) = 0; (b) x(O) = 0, x'(O) = (3; (c) x(O) = a where A = Vof + d, A < w, and the phase angles ¢ piecewise defined forcing function over [0, 2]:
x(O) = 0, x'(O) = 0
x'(O) = (3.
14. Find the solution of the differential equation
d x2
' and 0 are found with sin = cIIA, cos = czlA,
sin 0 = -2Ay
-1)2 + 4A2 -1'
' {I,
d-x
-+x=
Z
dt
2-1
'
OOStosl
1< t OS 2
using w2 = 6000, {3 = 6000, and F = 2. In each case,
plot and, if possible, play the solution. (Note: The pur-
pose of the high frequencies is to assist in heariu!; the
1
dx Y(W2 _
m dl 2 + C di + kx = F sin wI, c 2
- 4mk < 0 0, 1>2. solutions when they are played.)
and x(O) = a, x'(O) = b
~at satisfies the initial conditions: (a) x(O) = a, 2
coso= w _-I Graph the solution using the initial conditions a = I,
x (0) = 0; (b) x(O) = 0, x'(O) = (3; (c) x(O) = a
~~=~ , Y(wz _ -I)z + 4A z-l' b = l;a = O,b = I; and a = I;b= O.
*15. Find the solution to the initial-value problem 20. The steady-state solution (the approximation of the
solution for large values of I) of the differential equa·
d2~ + x = {I,D OS I OS '1T' _ , tIon In Exercise 19 isxh(t) = g(y) sin (-yt + 0), where
dr 0, I > '1T' ' x(O) - 0, x (0) = o. ~ Other Applications
g(y) = FIV(w - -If + 4Az-l. By differentiating
Z
(Hinl: Solve the initial-value problem over each in- g WIth respect to y, show that the maximum value (' L-R-C Circuits (' Deflection of a Beam
t~rval. Choose constants appropriately so that the func- of g( y) occurs when y = V WZ - 2A z. The quantity c
tIons x and x' are continuous.) 2
Y w - 2A /2'1T' is called the resonance frequency for
2
L-R-C Circuits
16. Find the solution to the initial-value problem the system. Describe the motion if the external force
Second-order nonhomogeneous linear ordinary differential equations arise in the study
2
d x {cos I, 0 os I os '1T' has frequency V wZ - 2A zI2'1T'.
d(2 + X 0, I> '1T' ' x(O) = 0, x'(O) = o. of electrical circuits after the application of Kirchhoff's law. Suppose that let) is the
= 21. Solve the initial-value problem x" + x = cos t,
current in the L-R-C series electrical circuit (shown in Figure 5.17) where L, R, and C
x(O) = 0, x' (0) = busing b = 0 and b = I. Graph the
17. Find the solution to the initial-value problem represent the inductance, resistance, and capacitance of the circuit, respectively.
solutions simultaneously to determine the effect that
d 2x R The voltage drops across the circuit elements in Table 5.2 have been obtain~d
.d12 + X = J(I), x(O) = O. x'(O) = 0, where the nonbomogeneous initial velocity has on the solu·
tion to the second initial-value problem as I increases. Figure 5.17 An L-R-C circuit. from experimental data, where Q is the charge of the capacitor and dQldt = I.
269
268 Chapter 5 Applications of Higher Order Differential Equations 5.4 Other Applications
d2Q dQ I
2 + R - + -Q = E(t)
TABLE 5.2 L-
Circuit Element
Inductor
Voltage Drop
L:!!..
1 Q(O)
dt dt C
= Qo, 1(0) = Q' (0) = 10
for the charge Q(t). The solution is differentiated to find the current I(t).
dt
l'f·"I,'.'
Resistor RI
Capacitor J.. Q
C Consider the L-R-C circuit with L = I henry, R = 40 ohms, C = 114000 farads,
and E(t) = 24 volts. Determine the current in this circuit if there is zero initial cur-
rent and zero initial charge.
Our goal is to model this physical situation with an initial-value problem so that
we can determine the current and charge in the circuit. For convenience, the terminol_ :.1 Solution Using the indicated values, the initial-value problem that we must solve is
ogy used in this section is summarized in Table 5.3.
d2Q + 40 dQ + 4000Q = 24
dt 2
Gustav Robert Kirchhoff
(1824-1887). Gennan physicist;
worked in spectrum analysis,
optics, and electricity.
TABLE 5.3
Electrical Quantities Units

1 Q(O) =
dt
0,1(0) = Q'(O) = o.
The characteristic equation of the corresponding homogeneous equation is r2 +
(Northwind Picture Archives) 40r + 4000 = 0 with roots rI.2 = -20 ± 60i, so a general solution of the corre-
Inductance (L) Henrys (H) 20
sponding homogeneous equation is Qh(t) = e- ,(c! cos 60t + C2 sin 60t). Because
Resistance (R) Ohms (0) the voltage is the constant function E(t) = 24, we assume that the partIcular solu-
2
tion has the form 0 (t) = A. Substitution into d 2Qldt + 40 dQldt + 4000Q = 24
yields 4000.4 = 24~r A = 3/500. Therefore, a general solution of the nonhomoge-
Capacitance (C) Farads (F)
Charge (Q) Coulombs(C) neous equation is

Current (I) Amperes (A)
Q(t) = Qh(t) + Qp(t) = e-
20I
(c! cos 60t + C2 sin 60t) + 5~0'
with derivative
The physical principle needed to derive the differential equation that models the 20I
Q'(t) = -20e- 20I(c! cos 60t + C2 sin 60t) + 60e- (-cl sin 60t + C2 cos 60t).
L-R-C series circuit is Kirchhoff's law.
After application of the initial conditions, we have
3 dQ (0) = -20c! + 60C2 = O.
Kirchhoff's law: Q(O) = CI + 500 = 0 and
dt
The sum of the voltage drops across the circuit elements is equivalent to the volt- Therefore, CI = - 3/500 and C2 = -11500, so the charge in the circuit is
age E(t) impressed on the circuit.
20
Q(t) = e- ' (- 5~0 cos 60t - 5~0 sin 60t) + 5~0'
Applying Kirchhoff's law with the voltage drops in Table 5.2 yields the differential and the current is given by
equation L dIldt + RI + (l/C)Q = E(I). Using the fact that dQldt = I, we also have
Q '(t) = _..l9... e - 2o'(-3 cos 60t - sin 60t) + ...&Q..e- 2o'(3 sin 60t - cos 60t)
d Qldt 2 = dIldt. Therefore, the equation becomes L d 2Qldt 2 + R dQldt + (l/C)Q ~
2
500 500
E(t), which can be solved by the method of undetermined coefficients or the method
of variation of parameters. If the initial charge and current are Q(O) = Qo and 1(0) ~ = 2 e - 201 sin 60t.
Q' (0) = 10 , we solve the initial-value problem 5
5.4 Other Applications 271
270 Chapter 5 Applications of Higher Order Differential Equations
These results indicate that in time the charge approaches the constant value of 3/500 Boundary conditions for this problem may vary. In most cases, two conditions are
which is known as the steady-state charge. Also, because of the exponential term' given for each end of the beam. Some of these conditions, which are specified in pairs
the current approaches zero as t increases. We show the graphs of Q(t) and I(t) ~ at x = p, where p = 0 or p = a, include s(p) = 0, s'(p) = 0 (fixed end); s"(p) = 0,
Figures 5.l8(a) and (b) to verify these observations. s"'(p) = 0 (free end); s(p) = 0, s"(p) = 0 (simple support); and s'(p) = 0, s"'(p) = 0
L- (sliding clamped end).
In Example 1, how is the charge Q(t) affected if E(t) = 48 volts? What happens to
Q (in C)
P Q(t) if R = 40vlO ohms?
Deflection of a Beam
"f',,"'·'
Solve the beam equation over the interval 0 :5 x :5 1 if E = 1= 1, w(x) = 48, and
0.008 An important mechanical model involves the deflection of a long beam that is sup. the following boundary conditions are used: s(O) = 0, s'(O) = 0 (fixed end at x =
ported at one or both ends, as shown in Figure 5.19. Assuming that in its undeflected 0); and
0.006 form the beam is horizontal, then the deflection of the beam can be expressed as a (a) sell = 0, s"(l) = 0 (simple support at x = 1)
function of x. Suppose that the shape of the beam when it is deflected is given by the (b) s"(l) = 0, s"'(l) = 0 (free end at x = 1)
0.004 (c) s'(1) = 0, s"'(l) = 0 (sliding clamped end at x = 1)
graph of the function y(x) = - sex), where x is the distance from the left end of the
beam and s the measure of the vertical deflection from the equilibrium position. The (d) s(1) = 0, s'(1) = 0 (fixed end at x = 1).
0.002
boundary value problem that models iliis situation is derived as follows.
t(in s) 4 4
Let m(x) equal the turning moment of the force relative to the point x, and w(x) , :1 Solution We begin by noting that the differential equation is d s/dx = 48, which
0.1 0.2 0.3 represent the weight distribution of the beam. These two functions are related by the . is a separable equation that can be solved by integrating each side four times to yield
(a) equation
sex) = 2x4 + CIX 3 + C2X2 + C3X + C4,
[(inA)
with derivatives s'(x) = 8x 3 + 3CIX2 + 2C2X + C3, s"(x) = 24x2 + 6CIX + 2C2,
0.25
and s"'(x) = 48x + 6CI' (Note that we could have used the method of undetermined
0.2 Also, the turning moment is proportional to the curvature of the beam. Hence coefficients or variation of parameters to find this solution.) We next determine the
(ll + (:rr
0.15 . EI dZs arbitrary constants for the pair of boundary conditions at x = O. Because s(O) =
0.1 2 C4 = 0 and s' (0) = C3 = 0,
m(x) = dx '
0.05 sex) = 2x4 + CIX 3 + C2X2.
0.2 0.3 t where E and I are constants related to the composition of the beam and the shape and (a) Because s(1) = 2 + Cl + C2 = 0 and s"(l) = 24 + 6CI + 2cz = 0, Cl = -5
and C2 = 3. Hence sex) = 2x4 - 5x 3 + 3x 2. We can visualize the shape of the beam
-0.05
size of a cross section of the beam, respectively. Notice that this equation is nonlinear.
This difficulty is overcome with an approximation. For small values of s, the denom- by graphing y = -sex) as shown in Figure 5.20(a). (b) In this case, s"(l) =
(b)
inator of the right-hand side of the equation can be approximated by the constant I. 24 + 6Cl + 2C2 = 0 and s"'(l) = 48 + 6Cl = 0, so Cl = -8 and C2 = 12. The de-
Figure 5.18 (a) Q(t) (b) let). Therefore, the equation is simplified to flection of the beam is given by sex) = 2x4 - 8x 3 + l2xz. We graph y = -sex) in
Figure 5.20(b). We see from the graph that the end is free at x = 1. (c) Because
d 2s
m(x) = EI dx 2 ' s'(1) = 8 + 3Cl + 2cz = 0 and s"'(1) = 48 + 6Cl = 0, C, = -8 and C2 = 8. There-
beam fore, sex) = 2x4 - 8x 3 + 8xz. From the shape graph ofy = -sex) in Figure 5.20(c),
This equation is linear and can be differentiated twice to obtain we see that the end at x = 1 is clamped as compared to the free end in (b). (d) In
x=o x=a
2 4
this instance s(1) = 2 + Cl + C2 = 0 and s'(1) = 8 + 3Cl + 2C2 = 0, so Cl = -4
(a) d m = EI d s and C2 = 2. Thus, sex) = 2x4 - 4x 3 + 2x 2 . Y = -sex) is graphed in Figure 5.20(d).
dx z dx 4 ' Notice that both ends are fixed. Finally, all four graphs are shown together in
x=o x=a
Figure 5.21 to compare the different boundary conditions.
which is then used with the equation above relating m(x) and w(x) to obtain the single
I............... !s(x)~ fourth-order linear nonhomogeneous differential equation
(b) 4
d s If we had used free ends at both x = 0 and x = 1 in Example 2, what is the dis-
EI dx 4 = w(x). placement? Is this what we should expect from the physical problem?
Figure 5.19
272 Chapter 5 Applications of Higher Order Differential Equations 5.4 Other Applications 273
-s -s -S 9. A beam of length lOis fIXed at both ends. Detennine (a) s(O) = 0, s"(O) = 0, (simple support at x = 0);
the shape of the beam if the weight distribution is the s(IO) = 0, s"(10) = 0 (simple support atx = 10)
0.4 0.6 0.8 \x .2 constant function w(x) = 8, with constants E and I
-0.05 -\ -0.5 (b) s(O) = 0, s"(O) = 0, (simple support at x = 0);
such that EI = 100, EI = 10, and EI = I. What is the s"(10) = 0, s"'(10) = 0 (free end at x = 10)
-2 -\
-0.\ displacement of the beam from s = 0 in each case?
-3 -1.5 (c) s(O) = 0, s"(O) = 0, (simple support at x = 0);
How does the value of EI affect the solution?
-0.\5 -2 s'(10) = 0, s"'(IO) = 0 (sliding clamped end at
-4 10. Suppose that the beam in Exercise 9 is fIXed at x = 0 x = 10)
-2.5
-0.2 -5 and has simple support at x = 10. Detennine the max-
-3 (d) s(O) = 0, s"(O) = 0, (simple support at x = 0);
imum displacement using constants E and I such that
-0.25 -6 s(10) = 0, s'(IO) = 0 (fixed end at x =.10)
-3.5 EI = 100, EI = 10, and E[ = I.
Discuss the differences brought about by these condi-
(a) *11. Consider Exercise 9 with simple support at x = 0 and tions. (See Exercise 23.)
(b)
Figure 5.21 x = 10. How does the maximum displacement com-
15. Consider the L-Cseries circuit in whichE(t) = 0 mod-
-s pare to that found in each case in Problems 9 and 10?
-s eled by the initial-value problem
12. Detennine the shape of the beam of length 10 with
.2 0.4 0.6 0.8 \x constants E and I such that EI = I, weight distribu- d2Q I
-0.02 L-+ -Q=O
-0.5 tion w(x) = x 2 , and boundary conditions: dt 2 C
-0.04 {
(a) s(O) = 0, s'(O) = 0, (fIXed end at x = 0); Q(O) = Qo,I(O) = Q'(O) = o.
-\ -0.06 s(IO) = 0, s"(10) = 0 (simple support at x = /0)
Find the charge Q and the current 1. What is the max-
-0.08
-1.5 (b) s(O) = O. s'(O) = 0, (fixed end at x = 0); imum charge? What is the maximum current?
-0.\ s"(IO) = 0, s"'(10) = 0 (free end at x = 10)
16. Consider the L-C series circuit in which E(t) = 0 mod-
-2 -0.12 (c) s(O) = 0, s'(O) = 0, (fixed end at x = 0); eled by the initial-value problem
s'(10) = 0, s"'(10) = 0 (sliding clamped end at
(e) x = 10) d2Q I
(d)
L7+c Q =0
Figure 5.20 (d) s(O) = 0, s'(O) = 0, (fixed end at x = 0); {
s(IO) = 0, s'(IO) = 0 (fixed end at x = 10) Q(O) = 0, 1(0) = Q'(O) = [0.
Discuss the differences brought about by these condi- Find the charge Q and the current 1. What is the max-
tions. (See Exercise 21.) imum charge? What is the maximum current? How do
*13. Determine the shape of the beam of length 10 with these results compare to those of Exercise 15?
EXERCISES 5.4
constants E and I such that EI = I, weight distribu- *17. Consider the L-R-C circuit modeled by
tion w(x) = x 2 , and boundary conditions:
d20 dQ I .
(a) s(O) = 0, s"(O) = 0, (simple support at x = 0); L-y+R-+-Q=Eosmwl
dt dt C
In Exercises 1-4, find the charge on the capacitor and the cur· s(10) = 0, s"(IO) = 0 (simple support at x = 10) {
6. Consider the L·R-C circuit given in Exercise 5 with Q(O) = Qo,/(O) = Q'(O) = 10 ,
rent in the L·C series circuit (in which R = 0) assuming that E(t) = I. Detennine the value of Q(I) as t approaches (b) s(O) = 0, s"(O) = 0, (simple support at x = 0);
Q(O) = 0 and 1(0) = Q'(O) = O. infinity. s"(IO) = 0, s"'(IO) = 0 (free end at x = 10) Find the steady-state current, Iim,....oo [(I), of this
problem. Note that in this fonnula, Lw - IICw
1. L = 2 henry, C = 1/32 farad, and E(t) = 220 volts. (c) s(O) = 0, s"(O) = 0, (simple support at x = 0);
*7. Consider the solution to the L·R·C series circuit is called the reactance of the circuit, and
2. L = 2 henry, C = 1/50 farad, and E(t) = 220 volts. s'(IO) = 0, s"'(IO) = 0 (sliding clamped end at
indicated in Exercise 5. In this case, let E(t):
x = 10)
V (Lw - I/Cwi + R2 is called the impedance of
*3. L = ~ henry, C = 1/64 farad, and E(t) = 16t volts. 126 cos t + 5000 sin t. Detennine the solution to this the circuit, where both of these quantities are meas-
4. L = ~ henry, C = 1/64 farad, and E(t) = 16 sin 4t initial-value problem. At what time does the charge (d) s(O) = 0, s"(O) = 0, (simple support at x = 0); ured in ohms.
volts. first equal zero? What are the steady-state charge and s(10) = 0, s'(IO) = 0 (fixed end at x = 10)
18. Compute the reactance and the impedance ofthe cir-
current?
5. Find the charge Q(t) on the capacitor in an L·R·C Discuss the differences brought about by these condi· cuit in Example I if E(t) = 24 sin 4t.
series circuit if L = 0.2 henry, R = 25 ohms, C = tions. (See Exercise 22.) 19. Show that the maximum amplitude of the steady-
8. If the resistance, R, is changed in Exercise 5 to
0.00 I farad, E(t) = 0, Q(O) = 0 coulombs, and 1(0) = R = 8 ohms, What is the resulting charge on the ca· 14. Detennine the shape of the beam of length /0 with state current found in Exercise 17 occurs when
Q'(O) = 4 amps. What is the maximum charge on the pacitor? What is the maximum charge attained, and constants E and [ such that E[ = I, weight distribu- w = I/YLC. (In this case, we say that electrical res-
capacitor? when does the charge first equal zero? tion w(x) = 48 sin mil 0 and boundary conditions: onance occurs.)
274 Chapter 5 Applications of Higher Order Differential Equations 5.5 The Pendulum Problem 275
20. (Elastic sbaft) The differential equation that models Discuss the differences brought about by these The momentum of the mass is given by m dsldt, so the rate of change of the mo-
the torsional motion of a weight suspended from an conditions. mentum is
elastic shaft is I d 2 0ldt 2 + c dOldt + kO = T(t), where
orepresents the amount that the weight is twisted at
22. Repeat Exercise 21 with the following boundary
conditions: d (ds)
dt m dt = m dt 2 '
2
d s
time t, I is the moment of inertia, c is the damping
constant, k is the elastic sbaft constant (similar to the (a) s(O) = 0, s"~O) = 0, (s~ple support at x'" 0);
s(10) = 0, s (10) = 0 (SImple support atx '" 10) where s represents the length of the arc formed by the motion of the mass. Then, be-
spring constant), and T(t) is the applied torque. Con-
(b) s(O) = 0, s"(O) = 0, (simple support at x'" 0). mg sin (J mg cos(J cause the force mg sin (J acts in the opposite direction of the motion of the mass, we
sider the differential equation with I = I, c = 4 and
W=mg
k = 13. Find O(t) if (a) TCt) = 0, 0(0) = 00 , and s"(IO) = 0, s"'(10) = 0 (free end at x = 10) , have the equation
axis
0' (0) = 0; (b) TCt) = sin 77X, 0(0) = 00 , and (c) s(O) = =
0, s"(O) 0, (simple support at x'" 0)' d 2s .
0' (0) = O. Describe the motion that results in each s'(10) = 0, s"'(10) = 0 (sliding clamped end a; Figure 5.23 A force diagram m dt 2 = -mgsm (J.
case. x = 10) for the swinging pendulum.
(d) s(O) = 0, s"(O) = 0, (simple support at x'" 0)' (Notice that the force mg cos (J is offset by the force of constraint in the rod, so mg
To better understand the solutions to the elastic beam problem, s(IO) = 0, s'(10) = 0 (fixed end at x = 10) , and mg cos (J cancel each other in the sum of the forces.) Using the relationship from
we can use a graphics device to determine the shape of the beam geometry between the length of the are, the length of the rod, and the angle (J, s = L(J,
under different boundary conditions. 23. Graph the solution of the beam equation if the beam
has length 10, the constants E and I are such that we have the relationship
21. Graph the solution of the beam equation if the beam EI = I, the weight distribution is w(x) '" 48 sin
has length 10, the constants E and I are such that EI .( 77X/I 0), and the boundary conditions are the same as
= I, the weight distribution is w(x) = x 2 , and the in Problem 22.
boundary conditions are 2
24. Repeat Exercise 21 with EI = 10. .
(a) s(O) = 0, s'(O) = 0, (fixed end at x = 0);
The displacement (J (t) satls
. f"les mL""dt2
d (J = -mg sm
. (J or
25. Repeat Exercise 22 with EI = 100. How do these
s(lO) = 0, s"(10) = 0 (simple support at x = 10) Figure 5.24
solutions compare to those in Exercise 22?
(b) s(O) = 0, s'(O) = 0, (fixed end at x = 0); 26. Repeat Exercise 23 with EI = 100. How do these
s"(IO) = 0, s"'(10) = 0 (free end at x = 10) solutions compare to those in Exercise 23?
(c) s(O) = 0, s'(O) = 0, (fixed end at x = 0); 27. Attempt to find solutions of the beam equation in which is a nonlinear equation. However, because we are only concerned with small
s'(10) = 0, s"'(10) = 0 (sliding clamped end at displacements, we note from the Maclaurin series for sin (J,
Exercises 21-23 using other combinations of bound·
x= 10)
ary conditions. How do the differing boundary condi·
(d) s(O) = =
0, s'(O) = 0, (fixed end at x 0); tions affect the solution? sin(J=(J-.f...+!.._ .. ·
3! 5! '
s(lO) = 0, s'(10) = 0 (fixed end at x = 10)
and that for small values of e, sin e = e. Therefore, we obtain the linear equation
mL d 2 eldt 2 + mg e = 0 or d 2eldt 2 + (gIL) (J = 0, which approximates the original
problem. If the initial displacement (position of the mass) is given by (J(O) = eo and
the initial velocity (the velocity with which the mass is set into motion) is given by
~ The Pendulum Problem e' (0) = va, we have the initial-value problem
Suppose that a mass m is attached to the end of a rod of length L, the weight of which d2e + ~e = 0
is negligible. (See Figure 5.22.) We want to detennine an equation that describes the
motion of the mass in tenus of the displacement 6(t), which is measured counter-
clockwise in radians from the vertical axis shown in Figure 5.22. This is possible if we
1 dt 2 L
(J(O) = eo, e' (0) = Va
are given an initial position and an initial velocity of the mass. A force diagram for to find the displacement function eu)·
2 2
this situation is shown in Figure 5.23. Suppose that o} = giL so that the differential equation becomes d (Jldt + o}e
Notice that the forces are detenuined with trigonometry using the diagram in = O. Therefore, functions of the form
Figure 5.23. In this instance, cos (J = mglx and sin (J = mgly, so we obtain the forces e(t) = CI cos wt + C2 sin wt,
axis
Figure 5.22 A swinging

x = mg cos (J and y = mgsin (J, where w = Viii, satisfy the equation d 2eldt 1 + giL e = O. When we use the condi-
pendulum. which are indicated in Figure 5.24. tions (J(O) = eo and e' (0) = va, we find that the function
Incorporating this force into the sum of the forces acting on the pendulum, we obtain
IJ(t) = IJo cos wt + Vo sin wt
w the nonlinear equation L d 2IJ/dt 2 + b dIJ/dt + g sin IJ = O. Again, using the approxi-
2 2
mation sin IJ "" IJ for small values of IJ, we use the linear equation L d IJ/dt + b dIJ/dt
satisfies the equation as well as the initial displacement and velocity conditions. As
did with the position function of spring-mass systems, we can write this function We
+ gIJ = 0 to approximate the situation. Therefore, we solve the initial-value problem
. fun· . Iudes a phase shift with
ctlOn that mc asa
l
cosme L d2IJ +b dIJ + IJ = 0
dt 2 dt g
IJ(t) =
2 V5
IJo + w2 cos (wt - </», IJ(O) = IJo, IJ' (0) = Vo
where </> = cos-I(IJO/Ve5 + V6/(2) and w = Viii. to find the displacement function IJ(t).
Note that the approximate period of IJ(t) is T = 271"/w = 271"VTJg.
"f."I,'.' ,'ii,,','N,
A pendulum ofiengthL = 8/5 ft is subjected to the resistive force FR = (32/5) dIJ/dt
Determine the displacement of a pendulum of length L = 8 feet if IJ(O) = 0 and
due to damping. Determine the displacement function if IJ(O) = I and IJ'(O) = 2.
IJ' (0) ~ 2. What is the period? If th.e pendulum is part of a clock that ticks once for
each tIme the pendulum makes a complete swing, how many ticks does the clock
make in one minute?
Solution The initial-value problem that models this situation is
:1 Solution
ation is
Because giL = 32/8 = 4, the initial-value problem that models this situ-
d2IJ
I I d2IJ
5 dt 2
+ B.
IJ(O) = I, IJ'(O) = 2.
dIJ
5 dt
2
+ 32IJ = 0
Simplifying the differential equation, we obtain d 2IJ/dt + 4 dIJ/dt + 20IJ = 0, which

7+4IJ= 0 has characteristic equation r2 + 4r + 20 = 0 with roots rl.2 = -2 ::': 4i. A general
1 7r/2 7r t
solution is IJ(t) = e -21(CI cos 4t + C2 sin 4t). Application of the initial conditions
IJ(O) = D, IJ' (0) = 2. -0.5 yields the solution IJ(t) = e -21( cos 4t + sin 4t). We graph this solution in Figure
A .general solution of the differential equation is IJ(t) = CI cos 2t + C2 sin 2t, so ap- 5.25. Notice that the damping causes the displacement of the pendulum to decrease
plication of the initial conditions yields the solution IJ(t) = sin 2t. The period of this Figure 5.25 over time.
function is l.
T= 271"
~g
fi.. = 271"J 328filsft 2
= 71"S.
When does the object in Example 2 first pass through its equilibrium position? What
is the maximum displacement from equilibrium?
(Notice that we can use our knowledge of trigonometry to compare the period with In many cases, we can use computer algebra systems to obtain accurate approx-
T = 271"/2 = 71".) Therefore, the number of ticks made by the clock per minute is imations of nonlinear problems.
c.alculat.ed with the conversion (l rev/71" sec) X (l tickll rev) X (60 s/I min) "" 19.1
trcks/mm. Hence the clock makes approximately 19 ticks in one minute.
How is. motion affected if the length of the pendulum in Example 1 is changed to o ,'i,,,,,'M'
L = 4?
Use a computer algebra system to approximate the solutions of the nonlinear
If the pendulum undergoes a damping force that is proportional to the instanta- problems
neous velocity, the force due to damping is given by
I
d2IJ . I d2IJ + 32 dIJ + 32 sin IJ = 0
7 +4 IJ = 0 5 dt 2
1
(a) Sill and (b) 5 dt
FR = b dIJ
dt· IJ(O) = IJ, IJ'(O) = 2 IJ(O) = I, IJ'(O) = 2
Compare the results to the approximations obtained in Examples I and 2. EXERCISES 5.5
I So/~t;on We show the results obtained with a typical computer algebra system
In FIgure 5.26. We see that as t increases, the approximate solution obtained in Ex-
ample I becomes less accurate. However, for small values of t, the results are nearly 1. Use the linear approximation of the model of the sim- 12. If L = I m, how many ticks does the clock make in
identical. ple pendulum to determine the motion of a pendulum 1 min if it ticks once for each time the pendulum
with rod length L = 2 ft subject to the following sets makes a complete swing?
We show the results obtained for (b) with a typical computer algebra system
in Figure 5.27. In this case, we see that the error diminishes as t increases. (Why?) of initial conditions: 13. Assuming that a clock ticks once each time the pen-
(a) e(O) = 0.05, e'(O) = 0 dulum makes a complete swing, how long (in meters)
does the pendulum need to be for the clock to tick
(b) e(o) = 0.05, e'(o) =1 once per second?
(e) e(O) = 0.05, e'(O) = -I 14. In Exercise 13, how long (in feet) does the pendulum
In each case, determine the maximum displacement need to be for the clock to tick once per second?
(in absolute value).
*15. For the undamped problem
Exact solution to 0.8 0.006 2. Consider the situation indicated in Exercise I. How-
the linear problem ever, use the initial velocity e' (0) = 2. How does the d2? + Ke = 0
0.005 maximum displacement (in absolute value) differ from dl- L
0.6 {
0.004 that in Exercise I(b)-(e)? e(O) = eo, e' (0) = va,
21T I 0.4 0.003 *3. Suppose that the pendulum in Exercise I is subjected what is the maximum value of e(t)? For what values
to a resistive force with damping coefficient b = 40. of t does the maximum occur?
-1 0.002
0.2 Solve the initial-value problems given in Exercise I, 16. For what values of I is the pendulum vertical in Exer-
Numerical solution to 0.001
-2 the nonlinear problem and compare the resulting motion to the undamped cise 15?
case.
27T t 1T1l2 1T16 I 17. Solve the initial-value problem
4. Verify that e(l) = C I cos wi + C2 sin wi, where w =
(a) (b) (e)
Viii satisfies the equation d 2eldl + (gIL)e = o.
2
L d2e + b de + oe = 0
dt2 dt
5. Show that e(1) = eo cos wi + ~ sin wi, where w =
0
Figure 5.26 . (a) The approximations are quite similar for small values of I. (b) The absolute {
value of the dIfference between the two approximations. (c) The absolute value of the differ· Viii is the solution of the initial-value problem e(o) = eo, e' (0) = va·
ence between the two approximations. .
Determine restrictions on the parameters L, b, and g
d2e + s.. e = 0 that correspond to overdamping, critical damping, and
dl 2 L
{ underdamping. Describe the physical situation in each
e(o) = eo, e' (0) = va· case.
e 6. Let e(l) = A cos (WI - ",). Use cos (a + b) = 18. Solve the initial-value problem
cos a cos b - sin a sin b with the solution in Exercise
0.12 d2e de
0.8
5 to find A so that e(l) satisfies the initial·value prob- L dl 2 + b dt + ge = F cos yt
0.1 lem in Exercise 5. {
0.6 e(0) = 0, e' (0) = 0,
0.08 *7. Show that the phase angle in e(1) = A cos (wi - "') is
0.4
'" = cos- (eoN ~ + v'oIw2).
I assuming that b2 - 4gL < O. Describe the motion of
0.06
0.2 the pendulum as 1--> "'.
0.04
In Exercises 8-11, approximate the period of the motion of the *19. Consider Van-der-Po!'s equation,
ro/2 1TI 0.02
-0.2 /' pendulum using the given length.
d 2x ., dx
Exact'solution to dt 2 + e(x- - 1) dt +x = 0,
the linear problem 7T'/2 1T I S.L=lm
(a) (b) 9. L = 2 m where e is a small positive number. Notice that this
10. L = 2 ft equation has a nonconstant damping coefficient. Be-
Figure 5.27 (a) The approximations are quite similar for nearly all values of I. (b) The ab-
solute value of the difference between the two approximations. cause e is small, we can approximate a solution of
*11. L = 8 ft
280 Chapter 5 Applications of Higher Order Differential Equations Chapter 5 Summary 281
Van-der-Pol's equation with x(t) = A cos wi, a solu- CHAPTER 5 SUMMARY

2 2
tion of d x/dl + x = 0 (the equation obtained when ~-:::::--'::--~"""'.:~:;-;:;;;::~::2.(.2...-::::::;-';:::-'5~ozs:.::;I~:C::::~;;:X:;:;--'-~-;:'':::-~~~;;::::'-:::~'..7:z.;-.:=.:::.c-",~--::::::=-:::::-.:::'::,:'::':;':~:-::;':-'::=:--:=D:-::::~::~::S7":-~.;,·;:;:::::::£':':"""",,: -..:..~
s = 0). This method of approximation is called har-
monic balance. i Concepts & formulas
subject to the initial conditions:
(a) Substitute x(t) = A cos wi into the nonlinear term
(a) 11(0) = l, 0'(0) = 0 5.3)
~tion 5.1)
in Van-der-Pol's equation to obtain
) Section
(b) 11(0) = -l, =0
,dx
s(X--I)-=-sAw-A
dl 4
- I smwt (I,) . (c) 0(0) = 0, 0'(0) = 1
0'(0)
Jlooke's Law Forced motion

(d) 0(0) = 0, 0'(0) = - i F~ ks
3
_.lA
4 sin 3wt. (e) 11(0) = 1, 0'(0) = 1
(f) 11(0) = I, 0'(0) = -I Simple harmonic motion
(b) If we ignore the term involving the higher har- The initial-value problem
(g) 0(0) = -1, 0'(0) = 1
monic sin 3wl, we have S(X2 + I) dx/dl =
-sAw(!A' - 1) sin wi = sOA' - I) dx/dl. Sub-
(h) 0(0) = -I, 8'(0) = -1 d'x + kx = 0 ISection 5.4)
(i) 0(0) = I, 0'(0) = 2
7
stitute this expression into Van-der-Pol's {
X(O; = ", x'(O) = (3 Kirchhoff's law
equation to obtain the linear equation (j) 0(0) = 1, 0'(0) = 3
The sum of the voltage drops across the circuit elements is
d'x/dl + sU A' - I) dx/dl + x = o. bas the solution
2
cI<) 0(0) = -1, 0'(0) = 2 equivalent to the voltage E(I) impressed on the circuit
(c) If A = 2 in the linear equation in (b), is the ap- (3 .
proximate solution periodic?
(I) 0(0) = -1, 8'(0) = 3. x(t) = " cos wi + -;;;- sm wi, L-R-C Circuit
23. Plot the solutions obtained in Exercise 22 individually
(d) If A '" 2 in the linear equation in (b), is the ap-
and then plot them simultaneously. Give a physical in-
where w = vw;;.; the amplitude of the solution is + REQ + J.. Q = E(I)
L1:.Jf-
proximate solution periodic? dldl C
terpretation of the results. A = V,,, + (3'/w'. {
20. Comment On the behavior of solutions obtained in Q(O) = Qo, 1(0) = Q'(O) = 10 ,
24. If the computer algebra system you are using has a
Exercise 19 if (a) A < 2 and (b) A > 2.
built in function that approximates the solution of non- Deflection of a Beam
21. Use a computer algebra system to solve the initial- linear differential equations, solve the problem d 4s
value problem I Section 5.2) EI dx 4 = w(x).
d'O + 1 dO + sin 0 = 0
dl' 2 dr Damped motion
dOO + 0= 0 {
dl' 11(0) = 00 , 0' (0) = va,
{
0(0) = 00 , 0' (0) = va with the initial conditions stated in Exercise 22. Com-
) Section 5.5)
pare the results you obtain with each.
subject to the following initial conditions: Motion of a Pendulum
*25. Use a built-in computer algebra system function to ap· The initial-value problem
(a) 0(0) = 0, 0'(0) =2 proximate the solution of the pendulum problem with Overdamped
(b) 0(0) = 2, 0'(0) = 0 a variable damping coefficient C'-- 4mk> O. d20 + E..O = 0
dl' L
(c) 0(0) = -2, 0'(0) = 0 {
dZO +1(0' _ l) dO + 0= 0 Critically damped 0(0) = 00 , 0' (0) = Va
(d) 0(0) = 0, 0'(0) = -I dl 2 2 dl c' - 4mk = O.
{ has solution
(e) 0(0) = 0, 0'(0) = -2 11(0) = 00 , 0'(0) = Vo
(f) 11(0) = I, 0'(0) = -I using the initial conditions stated in Exercise 22. Com- Underdamped 0(1) = 00 cos wi + ~ sin wi.
C'- - 4mk< O. w
(g) 11(0) = -I, 0'(0) = 1. pare these results with those of Exercise 24.
Plot each solution indiVidually and plot the seven so- 26. Repeat Exercise 25 using
lutions simultaneously. Explain the physical interpre-
d'O I dO.
tation of these solutions. 7+Z(o'-l)di+ smO =0
22. Solve the initial-value problem {
0(0) = 00 , 0' (0) = Va.
position? When does the pendulum first pass through 24. Consider a buoy in the shape of a cylinder of radius
CHAPTER 5 REVIEW EXERCISES r, height h, and density P where P :s 0.5 g/cm (the
3
the vertical position?
density of water is 1 g/cm 3 ). Initially, the buoy sits
18. If the initial conditions in Exercise 17 are 0(0) = 0 with its base on the surface of the water. It is then reo
and 0' (0) = -I, what is the maximum displacement
leased so that it is acted on by two forces: the force
of the pendulum from the vertical position? When does
1. An object weighing 32 Ib stretches a spring 6 in. If *7. If the object in Exercise 5 is subjected to the forcin of gravity (in the downward direction) equal to the
the pendulum 1mt return to the vertical position?
the object is lowered 4 in. below the equilibrium and function f(t) = 4 cos t, detemUne the displacemen; weight of the buoy, F\ = mg = pm- 2 hg; and the force
released from rest, determine the displacement of the Describe the physical phenomenon that occurs. Find *19. How does the motion of the pendulum in Exercise 17 of buoyancy (in the upward direction) equal to the
differ ifit undergoes the damping force FR = 8 dOldl? 2
weight of the displaced water, F2 = m- xg, where
objec~ assuming there is no damping. What is the the enveloping functions.
maximum displacement of the object from equilib- 20. Solve the model in Exercise 17 with the damping force x = x(t) is the depth of the base of the cylinder from
8. An object of mass 2 slugs is attached to a spring with
rium? When does the object first pass through the FR = SY3 dO/dt. How does the motion differ from the surface of the water at time t. Using Newton's
spring constant k = 5 lb/ft. If the resistive force is
equilibrium position? How often does the object re- that in Exercise 19? second law of motion with m = pm- 2 h, we have the
FR = 6 dxldt and the external force is f(t): 2
turn to the equilibrium position? differential equation pm- 2 hx" = -m- xg or phx" +
12 cos 2t, determine the displacement if x(O): 21. Undamped torsional vibrations (rotations back and
gx = O. Find the rlisplacement of the buoy if x(O) = I
2. If the object in Exercise I is released from a point x' (0) = O. What is the steady-state solution? What is forth) of a wheel attached to a thin elastic rod or wire
and x'(O) = O. What is the period of the solution?
3 in. above equilibrium with a downward initial ve- the transient solution? satisfY the differential equation loY' + kO = 0, where
ois the angle measured from the state of equilibrium, What is the amplitude of the solution?
locity of 1 fils, determine the displacement of the ob- 9. Find the charge and current in the L-R-C circuit if
ject, assuming there is no damping. What is the max- lo ·is the polar moment of inertia of the wheel about 25. A cube-shaped buoy of side length C and mass den-
L = 4 H, R = SO n, C = 1/436 farad, and E(t) = 100
imum displacement of the object from equilibrium? its center, and k is the torsional stiffness of the rod. sity p per unit volume is floating in a liquid of mass
·ifQ(O) = Q'(O) = O. Find limHoo Q(t) andlim,_,.,l(I).
When does the object first pass through the equilib- Solve this equation if kilo = 13.69 S-2, the initial density Po per unit volume, where Po > p. If the buoy
10. Find the charge and current in the L-R-C circuit in angle is 15° = 0.2168 rad, and the initial angular is slightly submerged into the liquid and released, it
rium position? How often does the object return to the
Exercise 9 if E(I) = 100 sin 2t. Find limH~ Q(I) and velocity is 10° s-' = 0.1745 rad' s-'. oscillates up and down. If there is no damping and no
equilibrium position?
limH~ J(I). How do these limits compare to those in air resistance, the buoy is acted on by two forces: the
*3. An object of mass 5 kg is attached to the end of a 22. Determine the displacement of the spring-mass
Exercise 9? force of gravity (in the downward direction), which is
spring with spring constant k = 65 N/m. If the object system with mass 0.250 kg, spring constant k =
*11. Find the charge and current in the L-R-C circuit if equal to the weight of the buoy, F, = mg = pC3 g ; and
2.25 kg/s 2 , and driving force f(t) = cos t - 4 sin I if
is released from the equilibrium position with an
L = I H,R = on, C= 10-4farad,andE(I) = 220 if there is no damping, zero initial position, and zero ini- the force of buoyancy (in the upward direction), which
upward initial velocity of 1 mis, determine the dis- is equal to the weight of the displaced water, F2 =
. Q(O) = Q'(O) = O. Find limHoo Q(t) and lim,...,.. J(I). tial velocity. For what frequency of the driving force
placement of the object assuming the force due to
12. Find the charge and current in the L-R-C circuit in would there be resonance?
e2xgpo, where x = x(t) is the depth of the base of
damping is FR = 20 dx/dt. Find lim,_oo x(t) and the the buoy from the surface of the water at time t. Then
quasiperiod. What is the maximum displacement of Exetcise II if E(I) = 100 sin lOt. Find lim t _ oo Q(I) 23. The differential equation by Newton's second law with m = pC3, we have
the object from equilibrium? When does the object and limHoo J(I).
I - t2 , 0 :s I :s 1 the differential equation pC2X" + gpox = O. Deter-
first pass through the equilibrium position? 13. Determine the shape of the beam of length 10 with yeO) = y'(O) = 0 mine the amplitude of the motion if Po = I g/cm ,
3
y" +Y = { 0, t > I '
2
4. If the object in Exercise 3 is released from a point constants E and J such that EJ = I, weight distribu· p = 0.25 g/cm 3 , e = 100 cm, g = 9S0 cmls , x(O) =
tion w(x) = x(1O - x), and fixed-end boundary con· can be thought of as an undamped system in which a 25 cm and x' (0) = O.
1 m below equilibrium with zero initial velocity, de-
ditions at x = 0 and x = 10. force F acts during the interval of time 0 :s I :s 1.
termine the displacement. Find lim Hoo x(t) and the 26. A rabbit starts at the origin and runs with speed a due
This is the situation that occurs in a gun barrel when
quasiperiod. What is the maximum displacement of 14. Determime the shape of the beam in Exercise 13 if north toward a hole in a fence located at the point
a shell is fired. The barrel is braked with heavy springs
the object from equilibrium? When does the object there are fixed-end boundary conditions at x = 0 and (0, d) on the y-axis. At the same time, a dog starts at
(see Figure 5.28). Solve this initial-value problem.
first pass through the equilibrium position? a sliding clamped end at x = 10. the point (c, 0) on the x-axis, running at speed b in
*15. Determine the shape of the beam in Exercise 13 if pursuit of the rabbit. (Note: The dog runs directly to-
5. An object of mass 4 slugs is attached to a spring with
there are fixed·end boundary conditions at x = 0 and ward the rabbit.) The slope of the tangent line to the
spring constant k = 16 lb/ft. If there is no damping
a free end at x = 10. dog's path is dyldx = -(at - y)lx, which can be writ-
and the object is subjected to the forcing function
ten as xy' = y - at. Differentiate both sides of this
f(t) = 4, determine the displacement function x(t) if 16. Determine the shape of the beam in Exercise 13 if equation with respect to t to obtain xy" = -a dtldx. If
x(O) = x'(O) = O. What is the maximum displacement there are fixed-end boundary conditions at x = 0 and s is the length of the arc from (c, 0) along the dog's
of the object from equilibrium? When does the object simple support at x = 10. path, then dsldt = b is the dog's speed. Also, dsldx =
1mt pass through the equilibrium position?
17. Use the linear approximation of the model of the sim· _~, where the negative sign indicates that s
6. If the object in Exercise 5 is subjected to the forcing pie pendulum to determine the motion of a pendulum increases as x decreases. By the chain rule,
functionf(t) = 4 cos 2t, determine the displacement. with rod length L = ~ ft subject to the initial condi·
Find limH~ x(t) if it exists. Describe the physical phe- tions 0(0) = 1 and dOldl(O) = O. What is the maxi· !!!.._!!!..ds=_..!.~
dx-dsdx b ),
nomenon that occurs. mum displacement of the pendulum from the vertical Figure 5.28
-
284 Chapter 5 Applications of Higher Order Differential Equations Differential Equations at Work 285
so substitution into the equationxy" = -a dt/dx yields 28. (Self-excited vibrations) The differential equation (a) Show that the roots of the characteristic equatioos
the equation xy" = k~, where k = a/b. Make of x" + [(e - P)/m] x' + (kim) x = 0 are
the substitution
xoi that describes the motion of a spring-mass system
~±J(~)2
with a single degree of freedom excited by the force
I
Px'is _.5....
p =y' or y" =dp
- I 2m 2m m
dx ~-+-I mx" + ex' + kx = Px', (b) Show that if P > e the motion of the system di-
I I I
verges (dynamically unstable), if P = e the so-
to find the path of the dog with the initial condition I I
which can be rewritten as lution is the solution for a free undamped system,
y'(e) = O. If a < b, when does the dog catch the
rabbit? Does the problem make sense if a = b? (See and if P < e the solution is the solution for a free
B x" + e - P x' + .5... x = O. damped system.
Figure 5.29.) m m
(0, d) Y
4J:]
F=/lW
w
!
(Motion to right)
I)jXf Differential Equations at Work: I
Figure 5.30
damping.
A system with coulomb (dry-friction) , ,
A. Rack-and-Gear Systems
I
(c,O) x Consider the rack-and-gear system shown in Figure 5.31. Let T represent the kinetic
I
Figure 5.29 and
energy of the system and U the change in potential energy of the system from its po-
tential energy in the static-equilibrium position. The kinetic energy of a system is a
27. (Coulomb damping) Damping that results from dry Motion to left x" + w~x =.:;.. function of the velocities of the system masses. The potential energy of a system con-
sists of the strain energy Ue stored in elastic elements and the energy U/:' which is a
friction (as when an object slides over a dry surface)
is called Coulomb damping or dry-friction damp- where F represents the friction force, m the mass, and function of the vertical distances between system masses.
ing. On the other hand, viscous damping is damping w;= klm.* The rack-and-gear system cO!lsists of two identical gears of pitch radius r and
that can be represented by a teon proportional to the (a) Find a general solution of x" + w~x = F/m.
centroidal mass moment of inertia I, a rack of weight W, and a linear spring of stiff-
velocity (as when an object such as a vibrating spring ness k, length I, and a mass of"Y per unit length. To determine the differential equation
I
vibrates in air or when an object slides over a lubri· (b) Solve the initial-value problem to model the motion, we differentiate the law of conservation, T + U = constant, to
cated surface).
obtain
Suppose that the kinetic coefficient of friction x"+W2x=£
is /1-. Friction forces oppose motion. Therefore, the
force as a result of friction F is shown opposite the
{ x(O) = :0, x'~)
= o.
d
dt (T+ U) = o.
direction of motion in Figure 5.30. However, because
F is a discontinuous function, we cannot use a single
differential equation to model the motion as was done
(c) Find the values of t for which the solution in (b)
is valid.
We then use Tmax = Urn"" to determine the natural circular frequency of the system,
where Tmax represents the maximum kinetic energy and Urnax the maximum potential
I
with previous damping problems. Instead, we have one (d) At the right-endpoint of the interval obtained in energy.
equation for motion to the right and one equation for (c), what is the displacement of the object? Because the static displacement Xs of any point on the spring is proportional to
motion to the left: its distance y from the spring support, we can write
(e) What must the force F be to guarantee that the
Motion to right object does not move under the given initial con- yx
ditions? xs=]·
'" M. L. James, G. M. Smith, 1 C. Wolford, and P. W. Whaley. Vibration of Mechanical and Structural Systems with Microcomputer Applications,
Harper & Row, New York (1989), pp. 70-72. • Robert K. Vierck, Vibration Analysis. Second Edition, HarperCollins, New York (1979). pp. 137-139.
T = if1- + 1.. W (riJl + 1'.

~ 2 g 2
II (_yr_iJ)2
0 1
dy
~~-~' ~
Gears
Rack Spring
kt;.,
or
k(x + t;.,)
where m3 = yl is the total mass of the spring.

The change in the strain energy Ue for a positive downward displacement x of
~----7----!--x the rack from the static-equilibrium position where the spring is already displaced t:..s
from its free length [see Figure 5.30(b)] is the area of the shaded region [see Figure
5.30(c)] given by
w
(a) (b) (e) Ue = '2I kx 2 + Wx.
Figure 5.31 (a) A rack-and-gear system:c(b) Static-equilibrium position. (c) Spring-force di.
agram. Similarly, the change in potential energy Ug as the rack moves a distance x below the
static-equilibrium position is
Ug = -Wx
When the rack is displaced a distance x below the equilibrium position, the gears have
the angular displacements 0, which are related to the displacement of the rack by and the total change in the potential energy is
O=~.
r
Then the angular velocity of each gear is. given (through differentiation of 0 with reo
spect to t) by or
6= ~r'
(Notice that we use a "dot" to indicate differentiation with respect to t. This is a com. Substitution of these expressions into the equation T + U = constant yields
mon practice in physics and engineering.) Because x = rO, we have
- I W 2 m3r2J;,'2 I 2
yrO [I + '2 -gr + 2(3) tJ + '2k(r/J) = constant
Xs = -1-'
Differentiating with respect to t then gives us the differential equation
so the velocity at any point on the spring is
2
. yr6 8+[ 2i + Wkrr2 + m3r2]0=0.*
xs = -1-'
g 3
The kinetic energy of a differential element of length dy of the spring is
dT= t/'Cxs)2 dy.

• M. L. James, O. M. Smith. 1 C. Wolford, and P. W. Whaley, Vibration of Mechanical and Structural
The total kinetic energy of the system in terms of 0 is Systems with Microcomputer Applications, Harper & Row, New York (1989), pp. 82-86.
I. Determine the natural frequency Wn of the system. d2x

2. Determine O(t) if 0(0) = 0 and 11(0) = Oown •
-2 + O.3x + 0.04x 3 = 0
dt
(
3. Find Tmax ~nd Urn"". Determine the natural frequency with these two quantif x(O) = "', X' (0) = f3
Compare this value of Wn with that obtained above. les.
for various values of '" and f3 in the initial conditions.
4. How does the natural frequency change as r increases?
2. How does increasing the initial amplitude affect the period of the motion?
B. Soft Springs D. Aging Springs

In the ~ase of a soft spring, the spring force weakens with compression or extension. In the case of an aging spring, the spring constant weakens with time. For springs of
For spnngs of thiS type, we model the physical system with the nonlinear equation this type, we model the physical system with
d2X dx
m-
d2X dx
2 + c- + kx - jx = f(t)
3 m-
2 + c - + k(t)x = f(t)
dt dt dt dt
(
(
. x(9) = "', x' (0) = f3, x(O) = "', x' (0) = f3.
where k(t) -> 0 as I -> co.
where j is a positive constant.
1. Approximate the solution to
I. Approximate the solution to
d2~ + 4e-tl4x = 0
d2X dx dt
dt2 + 0.2 dt + lOx - 0.2x3 = -9.8 (
x(O) = "', x' (0) = f3
(
x(O) = "', x'(O) = f3 for various values of '" and f3 in the initial conditions.
for various values of '" and f3 in the initial conditions. 2. What happens to the period of the oscillations as time increases? Is the motion os-
2. Finbd codndditions on '" and f3 so· that limHoo x(t) (a) converges and (b) becomes cillatory for large values of I? Why?
un oun e .
3. If .limr-.oo x(t) exists, does the motion of the spring converge to its equilibrium po-
sllion? Explam. E. Bode Plots
Consider the differential equation d 2xldt 2 + 2c dxldt + ex = Fo sin wI, where c
and k are positive constants such that c < k. Therefore, the system is underdamped.
C. Hard Springs To find a particular solution, we can consider the complex exponential form of
the forcing function, Foe''"', which has imaginary part Fo sin wI. Assuming a solu-
~ the case o~ a hard ~pring, the spring force strengthens with compression or exten- tion of the form zit) = Ae i ,"', substitution into the differential equation yields
sIOn. For spnngs of thiS type, we model the physical system with A[ -w2 + 2icw + e] = Foe''"'. Because e - w2 + 2icw = 0 only when k = w and
c = 0, we find that A = F o/(k 2 - w2 + 2icw) or A = Fo/(e - w2 + 2icw) .
(e - w2 - 2icw)/(k2 - w2 - 2icw) = (e - w2 - 2icw)/[(k 2 - w i + 4C W ] Fo =
2 2 2
d2X dx
m -d
2
t
+ c -dt + kx + jx 3 = f(t) H(iw)Fo. Therefore, a particular solution is zp(t) = H(iw)Foe''"'. Now we can write H(iw)
( in polar form as H(iw) = M(w)e i 4>{,"l, where M(w) = IrV(e - ( 2 )2 + 4c2 w2 and
x(O) = "', x'(O) = f3,
cf>(w) = cot- 1((w2 -k2 )/2cw) (-7r~cf>~0). A particular solution can then be
where j is a positive constant. written as zit) = M(w)Foe i'"'e,4>{,") = M(w)Foe,(w'+4>{w) with imaginary part
M(w)Fo sin(wl + cf>(w)), so we take the particular solution to be xp(t) =
1. Approximate the solution to M(w)Fo sin(wl + cf>(w)). Comparing the forcing function to xp, we see that the two
6
functions. have the same form but different amplitudes and phase shifts Th .
~e ampl~tude of~e particula:: solution (or steady state), M(w)Fo, to th~t ofe~:tio of
mg. function, F?, IS M(w) and IS called the gain. Also, xp is shifted in time b I fore.
radians to the nght, so </>(w) is called the phase shift Wh
h h· . . en we grap
h th ~ </>(w)I;",
e gam and tb
p ase s 1ft agamst w (using a loglO on the w axis) we obtain the Bode plots E . e
refer to the value of 20 [oglo M(w) as the gain in decibels. . ngIneers
~:~ + 2~ + 4x = sin 2t
I x(O) = 2' x' (0) = I.
2. Graph ~e solution simultaneously with the forcing functionf(t) = sin 2t.
3. Approximate M(2) and </>(2) using the graph in Number 2.
Systems of
Differential Equations
4. Graph the corresponding Bode plots.
5. Compare the values of M(2) and <p(2) with those obtained in N urnb er 2 .
~ Introduction
Suppose that we consider the path of an object in the xy-plane. For example, if we
launch a model rocket or kick a ball, we would describe the object's position in the
plane with the vector-valued function ret) = (x(t), yCt» = x(t)i + y(t)j. In other words,
the coordinates of the object at time t = T are given at the terminal point of the vec-
tor reT) = (x(T),y(T». Following a method similar to that discussed in Chapter 3 con-
cerning falling bodies, we can assume that the only force acting on the object is caused
by gravitational acceleration, g = (0, -g) = -gj. (Note that this force acts entirely in
the vertical direction as expected.) Then if we let the velocity of the object be repre-
sented by the function vet) = (vit), vit», where vit) is the horizontal component of
the velocity vector and vy(t) is the vertical component, and the acceleration function
with aCt) = v'Ct) = (v~(t), v~(t», then we can make use of Newton's Second Law, F =
ma, to model the situation. In this case, we have mg = ma so that the acceleration must
satisfy aCt) = g or (v~Ct), v~(t» = (0, -g). Equating the corresponding components of
these two vectors, we obtain the first-order system of differential equations
{
v~(t) = 0
v~(t) = -g.
291
292 Chapter 6 Systems of Differential Equations 6.1 Introduction 293
The equations in this system are uncoupled in that Vx does not appear in the equa.. dx
l
tion involving v~ and Vy does not appear in that describing v~. Therefore, we can solve dt = ax - bxy, x(O) = Xo
vo/1
this system by finding a solution of each individual first-order equation. If we assume
~VOSina dy
that the object is launched with an initial speed va making an angle a with the hOri. dt = -CY + dxy, yeO) = Yo
zontal (Figure 6.1), the initial velocity vector is
Vo cos a where the initial populations are x(O) = Xo and yeO) = Yo, and a, b, c, and d are posi-
v(O) = (vxCO), vy(O» = (va cos a, va sin a).
tive constants. Notice that when we have a system of two first-order equations, we must
Figure 6.1 Sketch to deter· This means that we detennine the components of vct) = (vit), vy(t» by solving the specify an initial condition for each of the unknowns, xct) andy(t). The tenns xy indi-
mine components of v(O). IVP cate an interaction between the two populations. For example, if the two popUlations
v~(t) = 0, viOl = va cos a
{
are foxes, denoted by y, and rabbits, denoted by x, then xy denotes a rabbit being eaten
v~(t) = -g, Vy(O) = va sin a· by a fox. This interaction hurts the rabbit population, as indicated by the negative co-
efficient of xy, - b, in the dx/dt equation, while it helps the fox popUlation because the
Through integration, we find that vit) = C I and vy(t) = - gt + C2, so that application coefficient d of the xy tenn is positive in the dy/ dt equation.
of the initial conditions viOl = Cl and viOl = C2 indicates that C l = va cos a and We may also encounter systems that involve second-order differential equations.
C2 = va sin a. Therefore, the solution is For example, we could consider a spring-mass system in which an object of mass ml
vet) = (va c?s a, - gt + va sin a) is attached to the end of a spring with spring constant kl that is mounted on a support
(as we did in Chapter 5). Then, if we attach a second spring with spring constant k2
with components vit) = va cos a and vyct) = -gt + Vo sin a. We detennine the posi. and mass m2 to the end of the first spring, we can detennine the displacement of the
tion function ret) using the relationship r' (t) = (x' (t), y' (t» = v(t). Equating the cor. first and second springs, xct) and yct), respectively, by solving the system
responding components as we did earlier, we obtain the system of first-order equations
d2~
l
Jx'(t) = Vo cos a ml = -klX + ko(y -
x)
dt -.
ty'(t) = -gt + va sin a·
d 2y
m2 = -k2 (y - x)
Figure 6.2 Sketch to deter- If we also assume ihat the object is laUnched from the position reO) = (0, h) (Figure df2
mine reO). 6.2), then we consider the IVP
Later, we discuss how this system is modeled using Hooke's Law and Newton's Sec-
Jx' (t) = va cos a, x(O) = 0 ond Law. In the case of a system of second-order equations, we must have two initial
ty'(t) = -gt + va sin a, yeO) = h" conditions for each unknown to define an initial-value problem. For example, we could
have x(O) = 0, x'(O) = -I, yeO) = I, and y'(O) = o.
As in the previous system, we can solve the individual differential equations. Integra.
Although we can solve and approximate solutions of many different types of sys-
tion yields xct) = (va cos a)t + C3 and yct) = _~gt2 + (va sin a)t + C4 . Then, appli.
cation of the initial conditions x(O) = C3 and yeO) = C4 indicates that C3 = and
C4 = h, so
° tems of differential equations, we will focus much of our attention on systems of first-
order linear differential equations. These systems can be written as
-± x; + adt)x2 + ... + alnCt)Xn + fl(t)
l
2 = all(t)xl
rCt) = «va cos a)t, gt + (va sin a)t + h)
y ~2 = a21 Ct)XI + a22(t)x2 + ... + a2nCt)xn + hU).
15 with components x(t) = (va cos a)t and yet) = _~gt2 + (va sin a)t + h.
12.5 For example, if va = 64 ft/s, a = 7T/6, g = 32 ft/s 2 , and h = 0, then x~ = ani (t)XI + a n2(t)x2 + ... + ann(t)xn + !n(t)
2
10 rCt) = (32tY'3, -16t + 32t). We graph this function (parametrically) in Figure 6.3 in·
In this system, there are n unknowns Xl (t), X2(t), ... , xn(t) and n differential equations.
7.5 dicating the orientation of the curve. The object moves in the direction of the orienta·
tion as t increases. Ifjj(t) == O,j = 1,2, ... ,n, then we say that the system is homogeneous. Otherwise,
2.5 the system is nonhomogeneous. The system
Of course, we may encounter systems of differential equations that are more dif·
+--2-+0-4-+0-6+0-S+0-100+-''---:X ficult to solve than the previous system. For example, we can consider a population X; = X2
problem involving the interaction between two populations, predator and prey. In this { Xl = -Xl
Figure 6.3 Path of projectile case, if we let x(t) represent the size of the prey population and yet) that of the preda·
with orientation. tor population, we can then model the situation with the IVP is an example of a first-order linear homogeneous system, while
294 Chapter 6 Systems of Differential Equations "'~
6.1 Introduction 295
Xl = X2 which is equivalent to the system. A general solution of this equation is x(t) =

{Xl = -XI + sin t CI cos 2t + C2 sin 2t. We can use y = !(x' - x) to detennine yet). This gives us
is a first-order linear nonhomogeneous system because h(t) = sin t. A system such as yet) = 1.[( -2cI sin 2t + 2C2 cos 2t) - (CI cos 2t + C2 sin 2t)]
S
{x~ :
X2 -
Xl -
-X2
XI -
+ XIX2
XtXz
= t( -C2 - 2cI)sin 2t + t( 2C2 - c,)cos 2t.
is nonlinear because of the nonlinear terms xi and Xl X2' Of course, variable names Therefore, we have solved the system of first-order equations by solving the equiva-
other than XI (I), xit), ... , xn(l) can be used. For example, the previous system can be lent second-order equation. Now, if we had considered the initial-value problem
written in terms of xCt) and yet) with
Figure 6.4 Graphs of
X' = X - X2 - xy x(l) = l sin 21 and X' = x + Sy, x(O) = 0 ,
{ y' = -y +xy .
y(l) = cos 21 - ~ sin 2t. { y' = -x - y, yeO) =I
then we would apply the initial conditions to a general solution of the system. In our
We mentioned in Chapter I that there is a relationship between a single higher order
case, we have x(O) = C], so CI = O. Then, yeO) = !C2C2), so C2 = ~, and the solution to
equation and a system of first-order equations. We revisit that topic now by consider. y
ing the second-order linear ODE with constant coefficients x" - 4x' + 13x = 2 sin t the IVP is
We can write this equation as a systeci of first-order equations by making the substi:
tution {x(t) = t sin 2t, yet) = cos 2t - +sin 2r}
X' = y
We can graph these two functions of t as done in Figure 6.4. The graph of x(t)
x
so that differentiation with respect to t yields is the solid curve, and that ofy(t) is the dashed curve. (We can recognize the functions
x" = yr. by referring to the initial conditions.) We can also plot them parametrically in the xy-
plane as done in Figure 6.S. Note: We should include the orientation of this curve. The
Solving x" - 4x' + 13x = 2 sin t for x!' gives us x" = 4x' - 13x + 2 sin t. Substitu. value of t = 0 corresponds to the point (0, I) while t = 7T/4 corresponds to (SI2, -1/2).
tion of x" = y' and x' = y into this equation then gives us y' = 4y - 13x + 2 sin t. Therefore, the orientation is clockwise because we move from (0, I) to (SI2, -1/2) as
Therefore, the second-order equation x" ~ 4x' + 13x = 2 sin t is equivalent to the sys. t increases.
tern of first-order linear differential equations Figure 6.S Parametric plot
of (x(t) = l sin 2t, Nonlinear differential equations can often be written as a system of first-order
X' = y yet) = cos 2t - ~ sin 2t}. equations as well.
{ y' = 4y - 13x + 2 sin {
We can also begin with a system of first-order equations and transform it into a
higher order equation. For example, consider the system
X' = X + Sy
o "f'';I".,The Van-der-Pol equation is the nonlinear ordinary differential equation
{ y' = -x - y' w" - iL(1 - w 2 )w' + w = O.
If we differentiate x' = X + Sy with respect to t, we obtain Write the Van-der-Pol equation as a system. Notice that this equation is similar to
those encountered when studying spring-mass systems, except that the coefficient
x" = x' + 5y'. of w' that corresponds to damping is not constant. Therefore, we refer to this situ-
Substituting y' = -x - y into this equation then gives us ation as variable damping and can observe its effect on the system.
x" = x' + S(-x - y) = x' - Sx - Sy. :.1 Solution Let Xl = wand X2 = W' = Xl. Then, Xl = w" = iL(1 - w )w' - W =
2
Solving x' = X + Sy for y (or in this case, Sy, yields Sy = X' - X so that substitution -3 , iL(l - Xi)X2 - x], so the Van-der-Pol equation is equivalent to the nonlinear system
into x" = x' - Sx - Sy gives us the second-order equation x" = x' - Sx - (x' - x) = Xl = X2
-4x, or Figure 6.6 Graph of x"
{ = iL(l - Xi)X2 - XI' •• . .•
x" + 4x = 0, x,(t) and X2(t). In Figure 6.6, we show graphs of the solutIOn III the Illitlal-value problem
296 Chapter 6 Systems of Differential Equations 6.1 Introduction 297
[:l : ~~ -
X' = y X' = Y Therefore, a general solution of the system is
+1 12. .
xT)X2 - Xl 11. { y' = -x { y' = -x + sm 21 X(t) = CJ sin I - c, cos I
{ y(l) = CI cos I + c, sin (
XI(O) = I, X2(0) = I
In Exercises 13-18, write each equation as an equivalent sys-
on the ~terval [0, 25]. Because we let X2 = XI, notice that X2(t) > 0 when XI(t) is tem of first-order equations. In Exercises 21-27, sol ve the system using the operator method.
mcreasmg and X2(t) < 0 when Xl (t) is decreasing. Note that these functions appe
to become periodic as t increases. ar 13. x" - 3x' + 4x = a X' = -2x - 2y +4 X' = x + 2y + cos t
21. { y' = -5x + y 22. { y' = 3x + 2y
The observation that these solutions become periodic is further confirmed b 14. x" + 6x' + 9x = a
a graph of Xl (t) (the horizontal axis) versus xzCt) (the vertical axis) shown in Figur~ *15. x" + 16x = I sin I X' = - 3x + 2y - e' x" = 2y - 5x
6.7, called the phase plane. We see that as t increases, the solution approaches 16. xl! +x= e!
23. { y' = _ 3y +1 24. { y" = 2x - 2y
certain fixed path, called a limit cycle. We will find that nonlinear equations ara 17. y'" + 3y" + 6y' + 3y = x * {X" + y" + x + Y = a
more easily studied when they are written as a system of equations. e 18. y(4) + y" = a 25. y" + 2x' - 2x - Y = a
*19. Rayleigh's equation is 3x + 2y
X' = - + 2z
26. y' = -2x + 3y + 2z
x" + IL[ t (x')' - I] x' + x = 0,
{
z' =x + y
X' = x + 2y - 2z + cos I
where IL is a constant. 27. y' = x - y + 2z
{
(a) Write Rayleigh's equation as a system. z' = -x-y
Xl
(b) Show that differentiating Rayleigh's equation and
setting x' = z reduces Rayleigh's equation to the In Exercises 28-30, solve the initial-value problem. For each
Figure 6.7 The graph of {Xl (I)) indicates that the
X2(t
Van-der-Pol equation. ·equatIons
probIem, graph X(I), yet), and the parametnc · {~0
y(t)'
solution approaches an isolated periodic solution, 20. (Operator Notation) Recall the differential operator
which is called a limil cycle. D = dldl, which was introduced in Exercises 4.4. 28. x' - x - 2y = 0, y' - 2x - Y = 0, x(O) = 0,
Using this notation, we can write the system yeO) = -I
y" + 2y' - x' + 4x = cos I
Graph the solution to the initial-value pr~blem {w" + (w - I)w' + w = 0 on the
2
{x" + x + y" - y' - 2y = et *29. x' - 3x + 2y = 0, y' - 2x +y = 10, x(O) = 0,
as
. w(O) = 1, w'(O) = 1 yeO) = a
interval [0, 25]. Cd + 2D)y - (D - 4)x = cos I 30. x" = y - x, y" = y - X + sin I, x(O) = 0, x'(O) = 0,
{ (d + I)x + (d - D - 2)y = e r yeO) = 0, y'(O) = a
Operator notation can be used to solve a system. For (Degenerate Systems) Suppose that the system of linear
. .. {LJX + L,y = frCI)
example, if we consider the system {DDX = y Or dIfferential equartons has the form L3 X + L.y = 12(1)' where
EXERCISES 6.1 y =-x
DX - y = a I' . f th . bl L" L20 L3, and L. are linear differential operators. For example,
{Dy + x = 0' we can e Iffimate one 0 e varIa es in the system
by applying the oper.ator ( - D) to the second equation
Dx-y=O
(D' + 2D)y - CD - 4)x = cos I
Solve each of the following uncoupled systems of equations. to obtain { _ Dx _ dy = O· When these two equations { (D' + I)x + (d - D - 2)y = e"
6. {x~ = -Xl + I, XI(O) = a
1.
X'
,
= I X' = X x, x"
= x,(O) = 1 are added, we have the second-order ODE - D'y - Y LI = -(D - 4), L, = (D' + 2D), L3 = (d + I), andL. =
{y 2.
= cos I = a or dy + Y = O. This equation has the character-
{ y' = 1
*3 {X'y' == a-2y X' = x' Solve each system by solving the equivalent second-order istic equation r' + I = a with roots r = ±i. There- (d - D - 2). If
£, L.
ILl L'I
= 0, we say the system is degener-
. 4. { y' = e' equation. fore, y(l) = CJ cos I + c, sin I. At this point, we can ate, which means that the system has infinitely many solutions
repeat the elimination procedure to solve for x, or we
Solve each of the following initial-value problems.
7. {X:y == 4x-3x- +y 6y 8 {x' = 8x - y
. y'=x+6y
can use the equation Dy + x = a or x = - Dy to find or no solutions. If I~~ ~:I = 11,i:~ ~:I = I~~ 1,i:~1 = 0,
x. Applying (-D) to y yields
then the system has infinilely many solutions. If at least one
5. {x~ = -3x" x,(O) = -I
x, = I, x,(O) = 1
*
9.
{X' =
y' =x
-x - 2y
+y
10
.
{x'y' == 4x-x++2y2y x(l) = -DY(I) = -D(CI cos 1+ c, sin I) of the other two determinants is not zero, the system has no
solutions.
= Cl sin t - C2 cos t.
298 Chapter 6 Systems of Differential Equations 6.2 Review of Matrix Algebra and Calculus 299
In Exercises 31 and 32, show that the system has no solution.

( b) {X" = -x + 2y + cos 31 ~ Review of Matrix Algebra and Calculus
y" = -2x + y
31 {x' + x+ y' = e'
. x' + x + y' = 2e' 40. Consider the initial-value problem
(' Basic Operations (" Determinants and Inverses
(" Eigenvalues and Eigenvectors (" Matrix Calculus
x" + y' = cos I x' -2x - Y
l
=
32. { 3x" + 3y' = sin I
When we encounter a system of linear first-order differential equations such as
Y ' =lx
4 .
In Exercises 33 and 34, show that the system has infinitely many x(O) = 1, yeO) = 0 X' = -x + 2y
solutions. { y' = 4x - 3y'
The solution obtained with one computer algebra sys-
+ y" =
x" 12 X" + y'/ - y' = tel tem was
we often prefer to write the system in terms of matrices. Because of their importance
33. {4x" + 4y" = 4/2 34. { 2x" + 2y" - 2y' = 21e' X(I) = G- i)e(-1-112), + G+ i)e(-1+I12), in the study of systems of linear equations, we now briefly review matrices and the ba-
{ yet) = ~ie(-I-iI2)t _ ~ie(-l+jl2)f , sic operations associated with them. Detailed discussions of the defInitions and prop-
In Exercises 35 and 36, determine the value of k so that the sys-
erties discussed here are found in most introductory linear algebra texts.
tem has infinitely many solutions. while the solution obtained with another was
X' + x - y' =I { y" - x = cos I X(I) = e -'(cosGI) - 2 sin(tt)). Basic Operations

*35. { 4y' - 4x' - 4x = kt 36. 4y" + Icc = 4 cos I { y(l) = ~e -, sinGI)
Show that these solutions are equivalent

In Exercises 37 and 38, determine restrictions on c so that the
system has no solutions. Often, we can use a computer algebra system Definition 6.1 n x m Matrix
to generate numerical solutions to a system of equa-
X + x' - cy = e-' An n X m matrix is an array of the form
tions, which is particularly useful if the system under
*37. { 2x' - 8y + 2x = 2e-'
consideration is nonlinear. (Numerical techniques for
x" + 4y' = cos I systems are discussed in more detail in Section 6.8.)
38. {2x" + 8y' = c cos I *41. Figure 6.8 shows the direction f'leld associated with
39. Suppose that we have a system of second-order equa- the nonlinear system of equations
tions of the form
dx=2x_xy
x" = F 1(1, x,y, x',y') dl with n rows and m columns. This matrix can be denoted A = (aij).
{ y" = F2 (1, x, y, x', y')'
[ El.=-3y+xy
dl
If we let Xl =x,x2 =x' =Xi,YI =y. andyz =y' =
y;, then we can transform the original system to a sys- for 0 S x S 6 and 0 S Y S 6. Use the direction field
tem of f'lrst-order linear equations. For example, with
these substitutions, we transform the system
to sketch the graphs of the solutions {;n that satisfy matrix
the initial conditions (a) x(O) = 2 and yeO) = 3; and
x" = -x + 2y
(b) x(O) = 3 andy(O) = 2. (e) How are the solutions
{ y" = 3x - y + 5 sin I Vn ) a row vector.
alike? How are they different?
into the f'lrst-order system
[ :~ ~Xl + 61' Definition 6.2 Transpose

: 2Yl
Yl = Y2 The transpose of the n X m matrix
Y2 = 3Xl - Yl + 5 sin I 3-
Transform each of the following systems of second- 1·,'

order equations into a system of f'lrst-order equations. ! ; I---
... 5 6x
X"=-3X+ y
(a) { y" = -x - 2y Figure 6.8
is the m X n matrix bu ...

...
AT = a~2
('"
a21
a22 '., )
an2
B = b~1
('"
bil
b 22
b,2
"")
b 2m . . .
: IS a; X m matnx,
b,m
aim a2m a nm au b 12 b lm
AB = a~1
Cani
azz
a n2
'")(''
a2i b21
ani bil
b 22
b,2
b 2m
bim
is the unique n X m matrix
Definition 6.3 Scalar Multiplication, Matrix Addition
Cl2
Let A = (aij) be an n X m matrix and c a scalar. The scalar multiple of A by c
is the n X m matrix given by cA = (caij).
If B = (bij) is also an n X m matrjx, then the sum of matrices A and B is the
C = C~I
('"
Cnl
C22
Cn2
' ")
G2m
Cnm
n X m matrix A + B = (aij) + (bij) = (aij + b;). i where
j
Cll = allb ll + a12b21 + ... + a1jbjl = L alJPklo

k~1
Hence, cA is the matrix obtained by multiplying each element of A by c; A + B is ob.
i
tained by adding corresponding elements of the matrices A and B that have the same
dimensions.
GI2 = allb 12 + al2b22 + ... + ali b,2 = L
k~1
aIJPk2,
and
"""p,'.' GUY = aulb lv + a u2b2v + ... + auibiv = L

i
k=1
aukbkv.
Compute 3A - 9B if A = (-~ ~ _~~) and B = (; In other words, the element Cuv is obtained by multiplying each member of the
uth row of A by the corresponding entry in the vth column of B and adding the
result.
: ) Solution Because 3A = (~~ 1~ _~~) and
-18
-9B = ( -63
36
-36
-72)
-18' 3A - 9B = 3A + (-9B) =
(-21
-45
48 -48·
-30
-78)
Definition 6.4 Matrix Multiplication

"i'"I".,
Compute AB If A =. (0 4 5) andB=-i
-5 -15 (-3 -:).
-4
: 1 Solution Because A is a 2 X 3 matrix and B is a 3 X 2 matrix, AB is the 2 X 2

matrix:
AB = (0
-5
4
-1
5)(=; _:)
5 1-4 det(A) =
all
a21
al2
a22
al n
a2n
=:t (_l)i+ia'i det(AiJ) =:t (-I),+iaij"IA,J
:
i=l i=l
0.-3+4.-5+5.1 o. 4 +
4 . -4 + 5 . -4 ) ani an2 ann
= ( -5·-3+-1·-5+5·1 -5·4+ -1 . -4 + 5·-4
The number (-Iy+i aij det(Aij) = (-l)i+iaijIAijl is called the cofactor of aij. The
= (-15 -36) cofactor matrix, AC, of A is the matrix obtained by replacing each element of
25 -36· A by its cofactor. Hence,
IAIlI
-IA2'!
-lAd
IA221
(-I)":'IA.")
(-I) IA2nl
Definition 6.5 Identity Matrix A' -(( -l)n+IIAnll (-I)"IAn2 1 IAnni
o 0
ThO" X" =trcr (1 1 0
o 0 'if'"I,'.'
by I or In. -2
-1)
Calculate IAI if A =
Ci -4
1 -2 -3 .
If A is an n X m matrix (an n X n is called a square matrix), then IA = ISo/ution

AI=A.
IAI = I-~ =; =il

Determinants and Inverses : 4 5
= (_1)2(_4)1- -31 + (-li(-2)15 -31 + (-1)\-1)1 -41
-2 5 -2 5 1
Definition 6.6 Determinant = -4((-4)(-2) - (-3)(1)) + 2((5)(-2) - (-3)(5)) - ((5)(1) - (-4)(5))
If A = (all), the determinant of A, denoted by det(A) or IAI, is det(A) = all; = -59.
if A = (all a J2 ), then
a21 a22
Definition 6.7 Adjoint and Inverse

~------------------------------~
B is an inverse of the n X n matrix A means that AB = BA = I. The adjoint,
Aa, of an n X n matrix A is the transpose of the cofactor matrix: A a = (Acl. If
. A=
More generally, If (:::
: ::: ::: :::). : .
IS an n X n matrix and.
Ay IS the
1
IAI 1= 0 and B = IAfAa , then AB = BA = I. Therefore, if IAI 1= 0, the inverse of
A is given by
anI a n2 ... ann
(n - 1) X (n - 1) matrix obtained by deleting the ith row andjth column from A-l=~Aa
IAI .
~~ J
'if 1,,','.' For convenience, we state the following theorem. The proof is left as an exercise.
Find A-I if A = ( 2
-3
-1) 1.
Theorem 6.1 Inverse of a 2 X 2 Matrix
:J Solution In this case, IAI = 1-; - iI = 2 - 3 = -1 *- 0, so A-I exists. More- Let A = (~ ~). IflAI = ad - be *- 0, then
C
over, A = G ;), so A = G ;) and A-I = I ! ( = (=~ =;).
a
A-I 1 (d -ab).
= ad - be -e
'-
"f'"i,I.,
FindA-lifA=
(
-2
;
-1
1
1
°1).
-1 ..,
,'ii,,','.,
Find A -I if A = (; -1)
3 .
: ) Solution We begin by finding IAI, which is given by
~I =
2 2 : 1 Solution Because IAI = (5)(3) - (2)(-1) = 17,
°1 + (_1)3(_1)1 °1 + (_1)4(1)1 .
-1 3 -1
(-2)(-1)
3
+ (1)(-2) + (1)(-1) =-1. A -I - - I ( 3 51) --
- 17 -2
(f,
-?,
,~
We then calculate the cofactors: We almost always take advan-
tage of a computer algebra
. 123 system to perform operations The inverse A -I can be used to solve the linear system of equations Ax = b. For
AlJ = (-1) 2111 °1 = -1 ' AI2 = (-I?
-1°1 = 2 '
(-~;) in which A = (-~;), we
-I on matrices. In addition, if
you have taken linear algebra, example, to solve (; -;)(;) = (; -;) and b =
A13=(-I)412
3 ~I =-1 , you can use techniques such
as row reduction to find the find x = A -Ib = (_~ ~)( -~;) = (-~). We can find several uses for the inverse
inverse of a matrix or solve
A 21 =(-I)3 1-11 _~I = 0,
2
A22 = (-1)4 1- 3 11 =-1 : systems of equations. in solving systems of differential equations as well.
-1 '
2
A 23 =(-1)5 1- 3 -11
1 =-1 Eigenvalues and Eigenvectors
~I =
2
A 31 =(-1)4 1-11 -1, A32 = (_1)5 1- 2 11 = 2 Definition 6.8 Eigenvalues and Eigenvectors
° '
2 -11 = O. A nonzero vector x is an eigenvector of the square matrix A if there is a num-
A33 = (-1)6 1- 2
1 ber A, called an eigenvalue of A, so that
Then Ax = Ax.
Note: By definition, an eigenvector of a matrix is never the zero vector.

l~
----------------------------------------------------------
"f''''''.
Show that (- ~) and G)
are eigenvectors of
(
-I 2) .
4 _ 3 WIth corresponding . .. . . { 9X I - 6YI = J
Slmphfymg Yields the system of equatIons 3XI _ 2YI = 0' so YI - ,Xl'
Th fi ° _
ere ore,
eigenvalues - 5 and I, respectively.
When finding an eigenvector
if Xl is any real number, then (/1) is an eigenvector. In particular, if XI = 2,
,. corresponding to the eigen· 2X I
i1Solution value A, we see that there is
actually a collection (or fam· then (;) is an eigenvector of A = (~ =~) with corresponding eigenvalue A = - 5.
ily) of eigenvectors correspon·
dina to A. In the study of dif· Similarly, if we let (;~) denote the eigenvectors corresponding to A = 2, then
- (~ n}(;~)
fer:ntial equations, we find
that we only need to find one
member of the collection of
{(~ =~) 2 = 0, which yields the system =
{~~ ~~ :~, so
If x is an eigenvector of A with corresponding eigenvalue A, then Ax = Ax. Be.

cause this equation is equivalent to the equation (A - AI)x = 0, x "*
0 is an eigenvec.
eiaenvectorS. Therefore, as we
did in Example 8, we usually Y2 = jX2' Ifx2 = 3, then (i) is an eigenvector of A = (; =~) with corresponding
tor if and only if det(A - AI) = 0.
eliminate the arbitrary con·
stants when we encounter eigenvalue A = 2. (Notice that constant (scalar) multiples of the eigenvectors G)
lfdet(A - AI) "* 0, what is the solution of (A - AI)x = O? Can this solution them in eigenvectors by select·
ing particular values for the and (D are also eigenvectors corresponding to A = -5 and A = 2, respectively.)
(vector) be an eigenvector of A?
constants.
Definition 6.9 Characteristic Polynomial 'ifi ,,',• •'

The equation det(A - AI) = °
is' called the characteristic equation of A;
det(A - AI) is called the char~cteristic polynomial of A.
Find the eigenvalues and corresponding eigenvectors of A = ( _ ~ ~).
1 Solution .. poIynorrua
In this case, the charactenstIc '1'IS I-A
-I II = A2+1 , so
_A
Notice that the roots of the characteristic polynomial of A are the eigenvalues of A.' the eigenvalues are the roots of the equation A2 + I = 0. These are the imaginary
numbers A = i and A = - i where i = v=l. The corresponding eigenvectors are
found by substituting the eigenvalues into the equation (A - AI)x = 0 and solving
.
for x. For A = i, thIS equatIon IS -I .. (-i I)(XI) (0) h' h. .
-i YI = 0' W IC IS eqUlvaIent t0 th e
Calculate the eigenvalues and corresponding eigenvectors of A = (; -6) {-ix l + :1 0. Notice that the second equation of this system is a constant
= °
-7' system =
-XI - rYI
i .J Solution The characteristic polynomial of A = (; -6) multiple (i) of the first equation. Hence, an eigenvector (;:) must satisfy ix YI = l·
-7 is
Therefore, (~:) = C)XI is an eigenvector for any value of XI' For example, if
-7-6 I
_ A = A2 + 3A - 10 = (A + 5)(A - 2). Xl = 1, then C) is an eigenvector. For A = -i, the system of equations is
Because the eigenvalues are found by solving (A + 5)(A - 2) = 0, the eigenvalues ~)(X2) = (0), which is + Y2 :: 0°. Because the second
are A = -5 and A = 2. Let (;:) denote the eigenvectors corresponding to the
( -1
i
r Y2 ° equivalent to {_ ix2
X2+ rY2-
eigenvalue A = -5. Then,

equation equals i times the first equation, the eigenvector (;~) must satisfy
-ix2· Hence, (_ ;:) = (_~ )X2 is an eigenvector for any value of

Y2 = Xl.
(-~::) = (-})ZI is an eigenvector for any value of z,.
Therefore, if X2 = 1, then ( _~) is an eigenvector.
If A = 2, we have C~ ! !)~~) (n = which is equivalent to
Recall that the complex conjugate
if = a - bi. Similarly, the

(i I I)~;) = m in reduced form. Thus, X2 = 0 and Y2 = -Z2, so
a, - b,z ( -;~) = (- !)Z2 is an eigenvector for any value of Z2'

- a2 - N'
olice th at the eigenvectors
b 2i . corresponding to the complex
(-~ -~ ~)(;:) = (~),
vector x = :.'
an - bnl If A = 5, then which is equivalent to

o
conjugate eigenvalues A = i and A ~ -i in the previous example are C) and (_~),
2 -I Z3 0
which are complex conjugates. This is no coincidence. We can prove that the eigen-
vectors that correspond to complex eigenvalues are themselves complex conjugates.
(i ~ -I)(~:) = m· Therefore, X3 =0 and Z3 = 2Y3, so (~J = mY3 is an
eigenvector for any value of Y3'
Calculate the eigenvalues and corresponding eigenvectors of the matrix Definition 6.10 Eigenvalue of Multiplicity m
A = (~ ~ r). Suppose that (A - AI)m where m is a positive integer is a factor of the charac-
teristic polynomial of the n X n matrix A, while (A - A,)m+1 is not a factor of
this polynomial. Then A = A, is an eigenvalue of multiplicity m. J
] Solution We begin by finding the characteristic polynomial of A with
o
3- A
2
~ I
4-A
3
= (1 - A) 1 - A
2 4-A
I 1 = (I - A)(A2 - 7A + 10)
We often say that the eigenvalue of an n X n matrix A is repeated if it is of
multiplicity m where m 2: 2 and m s; n. When trying to find the eigenvector(s)
corresponding to an eigenvalue of mUltiplicity m, two situations may be encountered:
= (1 - A)(A - 2)(A - 5).
either m or fewer than m linearly independent eigenvectors can be found that corre-
The eigenvalues of A are A = I, A = 2, and A = 5. For each eigenvalue, we frod spond to A.
the corresponding eigenvectors by substituting the eigenvalue A into the equation
(A - AI)x = 0 and solving for the vector x. If A = 1, we obtain
Example 11
(~ ~ ~)~:J m = Find the eigenvalues and corresponding eigenvectors of (a) A = (i =~ !); and
which is reduced to (I ~ -r)(~:) (~}= so XI = z, and y, = -kz. Hence, (b) B = (r -~ n

------ . - . . _. ---
J Solution (a) The eigenvalues are found with

Hence, A = 5 is an ei enVa1(U;I)Of multiPliC(i~ 2~~ ~)s(~~)se, w(~e)n we find a cor-
g
-3
-5 - A ~ I =(I_A)I-~-A 4-,1.
3 1_ (_3)1 63 4-,1.
3 I+ responding eigenvector v = ~; ,we solve b -; ~ ~; = ~,WhiCh is equiv-
-6 4 - A 6
3
3
6
-5 - AI
-6 =16+12,1.-,1.3==0.
alent to (~ ! ~)~:) = m. The components of v must satis~ XI = -2z 1 and
1
Note that A = -2 is a root of the characteristic polynomial. Using this root with Yl = O. We find only one eigenvector V =
(-2z1)
~I =
(-2)
~ ZI that corresponds to the
synthetic or long division, we find that ,1.3 - 12,1. - 16 = (A + 2)\,1. - 4) == O.
Hence, ,1.= -2 is an eigenvalue of mUltiplicity 2. When we try to find an eigen. eigenvalue A = 5 of mUltiplicity 2. We leave it to you to find an eigenvector corre-
~:) corresponding to A
sponding to ,1.=0.
vector v = = -2, we see that
Matrix Calculus
Definition 6.11 Derivative and Integral of a Matrix

all (t) adt)
must be satisfied. After the necessary row operations, we find that this system is
equivalent to The derivative of the n X m matrix A(t) = az)(t) azz(t)
(
anl(t) anz(t)
a",(t) is differentiable for all values of i and j, is
which indicates that XI - YI + ZI ='0 or XI = YI - ZI. If ZI = 0, then XI = YJ. d

dt adt)
Hence, VI = ~~) = (l)Yt is an eigenvector for any choice ofYI. On the other hand,
d
d
dt a22(t)
dt A(t) =
ifinxI = YI - Zt.YI = 0, then XI = -ZI' Therefore, Vz = ~ = (-I)

-ZI) ~ ZI (forany
(
choice of Zl) is another linearly independent eigenvector corresponding to A = -2
because V2 is not a constant multiple of VI' Therefore, we have found two linearly The integral of A(t), where aij(t) is integrable for all values of i and j, is
independent eigenvectors that correspond to the eigenvalue A = - 2 of multiplicity
2. We leave it to you to find an eigenvector corresponding to A = 4.
(b) The eigenvalues ofB are determined by solving
Jal1(t) dt Jadt) dt J alm(t) dt
J Ja2I(t) dt Ja22(t) dt J a2m(t) dt .
=: ~ 1=(5-A)I-~ ~ 1_(_4)1
1
A(t) dt =
2 5-,1. - 5 A 0
II Janl(t) dt Jadt) dt J anm(t) dt

,'f''''''.') -2
14.A= _~
(, 2
-3
-2
-5
2
0
-5
-3
-2
')
4 and
-3
*33. A = C41
4~
-38
44
6
18)
-21
-3
: 'I Solution We find
.E... ACt) =
~ A(t) by differentiating each element of ACt). This yields
(-3 sin 3t 3 cos 3t -e

-')
sf -3
5
-2
-5
3
-3
-2
2
-3
~l)
-5
34. A = (( 0
-I
I
1)
In Exercises 35-40, find dldl A(t) and JA(I) dt.
dt I sin( 2 ) + 2t2 cos(t2 ) e
t .
In Exercises 15 -19, find the determinant of the square matrix.
J A(t) dt by integrating each element of AU).

21
e 51) 36. A = (Sin 31)
-~) A=(_~ -!)
Similarly, we find 35. A = ( e-
15. A = (-I5 16. cos 31
J ACt) dt =
"31 sm
1 2
("2 t
. 3
t + Cll
+ C21
-1
-3- cos 3t
-1
2 cos(t )
2
+ C12
+ C22 e
-e
I
_,)
+ cn
+ C23
-!
*17. A = (
-2
-2 ~) 18. A = ( ~
3
-~)
-3 2 -3
I
*37. A = (cos 1 1 cos I)
sm t (sm t
38. A = (e-' te-')
1 12
0 -2
~:) In I)
'-
where each cU" represents an arbitrary constant.
"W(=! -2
-I
0
0
-3
0
-3
I
*39. A = cos 31
(
sm 3t
e4t )
40. A = ( tl~ t
In 1
In Exercises 20-23, find the inverse of each square matrix. *41. If A = (: ~) and IAI = ad - be *- 0, then show
EXERCISES 6.2
20. A = (~ -D *21. A = (~ -D that A-I = -1d d
a - c -c
b ( -b).a
-~)
3
22. A= C3~ -2
In Exercises 1-4, perform the indicated calculation if Many computer software packages and calculators contain
(-~ -~) and
-3 -4
A= (~ -!), B = (-~ ~), and C = (_~ ~). 9. A = -I
-I
-2
0 -4 *23. A = CI~
-I
2 -;) built·in functions for working with matrices. Use a computer
or calculator to perform the following calculations to become
familiar with these functions.
(~;
~;)
1. B - A 2. 2A +C 2 0 -2
_~)
+ C) -5 0 0 0-2
3. (B - 4A 4. (A - 3C) + (C - 5B) B=
0 4 3 -I 7
In Exercises 24-34, find the eigenvalues and corresponding 42. If A = -6 0 5 and
4 -4 eigenvectors of the matrix. (
In Exercises 5-8, perform the indicated calculation if
A= (~ -I
3
-I -3
B= ~),
5 -3)
4 -5
(~
; , and
-2
10. A = (-2
-3
4 3
2 4
3 -2
_3)andB= -2
(, -2
-4
0
1)
24. A = (-~ =~) *25. A = (=~
B= (-~
-6
=!
0
-~ -~),
-2
compute (a) 3A - 2B;

-I -4 26. A=(~ ~) 27. A = ( 0
-3
~).
-4 5 6 3 6
C= C25 -I
-I -3 -3
C
*11 . A = 3 ;) andB = (_: -I)
-5 28.A=(=; _~) *29. A = (=~ (b) BT; (e) AB; (d) IAI, IBI, and IABI; (e) A-I.
5. 7A + 3B
7. 2A - 4(B + C)
6. 8B - 9A
12. A= (-4I _;) andB = (~ ~) 30. A = (=~ _~) 31. A = ( 5
-I *43. Find the eigenvalues and corresponding eigenvectors
8. 4C + 2(2A - 5B) of the matrices
-I-2)3 C~ r -~) (-~5 =:) ~ - ~ ~).

-I -3
In Exercises 9-14, compute AB and BA, when defined, using
13. A = (-: 5 -I)
-2 andB= C3
-I 4 32. A =
A= ;0 and B = (
the given matrices. -4 -I -I 5 -I -5 -2 3
314 Chapter 6 Systems of Differential Equations 6.3 Preliminary Definitions and Notation 315
-1t A(t) and f A(t) dt if eigenvectors of an n X n matrix for n = 2, 3, and 4 Xl(t) = all(t)xl(t) + adt)x2(t) + ... + aln(t)Xn(t) + fi(t)
l
44. Find
(a) Use a computer algebra system to compute th' X2(t) = a21 (t)XI (t) + adl)x2(t) + .. , + a 2n (t)xn(t) + f2(t)
te- l t 2 sin 2t
9
I
+ 4t 2 :~::~v::::: :e=el(·g~rl~eI
:: ·. ;
and; co:r)po:mn x~(t) = anl(t)xI(t) + a n2Ct)xzCt) + ... + ann(t)xn(t) + !n(t)
A(t) = 6
cos t 3
sec 2t -33 -2 I nte
t(t- I) is equivalent to X' (t) = A(t)X(t) + F(t).
o I 2 0
4 3 down a portion of the result (or print it and staple it
sins t cos t t sin t 2
VI-t 2 to your homework). (b) Compute approximations of
"i,,,','N.
*45. Calculate the eigenvalues of
0
(I
-1-k2)
2k
(~ -1 -
2
. How do the eigenvalues change for
0) and
:e= e(rrn)SP::~i:~~:e::::::s:: Ca) Write the homogeneous system,
X' = -5x + 5y . . C
{ y = - 5x + y m matnx ,orm.
(b) W't th
nee
-00 <k< +oo? o I 2 0 X' = X + 2y - sin t . .

more meaningful to you? nonhomogeneous system { y' = 4x _ 3y + 12 m matnx form.
46. Nearly every computer algebra system can compute
exact values of the eigenvalues and corresponding
X' = - 5x + 5y . .
] Solution (a) The homogeneous system,
{ y=-x+y
5 IS eqUivalent to the sys-
tem(»=(=; ;)(~).
X' = X + 2y - sin t . .
Cb) The nonhomogeneous system { y' = 4x _ 3y + t2 IS eqUivalent to
~ Preliminary Definitions and Notation (~:) = (~ _;)(~) + (-:~n t).

We first encounter systems of equations in elementary algebra courses. For example,
3x - 5y = -13 At this point, given a system of ordinary differential equations, our goal is to
{ -3x+6y=15 construct an explicit, numerical, or graphical solution of the system.
We now state the theorems and tenninology used in establishing the fimdamen-
is a system of two linear equations in two variables with solution (x, y) = (-1,2). In tals of solving systems of differential equations. All proofs are omitted but can be found
the same manner, we can consider a system of differential equations. in advanced differential equations textbooks. In each case, we assume that the matrix
We begin our study of systems of ordinary differential equations by introducing A(t) in the systems X' (t) = A(t)X(t) + F(t) and X' (t) = A(t)X(t) is an n X n matrix.
several definitions along with some convenient notation. Let
all(t) adt) ... aln(t») Definition 6.12 Solution Vector

A(t) = a2~(t) a2~(t) ... a 2:(t) ,
and A solution vector of the system X' (I) = A(I)X(t) + F(t) on the interval I is an
( n X 1 matrix of the form
an I (t) anzCt) ann(t)
(:~~:~),
Then, the homogeneous system of first-order linear differential equations
+ adt)x2(t) + '" + aln(t)Xn(t) X(I) =
l
Xl(t) = all(t)xl(t)
X2(t) = a21(t)xl(t) + adt)x2(t) + '" + a2n(t)Xn(t) xn(t)
where the x/(t) are differentiable fimctions on I, that satisfies X'(t) =

x~(t) = anl(t)Xl(t) + an2(t)x2(t) + '" + ann(t)xn(t)
A(t)X(t) + F(I).
is equivalent to X' (t) = A(t)X(t), and the nonhomogeneous system
316
Chapter 6 Systems of Differential Equations 6.3 Preliminary Definitions and Notation 317
,'i',,',I., 1
(2
St 2t
2t
2t
2t
6 c,e + 2c2eSt = 2(-2c,e - C2eSt) + 6(c,e + 2c2eSt)
2t
St
-2)(-2C,e2t - c2e ) (( -2c,e - C2eSt ) - 2(c,e + 2C2e ))
Show that X(I)

2t
= ( -2ee 2t ) IS. a solutIOn
. of X' = (12 -2)6 X. 2t
-4c,e - 5c2e )
St
= ( 2c,e 2t + IOc2 eSt .
Therefore, the linear combination of the solutions is also a solution .

.1 Solution NoticethatX'(I) = (-4e~:) and
2e
(12 6 e
2t
-2)(-2e 2t) = (-2e 2t2t -
-4e
2e 2t)
+ 6e2t " .
..
We define linear dependence and independence of a set of vector-valued func-
(-~;~:). Then, because X' = AX, X(I) = (-2;::) is a solution of the system. tions (X, (I), Xit), ... , XnCt)} in the same way we defined linear dependence and in-
'- dependence of sets of real-valued functions. Then, the set (X,(t), X 2(t), ... , Xn(t)} is
linearly dependent on an interval I if there is a set of constants {c" C2, ... , cn} not
all zero such that
c,X,(t) + C2X2(t) + ... + cnXnCt) = 0;
Theorem 6.2 Principle of Superposition
otherwise, the set is linearly independent. (Note that 0 is the zero vector with the same
~uppose that XI (I), X 2(1), ... , Xn(t) are n solutions of the system of first-order dimensions as each of the Xj{I), j = 1, 2, ... , n.) As with two real-valued functions,
Imear homogeneous differential equations X' (t) = A(I)X(I) on the open interval two vector-valued functions are linearly dependent if they are scalar multiples of each
I. Then, the linear combination other. Otherwise, the functions are linearly independent. For more than two vector-
valued functions, we often use the Wronskian to determine if the functions are linearly
X(I) = cIXI(t) + C2X 2 (1) + ... + cnXn(I), independent or dependent.
where c], C2, . .. , Cn are arbitrary constants, is also a solution ofX'(t) = A(I)X(I)
on I.
Definition 6.13 Wronskian of a Set of Vector-Valued Functions
The Wronskian of (X,(t), X2(t), ... , Xn(t)} is defmed as the determinant of the
matrix with columns XI (t) ("(0) (Xd'l) c«))

= X2) (t) , X 2(1) = X2t(l) , ... , Xn(l) = X2~(t) :
,'ii,,',"w' Xn, (t) xn2(1) xnn(t)
X'I XI2
... X'n
_eSt) (-2e2t)
Show that X(t) = ( 2eSt and X(t) = CI e 2t + C2(_eSt)
2e St are solutions of
X21 X22 ... X2n
[
(1 -2)6
I
W(X" X 2 , ••. , Xn) =
X' = 2 X.
Xn , Xn2 ... Xnn
I
: X(I) = (
. ear combmatlOn
~;St) satisfies X' =
. . of the two solutIOns
.
2t
a -~)x.
: ) Solution syve let the reader follow the procedure used in Example 2 to show that
(-2ee ) and (_eSt)

By the Principle of Superposition, the lin.
Theorem 6.3 The Wronskian of Solutions
Suppose that XI(t), XiI), ... , Xn(l) are n solutions of the system of first-
order linear homogeneous differential equations X' (I) = A(t)X(t) on the open
2t 2e St is also a solution. We
interval I, where each component of A(t) is continuous on I. If the set
. .
venfY thIS now by first writing X(t) as X(t) =
21
(-2c,e+2t - C2e St) .
Cle
2t
2C2e51
Then X'(t) ~
,
(XI(t), X2(t), ... , Xn(t)} is linearly dependent, then W(X" X 2 , . . . , Xn) =; 0
on I. If the set {XI(t), X2(t), . . . , Xn(t)} is linearly independent, then
-4c,e - 5c?e St W(Xh X 2 , . . . , Xn) "" 0 for aU values on 1.
( 2c, e 2t + IOc;eSt) and
6.3 Preliminary Definitions and Notation 319
318 Chapter 6 Systems of Differential Equations
both -2 sin 2t. + 2 cos 2t) and ( . 42 cos 2 t2) are solutIOns
. to the system. Com-
"fi,,"'.' ( sm2t
puting the Wronskian, we have
sm t-cos t
. (-2e2t) (_eSt)..
Venfy that XI(t) = e 2t and Xit) = 2e St are Imearly mdependent solutions -2 sin 2t + 2 cos 2t 4 cos 2t \
of X' = (2I -2)6 X. \ sin 2t sin 2t - cos 2t

= (-2 sin 2t + 2 cos 2t)(sin 2t - cos 2t) - (4 cos 2t)(sin 2t)
'I Solution We showed earlier that XI(t) and X 2(t) are solutions of the given sys- = -2 cos 2 2
2t - 2 sin 2t = -2.
_2e 2t _eSt\ 7t
tern. Therefore, we calculate W(Xh X 2 ) = e2t 2e St = _3e . The vector_ Thus the set -2 sin 2t. + 2 cos 2t) , ( . 42 cos 2t 2 ) }IS. linearly
. .
mdependent
\ , sm 2t sm t - cos t
valued functions are linearly independent because W(Xh X 2 ) = - 3e 7t 0 for all * {(
and is consequently a fundamental set of solutions.
values of t.
L
Show that any linear combination oj (

-2 sin 2t + 2 cos 2t)
sin 2t
(4 cos 2t )
and sin 2t - cos 2t
is also a solution oj the system.
Definition 6.14 Fundamental Set of Solutions
Any set {XI(t), X 2(t), . .. , Xit)} ofn linearly independent solutions ofX'(t) =
A(t)X(t) on an open interval I is called a fundamental set of solutions on I. Theorem 6.4 General Solution of a Homogeneous System
Suppose that XI(t), Xit), ... , Xit) are n solutions of the system of first-order
linear homogeneous differential equations X' (t) = A(t)X(t) on the open interval
I, where each component of A(t) is continuous on I. Then every solution of X' (t)
"i·,,·,',.. = A(t)X(t) is a linear combination of these solutions. Therefore, a general so-

lution of X' (t) = A(t)X(t) on lis
Which of the following is a fulldamental set of solutions for X(t) = CIXI(t) + C2X2(t) + ... + cnXn(t).
X:(t)) = (-2 -8)(X(t))? (a) {(cos 2t) (sin 21)}
"f·,,',I.,
( y (I) I 2 Y (t) sm 21 ' cos 2t
(b) {(-2 sin 2: + 2 cos 2t), ( . 4 cos 21 )}
sm 21 sm 21 - cos 21
Solution We first remark that the equation (~: ~~~) = (- i -8)(X(t)). .

2 y (I) IS eqwva·
Find a general solution of X' = G -2)6 X.
lent to the system
X' = -2x - 8y.
, ] Solution In previous examples, we have shown that XI(t) = (-2:~:) and
(a) Differentiating we see that

{ y' =x + 2y
X 2 (t) = (~:::) are linearly independent solutions of X' = G -~)X. Because
A(t) is a 2 X 2 matrix, we need two linearly independent solutions to form a gen-
C~S 21)'
( sm 2t = (-2 sin 21) * (-2 cos 21 - 8. sin 21) eral solution. Therefore,
2 cos 21 cos 21 + 2 sm 21
· h sows
w h IC h us that (cos 2t) IS
sin 21 . not a solution
. to the system. Therefore,
X(t) = CI( -2:::) + C2( ~:::) = (~~:~~: ;c::~:t)
is a general solution.
{ (~~: ~:), (~: ~I)} is not a fundamental set of solutions. (b) You should verify that L
320 Chapter & Systems of Differential Equations 321
&.3 Preliminary Definitions and Notation
~~~~~~~==========~~
Definition 6_15 Fundamental Matrix
Graphs of several solutions are shown in Figure 6.9. (a) Identify the gral!h of the
solution that satisfies the initial conditions x(O) = 0 and yeO) = 1. (b) Fznd Cl and C2
Suppose that Xl (t), X2(t), ... , X.(t) are n solutions of the system of first-order e -2t)
5t
(- 3e ) .
so that X(t) = Cl ( 2e -2t + C2 e5t satIsfies X(O) - I·
_ (0)
linear homogeneous differential equations X' (t) = A(t)X(t) on the open interval
I, where each component of ACt) is continuous on I. The matrix
(t) = (Xl X2 ••• X.)
is called a fundamental matrix of the system X'(t) = A(t)X(t) on I. Thus, a gen-
om! ,,1""00 = bo writt= ~ X(Q • ~,)C. wh= C • (l:)

,if·,,1,".' Figure 6_9
Show that (t) = (2: =~: - 3e5t)

e5t is a fundamental matrix for the system
X'(t) = (_; =:)X(t). Use the matrix to find a general solution of
X'(t) = (_~ =~)X(t). EXERCISES 6_3
: 1 Solution Because
-2t)'
( 2:- 2t =
(2
=4:--2t)2t (4
= -2 =13)(2:--2t)2t and
in Exercises 1-8, write the system of first-order equations in
matrix form. 12_ (
COS 2t ) ( 1 - 2 sin t )
-2 sin 2t' -4 sin I cos t
2
t
fx' = x + 4y fx' = 2x - 3y t
-e3ee (e-ee=:
St 5t 5 1
-3e )' = (-15e ) = ( 4 -3)(-3e \ -V=2x-y 2. V =x + Y *13. eu)
e~r, (
t)
,
)
( e 5t 5e 5t -2 -I 5t
e )' ( 2e-' t
X' = y + 2x fx' = Y - 4x
The theorems and definitions
e-2t) and (-3e
both ( 2e-2t
5t
4
e5t ) are solutions to the system X'(t) = ( -2 -3)
-1 XCI).
3_ { y' = x - 5y
Xi - X = et
4.
6
V =y-x
fx' = Y
14.
6
( 3e4t
4t) ( -2t)
6: 41 , : -2t ,
-e -21
(2e 41
2e
4')
e4t
introduced in the section in- The solutions are linearly independent because 5_ { y'=x+y 41,y' =x+cost
dicate that when solving an
n X n homogeneous system
W ((2:-2t, -e5~
-2t) ( 3 5t))
=
I2:--2t2t 3e5t
- e 5t l = 7e 3t 0/= O. 7.
fx' - Y = °
V + x = 2 sin t 8.
fx' +x -
11' - x =
3y =
e- I
° In Exercises 15-20, determine whether the given matrix is a
fundamental matrix for the system.
of linear equations X' (I) =
A(I)X(t), we find n linearly
independent solutions. After
A general solution is given by 15. cI>(t) = (
-2e -8t
5e
e-8t e-' and
-t)
-2t In Exercises 9-14, determine if the given vectors are linearly
(2: -2t
finding these solutions, we
form a fundamental matrix X(t) = (t)C = independent. X'(t) = (-~ ~~)X(t)
that can be used to form a
e-lO' _3e St)
general solution or solve an 16_ cI>(t) = ( 2e -lOt eS, and
initial-value problem.
*
11.
(cos2t)(Sin2t)
-2 sin 2t' 2 cos 21
X'(t) = (_; _1~)X(t)
322 Chapter 6 Systems of Differential Equations 323
6.4 First-Order Linear Homogeneous Systems with Constant Coefficients
*17. «1>(1) = (cos I :- 2 sin I 2 cos I - sin I) First-Order Linear Homogeneous Systems with
sml cos I X" = ~ ( CMW"Y'([Xf + [Y'f)(b-l)/2 _
and X'(I) = U =;)X(t) 1
2CDpAX'~)
Constant Coefficients
(' Distinct Real Eigenvalues (' Complex Conjugate Eigenvalues (' Alternate
;:~:) and X'(I) = (=~ ~)X(t) Method for Solving Initial-Value Problems (' Repeated Eigenvalues
~ ( CMW"X([Xj" + [Y'fP-l)12 + .'
18. «1>(1) = (::
Y" = -g -
0) Now that we have covered the necessary terminOlogy(~I:e ~~ o~~. att:~:)ion to solving
27e-6'
*19. «1>(1) = -2ge- 6,
( 36e- 6 ,
-e'
-e-' e and
"
f CDPAY"V[X'f + fry)
e-' 0 linear systems with constant coefficients. Let A = af2 af2 '. . . afn be an n X n
Write this system of two second-order nonlinear
-2 0 -3) ti
ons as a system of four fiIst-order equations.
equa·
anI Q n2 ann
X'(I) =
( 3
-4
3
0
5 X(I)
-3
25. (principle o~ Superposition) (a) Show that any lin.
ear combmation of solutions of the homogeneou
matrix with real components. In this section, we will see that a general solution to the
homogeneous system X' = AX is determined by the eigenvalues and corresponding
tern X'(t) = A(t)X(t) is also a solution of the h~:~:
eigenvectors of A. We begin by considering the cases when the eigenvalues of A are
geneous system. (b) Is the Principle of Superposition
2e-' sin I ) ever vahd for nonhomogeneous systems of eq f
distinct and real or the eigenvalues of A are distinct and complex, and then consider
e-'(cos I + 2 sin I) Explain. ua IOns? the case when A has repeated eigenvalues (eigenvalues of multiplicity greater than one).
-3e- t sin t
26. Show that each matrix is a fundamental matrix for the
gl.ven system. In each case, use the fundamental rna.
tnx to construct a general solution to the system. Distinct Real Eigenvalues
(a) (cos 3t - sin 3t -sin 3t - cos 3t). In Section 6.3, we verified that a general solution of the 2 X 2 system X' =
In Exercises 21-23, use the given fundamental matrix to ob-
tam agenera; solution to the system of first-order homogeneous
sm 3t
'. X'(t) = ( - ; -~)X(I)

cos 3t '
(_~ =Dx, where the eigenvalues of (_~ =D are Al = -2 and A2 = 5 and
equatl?n'. X = AX. Also, find the solution that satisfies the
gIven mInal condition.
(b) (e'(sin 31 - 3 cos 3t) 5e' cos 31 )
d. .
correspon mg eIgenvectors are VI = (I)2 and V2 (":'3I)' respective
= . Iy, . IS
21. «1>(1) = (~::: _:=:), A G ;), X(O)

= = m 2e' sin 31
~'(t) = (_~ ~)X(t)

e'(cos 31 - 3 sin 3t) ;
XU) = (t)C = Cl 2e -2,
5
( e-2') + C2(-3ee5' ') = Cl (1)2 e _",- + C2(-3)1 e .
5,
22. «1>(1) = (3e4<

2e4<
l
e ')
e " A = (62 -3)I' X(O) = (-46) (c) (e'(sin 31 + 3 cos 3t) 5e-' cos 31 )
' More generally, if the eigenvalues of the n X n matrix A are distinct, we may expect a
2e-' sin 31 e-'(cos 3t + 3 sin 3t) ;
23. «1>(1) = (2 cos 21 + 2 sin 21 2 cos 21 - 2 sin 21) general solution of the linear homogeneous system X' = AX to have the form
-sm 21 -cos 21 ' X'(t) = (~ =~)X(t)
XU) = ClvleA" + c2vzeA2t + .. , + Cnvne A"',
A = (_~ _~), X(O) = (-~) (d) Use the results of (a), (b), and (c) to solve (a), (b),
24. (Mo~eling the :"otion of Spiked VoUeybaU) Under and (c) subject to X(O) = (n In each case, graph
where A" Az, ... , An are the n distinct real eigenvalues of A with corresponding eigen-
vectors Vb Vz, . . • , v"' respectively. We investigate this claim by assuming that X =
certam assumptions, the position of a spiked volley- the resulting solution for 0 :S t :S 2?T. What sim. veAl is a solution of X' = AX. Then, X' = AveA< must satisfy the system of differen-
ball can be modeled by the system of two second- i1arities and differences between the graphs do you
order nonlinear equations tial equations, which implies that
observe?
Using Iv = v yields
AlYeA< = AveA<
AveA< - AlYeA< = 0
• Shawn S. Kao, Richard W Sellens, and Joan M. Stevenson "A Math ma Ii
Applications," Journal of Applied Biomechanics Human Ki 'to P bleh cal Model for the Trajectory of a Spiked Volleyball and Its Coaching
, ne lCS U IS ers, Inc. (1994), pp. 95-109. (A - AI)veA< = O.
324
Chapter 6 Systems of Differential Equations
6.4 First·Order Linear Homogeneous Systems with Constant Coefficients 325
---
*"
Then, because. eAt 0, we have (A - AI)v = O. In ?rder for this system of equations ,
to have a solu1:J.on other than v = 0 (remember that eIgenvectors are never the zero Ve . cle + 3t X -- cle 3t3+ C2eSt . Several solutions along with the direction
3t C2 e St) or {(t)
( 2cle yet) = 2cle t
~-~~ ~
field for the system are shown in Figure 6.10. (Notice that each curve corresponds
IA - All = O.
to the parametric plot of the pair {X(t) for particular values of the constants CI and
A solution A to this. equat~on is ,m
eigenvalue of A while a nonzero vector v satisfy. yOO ..
C2.) Because both eigenvalues are positive, all solutions move away from the ongm
u:
g (A - AI)v = 0 IS an eIgenvector that corresponds to A. Hence, if A has n distinct
as I increases. The arrows on the vectors in the direction field show this behavior.
eIgenvalues {AI> ,1.2, ... , An}, we can fmd a set of n linearly independent eigenvec.
tors {VI> V2,· .. , v n }. From these eigenvalues and corresponding eigenvectors we fi
the n linearly independent solutions ' onn
Complex Conjugate Eigenvalues
If A has complex conjugate eigenvalues Al = a + {3i and ,1.2 = a - (3i with corre-
Figure 6.10
Therefore, if A is an n X n matrix with n distinct real eigenvalues {A }" sponding eigenvectors VI = a + bi and V2 = a - bi, one solution of X' = AX is
a general solution of X' = AX is the linear combination of the set of sol~~~ You may notice that general X(I) = vle"t = (a + bOe(a+/3.)t = e"t(a + bOe i /3t = e"t(a + bi)(cos {31 + i sin (3t)
{XI> X 2, ... , Xn}, -
solutions you obtain when
solving systems do not agree = eW(a cos (3t - b sin (31) + ie"t(a sin (3t + b cos (3t)
with those of your classmates = XI(t) + iX2(t).
or those given in the Exercise
,'i',,"'IM' 6.4 solutions. Before you be·

come alarmed, realize that
your solutions may be correct.
Because X is a solution to the system X' = AX, we have that x;(I) + ix:,(t) = Axl(t)
+ iAx 2 (t). Equating the real and imaginary parts of this equation yields xW) = Axl(t)
and x:,(t) = Ax2(t). Therefore, XI(I) and x2(1) are solutions to X' = AX, and any linear
Find a general solution of X' = (; - ~ ) X. You may have simply selected combination of XI(t) and X2(t) is also a solution. We can show that XI(t) and X2(t) are
different values for the arbi· linearly independent (see Exercise 41), so this linear combination forms a portion of a
trary constants in finding general solution of X' = AX where A is a square matrix of any size.
:.1 Solution The eigenvalues of A (5o -1)3 are found with 15 -0 A 3-,1.
= -1 I =
eigenvectors.
(5 - ,1.)(3 - A) = o. The eigenvalues bf A are AI = 3 and ,1.2 = 5. An eigenvector

VI =. G:) co~esponding to Al = 3 satisfies the system (~ - ~ ) (;:) = (~), so Theorem 6.5 - - - - - - - - - - - - - - - - - - - - - - ' 1
Let A be a square matrix. If A has complex conjugate eigenvalues Al = a + {3i
solutIOns to this system satIsfy 2x1 - YI = 0 or YI = 2x1 • Choosing XI = 1, we
*"
and ,1.2 = a - {3i, (3 0, and corresponding eigenvectors VI = a + bi and
obtain the eigenvector VI = (;). Similarly, an eigenvector V2 = (;~) correspond. V2 = a - bi, two linearly independent solutions of X' = AX are XI(t) =
e"t(a cos (3t - b sin (3t) and X2(t) = e"t(a sin (3t + b cos (31).
ing to ,1.2 =5 satisfies (~ =; )G~) = (~), which indicates that Y2 = o. Notice that in the case of
Hence, V2 = (X~) is an eigenvector for any value of X2. If we let X2 = I, complex conjugate eigenval·
ues, we are able to obtain two If A is a 2 X 2 matrix with complex conjugate eigenvalues Al = a + {3i and ,1.2
V2 = (~). Therefore, a general solution of the system X' = (; . - ~ )x is
linearly independent solutions
from knowing one of the
= a - (3i and corresponding eigenvectors VI
lution of X' = AX is
= a + bi and V2 = a - bi, a general so-
eigenvalues and an eigenvector
that corresponds to it. XU) = CIXI(t) + C2~2(t) = cle"t(a cos {31 - b sin (3t) + c2e"t(a sin {3t + b cos (3t).
X(t) = clvle"t + C2v2e"t = CI (~) e 3t + C2 (~ ) eSt.
Remember that the system X' = (; - 31) X is the same as the system {X', -- 5x - Y. "ii"i,IN'
Y = 3y _._
Thus, we can write the general solution obtained here as X(t) = (X(t») =
y(t)
Find a general solution to X' = (! -2)
-I X.
326
Chapter 6 Systems of Differential Equations 6.4 First-Order Linear Homogeneous Systems with Constant Coefficients 327
Solution The eigenvalues of A = (! =D are Al = I + 2i and ..1.2 = I - 2i. 5 - ..I.

-6
5
-6 - ..I. -5
2 I
An eigenvector VI = (;;) that corresponds to Al = I + 2i satisfies 1 6 6 5-..1.
(2 ~ 2i -2-~ 2i)(;;) = (~). The first equation, (2 - 2i)xI - 2YI = 0, is equiv.

6
= (5 - ..1.)1- 6- A 5 -::..\1- 51~6 5-::..\1 + 21~6 -66- AI
2 - 2i
alent to -4-- times the second equation, 4xI + (-2 - 2i)YI = O. We can Write = (5 - ..1.)(..1.2 + A) - 5(6..1.) + 2(6..1.) = -..1.(..1.2 - 4..1. + 13) = O.
the system as e~ i -~)(;;) = (~). which indicates thatYI = (1 - i)xI. Chaos. Hence, Al = 0 and ..1.= (4 ± ~)/2 = 2 ± 3i, so ..1.2 = 2 + 3i and
ing XI = I, we obtain VI = (;;) = (I ~ i) = G) + i( _~). Therefore, in the nota.

..1.3 = 2 - 3i. An eigenvector VI = ~:) corresponding to Al = 0 satisfies the sys-
tion used in Theorem 6.5,
a = G) and
tem (-~ -~ -~)~:) = (n which, after row operations, is equivalent to
With a = I and f3 = 2 from the eigenvalues, a general solution to the system is
XU) = cle'[G) cos 2t- (_~) sin 2tJ + c2e [G) sin 2t+ (_~) cos 2tJ
l
(~ ~ !)~:) = (n Thus, XI + YI = 0 and ZI = O. If we choose YI = I,
cle' cos 2t + coe' sin 2t ) VI = (-~;) = (-l} One solution of the system of differential equations is
(
= cle'(cos 2t + sin 2t) + c;e'(sin 2t - cos 2t) .
Figure 6.11 shows the graph of several solutions along with the direction field for
the equation. Notice the spiraling motion of the vectors in the direction field. This
XI = vle"" = (-l)e(O)' = (- i} We find two solutions that correspond to the com-
Figure 6.11 We see that all is due to the product of exponeniial and trigonometric functions in the solution' the
(nontrivial) solutions spiral away
from the origin.
exponential functions cause x(t) and Y (t) to increase, while the trigonometric
tions lead to rotation about the origin.
func- plex conjugate pair of eigenvalues by finding an eigenvector V2 = (~:) corresponding
'-
Sketch the graphs ofx(t), y(t), and t% in Example 2 ijx(O) = 0 and yeO) = I. to ..1.0
-
= 2 + 3i. This vector satisfies the system (3 =t 6
5
_8 _ 3i
6
!5
3 - 3i
)(;~) =
Z2
Calculate limHoo x(t), lim Hoo y(t), limH x(t), and IimH yet).
! -t(1~ i))~:)
_ oo _
oo
Initial-value problems can be solved through the use of eigenvalues and eigen'
vectors as well. m, which can be reduced to (~ + = m. Therefore, the
"F',,"'. components ofv2 must satisfy X2 = 1-c1 + i)Zl andY2 = -Z" If we let z, = 2, then
-~ (I ~2+ i) = (I)
~2 + (I) . .mdependent solutIOns
.
-~) X subject to X(O) = m.
5 X2)
~~ = ~ i = a + hi. Thus, two lmearly
Solve X' = ( -6 (
6
that correspond to the eigenvalues ..I. = 2 ± 3i are
(-~ -~ -~) satisfy I) (I) 1 21

(e (cos 3t - sin 3t))
-2e 2 , cos 3t
1 Solution The eigenvalues of A = X, = e'l
[( -2 cos 3t -
2
0 sin 3t =
0 2e 21 cos 3t
328 Chapter & Systems of Differential Equations 6.4 First-Order Linear Homogeneous Systems with Constant Coefficients 329
and
Alternate Method for Solving Initial-Value Problems
2t
X3 =e
2t
[(
1) (1)
-2 sin 31 + 0 cos 31
1= (e (Sin 31t + cos 31»)
-2e: ~in 31 ,
We can also use a fundamental matrix to help us solve homogeneous initial-value prob-
lems. If «I>(t) is a fundamental matrix for the homogeneous system X' = AX, a gen-
2 0 2e 2 SIn 31 eral solution is X(t) = «I>(t)C, where C is a constant vector. Given the initial condition
X(O) = Xo, then through substitution into X(t) = «I>(t)C,
so a general solution is
X(O) = «I>(O)C
Xo = «I>(O)C
_ (-1) +
- Cl 1 C2
2t
(e (cos ~It - sin 31»)
-2e- cos 31 + C3
2t
(e (Sin 31 + cos 31»)
-2e 2t sin 31
C = «I>-\O)Xo.
40 2t = AX, X(O) = Xo is X(t) =
o 2e cos 31 2e 2t sin 31 Therefore, the solution to the initial-value problem X'
«I>(I)«I>-I(O)Xo.
20 2
_ (-Cl + c2e t(cos 31;: sin 31) + c3e2t(s~ 31 + cos 31»)
-20
- Cl - 2c2e cos 31 - 2c3e2t SIn 31
2c2e2tl1os 31 + 2c3e2t sin 31

.
"f·,,"'.'
Use a fundamental matrix to solve the initial-value problem X' = (! -Dx sub-
-40
Application of the initial condition X(O) = (~) yields
zCe) ject to X(O) = (_;).
CC~I~::C: C3) (~).

Ca)
X(O) =
+ C2 + C3
= which gives us the system of equations
:1 Solution The eigenvalues of A = (! -D are Al = 2 and A2 = -3 with
10 40 Y
[
-Cl
Cl -
2C2 = 2
= 0
2C2 = 0 with solution Co
•
=1
,
Cl = 2, and C3 = 1. Therefore, the corresponding eigenvectors VI = (D and V2 = (_!), respectively. A fundamental
x
(:~: _~~~3,). We calculate «1>-1(0) by observing
Ca)
-20 solution of the initial-value problem is matrix is then given by «I>(t) =
y
G _!), so «1>-1(0) = _41_ 1 (=~ -D= ( -t)(=~

2
-40
X(t) =
-2 + e t(cos 3t - sin 3t) + e 2t(sin 31 + cos 31»)
2 - 2e 2t cos 31 - 2e 2t sin 31
that «1>(0) = -~).
( 2t Hence,
2e cos 3t + 2e 2t sin 3t
Cb) 2t -3 -2 -1 / 2 3x
-2 + 2e cos 31 )
2 - 2e 2t cos 31 - 2e 2t sin 31
//-1
Figure 6.12 (a) Graph of X(I), = /
(
2e 2t cos 31
/
Y(I), and Z(I). Identify each + 2e 2t sin 31 /
/
graph. (b) Graph of {;rl), In Figure 6.12, ;Ve graph. x(t), :,(1), and z(t) and we show the parametric plot of
{x(t),y(t), z(I)} In three dunenslOns. Notice that limt--+~x(l) = 00, limH~y(t) = 00,
,/
(b)
= (-t)(:::
o :S I :S 7T/2. Z(I)
and hmH~ z(t) = 00, so the solution is directed away from the initial point (0, 0,2).
L Figure 6.13 (a) Graph of x(t) Figure 6.13 shows a graph of the solution. Notice that as t ~ 00, the values of x(t)
and yet). Identify each graph.
2t
-2 + 2e cos 31 ) (b) Graph of {;n along with
andy(t), where X(I) = ~~~~). are both close to ~e2t because limt--+~ e- 3t = O. This
2t
~
...,." Verifo that X(I) = ( 2 - 2e cos 3t - 2e 2t sin 3t satisfies the initial conditions in
2t 2t the direction field associated means that the solution approaches the line y = x because for large values of t, x(t)
o ,,"~." 2e cos 3t + 2e sin 3t with the system of equations. and y(t) are approximately the same.
Example 3.
330 Chapter 6 Systems of Differential Equations 6.4 First-Order linear Homogeneous Systems with Constant Coefficients 331
Find the solution that satisfies the initial condition X(O) = (~). Find conditions on xo, Yo, and zo, ifpossible, so that the limit as t -> '" of the
Repeated Eigenvalues solution X(t) that satisfies the initial condition X(O) = (~:) is O.
~o
We now consider the case of repeated eigenvalues. Tills is more complicated than the
other cases because two situations can arise. As we discovered in Section 6.2, an eigen_ If an eigenvalue of multiplicity m has fewer than m linearly independent eigen-
value of multiplicity m can either have m corresponding linearly independent eigen. vectors, we proceed in a manner that is similar to the situation that arose in Chapter 4
vectors or have fewer than m corresponding linearly independent eigenvectors. In the when we encountered repeated roots of characteristic equations. Consider a system with
case of m linearly independent eigenvectors, a general solution is found in the same the repeated eigenvalue Aj = A2 and corresponding eigenvector Vj. (Assume that there
manner as the case of m distinct eigenvalues. is not a second linearly independent eigenvector corresponding to Aj = A2') With the
eigenvalue Aj and corresponding eigenvector v], we obtain the solution to the system
"i''''''.' X j = vje"". To fmd a second linearly independent solution corresponding to AJ, in-
stead of multiplying X j by t as we did in Chapter 4, we suppose that a second linearly
independent solution corresponding to Aj is of the form
Solve X' = (~ =~ !)X' Xl = (Vlt + w2)e A".

To find the vectors V2 and Wz, we substitute X2 into X' = AX. Because X, =
Aj(vlt + wl)e A" + vle"", we have
:~ :o:u(rn=tr)O;:dE::;~:n;~I;:SS:~:O:l6:. -~ea:~s:3f~:dO~t the eigen-

Aj(Vlt+ wl)e A" + V2eA"
X, = AX2
= A(v2t + wl)e A"
6 -6 4
AjV,t + (AjWl + v,) = AVlt + AWl'
value Aj = A2 = -2 of mUltiplicity 2 has two corresponding linearly independent
eigenvectors, Vj = (l) and Vl =Cn An eigenvector V3 = (m that corresponds

Equating coefficients yields AjV2 = AVl and AjWl + Vl = Aw2. The equation AjV2 =
AVl indicates that Vl is an eigenvector that corresponds to AJ, so V, = Vj. Simplifying
AjW2 + V2 = AWl, we find that
to A3 = 4 satisfies the system C~ =~ ~)(m = m, which can be reduced

Ajw, + V, = AWl
V2 = AW2 - AjW,
I =l)(m
V2 = (A - AjI)w"
with row operations to the system (~ = m· Hence, X3 = ~ and Hence, W2 satisfies the equation
(A - AjI)w2 = Vj.
Y3 = T' Choosing Z3 = 2, we obtain V3 = (1). A general solution is Therefore, a second linearly independent solution corresponding to the eigenvalue Aj
has the form
X 2 = (vjt + wl)e A",
X(t) = cjvje A" + C2v2eA" + C3 v 3 e "3'
where W2 satisfies (A - AjnWl = Vj.
-4 6 x = Cj(~)e-2' + C2(-~)e-" + C3(~)e4' = ((Cj ~eC~;~~"c:e2e4').

o I 2 cle -2' + 2c3e4' Theorem 6.6 Repeated Eigenvalues with One Eigenvector
Figure 6.14 shows graphs of several solutions to the system along with the di- Let A be a square matrix. If A has a repeated eigenvalue A, = A2 with only one
rection field. Notice that if C3 "" 0, then limH~ x(t) = "', limH~ y(t) = 00, and corresponding (linearly independent) eigenvector v" two linearly independent so-
limH~ z(t) = 00. Therefore, many of the vectors in the direction field are directed lutions of X' = AX are X, = vje"" and X, = (v,t + wl)e"", where W2 satisfies
Figure 6.14 away from the origin. Ii (A - Aj I)w2 = v,.
======================~
332 Chapter 6 Systems of Differential Equations 6.4 First-Order linear Homogeneous Systems with Constant Coefficients 333
If A is a 2 X 2 matrix with the repeated eigenvalue Al = A2 with only one co Figure 6.15 shows the graph of several solutions to the system along with the di-
responding (linearly independent) eigenvector VI> a general solution to X' = AX is r· rection field. Notice that the behavior of these solutions differs from those of the
X(t) = clvle A,! + C2(V l t + w2)e A,! systems solved earlier in the section due to the repeated eigenvalues. We see from
the formula for XU) that lim,->~ x(t) = 0 and limt->~ yet) = 0 (why?). In addition,
where W2 is found by solving (A - AII)W2 = VI. these solutions approach (0, 0) tangent to y = -4x, the line through the origin that
Suppose that A has a repeated eigenvalue Al = A2 and we can find only one corre_
sponding (linearly independent) eigenvector v I. What happens if you try to find a
is parallel to the vector VI = ( _!). (We discuss why this occurs in Section 6.6.)
second linearly independent solution of the form V2teA" as we did in solving higher L.
order equations with· repeated roots of the characteristic equation in Section 4.4?
I'f',,','., o !'f·,,',',*,
[ Find a general solution to X'(t) = (~~ -1)o XU)· Solve X' = (
-4
; ! -3)
2 -3
-5 X.
= (~~ -~)
:lSolution The eigenvalues of A are Al = A2 = -4. An
: 'I Solution The eigenvalues of A = ( ;
-4
! =~)
2 -3
are determined with
eigenvector VI = (;:) that corresponds to Al = -4 satisfies the system
(~: -!)(;:) = (~), which is equivalent to (~ ~)(;:) = (~). Hence,YI = -4xJ, 5; A 4 ~ A =~ 1= _A3 + 6A2 + 15A + 8 =-(A + 1)(A + I)(A - 8)
1-4 2 -3 - A
so if we choose XI = I,VI = (_!) and one solution to the system is XI = (_!)e- 4'.
= 0 to be Al =A2 =-I and A3 =8. An eigenvector VI = (:J that corresponds to
To find W2 = (;~) in a second linearly independent solution X2 = (Vlt + w2)e A",
we solve (A - AII)W2 = VI> which in this case is 6 3 -3)(XI) _(0)o.
Al = -I satisfies
(-42 25 -2
-5
0
YI
ZI
- This system is equivalent to
This system is equivalent to

(~ : -!)(::) =m. Hence, XI =0 and YI - ZI = o. If we let YI =1, then
VI = (r)- Therefore, one solution to the system is XI = (r)e-'. A second linearly
independent solution X 2 = (VI! + w2)e A,( is found by solving (A - AII)W2 = VI

which indicates that 4X2 + Y2 = -1. If we let X2 = 0, then Yz = -1, W2 = (_ ~).
and a second linearly independent solution is given by ( ~ ~ =~)(;~) (~) for the vector W2 (;~). This system is equiv-
= =
-4 2 -2 Z2 I· Z2
Hence, a general solution is

alent to (~ ~ - ~)(;~)
o 0 0 Z2
= (-t),
0
which indicates that X2 = -k and Yz - Z2 = *.
Figure 6.15
Hence, if Z2 = 0, Y2 = t so W2 = (-1)- A second linearly independent solution is
6.4 First-Order Linear Homogeneous Systems with Constant Coefficients 335
334 Chapter (; Systems of Differential Equations
((1} + n)~ -'.

Similar to the previous case, V3 = V2 = Vb W2 = W3, and the vector U3 is found by
fum X, • A iliffd lin=ly ind'Pmdd>' 00•• "00 • fonnd "",' solving the system
(A - A1I)U3 = W2'
an eigenvector V3 = ~~) corresponding to A3 = 8, which satisfies The three linearly independent solutions have the form
and X3 = (VI ~ + W2t + U3 )eA".

(-; -! =;)(~:) = (~). Because this system is equivalent to
Notice that this method is generalized for instances when the multiplicity of the re-
-4 2 -11 Z3 0
peated eigenvalue is greater than 3.
(
~1 0~ J~)(X3)
~: =(0)~' the components of V3 must satisfy X3 + ~Z3 = 0 and
Y3 + ~Z3 = O. If we let Z3 = -2, then X3 = 9 and Y3 = 7. Hence, V3 = (_~) and o

X3 = (_~)e81. A general solution';is then given by
Solve X' = (j -l)x. 2
: ]I Solution The eigenvalues are found by solving
1- A -1
1
4-A
I
2
1- A
-1 \ \ 2 -1 \
= (1 - A) 2 4 - A - (1) -3 4 - A + (1) \ -32
\
Figure 6.16 shows the graph of several solutions to the system along with the di· = _A3 + 6A2 - 12A + 8 = -(A - 2)(A2 - 4A + 4) = -(A - 2)' = O.
Figure 6.16 rection field.
Hence Al = A2 = A3 = 2. We find eigenvectors VI = ~:) corresponding to A = 2

0MP Using the formula for X in Example 7 and the direction field in Figure 6.16, deter-
~'
mine if the solutions approach the origin as t increases if C3 oF O. What happens to
the solutions if C3 = 0 and either Cl oF 0 or C2 oF O?
A similar method is carried out in the case of three equal eigenvalues Al = A2
with the system C~ -i -i)~:) m· = With elementary row operations, we
(~ ! !)~J (n
= A3 , where we can find only one (linearly independent) eigenvector VI' When we en·
counter this situation, we assume that reduce this system to = Thus, Xl =0 and Yl + Zl = O. If
= vleA", X2 = (V2t + w2)e A", = (V3~ + W3t + U3 )eA".

Xl and X3
Substitution of these solutions into the system of differential equations yields the fol-
we select Zl = 1, then Yl = -1 and VI =( -1} Therefore we can find only one
lowing system of equations, which is solved for the unknown vectors V2, W2, V3, w" (linearly independent) eigenvector corresponding to A = 2. Using this eigenvalue
and U3:
A1V2 = AV2, (A - A1I)W2 = V2, A1V3 = AV3, (A - A1I)W3 = V3, and eigenvector, we fmd that one solution to the system is Xl = vle
2
! = (-1)e 2t
•
and (A - All)U3 = W3'

The vector W2 = ~~) in a second linearly independent solution of the form X 2

'" .. EXERCISES 6.4
(=i + r-li~r ~ir:: i~:~:~ :(l~flX~):~:) In Exercises 1-12, use the given eigenvalues and correspon-
3 2 -3) ( 3+i )
ding (linearly independent) eigenvectors of the matrix A to Tmd 12. A = 1 1 1; Al = -2 + i, VI = -2 - i ;
(
a general solution of X' = AX. o -4 -4 4
Hence X2 = -I and Y2 + Z2 = -I, so if we choose Z2 = 0, then Y2 = -I. There.- l.A=G -:);AI =2,VI=G}A2=4,V2=(:) (3-
A2 = -2 - i, v, = -2 +i i) ; A, = 4, V, = (-7)
-i
~ (=iJ. ~ ((-1)'+ (=i)),,' -m~"'~,ho_
4
f=, w, wX, F_y, 2. A=(_~ ~);AI=0'VI=m;A2=4,v,=(-~)
In Exercises 13-32, find a general solution of the system.
tor U3 = (~~) in the third linearly independent solution X3 = ~ + wzI + U3 )eA" (VI
3. A = (=~ ~); Al = -4, VI = (~} A2 = 2, v, = m 13 X' = ( 1 -10)x
. -7 10 14.
X'
{ y'
= X - 2y
= 2x + 6y
4. A=(~ ~}AI=A'=5'VI=(~)'V2=m
~)(~:) = (= :),
l
dx =6x-y
by solving the system (A - AI)u3 '" w2 . This yields (-; -I _ *S.A=(_: ~}AI=A'=2'VI=(:) *15. :;
-=5x
16. X' = (4
-5 -4
3)x
-3 2 2 Z3 0
(~o 0~ 0~)(~:) = (=;). 6. A = (~ -~); Al = 6i, VI = U} A, = -6i,

dt
!
which is equivalent to Therefore, X3 = -2 and Y3 + Z3 =
{x' = 7x ~=8X+9Y
Z3 0 V,=(~I") 17
. y' = 5x - 8y 18. dy
dt = -2x - 3y
- 3. If we select Z3 = 0, then 13 = -3. Hence U3 = (=~). so a third linearly inde. 7. A= (_; _~). Al = I+2i, vI = (-i: I}
*19. X' (-~ _1~)X X' = 8x + 5y
A, = 1- 2i, v, = (i ~ 1)
= 20. { y' = -lOx - 6y
p~don,wlutiMi'X, ~
gIven by
(( -if, + (=i)' + (=m'>A,""~WIUtiUUti~ 8. A= (=~ : -~); Al = 1, VI = (:); A2 = 2,
21.\ :dt : ~: ~ ::y 22. X' (I-2 -78)x
=
-3 1 3 1
V2 = (~} A3 = 3, V, = m *23.
X'
{ y'
= -6x + 2y
= -2x - lOy 24 . X' = (02 0 8)x
3 0
*9. A= 0 2 25. X' = (_~ ~)X 26 {x'= y
. y'=-13x-4y
(
o 0
- C2e21 + C3( - 1- 2)e zl
V3=m
*27. X' (~-~=
o 0-1
~)X
21
+ C3 ( -~ )e
+ C2( -I - l)e zl
28. X' (-~o -5-~ -4
-~)X
-cIe - t- 3 Zt •
=
cle 21
+ C2 te + C3
21 (t2" e
2
) 21
(-~ -~)X
-I
29. X' = -1
-4 2 -3
5 4
-5)OX
30. X' =
C ~ -1
I
-,

X' = X + 2y + 3z Group A Group B Group A Group B and Yo so that at least one of x(l) or y(l) approaches
*31. y' = Y + 2z =x
{ X' zero as I approaches infinity. (e) Is it possible to choose
z' = -2y+z 43. {y , 2
= y
(a) X' = x- Y
(I) Xo and Yo so that both X(I) and yet) approach zero as I
48. ( y' = a
X' =Y approaches infinity?
32. y' = z 54. (a) Find a general solution of X' = AX if
{
z' =x - y + z
In Exercises 33-40, solve the initial-value problem.
(~ (~)
-I' --0.5
-0.'S
0,' 1 x
and
00 A - i
(1 1!)
~ ~ ~), (~ ~ ~),
-I
33. X' = DX, X(O) = X' =-x
(b)
44. { y' = 2y ,f':":': c' XU) = e 2'(CI COS I - C2 sin I) (iii) A = :
--I ..
34. X' = (-~ ~)x, X(O) = (~) ..'. ~:'D~'i .~ 49. Show that both ( y(l) = e"'(cI sin I + C2 cos I) and OOA 000"
c
t
l
+ e 2i') + c2i(e 2"
~d (1 : ; 1)~) err,~ ,~-
.. ""..<'- -' I
X(I) = e (2-"'[c, (1 - I)]
(~ ~)x, X(O) = (~) ~r~;~~~~'~:+~7-:~~~~~1~ m

*35. X' = -, - ,.
,,",","'
y(l) = te(2-"'[C,i(1 - e 2i') + C2(1 + e 2it )] Q,) A -
(~ m -~T are general solutions of the system

(a), find the solution that satisfies the initial condi-
n) ;, ,-
36. X' = !)X, X(O) = X'(I) = a(l) - y(l)
( y'(I) = X(I) + 2Y(I)'
37. X' = (~ -~)x, X(O) = (~) 45. {

X' =-x
, 2
Y = - Y
(e)
50. Solve the systems (a) (X; = a -+Y3 ;
y = -x y
tioo X(O) - -I '"" .00 ,",h <,V),
X' = a (X' = x + 4y . X2(1), x3(1), and X.(I) for 0 :5 t :5 10. How are the so-
38. X' = (_ ~ ~!)x, X(O) = (~) (h) {y' = 3x + 2y; (e) y' = -a _ y subject to
lutions similar? How are they different? (e) Indicate
x(O) = I and yeO) = 1. In each case, graph the solu-
-I how to generalize the results obtained in (a). Explain
tion parametrically and individually.
(-~ ! ~)x, Ci) *51. Find a general solution of the system X' =
how you would find a general solution of X' = AX if
~ ~ I ~ ~).
*39. X' = X(O) =
(o~ -~ ~) X. Graph the solution for various values

-2
A = How would you find a
-~o -2
-~)x, X(O) = (~)8
X' =-y -4 1 , 0 0 0 A 1
(d)
40. X' = (-: 46. { y' = a of the constants.
-1
o 0 0 0 A
52. How do the general solutions and direction field of general solution of X' = AX for the 5 x 5 matrix
41. Show thatxI(I) = ea'(a cos f31 - b sin f31) and X2(1) =
(~ ~)X
(~ ~ I ~ ~?
the system X' = change as A goes from
ea'(a sin f31 + b cos f31) are linearly independent func-
tions. -2 to O? Solve the system for values of A between A = How would you find a gen-
- 2 and 0 and note how the solution changes. 000 A 1
42. Show that in the 3 x 3 system X' = AX with the
eigenvalue A of multiplicity 3 with one correspond-
53. Consider the initial-value problem o 0 0 0 A
eral solution of X' = AX for the n x n matrix
X' 1 -:)X
l
ing eigenvector VI, the three linearly independent
X' = -x - Y = (
solutions are XI = vIe).", X 2 = (VII + w2)e).", y' a (e) -2 -, A 0 0 0
47. ( =
0 A I ... 0 0
and X3 = (VI~ + W21 + U3 )e).", where U3 satisfies x(O) = Xo, yeO) = Yo
A=
(A - AI)u3 = W2 and W2 satisfies (A - AI)w2 = VI' (a) Graph the direction field associated with the
system X' = (
-2
I -f)x.
-,
(b) Find conditions on Xo
0
0
0
0
0 A
0 ... 0 A
In Exercises 43-49, without solving each system, match each
system in Group A with the graph of its direction field in
Group B.
340 Chapter & Systems of Differential Equations
&.5 First-Order Linear Nonhomogeneous Systems: Undetermined Coefficients 341
and Variation of Parameters
First-Order Linear Nonhomogeneous Systems: sume that there is a particular solution of the form Xp(t) = ae 3t + bt + c. Then,
Undetermined Coefficients and Variation of Parameters Xp(t) = 3ae 31 + b and substitution into the nonhomogeneous system X' =
r Undetermined Coefficients r Variation of Parameters . AX + (~)e3t + mt, where A = (~ ~). yields
In Chapter 4, we learned how to solve nonhomogeneous differential equations through
the use of undetennined coefficients and variation of parameters. Here we approach 3ae
31
+b = Aae
3t
+ Abt + Ac + (~)e31 + (~)t.
the sO~e:o: ~ :;:e:S(::i~,)~~~og(~':~":ti:~)~(:~~)'::~(" 00 ,

Collecting like terms, we obtain the system of equations
Xn(t) anI a n2 ... ann ~(t) 3a = Aa + (~) (Coefficients of e

3
')
fundamental matrix of the system X' = AX. Then a general solution to the homo-
b = Ac (Constant terms)
g'=~ 'Y""" X' = AX " X= ~(~C ""= C = (:) " m • X I ,_"'" """"
1 Ab + G) = 0 (Coefficients of t)
To find a general solution to the linear nonhomogeneous system X' = AX + F(t), we

proceed in the same way we did with linear nonhomogeneous equations in Chapter 4.
.. 3t
From the coeffIcIents of e , we find that (A - 31)a = (-1) (-3 8)(a
0 or 2 -3 a2l ) -_
If X p is a particular solution of the nonhomogeneous system, then all other solutions
X of the system can be written in the fonn X = <fI(t)C + Xp (see Exercise 36). (- ~), where a = (:~). This system has the unique solution a = (=0. Next, we
Undetermined Coefficients solve the system Ab + m = 0 or (~ ~)(~~) = (_ n for b. This yields the unique
We use the method of undetennined coefficients to find a particular solution Xp to a

nonhomogeneous system in much the s:qne way as we approached nonhomogeneous
solution b = (_1)
~. Finally, we solve b = Ac or (02 8)(C)
0 c~ = (-!) .
~ for c, whIch
higher order equations with constant coefficients in Chapter 4. The main difference is gives us c = ( _~). A particular solution to the nonhomogeneous system is then
that the coefficients are constant vectors when we work with systems. For example, if
= (e~2t) = (~)e-2t + (~) (i) + (0) _(-~e31 it)
+ bl + c = (-~)
we consider X' = AX + F(t) where F(t) and none of the - l'
Xp(t) = ae 31 ,e31 + 0 t I - 2 31
tenns in F(t) satisfy the corresponding homogeneous system X' = AX, we assume that -7 -16 -7e - 16
a particular solution has the fonn XP(t) = ae -2t + b where a and b are constant
vectors. On the other hand if A = -2 is an eigenvalue of A, we assume that Xp(t) =
ate- 2t + be- 2t + c.
so a general solution to X' = AX + (~)e31 + (~)I is
"f''''''.' 31
Solve X' = 2 (0 08) X + (et ) . To give another illustration of how the form of a particular solution is selected,
:J Solution In this case, F(t) = (e;t) = (~) e 3t + (n t and a general solution to suppose that F(t) = (4sine -I (4).
2t) = 0 sm 2t + (0)1 e - t ·m E le
xamp l. In this case, we
21 21
assume that Xp(t) = a cos + b sin + ce -t and find the vectors a, b, and c through
the corresponding homogeneous system X' = (~ ~)x is Xh(t) = cI (De 4t
+ substitution into the nonhomogeneous system.
C2 (-~) e -41. Notice that none of the components of F(t) are in Xh(t), so we as- . .
Fmd a partlcular solutIOn to X = 2 . ,(0 08) X + (4 sin
e -t 21) .
342 Chapter 6 Systems of Differential Equations 6.5 First-Order Linear Nonhomogeneous Systems: Undetermined Coefficients 343
Variation of Parameters
Variation of parameters can be used to solve linear nonhomogeneous systems as well.
x
2 In much the same way that we derived the method of variation of parameters for solv..;
1.5 ing higher order differential equations, we assume that a particular solution of the non-
homogeneous system can be expressed in the form
05
1.5 2 t
XP(t) = cI>(t)V(t), where Vet) = (::~~~). Therefore,
vn(t)
Vet) = f cI>
-I -
(t)F(t) dt -
f (:f e9'
;1 21
+ ~ell') _ (ife9' + f,e )
+ 17e 4' dt - 17e 2, + 1.
2S e
41 .
lll
Notice that Xp = cI>(t)V' (t) + cI>' (t)V(t). Then ifXp satisfies X' = AX + F(t), we have 7e
(a) cI>(t)V'(t) + cI>'(t)V(t) = AcI>(t)V(t) + F(t). By variation of parameters, we have the particular solution
However, the fundamental.matrix cI>(t) satisfies X' = AX, so cI>'(t) = AcI>(t). Hence,
Xp(t) = cI>(t) fcI>-I(t)F(t) dt = ( Te~II'
-I -II' 3e- ')(if e91
e- 4t
4
+ f,e
~e21+.ge4t
llt
) = (f.i4++ 2
e- ')
e--' t '
cI>(t)V'(t) + AcI>(~)V(t) = AcI>(t)V(t) + F(t)
cI>(t)V' (t) = F(t). Therefore, a general solution is given by
05 Multiplying both sides of this equation by cI> -I(t) yields
X(t) = cI>(t)C + Xp(t) = ( :~Ilt
-I III 4
3e- ')(c I )
e
-41 +
(f. ++ ~e-21)
5 e-2' .
1
1.5 2 t cI>-I(t)cI>(I)V' (t) = cI>-I(t)F(t) C2 44"9
-0.5
-1 V'(t) = cI>-I(t)F(t). x(t)
In Figure 6.17, we graph x(t), y(I), and {yet) for several values of CI and C2'
-1.5 Therefore, V(t) = fcI>-I(I)F(t) dt, so a particular solution is
-2
(b)
Xp(t) = cI>(t) f cI>-I(t)F(t) dt, Evaluate liml __ x(t) and limt->oo y(t). Does the choice of CI or C2 affect the limit?
and a general solution to the system is
Variation of parameters allows us to solve systems that we cannot solve using
X(t) = cI>(t)C + Xp(t) = cI>(t)C + cI>(t) f cI>-I(t)F(t) dt. the method of undetermined coefficients.
4
2
l'fiuN.
Solve X' =
-5
( 2 -I ~) X + (e -2:) using variation of parameters.
o lif',,',".' X' = 2x - 5y + esc t
Solve the initial-value problem y' = X - 2y + sec t , 0< t < 7T/2.
{
x( 7T/4) = 5, y( 7T/4) = I
: ] Solution To apply variation of parameters, we first calculate a fundamental
_1~)X'
-5
(c)
matrix for the corresponding homogeneous system X' =
(
2 The
] Solution First, we write the system as X' = (~ =;)x + (~:~ ~) and solve the
Figure 6.17 (a) Graph ofx(t)

for various initial conditions, (b)
eigenvalues of A = (-; _I~) are Al = -4 and A2 = -II with correspond- corresponding homogeneous system X' = (~ =;)x to obtain Xh(t) =
Graph of yet) for various initial ing eigenvectors VI = (~) and V2 = (-t), respectively, A general solution to C (cos t :- 2 sin t) + c, ( - 5 sin t ). A fundamental
smt -cost-2smt
matrix is cI>(t) =
conditions. (e) Graph of {;~h
I
the homogeneous system is Xh(t) = CI(-t)e- II ' + c2(De- 41 , so a fundamental Cost+2sint -5sinl) 'thO
for various initial conditions. ( sin I cos I - 2 sin t WI mverse
344 Chapter 6 Systems of Differential Equations 6.5 First-Order linear Nonhomogeneous Systems: Undetermined Coefficients 345
<1>-1(1) = (cos I -.2 sin t

-sm I
5 sin I. ). Then
cos I + 2 sm I ' .. EXERCISES 6.5
V(I) = f<l>-l(I)F(I) dl = f(~~:: -2 2.+

sm t
5c~~n/)dl =
In Exercises 1-24, solve the system by undetermined coeffi-
cients or variation of parameters. 18. y'
{
X' = -2,; - 2y - 2z + I
= -2y + z + t 2
Z' = -2y - 5z +t
cos I
1. X' = (-~ _~)x + CI) = -x + y - z + e- 4'
X'
In (sin I) - 21 - 5 In (cos I)) *19. y' = 4x + y - 5z + te -2,

(~ ~)x + C:)
{
( z' = -x - 2y - 2z + t
-2 In (cos I) , 2. X' =
so a particular solution is Jx' = -6x + t

*3. )y' = -x - 6y + t 2
20. X' = (_; _~ -~)x + (t~41)
o 0 6 e6'
f , _(2 2)X + (e-",\

(~ ~)x + (:,)
Xp(t) = (t) -l(t)F(I) dt 1
4. X - 3 I e') 0
21. X' =
2 3
= (cos t + 2 sin I -5 sin /' )(In (sin I) - 2t - 5 In (cos t)) 5. X' = (=~ _I~)X + (:1 :,)
(~
0
sml cost-2smt -2 In (cos t)
(-~ ~)x + (~:)
22. X' = 2 -~)x + (e~)
= (cos I In (sin I) - 2t cos t - 5 cos I In (cos t) + 2 sin 1 In (sin I) - 4t sin I) 6. X' = 0
(! 4t
sin I In (sin t) - 2t sin 1 - sin I In (cos I) - 2 cos I In (cos I) . * {X' = - 3x + 7y + cos t
7. y' = -x + 5y + e- 41 *23. X' =
2
1 -3) X + (e0
I )
-4 -4 t
A general solution to the nonhomogeneous system is 8. X' (-6 -3)X + (
= te
2
(=1-3 -2; -~)X + (:i:'4;)

' )
4 I sm 2t
24. X' =
9. X' =:)x + (:=::)
= (-:
0 0
X' = -3x + 2y + e-' sec 2t In Exercises 25-30, solve the initial-value problem.
COS 1 (Cl + In (sin I) - 21 - 51n (cos t)) + sin t (2C1- 5C2 + 21n (sin t) - 41)) 10. { y'=-lOx+5y+e-'csc21
(
cos 1 (C2 - 21n (cos t)) + sm t (Cl - 2C2 + In (sm t) - 21 - In (cos t))
X'=(-~ ~)X+(~),X(O)=(~)
.
*11. X' = (_~ _~)x + (e" ;~ 2/) 25.
7.5 XCI) To find the solution that satisfies the initial conditions x( 1714) = 5 and y( 1714) = I,
we solve the system of equations
12. \y' =
Jx' = 6x - 5y + e"
X + 4y + e" cos 21
26. X' = (_~ _~)x + (s~ I)' X(O) = m
2.5
X'=(~ 1)X+(~}X(0)=(-D
yCr)
Jx(f) t = Cl Y2 - i C2 Y2 + + Y2 In 2 - i 17Y2 = 5
13. Jx' = -y
II
+ sec t *27.
-2.5
-5
1tI4 1tI2
I
I
I
I
lY(f) = +CI Y2 - +C2 Y2 + +Y2ln2 - t 17Y2 = 1

= x
Jx' = -y
14. )y' = x
+ 4 tan 1 28. X' = (~ -Dx + (/ 2 ~ I)' X(O) = m
-7.5
I
I *15 {x'
= y
, _(-53 2)I X + (10'
29. X - 0) X(O) -_(-21)
I for Cl and c" which yields Cl = i17 - 2 In 2 = 0.184501966 and C2 = -Yz- . y' = -x - 2 cot t
In 2 = -2.107360743. We use the values of Cl and C2 to graph x(t) and yet) for X' =-y 30. X' = (34 -I)X + (cos I), X(O) = (0)
Figure 6.18 0< t < 17/2 in Figure 6.18. -I sm 0 1
16. { y' = x - 2 esc 1
C~ -~ _~)x + (2)
In Exercises 31-35, without solving each of the following
Find limt->o· x(t), limHo• y(t), limH.".;z- x(t), and limH.,,-/z- y(t) for the solution to initial-value problems, match each problem with the graph of
17. X' =
Ihe initial-value problem in Example 3. its solution.
346 Chapter 6 Systems of Differential Equations 6.6 Phase Portraits 347
X(t) 38. (a) Show that
~ XV, ~ (:i::)
Graph of x(t) (in dark blue) and y(t)
Initial-value problem (in light blue)
Graph of { y(t)
. , . Lcl X x -- L3 e -41
-'"- + '"-e 9
13 ' -
9
l3 e '
1. I'
0 e -9)(S) ds +
= 2x + Y
X'
I: e '/(s) ds
~~(t) a l")
(a) 4
31. y' = -8x - 2y ,yet) ¥,e -4,
{
x(O) = 0, yeO) = I all al2 (fi(t)) + 13
.l.&e 9r _ ..!i.e9t It e +
A = afl a~2 ":.' af", F(t) = J,~t) ,and y -- _ l e -41
i3 13 0
-9s/(S) ds
(
anI Q n2 ... ann hCt) f,e -41 f
o
e 4 )(s) ds
(t) be a fundamental matrix of the system X' = AX.
is the solution to the initial-value problem
(a) Show that if XI and X 2 are any two solutions of
X' = AX + F(t), then XI - X 2 is a solution to X' = dx = -7x + 6y + let)
X' = 2x + y + I AX. (b) Show that if Xp is a particular solution to dt
32. y' = -8x - 2y + I (h) X' = AX + F(t) and. X=y is any solution to X' =
{ dy = -8x + 12y
AX + F(t), then there is an n X I constant vector
x(O) = 0, yeO) = I
1 dt
C~ m~ ." ~ "','C
x(O) = 0, yeO) = I
x.., + "" (b) Is it possible to choose let) so that

lim,_oo X(I) = O? So that lim,_ oo y(l) = O? So
-I that both lim,_ oo X(I) = 0 and lim,_oo y(l) = O?
*37. Use a computer algebra system to solve each of the
Why or why not? Provide evidence of your resuIts
following initial-value problems. In each case, graph
by graphing various solutions.
X' = 2x + y + sin I . . {X(I)
X(I), y(I), and the parametric equatIOns y(l) for the
33. y' = -8x - 2y (c) In Exercises 39-41, graphically compare solutions to each
{
x(O) = 0, yeO) = I indicated values of I.
nonhomogeneous system and the corresponding homogeneous
(a) {(;:) = (=~ =D(;) + (te~) ~ t~ 0 10

system.
x(O) = 0, yeO) = I 39. X'(I) = (-~ _~)X(t) + (~), X(O) = (_:)

X' = (8 IO)X + (t2e-~'); 0~ t ~ 3 X'(t) = -6x(l) + t
(b)
{ x(O)-7= 1,-9 yeO) = 0-M 40. y'(t) = -X(I) - 6y(t) + 12
{ x(O) = 0, yeO) = I
= 2x + y + sin 21
m
X'
dx = 2x - 5y + sin 41
= (_~ _DX(t) + ( ' ;~~ 2t). X(O) =
34. y' = -8x - 2y (d)
{ dt 41. X'(t)
x(O) = 0, yeO) = I
(e) dy = 4x _ 2y - te-'; 0 ~ t ~ 27T
-2
-6
yO) 1 dt
x(O) = 1, yeO) = I
-,
-10
+ y + sin 31
~ Phase Portraits
X' = 2x
(e) 30
35. y' = -8x - 2y
{
x(O) = 0, yeO) = I 20
r Real Distinct Eigenvalues C Repeated Eigenvalues
10
C Complex Conjugate Eigenvalues '.~ Stability
-10
In this section, we consider solutions of
-20
-30 X' = ax + by
{ y' = ex + dy
6.6 Phase Portraits 349
from a geometric point of view. Notice that (0, 0) is a solution of the system of equa.
2. If both eigenvalues are positive, suppose that 0 < A2 < AI' Then e A1 ' and e A" both
tions obtained by setting the right sides of the differential equations equal to zero, become unbounded as t increases. If CI "*
0, then X(t) becomes unbounded along
ax+bY=O the line through (0, 0) in the direction of VI' If c, = 0, then X(t) b.ecomes unbo~ded
{ ex+dy=O along the line through (0, 0) in the direction given by V2' In this case, (0, 0) IS an
unstable improper node.
for all values of a, b, c, and d. In fact, if IAI = I~ ~I"* 0, then the only solution of 3. If the eigenvalues have opposite signs, suppose that A2 "': 0 < AI. and CI *" O. The~,
X(t) becomes unbounded along the line through (0, 0) m the dire~tlOn of v I as It
this system is the origin. Note: We make the assumption that IAI "*
0 throughout this did in (2). However, if CI = 0, then due to the fact :~~ A2 ":: 0, lim,coo X(t) = 0
section. We call (0, 0) an equilibrium point of the system of ODEs because (0, 0) sat. along the line through (0, 0) determined by V2' ~f the .mltral pom.t X(O) IS not on the
isfies the corresponding system of algebraic equations. In this section, we investigate line through (0, 0) determined by V2, then the line gIven by VI IS an asymptote for
the properties of the equilibrium point based on the eigenvalues and corresponding the solution. We say that (0, 0) is a saddle point in this case.
eigenvectors of the coefficient matrix A by viewing the phase portrait, a picture of a
set of solutions (trajectories) and equilibrium point(s) of the system. As we saw in Sec.
tion 6.4, solutions of these systems vary greatly.
Before we move on to nonlinear systems, we first investigate properties of sys.
terns of the form
o """.,'.' .
Classify the equilibrium point (0, 0) m the systems (a) y' {x' == x2y and (b) {x'y' == yx.
X' = ax + by
{ y' = ex + dy'
: 1 Solution (a) The coefficient matrix in this case is A = (~ ~), so we identify

where I; ~I = ad - bc "* 0, which has only the one equilibrium point (0,0). We have . 11 -0 A I )-
the eigenvalues of A by solvmg 2 _0 A = (1 - A)(2 - A - O. Th '
e eigen-
solved many systems of this type using the eigenvalues and corresponding eigen.
vectors of A = (~ ~) and have seen that the solutions vary greatly. values A, = I and A2 = 2 are real and unequal, so we classify (0, 0) as an unstable
The behavior of the solutions of this system and the classification of the equi-
improper node. Substitution of A, =I into (A - AI)v = 0 indicates that v I = (~)
librium point depend on the eigenvalues and corresponding eigenvectors of the system.
We more thoroughly investigate the cases that can arise in solving this system by con- is an eigenvector associated with Al = I. Similarly, we find that V2 = (n is an
sidering the classification of the equilibrium point (0, 0) based on the eigenvalues and eigenvector corresponding to A2 = 2. Therefore, a general solution is
'.
correspondmg eIgenvectors of c d ' (a b)
X(t) = cI(~)e' + c2G)e2' = (~'::,).
Real Distinct Eigenvalues
Notice that both components of the solution move away from zero as t -> 00 while
Suppose that Al and A2 are real eigenvalues of A = (; ~) where A2 < Al with both approach zero as t -> - 00. • •
There is a relationship between solutions of the system and the eIge.nlmes,

corresponding eigenvectors VI and V2, respectively. Then, a general solution of
X' = AX is lines through the equilibrium point in the direction of the eigenvecto:s. In this .case,
the eigenlines corresponding to VI and V2 are the y-axis and the X-~IS, respect.Ively.
X(t) = (;~%) = clvle A1 ' + C2v2eA" = eA1'[cIVI + CZv 2e(A,-A 1)1. To see the relationship between solutions and the eigenlines, we fmd trajectorIes of
the system by writing the system as the first-order equation
I. If both eigenvalues are negative, suppose that Az < Al < 0 so A2 - Al < O. This dy dy/dt 2y
means that e(A,-A1)1 is very small for large values of t, so X(t) = clvle A1 ' is small = dx/dt = ~.
for large values of t. If CI "*
0, then lim,_oo X(t) = 0 along the line through (0, 0)
dx
in the direction of VI. If CI = 0, then X(t) = czvze
A
". Again, because A2 < 0,
Separating variables, integrating, and simplifying gives us
limt_ oo X(t) = 0 along the line through (0, 0) in the direction of V2. In this case,
(0, 0) is a stable improper node_
y = ex 2 ,
351
350 Chapter 6 Systems of Differential Equations G.G Phase Portraits
c:
where. is an arbitrary constant, so solution curves are parabolas with vertices at
To determine the behavior of solutions as I ..... 00, notice that if CI *" 0, then X(t)
the ongm. (We can obtain the same result by eliminating the parameter from the so. becomes like CI(I)1 e' because lim e -I = O. In other words, solutions become asymp-
t-t1X>
lution for X(t) found earlier.) To sketch the phase portrait, we begin by graphing the
elgenhnes. Because each is associated with a positive eigenvalue, we place arrows totic to the eigenline associated with the positive eigenvalue y = x as t ..... -00. In a
directed away from the origin on each eigenline. Next, we graph several parabolas similar manner, we determine the behavior of solutions as I ..... -00. In this case,
(Figure 6.19). These trajectories are also directed away from the origin because of
th~ positiv.e eigenvalues. We can also graph several members of this family along
solutions are dominated by the term c2(_De-' (if C2 *" 0) because,.!.irnro e' = O.
With the directIOn field of the system, as done in Figure 6.20, to view the orienta. Therefore, solutions approach the eigenline associated with the negative eigenvalue
tion of trajectories. Based on our earlier observation, solutions become tangent to y = -x as I ..... -00. Also, if C2 = 0 and CI *"
0, then the solution approaches the
one eigenline, the x-axis, as I ..... -00 (as we move against the direction of the ar. origin along the eigenline y = x as I ..... 00. When we solve
rows) and become parallel to the other eigenline, the y-axis, as I ..... 00. (We will find dy dy/dl x
that this behavior holds in other systems that include a node.) Notice that the tra. dx = dx/dl = Y'
jectories are tangent to the eigenline associated with the smaller eigenvalue (in
absolute value). They become parallel to the eigenline associated with the larger we obtain x 2 - y2 = C, which represents a family of hyperbolas (for C 0) and *"
eigenvalue (in absolute value). the Iinesy = ::':x (for C = 0). When we sketch the phase portrait in Figure 6.21, we
(b) By solving begin by drawing the eigenlines. The line y = x is associated with a positive eigen-
value so we place arrows directed away from the origin on this line. In contrast,
-A
I[ = ,1.-? - 1=0
. y = -x corresponds to a negative eigenvalue, so the arrows are di~ected to,,":ard ~e
origin on this line. Next, we sketch hyperbolas centered at the ongm and assign on-
we find that the eigenvalues are A I = I and ,1.2 = -1. In this case, one eigenvalue entation based on that of the eigenlines (asymptotes of the hyperbolas). Observe that
is positive and one is negative, so we classify (0, 0) as a saddle point. Note that the all solutions approach the eigenlines. They become asymptotic to the eigenline
y = x associated with AI = 1 (the positive eigenvalue) as I ..... 00 an~ be:ome
eigenvectors corresponding to AI = 1 and ,1.2 = - I are VI = G) and V2 = (- n. Figure 6.21
X' = y
Phase portrait of asymptotic to the eigenline y = -x associated with ,1.2 = -1 (the negallve eigen-
respectively, so the eigenlines associated with VI and V2 are Y = x and y = -x, re- { y' = x
value) as I ..... -00. This is a property linked to other saddle points we encounter.
spectively. A general solution is.
Note: Eigenlines are trajectories of the system because each can be obtained by
setting one of the arbitrary constants c, or C2 in the general solution e~ual to zero.
Therefore, other trajectories cannot intersect the eigenlines because dOIng so would
contradict the uniqueness of solutions of IVPs involving a first-order linear homo-
y
geneous system of ODEs.
y
[3 "fi"i,INj
Classify the equilibrium point (0, 0) of the systems of differential equations
. {x' = 5x + 3y {x' = x - 2y .
x and sketch the phase portrait: (a) y' = -4x _ 3y; (b) y' = 3x - 4y'
X' = -x - 2y
(c) { y' = 3x + 4y
: .) Solution (a) The eigenvalues are found by solving

Figure 6.19 Phase portrait of Figure 6.20 Solutions with direction 5- A
X' =x 3 [= ,1.2 _ 2,1. - 3 = (A - 3)(,1. + 1) = O.
field of {x: = x? [ -4 -3 - A
{ y' = 2y y =-y
y
Because the eigenvalues Al = 3 and A2 = -I have opposite signs, (0, 0) is a y dy dyldt 3x - 4y
saddle point. ~ote that the eigenvectors corresponding to Al and A2 are v, = (-;) dx = dxldt = x - 2y ,
and we observe that trajectories have horizontal tangent lines, where dyldx = 0 (when
and V2 = (-2). respectively. To obtain an idea about the behavior of solutions of' 3x - 4y = 0). In other words, trajectories cross the line y = 3x/4 horizontally. This
this syste~, we fmd equ~tions of the eigenlines. One way to find the equation of fact and the vectors in the direction field indicate that trajectories approach the eigen-
an elgenlme IS to determIne a point in addition to (0, 0) on the line. For example line y = x as t --> 00 and are parallel to y = 3xl2 as t --> -00. (Remember that tra-
for v, = (-3)2' another point is (- 3, 2), the terminal point of the vector, so the
, jectories cannot cross eigenlines.) We also recall from Example 1 that trajectories
are tangent to the eigenline associated with the smaller eigenvalue (in absolute value),
eigenline corresponding to v, has slope m = (2 - 0)/(-3 - 0) = -2/3 and y = x, and becomes parallel (as t --> ±oo) to the eigenline associated with the larger
f - 2x/3 S' . '" equa- eigenvalue (in absolute value),y = 3x12. We graph the phase portrait in Figure 6.23.
Ion y - - .. u:mlru:ly, the elgenlme assocIated with V2 is y = -2x. We sketch
Figure 6.22 Phase portrait for The line y = 3x/4 is dashed. Both eigenlines have associated with them arrows di-
Example 2, solution (a).
th~ phase p~rtralt In F:gure 6.22. First, we draw the eigenlines, y = - 2x/3 and Figure 6.23 Phase portrait for
y - - 2x, WIth arrows dIrected toward and away from the origin, respectively. Th Example 2, solution (b). rected toward the origin because both eigenvalues are negative.
we draw. curves .using the eigenlines as asymptotes and follow the directions as:: (e) Because
c.Iated ~th the ~Igenlmes to ~sign the proper orientation to the trajectories. As men-
honed ~ a p~eVlous e~ample .Involving a saddle point, solutions become asymptotic -1 - A -2 I = A2 - 3A +2= (A - 2)(A - 1) = 0,
3 4-A
to the el~enlIne assocIated wIth the;positive eigenvalue as t --> 00 and they become 1
asymptotrc to that assocIated with the negative eigenvalue as t --> _00. the eigenvalues AI = 2 and A2 = 1 are real and unequal. Therefore, (0, 0) is an
For another interpretation of the behavior of solutions as t --> 00, notice that
in a general solution of this system, X(t) = CI( -;)e 31 + -De-"
C2( if c, =f. 0, then
improper node. Note that corresponding eigenvectors are VI = ( _;) and V2 = ( - D,
respectively. The solutions become unbounded along the lines through (0, 0) deter-
XU) becomes like cI(-;)e31 because E,r;! e- I= O. In other words, solutions mined by these vectors, y = -~ and y = -x. In this case, the eigenlines are
y = - 3xl2 and y = -x. We graph these lines in Figure 6.24 and place arrows on
become asymptotic to the eigenline associated with v, = (-;) as t --> 00. In a them directed away from the origin due to the positive eigenvalues. Based on our
earlier findings, trajectories are tangent to y = -x as t --> -00 and become parallel
sImIlar manner, we can determine the behavior of solutions as t --> -00. In this case,
Figure 6.24 Phase portrait for to y = - 3xl2 as t --> 00. Following a method similar to that in (b), we see that tra-
solutions are dominated by the te~ c2(_I)e-' (if Cz =f. 0) because lim e31 = O. Example 2, solution (c). jectories cross the line y = - 3x/4 horizontally (this line is dashed in Figure 6.24).
2 t-:,-o:;l
Therefore, solutions approach the eigenline associated with V2 = (-D (or

Repeated Eigenvalu es
A~ = -I), y = -2x, as t --> Notice that the behavior of the solutions agrees
-00.
WIth what we observed in our earlier discussion. We recall from our previous experience with repeated eigenvalues of a 2 X 2 system
(b) Because the characteristic equation is that the eigenvalue can have two linearly independent eigenvectors associated with it
or only one (linearly independent) eigenvector associated with it. We investigate the
1- A -2 / behavior of solutions in the case of repeated eigenvalues by considering both of these
/
3 -4 _ A = A2 + 3A + 2 = (A + 1)(A + 2) = 0, possibilities.
the eigenvalues AI = -1 and A2 = - 2 are distinct and both negative; (0, 0) is a I. If the eigenvalue A = A, = A2 has two corresponding linearly independent eigen-
vectors v, and V2, a general solution is
stable ~proper node. In this case, corresponding eigenvectors are VI = G) and XU) = c,vle At + C2v2eAI = (c,v, + c2v2)eAt.
V2 = (3). so the solutions approach (0, 0) along the lines through (0, 0) in the di-
If A > 0, then XU) becomes unbounded along the lines through (0, 0) determined
rection of these vectors, y = x and y = ~. Based on our earlier findings concern- by the vectors c,v, + C2V2, where c, and C2 are arbitrary constants. In this case, we
mg nodes, we know that trajectories are tangent to one eigenline and become par- call the equilibrium point an unstable star node. However, if A < 0, then XU) ap-
allel to the other. To better understand the shape of the phase portrait, we convert proaches (0, 0) along these lines, and we call (0, 0) a stable star node_ Note that
the system to the first-order equation the name "star" was selected due to the shape of the solutions.
......
355
354 Chapter G Systems of Differential Equations G.G phase Portraits
2. If A = Al = ~2 h~s only one corresponding (linearly independent) eigenvecto

general solutIOn IS r Vlo a
two linearly independent vectors such as v, = (~) and V2 = (~). we obtain two
linearly independent eigenvectors corresponding to A, = A2 = 2. (Note: The choice
X(/) = clvleA, + C2[VIt + w2le At = (CIVI + c2w2)eA, + C2vlteAt of these two vectors does not change the value of the solution, because of the form
where (A - AI)W2 = VI' If we write this solution as of the general solution in this case.) Because A = 2 > 0, we classifY (0, 0) as a de-
generate unstable star node. A general solution is X(t) = CI(~)e2' + C2(~)e2' =
X(t) = teAt[+(CIV I + C2W2) + C2V I ].
we can more easily investigate the behavior of this solution. If A < 0, theu (C2Cle~:),
e
so if we eliminate the parameter, we obtainy = C2xlc" Therefore, the tra-
lim,~oo teA' -- 0and\'Im,_oo[It ( CIVI + C2W2) + C2V I ]= C2VI' The solutions ap.
jectories of this system are lines passing through the origin. In Figure 6.26, we graph
the eigenlines (the x- andy-axes) as well as several trajectories. Because of the pos-
proach (0, O! along the Ime through (0, 0) determined by v J, and we call (0, 0) a Figure 6.26 Phase portrait for itive eigenvalue, we associate with each an arrow directed away from the origin.
stable defiCient node. If A > 0, the solutions become unbounded along \hi \' Example 3, solution (b). L.
and we say that (0, 0) is an unstable deficient node. s me,
Complex Conjugate Eigenvalues
o I.fi,,','.' We have seen that if the eigenvalues of the system of differential equations are AI =
'" + {3i and Az = '" - (3i with corresponding eigenvectors v, = a + bi and V2 =
a - bi, then two linearly independent solutions of the system are
Xit) = ea'(cos {3t b + sin (3t a).
ClassifY the equilibriwn point (0, 0) in the systems: (a) {x;Y :- x-x+ -9y5y; (b) {x'y' == 2y'2x and
A general solution is X(!) = CIX,(t) + C2Xz(t), so there are constants A" A2, BJ, and
: J Solution (a) The eigenvalues are found by solving B2 such that x and yare given by
at
X(t») (Aleat cos (3t + A2e sin (3t)
1- A X(t) = ( yet) = B,eat cos {3t + B 2e at sin (3t .
-5 9_ AI = A2
' + 4A + 4 = (A + 2)2 = O.
1
-I
I. If", = 0, a general solution is
Hence, Al = A2 = -2. In this case, an eigenvector v, = (Xl) satisfies X(t) = (x(t») = (A I cos (3t + A2 sin (3t).
(-13 -39)(XI) _(0)0' which . .eqUIvalent

. to (I0 3)(X) (0)0' soY,there is only y(t) B I cos {3t + B2 sin (3t
YI - IS 0 Y: =
Both x and yare periodic. In fact, if A2 = BI = 0, then
one corresponding (linearly independent) eigenvector v (-3Y,Y (-3)I Y" B.e I
= I
) =
X(t) = (X(t») = (AI c~s (3t).
y _ A = -2 < 0 ' (0 '. 0) .I'S a. d egenerate stabl e nod e. I
cause ' case, the elgenlme
n thiS . . IS. y(t) B2 sm {3t
Y - - x13. We graph this Ime m Figure 6.25 and direct the arrows toward the origin
In rectangular coordinates this solution is
because of ~e negative eigenvalue. Next, we sketch trajectories that become tan·
x2 y2
gent. to the elgenhne as I -> co and associate with each arrows directed toward the
ongm. A7 + B/ = 1
x (b) Solving the characteristic equation where the graph is either a circle or an ellipse centered at (0, 0) depending on the
value of A I and B 2 • Hence, (0, 0) is classified as a center. Note that the motion
2- A 0 1 around these circles or ellipses is either clockwise or counterclockwise for all so-
o 2 _ A = (2 - Ai = 0,
1 lutions.
-6 2. If '" *- 0, eO't is present in the solution. The eat term causes the solution to spiral
we have A, = A2 = 2. However, because an eigenvector VI = (Xl) satisfies the around the equilibrium point. If", > 0, the solution spirals away from (0, 0), so we
(~ ~)(~:) (~). any nonzero choice of v, is an eigenve::or. If we select

Figure 6.25 Phase portrait for classifY (0, 0) as an unstable spiral point. If '" < 0, the solution spirals toward
Example 3, solution (a). system = (0, 0), so we say that (0, 0) is a stable spiral point.
o "fI"i,'.' Stability
We now understand the behavior of solutions of systems of linear equations based on
Classify the equilibrium point (0,0) in each of the following systems: (a) {x: -yo
'= the eigenvalues of the coefficient matrix. In addition, we would like to comment on
y =x ' the stability of the equilibrium point of the system using the same information. Look-
(b) x' = x - 5y.
{ y' =x - 3y ing back, we classified equilibrium points as saddle points, nodes, spirals, or centers.
In the case of saddle points, some nodes, and some spiral points in which at least one
eigenvalue is positive or has a positive real part, trajectories move away from the equi-
i J Solution (a) The eigenvalues are found by solving
librium point as t --> 00. (Note: In the case of a saddle point, all solutions except the
-1/
-A =,\.2+1=0 .
eigenline associated with the negative eigenvalue move away from the equilibrium point
as t --> 00.) Each of these classes of equilibrium points is unstable. Then, when we con-
sider nodes and spiral points in which both eigenvalues are negative or have a nega-
Hence, A = ::!:.i. Because these eigenvalues have a zero real part (that is they tive real part, solutions converge to the equilibrium point as t --> 00. These equilibrium
purely imaginary), (0, 0) is a center. We can view solutions of this ;ystem ~ points are asymptotically stable because of this convergence. Finally, centers are sta-
.. th ~ ~ ~
ble if a solution that starts sufficiently close to the equilibrium point remains close to
wntmg e system as a first-order equation with - = - - = - Separaf the equilibrium point. Note: An equilibrium point that is not stable is considered un-
. dx dxldt y' mg
y stable. We summarize our findings in Table 6.1.
van~bles, we have y dy = -x dx:'so that integration yields /12 = -x 2 /2 + C
or x + y2 = K (where K = 2C). Therefore, solutions of the system are circles (if
K> 0) c?nte~ed at the origin. The solution is (0, 0) if K = O. Several solutions are TABLE 6.1 Classification of Equilibrium Point in linear System
graphed m Figure 6.27(a). The arrows indicate that the solutions move counter-
c~o~kwlse aro~d (0, 0). We. see this by observing the equations in the system, Eigenvalues Geometry Stability
~ -:- -y and ~ = x. In t~e first quadrant where x ~ 0 ~d y > 0, these equations At. A2 real; A, > A2 > 0 Improper node Unstable
mdlcate th~t x < 0 and y > O. Therefore, on solutIOns In the first quadrant, x de-
creases while y Increases. Similarly, in the second quadrant where x < 0 and y > 0 At. A2 real; A, = A2 > 0; 1 eigenvector Deficient node Unstable
x' < 0 and y' < 0, so x and y both decrease in this quadrant. (Similar observatio~ A,. A2 real; A, = A2 > 0; 2 eigenvectors Star node Unstable
are made in the other two quadrants.) .
Ca) (b) Because the characteristic equation is A" A2 real; A2 < A, < 0 Improper node Asymptotically stable
Figure 6.27 (a) Phase portrait At. A2 real; A, = A2 < 0; 1 eigenvector Deficient node Asymptotically stable
1- A -5 / .
for Example 4, solution (a). 1 -3 _ A = A2 + 2A + 2 = 0,
/ At. A2 real; A, = A2 < 0; 2 eigenvectors Star node Asymptotically stable
At. A2 real; A2 < 0 < A, Saddle point Unstable
th~ eigenvalues are A '= (-2 ± Y4=8)/2 = -1 ± i. Thus, (0, 0) is a stable spiral
pomt because a = -1 < O. Al = a + f3i, A2 = a - f3i, f3 *- 0, a > 0 Spiral point Unstable
. To better understand the behavior of solutions, consider a general solution of
this system, A, = a + f3i, A2 = a - f3i, f3 *- 0, a < 0 Spiral point Asymptotically stable
A, = f3i, A2 = -f3i, f3*- 0 Center Stable
X(t) = e-'(-CI ~os t - 2cI sin t + 5C2 sin t).
CI Sin t + C2 cos t - 2C2 sin t
The trigonometric functions cause solutions to rotate about the origin, while e-' To better understand the meaning of the terms unstable, asymptotically stable,
forces. both compone?ts to zero as t ~ 00. In Figure 6.27(b), we graph the phase and stable, we take another look at graphs of solutions to several systems in the ex-
portrait. Solutions spITal in toward the origin. ample problems. Instead of graphing these solutions in the phase plane, however, we
Cb) graph them in the tx- and ty-planes. Consider the system in Example 2 that included a
saddle point with general solution X(t) = cl(-~)e3t + c2(_De- . When we graph
t
Figure 6.27(b) Phase portrait Note: In the case of centers and spiral points, the direction field may be particu·
for Example 4, solution (b). larly useful In determining the shape of solutions as well as their orientation. x(t) = - 3cle3t + C2e -t and y(t) = 2cle3' - 2c2e -t for various values of CI and C2 in
-----. - ._---------
358 Chapter b Systems of Differential Equations b_b Phase Portraits 359
Figures 6.28(a) and 6.28(b), we see that all solutions diverge from xCt) = 0 and
yet) = 0, which are the coordinates of the equilibriwn point of this system. We there_
fore call the equilibriwn point in this case unstable. In contrast, in Figures 6.29(a) and
... EXERCISES 6.6
6.29(b), when we graph xCt) = e -t( -Cl cos t - 2Cl sin t + 5C2 sin t) and yet) '"
t
e- (cl sin t + C2 cos t - 2C2 sin t), solutions in Example 4(b) that involved a stable In Exercises 1-12, classify the equilibrium point (0, 0) of the In Exercises 17-20, use the given eigenvalues to (a) develop a
spiral point, we see that all solutions asymptotically approach xCt) = 0 and yet) '" O. system. corresponding second-order linear equation; (b) develop cor- a
Therefore, we classifY the equilibrium point in a system possessing this property as responding system of first-order linear equations; (e) classify
X' = -5x - 4y X' = -x - 7y
asymptotically stable. Finally, when we graph solutions to the system in Example 4(a) the stability of the equilibrium point (0, 0).
1. { y' = _ 9x - 5y 2. { y' = _ 3x + 3y
containing the center x(t) = Cl cos t + Cz sin t and yet) = -Cl sin t + C2 cos t in 17.A I =2,A 2 =-I
X'= 8x - 4y X' = 4x - 2y
Figures 6.30(a) and 6.30(b), we notice that solutions neither converge to nor diverge
*3_ { y'=9x-4y 4. { y' = -6x + 8y 18.)'1 = -3,A 2 =-1
from x(t) = 0 and yCt) = O. Hence, the equilibrium point is stable.
= -lOx - 2y
X' X' = 6y *19. AI = -3, A2 = -3
y 5. { y' = -2x - lOy 6. { y' = -3x - 9y
y 20. AI = 3i, A2 = -3i
*7. {x'y' =-y
-x=
8. {;::;; 21. (Linear Nonbomogeneous Systems) Consider the
X'=X-I
X' = -9y X' = -8x + 5y system { y' = X + y'
-I 9. { y' = 6x + 4y 10. { y' = -2x - 2y (a) Show that (I, -I) is an equilibrium point.
-2
X' = -3x + 5y = -4x - 4y
X' (b) Show that with the change of variables u = x-I
*11. { y' = -lOx + 3y 12. { y' = 8x and v = y + I, we can transfonn the system to
0.5 x Figure 6.28(a)
-I -3 Figure 6_28(b) 13. (Linear Systems with Zero Eigenvalues) u' = u . (Notice that if the equilibrium point
Graphs of x(t) = Graphs ofY(I) = { v' = u+v
- 3CI e + cze -to
3l
2Cl e
3r - 2c2e -t, (a) Show that the eigenvalues of the system is (xo, Yo), then the change of variables is u =
Xi = -2x
are AI = 0 and A2 = -2. X - Xo and v = y - Yo·)
y y { y' = -4x
(e) Classify the equilibrium point (I, -I) of the orig-
(b) Show that all points on the y-axis are equilibriwn inal system by classifying the equilibrium point
points. (0, 0) of the transfonned system.
(e) By solving dyidx = -4xi-2x = 2, show that the (d) Sketch the phase portrait of the original system
trajectories are the lines y = 2x + C, where C is by translating that of the transfonned system.
arbitrary.
4 6 x (d) Show that a general solution of the system is For each of the following systems, follow steps similar to those
4 6 x Figure 6.29(a) -I
Figure 6_29(b) X(I) = (
2 -2')'
+e e2 -
2t
in Exercise 21 to locate the equilibriwn point of the system,
Graphs of x(t) = Graphs of y (t) = Cl e2 e classify the equilibrium, and sketch the phase portrait.
e -'( -CI cos t - 2Cl -2
e -'(el sin I + e2 cos (e) Find lim X(t).
sin f + 5e2 sin f). t - 2e2 sin t). X' = 2x + Y X' = -2x + y - 6
(f) Use ;;ooinfonnation in (c) and (e) to sketch the 22. { y' = _y + 6 *23 {
.. y' =x - 2y
phase portrait of the system.
4
X' =x - 2y - I
4 24. { y' = 5x - y - 5
Follow steps similar to those in Exercise 13 to sketch the phase
2 portrait of each of the following systems.
X' = ax + by
25_ Consider the system { y' = ex + dy' where a, b, e, and
x
= -2x - Y
X'
14. { y' = -4x - 2y d are real constants. Show that the eigenValues of
-2 -2 this system are found by solving A2 - (a + d)A +
= 2x
X'
-4 -4 15. { y' = -4x (ad - be) = O. If we let p = a + d and q = ad - be,
Figure 6.30(a) Figure 6_30(b) show that the eigenvalues are A'.2 = l/2(p ± v'il:)
Ca) Graphs of x(t) = CI
Graphs of yet) = X' = 12x - 6y where!!. = p2 - 4q. Show that each of the following
cos t + C2 sin t. Cb)
-CI sin t + Cz CDS t. 16. { y' = 18x - 9y is true of the equilibrium point.
,~ ,
360 Chapter 6 Systems of Differential Equations 6.7 Nonlinear Systems 361
(3) (0, 0) is a node if q > 0 and A. 2: O. It is asymp-
!
totically stable if p < 0 and unstable if p > O. ~= -k,x can quickly show that the linear system has a saddle point at (0, 0). We show several
trajectories of this system together with its direction field in Figure 6.32. Next, in Fig-
(b) (0, 0) is a saddle point if q < O. dy ure 6.33(a), we graph several trajectories of the nonlinear system along with its direc-
(c) (0, 0) is a spiral point if p *
0 and A. < O. It is
dt = k,x - k2 y
tion field. We see that the nonlinear term affects the behavior of the trajectories. How-
asymptotically stable if p < 0 and unstable if ever, the behavior near (0, 0) is "saddle-like" in that solutions appear to approach the
p>O. is solved t? .de:ermine x(t) and y\t). Show that (0; 0)
,s the equtltbnum solutIon of thiS system. Find the origin but eventually move away from it. When we zoom in on the origin in Figure
(d) (0, 0) is a center ifp = 0 and q > O. eigenvalues of the system and classify the equilibrium 6.33(b), we see how much it resembles a saddle point. Notice that the curves that sep-
(e) Place the classifications in parts (a)-(d) in the re- solution. (Note: k, and k2 are positive constants.) Also arate the trajectories resemble the linesy = x andy = -x, the asymptotes oftrajecto-
gions shown in Figure 6.3l. The parabola is the solve the system. Find limr-.oo x(t) and lim,_ yet). D~ ries in the linear system, near the origin. These curves, along with the origin, form the
graph of A. = p2 - 4q = O. these limits correspond to the physical situation? separatrix, because solutions within the portion of the separatrix to the left of the
27. In the circular chemical reaction, compounds A, and y-axis behave differently than solutions in other regions of the xy-plane. In this sec-
A2 combine at rate kA to produce compound B,. This tion, we discuss how we classify equilibrium points of nonlinear systems by using an
compound reacts with compound B2 at rate ka to form associated linear system in much the same way as we analyzed these two systems.
C" which in turn combines with compound C2 at rate When working with nonlinear systems, we can often gain a great deal of infor-
kc to form A ,. Therefore, the system of equations mation concerning the system by making a linear approximation near each equilibrium
point of the nonlinear system and solving the linear system. Although the solution to
p
~' = keG,C2 - kA A,A 2 the linearized system only approximates the solution to the nonlinear system, the gen-
Figure 6.31 Graphs of A. = p2 - 4q = O. d!, kA,A 2 - kaB ,B2

= A
eral behavior of solutions to the nonlinear system near each equilibrium point is the
same as that of the corresponding linear system in most cases. The first step toward
26. Suppose that substance X decays into substance Y at

rate k" which in turn decays into another substance at
1
dC,
dt = kaB ,B2 - kCC,C2
models this situation. Let K, = kAA2' K2 = koB2, and

approximating a nonlinear system near each equilibrium point is to find the equilib-
rium points of the system and to linearize the system at each of these points.
Recall from multivariable calculus that if z = FC;c, y) is a differentiable function,
rate k2 • If x(t) and y (t) represent the amount of X and the tangent plane to the surface S given by the graph of z = F(x, y) at the point (xo, Yo)
K J = keG2, and determine the equilibrium solutions
Y, respectively, then the system of the system. is
z = FxCxo, Yo)(x - xc) + FyC;co. Yo)(y - Yo) + F(xo, Yo)·
Near each equilibrium point (xo, Yo) of the nonlinear system
~ Nonlinear Systems
We now turn our attention to systems in which at least one of the equations is not lin. y y
ear. These nonlinear systems possess many of the same properties as linear systems.
In fact, we rely on our study of the classification of equilibrium points of linear sys.
terns to assist us in our understanding of the behavior of nonlinear systems. For ex-
ample, consider the system
X' = y
{ y' = X + x2
2
which is nonlinear because of the x term in the second equation. If we remove this
nonlinear term, we obtain the linear system
X' =y
{ y' =x
Figure 6.32 Trajectories of Figure 6.33 (a) Trajectories of Figure 6.33(b) Phase portrait
This system has an equilibrium point at (0, 0), which is also an equilibrium point of of nonlinear system near the
corresponding linear system nonlinear system with direction
the system of nonlinear equations. Using the techniques discussed in Section 6.6, we with direction field. field. origin.
X' = f(x,y) @ If (xo, Yo) is classified as a center in the associated linear system, (xo, Yo) may
{ y' = g(x,y)' be a center, unstable spiral point, or asymptotically stable spiral point in the non-
linear system, so we cannot classify (xo, Yo) in this situation (see Exercise 28).
the system can be approximated with ® If the eigenvalues of J(xo, Yo) are real and equal, then (xo, Yo) may be a node or
X: = fx(xo, Yo)(X - Xo) +h(xo, Yo)(Y - Yo) + f(xo, Yo) a spiral point in the nonlinear system. If A, = ..1.2 < 0, then (xo, Yo) is asymptoti-
{ y = gx(xo, Yo)(x - xo) + gy(xo, Yo)(Y - Yo) + g(xo, Yo) cally stable. If A, = ..1.2 > 0, (xo, Yo) is unstable.
where we have used the tangent plane to approximate f and g in the two differential These findings are summarized in Table 6.2.
equations. Because f(xo, Yo) = 0 and g (xo, Yo) = 0 (why?), the approximate system is
x: = fx(xo, Yo)(x - xo) + h(xo, Yo)(Y - Yo) . TABLE 62 Classification of Equilibrium Point in Nonlinear System
{ y = gxCxo, Yo)(x - xo) + gy(xo, Yo)( y - Yo) I Geometry
Eigenvalues of }(xo, Yo) I Stability
which can be written in matrix fonn as A" A2 real; Al > A2 > 0 Improper node Unstable
I
x:) = (fx(xo, Yo) h(xo, Yo) )(X - xo). A" A2 real; Al = A2 > 0 I
Node or spiral point Unstable
(y \g-x(xo,,JIo) gy(xo, Yo) y - Yo A" A2 real; A2 < A, < 0 Improper node Asymptotically stable
Note that we often call this system the linearized system corresponding to the nOD- A" A2 real; A, = A2 < 0 Node or spiral point Asymptotically stable
!
linear system or the associated linearized system due to the fact that we have re-
moved the nonlinear tenns from the original system. A" A2 real; A2 < 0 < A, Saddle point Unstable
The equilibrium point (xo, Yo) of the system {x: = f«X, y)) is classified by the
y = g x,y
A, = a + {3i, A2 = a - {3i, {3 * 0, a> 0 ! Spiral point Unstable
eigenvalues of the matrix A, = a + {3i, A2 = a - {3i, {3 * 0, a < 0 Spiral point Aymptotically stable
J( ) - (fx(xo, Yo) h(xo, Yo) )

A, = {3i, A2 = -{3i, {3 *" 0 Center or spiral point Inconclusive
Xo, Yo - \g-x(xo, Yo) gy(xo, Yo) ,
Notice that the linearization
must be carried out for each
equilibrium point.
which is called the Jacobian matrix. After determining the Jacobian matrix for each
equilibrium point, we find the eigenvalue of the matrix in order to classify the corre-
sponding equilibrium point according to the following criteria.
o I'''"....'
Find and classify the equilibrium· points of {< = 1 2- Y 2'
Y =X - Y
Classification of Equilibrium Points of a Nonlinear System ) Solution We begin by finding the equilibrium points of this nonlinear system by
Let (xo, Yo) be an equilibrium point of the system {x: = f(X( ,y)) and let A, and .\2 solving { 21 - ~ = O. Because y = 1 from the first equation, substitution into the
y = g x,y x - y = 0
be the eigenvalues of the matrix second equation yieldsx 2 - 1 = O. Therefore, x = :!:: I, so the two equilibrium points
are (1, 1) and (-I, 1). Because f(x, y) = I - Y and g(x, y) = x 2 - y2,
(fx(xo, Yo) h(xo, Yo) ) fx(x, y) = 0, hex, y) = -1, gxCx, y) = 2x, and gy(x, y) = -2y, so the Jacobian
\gxCxo, Yo) gy(xo, Yo)
matrix is J(x, y) = (~ ~;y). Next, we classify each equilibrium point by finding
of the associated linearized system about the equilibrium point.
the eigenvalues of the Jacobian matrix of each linearized system.
CD If (xo, Yo) is classified as an asymptotically stable or unstable improper node
(because the eigenvalues of J(xo, Yo) are real and distinct), a saddle point, or an
For (1, 1), we obtain the Jacobian matrix J(I, 1) = (~ =;) with eigen-
asymptotically stable or unstable spiral point in the associated linear system,

(xo, Yo) has the same classification in the nonlinear system.
values that satisfy 1-;..1. _;~ AI = ,12 + 2..1. + 2 = O. Hence, Al = -1 + i and
A2 = -I - i. Because these eigenvalues are complex-valued with negative real Part, . The Jacobian matrix is
we classify (1, I) as an asymptotically stable spiral in the associated linearized sYs-
tem. Therefore, (I, 1) is an asymptotically stable spiral in the nonlinear system.
J(x, y) = (
7- 2x_y- 2y -2x)
5 - x - 2y .
For (-I, I), we obtain J(-I, I) = (_~ =D. In this case, the eigenvalues
We classify each of the equilibrium points (xo, Yo) of the associated linearized sys-
-A
are solutions of -2 -2-1 I
_ A = A2 + 2A - 2 = O. Thus, Al = -2 +2 2v3
_" tem using the eigenvalues of J(xo, Yo):
I
= (~ ~} Al = 7, A2 =
_ r;; -2 - 2V3 _ r;;
-1+ v3>OandA2 = 2 =-1- v3<0,so(-1,I)isasaddlepoint J(O, 0) 5; (0, 0) is an unstable node.
in the associated linearized system and this classification carries over to the non.
-3
linear system. In Figure 6.34, we graph solutions to this nonlinear system approxi. J(O, 5) = ( -5 _ 0); Al = -3, A2 = -5; (0, 5) is an asymptotically stable
Figure 6.34 mated with the use of a computer algebra system. We can see how the Solutions Figure 6.35 5 improper node.
move toward and away from the equilibrium points by observing the arrows on the
-7
L
vectors in the direction field. J(7, 0) = ( 0 - ~~} Al = -2, A2 = -7; (7, 0) is an asymptotically stable
Improper node.
Note: In Example 1, the linear approximation about (1, 1) is -3
J(3, 2) = ( -2 =~); Al = 1, A2 = -6; (3, 2) is a saddle point.
-1)(X -1) = (-y + 1)
(YX:) = (02 -2 or {
X' = -y + 1 .
Y- 1 2x - 2y y' = 2x - 2y In each case, the classification carries over to the nonlinear system. In Figure 6.35,
Notice that this (nonhomogeneous) linear system has an eqUilibrium point at (1,1), we graph several approximate solutions and the direction field to this nonlinear sys-
With the change of variable u = X - 1 and v = y - 1, where the change of variable tem through the use of a computer algebra -system. Notice the behavior near each
for the equilibrium point (xo, Yo) is u = X - Xo and v = y - Yo, we can transform this equilibrium point.
[u' -v
system to v' == 2u - 2v with equilibrium point (0, 0). Finding the eigenvalues of the
Identify the eqUilibrium points in Figure 6,35,
matrix of coefficients indicates that (0, '0) is a saddle point, and we can sketch the
phase portrait of the linearized system by translating the axes back to the original
variables.
"f'"I,'.
o "i',jI,'.j Investigate the stability of the equilibrium point (0, 0) of the nonlinear system
f x' = y + ~(X2 + y2) .

' d d I 'fy th
Fm 'I'b . , f {x' = x(7 - x - 2y)
an c asSI e equII numpomts 0 y' = y(5 -x _ y) .
1y' = -x + f(x 2
+ y2)
:, .IS' .
o.ution The eqUl' l i b
umrpomts
i' 0
f thoIS system satls
, fy {X(7
y(5-_x x- _2y) =0
y) = O·
: ] Solution First, we find the Jacobian matrix,
Thus, {x = 0 or 7 - x - 2y = O} and {y = 0 or 5 - x - y = O}. If x = 0, then
y(5 - y) = 0 so y = 0 or y = 5, and we obtain the equilibrium points (0, 0) and
_ (-IX2 + +y2 I + xy )
(0, 5). If y = 0, then x(7 - x) = 0, which indicates that x = 0 or x = 7, The corre- J(x, y) - I 3'
sponding equilibrium points are (0, 0) (which we found earlier) and (7,0). The other -l+xy _X 2 +_y2
possibility that leads to an equilibrium point is the solution to {7 5-: }~
x- : ~
Then, at the equilibrium point (0, 0), we have J(O, 0) = ( _ ~
2 2
b)' so the linear

which is x = 3 and y = 2, The equilibrium point associated with this possibility is
(3,2). approximation is
36& Chapter & Systems of Differential Equations 6.7 Nonlinear Systems 367
Group A Group B
X' = y X' = 1 - xy X' = y - ~
{ y' = - X *19. { y' = y _ x3 20. 2 X' = y2 + 2xy
{
y' = 1 - xy 24. { y' = x 2 - 4x - 5
with eigenvalues Al = i and A2 = -i. Therefore, (0, 0) is a (stable) center in the lin. 21. (population) Suppose that we consider the relation-
earized system. However, when we graph the direction field for the original (non. ship between a host population and a parasite popula-
linear) system in Figure 6.36, we observe that (0, 0) is not a center. Instead, trajec. tion that is modeled by the nonlinear system ·I--.-l-~-+ r - I', I
:: ".2 4 '6 8 __ lOX
tories appear to spiral away from (0, 0), so (0, 0) is an unstable spiral point of the Jx' = (al - blx - elY)x
-2t
nonlinear system. The nonlinear terms not only affect the classification of the equi. \y' = (-a2 + C2 X )Y ,
-4~'
librium point but also change the stability. Note: This is the only case in which We
cannot assign the same classification to the equilibrium point in the nonlinear sys.
where x(t) represents the number in the host popula-
tion at time t, y (t) the number in the parasite popula-
i
tern as we do in the associated linear system. When an equilibrium point is a center tion at time t, al the growth rate of the host popula- X' = -5xy _ y3
in the associated linear system, then we cannot draw any conclusions concerning its tion, a2 the starvation rate of the parasite popUlation, *25. { y' = 3xy + Y + 2x
Ce)
classification in the nonlinear system. and b l the interference when the host population over-
populates. The constants CI and C2 relate the interac-
y tions between the two populations leading to decay in
-4 -2 4 x
the host population and growth in the parasite popu-
2 lation, respectively. Find and classify the equilibrium -2
point(s) of the system if all constants in the system
are positive.
22. (Economics) Let x(t) represent the income of a com-
x pany andy(t) the amount of consumer spending. Also X' = -3y - 3xy fY
-2 -I Cd) 1.5
suppose that z represents the rate of company expen- 26. { y' = 5y2 - X +5 i
-I
Figure 6.36 Direction field for
ditures. The system of nonlinear equations that mod- 1"i
els this situation is o.st
-2
X' = y + ~(x2 + y2)
' =x- ay -=i.5-~~~-O'5 ~1--~~i~r-~5i
. { y' = -X: 1'(x + 2 y2) t , = b(x - y - z)
, where I < a <
(a) If z = Zo is constant, fmd and classify the equi-

00, and I :5 b.
.
. ,
.
-0.5
-I
.
..
__ 1.5.L '.
librium point ofthe system. Also consider the spe- I
cial case if b = 1.
(b) If the expenditure depends on income according ce)1.5t'
to the relationship z = Zo + ex (c > 0), find and I
classify the equilibrium points (if they exist) of

.1+
"
EXERCISES 6.7 the system. : O.SJ.
-::i:::t.5~:='i-.:t5~-I~-o.'.s. -.~- x
In Exercises 23-27, match each of the following systems in
Group A with its direction field in Group B.
-oA .
- ,~-J.'+
In Exercises 1-20, classify the equilibrium points of the non· X' = -2x + x 2 - xy X' = lOx - 2x 2 - 2xy
linear system. 9. { y' = -4y + y2 + xy 10. { y' = -4y + 2xy
Group A Group B
'~L~t >
X' = 4y2 + 3x +2
X' = Y
2. {< = x +Y
* 11. {x'Y , =y-3y - xy -
=
2 2
4 =y
X'
12. { y' = sin X
23. { y' = 4x2 - 4 Ca) sf
, 28. Consider the nonlinear system
1. { y' = x _ x2 Y =x +xy -X 2.5+
{x'y' =x {X' =x 2- + 16 {X' = --+--I-d--+--+---+-- dx = y + fJX(x 2 + y2)

~ + y2
*3 = y + y2 13. lOx 14 x(3 - y) -7.5 -,5 -2.5 2.5 5 7.SX
4. {;:: y' = y+ 1 . y'=x+y+1 dt
. +Y -2.5! -
2
5.
X'
{ y'
=Y
= -x +xy 6. {
X' = 4y
y' = -4x + y3
*15
.
{x'y' = y -+ y -
= X
X
2xy
16. {<=I-X
y =y
X' = 2x + y2
-5+
-75+
.-10+
1 dy
dt
,.
- = -x + FLY(X- + y2)
(a) Show that this system has the equilibrium point

*7.
X' = x + Y + y2 y + xy
X' = = xyX'
-12.5+
{ y' =y 8. { y'=x+y 17. { y' = x2 + y2 _ 4 18. { y' = -2x + 4y (0, 0) and the linearized system about this point
6.8 Numerical Methods 369
°
J: ~ ~ ~
librium point (b) If < r < 1, a> 0, and b > 0
show that the system is asymptotically stable a;
Numerical Methods
is x. Also, show that (0, 0) is classified
l dt
as a center of the linearized system.
(0,0,0).
*31. An exciting system to explore is the system
(" Euler's Method (" Runge-Kutta Method (' Computer Algebra Systems
Jx' = }.LX + Y - x(x 2 + y2) Because it may be difficult or even impossible to construct an explicit solution to some
(b) Consider the change of variables to polar coordi- systems of differential equations, we now turn our attention to some numerical meth-
jy' = J.Ly - x - y(x2 + y2)
nates x = r cos 0 and y = r sin O. Use these equa- ods that are used to construct solutions to systems of differential equations.
tions with the chain rule to show that dxldt = drldt for fixed values of J.L.
cos 0 - r dOldt sin 0 and dyldt = drldt sin 0 + (a) Show that the only equilibrium point of the sys-
r dOldt cos O. Use elimination with these equa- tem is (0, 0). Euler's Method
tions to show that drldt = dxldt cos 0 + dyldt sin 0
and r dOldt = -dxldt sin 0 + dyldt cos O. Trans- (b) Show that the eigenvalues of the Jacobian matrix Euler's method for approximation, which was discussed for first-order equations, can
form the original system to polar coordinates and ( '::-1 ~) are J.L ± i and classify the equilibrium be extended to include systems of first-order equations. The initial-value problem
substitute the equations that result in the equations
~
I
point (0, 0).
for drldt and r dOldt to obtain the system in polar = l(t, x, y)
coordinates (c) How do you think the direction field of the sys.
tern changes as J.L goes from - I to I? dy
Yt = get, x, y)
dr = J.Lr'
dt (d) Graph the direction field and various solutions of
x(to) = xo, y(to) = Yo
1
the system for several values of J.L between -1
dO = -I' and!.
dt is approximated at each step by the recursive relationship based on the Taylor expan-
(e) How do your results in (d) differ from your pre. sion of x and y up to order h,
(e) What does the equation dOldt = -I indicate dictions in (c)?
about the rotation of solutions to the system? Ac- Xn+l : Xn + hl(tm Xm Yn),
32. The following figure shows the direction field asso-
cording to drldt = J.Lr', does r increase or de-
ciated with the nonlinear system of equations
{Yn+l - Yn + hg(tm Xm Yn)
crease for r > o? Using these observations, is the
equilibrium point (0, 0), which was classified as where tn = to + nh, n = 0, 1, 2, ....
l
dx=2x-XY
a center for the linearized system, also classified dt
as a center for the nonlinear system? If not, how dy
would you classify it?
29. (a) Find a general solution of the system {X:Y =y+x

= -x 2
by solving the first equation and by substituting the

for ° :0: x :0: 6 and °
to sketch the graphs of the solutions
- = -3y +xy
dt
:0: Y :0: 6. Use the direction field

tr~ that satisfy
o 1'*,,,"'·'
Use Euler's method with It = 0.1 and h = 0.05 to approximate the solution to the
result into the second. (b) Sketch the phase plane and initial-value problem:
the initial conditions (a) x(O) = 2 and y (0) = 3; aod
determine if the origin is a stable equilibrium point
(b) x(O) = 3 and yeO) = 2. (c) How are the solutions dx
l
(e) Solve the linearized system about (0, 0) and com-
alike? How are they different? Yt=x-y+I
pare the result to that found in (b). Is (0, 0) assigned
the same classification as in (b)? dy _
Yt = x + 3y + e '.
30. Consider the Lorenz system
x(O) = 0, yeO) = 1
X' = -ax + OJ'
y' = rx - y - X2, Compare these results with the exact solution to the system of equations.
{
z' = -bz + xy
which was developed in the 1960s by Edward Lorenz
'] Solution In this case, I(x, y) = x - y + 1, g(x, y) = x + 3y + e-', to = 0,
..--1- 1 I : _ Xo = 0, and Yo = 1, so we use the formulas
to study meteorology. (a) Show that (0, 0, 0) is an equi- I 3 4 5 6 ~
rX n+l : Xn + h(xn - Yn + 1) _'" '

lYn+l - Yn + h(xn + 3Yn + e )
where tn = (O.I)n, n = 0, 1, 2, ....
370 Chapter 6 Systems of Differential Equations 6.8 Numerical Methods 371
Ifn = 0, then TABLE 6.4
fXl = Xo + h(xo - Yo + 1) = 0 tn Xn (approx) Xn (exact) Yn (approx) Yn (exact)

LYI = Yo + h(xo + 3yo + e- t ,) = 1.4· 0.0 0.0 0.0 1.0 1.0
0.05 0.0 -0.00532 1.2 1.21439
The exact solution of this problem, which can be determined using the method f
variation of parameters, is 0 0.10 -om -0.02270 1.42756 1.46032
0.15 -0.03188 -0.05447 1.68644 1.74321
0.20 -0.06779 -0.10335 1.98084 2.06545
0.25 -0.12023 -0.17247 2.31552 2.43552
0.30 -0.192013 -0.26543 2.69577 2.85904
0.35 -0.28640 -0.38639 3.12758 3.34338
0.40 -0.40710 -0.54011 3.61763 3.89682
TABLE 6.3 0.45 -0.55834 -0.73203 4.17344 4.52876
tn Xn (approx) Xn (exact) Yn (approx) 0.50 -0.74493 -0.96841 4.80342 5.24975
I Yn (exact)
0.0 II 0.0 0.0 1.0 1.0 0.55 -0.97234 -1.25639 5.51701 6.07171
0.1
I 0.0 , -0.02270 1.4 1.46032 0.60 I -1.24681 -1.60412 6.32479 7.00806
0.2 -0.04 -0.10335 0.65 -1.57529 -2.02091 7.23861 8.07394

i 1.91048 2.06545
0.70 -1.96609 -2.51737 8.27174 9.28638
0.3 -0.13505 -0.26543 2.5615 2.85904
0.75 -2.42798 -3.10558 9.43902 10.6645
0.4 -0.30470 I -0.54011 3.39053 3.89682
I
I 0.80 -2.97133 -3.79926 10.7571 12.23
0.5 -0.57423 -0.96841 4.44423 5.24975
0.85 -3.60776 -4.61405 12.2446 14.0071
0.6 -0.97607 -1.60412 5.78076 7.00806 0.90 -4.35037 -5.56767 13.9222 16.0232
0.7 -1.55176 -2.51737 7.47226 9.28638 0.95 -5.214 -6.68027 15.8134 18.3088
!
0.8
,
-2.35416 -3.79926 9.60842 12.23 1.00 -6.21537 -7.97468 17.944 20.8987
I
0.9 -3.45042 -5.56767 12.3005 16.0232
1.0 -4.9255 -7.97468 15.6862 20.8987
Runge-Kutta Method
Because we would like to be able to improve the approximation without using such a
In Table 6.3, we display the results obtained with this method and compare them to small value for h, we seek to improve the method. As with first-order equations, the
the actual function values. Runge-Kutta method can be extended to systems. In this case, the recursive formula at
Because the accuracy of this approximation diminishes as t increases, we at· each step is
tempt to improve the approximation by decreasing the increment size. We do this
by entering the value h = 0.05 and repeating the above procedure. We show the reo
suits of this method in Table 6.4. Notice that the approximations are more accurate
with the smaller value of h.
372 Chapter 6 Systems of Differential Equations 6.8 Numerical Methods 373
where k! = Xo - Yo + 1= 0- 1+ 1= 0
m! = Xo + 3yo + e -to = 0 + 3 + 1=4
h hk! hm!)
k2 = / ( In + 2' Xn + 2' Yn + -2- k2 = ( Xo + 2hk!) - (hm!) 4(0.1)
Yo + -2- + 1 = -1 - - 2 - + 1 = -0.2
_ (
m3 - g In
h
+ 2' hk2 hm2)
Xn + 2' Yn + -2- m2 = (xo + h~!) + 3 ~o + h;!) + e-(to+hI2) = 3(1 + 4(~.1)) + e-o.os
"" 4.55123
hk2) ( hm2) (0.1)(0.2) (0.1)(4.55123)
o "fi""'PJ
Use the Runge-Kutta method to approximate the solution to the initial-value
k3 = ( Xo
"" -0.23756
+2 - Yo + -2- + 1 = 2 - 1- 2 +1
problem
dx
7t=x-y+1
j
= (0.1)(0.2)
2 + 3( 1 + (0.1)(4.55123))
2 +e
-0.05"" 4 2391
.6
dy _I
7t = x + 3y + e
k4 = (xo + hk3) - (Yo + hm3) + 1 = (0.1)(-0.23756) - 1 + (0.1)(4.62391) + 1
x(O) = 0, yeO) = 1
"" -0.48615
for h = 0.1. Compare these results with the exact solution to the system of equa-
tions as weI! as those obtained with Euler's method. m4 = (xo + hk3) + 3(yo + hm3) + e-(to+h)
= (0.1)(-0.23756) + 3(1 + (0.1)(4.62391)) + e- O ! "" 5.26826.
:lSolution Because/(x, y) = x ~ y + 1, g(x, y) = x + 3y + e- I, 10 = 0, Xo =0,
and Yo = 1, we use the formulas Therefore,
f Xn+! ~ Xn + t (k! + 2k2 + 2k3 + k4)

lYn+! - Yn +"6 (m! + 2m2 + 2m3 + m4) = 0 + °i/ [0 + 2(-0.2) + 2(-0.23756) + -0.48615] "" -0.0226878
where
and
k2 = ( Xn + 2hk!) - (Yn + -2-

hm!) + 1
= 1 + 0(/ [4 + 2(4.55123) + 2(4.62391) + 5.26826] "" 1.46031.

k3 = ( Xn + 2hk2) - (Yn + -2-
hm2) + 1 m3 _- ( Xn + 2hk2) + 3 (Yn + -2-
hm2) + e _(/"+hIlI"c.
In Table 6.5, we show the results obtained with this method and compare them to
m4 = (xn + hk3) + 3(Yn + hm3) + e-(I,+hl the exact values. Notice that the Runge-Kutta method is much more accurate than
Euler's method. In fact, the Runge-Kutta with h = 0.1 is more accurate than Euler's
For example, if n = 0, then method with h = 0.05. (Compare the results here to those given in Table 6.4.)
.~
"
374 Chapter 6 Systems of Differential Equations 6.B Numerical Methods 375
TABLE 6.5
tn Xn (approx) Xn (exact)
I Yn (approx) --
Yn (exact)
TABLE 6.6
tn xn (R-K) Xn (linear) Yn (R-K) Yn (linear)

I
0.0 I
I
0.0 0.0 1.0 1.0 0.0 0.0 0.0 1.0 1.0
0.1 -0.02269 -0.02270 1.46031 1.46032 0.1 0.09983 0.09983 0.99500 0.99500
0.2 -0.10332 -0.10335

I 2.06541
-
2.06545 0.2 0.19867 0.198669 0.98013 0.98007
0.3 -0.26538 -0.26543 2.85897 2.85904 0.3 0.29553 0.29552 0.95566 0.95534
0.4 -0.54002 -0.54011 3.8967

-
3.89682 0.4 0.38950 0.389418 0.922061 0.92106
~
0.5
0.6
-0.96827
-1.60391
-0.96841
-1.60412
I 5.24956 5.24975 0.5
0.6
0.47966
0.56523
0.47943
0.56464
0.87994
0.83002
0.87758
0.82534
7.00778 7.00806
i
0.7 -2.51707 -2.51737 9.28596 9.28638 0.7 0.64544 0.64422 0.77309 0.764842
I
0.8 -3.79882 -3.79926 12.2294 12.23 0.8 0.71964 0.71736 0.70999 0.69671
---
0.9 -5.56704 -5.56767 16.0223 16.0232 0.9 0.78726 0.78333 0.641545 0.62161
-
1.0 -7.97379
.
-7.97468
I 20.8975 20.8987
-
1.0 0.84780 0.84147 0.568569 0.54030
The Runge-Kutta method can be extended to systems of first-order equations so Detailed discussions regarding Computer Algebra Systems
it can be used to solve higher order differential equations. This is accomplished by the error invo Ived in using
Euler's method or the Runge- Numerical and graphical solutions generated by computer algebra systems are useful
transforming the higher order equation into a system of first-order equations. We il· in helping us observe and understand behavior of the solution(s) to a differential equa-
Kutta method to approximate
lustrate this with the pendulunl equation that we have solved in several situations (by tion, especially when we do not wish to utilize numerical methods like Euler's method
solutions to systems of equa-
using the approximation sin x = x). tions can be found in ad- or the Runge-Kutta method.
vaoced numerical analysis
o """'M#I
Use the Runge-Kutta method with h = 0.1 to approximate the solution to the non·
texts.
o "fi"I.,., Rayleigh's equation is the nonlinear equation x" + G(x'f - l)x' + x = 0 and
linear initial-value problem x" + sin x = 0, x(O) = 0, x' (0) = 1. arises in the study of the motion of a violin string. We write Rayleigh's equation as
a system by letting y = x'. Then,
:1 Solution We begin by transforming the second-order equation into a system of
first-order equations. We do this by letting x' = Y, so y' = x" = -sin x. Hence,
f(t, x, y) = y, and get, x, y) = -sin x. With the Runge-Kutta method, we obtain the
y' = x" = -(t(X')2 - I)X' - = -(t x y2 - I)Y - x
approximate values given in Table 6.6 under the heading R-K. Also in Table 6.6 so Rayleigh's equation is equivalent to the system
under the heading "linear," we give the corresponding values of the initial-value
X' =y
problem x" + x = 0, x(O) = 0, x'(O) = 1. We approximate the nonlinear equation
x" + sin x = 0 with x" + x = 0 because sin x = x for small values of x. The solu· {
, (I
Y = - TY 2 - I ) y-x .
tion of x" + x = 0, x(O) = 0, x'(O) = I is found to be x = sin t, so y = x' = cos t.
Because the use of the approximation sin x = x is linear, we expect the approxima· We see that the only equilibrium point of this system is (0, 0). (a) Classify the
tions obtained with the Runge-Kutta method (which is a fourth-order method) to be. '*
equilibrium point (0, 0). (b) Is it possible to find Yo 0 so that the solution to the
more accurate than those obtained by solving x" + x = 0, x(O) = 0, x'(O) = 1. initial-value problem
~~~--~~-----------------------------------------------------_.II_;I' ~_':____________________________
' __
6.8 Numerical Methods 377
x' =y
l
EXERCISES 6.8
y' = - (ty2 - l)Y - x
x(O) = 0, yeO) = Yo
-3
is periodic? InExercises 1-6, use Euler's method with h = 0.1 to approxi- 14. x" + 4x' + 4x = 0, x(O) = 4, x'(O) = 0, 1 = I
IDate the solution of the initial-value problem at the given value
*15. x" + 9x = 0, x(O) = 0, x'(O) = 3, 1 = I
-3 ] Solution (a) The associated linearized system about the point (0, 0) is oft.
Figure 6.37 X' = (--' ~ Dx and the eigenvalues of (_ ~ Dare '\1.2 = t::': V; i so (0,0) X' = -x
1. { y' = 2x _ y
+ 2y + 5 _
+ 4 ,x(O) -
__
I, yeO) - 0, 1 - I
16. x" + 4x' + 13x = 0, x(O) = 0, x'(O) = 12,1 = I
17. 1 2X" + tx' + 16x = 0, x(l) = 0, x'(I) = 4, I = 2
is an unstable spiral point. This result is conflrmed by the direction field of the sys- 18. 12x" + 3tx' + x = 0, x(l) = 0, x'(l) = 2, t = 2
X' =x + Y + 1
tem for -3 :=; x:=; 3 and -3 :=; Y :=; 3, shown in Figure 6.37. 2 {. , ,x(O)=I,y(O)=-I,I=1
y =x - y
In the direction field, we see that all solutions appear to tend to a closed curve In Exercises 19-24, use the Runge-Kutta method with h = 0.1
L. Indeed, we see that the graph of the solution that satsifies the initial conditio~ *3.,
X' = 3x - 5y to approximate the solution of the initial-value problem in the
{ y =x- 2y+1 2, x(O) = -I, y (0) = 0, 1 = I
x(O) = 0.1 and yeO) = 0 in Figure 6.37 tends toward L. Choosing initial conditions earlier exercise given. Compare the results obtained to those in
outside of L yields the same result; Figure 6.38(b) shows the graph of the solution Exercises 13-18.
X' = x + 2y + cos 1 (0) = 0 (0) = I 1 = I
that satisfies x(O) = 0 and yeO) =s 3. 4. { y' = 5x _ 2y ,x ,y ,
19. Exercise 13
2
5. {x; = x - y, x(O) = -I, yeO) = 0, t = I 20. Exercise 14
Y =x +y *21. Exercise IS
6. {x; = xy , x(O) = I, yeO) = I, t = I 22. Exercise 16
Y =x - Y
23. Exercise 17
-3 In Exercises 7-12, use the Runge-Kutta method with h = 0.1 24. Exercise 18
to approximate the solution of the initial-value problem at the
given value at I.
Find the exact solution of the following initial-value problems
-3
-3 -3 = x - 3y + e'
X' _ __ and then approximate the solution with the Runge-Kutta method
7. { y' = -x + 6y , x(O) - O,y(O) - I, t - I using h = 0.1. Compare the results by graphing the two solu-
Cal Cbl tions.
X' = 4x - y _ __
-3 Figure 6.38 All solutions tend to L 8 {.y, = -x +5y +6'
sm t,x(O)--I,y(O)-O,I-1
25. {x; = Y2x ' x(O) = 0, yeO) = I
y = - y
Figure 6.39 An isolated peri- * {x' = x- 8y _ __
9. y' = 3x _ y + e' cos 21' x(O) - 0, yeO) - 0, I - I X' =y
(b) Can we find L? We use Figure 6.37 to approximate a point at which Lin. *26.
odic solution like this is called { y, = -4y - 13 ,x(O)
x
= -1,y(O)
,
= I
a limit cycle. This limit cycle is tersects the y-axis. We obtain y = 1.9. We find a numerical solution that satisfies
slable because all solutions spi- the initial conditions x(O) = 0 and yeO) = 1.9 and graph the result in Figure 6.39.
10. {x' 5x + Y
y' =x - 2y
= + v't+l, x(O) = 0, yeO) = 0, I = 2 X' - Y
ral into it." 27. { y ,=
- 2 ' x(O) = 2, yeO) = 0
The solution does appear to be periodic. -y-x
X' = _2y __
11. { y' = _xy' x(O) - 0, yeO) - I, 1 - I 28. (a) Graph the direction field associated with the non-
X' = y
Approximate the period of the periodic solution in Example 4. Hint: Graph x(t) and linear system { y' = _ sin x for -7 :5 x :5 7 and
12. {x' =xVy ,x(l) = l,y(l) = 1,1= 2
y (t) together for 0 :=; t :=; T for various values of T. y' = x - y -4:;; y:5 4.
::::::m{ate;f:s~:t:o:1 to the initial-value

(b) (i)
In Exercises 13-18, use Euler's method with h = 0.1 to ap-
proximate the solution of the initial-value problem by trans-
fOrming the second-order equation to a system of first-order XI(O) = 0, YI(O) = I
+ See texts like Nonlinear Ordinary Differential Equations (Second edition) by D. W. Jordan and equations. Compare the approximation with the exact solution
P. Smith, which is published by Oxford University Press (I987), for detailed discussions about limit cycles at the given value of I. (ii) Graph {XI(i) for 0 :5 1:5 7 and display the
and their significance. yI(I)
13. x" + 3x' + 2x = O. x(O) = 0, x'(O) = -3,1 = I graph together with the direction field. Does
Chapter 6 Systems of Differential Equations Chapter 6 Summary 379
378
the vector field are tangent to this curve? Is a II (I) a,z(t) ... alm(t)) General Solution
it appear as though the vectors in the vector
field are tangent to the solution curve? XI(t) + xit) . aZI (I) azzCt) ... a2m(t) A general solution of X'(t) = A(I)X(t) is X(t) = e,,(t) +
{ YI(t) + Yo(t) a solution to the system A(t) = :: :' e z<l>2(1) + ... + cn<l>n(t) where (,}7~1 is a set ofn linearly
. . .
(iii) ::::::lln{ate~r:S;:;~0:2to th.e initial-vruue X' = Y ? E I'
(
an I (t) anzCt) anm(t)
independent solution vectors of the system.
{ y' = -sin x· xp am.
where aU(t) is differentiable for all values of i and}, is Fundamental Matrix
X2(0) = 0, Y2(0) = 2 (c) {SOI;;~; initial-value problem
d d d <P1I<P,z ... <p,n)
(iv) Graph {~~~h for 0 ,;: t ,;: 7 and display the y' = -sin X and graph the solution para.
diall(l) dia,z(t) dia'm(t)
(t) = (, <l>z ...
<PZI <PZ2
<1>") = : :
.. . <P2n
:
graph together with the direction field. Does x(O) = 0, yeO) = 3 d d d ( .. .
i£.A(t) = diaz,(t) di aZ2(t) diaZm(t) clJ n1 ¢ln2 ¢nn
it appear as though the vectors in the vector metrically for 0 ,;: t ,;: 7. Is this the graph of dt
field are tangent to the solution curve? xI(t) + xz(t)? is called a fundamental matrix of the system X'(t) =
{YI(t) + Y2(t)"
d d d A(t)X(I) where {,}7~, is a set of n linearly independent so-
(v) Graph {XI(t) + X2(t) for 0 ,;: t ,;: 7 and dis- dia",(t) di adt) dianm(t)
YI (t) + Y2(t) (d) Is the Principle of Superposition valid for nonlin. lution vectors of the system.
play the graph together with the direction ear systems? Explain.
The integral of A(t), where ay(l) is integrable for all values
field. Does it appear as though the vectors in ofi and}, is
Iall(t)dl fadl)dt Ia,m(t)dt) 1Section 6.41
f A(t)dt = f a 21 ,(I)dl f a22(I)dl f a2m(l)dl
Distinct Real Eigenvalues
( If A is an n X n matrix with n distinct real eigenvalues
fan,(t)dt .-:'. janm(t)dl'
CHAPTER 6 SUMMARY {Ak}k~" a general solution to X' = AX is the linear combi-
nation of the set of solutions {X" X2 , ••• , X n },
:, Concepts & Formulas [section 6.31 X(t) = cIVleAlt + c2v2eAzt + ... + cnv"e Ant •
Inverse of a 2 x 2 Matrix Solution Vector Complex Conjugate Eigenvalues

1 Section 6.1 1 A solution vector of the system X'(I) = A(I)X(I) + F(t) on If A has complex conjugate eigenvalues A, = " + {3i and
If A .~ (~ ~) and IAI = ad - be *" 0, then the interval I is an n X I matrix of the form A2 = " - {3i and corresponding eigenvectors v, = a + bi
System of Ordinary Differential Equations and V2 = a - bi, two linearly independent solutions of
A system of ordinary differential equations is a simultaneous A-' I (d -b ). X'(t)) X' = AX are X,(I) = eQ'(a cos {31 - b sin {31) and X 2 (t) =
set of equations that involves two or more dependent vari- = ad - be -e a X(t) = xz~t) eQ'(a sin {31 + b cos {3t).
ables that depend on one independent variable. A solution to
Eigenvalues and Eigenvectors (
the system is a set of functions that satisfies each equation on xn(t) Repeated Eigenvalues
A nonzero vector x is an eigenvector of the square matrix A
some interval I. If the system X' = AX has the repeated eigenvalue A, = Az
if there is a number A, called an eigenvalue of A, so that Ax where the x,(I) are differentipble functions that satisfy
with only one corresponding eigenvector v" two linearly
= Ax. X'(t) = A(t)X(t) + F(t) on I.
independent solutions corresponding to AI = A2 are
1Section 6.21 Characteristic Polynomial Fundamental Set of Solutions X, = Vie'"~ and X2 = (v,t + wz)e'" where W2 satisfies
The equation det(A - AI) = 0 is called the characteristic (A - A,l)wz = VI.
Matrix, Transpose of a Matrix
equation of A; det(A - AI) is called the characteristic poly-
A set {<P,}7~, = ~21
<p'll]n
of n linearly independent
Scalar Multiplication, Matrix Addition, Matrix nomial of A; the roots of the characteristic polynomial of A [(
are the eigenvalues of A. q,m
Multiplication
ir::cl
solution vectors of X' (t) = A(t)X(t) on an interval I is caUed 1Section 6.51
Eigenvalue of Multiplicity In • fundamental set of solutions on I.
Identity Matrix (2 X 2)
Suppose that (A - Adm where m is a positive integer is a fac· Variation of Parameters
I = (~ ~) tor of the characteristic polynomial of the n X n matrix A
while (A - A,)m+1 is not a factor of this polynomial. Then
Wronskian A general solution to X' = AX + F(I) is
Determinant of a 2 x 2 Matrix A = Al is an eigenvalue of multiplicity m. XCt) = (t)C + Xp(t) = (t)C + (t) f<1>-' (t)F(t) dl.
Derivative and Integral of a Matrix
where <1>(1) is a fundamental matrix of the system X' = AX.
The derivative of the n x m matrix
where
\ Section 6.6\
~
Euler's Method
The initial-value problem
Eigenvalues Geometry Stability
l
A" A2 real; Al > A2 > 0 Improper node Unstable ~ =f(l,x,y)
h hkz hm2)
k3 = f ( In + 2' Xn + T' Yn + -2-
A" A2 real; A, = A2 > 0; I eigenvector Deficient node Unstable dy
dt = g(t, x, y)
A" Az real; Al = A2 > 0; 2 eigenvectors Star node Unstable
x(lo) = xo, y(to) = Yo
A" A2 real; A2 < A, < 0 Improper node Asymptotically stable
is approximated at each step by the recursive relationship and
A" A2 real; A, = A2 < 0; I eigenvector Deficient node Asymptotically stable based on the Taylor expansion of x and y up to order h:
A" A2 real; A, = A2 < 0; 2 eigenvectors Star node Asymptotically stable Xn+ I :: Xn + hf(tm Xm Yn)
{Yn+1 - Yn + hg(ln. Xm Yn) h hkl hml)
A" A2 real; Az < 0 < Al Saddle point Unstable mz = g ( In + 2' Xn + T' Yn + -2-
where In = 10 + nh, n = 0, I, 2, ....
A, = ex + {3i. Az = *' 0, ex> 0
ex - {3i, {3 Spiral point Unstable
Runge-Kutta Method h hko hmz)
+ 2' x" + 2' Yn + -2-
Al = ex + {3i, A2 = ex - {3i, {3 *' 0, ex < 0
m3 = g ( tTl
Spiral point Asymptotically stable The Runge-Kutta method for systems uses the recursive for-
II mula at each step
Al = {3i, A2 = - {3i, (3 *' 0 Center Stable m4 = g(ln + h, Xn + hk3, Yn + hm,).
f Xn+1 = Xn + t(k, + 2kz + 2k, + k4)
lYn+, = Yn + "6(ml + 2m2 + 2m3 + m4)
\ Section 6.7\
Eigenvalues of J(xo, Yo) Geometry Stability
A" Ao real; Al > A2 > 0 Improper node Unstable CHAPTER 6 REVIEW EXERCISES
A" A2 real; Al = A2 > 0 Node or spiral point Unstable
A" Az real; Az < Al < 0 Improper node Asymptotically stable
-~)
In Exercises 1-9, find the eigenvalues and corresponding eigen-
AI, A2 real; Al = Az < 0 Node or spiral point Asymptotically stable -2
vectors of A. ~
9. A = C4
-3
A" A2 real; A2 < 0 < Al Saddle point Unstable
(-16 ~) 2. A= (-34 -~)
Al = ex + {3i, A2 = ex - {3i, {3 *' 0, ex > 0 Spiral point Unstable
1. A =
In Exercises 10-34, find a general solution to each system or
AI = ex + {3i, A2 = ex - {3i, {3 *' 0, ex < 0 Spiral point Aymptotically stable *3 A =

. (-I-I -~) 4. A = (~ -;) solve the initial-value problem.
*' 0
I
-5)6
Al = {3i, A2 = -{3i, (3 Center or spiral point Inconclusive
5. A=( b
-I i) -2
0
6. A = (-~
-2
0
0
-i)
-3
10. X' = (-;
11. X' = (_~ -4)

5
X
(-~ -~)
-3
*7. A = (-1 i) 1
3
0
8. A =
-3
3
0 -3
12. X' = (-b -6)
-7 X, X(O) = (-I)2
382 Chapter 6 Systems of Differential Equations Differential Equations at Work 383
G _~)x, m x' = -3x - ~y + ~ sin 2t
l . Differential Equations at Work:

*13. X' = X(O) = IX/XC'
28., 2 2
Y = 4x + 3y + 2 cos 2t - 3 sin 2t ,
14. X' = (=; ~)X x(O) = -1,y(O) = I A. Modeling a Fox Population in which Rabies Is Present
15. X' = (=; ~)x *29. X' = (~ -I)X + (-sin

o cost'
t) X(O) = (I)0 Under various assumptions, the nonlinear system of differential equations
dX
30. (Modeling Testosterone Production) The level of dt = rX - yXN - f3XY
16. X' (4 -S)x
=
4 -4 testosterone in men can be modeled by the system of
dl
delay equations dt = f3XY - (0" + b + yN)1
*17. X' = (-~ -1)
-2 X dR dY
dt = f(T) dt 0"1 - (a + f3 + yN)Y
I
- b,R =
18. X' = (_~ ;)X

dL
dt = g,R - bzL .'
dN
dt = aX - (b + yN)N - aY
19.X'=(_~ ~)x
dT
has been successfully used to model a fox population in which rabies is present. * Here,
I
I
dt = g2L (t - T) - b3 T
X(t) represents the population of foxes susceptible to rabies at time t, let) the popula-
tion that has contracted the rabies virus but is not yet ill, yet) the population that has
20. X' = (_; _:)X, X(O) = (-D Show that if f(O) > 0 and f(T) is a one-to-one de-
creasing function, then the equilibrium point of this developed rabies, and N(t) the total population of the foxes. The symbols a, b, r, ')I, 0",
a, and f3 represent constants and are described in the following table.
=!)x, X(O) = m
system is (Ro, Lo, To), where
*21 X' = ( 2
· 18
Constant Description Typical Value(s)
I
22. X' = (_~ -42)X a a represents the average per capita birth rate of foxes.
b lib denotes fox life expectancy (without resource limita- 0.5
23 · X' = (8I -1) tions), which is typically in the range of 1.5 to 2.7 years.
6 X
and that the associated linearized system is
e -3) (-2e2e4'4')
r = a - b represents the intrinsic per capita population 0.5
II
24 · X' = 1 growth rate.
5 X+ d:c
l
dt = j'(To): - b,x
I' K = ~ represents the fox carrying capacity of the Varies
*25. X' = (~ -9)X _ 8(
-2
l)e
-1
7'
!!l.._
dt -g,x-b2 y .
"y
defined area, which is typically in the range of 0.1 to 4
foxes per \an2 • We will compute K and r and then
26. {X:Y == 3x-4x+ -3ylOy- 1+ 1 dz
dt = g2y(t - T) - b3 z
approximate "y.
1/0" represents the average latent period. This 12.1667
27. {x:y=2x+y-l
= 3x + 4y + 1
represents the average time (in years) that a fox can
carry the rabies virus but not actually be ill
with rabies. Typically, I/O" is between 28 and 30 days.
a represents the death rate of foxes with rabies. 11 a 73
is the life expectancy (in years) of a fox with rabies
and is typically between 3 and 10 days.
f3 f3 represents a transmission coefficient. Typically, 80

1/ f3 is between 4 and 6 days.
• 1. D. Murray, Mathematical Biology. Springer-Verlag (1990). pp. 166 175.

• Roy M. Anderson, Helen C. Jackson, Robert N. May & Anthony M. Smith, "Population dynamics of
fox rabies in Europe," Nature. Volume 289 (February 26,1981), pp. 765-771.
II
I
I
'"
384
1. Generate a numerical solution to the system that satisfies the initial conditio into consideration. This model might be used to model diseases that are epidemic to
XC?) = 0.93, 1(0) ~ 0.~35, YeO) ~ 0.035, and N(O) = 1.0 valid for 0 :s; t :s; : a population: those diseases that persist in a population for short periods of time (less
usmg the values gIVen m the preVIous table if K = 1,2, 3, 4, and 8. In each cas
graph xct), I(t), Yct), and NCt) for 0 :s; t :s; 40. e, than one year). Such diseases typically include influenza, measles, rubella, and chicken
pox.
2. Repeat (1) using the initial conditions XCO) = 0.93, 1(0) = 0.02 YeO) = 0.05 d
N(O) = 2.0. ' , an
3. For. b~th (1) ~d (2), estimate the smallest value of K, say K r , so that yet) is
1. Show that if So < t, the disease dies out, while an epidemic results if So> f.
penodic functIOn. a
dl
4. What happens to yet) for K < Kr? Explain why this result does or does not mak 2. Show that dS = (AS - y)1 = -1 + E. where p =.1 has solution
ASI S' A'
~~ e
5. Define the basic reproductive rate R to be 1+ S - pinS = 10 + So - p In So.
R= U"{3K 3. What is the maximum value of n

(U" + a)(a + a) When diseases persist in a population for long periods of time, births and
and deaths must be taken into consideration. If a person becomes immune to a disease
after recovering from it and births and deaths in the popUlation are taken into ac-
count, then the percentage of persons susceptible to becoming infected with the
Kr'; (U" + a)(a + a) disease Set) and the percentage of people in the population infected with the dis-
U"{3
ease l(t) can be modeled by the system
Show that if R > 1 then K > Kr and if R < I then K < K r.
S'(t) = -ASI + J.L - J.LS
6. Use the values in the table to calculate Kr = [(U" + a) (a
+ a)]/U"{3. Compare the re-
sult to your approximations in (3). How do they compare? I'(t) = ASI - yl - pi .'
{
S(O) = So, 1(0) = 10
7. Predict how the solutions would change if the transmission coefficient {3 were de-
~reased or the de~th rate a were increased. What if the average latent period U" were This model is called an SIR model with vital dynamics because once a person
mcreased? Expenment with different conditions to see if you are correct. has had the disease he becomes immune to it, and because births and deaths are
taken into consideration. This model might be used to model diseases that are en-
demic to a popUlation: those diseases that persist in a population for long periods
B_ Controlling the Spread of a Disease
of time (ten or twenty years). Smallpox is an example of a disease that was en-
demic until it was eliminated in 1977.
See the subsection Modeling the Spread of a Disease at the end of Chapter 2 for an
introduction to the terminology used in this section. . {S'(t) = -ASI + J.L - J.LS
4. Show that the equilibrium pomts of the system ret) = ASI _ yl _ pi are
If a person becomes immune to a disease after recovering from it, and births and
y+J.L J.L[A-(y+J.L)]
deaths in the population are not taken into account, then the percentage of persons sus- S= 1,1= 0 andS= -A-' 1= A(y + J.L)
ceptible to becoming infected with the disease Sct) the percentage of people in the pop-
ulation infected with the disease let) and the percentage of the population recovered Because Set) + I(t) + R(t) = 1, it follows that Set) + I(t) :s; 1.
and immune to the disease RCt) can be modeled by the system 5. Use the fact that sct) + ICt) :s; 1 to determine conditions on y, J.L, and A so that the
S'ct) = -ASI system {S;(t) = -SASI +f -'.if
I (t) = A 'I - y1 - ,....
has the equilibrium point S = Y ~ J.L,
1\
ret) = ASI - y J
{ R'(t) = yl + J.L)] In this case classify the equilibrium point.
1 = J.L[A - (y
A(y+J.L)· ,
S(O) = So, 1(0) = 10, R(O) = 0
Because Sct) + I(t) + R(t) = I, once we know S and I, we can compute R with
• Herbert W Hethcote, "Three Basic Epidemiological Models," Applied Mathematical Ecology, edited by
RCt) = 1 - Set) - ICt). SimonA. Levin, Tbomas G. Hallan, and Louis J. Gross, Springer-Verlag (1989), pp. 119-143. Roy M.
Anderson and Robert M. May, '"Directly Transmitted Infectious Diseases: Control by Vaccination,"
This model is called an SIR model without vital dynamics because once a person bas
Science. Volume 215 (February 26, 1982), pp. 1053-1060. J. D. Murray, Mathematical Biology, Springer-
had the disease, he becomes immune to it, and because births and deaths are not taken Verlag. 1990, pp. 611-618.
6. Use the fact that Set) + I(t) :s; 1 to detennine conditions on y, p., and A so that the 0"(1 - R) < 1
S'(t) = -ASI + f.L - jLS
system { I' (t) = ASI - yI _ jLl does not have the equilibrium point O"-O"R<1
S= y+f.L 1= f.L[A-(y+f.L)]
A' A(y + f.L)
The following table shows the average infectious period and typical Contact
-O"R < 1- 0"
R> 0"-1 =1-1...

I
numbers for several diseases during certain epidemics.
Infectious Period
0" 0"
8. For each of the diseases listed in the following table, esti~ate the minim~ per-
centage of a population that needs to be vaccinated to achIeve herd unmumty.
I
I
Typical contact
(Average) number
Disease 1/')' Minimum Value of
')' a
Measles 6.5
R to Achieve Herd
0.153846 14.9667 Disease Immunity
Chicken pox 10.5
Mumps 19
0.0952381
0.0526316
11.3
8.1
Measles
I
I
Chicken pox
Scarlet fever 17.5 0.0571429 8.5 Mumps
Scarlet fever
Let's assume that the average lifetime IIf.L is 70 years so that f.L = 0.0142857.
II
7. For each of the diseases listed in the following table, use the formula (J" = 9. Using the values obtained in the previous exercises, for each disease in the tables
AI( y + f.L) to calculate the daily contact rate A.
.
graph the direction field and several solutIOns ( I(t), Set), R(t), and {set») . both
I(t) usmg
Disease A models. Discuss scenarios in which each model is valid and note any significant
differences between the two models.
Measles
10. What are some possible ways that an epidemic can be controlled?
Chicken pox I
Mumps
Scarlet fever
c. FitzHugh-Nagumo Model
Under certain assumptions, the FitzHugh-Nagumo equation, which a~ses in.the stu~y
II
II
of the impulses in a nerve fiber, can be written as the system of ordmary differential
Diseases such as those listed above can be controlled once an effective and equations
inexpensive vaccine has been developed. It is virtually impossible to vaccinate
evel!'body against a disease; we would like to determine the percentage of a pop- dV = W
ulatIOn that needs to be vaccinated to eliminate a disease from the population un- dg
j
der consideration. A popUlation of people has herd immunity to a disease when
enough people are immune to the disease so that if it is introduced into the popu-
: = F(V) +R - uW,
lation, it will not spread throughout the popUlation. To have herd immunity, an in- dR €
fected person must infect less than one uninfected person during the time the per- - = -(bR - V - a)
son is infectious. Thus, we must have
dg u
where F(V) = tv 3 - V.*

O"S < 1.
Because I + S + R = 1, when I = 0 we have that S = 1 - R; consequently, herd • J. D. Murray, Mathematical Biology, Springer-Verlag (1990), pp. 161 166.Alwyn C. Scott, "The elec-
immunity is achieved when trophysics ofa nerve fiber," Reviews oj Modern Physics, Vol. 47, No.2 (Apol 1975), pp. 487-533.
388
1. 0~ph the solution to the FitzHugh-Nagumo equation that satisfies the initial
7
ditlOns V(O) = I, W(O) = 0, R(O) = I if e = 0.08, a = 0.7, b = 0, and u = I Con·
2. Graph .the solution that satisfies the initial conditions V(O) =I W(O) = 0 5·
= 0.5 If e = 0.08, a = 0.7, b = 0.8, and u = 0.6. ' . , R(O)
3. A(IP)proxinIate the maxinIum and minimum values, if they exist of v: W and R .
and (2). ' " In c.
Applications of Systems of
Ordinary Differential Equations
Mechanical and Electrical Problems with First-Order

Linear Systems
(" L-R-C Circuits with Loops (" L-R-C Circuits with One Loop
(" L-R-C Circuits with Two Loops (" Spring-Mass Systems
L-R-C Circuits with Loops

As indicated in Chapter 5, an electrical circuit can be modeled with a linear ordinary
differential equation with constant coefficients. In this section, we illustrate how a cir-
cuit involving loops can be described as a system of linear ordinary differential equa-
tions with constant coefficients. This derivation is based on the following principles.
Kirchhoff's Current Law:

The current entering a point of the circuit equals the current leaving the point.
389
390 Chapter 7 Applications of Systems of Ordinary Differential Equations 391
7.1 Mechanical and Electrical Problems with First·Order Linear Systems
l ~ :~1...0-!'I+
Kirchhoff's Voltage Law:
II The sum of the changes in voltage around each loop in the circuit is zero.
E(O
dt LC- L L
with initial conditions Q(O) = Qo and 1(0) = 10 on charge and current, respectively. TI:e
As was the case in Chapter 5, we use the following standard symbols for the
components of the circuit: method of variation of parameters (for systems) can be used to solve problems of thIS
type.
dQ . .
I(t) = current where I(t) = d("(t), Q(t) = charge, R = resIstance, C = capacItance,
E = voltage, and L = inductance. "ii,,','.'

The relationships corresponding to the drops in voltage in the various compo. Determine the charge and current in the L-R-C circuit with L = I henry, R =
nents of the circuit are restated in Table 7.1. 2 ohms, C = 4/3 farads and E(t) = e - I if Q(O) = I and 1(0) = I.
,-, ] Solution We begin by modeling the circuit with a system of differential equa-
TABLE 7.1
tions. In this case, we have
I~:~~Q
Circuit Element Voltage Drop
Inductor L dl
dl
- 2I + "
Resistor dt 4
RI
WIt. h'mltla
.. I condiu' ons Q(O) -- I and 1'(0) = I . We can write this nonhomogeneous
Capacitor '!'Q system in matrix form as
C
m~(Ot _:)(~HO,)
Voltage Source -E(I)
~~~=J;;:r.."
L-R-C Circuit with One Loop . homogeneous systern are AI = -1/2 and
The eigenvalues of the corresponding
In determining the drops in voltage around the circuit, we consistently add the volt.
ages in the clockwise direction. The positive direction is from the negative symbol ta-
A2 = -312, with corresponding eigenvectors VI = (-i) and V2 = (_;). respec-
R
ward the positive symbol associated with the voltage source. In summing the voltage tively. Thus a fundamental matrix is
$1 ; I~'
Ca) C .
drops encountered in the circuit, a drop across a component is added to the sum if the
positive direction through the component agrees with the clockwise direction. Other.
wise this drop is subtracted. In the case of the L-R-C circuit with one loop involving
-2e- ,/2
(t) = ( e -112
each type of component shown in Figure 7.1, the current is equal around the circuit by and
Kirchhoff's current law.
Also, by Kirchhoff's voltage law, we have _1.4 e '/2 _2.2 e '/2 )
-+
-'(t) = .
Cb)
dI
RI + L dt + CQ -
I
E(t) = O. ( -t e
31/2
e
31/2
Figure 7.1 (a)-(b) A simple

L-R-C circuit.
Solving this equation for dIldt and using the relationship dQldt = I, we have the sys·
tern of differential equations Then, by variation of parameters, if we let X(t) = (9g1)' we have
392 Chapter 7 Applications of Systems of Ordinary Differential Equations 7.1 Mechanical and Electrical Problems with First-Order linear Systems 393
X(t) = 4>(t)4>-\O)X(O) + 4>(t) fo 4>-\u)F(u) du In Example 1, do the limits limHoo Q(t) = limt_
initial conditions?
oo I(t) = 0 hold for all choices of the
= (-2e-
e- 112
tI2
2e-
3t12
- 3e -3t12
)(-t -+)(1)
I I I
L-R-C Circuit with Two Loops
The differential equations that model the circuit become more difficult to derive as the
-"4 -2
number of loops in the circuit increase. For example, consider the circuit that contains
two loops, as shown in Figure 7.3.
+ (-2e- tI2
~;:3UI2 C~u)dU
I U12)
e- tl2
2e-
3t12
_3e- 3tI2
)(-t) _1
+ (-2e-tl2
e- tl2
2e- 3t12 ) It -2I e
Ie
- 3 e -3tl2 0 ( _ _ u/2
-UI2)
du
4 2
le-tl2 _le-3t12 ) (b)

Ca)
+ -4e- t + 2e-t12 + 2e- 3t12
=
(-t t
2 2
e -tl2 + e - 3t12 (4e- t - e- tl2 - 3e- 3t12 ) Figure 7.3 (a)-(b) A two-loop circuit
= ( t e -tl2 + +e-
3tI2
- 4e. -t ) In this case, the current through the capacitor is equivalent to II - 12 • Summing
the voltage drops around each loop, we obtain the system of equations
_.'2.e - tI2 _le- 3t12 + 4e-t .
4 4 .
+ ~Q -
l
RIII E(t) = 0
We plot the solution X(t) = (9(~i) parametrically in Figure 7.2(a). Notice that dI2 I·
LTt + R2I2 - C Q = 0
limHoo Q(t) = limHoo I(t) = 0, so the solution approaches (0, 0) as t increases. We
also plot Q(t) and I(t) simultaneously in Figure 7.2(b). Finally, in Figure 7.2(c), we Solving the first equation for II yields
graph X(t) = (~~i) for other initial conditions. I I
II = R; E(t) - RIC Q.
Using the relationship dQldt = I = II - 12 gives us the system
l
dQ = __I_Q_I + E(t)
Q(') dt RIC 2 RI
-2 -1 dI2 I R2
Tt= Lc Q I2 -r:
-I
(a)
Figure 7.2 (a)-(c).

-2
(b) (e)
-2
o ·*fi"....,
Find Q(t), I(t), II (t), and lz(t) in the L-R-C circuit with two loops given that RI =
R2 = I ohm, C = 1 farad, L = 1 henry, and E(t) = e - t if Q(O) = 1 and 12 (0) = 3.
394
Chapter 7 Applications of Systems of Ordinary Differential Equations 7.1 Mechanical and Electrical Problems with First-Order Linear Systems 395
.1 Solution The nonhomogeneous system that models this circuit is

= (e- ' cos I - 3e=' s~n I) + (_~-t sin I e=' C?S I)(COS.I - I)
3e -t cos I + e t SIn l e t cos l e t SIn I SIn I
=(e-'cosl-3e=ts~I)+( _te-ts~~1 )
3e- t cos t + e t SIn I e - e cos I
e-t cos I - 2e-' SID t )

with initial conditions Q(O) = I and 12(0) = 3. As in Example 1 we use the th
.. f ' me od = ( 2e- 1 cos 1+ e- I + e- t sin I .
o f vanatlOn 0 parameters to solve the problem. In matrix form this system is
Because dQ/dl = I and Q(I) = e -t cos I - 2e -t sin I, differentiation yields
I(t) = -e- t cos 1- e- I sin t + 2e- t SID I - 2e- t cos t = -3e- t cos 1+ e- t sin I.
Also, because II (t) = 1(1) + h(t),
11(1) = I(t) + 12 (t) = -3e- t cos I + e- I SID t + e- t sin I + 2e- t cos I + e- t
The eigenvalues of the corresponding homogeneous system are ,\ 1.2 = -1 :t i, and = -e- t cos I + 2e- 1 sin I + e- t.
an eigenvector corresponding to .Al = - j + i is VI = (i) I + i(l)
1 = (0) T I·
O·worn- We graph Q(t), II(t), lz(t), and I(t) together ID Figure 7.4(a). In Figure 7.4(b), we
early independent solutions of the corresponding homogeneous system are graph {~~:~ parametrically to show the phase plane for the system of non-
= (~gD = e- I(~) = (~~~~~~n/)

I I homogeneous equations uSIDg several different initial conditions. Notice that some
XI(t) cos I(n - e- sin 4 5 6' of the solutions overlap, which does not occur if a system is homogeneous.
and
Find Ihe limit of Q(t), II (t), 12(/), and 1(1) as I -> 00. Does a change in initial condi-
X 2 (t) = (Q(I)) = e - I SID 1(0) + e -t cos t (1) = (e -tt cos I) lions affeci Ihese limits?
IiI) I 0 e- sin I '
(al
so a fundamental matrix is
Spring-Mass Systems
(t) = (_~-t SID I The displacement of a mass attached to the end of a spring was modeled with a second-
e 1 cos I order linear differential equation with constant coefficients in Chapter 5. This situation
and can be expressed as a system of first-order ordIDary differential equations as welL Re-
call that if there is no external forcing function, the second-order differential equation
<1>-\1) = -; tSIn
·
I e 1 C?S I) .
t
that models the situation is
( e cos I e SIn I
d 2x dx
md(2 + c
Therefore,
dt + kx = 0,
X(I) = (t)-\O)X(O) + (t) fo -\u)F(u) du where m is the mass of the object attached to the end of the spring, c is the damPIDg
coefficient, and k is the spring constant found with Hooke's law. This equation is trans-
= (-e- t
sin I
e-tcosl
(b)
Figure 7.4 (a)-(b).

formed into a system of equations with the substitution dx/dl = y. Then, solving the
differential equation ford 2x/dt 2 , we have dy/dl = d 2x/dt 2 = -klmx - elm dx/dt, which
yields the system
+ (-e -t sin I e -t cos
e - t cos I e -I sin I
t) It (-ee UcosSIDuu
U
U
e cos u)(e -U) du
e" SIn u 0 dx
0 -=y
t
= (-e-t sin t
t
e- t cos 1)(3)
t
(-e- ' sin I e- cos I) It (-SID u) ;; = -~x -~y"
e- cos I e- SID I I + e-tcos I e -t· SInt 0 cosu du 1dl m m
396 Chapter 7 Applications of Systems of Ordinary Differential Equations 7.1 Mechanical and Electrical Problems with First-Order Linear Systems 397
In previous chapters, the displacement of the spring was illustrated as a functio

of tim~. Problems of this type may also be investigated using the phase plane. In
followmg example, the phase plane corresponding to the various situations encoun_
th: ... EXERCISES 7.1
tered by spring-mass systems discussed in previous sections (undamped, damped, Over_

damped, and critically damped) are determined.
solve each of the following systems for charge and current us-
ing the procedures discussed in the example problems.
"i'd',·i.1 *1. Solve the one-loop L-R-C circuit with L = 3 henrys,

R = 10 ohms, C = 0.1 farad, and (a) E(t) = 0;
Solve the system of differential equations to find the displacement of the mass if (b) E(I) = e- I • (Q(O) = 0 coulombs, 1(0) = 1 amp.)
m = I, c = 0, and k = I. 2. Solve the one-loop L-R-C circuit with L = 1 henry,
R = 20 ohms, C = 0.01 farad, and (a) E(I) = 0;
J Solution In this case, the system is (b) E(I) = 1200. (Q(O) = 0 coulombs, 1(0) = 1 amp.)
(a) (b)
3. Find Q(I), let), I,(t), and 12(t) in the two-loop L-R-C
dx circuit with L = 1, R, = 2, R2 = 1, C = 1/2, and Figure 7.6 (a)-(b) A three-loop circuit.
Tt=y (a) E(I) = 0; (b) E(t) = 2e- 1/2 • (Q(O) = 10- 6 cou-
.• dy . lomb, 12(0) = 0 amps.) In Exercises 7-11, solve the three-loop circuit using the given
- = -x
I dt 4. Find Q(t), let), I,(t), and I,(t) in the two-loop L-R-C
circuit, with L = 1, R, = 2, R2 = 1, C = 1/2, and
values and initial conditions.
7. L2 = L, = I henry, R, = I ohm, R2 = 1 ohm, C =
1=; _\.1 = A
(a) E(I) = 90; (b) E(t) = 90 sin t. (Q(O) = 0 coulomb,
I farad, and (a) E(I) = 0 volts; (b) E(t) = e- I volts.
The eigenvalues are solutions of 2
+ I = 0, so Au = ~i. An lz(O) = 0 amps.)
(Q(O) = 10- 6 COUlomb, 12(0) = 1,(O) = 0 amps.)
*5. Find Q(I), l(t), 1,(t), and 12(t) in the two-loop L-R-C
eigenvector corresponding to A, = i is v, = C) = (~) + i(~). so two linearly inde- circuit, with L = 1, R, = 1, R2 = 3, C = 1, and
(a) E(I) = 0, Q(O) = 10- 6 , and 12(0) = 0; (b) E(I) =
8. L2 = L, = I henry, R, = R2 = 1 ohm, C = 1 farad,
and (a) E(t) = 90 volts; (b) E(I) = 90 sin I volts.
pend ent solutions

· are X,(t) = (I)
0 cos t - (0).I sm t = ( t)
cos t and
-sin 90, Q(O) = 0, 1,(0) = O.
6. Consider the circuit made up of three loops illustrated
(Q(O) = 0 coulomb, loCO) = 1,(O) = 0 amps.)
*9. L2 = I henry, L, = I henry, R, = R2 = 1 ohm, C =
I farad, and (a) E(I) = 90 volts; (b) E(t) = 90 sin I
(I). (0) cos t)

in Figure 7.6. In this circuit, the current through the
(a) X 2(t) = 0 sm t + I cos t = (Sin t . resistor R2 is 12 - I" and the current through the ca-
volts. (Q(O) = 0 coulombs, 12(0) = I amp, 1,(O) =
Damps.)
pacitor is It - 12 , Using these quantities in the
A general solution is voltage-drop sum equations, model this circuit with 10. L2 = 3 henrys, L, = I henry, R, = R2 = I ohm, C =
the three-dimensional system: I farad, and (a) E(t) = 0 volts; (b) E(t) = 90 sin t
volts. (Q(O) = 0 coulombs, 12(0) = 1 amp, 1,(0) = 0
X(t) = (x(t)) = c,X,(t) + C2 X 2(t) = ( c, co.s t + C2 sin t ). amps.)
yet) -c, sm t + C2 cos t
11. L2 = 4 henrys, L, = 1 henry, R, = R2 = 1 ohm, C =
Notice that this system is equivalent to the second-order differential equation d 2x1dt2 I farad, and (a) E(I) = 90 volts; (b) E(t) = 90 sin t
+ x = 0, which we solved in Chapters 4 and 5. At that time, we found a general so- volts. (Q(O) = 0 coulombs, 12(0) = 1 amp, 1,(O) =
lution to be x(t) = c, cos t + C2 sin t, which is the same as the first component of Damps.)
X(t) = (;i~~) obtained in this instance. We graph this function for several values of 12. Show that the system of differential equations that
models the four-loop circuit shown in Figure 7.7 is
the arbitrary constants in Figure 7.5(a) to illustrate the periodic motion of the mass. Using the relationship dQldl = I, - 12 and solving the
(b) Also notice that (0, 0) is the equilibrium point of the system. Because the eigen- first equation for I" show that we obtain the system L, dl, = -(R, + R 2)1, + R212 + R,f. + E(t)
dt
values are A = ~ i, we classify the origin as a center. We graph the phase plane of
Figure 7.5 (a)-(b). this system in Figure 7.5(b). 2
L2 dl = R21, - (R2 + R,)12 + R,l,
dl
By observing the phase plane in Figure 7.5 and the corresponding system of differ- L, dI, = R,lz - (R, + R.)l, + R.I.
dt
ential equations in Example 3, describe the motion of the object in each quadrant
and determine the sign on the velocity dxldt of the object in each quadrant. L4 ~; = R,l, + R.I, - (R, + R.)f.
398 Chapter 7 Applications of Systems of Ordinary Differential Equations 7.2 Diffusion and Population Problems with First-Order Linear Systems 399
Diffusion and,Population Problems with First-Order

Linear Systems
C" Diffusion Through a Membrane C" Mixture Problems
C" Population Problems
(b) Fi~d restrictions on m, c, and k to classify this
pomt as a center, stable node, or stable spiral. Diffusion Through a Membrane
(c) Can the equilibrium point be unstable for an
choice of the positive constants m, c, and K! Is ~
Solving problems to determine the diffusion of a substance (such as glucose or salt) in
saddle possible? a medium (such as a blood cell) also leads to fIrst-order systems of linear ordinary dif-
Figure 7.7 A four-loop circuit.
ferential equations. For example, consider the situation shown in Figure 7.8 in which
two solutions of a substance are separated by a membrane. The amount of substance
In Exercises 13-20, transform the second-order equation to a Concentration: x~:)
Use a graphing device to graph the solutions to Exercises that passes through the membrane at any particular time is proportional to the differ-
system of rITst-order equations and classify the system as un·
II
2S-28 SImultaneously and parametrically. Also determine th ence in the concentrations of the two solutions. The constant of proportionality, P, is
damped, overdamped, underdamped, or critically damped by
rmding the eigenvalues of the corresponding system. Also clas-
sify the equilibrium point (0, 0).
limit of these solutions as t ..... "'.
25. Solve the one-loop L-R-C circuit with L = I R = 40

'
g;'il Concentration: ~
called the permeability of the membrane and describes the ability of the substance to
permeate the membrane (where P > 0). If we let x(t) and yet) represent the amount
of substance at time t on each side of the membrane, and V, and V2 represent the
*13. !x" + ~x = 0 ·cC = 0.004, andE(t) = 120 sin t. (Q(O) = 0 c~ulamb;
1(0) = 0 amps.) , (constant) volume of each solution, respectively, then the system of differential equa-
14. tx" + 4x = 0 Figure 7.8 tions is given by
26. Solve the one-loop L-R-C circuit with L = 4, R = 80,
15. 4X 'l + 5x' + ~x = 0 C= 0.08,andE(t) = 120e-'sint.(Q(0) = 1O-6 cau_
lomb, 1(0) = 0 amps.)
16. !x" + x' + ix = 0
27. Solve the two-loop L-R-C circuit with L = I, R, =
*17. !x" + Sx' + 2Sx = 0 R2 = 40, C = 0.004, and E(t) = 220 cos t. (Q(O) =
18. ~Xlf + tx' + ¥x = 0
o coulombs, 12(0) = 0 amps.) where the initial amounts of x and y are given with the initial conditions x(O) = Xo and
28. Solve the two-loop L-R-C circuit with L = I, R, = yeO) = Yo. (Notice that the amount of the substance divided by the volume is the con-
19. !x" + 5x' + ¥x = 0
40, R z = 80, C = 0.004, and E(t) = IS0e-' cost. centration of the solution.)
20. x" + 6x' + 9x = 0 (Q(O) = 10- 6 coulomb, 12(0) = 0 amps.)
In this system, ify(t)1V2 > x(t)/Vlo is dxldt > 0 or is dxldt < O? Also, is dyldt> 0 or
*21. Solve Exercises 13, IS, and l7 with the initial condi- 29. Use the system derived in Exercise 6 to solve the three- is dyldt < O? Using these results, does the substance move from the side with a
tions x(O) = 1, x'(O) = yeO) = o. loop circuit shown in Figure 7.6 if R, = R2 = 2, L, = lower concentration to that with a higher concentration, or is the opposite true?
L2 = L3 = 1, C = 1, E(t) = 90, and the initial condi-
22. Solve Exercises 17-20 with the initial conditions tions are Q(O) = 0, 12(0) = 13(0) = O. Find I, (t) and
x(O) = 0, x'(O) = yeO) = 1. determine lim,_oo Q(t), limHoo I,(t), limHoo 12(t), and
23. Find the eigenvalues for the spring-mass system

lim,_oo I,(t). "f',,','.'
30. Solve the system of differential equations to fwd Suppose that two salt concentrations of equal volume V are separated by a mem-
the displacement of the spring-mass system if m = I, brane of permeability P. Given that P = V, determine the amount of salt in each
e = I, and k = 112. Graph several solutions in the concentration at time t if x(O) = 2 and yeO) = 10.
phase plane for this system. How is the equilibrium
point (0, 0) classified?
: 'J Solution In this case, the initial-value problem that models the situation is
How do these values relate to overdamping, critical *31. Solve the system of differential equations to rmd the
damping, and underdamping? displacement of the spring-mass system given thai
m = 1, e = 2, and k = 3/4. Graph several solutions in x(O) = 2, yeO) = 10.
24. (a) Find the equilibrium point of the spring-mass sys- the phase plane for this system. How is the equilib·
tem rium point (0, 0) classified?
...
The eigenvalues of A = (- ~ _ ~) are A1 = 0 and A2 = - 2. Corresponding eigen.

vectors are found to be Vl = G) and v; = (- n, so a general solution is Similarly, we find dyldt to be
dy Rx Ry
= (;~h) = ClG) + C2( - = (~: ~ ~~:=~)
2t
y X(t) De- dt"=V;- V2 ·
10
= (~: ~ ~~) = (l~), we have that Cl = 6 and C2 = 4, so the solu-

We use the initial conditions x(O) = Xo and yeO) = Yo to solve the nonhomogeneous
Because X(O) In deriving this system of system
equations, we used rates in
tion is
dx Rx
l
.allmin and concentrations in
-=RC--
(;~~D = (66 ~ ~':-2~)
ib/gal. In general, rates are dt Vl
4 X(t) = given by (volume of
2
We graph this solution parametrically in Figure 7.9(a). We then graphx(t) andY(I)
liquid)/(time) and concentra-
tions are given by (amount of
d:=~-~
together in Figure 7 .9(b). Notice that the amount of salt in each concentration ap- salt)/(volume of liquid). for X(I) and yet).
2 4 6 8 x
proaches 6, which is the average value of the two initial amounts.
(a)
l-
In Example 1, ijx(O) = Xo and yeO) = Yo, does limt_oo x(t) = limt_oo y(t) = "ii"i,'.,
10
1(xo + Yo)? Determine the amount of salt in each tank in Figure 7.10 if Vl = V2 = 500 gal,
R = 5 gal/min, C = 3 Ib/gal, Xo = 50 Ib, and Yo = 100 lb.
y(t)
Mixture Problems
:. Solution In this case, the initial-value problem is
Consider the interconnected tanks that are shown in Figure 7.10 in which a salt solu-
x(t)
4
tion is allowed to flow according to the given information. Let x(t) and y(l) represent
the amount of salt in Tank 1 and Tank 2, respectively. Using this information, we set
2 up two differential equations to describe ·the rate at which x and y change with respect x(O) = 50, yeO) = 100,
to time. Notice that the rate at which liquid flows into each tank equals the rate at which.
4 8 t it flows out, so the volume ofliquid in each tank remains constant. Ifwe consider Tank
= (-~ _~) X + C~) = AX + F(t), X(O) = (l~~).

(b) 1, we can determine a first-order differential equation for dxldt with
which in matrix form is X'
Figure 7.9 (a)-(b). : = (Rate at which salt enters Tank 1) - (Rate at which salt leaves Tank 1),
The matrix A has the repeated eigenvalue Al = A2 = -11100, for which we can
where the rate at which salt enters Tank 1 is R gal/min x C Ib/gal = RC lb/min, and
the rate at which it leaves is R gal/min x xiVl lb/gal = RxIVl lb/min, where XIVl is the find one (linearly independent) eigenvector Vl = (~). Therefore, one solution to the
salt concentration in Tank 1. Therefore,
corresponding homogeneous system X' = AX is X 1(t) = (~)e-t/lOO, and a second
solution is X 2(t) = [(nt ]
+ CO~) e-tll00, so
X h (t) = 0)
Cl ( 1 e
-tIlOO + C2
[(0)
1 1+ e°
(100)] -1/100 ( 100C2e-tll00 )
= Cle-tiIOO + c2te-I/lOO .
Rate: ' -_ _ _ _-" Notice that F(t) = C~) is not contained in Xh(t), so with the method of undeter-
R
Figure 7.10 mined coefficients we assume a particular solution has the form Xp(t) = a = (:~)
I,jl. __
and substitute into the nonhomogeneous system X' = AX + F(t). This yields dx
dt = alx - a2x + b 2y = (al - a2)x + b2y,
(o0) = (-1/100
11100
and the rate at which population Y changes is
dy
dt = bly - b2y + azx = (b l - b2)y + a2 x .
with solution al = 1500 and az = 1500. Therefore, Xp(t) = G;~~) and
Therefore, the system of equations that must be solved is
X(t)
100c e-tlIOO + 1500
= Xh(t) + Xp(t) = ( cle- dlOO ~ czte-tllOO + 1500
)
. : = (al - az)x + b2y
1
dy ,
- - -- - dt = a2X + (b l - b2)y
(lOc~C~ ~5~~0) = (I~~),
1500 -::.;-==--=~-
1400 Application of the initial conditions then gives us X(O) =
so CI = -[400 and Cz = -[450/100 = -29/2. The solution to the initial-value where the initial populations of the two territories x(O) = Xo and yeO) = Yo are given.
problem is
X(t) = (x(t)) = (.;

t
-1450e- ;;o + 1500 ) l'fi,,',IMI
yet) -1400e-tIIOo - Tte-dlOO + 1500 .
Determine the populations x(t) and yet) in each territory if al = 5, az = 4, b l = 5,
200
and b z = 3, given that x(O) = 14 and yeO) = 7.
In Figure 7.11, we graphx(t) andy(t) together. Notice that each function approaches
100 200 300 400 500 600 c y(c)
a limit of 1500, which means that the amount of salt in each tank tends toward a 140 : J Solution In this case, the initial-value problem that models the situation is
Figure 7.11 value of 1500 lb. 120 X(c)
100 ~= x + 3y
In Example 2, is there a value of t for which x(t) = yet)? If so, what is this value? 80 t ,x(O) = 14 and yeO) = 7.
Which jUnction increases most rapidly hI' smaller values of t? 60
40
1
: = 4x + 2y
Population Problems
In Chapter 3 we discussed population problems that were based on the simple princi-
20 The eigenvalues of A = (! ;) are Al = -2 and Az = 5. Corresponding eigen-
ple that the rate at which a population grows (or decays) is proportional to the num-
ber present in the population at any time t. This idea can be extended to problems in-
0.1 0.2 0.3 0.4 0.5 c
vectors are found to be VI = (_ D and V2 = (!), so a general solution is
volving more than one population and leads to systems of ordinary differential
equations. We illustrate several situations througb the following examples. Note that in; ,
X(t))
X(t) = ( y(t) = CI -I
(I) e
-2t (3) 5t _ ( cle- 2t + 3c2e5t )
+ C2 4 e - -cle- 2t + 4c2e51 .
each problem we determine the rate at which a population P changes with the equa-
tion Application of the initial condition X(O);" (;~~) = e~) yields the system
dP = (rate entenng)
dt . - (rate I ' )
eavmg. CI + 3cz = ~4 , so CI = 5 and Cz = 3. Therefore, the solution is X(t) = (X((tt))) =
{ -CI + 4cz - 7 Y
5t
We begin by determining the population in two neighboring territories where the .', 5e ~Z:, + ge 5t). We graph these two population functions in Figure 7.12(a). In
populations x and y of the territories depend on several factors. The birth rate of x is ( -5e + I 2e
aJ, and that ofy is b l . The rate at which citizens ofx move to y is a2, and that at which .. Figure 7 .12(b), we graph several solutions to the system of differential equations for
citizens move from y to x is b 2 • After assuming that the mortality rate of each territory various initial conditions in the phase plane. As we can see, all solutions move away
80 100 X
is disregarded, we determine the respective populations of these two territories for any
time t.
from the origin in the direction of the eigenvector Vz = (!), which corresponds to
Using the simple principles of previous examples, the rate at which population the positive eigenvalue Az = 5.
x changes is
404 Chapter 7 Applications of Systems of Ordinary Differential Equations
7.2 Diffusion and Population Problems with First-Order Linear Systems 405
Population problems that involve more than two neighboring populations can be
solved with a system of differential equations as well. Suppose that the population of "
three neighboring territories, x, y, and z, depends on several factors. The birth rates of
x, y, and z are aj, bj, and Cj, respectively. The rate at which citizens of x move to y is
o "fF,,'·'.' Determine the population of the three territories if al = 3, a2 = 0, a3 = 2, b 1 = 4,
b 2 = 2, b 3 = I, Cl = 5, C2 = 3, and C3 = 0 if x(O) = 50, y(O) = 60, and z(O) = 25.
a2, and that at which citizens move from x to z is a3' Similarly, the rate at which citi_
zens of y move to x is b 2 , and that at which citizens move from y to z is b3 . Also, the
rate at which citizens of z move to x is C2, and that at which citizens move from z to y J Solution In this case, the system of differential equations is
is C3" (This information is summarized in Table 7.2.) If the mortality rate of each ter-
ritory is ignored in the model, we can determine the respective populations of the three
territories for any time t.
: = x + 2y + 3z
dy
Tt=y
1
TABLE 7.2 dz
dt = 2x + Y + 2z.
From To Birth Rate
Because the characteristic polynomial is
x
1 y z
x -
JI
a2 aj' at 2
I-A 30 I
=(I-Ai(2-A)+3(-2)(l-A)=
y b2 - b3 bt 1 2- A
i
(I - A)(A + I)(A - 4) = 0,
Z -
C2
I c) C1
the eigenvalues are A1 = 4, A2 = I, and A3 = -I, with corresponding eigenvectors
The system of equations to determine x(t), y(t), and z(t) is similar to that derived
in the previous example. The rate at which population x changes is
VI = G} -D,
Vz =( and V3 = (-~} respectively. A general solution is
X(t) = G~) = C1G)e
4t
+ C2( -De + c3CDe-
t t
=
and the rate at which population y changes is cle4t + C2et - 3C3e-t)

- 6c2 et .
( Cl e4t + 4c2et + 2c3e-t
400 Cl + C2 - 3C3 = 50
and that of z is Using the initial conditions, we find that -6C2 = 60, so Cl = 63,
{
300 Ct + 4C2 + 2C3 = 25
C2 = -10, and C3 = 1. Therefore, the solution is

200
t
X(t)) (63e4t - 10e: - 3e- )
100 X(t) = y(t) = 60e .
Y(I) ( z(t) 63e 4t - 40e t + 2e- t
0"1 0.2 0.3 0.4 05 I

We graph these three population functions in Figure 7.13. Notice th~t although pop-
ulation y is iIritially greater than popUlations x and z, these populatIOns mcrease at
Figure 7.13 a much higher rate than does y.
"-
In Example 4, does population y approach a limit or do all three populations in-
where the initial popUlations x(O) = Xo, y(O) = Yo, and z(O) = Zo are given. crease exponentially?
7.2 Diffusion and Population Problems with First-Order Linear Systems
EXERCISES 7.2
- 18. Solve the problem described in Exercise 17 using the

unequal volumes V, = 100, V2 = 50, and V3 = 25.
How do the limiting values of x(I), yet), and z(t)
differ from those in Exercise l7?
30. In Exercise 27, what is the half-life of substance XI
31. In the reaction described in Exercise 27, show that if
k units of X are added per year and h units of Z are
removed, then the situation is described with the
In Exercises 1-6, solve the diffusion problem with one per-
meable membrane with the indicated initial conditions and x(l) D
Rale:R2 - - -
y(t)
In Exercises 19-22, solve the initial-value problem using the
system
l
parameter values. Find the limiting concentration of each
Volume: VI Volume: V2 given parameters to find the population in two neig~borin~ ter- dx = -ax +k
Concentration Tank 1 Tnnk2 dt
solution. C Rate
ntories. Do either of the populations approach a fmlte hm,t? If
--Rate:RJ RI so, what is the limit? dY=ax-bu
1. P = 0.5, V, = V2 = 1, x(O) = 1, yeO) = 2 dt "
19. a, = lO,a2 = 9,b, = 2,b2 = l;x(O) = 10,y(0) = 20
2. P = 0.5, V, = V2 = 1, x(O) = 2, yeO) = 1 Figure 7.14 dz =by-h
20. a, = 4, a2 = 4, b, = 1, b2 = 1; x(O) = 4, yeO) = 4 dt
*3. P = 2, V, = 4, V2 = 2, x(O) = 0, yeO) = 4
14. Solve the tank problem described in Exercise 13 us- *21. a, = 2, a2 = 0, b, = 2, b2 = 3; x(O) = 5, yeO) = 10
where a, b, k, and h are positive constants.
4. P = 2, V, = 4, V2 = t x(O) = 8, yeO) = 0
s. p = 6, V, = 2. V2 = 8, x(O) = 4, yeO) = 1
ing the initial conditions x(O) = 0 and yeO) = 6. For
how many values of I do the functions x(t) and yet)
22. a, = 1, a2 = 1, b, = 1, b, = 1; x(O) = 5, yeO) = 10
32. Solve the system described in Exercise 31 if a b, *
x(O) = xo, yeO) = Yo, and z(O) = zoo Find limt-oo x(t)
6. P = 6, V, = 2, V, = 8, x(O) = 1, yeO) = 4 agree? In Exercises 23-26, solve the initial-value problem using the
and lim,_oo Y(I). How do these limits differ from
*15. Find the amount of salt in each tank in Figure 7.15 if given parameters to find the population in three neighboring
those in Exercise 27? If k = h, then find lim,_oo Z(I).
In Exercises 7-8, use the tanks shown in Figure 7.10 with .. : R = 1 gal/min, e = lib/gal, V, =; 1 gal, V, = ) ga~ territories. Which population is largest at I = I?
If k > h, then find lim,_oo z(t). Describe the corre-
R = 4 gal/min, e = 4 Ib/gal, V, = V, = 20 gal, and the given x(O) = 2, andy(O) = 4. What is the maximum amount 23. a, = 10, a2 = 6, a3 = 7, b, = 6, b2 = 3, b3 = 3, sponding physical situation.
initial conditions. (a) Determine x(l) and y(I). (b) Find of salt (at any value of t) in each tank? Find
lim,_oo x(t) and lim,_oo yet).
c, = 5,c2 = 7,C3 = l;x(O) = l7,y(0) = O,z(O) = 34
lim,_oo x(l) and lim,_ooy(I). (e) Does one of the tanks contain 24. a, = 2, a, = 1, a3 = 4, b, = 6, b2 = 4, b3 = 5, In Exercises 33-36, solve the radioactive decay model in Ex-
more salt than the other tank for all values of t? c, = 2, C2 = 8, C3 = 4; x(O) = 0, yeO) = 4, z(O) = 2 ercise 27 with the given parameter values and initial conditions.
D
ConCenlr.ltioD
r--~~-~-:~R-_~-_~-_-_------' C If z(t) represents the amount (grams) of substance Z after I hours,
*7. x(O) = yeO) = 0 *25. a, = 7, az = 2, a3 = 4, b, = 7, b2 = 5, b3 = 8,
how many grams of Z are eventually produced?
x(l) y(,) c, = 7, C2 = 1, C3 = 2; x(O) = 8, yeO) = 2, z(O) = 0
8. x(O) = 0, yeO) = 2 Volume: VI Volume: V2
Tank 1 Tank 2 26. a, = 7, a2 = 2, a3 = 4, b, = 7, b2 = 5, b3 = 8, 33. a = 6, b = 1, Xo = 7, Yo = I, =0 = 8
9. Suppose that pure water is pumped into Tank I (in Fig- ~'e C, = 7, C, = 1, C3 = 2; x(O) = 0, yeO) = 0, z(O) = 16 34. a = 4, b = 2, Xo = 10, Yo = 2, Zo = 4
ure 7.10) at a rate of 4 gal/min (that is, R = 4, e = 0) - - Rate:R R
and that V, = V2 = 20 gal. Determine the amount of 27. Suppose that a radioactive substance X decays into an- 35. a = 4, b = 2, Xo = 1, Yo = I, Zo = 1
salt in each tank at time I if x(O) = yeO) = 4. Calcu- Figure 7.15 other unstable substance Y, which in turn decays into
36. a = 1, b = 4, Xo = 2, Yo = 2, Zo = 2
late lim,_oo X(I) and lim,_oo y(I). Which function de- a stable substance Z. Show that we can model this sit-
creases more rapidly? 16. Determine the amount of salt in each tank described uation through the system of differential equations
In Exercises 37-40, solve the radioactive decay model in Ex-
10. If x(O) = yeO) = 0 in Exercise 9. how much salt is in in Exercise 15 using the initial conditions x(O) = 4 and dx ercise 31 with a = 1, b = 1, and the given parameter values and
l
-;::: -ax
each tank at any time I? yeO) = O. How do these functions differ from those in dl initial conditions. Describe what eventually happens to the
Exercise l5?
*11. Use the tanks in Figure 7.10 with R = 5 gal/min,
e = 3 Ib/gal, V, = 100 gal, Vz = 50 gal, and the ini- 17. Consider the three tanks in Figure 7.16 in which the It
d ..
= ax - by, where a and b are posItive constants.
amount of each substance.
37. k= 2,h = l,xo = 2,yo= l,zo= 2
amount of salt in Tanks 1, 2, and 3 is given by x(t),
tial conditions x(O) = yeO) = O. (a) Determine X(I) dz 38. k = 0, h = 10, Xo = 4, Yo = 2, =0 = I
y (t), and Z(I), respectively. Find x(t), y (t), and z(t) dt = by
and y(I). (b) Find limHoo X(I) and lim,_oo y(I).
if R = 5 gal/min, e = 2 Ib/gal, V, = V, = V3 = *39. k = 10, h = 0, Xo = 8, Yo = 2, Zo = 2
12. .solve Exercise 11 with V, = 50 and V, = 100. How 50 gal, and x(O) = yeO) = =(0) = O. Find limH~x(I), (Assume that one unit of X decomposes into one unit
do the limiting values ofx(l) andy(l) compare to those
40. k = 0, h = 5, Xo = 1, Yo = 10, Zo = 5
lim,_oo yet), and lim,_oo z(t). of Y, and one unit of Y decomposes into one unit of
in Exercise 11? *41. Solve the initial-value problem to find the concentra-
Z.)
tion of a substance on each side of a permeable mem-
13. Consider the tanks shown in Figure 7.14, where R, = 28. Solve the system of differential equations in Exercise
3 Limin, R, = 4 Limin, R3 = 1 Limin, e = 1 kglL, *
27 if a b, x(O) = Xo, yeO) = Yo, and z(O) = zoo Find
brane modeled by the system
V, = V, = 50 L. If x(O) = yeO) = 0, determine the

amount of salt in each tank at time t. Find lim,_oo X(I)
limt _ oo x(t), lim,_oo yet) and lim,_oo =(t).
*29. Solve the system in Exercise 27 with x(O) = Xo, ~ = pu, - ;J, x(O) = a, yeO) = b.
and limt-oo y(I). Is there a time (other than 1= 0) at
which each tank contains the same amount of salt? Figure 7.16
yeO) = yo, and z(O) = =0, if a = b. Find lim,_oo x(t),
lim,_oo y(t), and lim,_oo =(t). 1 dy P (Xv;- - VY)
di= z
408 Chapter 7 Applications of Systems of Ordinary Differential Equations 7.3 Nonlinear Systems of Equations 409
(a) Find lim,_~ x(t) and limH~ yet). taneously. Describe the effect that the value of P has populations indicates that a predator overtakes a member of the prey population. To in-
(b) Detennine a condition so that x(t) > yet) as on the corresponding solution. clude these interactions in the model, we assume that the number of interactions is di-
t --> "'. When does lim,_~ x(t) = limt-~ yet) as *43. Solve the initial-value problem rectly proportional to the product of x(t) and yet). Therefore, the rate at which x(t)
t~ oo? changes with respect to time is
l
tb:
(e) Find a condition so thatx(t) is an increasing func- dt = (al - a2)x + bly
tion. Find a condition so that yet) is an increasing
function. Can these functions increase simultane-
dy , x(O) = xo, yeO) = Yo
.
dx
di = ax - bxy,
dt = a2 x + (b l - b2)y
ously? where b > O. Similarly, the rate at which y(t) changes with respect to time is
(d) If VI = V2 and a = b, describe what eventually Are there possible parameter values so that the func-
happens to x(t) and y (t). tions x(t) andy(t) are periodic? Are there possible pa- dy
rameter values so that the functions x(t) and yet) ex-
di = -cy + dxy,
42. Investigate the effect that the membrane penneability
has on the diffusion of a substance. Suppose that perience exponential decay? For what parameter
where d> O. These equations and initial conditions form the Lotka-Volterra prob-
VI = V2 = I, x(O) = I, yeO) = 2, and (a) P = 0.25, values do the two populations experience exponential
growth? lem
(b) P = 0.5, (e) P = 1.0, and (d) P = 2.0. Graph the
dx
ax - bxy
l
solution in each case both parametrically and simul- Alfred James Lotka (1880-
1949), American biophysicist,
di =
born in Ukraine; wrote first text dy , x(O) = Xo, yeO) = Yo·
on mathematical biology. di = -cy + dxy
~ Nonlinear Systems of Equations
(UP//Bellman)
r Biological Systems: Predator-Prey interaction

r Physical Systems: Variable Damping
Find and classifY the equilibrium points of the Lotka-Volterra system.
Several special equations and systems that arise in the study of many areas of applied
mathematics can be solved using the techniques of Chapter 6. These include the
predator-prey popUlation dynamics problem, the Van-der-Pol equation that models : ] Solution
. {ax -+ bxy
Solvmg -0'
x(a - by) = 0
= __
dxy = y( -c + dx) = 0' we have x - 0 or y - alb and
variable damping in a spring-,masssystem, and the Bonhoeffer-Van-der-Pol (BVP)
oscillator. y = 0 or x = c/d. The equilibrium points are (0, 0) and (c/d, alb). The Jacobian
Biological Systems: Predator-Prey Interaction

matrix of the nonlinear system is J(x, y) = (a ~/y _;!X dx). Near (0,0), we
Let x(t) and y(t) represent the number of members at time t of the prey and predator
haveJ(O,O) = (~ ~c) with eigenvalues At = -candA2 = a. Because these eigen-
populations, respectively. (Examples of such populations include fox-rabbit and shark- values are real with opposite signs, we classifY (0, 0) as a saddle. Similarly, near
seal.) Suppose that the positive constant a is the birth rate of x(t), so that in the ab-
sence of the predator
near (c/d, alb), we have J(c/d, alb) = (a~/b -b~/d), with eigenvalues AI ,2 =
~ i V;;;, so the point (cld, alb) is classified as a center.
-=ax
dx
dt
In Figure 7.17, we show several solutions (which were found using different
initial populations) parametrically in the phase plane of this system with a = 2,
and that c > 0 is the death rate of y, which indicates that b = I, c = 3, and d = 1. Notice that all of the solutions oscillate about the center.
These solutions reveal the relationship between the two popUlations: prey, x(t), and
dy
di= -cy predator, yet). As we follow one cycle counterclockwise beginning, for example,
near the point (3/2, 1) (see Figure 7.18), we notice that as the prey population, x(t),
in the absence of the prey population. In addition to these factors, the number of in- increases, the predator population, yet), first slightly decreases (is that really possi-
4 10 12 x ble?), and then increases until the predator becomes overpopulated. Then, because
teractions between predator and prey affects the number of members in the two popu-
lations. Note that an interaction increases the growth of the predator population and Figure 7_17 Typical solutions the prey population is too small to supply the predator population, the predator pop-
decreases the growth of the prey population, because an interaction between the two of the Lotka-Volterra system. ulation decreases, which leads to an increase in the population of the prey. At this
point, because the number of predators becomes too small to control the prey po _
ulation, the number in the prey population becomes overpopulated and the cycle r~_
peats itself.
x(t) YCt)
:~
(Prey) (Predator)
6
\!~!~t~!t~
x
4
2
2 4 6 8 10 12 14 x 2 4 6 8 10 12 14 t
Ca) (b)
(b) (e)
Figure 7.18 Ca)-(b).
'~
I Solution We find the equilibrium points by solving

How does the period of a solution with an initial point near the equilibrium point
(c/d, alb) compare to that of a solution with an initial point that is not located near y=O
this point? { -x - /-L(x 2 - I)y = O·
From the first equation, we see that y = O. Substitution of y = 0 into the second
Physical Systems: Variable Damping
equation yields x = 0 as well. Therefore, the equilibrium point is (0, 0).
In some physical systems, energy is fed into the system when there are small oscilla- The Jacobian matrix for this system is
tions, and energy is taken from the system when there are large oscillations. This in-
dicates that the system undergoes "negative damping" for small oscillations and "pos-
itive damping" for large oscillations. A "differential equation that models this situation
J(x, y) = (-1 -\.I.X}' - /-L(X~ 1»)' -
(d)
is Van-der-Pol's equation At (0, 0), we have the matrix
Figure 7.19 Ca)!1- = ~
+ /-L(x 2 - I)x' + X = 0,
x" (b)!1-= 1 CC)!1-=~
(d)!1- = 3.
0
J(O, 0) = ( -1 ~,).
r-
where /-L is a positive constant. We can transform this second-order differential equa-
tion into a system of first-order differential equations with the substitution x' = y. We find the eigenvalues of J(O, 0) by solving
2
Hence y' = x" = -x - /-L(x - I)x' = -x - /-L(x 2 - I)y, so the corresponding sys-
tem of equations is
I-A
-1
I 1= A2 -
/-L - A
/-LA +1 =0,
X' = y
Balthasar Van-der-Pol, Dutch which has roots AI ,2 = (/-L ± V /-L2 - 4)/2, Notice that if /-L
> 2, then both eigen-
{ y' = -x - /-L(x 2 - I)y'
applied mathematician and engi- values are positive and real, so we classifY (0, 0) as an unstable node. On the other
neer. (UPIIBettman) which is solved using an initial position x(O) = Xo and an initial velocity yeO) = x'(O) hand, if 0 < /-L < 2, the eigenvalues are a complex conjugate pair with a positive
2
= Yo· Notice that /-L(x - 1) represents the damping coefficient. This system models real part, Hence (0, 0) is an unstable spiral point. (We omit the case when /-L = 2
variable damping because /-L(x 2 - 1) < 0 when -1 < x < 1 and /-L(X2 - 1) > 0 when because the eigenvalues are repeated,)
r:1 > I. Therefore, damping is negative for the small oscillations, - 1 < x < 1, and pos- In Figure 7.19, we show several curves in the phase plane that begin at dif-
itive for the large oscillations, Ixl > 1. ferent points for various values of /-L, In each figure, we see that all of the curves
approach a curve called a limit cycle. Physically, the fact that the system has a limit
cycle indicates that for all oscillations the motion eventually becomes periodic, which
"f',,','. is represented by a closed curve in the phase plane,
On the other hand, in Figure 7.20 we graph the solution that satisfies the ini-
Find and classifY the equilibrium points of the system of differential equations that tial conditions x(O) = 1 and yeO) = 0 parametrically and individually for various
is equivalent to Van-der-Po!'s equation. values of /-L, Notice that for small values of /-L the system more closely approximates
that of the ~armonic oscillator because the damping coefficient is small. The curves
are more cITcular than those for larger values of f.L. 4 4
'- yet)
y y -4
4 4 4
-4 -4 yet)
-4 (j) (k) (1)

-4 4 x
Figure 7.20 continued (j) f.L = 2 (k)-(I) f.L = 3
-4 -4 -4
Graph several solution curves in the phase plane for the Van-der-Pol equation if
(a) (b)
0~, f.L = 0.1 and f.L = 0.001. Compare these graphs to the corresponding solution curves
(c) ~~:I'
for the equation x" + x = O. Are they similar? Why?
y
4 4 4
EXERCISES 7.3
(t)
-4 ot
1. The phase paths of the Lotka-Volterra model c,
b, = = C2 = 1, locate and classify the three equi-
librium points. (b) If a2 = 2, a, = b, = c, = C2 = 1,
l
dx=ax-bXY classify the two equilibrium points.
-4 -4 -4 dl
*3. The national economy can be modeled with a nonlin-
(d)
dy = _cy + dxy ear system of diffurential equations. If I represents the
(e) (I) dl national income, C the rate of consumer spending, and
. dy dy/dl -cy + dxy . G the rate of government spending, a simple model of
are gIven by dx = dx/dl = ax _ bxy . Solve thIS the economy is
y
Y
4 4 separable equation to find an implicit equation of the
l
phase paths. EI.
dl
= 1- aC
2. Consider the predator-prey model dC
dt = b(l - C - G),
~
I
= (a, - b,x - c,y)x,
-4 -4 4 x where a and b are constants. (a) Suppose that a =
dy b = 2 and G = k (k = constant). Find and classify the
dt = (-a2 + C2 X )y
equilibrium point. (b) If a = 2, b = I, and G = k, f'md
-4 where x(t) represents the prey population, y (I) the and classify the equilibrium point.
predator popu1ation, and the constants a\, az, bi! ell 4. If a = b = 2 and G = k, + k2! (k2 > 0), show that
(g) (b) (i) C2 are all positive. In this model, the term -b,x
2 there is no equilibrium point if k2 2: ! for the national
represents the interference that occurs when the prey economy model in Exercise 3. Describe the economy
Figure 7.20 (a)-(b) f.L =± (c)-(d) f.L =! (e)-(f) f.L =1 (g)-(h) f.L =~ (i) f.L = 2 popUlation becomes too large. (a) If a, = 2, a2 = under these conditions.
----~.----
414 7.3 Nonlinear Systems of Equations 415
Chapter 7 Applications of Systems of Ordinary Differential Equations
5. The differential equation x" + sin x = 0 can be used 17. In a mechanical system, suppose that x(t) represents
l
(c) Show that this system has solution dx = xy2
to describe the motion ofa pendulum. In this case, X(I) position at time t, K kinetic energy, and V potential
r(1) = lfY I + ae -2', 8(1) = I + b so that (c) dl energy, where K = 1/2 m(x) (dxldt)2 (m is a positive
represents the displacement from the position x = O.
X(I) = UYI + ae- 2, cos (I + b), dy = x2 + Sy function) and V = Vex). If the system is conservative,
Represent this second-order equation as a system of dl
first-order equations. Find and classifY the equilibrium y(l) = IIVI + ae- 2t sin (I + b). then the total energy E of the system remains constant
points of this system. Describe the physical signifi- (d) Calculate lim,~oox(l) and Iim,_ooy(l) to show that 14. Let E and s be positive constants and suppose thatf during motion, which indicates that 1/2 m(x)(dxldt)2 +
cance of these points. Graph several paths in the phase all solutions approach the circle x 2 + y2 = I as is a continuous odd function that approaches a finite Vex) = E. Show that with the cbange of variable u =
plane of the system. 1-> 00. Thus x 2 + y2 = I is a limit cycle. limit as x ---+ 00, is increasing, and is concave down for
x> O. The voltages over the deflection plate in a
I V;;;W dx this equation becomes d 2uldt 2 + Q'(u) =
6. Show that tbe paths in the phase plane of x" + *11. Consider Lienard's equation d'xldP + f(x) dxldt + 0, where Q'(u) = V(x)/V;;;W. This means that an
sweeping circuit for an oscilloscope are determined by
sin x = 0 satisfY the first-order equation dyldx = g(x) = 0, wheref(x) and g(x) are continuous. Show equation of the form d 2x/dt 2 = f(x) is a conservative
that this equation can be written as the system solving
-(sin x)ly. (See Exercise 5.) Use separation of vari- system, wbere f is a function representing force per
ables to show that paths are t y2 - cos X = e, where
~ =y -F(x)
unit mass that does not depend on dxldt.
j l
dV,
e is a constant. Graph several paths. Do these graphs dt = -sV, + feE - V2 ) 2 2
agree with your result in Exercise 5? 18. Consider the conservative system d xldt = f(x) in
dy dV2 whichfis a continuous function. Suppose that Vex) =
*7. (a) Write the equation d 2xldl 2 + k 2x = 0, which mod- dl = -g(x), dt = -sv2 + feE - V,).
els the simple barmonic oscillator, as a system of -I f(x) dx so that V'(x) = -f(x). (a) Use the substi-
first-order equations. (b) Show that paths in the phase
plane satisfY dyldx = -k 2xly. (See Exercise 6.)
w,here F(x) = f feu) duo Use Bendixson's theorem (see Exercise 13) to show
that this system has no limit cycles.
tution dxidt = y to transform d 2x/d/ 2 = f(x) into the
(c) Solve the equation in (b) to find the paths. (d)

What is the equilibrium point of this system and how
is it classified? How do the paths in (c) compare with
12. (See Exercise Il.) Lienard's theorem states that if
(i) F(x) is an odd function, (ti) F(x) is zero only at
x = 0, x = a, x = -a (for some a > 0), (iii) F(x) ...
*15. Consider the relativistic equation for the central orbit
of a planet, d 2uldOZ + u - ku 2 = ", wbere k and"
are positive constants (k is very small), u = IIr, and r
system of first-order equations 1~ = y
dy = -Vex)
.
what you expect to see in the pbase plane? :too monotonically for x> a, and (iv) g(x) is an odd dt
function where g (x) > 0 for all x > 0; then Lienard's and 8 are polar coordinates. (a) Write this second-
8. Repeat Exercise 7 for the equation d 2xldl 2 - k 2x = O. (b) Find the equilibrium points of the system in (a).
equation has a unique limit cycle. Use Lienard's the- order equation as a system of first-order equations.
9. Suppose that a satellite is in flight on the line between
a planet of mass M, and its moon of mass M 2 , which
orem to determine which of the following equations «
(b) Show that I - v'1 - 4k,,)/2k, 0) is a center in What is the physical significance of these points?
(c) Use the chain rule and dx/dt = y to show that
has a unique limit cycle: the linearized system.
are a constant distance R apart. The distance x between dy/d/ = Y dy/dx. (d) Show that the paths in the phase
the satellite and the planet satisfies the nonlinear d'x dx plane are )y2 + Vex) = e.
(a) d? + e{l - x 2 )di + x = 0, e> 0;
. .
dx = P(x,y)
dl
second-order equation 16. Cons,der the system of equatIOns .
1
*19. (See Exercise IS.) What are the paths in the phase
dy
x" = _gM, + gM2 (b) d'x + 3x 2dx +x' = 0 di= Q(x,y), plane if Vex) = tx 2? How does this compare with the
dl' dl .
x2 (R-x),' classification of the equilibrium point of the corre-
where P and Q are polynomialS of degree n. Finding
13. Consider the system of autonomous equations sponding system of first-order equations? Notice that
where g is the gravitational constant. Transform this the maximum number of limit cycles of this system,
if V has a local minimum at x = a, the system has a
~ =f(x,y).
j
equation into a system of first-order equations. Find called the Hilbert number H., has been investigated
center at the corresponding point in the phase plane.
and classifY the equilibrium point of the linearized sys- as part of Hilbert's 16th problem. Several of these
tem. dy numbers are !mown: Ho = 0, HI = 0, H2 '" 4, Hn '" 20. (See Exercise LS.) What are the paths in the phase
dl =g(x,y) (n - 1)/2 (if n is odd), and Hn < 00. Determine the plane if Vex) = -!x'? How does this compare with
10. Consider the nonlinear autonomous system
Hilbert number (or a restriction on the Hilbert num- the cLassification of the equilibrium point of the cor-
Bendixson's theorem (or negative criterion) states
j
dx = _y + x(1 _ x2 _ y') ber) for each system: responding system of first-order equations? Notice
that if.fx(x, y) + gy(x, y) is a continuous function that
dt . that if V has a locaL maximum at x = a, the system
is either always positive or always negative in a par-
l~
l
dy dx=1O has a saddle at the corresponding point in the phase
di = x + y(1 - x' - y2) ticular region R of the phase plane, then the system dt (b) = x - Y
has no limit cycle in R. Use this theorem to determine (a) pLane.
dy
if the given system has no limit cycle in the phase dy =-5 di=y+1
(a) Show that (0, 0) is an equilibrium point of this 21. Use the observations made in Exercises 19-20 con-
dl
system. ClassifY (0, 0). plane. cerning the reLationship between the local extrema of
V and the classification of equilibrium points in tbe
(b) Show that x = r cos 8, y = y sin 8 transforms the
dx jdx- dx = y2 _ 2xy dx 7
j
di = x 3 + x + 7y = y - x di =x + 5y phase plane to classifY the equilibri urn points of a con-
system to the (uncoupled) system (a) (b) dt (c) d/ (d) .
1
2
dr/dt = r{l - r')
{ d8ldl= I .
j dy 2
dl=XY
dy
di=3x- y
dy
_=x2
dt
+ y2 dy 7
- = lOx +y
dt
4 servative system with (a) V'(x) = x - I; (b) Vex) =
X _x 3 ,
Chapter 7 Summary
22. Which of the following physical systems are conser-

(a) Find and classifY the equilibrium points of this SYs-
vative? (a) The motion of a pendulum modeled by
2 2 tem if l(t) = 0, a = c = I and b = 1. (b) Graph the
d x/dt + sin x = O. (b) A spring-mass system that
direction field associated with the system and then ap.
disregards dsmping and external forces. (c) A spring-
proximate the phase plane by graphing severa! solu_
mass system that includes damping.
tions near each equilibrium point.
*23. Consider a brake that acts on a wheel. Assuming that Population of Two Neighboring Territories
27. Find Xo > 0 s.o that the solution to the initial-valUe
the force due to friction depends only on the angular problem [ECtion 7_ 1 1
I
velocity of the wheel, dOldt, we have I d 20ldt 2 = dx = (al - a2)x + bly
- FR sgn (dOldt) to describe the spinning motion of X' = y L-R-C Circuit with One Loop dt
the wheel, where R is the radius of the brake drum, F y' = -x- (x 2 -l)y
{
dQ =1 dy = a2 x + (b l - b2)y
is the frictional force, I is the moment of inertia of the x(O) = Xo, yeO) = 0 dt
dt
I, x> 0
is periodic. Confinn your result graphically. dl R E(t)
wheel, and sgn (x) = 0, x = O. I
{ 71= - LC Q -y1+T Population of Three Neighboring Territories
-I, x<O
(a) Let dOldt = y and transfonn this second-order
28. We saw that solutions to the Latka-Volterra problem:
X'
{y'
= x(a - by), x(O) = Xo
+ dx), yeO) = Yo'
1 Q(O) = Qo, 1(0) = 10 dx = (a I
dt
- a2 - a3)x + b2y + C2Z
!
equation into a system of first-order equations. = y( -c
(b) What are the equilibrium points of this system? L-R-C Circuit with Two Loops dy = a2 x + (b l - b2 - b3)y + C3Z
oscillate periodically, where the period T and ampli.
(c) Show that d 20ldt 2 = dOldt didO (dOldt). (d) Use dt
tude of x(t) and yet) depend on the initial conditions.
the relationship in (c) to show that the paths in the dQ = __I_Q - 12 + E(t)
Volterra's principle states that the time averages of dt RIC RI dz = a3x + b3y + (CI - C2 - C3)Z
phase plane are I121(dOldt}' = -FRO + C, dOldt > 0 dt
x(t) and yet) remain constant and equal the corre-
and 1/2 I(dOldt)' = FRO + C, dOldt < O. What are
x
sponding equilibrium values, = cld and y = alb. In dI2 = Q
.J..
_ R2 I
these paths?
24. The equation d 2xldt 2 + k 2 sin x = 0 that models
the motion of a pendulum is approximated with
2 2
symbols, this statement is represented with
I
x=T iT I
ox(t)dtandY=T oy(t)dt. iT 1 dt LC L 2
Q(O) = Qo, 1,(0) = 10
1Section 7.31
d xldt + k2(X - 1/6 x 3) = O. Why? Write this equa-
Spring-Mass System
tion as a system, then locate and classifY its equilib-
rium points. How do these fmdings differ from those Follow the steps below to prove that y- = TI iTyet) dt
0 dx
Predator-Prey (Latka-Volterra system)
2
of d xldt 2 + k' sin x = O? di= y
I
. d dx = ax - bxy
25. Consider the system
X'
y'
= y - x(l - x 2 - y2 _ Z2)
= -x + y(1 - x 2 - y2 - Z2).
(a) Show that 71 (In x) = a - by.
(Hint: -it (In x) = f.) I dy = _.!..x _.E.. y

dt m m
dt
dy = _cy
dt
+ dxy
{
Z' = 0 T d
(b) Integrate each side of (
-0 71 (In x(t)) dt = Van-der-PoFs Equation (System)
(a) Show that Yr(x 2 + y2 + Z2) = 0 if x2 + y2 + 1Section 7.21 =y
rea - by(t)) dt to obtain In(x(T)) - In(x(O)) = x" + p.(X2 - I)x' +x =
X'
0; { y' = -x _ p.(X2 _ I)y
Z2 = I, 2
so x + y2 + z2 = I is called an invariant o
Diffusion through a Membrane
set. (b) Show that a solution with initial point aT - b rY(l) dt. Recall that because x(t) is peri-
o =p(~-~)
I
(xo, Yo, zo) remains in the plane z = zoo Therefore, dxl
z = Zo is an invariant set. (c) Show that the limit odic, x(T) = x(O). Use this fact to solve the pre· dt V2 VI
cycle for the solution with initial point (1/2, 1/2, 1/2)
is the circle x 2 + y2 = 3/4 in the plane z = I12.
vious equation for ~ fy(t) dt. dx2 =
dt
p(~
VI
_.2)V2
26. The Bonhoeffer-Van-der-Pol (BVP) oscillator is the (c) Follow steps similar to those outlined in (a) and
system of ordinary differential equations (b) to prove that x = TI iT0 x(t) dt. Mixture Problem with Two Tanks
: =x- ;3 -y + 1(1)
I
dx Rx
-=RC--
dt VI
:=~-~
+ a - by)
I : = c(x .
418 Chapter 7 Applications of Systems of Ordinary Differential Equations Differential Equations at Work 419
CHAPTER 7 REVIEW EXERCISES Experimental data indicate that the formation of The rate constants k, and k2 have been deter-
trichlorobenzene is small. Therefore we will neglect mined experimentally. We give these constants in the
it, so the rate equations are given by followingtable.* IfxA(O) = 1 andxB(O) = 0, solve the
1. Solve the one-loop L-R-C circuit with L = I henry,

R = 5/3 ohms, C = 3/2 farads, Q(O) = 10-· coulomb,
*11. Solve the three-dimensional population problem with
c,
-~: = k,XA
system for each of these three temperatures. Graph
XA(t) andxB(t) simultaneously for each temperature. Is
there a relationship between the temperature and the
1(0) = 0 amperes, and (a) E(t) = 0 volts; (b) E(t) =
e- r volts.
2. Solve the one-loop L-R-C circuit with L = 3 henrys,
a, = 2, a2 = I, a, = 4, b, = 6, b2 = 4, b, = 5, = 2
C2 = 8, C, = 4, x(O) = 4, yeO) = 4, and z(O) = 8. '
12. The nonlinear second-order equation that describes the
1 dx
dtB = k,XA - k2xB
'
where XA is the mole fraction (dimensionless) of ben-

time at which XB(t) > XA(I)?
40°C I 55°C 70°C

motion of a damped pendulum is d 2x/dl 2 + b dx/dt +
R = 10 ohms, C = 0.1 farad, Q(O) = 0 coulomh, zene and XB is the mole fraction of monochloroben-
k, (hr-') 0.0965 0.412 1.55
giL sin x = O. (a) Make the substitution dx/dl = y to zene. The formation of trichlorobenzene is neglected,
1(0) = 0 amperes, and (a) E(t) = 120 volts;
write the equation as the system so we assume that xc, the mole fraction of
(b) E(t) = 120 sin t volts. kz(hr-') 0.0045 0.055 0.45
dichlorobenzene, is found with Xc = I - XA - XB' ~?'Il'"'r£"
I .......,
j
*3. Solve the two-loop L-R-C circuit with L = I henry, dx
R, = R2 = 1 ohm, C = 1 farad, Q(O) = 10-· cou- Tt=y
lomb; 12 (0) = 0 amperes, and (a) E(t) = 0 volts;
(b) E(t) = 120 volts.
dy = -by _.!I sin x·
dt L
4. Solve the two-loop L-R-C circuit with L = 1 henry,
R, = R2 = I ohm, C = I farad, Q(O) = 0 coulombs, tb) Show that the equilibrium points of the system are I~rxf
. .. - D,if.ferent~al Equ'atio~s 'at Work:
I
12(0) = 0 amperes, and (a) E(t) = 120e -,/z volts;
(b) E(I) = 120 cos I volts.
(mT. 0), where n is an integer. (c) Show that if n is
odd, then (n7T, 0) is a saddle. (d) Show that if n is A.
,
Competing Species I
even, then (n7T, 0) is either a stable spiral point or a
5. Transform the second-order equation to a system stahle node. The system of equations
I
of first-order equations and classify the systems as
undamped, overdamped, underdamped, or critically
damped by finding the eigenvalues of the corre-
sponding system. Solve with the initial conditions
*13. Suppose that the predator-prey model is altered so
that the prey population X(I) follows the logistic equa.
tion when the predator population y (I) is not present.
: = x(a - b,x - b 2 y)
x(O) = I, dx/dl(O) = yeO) = O. (a) x" + 2x' + x = 0;

(b) x" + 4x = 0
6. Transform the second-order equation to a system
With this assumption, the system is
1d: = y(e - d1x - d 2 y),
where a, b" b2 , e, d" and dz represent positive constants, can be used to model the I
II
of first-order equations and classify the system as popUlation of two species, represented by X(I) and y (I), competing for a common food
undamped, overdamped, underdamped, or critically
supply.
damped by finding the eigenvalues of the corre-
sponding system. Solve with the initial conditions
where a, b, c, d, and k are positive constants, a k,'* I. (a) Find and classify the equilibrium points of the system if a = I, b l = 2, b~ =:'.1,
x(O) = 0, dx/dt(O) = yeO) = 1. (a) x" + 4x = 0; = 1 d = 0 75 and d = 2. (b) Graph several solutions by using dIfferent millal
(b) x" + 4x' + :Ix = 0
and the ratio a/kis much larger than c/d. Find and clas-
~opul~ti~ns p~a:netricahy in the phase plane. (e) Find lim,_ro x(t) and lim,_ro y(t)
*7. Solve the diffusion prohlem with one permeable
sify the equilibrium points of this system.
14. (production of Monochlorobenzene) The Ajax Phar-
if both x(O) and yeO) are not zero. Compare your result to (b). I
membrane with P = 0.5, V, = V, = I, x(O) = 5, and
yeO) = 10.
maceutical Company has often thought of making its
own monochlorobenzene C6 Hs CI from benzene CoH.
2. (a) Find and classify the equilibrium points of the system if a = I,. b l ~ I, b2 ~ ~,
e = 0.67, d l = 0.75, and d z = I. (b) Graph several solutions by.usmg dIfferent ffil- I
8. Solve the mixture problem with two tanks (see Figure
7.'10), with R = 4 gal/min, V, = V2 = 80, C =
and chloride CI2 (with a small amount of ferric chlo·
tial populations parametrically in the phase plane. (e) Determme the fate of the
species with populationy(t). What ha:ppens to the species with ~opulation ~(I)? Wh~t I
I
ride as a catalyst) instead of purchasing it from an-
3 Ib/gal, x(O) = 10, andy(O) = 20. happens to the species with population y (I) If the species With populatIOn x(t) .:'
other company. The chemical reactions are
suddenly removed? (Hinl: How does the term -d,xy affect the equatIOn dy/dl-
9. Investigate the behavior of solutions of the two-
dimensional population prohlem with a, = 3, a2 = I, C 6H6 + CI 2 --> C6H sCl + HCI y(e - d1x - dzy)?
b, = 3, b2 = I, x(O) = 20, and yeO) = 10. C6 HsCl + CI 2 --> C6H.C1 2 + HCI 3. Find conditions on the positive constants a, bh b2 , e, d" and d z so that (a) the sys-
10. Investigate the behavior of solutions of the two- C.H4 CI2 + CI 2 --> C.H.CI, + HCI, tem has exactly one equilibrium point in the first quadrant and (b) the system has I
dimensional population problem with a, = I, a2 = 2,
b, = I, b2 = 1, x(O) = 4. andy(O) = 8.
where HCI is hydrogen chloride. C.H.CI2 is
dichlorobenzene, and C.H.Cl, is trichloroben2ene.
+ Samuel W BodmaI4 The Industrial Practice of Chemical Process Engineering. MlT Press, Cambridge,
I
~
MA (1968), pp. 18-24.
[2
~
~
------------------------------ r:
no equilibriwn point in the first quadrant. (e) Is it possible to choose positive c Xl = axI - bxi - CXIX2
~tants a, bj, b2, C, d I , and d2 so that each population grows without bound? If~n {X2 = -dx2 + ex,X2 - h.
Illustrate graphically. 0,
10. Let a = 36, b = 4, C = 2, d = 1, h = 1, XI(O) = 1, and X2(0) = 2. Compare the
B. Food Chains*
solution to that obtained in (6). How does harvesting affect the populations?
11. Let a = 36, b = 4, C = 2, d = 1, h = 4, XI(O) = 1, and xzCO) = 2. Compare the
I
The food chain in the ocean can be divided into five trophic levels, where the base
level is made up p~marily of seaweeds and phytoplankton. Suppose that XI(t) repre_
sents the SIze of thIS base level population and let X2(t) represent that of the second
solution to that obtained in (10).
12. Let a = 36, b = 4, C = 2, d = 1, h = 4.5, XI(O) = 1, and X2(0) = 2. Compare the
solution to that obtained in (10) and (11). Suppose that we consider all five trophic
levels in the food chain, where we let X3(t) represent the size of the population at
II
level in the food chain, which feeds on the base level. In the absence of the second
level, asswne that Xl (t) follows the logistic equation. However, an interaction between
the third level, xit) that at the fourth level, and xs(t) that of the fifth (or top) level.
This situation is modeled with the system I
members of the first two levels of the food chain prohibits growth of X I(t), and it en-
hances that of X2(t). In addition, asswne that X2(t) decays exponentially without the food
X; = aIXI - a2xi - a3xIX2 I
supply of the base level. Therefore, we model this situation with
X2 = -bIX2 + b2xIX2 - b3X2X3
X3 + C2X2X3 - C3 X 3X 4 I
j = -CIX3
I
Xl = axI - bxi - CXIX2 x~ = -dIX4 + d2X3X4 - d3X4XS
{X2 = -dx2 + ex,X2' = - el Xs + e2x4 x S-
,I
Xs
1. Show that the equilibriwn solution of this system is (die, (ae - bd)/(ce)). Notice that we asswne each level benefits from an encounter with a member of
the next lower level.
2. Suppose that a = 36, b = 4, c = 2, and d = 1. What is the equilibrium solution in
this case? 13. Suppose that al = a2 = a3 = 1, b I = b2 = b3 = I, CI = C2 = C3 = 1, d 1 = d 2 =
d3 = 1, e, = e2 = e3 = 1, XI(O) = 2, X2(0) = 3, X3(0) = 8, X4(0) = 10, and
3. Nwnerically solve the system using the initial popUlations XI(O) = 3 and X2(0): xs(O) = 12. Nwnerically solve the system using the initial populations and graph
2 and graph these functIons on the same set of axes. What is limt_~ X I(t)? What these functions on the same set of axes. Detenrune limt_~ x, (t), limH~ X2(t),
is limH~ X2(t)? Describe the .corresponding physical situation. At how many val-
ues of t does XI(t) = X2(t)? .
limt_~ X3(t), limH~ X4(t), and limr_~ xs(t)? Describe the corresponding physical
situation. I
4. Repeat part 3 using XI(0) = 1 and X2 (0) = 2. Suppose that we harvest the base
level at the rate h. The system becomes
14. Repeat (13) with a, = 2. ;
15. Repeat (13) with a, = 10. If we fish at the base level at rate h, we obtain the I
Xl = axI - bxi - CXIX2 - h
{X2 = -dx2 + eXlx2.
system
I
X; = alxI - a2x i - a3 x l x 2 - h g
5. Let a = 36, b = 4, c = 2, d = I, h = 10, XI(O) = I and X2(0) = 2. Compare the
solution to that obtained in 4.
6. How does harvesting affect the popUlations?
7. Repeat (5) with h = 20. Compare the results with those of (5).
l X2 = -bIX2
X3
x~ =
Xs
= -C}X3
+ b2xIX2
+ C2X2X3
-dIX4 + d2X3X4 - d3X4XS
= -elx5 + e2x4 xS-
- b3X2X3
- C3 X 3 X4
I
I
8. Repeat (5) with h = 25. Compare the results with those of (5) and (7). 16. Suppose that al = 10, a2 = a3 = 1, b I = b2 = b3 = 1, c, = C2 = C3 = 1, d, =
d 2 = d3 = 1, e) = e2 = e3 = 1, XI(O) = 2, X2(0) = 3, X3(0) = 8, X4(0) = 10, and
9. Repeat (5) with h = 26. Compare the results with those of (5), (7), and (8). Sup- xs(O) = 12. Nwnerically solve the system using the initial populations and graph
pose that we harvest the second level at the rate h. Then the system becomes these functions on the same set of axes. Determine limr_~ Xl (I), limr_~ xzCt),
limr_~ X3(t), limr_~ X4(t), and limr-ro xs(t)? Describe the corresponding physical
situation.
• J. Raloff, "How Long Will We Go Fishing for Dinner?" Science News. Volume 153 (Feb. 7, 1998),
p.86.
17. Suppose that we fish at the third level at rate h = 8. How does this form of har-
vesting affect the popUlation sizes?
~
~
rl
k
~
II
422 Chapter 7 Applications of Systems of Ordinary Differential Equations
Figure 7.21
Flow Raleq
Concentration cin -
Temperature Tin
Continuous-Flow Stirred Tank Reactor

B-
Volume V
Tonk
Flow Rale q
Concentnltion c
Tcmperarure T
VC
dT
Pdt = qCpTin - qCpT- VS(T) + HVk(T)c,
where Tin is the temperature at which the chemical flows into the tank. Solve this
equation for dT/dt.
3. The equations in (1) and (2) form a system of nonlinear ordinary differential equa-
I•
C. Chemical Reactor
Consider the continuous-flow stirred-tank reactor shown in Figure 7.21. In this reac.
tions. Show that the equilibrium point of this system satisfies the equations
V
c' - c = -ck(T)
In q and
HV
T - Tin = Cck(T),
q p
II
tion, a stream of chemical C flows into the tank of volume V, and products as well as
residue flow out of the tank at a constant rate q. Because the reaction is continuous
the composition and temperature of the contents of the tank are constant and are th~
if we assume there is no cooling system. Solve the first equation for c and substi-
tute into the second equation to find that I
same as the composition and temperature of the stream flowing out of the tank. Sup.
HVCin I
pose that a concentration Cin of the chemical C flows into the tank and a concentration
C flows out of the tank. In addition, suppose that the chemical C changes into products
I
I
at a rate proportional to the concentration c of C, where the constant of proportional.
ity k(T) depends on the temperature T,
k(T) = Ae- B1T
Because
( of amount ol
Rate of chan e ) . (Rat? that
C = (Rate m of C) - (Rate out of C) - C dlsappear~ ,
we bave the differential equation'

by the reaction
)
II
d
dt (Vc) = qCin - qc - Vk(T)c.· I
1. Show that if V is constant, then this equation is equivalent to dc/dt =
q/V(Cin - c) -k(T)c. I
2. In a similar manner, we balance the heat of the reaction with
Rate of Change)
( of heat content
= (Rate in of heat) - (Rate out of heat) _ (b

Heat
rel~oved) + (bHeat pr~duced).
I
I
y coo mg y reac"on
Let Cp be the specific heat so that the heat content per unit volume of the reaction
mixture at temperature T is CpT. If H is the rate at which heat is generated by the
reaction and VS(T) is the rate at which heat is removed from the system by a cool·
ing system, then we have the differential equation
'" Applications in Undergraduate Mathematics in Engineering. Mathematical Association of America, I

Macmillan Company, New York (1967), pp. 122-125. W Fred Ramirez, Computational Methods for
Process Simulation. Butterworths Series in Chemical Engineering, Boston (1989), pp. 175-182. I
I
I
I
""
8,1 The Laplace Transform: Preliminary Definitions and Notation 425
8 is usually difficult.
In this chapter, we discuss a technique that transforms the differential equation,
anin) + an_lin-I) + . , . + aI/ + aoy = f(t),
into an algebraic equation that can often be solved to obtain a solution of the differ-
ential equation, even ifl(t) it not a smooth function.
Introduction to the [I!J The Laplace Transform: Preliminary Definitions

LapLace Transform and Notation
r Definition of the Laplace Transform r Exponential Order
r Jump Discontinuities and Piecewise Continuous Functions
r The Inverse laplace Transform
We are already familiar with several operations on functions. In previous courses, we

learned to add, subtract, multiply, divide, and compose functions. Another operation on
functions is differentiation, which transforms the differentiable function F(t) to its de-
rivative F'(t),
D,(F(t)) = F' (t).
Similarly, the operation of integration transforms the integrable function f(t) to its in-
tegral. For example, iff is integrable on an interval containing a, then
I
n previous chapters we investigated solving the equation
an(t) yfn) + an-l (t) yln-l) + ... + a , (t) y' + ao(t) y = f(t) [f(X) dx
for y. We saw that if the coefficients an, an- 1, ... , a o are constants, we can find Pierre Simon de Laplace
(1749-1827) French mathemati- is a function of t.
a general solution of the equation by first finding a general solution of the cor- In this section, we introduce another operation on functions, the Laplace trans-
responding homogeneous equation
cian and astronomer
(North Wind Picture Archives) form, and discuss several of its properties.
any!n) + an-l ytn-ll + ... + a , y' + aoY = 0,
Definitiol1l of tlte laplace Tral1lsform
and then finding a particular solution of
anyfn) + an_1y ln-l) + ... + a,y' + aoy = f(t).

The French mathematician Definition 8_1 Laplace lransform
If the coefficients an, an- 1, .. _, ao are not constants, the situation is more diffi- Pierre de Laplace (1749- Let f(t) be a function defined on the interval (0, +00), The Laplace transform
cult. In some cases, as when the equation is a Cauchy-Euler equation, similar 1827) introduced this integral
of f(t) is the function (of s)
techniques can be used. In other cases, we might be able to use power series transform in his work Theorie
to find a solution. In all of these cases, however, the function f(t) has typically .nalytique des probabilites,
published in 1812. However, I ;e{f(t)} = fe-s'[(t) dt,
been a smooth function. In cases when f(t) is not a smooth function, as when o
laplace is probably most fa- I
f(t) is a piecewise-defined function, implementing methods like variation of pa-
rameters to solve the equation
mous for his contributions to I provided that the improper integral exists.
astronomy and probability.
424
8.1 The Laplace Transform: Preliminary Definitions and Notation 427
426 Chapter 8 Introduction to the Laplace Transform
The formula found in Example 2 can be used to avoid using the definition, as illus-
Because the Laplace transform yields a function of s, we often use the notation
5£{f(t)} = F(s) to denote the Laplace transform off(t). We use the capital letter to de. trated in Example 3.
note the Laplace transform of the function named with the corresponding smalliette
The Laplace transform is defined as an improper integral. Recall that we eVa!u~
ate improper integrals of this form by taking the limit of a definite integral: "Ei"",••1
f~ f(t) dt =
D
lim fM f(t) dt.
M-YX> 0
Compute (a) 5£{e- 3t } and (b) 5£{e 5/}.
: ] Solution (a) 5£{e- 3t } = 1/(s - (-3» = I/(s + 3), s> -3, and (b) 5£{e 5t } =
"i·,,'.'
Compute 5£ {f(t)} if f(t) = 1.
1/(s - 5), s > 5.
: JSolufion
"ii,,','.'
Compute 5£ {sin t}.
f
~
F(s) = 5£{f(t)} = e- st . l.dt = lim fMe- st . 1 dt = lim [ __e_
-st]t=M
o . M-7~ 0 M-->oo S 1=0 : ] Solufion To evaluate the improper integral that results, we use a table of inte-
grals or a computer algebra system. Otherwise, we would have to use integration by
I l'1m [-sM
= -- e -1]
. parts twice.
SM-).oo
If s > 0, limM_~ e -sM = O. (Otherwise, the limit does not exist) Therefore, 5£{sin t} = f~o e- st sin t dt = M--)oo
lim fMe - st sin t dt
0
F(s) = 5£{f(t)} = lim [e- sM -

-1.SM--)CQ, I] = -1.S (0 - 1) = 1.S' s > O. e-s' ]t=M
= lim [ - - 2 - - (s sin t + cos t)
M--)oo S + 1 1=0
Notice that the limit in Example I does riot exist if s < O. This means that 5£{ I} is de- = --2-1- lim [e-sM(s sin M + cos M) - 1]
s + 1 M~~
fined only for s > 0, so the domain of 5£{I} is s > O. We will fmd that the domain of
most functions is s > a, where a is a constant. If s > 0, limM_~ e -sM(S sin M + cos M) = O. (Why?) Therefore,
5£{sin t} = --2-1- lim [e-sM(s sin M + cos M) - 1] = -2-1- , s > O.

s + 1 M-?~ +1
"F',,'.I S
Compute 5£{f(t)} iff(t) = eat.

lif',,',I.)
; ) Solution
F(s) = 5£{f(t)} = r o
e-s'f(t) dt = r
0
e-ste at dt = f~ e-(s-a)t dt
0
Compute 5£ {f(t)} if f(t) = t.
To compute F(s) = 5£{f(t)} = f~ e-st t dt we use integration by parts

. [ e -(s-a)t]t=M
= hm ---- = . (e -(s-a)M
-hm - - - - - -
1) : ") Solution
. with u = t and dv = e -st dt. Then du = dt and v = -e -stls, so
M-'1 OO S - a t=O M-).oo S - a s - a .
If s - a > 0, then limM_~ e -(s-a)M = O. Therefore, F(s) = 5£{f(t)} = f~ e- st t dt = lim fM e- s, t dt
o
. (e -(s-a)M
F(s)=5£{f(t)}=-hm - - - - - - = - 0 - - - =--,s>a.
s-a
I) ( 1) s-a
s-a
1 _te-
st
M--)oo
I'=M) + lim -1 fM e- st dt
0
M-'1 00 S-Q = lim - --

M--)oo ( S t=O M---+oo S 0
[-
428 Chapter 8 Introduction to the Laplace Transform 8.1 The Laplace Transform: Preliminary Definitions and Notation 429
1 I' (-stlt~m) : I Solution (a) .:e{6} = 6.:e{l} = 6(l/s) = 6/s. (b) .:e{5 - 2e- t } =
= O -21m e t=O'
S M--+oo . 5.:e{1} - 2.:e{e- t } = 5(1/s) - 2(I/[s - (-1)]) = 5/s - 2/(s + 1).
If s > 0, limM~oo e- sM = O. Therefore,
F(s) = .:e{f(t)} = -~ lim (e- sM - 1) = ~,s > O.

0~
~j',
For what values of s do the trallsforms in Example 6 exist?
s M-+r:T.> S
Exponential Order
In calculus we saw that in some cases improper integrals diverge, which means that
As we can see, the definition of the Laplace transform can be difficult to apply.
the limit of the definite integral f':ff(x) dx does not exist as M -> 00. Hence, we may
For example, if we wanted to calculate .:e{ In} with the defiilltion, we would have to
believe that the Laplace transform may not exist for some functions. For example,
integrate by parts n times; a difficult and time-consuming task. We will investigate
other properties of the Laplace transform so that we can determine the Laplace trans-
fCt) = I/t grows too rapidly near t = 0 for the improper integral r;
e -s'f(t) dt to exist
and f(t) = e t' grows too rapidly as t -> 00 for the improper integral fa e -s'J(t) dt to
form of many functions more easily. We begin by discussing the linearity property,
exist. We present the following definitions and theorems to better understand the types
which enables us to use the transforms that we have already found to find the Laplace
of functions for which the Laplace transform exists.
transforms of other functions.
We are familiar with linearity properties through our work in calculus with derivatives
and integrals. For derivatives, this property is
II Definition 8.2 Exponential Order
A functionf(t) is of exponential order (of order b) if there are numbers b, C>
0, and T > 0 such that
and for integrals, it is I If(t) I S Ce
bl
J [af(x) + bg(x)] dx = a J f(x) dx +b J

g (x) dx. I fort>T.
f(x)
6
We can use Definition 8.2 to show that if limt_oof(t)e -bl exists and is finite,f(t)
Theorem 8.1 Linearity Property of the Laplace Tran_sf'_o_r_ID_ _ _ _ _ _..., is of exponential order; if lim,_ oo f(t)e -bl = +00 for every value of b > 0, f(t) is not
Let a and b be constants, and suppose that the Laplace transform of the func- , of exponential order. (See Exercise 64.)
tions fCt) and g (t) exists. Then,
Jump Discontinuities and Piecewise Continuous
.:e{af(t) + bg(t)} = a.:£{f(t)} + b.:e{g(t)}. Functions
===;;;;;;=!J
In the next sections, we will see that the Laplace transform is particularly useful in
PROOF OF THEOREM 8.1 solving differential equations involving piecewise or recursively defined functions. For
(a)
example the function f(x) = { X2 + I, . if x S 1 WIth
. the graph shown In . F'Igure 8 . I(a,
)
.:e{af(l) + bg(t)} = fe-st(af(t) + bg(t)) dt , 4 - x, If x > I
o g(x)
is a piecewise defined function. Because
= a (OOe-s'f(t) dt
Jo
+b fe-Slg(l) dt = a.:e{f(t)}
0
+ b.:e{g(t)}. lim (x 2
lim f(x) = x--tl-
x--+l-
+ 1) = 2 '* 3 = lim+ (4 - x) = limJ(x),
x-)l x-Jol
2 4 6 x
the limit limx_l f(x) does not exist andf(x) is not continuous if x = 1. Similarly, the
function g(x) defined by g(x) = { ...' 0 S1 Sx <

I I, if if I th·
x < 2, on e Interva
I [0 , 2) and then
(b)
recursively by the relationship g(x) = g(x - 2) (shown in Figure 8.l(b)), is discontin-
Calculate (a) .:e{6}; (b) .:e{5 - 2e- t } Agure 8.1 (a)-(b). uous when x = n, where n is any nonnegative integer. For both f and g, we say that the
discontinuities are jump discontinuities because the left- and right-hand limits both " the sum of integrals over which e -s'J(t) is continuous. The fact that e -s'J(t) is
exist at the points of discontinuity but are unequal. piecewise continuous on [T, co) is also used to show that the second integral,
I~ e -s'J(t) dt, converges. Because there are constants C and b such that If(t) I :=:;
bf
Ce , we have
I Definition 8.3 Jump Discontinuity
Letf(t) be defined on [a, b].fhas a jump discontinuity at t = c, a < c < b, if

the one-sided limits limr-c+ f(t) and lim f-+ c - f(t) are finite, but unequal, values. M [ e-(S-b)f]f~M
=
f(t) has a jump discontinuity at t a if limHa+ f(t) is a finite value different
=
fromf(a).f(t) has a jump discontinuity at t h ifJim'~b- f(t) is a finite value
= C lim
M~oo T f e-(s-b)1 dt = C lim - - - -
M-YX> S - b l=T
different fromf(b). = _~ lim (e-(s-b)M _ e-(s-b)T).

s - bM-+oo
Then, if s - b> 0, limM~oo e-(s-b)M = 0, so
e -(s-b)T
Definition 8.4 Piecewise Continuous
A functionf(t) is piecewise con1;inuous on the finite interval [a, hI iff(t) is .
IfT e -s'J(t) dt:=:;
I : _ b 's > b.
i
continuous at every point in [a, b except at finitely many points at which f(t) : Because both of the integrals J6e -Sf f(t) dt and I'Te -s'J(t) dt exist, r;e -s'J(t) dt
has a jump discontinuity. A functionf(t) is piecewise continuous on [0, ,,) if I also exists for s > b.
f(t) is piecewise continuous on [0, N] for all N> O. '
lif'"i,'••
1I(t- 1), O:=:; t< 1 . . .
Is f(t) = { t, I:=:; t :=:; 2 pIecewIse contmuous on [0, 2]7 Is f(t) =
Find the Laplace transform of f(t) = { I, 't
-I
>;;-
0 t 4
<
- . <
t, O:=:; t < I . . .'. [D)?

{ 1/(t - Ii, I:=:; t pIecewIse contmuous on , +co .
: ] Solution Because f(t) is a piecewise continuous function on [0, co) and of ex-
Note: In this textbook, typi-
ponential order, F(s) = .:e{f(t)} exists. We use the defmition and evaluate the im-
cally we work with functions
that are piecewise continuous proper integral using a sum of two integrals.
Theorem 8.2 Sufficient Condition for Existellce of .:elf(t)}
and of exponential order.
However, in the exercises we
explore functions that mayor
Suppose that f(t) is a piecewise continuous function on the interval [0, (0) and·
that it is of exponential order b for t > T. Then, .:e{f(t)} exists for s > b.
F(s) = .:e{f(t)} = {f(t)e- Sf dt =
0 0 4
f(-l)e-
Sf
dt + fe-si dt
[-s-
may not have these properties. e-Sf]f~4 . [ e-Sf]t~M
= + hm - - -
t=O M-+C1:J S r=T
We need to show that the integral I;;' e -s'J(t) dt converges for s > b, assuming that
f(t) is a piecewise continuous function on the interval [0, co) and that it is of ex- Ifs > 0, lim e- Ms = 0, so
M-+oo
ponential order b for t > T. First, we write the integral as
F(s) = .:e{f(t)} = 1s (2e- 4S - 1), s > O.
{e-s'JU) dt = {e-s'J(t) dt + fe-s'J(t) dt,
o 0 T
bf
where T is selected so that If(t)1 :=:; Ce for the constants b and C, C> O.
Notice that because f(t) is a piecewise continuous function, so is e -$1/(1). Notice that Theorem 8.2 gives a sufficient condition and not a necessary condi-
The first of these integrals, J6e -s'J(t) dt, exists because it can be written as tion. In other words, there are functions such as f(t) = t- 1/2 that do not satisfy the
hypotheses of the previous theorem for which the Laplace transform can be found. .;e{r3} = 3!/s 4 = 6/s 4 ,.;e-l{6/s4 } = t 3.(d)NoticethatF(s) = 6/(s + 2)4 is obtained
(See Exercises 62 and 63.) from F(s) = 6/s 4 by substituting (s + 2) for s. Therefore by the shifting property,
6_} = e- 2tr3.
.;e{e-2tt3} = _ _6_ so .;e-l{_ _
(s + 2)4' (s+2t
The Inverse Laplace Transform
L
Up to this point, we were concerned with finding the Laplace transform of a given
function. At this point we reverse the process: Given a function F(s) we want to fmd Find .;e-l{(S ~ 4)} and .;e-ts2 : 4J
a function fCt) such that .;e {f(t)} = F(s), if possible.
Just as we use the linearity of the Laplace transform to fmd .;e {f(t)} for many
functions!, the same is true for computing .;e-l{F(s)}.
Definition 8.5 Inverse Laplace Transform
The inverse Laplace transform of the function F(s) is the functionf(t), if such Theorem 8.4 Linearity Property of the Inverse Laplace Transform
a function exists, that satisfies .;e{fCt)} = F(s). We denote the inverse Laplace
Suppose that .;e-l{F(s)} and .;e-l{G(S)} exist and are continuous on [0, co) and
transform of F(s) with
that a and b are constants. Then,
f(t'f= .;e-l{F(s)}.
.;e-l{aF(S) + bG(s)} = a.;e-l{F(s)} + b.;e-l{G(S)}.
As we mentioned, there is a one-to-one correspondence between continuous functions If the functions are not in the forms presented in the table on the inside cover of
and their Laplace transforms. With this in mind, we can compile a list of functions the book or in Table 8.1 at the end of this section, we can make use of the linearity
with the corresponding Laplace transforms. Such a table of Laplace transforms, given property to determine the inverse Laplace transform.
on the inside cover of this textbook as well as in Table 8.1 at the end of this section,
is useful in finding the inverse Laplace transform of a given function. In the table,
we look for F(s) in the right-hand column to findf(t) = .;e-l{F(s)} in the correspon-
ding left-hand column. For example, the second row of the table indicates that '
'if',,','M'
.;e-l{4!1s5} = t 4 and the ninth row shows that .;e-l{(S - 1)/[(s - 1/ + l6]} = Find the inverse Laplace transform of (a) F(s) = lIs 3; (b) F(s) = -7/(S2 + 16);
et cos 4t. This example illustrates shifting in the variable s. (e) F(s) = 51s - 2/(s - 10).
Theorem 8.3
(Shifting in s) If .;e{fCt)} = F(s) exists, then .;e-l{F(s - a)} = eatf(t).
"F''''''N;I
Find the inverse Laplace transform of (a) F(s) = l/(s - 6); (b) F(s) = 2/(i + 4); .
(e) F(s) = 6/s 4 ; (d) F(s) = 6/(s + 2)4. Ii ",,',i.'/1
: J Solution (a) Because .;e{e 6t } = l/(s - 6), .;e-l{lI(s - 6)} = e6t. (b) Note that Find the inverse Laplace transform of F(s) =
2s - 9
S2 + 25·
.;e{sin 2t} = 2/(s2 + 22) = 2/(S2 + 4), so .;e-l{2/(S2 + 4)} = sin 2t. (e) Because.
:1 Solution The following theorem is useful in showing that the inverse Laplace transform
of a function F(s) does not exist.
;e-I{~} = 2;e-I{_S} _ 9;e-1{_1}
~+~ ~+~ ~+~
= 2;e-l{_S} _ 2-;e-1{_5} Theorem 8.5

S2 + 25 5 S2 + 25
Suppose that/(t) is a piecewise continuous function on [0, (0) and of exponen-
= 2 cos
9 sm
5t - "5 . 5t. tial order b. Then,
L lim F(s)
S--)"XI
= lim ;e{f(t)} = O.
8-')00
Find ;e-l{(3s + 1)/(S2 + 16)}.

Sometimes we must complete the square in the denominator of F(s) before fmd.
ing ;e-l{F(s)}.
"",,1,'.'1
"E·,,"'." For the following functions, determine if F(s) is the Laplace transform of a piece-
.. . .
WIse contmuous functlOn of exponenbal order. (a) F(s) = - - ;
s-6
b
Determine ;e-l{, s }. S3
s- + 2s + 5
(b) F(s) = S2 + 16'
: ] Solution Notice that many of the forms of F(s) in the table of Laplace trans·
. forms on the inside cover of this book involve a term of the form S2 + in the de· e J Solution (a) lim F(s) = lim 2s/(s - 6) = 2 "* 0; (b) lim F(s)
3
= lim _,_s_ _ =
nominator. Through shifting, this term is replaced by (s - a)2 + k 2. We obtain a S-+IXl s--+<Xl S-')OO s- + 16
S--+OO
term of this form in the denominator by completing the square. This yields 00 "* "*
O. In each case, lim F(s) 0, so F(s) is not the Laplace transform of a piece-
s-->oo
s s s wise continuous function of exponential order.
S2 + 2s + 5 (S2 + 2s + I) + 4
Because the variable s appears in the numerator, we must write it in the form In later sections, we may use Theorem 8.5 to determine if we have made an er-
(s + I) to find the inverse Laplace transform. Doing so, we find that ror in our solution process.
s s (s + I) - I
Does ;e-l{ 2 S2 } exist?
S2 + 2s + 5 (s + Ii + 4 (s + 1)2 +4' s + 8s + 16
Therefore, Using the properties of the Laplace transform discussed here and in the exercises,
we can compute the Laplace transforms of a large number of frequently encountered
;e-I{ s } _ ;e-l{ (s + 1) - I } functions. Table 8.1 lists the Laplace transforms of several of these frequently en·
S2 + 2s + 5 - (s + 1)2 + 4 countered functions. A more comprehensive table is found on the inside cover of
this textbook.
-;e-I{ (s+l) }_.!.;e-l{ 2 }
- (s + 1)2 + 4 2 (s + Ii + 4
TABLE 8.1 laplace Transforms of frequently Encountered Functions
= e- t cos 2t - 2"I e- t sin 2t.
f(t) F(s) = X{f(t)) itt) F(s) = X{f (t))
L
n!
s>o s>o
Find;e- '{ 2 2s
s - 4s + 8 }. s 7"
436 Chapter 8 Introduction to the laplace Transform 8.1 The laplace Transform: Preliminary Definitions and Notation 437
TABLE 8.1 cont. Laplace Transforms of Frequently Encountered Functions ~ (-I)"lln+1

(b) :£(e a' sin kt) = (s - k
a)2 + k2 57. Use the Maclaurin series sin I = ~o (2n + I)! to
f(t) F(s) = :£{f(O} f(t) F(s) = ~{f(m
verify that ~(sin t) = I/(S2 +
--
19. (Shifting Property) Suppose that ~(f(t)) = F(s) ex- I).
e a' tne at , n = 1,2, ...
n! ists for s > a. Use the definition of :£ (ea'j(I)) to show ~ (-I)"Iln
s-a
s>a
(s - a)"+1 that :£(ea'j(I)) = F(s - a). 58. Use the Maclaurin series cos I = "?o (2;;j! to
k k
verify that ~(cos t) = s/(s2 + I).
sin kt eat sin fa In Exercises 20-25, use the result of Exercise 19 to compute
S2 +~ (s-a)~ 59. Find ~(COS2 kt).
the following. Compare your results with those found in Table
8.1 or the table on the inside cover of this textbook. 60. Figure 8.2(a) shows tbe graph of a functionf(t) and
s S-a
cos kt eat cos let Figure 8.2(b) shows the graph of its Laplace transform
S2 + k2 (s - a)2 +-Ie' 20. ~(e' sin 21) 21. :£(e' cos 31) F(s) = ~(f(I)). Use Figure 8.2 to sketch the graphs
k 22. ~(e-2' cos 41) 23. :£{ e -, sin 5,t} of ~{e-2,!(lll and :£(e''j(t)}.
sinh kt e a, sinh kt k
S2 - k 2 (s - a)' - Ie' 24. ~(e3tt) 25. ~(e -tl2 t 2)
s S-a
2t=(f)
cosh kt S2 _ k 2 eat cosh kt
(s - a)' - Ie' In Exercises 26-55, use properties of the Laplace transform and
Table 8.1 or the table on the inside cover of this textbook to f(t)
1
compute the Laplace transform of each function.
26. f(l) = 28e' 27. f(t) = -18e"

1 2. 3 r
EXERCISES 8.1 28. f(l) = j(2t - sin 2t) *29. f(l) = cos 51
Ca)
30. f(t) = I + cos 21 31. f(l) = I + sin 51
32. f(l) = 18e" 33. f(l) = -16e 2 ' F(s)
34. f(l) = -61 4 e- 7 ' 35. f(l) = 17

In Exercises 1-16, use the definition of the Laplace transform
to compute the Laplace transform of each function.
13. f(t,) = {I,
-1,
O:S t < 10,
I? 10
36. f(l) = I'e
4t
*37. f(l) = t 2 e- 3t
4
1. f(t) = 21t 38. f(t) = t cos t 39. f(l) = 15 e- 4 '
2. f(l) 7e-'
*3. f(t) = 2e'
=
4. f(l) = -8 cos 3t
14. f(l) = It, O:S 1< 2,
40. f(t) = 8 cosh 4t 41. f(t) = I cos 31 5 ,
3, I? 2
42. f(t) = I sin 71 43. f(l) = 31 sin I Cb)
5. f(t) = 2 sin 2t *15. f(l) = sin kt
I if 0 :S t :S 2 16. f(t) = cos kt 44. f(l) = 12 sinh 61 *45. f(l) = t sinh 7t Figure 8.2 (a)-(b).
6. f(t) = { 0 if t > 2 47. f(l) = e 7 ' cosh
17. Use the definitions cosh kt = ~ (ela + e- la ) and 46. f(t) = e' sinh I I
sinhkt = 1(ela - e- la ) to verify each ofthefollowing. 61. Use the identities sin(A + B) = sin A cos B + cos A
48. f(l) = -7 e 6 ' sin 21 49. f(l) = e -2f cos 4t
{~
ifO:St:S1
*7. f(t) = sin Band cos(A + B) = cos A cos B - sin A sin B to
ift> I (a) ~{cosh kt} = ~ 50. f(t) = e- 2' sin 71 *51. f(t) = e 5' cos 71 assist in finding tbe Laplace transform of the follow-
s - k
I - I if 0 :S I :S I 52. f(l) = 2e- 2 ' sinh I ing functions.
8. f(t) = { 0 if I > I k
(b) :£{sinh kt} = S2 _ ~ 53. f(t) = sin 41 + 41 cos 4t (a) f(l) = sin(1 + 1'14)
COS I if 0 :S I :S 1'/2 54. f(l) = (' sin 2t (b) f(l) = cos(t + 1'14)
9. f(t) = { 0 if I > 1'/2 (c) :£{ e a ' cosh ":t} _ s - a
- (s - a)2 - k 2 55. f(l) = il sin 31 (c) f(l) = cos(t + 1'16)
sin lifO :S t:S 21' k (d) f(t) = sin(t + 1'16)
10. f(t) = { 0 if I> 21' (d) ~(ea, sinh kt) 56. Recall that the sum of the geometric series a + ar +
(s - a)2 - k 2 rex), rex) =
ar2 + ar' + ... = al(1 - r) if \r\ < 1. (The series
r
62. The Gamma function, is defined by
*11. f(t) = {
I - I, 0 O. Show that ~(I·) =
0, t? 3 sin kt = 1I2i(e 'la - e- ik,) to verify each of the series for f(l) = e f is e' = 1 + I + 12/2! + l'/3! + o
following. ... + 1"ln! + .. '. Take the Laplace transform of rea + l)/s·+I, a> -1.
3t + 2, 0< I< 5 each term in the series and show that the sum of the *63. Use the result of Exercise 62 to find the Laplace trans-
12·f(l) = { s-a
0, t? 5 (a) ~(ea, cos kt) = 2 2 resulting series converges to ~ (e'). form of (a)f(l) = t- 1I2 ; (b)f(t) = t l12 •
(s - a) + k
438 Chapter 8 Introduction to the Laplace Transform 8.2 Solving Initial-Value Problems with the Laplace Transform 439
64. (a) Use Dermition 8.2 to show that if lim,_oof(t)e -br (b) Show that ~-l{tan-l(als)} = (sin at)lt. 18.21 Solving Initial-Value Problems with the Laplace
exists and is finite, f(t) is of exponential order and if (c) Show that ~-I{l12 tan-leta
+ b)ls) + Transform
lim,_oof(t)e -bl = +00 for every value b,f(t) is not of
exponential order. (b) Give an example of a function
112 tan-leta -
b)/s)} = (sin at cos bt)lt.
91. (a) Use the linearity of the Laplace transform to In this section we show how the Laplace transform is used to solve initial-value prob-
f(t) of exponential order b for which limr-oo f(l)e -bl
compute ~{a sin bl - b sin at} and lems. To do this we first need to understand how the Laplace transform of the deriva-
does not exist.
~{cos bt - cos al}. tives of a function relates to the function itself. We begin with the first derivative.
65. Letf(t) be a piecewise continuous function on the in·
(b) Use these results to find
terval [0, 00) that is of exponential order b for t> T.
~-I{1/[(s2 + a 2 )(s2 + b 2 )]} and
Show that h(t) = f~ feu) du is also of exponential
~-1{s/[(S2 + a 2 )(s2 + b 2 )]}. (These formulas
order.
can be verified in a table of Laplace transforms Theorem 8.6 Laplace Transform of the First Derivative
or with a computer algebra system.)
In Exercises 67-89, compute the inverse Laplace transform of Suppose that f(l) is continuous for all I 2: 0 and is of exponential order b for
the given function. 92. Use the results of Exercise 91 to rmd
I > T. Also, suppose that f' (I) is piecewise continuous on any closed subinterval
66. J,
s
67. J,
s
(a) ~-I{(S2 + 1~~:2 + 16J of [0, ee). Then, for s > b
I
68. (s _ 5)6 *69. (s ~ 2)2 .; (b) ~-ts2 + 10~)(S2 + IJ :£{f'(I)} = S:£{f(I)} - f(O) .
I 1_
71. _ _ 93. Determine if F(s) is the Laplace transform of a piece.
70. s + 5 (s + 6)3 wise continuous function of exponential order.
72. _2_1- 4
73. (s - 3)' (a) F(s) = - s4
-s
(b) F(s)=~
s +9 s+ I
Using integration by parts with u = e- si and dv = /'(t) dl, we have
(d)F(S)=~
I S2
74. S2 + 15s + 56 *75. s2: 16 (c) F(s) = 4s + 10
S2 + I
I 94. (a) Show thatf(t) = t 3 is of exponential order.
77. 2 s
76. S2 + IZs + 61 s - 5s - 14 (b) Show thatf(l) = e " is not of exponential order.
s+3 (c) Isf(t) = tn, where n is any positive integer, of ex· = -j(0) + s:£{f(t)} = S:£{f(I)} - f(O).
78. S2 + 6s + 5
*79.~
S2 - 4s ponential order? Why?
s-7 95. Determine values of C and T that show that f(l) = Proof of Theorem 8.6 assumes thatf' is a continuous function. If we use the as-
80. s2 _ IZs + 35 81. s2 _ 14s + 48 (sin t cos 2t)lt is of exponential order t. sumption that/' is continuous on 0 < 11 < 12 < ... < In < ee, we complete the proof
I
82. s2+Zs-24 *83. S2 _ 4~ - 12 96 . L et 'C)t
In =
{lin,
0, if if 0S
I> n
tS n. (a) Graph '()t ~or
In
by using
I I n = 10, I, -/0, and ~. Is In(t) of exponential order?

84. S2 + IZs + 37 85. S2 _ Zs +2 Why? (b) Evaluate foln(t) dt and limn_ o+ foln(t) tit.
s+I s- I (c) Describe limn_o+ In(t). Is this limit a function? "
86. S2 + Zs + 37 *87. S2 - Zs + 50 (d) Calculate Fn(s) = ~{.fn(t)} and then limn_o·F,(s). This is the same integration by parts formula shown in the proof of Theorem 8.6 for
Are you surprised by the result? Explain. (e) Is every each integral. Now we make the same assumptions of/, and!" as we did offand/"
s+2 *89. s- 7
88. S2 + 4s + 8 S2 - 14s + 50 function F(s) the Laplace transform of some function respectively, in the statement of Theorem 8.6 and use Theorem 8.6 to develop an ex-
f(t)? Why? pression for :£{f"(I)}:
90. (a) Computef(t) = "e-l{tan-l(lls)} with
n
~{I"f(I)} = (-I)" dnlds F(s) in the form 97. Graphically determine if the Gamma function is of ex·
(n= I) ponential order. Can you prove your result analyti· :£{f"(t)} = s:£{f'(t)} - /,(0) = s[s:£{f(t)} - f(O)] - f'(O)
cally? Does the Gamma function have a Laplace tranS·
f(t) = _+~-l{~ F(S)} form? Explain. = S2.;e{f(t)} - sf(O) - /,(0).
= _+~-l{~ tan-1m} Continuing this process, we can construct similar expressions for the Laplace trans-
form of higher-order derivatives. This leads to the following corollary to Theorem 8.6.
440 Chapter 8 Introduction to the laplace Transform 8.2 Solving Initial-Value Problems with the laplace Transform 441
Corollary 8.7 Laplace Transform of Higher Derivatives -:

---..,
More generally, if 1'1}(t) is a continuous function of exponential order b on i
[0, +00) for i = 0, I, ... , n - 1 and1"l(t) is piecewise continuous on any closed i
subinterval of [0, 00), then for s > b I
How is the solution changed if yeO) = I?
!£{f(')(t)} = s'!£{f(t)} - s·-I/(O) - ... - s/(n-2\0) - 1'n-l)(O). In many cases, we must determine a partial fraction decomposition of Yes) to obtain
terms for which the inverse Laplace transform can be found.
We now show how the Laplace transform can be used to solve initial-value problems.
Typically, when we solve an initial-value problem that involves yet), we use the fol- "fi"...· '
lowing steps: Solve the initial-value problem y" - 4y' = 0, yeO) = 3, y'(O) = 8.
1. Compute the Laplace transform of each term in the differential equation;
• ] Solution Let Yes) = !£{y(t)). Then !£{y" - 4y') = !£{O}. Because
2. Solve the resulting equation for !£{y(t)} = Yes); and
3. Determine yet) by computing the inverse Laplace transform of Yes). !£{y" - 4y'} = !£{y"} - 4!£{y'} = s2y(S) - sy(O) - y'(O) - 4(sY(s) - yeO))
, ~
The advantage of this method is that through the use of the property !£{y"} !£{y'}
!£{f(")(t)} = sn!£{f(t)} - S"-I/(O) - ... - S1'"-2\0) - /(n-I)(O) and !£{O} = 0 (why?), the equation becomes
we change the differential equation to an algebraic equation that can be solved for s2y(S) - sy(O) - y'(O) - 4sY(s) + 4y(0) = o.
!£{f(t)}. Applying the initial conditions yeO) = 3 and y'(O) = 8 results in the equation
s2y(S) - 3s - 8 - 4sY(s) + 4(3) = 0 or s(s - 4)Y(s) = 3s - 4.
Solving for Yes), we find that
Solve the initial-value problem y' - 4y = e 4', yeO) = o. 3s - 4
Yes) = s(s - 4)·
: 1 Solution We begin by taking the Laplace transform of both sides of the differ. If we expand the right-hand side of this equation in partial fractions based on the
ential equation. Because !£{ y'} = sY(s) - yeO) = sY(s), we have two linear factors in the denominator of Yes), we obtain
!£{y' - 4y} = !£{e 4t } 3s-4 =.:!+_B_
s(s-4) s s-4·
1
!£{y'} - 4!£{y} = s - 4 MUltiplying both sides of the equation by the denominator s(s - 4), we have
I 3s - 4 = A(s - 4) + Bs.
~Y(s) -:- yeO) - 4~ = s _ 4
If we substitute s = 0 into this equation, we have 3(0) - 4 = A(O - 4) + B(O) or
!£{y'} !£{y} -4 = -4A so that A = 1. Similarly, if we substitute s = 4, we find that 8 = 4B so
1 B = 2. Therefore,
(s - 4)Y(s) = s - 4·
Solving for Yes) yields

so
I
yes) = (s _ 4)2·
By using the shifting property with !£(t) = Ils2, we have

8.2 Solving Initial-Value Problems with the laplace Transform 443
442 Chapter 8 Introduction to the laplace Transform
A partial fraction decomposition involving a repeated linear factor is iJlustrated in th Therefore,

following example. e yet) = 5£-I{i _ _
s
1_ + __
s +I (s + 1)2
9_}
= 6 - e-' + 9te-'.
,'ii".",*,
Solve the initial-value problemy" + 2y' + Y = 6, yeO) = 5, y'(O) = 10. In some cases, F(s) involves irreducible quadratic factors as we see in the next
example.
:] Solution If Yes) = 5£{y(t)}, then
5£{y" + 2y' + y} = 5£{6} "ii"I,I.'
5£{y"} + 25£{y'} + 5£{y} = i Solve the initial-value problemy'" + 4y' = -IOe',y(O) = 2,y'(0) = 2,y"(0) = -10.
s
: Holation Let yes) = 5£{y(t)}. Then
s2y(S) - sy(O) - y'(O) + 2[sY(s) - yeO)] + Yes) = i
~~ __~~~~~, ~ s
5£{y"'} + 45£{y'} = 5£{-IOe'}
5£{y"} 5£{y'}
3 2 10
(S2 + 2s + I)Y(s) = i + 5s + 20 s Yes) - s yeO) .- sy'(O) - y"(O) + 4~ = -~
s
5£{y"'} 5£{y'}
2
(S2 + 2s + I)Y(s) = 6 + 5s + 20s
s (S3 + 4s)Y(s) = _--..!.Q.... + 2S2 + 2s - 2
s- I
Solving for Yes) and factoring the denominator yields
10 + 2S2 + 2s - 2
2 yes) =
Yes) = 5s + 20s + 6 (s - 1)(s3 + 4s) S3 + 4s
. s(s + If
2s 3 - 4s - 8
In this case, the denominator contains the nomepeated linear factor s and the re- s(s - 1)(s2 + 4)
peated linear factor (s + I). The partial fraction decomposition of Y(s) is
Finding the partial fraction decomposition of the right-hand side, we obtain
2
5s + 20s + 6 =:i +~ + __
C_
s(s+I)2 s s+1 (s+I)2' 2s 3 - 4s - 8 A B Cs + D
-==----"'=--"-- = - + ---- + - - -
s(s - 1)(s2 + 4) s s- I S2 + 4 .
so mUltiplication on each side of the equation by s(s + 1f gives the equation
MUltiplying each side of the equation by s(s - 1)(s2 + 4) gives us
5s 2 + 20s + 6 = A(s + 1)2 + Bs(s + 1) + Cs or
2s 3 - 4s - 8 = A(s - 1)(s2 + 4) + BS(S2 + 4) + (Cs + D)s(s - 1)
5s 2 + 20s + 6 = (A + B)S2 + (2A + B + C)s + A.
= (A + B + C)S3 + (-A - C + D)S2 + (4A + 4B - D)s - 4A.
Equating like coefficients, we obtain the system
Equating coefficients yields the system of equations
A+B=5
2A +B + C= 20, A+B+C=2
[ A = 6 -A - C+D= 0
4A + 4B - D = -4'
which has solution A = 6, B = -I, and C = 9, so
5s 2 + 20s + 6 6 I 9
+ ------
1 -4A =-8
s(s + 1)2
= - - ----
S S+ 1 (s + 1)2' with solution A = 2, B = -2, C = 2, and D = 4, so
---------------
444 Chapter 8 Introduction to the Laplace Transform 8.3 Laplace Transforms of Several Important Functions 445
4s - 8 = ~ + ~ + Cs + D ~ _ _2_ + 2s + 4
3
2s - = there a relationship between the power of the inde- fy" + 2y' + 4y = t - e- t
s(s - 1)(S2+ 4) s s - 1 S2 + 4 s s - 1 S2 + 4 . pendent variable in the original equation and the or- 24. \.y(0) = I, y'(0) = -1
der of the differential equation involving Y? 2t
Therefore, 25 fy" + 4y' + 13y = e- cos 3t +I
21. Show that application of the Laplace transform . \.y(0) = 1, y'(O) 1 =
yet) = X-l{~
s
__2_ +
s- 1
2s
S2
+
+4
4} method to the initial-value problem (involving a
Cauchy-Euler equation)
y" + 2y' + Y = 2te- t - e- t
26. {yeO) = 1, y'(O) = -1
at 2y" + bty' + cy = 0, yeO) = ex, y'(O) = fJ 27. In Section 4.S we saw that Bessel's equation is the
= 2X-l{~} - 2X-t ~ I} + 2X-t 2 : 4} + 2X-t ~ 4}2 yields as 2 y" + (4a - b)sY' + (2a - b + c)Y =
0, equation
= 2- 2e + 2 21 + 2 21.
t
cos sin
where Yes) = .:e{y(t)). Is the Laplace transform
method worthwhile in the case of Cauchy-Euler equa-
where Jl. 2:
x"y" + xy' + (x 2 - Jl.2)y = 0,
a is a constant, and a general solution is
tions?
given by
22. Consider the ioitial-value problem (involving Bessel's
equation of order 0) y = c,J,,(x) + C2Y,,(X).
ty" + y' + ty = 0, yeO) = I, y'(O) = 0, (a) Fiod ex and fJ
so that the solution of the initial-
x 2y" + xy' + (x 2 - Jl.2)y = a .
with solutiony = Jo(t), the Bessel function of order O. value problem {yeO) = ex, y'(O) = fJ IS
EXERCISES 8.2 Use Laplace transforms to show that yet)) =.:e{ y=J,,(x) .
.:e{Jo(t)) = k1(v?+l), where k is a constant.
(b) Compute the Laplace transform of each side of
(Hint: .:e{ty") = -dlds [S2y(S) - s] and .:e{ty} =
the equation x 2y" + xy' + (x 2 - Jl.2)y = 0, sub-
-dlds [yes)].
stitute .:e{y) = Yes), and solve the resulting
In Exercises 1-14, use Laplace transforms to solve the initial- *13. y" - y' - 12y = e 2t - sin(t), yeO) = -1, y'(O) = I, 23. (See Exercise 22.) Use the binomial series to expand second-order differential equation for Jl. = 1,2,3,
value problem. In cases in which the equation could be solved
by other methods, describe these methods. Graph the solution
[-I, I]
14. y" + 25y = 10 cos 5t, yeO) = 0, y'(O) = 0, [-7T, 21T]
.:e{Jo(t)) = . ~ =!
y S2 + ISS-
(1 + .1,)-1/2 Take the 4,5.
on the indicated interval. (c) Use the results obtaioed in (b) to determine the
15. (Nonconstant CoeffiCients) Use the property that inverse transform of each term in the expansion to Laplace transform of J,,(t) for Jl. = 1,2, 3,4, 5.
1. y" + lIy' + 24y = 0, yeO) = -I, y'(O) = 0, (_I)"t2n
I
00
.:e{if(t)) = -pes) to solve the initial-Value problem (d) Can you generalize the result you obtained in (c)?
[-112,1] show that Jo(t) = k ---;;;--z. Use the initial con-
n-O 2 (n!)
2. y" + Sy' + 7y = 0, yeO) = 0, y'(O) = 1, [-1/2, I]
y" + 4ty' - Sy = 4, yeO) = 0, y'(O) = O. dition Jo(O) = I to show that k = 1. Therefore, 28. Consider the defmition of the Laplace transform,
*3. y" + 3y' - lay = O,y(O) = -1,y'(O) = I,
(Hint: Applying the Laplace transform to each term .:e{Jo(t)) = I/(Ys2+ I). F(s) = .:e{f(t)) = fe-s'l"(t) dt. If/satisfies the con-
in the equation yields Y'(s) + (3/s - s/4)Y(s) =
[-112,3/2]
- IIs2, where pes) = 3/s - s/4. Use the integrating Use the method of Laplace transforms to solve the following ditions given in Theorem S.6 (The Laplace Transform
4. y" - 13y' + 40y = 0, yeO) = 0, y'(O) = -2, factor Jl.(s) = efp(S) >is to find that the solution of this initial-value problems. In each case, use a computer algebra sys- of the Derivative), we can differentiate within the
[-112,1/4] first-order equation is Yes) = s-\4 + Ce s)'8). For tem to assist in computing the inverse Laplace transform either integral with respect to s when s > a. Show that
5. y" + 4y = 0, yeO) = 2, y'(O) = 0, [0, 27T] what value of C does I
.:e-
{Y(s)} exist? Substitute this by usiog a package command for performing the task or by de- F'(s) = .:e{ 2
-if(t)) and F"(s) = .:e{t /(tll. Use these
6. 16y" + Sy' + 65y = 0, yeO) = 0, y'(O) = 2, [0, 47T]

value and findy(t) = .:e-I{y(s)) = 2t 2 .) termining the partial fraction expansion. Graph the solution of results to generalize F(n)(s) = .:e{
(-Ipn/(t)} so that
each problem. .:e{tn/(t)} = (-I)"F(n)(s).
*7. ym - 4y" - 9y' + 36y = 0, yeO) = I, y'(O) = 0, In Exercises 16-19, solve the initial-value problem iovolving
y"(O) = -I, [-I, I] nonconstant coefficients.
8. ym + 6y" + 9y' + 4y = 0, yeO) = 0, y'(O) = -3, 16. y' + 3ty' - 6y = 3, yeO) = 0, y'(O) = a "U(t-a)
y"(O) = 2, [-1,4]
17. y" + ty' - 2y = 4, yeO) = 0, y'(O) = a ~ Laplace Transforms of Several Important Functions
9. ym + y" + 4y' + 4y = 0, yeO) = -5, y'(O) = 0,
y"(O) = 0, [0, 37T] 18. y" + 2ty' - 4y = 2, yeO) = 0, y'(O) = a
(" Piecewise Defined Functions: The Unit Step Function (" Periodic Functions
*19. y" + 2ty' - 4y = 4, yeO) = 0, y'(O) = a (" Impulse Functions: The Delta Function
10. y" - y' - 2y = e-t,y(O) = 2,y'(0) = I, [-I, I]
20. Find the differential equation satisfied by Yes) for
*11. y' - 12y' + 40y = sin(2t), yeO) = I, y'(O) = 0, (a)y' - ty = O,y(O) = I,y'(O) = a (Airy'sequatioD) Piecewise Defined Functions: The Unit Step Function
[-7T, 7T/4]
and for (b) (I - t 2 )y" - 2ty' + n(n + I)y = 0, t=a
12. y" - 2y' + 37y = et + cos(3t), yeO) = 0, y'(O) = I, yeO) = 0, y'(O) = I (Legendre's equation). What is An important function in modeling many physical situations is the unit step function
[-1,1] the order of each differential equation involviog Y? Is Figure 8.3 au,shown in Figure 8.3 and defined as follows.
Definition 8.6 Unit Step Function ----, Theorem 8.8

r-----------~~~~~==~------------------------ __
The unit step function ilU(t - a), where a is a given number, is defined by
~-i
I Suppose that F(s) = ::£{f(t)} exists for s > b ;;, 0. If a is a positive constant, then
I
::£{f(t - a)ilU(t - a)} = e-asF(s).
ilU(t_a)={O,O:t<a. I
I, t _ a
-- I
Using the definition of the Laplace transform, we obtain

"fi"FilM'
Graph (a) ilU(t - 5) and (b) ilU(t). ::£{f(t - a)ilU(t - a)} = fe-s'[(t - a)ilU(t - a) dt
o
o:=;5t < 5' so the Jump = fa e-S'[(t - a)ilU(t - a) dt + J~e-s'[(t - a)ilU(t - a) dt

"lJ.(I-5) : ] Sit'
: 0 U Ion a In this case, ilU(t - 5) = {O,
( ).
I, t;;,
. occurs at o I a '-v---"
='0 = I
t = 5. We graph ilU(t - 5) in Figure 8.4.
(b) Here ilU(t) = ilU(t - 0), so ilU(f) = 1 for t ;;, 0. We graph this function in Figure
8.5. = re-s'[(t - a) dt
a
Changing variables with u = t - a (where du = dt and t = u + a) and changing

The unit step function is useful in defining functions that are piecewise con.
1= 5 °
tinuous. Consider the functionf(t) = ilU(t - a) - ilU(t - b). If :=; t < a, thenf(t) =
the limits of integration, we have
Figure 8.4 °-°

0.
= 0. If a :=; t < b, thenf(t) = 1 - °
= 1. Finally, if t ;;, b, thenf(t) = 1 - 1 = fo~e-S(u+a)f(u) du = e- as fe-SUf(u) du = e-as::£{f(t)} = e-asF(s).
0
O, O:=; t < a
"lJ.(I) Hence, ilU(t - a) - ilU(t - b) = 1, a:=; t < b, and we can define the function
{
0, t;;, b lif',,','.,
O, O:=;t<a . Find (a) ::£{ilU(t - a)}, a > 0; (b) ::£{(t - 3)silU(t - 3)};
get) = h(t), a:=; t< b : (c) ::£{sin(t - 1'/6)ilU(t - 1'/6)}.
{
0, t;;, b
: J Solution (a) Because ::£{ilU(t - a)} = ::£{I . ilU(t - a)}, f(t) = 1. Thus
as get) = h(t)[ilU(t - a) - ilU(t - b)], which is shown in Figure 8.6. f(t - a) = I, and
1-0
Similarly, a function such as
Figure 8.5
f(t) = {g(t), o:=; t < a
::£{ilU(t - a)} = ::£{I' ilU(t - a)} = e-as::£{l} = e-as(+) = e~as.
h(t), t;;, a
g(l)
(b) In this case a = 3 and f(t) = tS. Thus,
can be written as
5' 120
::£{(t - 3)'ilU(t - 3)} = e- 3s::£{tS} = e- 3s -i = - 6 e- s.
3
f(t) = g(t)[ilU(t - 0) - ilU(t - a)] + h(t)ilU(t - a) S s
= g(t)[1 - ilU(t - a)] + h(t)ilU(t - a). (c) Here a = 1'/6 andf(t) = sin t. Therefore,
The reason for writing piecewise continuous functions in terms of unit functions
is because we encounter functions of this type in solving initial-value problems. Using , ::£{Sin(t - i)ilU(t - i)} = e-=16::£{sin t} = e-=16 S2 ~ ;::1:.
1=
I-a ~e methods in Chapters 4 and 5, we solved the problem over each piece of the func·
tlon. However, the method of Laplace transforms can be used to avoid those compli-
Figure 8.6 cated calculations. We state the following theorem.
448 Chapter 8 Introduction to the laplace Transform 8.3 laplace Transforms of Several Important Functions 449
In most cases, we must calculate

Itt) ;£-1 _e-~} I - 4)2OU(t - 4).
_ = J(t - 4)OU(t - 4) = -(t
;£{g(t)OU(t - a)}
"
4 { S3 2
instead of ;£{f(t - a)OU(t - a)}. To solve this problem, we let g(t) = J(t - a) so This ftmction is shown in Figure 8.7(a). For what values of t isJ(t) > O?
J(t) = get + a). Therefore, '
(b) In this case, a = 7T/2 and F(s) = s2116' Then J(t) = ;£-IL2116} =
;£{g(t)OU(t - a)} = e-as;£{g(t + a)}.
~ sin 4t and
6 r
(a)
;£-t;-+=~6} = J(t - f)OU(t - f) ± = sin 4(t - f)OU(t - f)
Calculate (a) ;£{t 2OU(t - I)}; (b) ;£{sin tOU(t - 7T)}.
,-
tsin4r"lL(t-rr.I2) = t sin(4t - 27T)OU(t --f) t = sin 4t ou(t --f).
.1 Solution (a) Because get) = t 2 and a = 1, 1tf2 1t We graph this ftmction in Figure 8.7(b). For what values of t doesJ(t) = O?
-1
;£{t2OU(t - I)} = e-s;£{(t + l)2};= e- s;£{t2 + 2t + I} = e- s U3 + s~ +~). ,;
(b) With the unit step ftmction, we can solve initial-value problems that involve piece-
(b) In this case, g(t) = sin t and a = 7T. Notice that sin(t + 7T) = sin t cos "IT + figure 8.7 (a)-(b) wise continuous ftmctions.
cos t sin 7T = - sin t. Thus
;£{sin tOU(t - 7T)} = e-=;£{sin(t + 7T)} = e-=;£{ -sin t}
I -""S
=-e -= - - = - e- -
,'f',,',I.,
S2 +I S2 + l' Solve y" + 9y = {Ol', o:=; t < 7T, subject to yeO) = y'(O) = O.
L t"" 7T
Find ;£{cos tOU(t - 7T)}. :1 Solution To solve this initial-value problem, we must compute ;£{f(t)}, where
Theorem 8.9 follows directly from Theorem 8.8. , J(t) = {I, o:=; t < 7T. Because this is a piecewise continuous ftmction, we write
0, t"" 7T
it in terms of the unit step ftmction as
Theorem 8.9 J(t) = 1[OU(t - 0) - OU(t - 7T)] + O[OU(t - 7T)] = OU(t) - OU(t - 7T).
Suppose that F(s) = ;£{f(t)} exists for s > b "" O. If a is a positive constant and Then
J(t) is continuous on [0, 00), then . 1 e-=
;£{f(t)} = .:£{I - OU(t - 7T)} = -; - -s-·
;£-1 {e -asF(s)} = J(t - a)OU(t - a).
Hence
.:£{y"} + 9;£{y} = ;£{f(t)}
,"',,'4.' s2y(S) - sy(O) - y'(O) + 9Y(s)

1 e-=
= -; - -.-s-
i: { -4S}
Find (a) ;£-1 ~ ; (b) ;£-1 _~_ _ .{ -""S/2} 1 e- m
: s s + 16 (S2 + 9)Y(s) = -; - -s-
: J Solution (a) Ifwe write the expression e- 4s/s 3 in the form e-asF(s), we see that 1 e-=
a = 4 and F(s) = I/s 3 • HenceJ(t) = .:£-1 {l/s3} = t 2 /2 and Yes) = S(S2 + 9) - S(S2 + 9)'
450
Chapter 8 Introduction to the Laplace Transform
8.3 Laplace Transforms of Several Important Functions 451
Then
Definition 8.7 Periodic Function

yet) = ~-I{y(S)} = ~-1{-2_1_}
s(s + 9)
_~_I{ ~-m }.
s(s + 9) A functionJ(t) is periodic if there is a positive number T such that
Consider ~-I{e-=/[s(s2 + 9)]}. In the form of ~-I{e-asF(s)}, a = 'IT and

J(t + T) = J(t)
F(s) = 1I[s(s2 + 9)]. Now J(t) = ~-I{F(s)} can befound with either a partial fra • for all t:2: O. The minimum value of T that satisfies this equation is called the
tion expansion or with the formula C period of J(t).
J(t)=~-t(S2~9)}= f~-t2~9}da= ff sin 3ada The calculation of the Laplace transform of periodic functions is simplified
= _1.[~]t
3 3 0
=..!.-1.cos3t
9 9 .
through the use of the following theorem.
Then with cos(3t - 3'IT) = cos 3t cos 3 'IT + sin 3t sin 3 'IT = -cos 3t, we have Theorem 8.10 Laplace Transform of Periodic Functions
CO-I{ e-= } [I 1 ,.; ] Suppose thatJ(t) is a periodic function of period T an~ thatJ(t). is pie~ewise con-
J..- S(S2 + 9) = 9" - 9" cos 3(t - 'IT) OU(t - 'IT) tinuous on [0, 00). Then ~{J(t)} exists for s > 0 and IS detemuned WIth the def-
inite integral
= [t -t cos(3t - 3'17) ]OU(t - 'IT) = [t + t cos 3t]OU(t - 'IT).
Combining these results yields the solution
yet) = ~-I{y(S)} = ~-1{-2-1_}

, s(s + 9)
_~-I{~}
+ s(s- 9)
PROOf Of THEOREM 8.10
We begin by writing ~{J(t)} = fe-s'f(t) dt as the sum

o
= 1.
9 - 1.9 cos 3t - [1.9 + 1.9 cos 3t]OU(t - 'IT)
.
~{J(t)} = rre-s'f(t) dt + fe-s'f(t) dt.
Notice that we can rewrite this solution as the piecewise defined function
Jo T
If we change the variable in the second integral to u = t - T where du = dt, we

II obtain
- - - cos 3t, 0 s t< 'IT
IIY y(t) = 9 9 ,
[ e-s'f(t) dt = r~e-S(U+TlJ(u + T) du = e- sT fe-suJ(u) du = e-sT~{J(t)}.
-c==:===-. ~;;;:;;:"""-""'i1t { 2
~ ~
-I
l
-9 cos 3t, t:2: 'IT T 0
= feu)
Then,
r
Figure 8.8 which is graphed in Figure 8.8. Can you determine where the behavior of yet)
changes? Are the initial conditions satisfied?
~{f(t)} = rT e-s'f(t) dt + r~ e-s'f(t) = e-s'f(t) dt + e-sT~{J(t)},
Jo Jr 0
Periodic Functions which can be solved for ~ {J(t)} to yield

T
Another type of function that is encountered in many areas of applied mathematics is
the periodic function. ~{J(t)} = ~ r e-s'f(t) dt.
1- e Jo
452 Chapter 8 Introduction to the laplace Transform 8.3 laplace Transforms of Several Important Functions 453
0~. Is the Laplace transform of a periodic function a periodic function? 2
.~"
""".,'.,1
(1 - e "")(S2 + I) .
Taking the Laplace transfonn of both sides of the equation and solving for Yes)
gives us
Find the Laplace transfonn of the periodic functionf(t) = t, 0 :5 t < 1,/(t + I) '"
f(t), t"" 1. ~{y"} + ~{y} = ~{f(t)}
2
:) Solu~ion
f(t)
The period off, which is graphed in Figure 8.9, is T = 1. Through in. s2y(S) - sy(O) - y'(O) + Yes) = (I _ e =)(s2 + I)
tegratlon by parts or a computer algebra system,
2
~{f(t)} =
leo
~
II'e-stt dt I-e -~s
-S']I + I1_es- dt
1 - s {[
= ---
0 0
-sc } Yes) = (I _ e m)(s2 + 1)2 .
Recall from your work with the geometric series that if Ixl < I, then
=~{_e-S _[e-S']I}=_I _ [_~+~]
3 t
Figure 8.9 eI - s e s S2 I - S2
0 S
_1_ = I + x + x 2 + x 3 + ....
I-x
1 - (s + I)e- S
s2(1 - e S) . Because we do not know the inverse Laplace transfonn of 2/[(1 - e-=)(s2 + I)],
we must use a geometric series expansion of 1/(1 - e-=) to obtain tenns for which
we can calculate the inverse Laplace transfonn. This gives us
Laplace transfonns can be used more easily than other methods to solve initial.
value problems with periodic forcing functions.
o 'ifF,,"'"'' so
Solve y" +y = = =
f(t) subject to yeO) y; (0) 0 if f(l) ={2
sin t, 0:5 t < 7T
0, 7T :5 t < 27T
and
f(t + 27T) = f(t), (f(t) is known as the half-wave rectification of 2 sin t.)
1 Solution We begin by finding ~{f(I)}. Because the period is T = 27T, we have Then
~{f(t)} = I I
- e
2"" 1
0
2
" e -S'J(t) dl y(t) = 2~-1 L2 ~ 1)2 + (S~~":)2 + (X:":)2 + (s~:~? + .. -}
= 2~
-I
{(S2
I
+ 1)2 }+ 2~ { -=}
-I
+ + 2~
(S2
e
1)2
-I _e_ _}
{-2m
(S2 + 1)2
+
-- I - e2 -2"" I"e
0
-sC'
sm I I.
d 2~-ILX:~)2} + ....
Using integration by parts, a table of integrals, or a computer algebra system yields Notice that ~-I{I/(s2 + If} is needed to find all of the other tenns. Using a com·
puter algebra system, we have
~{f(t)} = 2 [e-SC(-s ~in t - cos I)]" = 2 [e-"" + _1_]
1- e s· + I 1 - e -2= S2 + 1 S2 + I
~-I{-2-1_-o} =
2"s 0
.!.(sin I - t cos t).
2(e-= + 1) _ 2(e-= + 1) (s + It 2
(l - e 2=)(S2 + 1) - (1 - e "')(1 + e =)(S2 + 1) Then
454
Chapter 8 Introduction to the Laplace Transform
8.3 Laplace Transforms of Several Important Functions 455
y(t) = (sin t - t cos t) + [sin(t - 7T) - (t - 7T) cos(t - 7T))OU(t - 7T) +

We now try to create the idealized impulse function by requiring that 8a (t - to) act on
[sin(t - 27T) - (t - 27T) cos(t - 27T))OU(t - 27T) +
smaller and smaller intervals. From the integral calculation, we have
[sin(t - 37T) - (t - 37T) cos(t - 37T))OU(t - 37T) +
lim J(t) = I,
[sin(t - 47T) - (t - 47T) casU - 47T))OUU - 47T) + ... a--.O
We can write y as the piecewise defined function We also note that

yet)
6
4
sin t - t cos t, 0 S t < 7T lim 8a (t - to) = 0, t
a--.O
*" to.
-7T cos t, 7T S t < 27T
2
y(t) = sin t - tcos t + 7TCOS t, 27T S t< 37T, We use these properties to define the idealized unit impulse function. Notice that this
-2 { -27T cos t, 37T S t < 47T idealized function represents an instantaneous impulse of magnitude one that acts at
-4 t = to.
-6
Figure 8.10
which is graphed in Figure 8.10. Is y a smooth function? (That is, is y' continuous?)
Is y periodic? Ioefinition 8.8 Unit Impulse Function
The (idealized) unit impulse function 8 satisfies
. f() {2sint, O'StS7T
G·"
< ."
G
rap hthfi
e ,mctlOn t = 0 , 7T <::
- t < 27T
andf(t + 27T) = f(t) on the interval 8(t - to) = 0, t *" to
-::--::. [0, 67T).
f
+~
_ro 8(t - to) dt = I,
Impulse Functions, The Delta Function
In some situations, we consider a forcef(t), known as an impulse, that acts only over We now state the following useful theorem involving the unit impulse function.
a brief period of time, to - a < t < to + a. For example, we may strike a pendulum
(or spring-mass system) or there- may be a voltage surge in a circuit. In these cases,
the impulse delivered by the force is given by Theorem 8.11
50
'o+a
J=
I f(t) dt. Suppose that g (t) is a bounded and continuous function. Then
25
to-a
To better describe f(t) Suppose that this function has an impulse of lover the interval
to - a < t < to + a, so that the impulse begins at t = to - a and ends at t = to + a.
r: 8(t - to)g (t) dt = g (to).
Under these assumptions, we let
{ 2~'
The "function" 8(t - to), known as the Dirac delta function, is an example of ~ gen-
to - a < t < to +a eralized function. It is not a function of the type studied in calculus. This functlO~ IS
to- 0.05 t: to to+0.05 I f(t) = 8a (t - to) = .
0, otherwise useful in the definition of impulse-forcing functions that arise in many areas of applied
Figure 8.11 mathematics. Although this function does not possess the properties required to apply
We graph 8a U - to) for several values of a in Figure 8.11,
the Laplace transform, we can determine 5/:{ 8(t - to)} formally.
Graphf(t) = 8j(t - I). What is the maximum value off(t)?
The impulse off(t) = 8a (t - to) is given by I
- I'o+af(t) dt = I'o+a -21 dt = -21 ((to + a) - 1

I Theorem 8.12
J- (to - a)) = -(2a) = 1. For to 2: 0,
to-a to-a ex ex 2ex
Notice that the value of this integral does not depend on a as long as a is not zero.
L
PROOF OF THEOREM 8.12 5£{y"} + 5£{y} = 5£{o(t- 1')}
Because the Laplace transform is linear, we find 5£{ o(t - to)} through the follow_ s2y(S) - sy(O) - y' (0) + Yes) = e-
m
ing calculations: (S2 + I)Y(s) = e- m

-"s
Yes) =_e_ _ .
S2 + 1
We can represent the delta function oit - to) in terms of the unit step function as Hence
I m
oa(t - to) = 2a [OU(t - (to - a)) - OU(t - (to + a))]. y(t) = 5£-1{ e-+ I
S2
}.
Hence
Becausef(t) = 5£-I{l/(S2 + I)} = sin t,
5£{oa(t - to)} = 5£{2~ [OU(t - (to - a)) - OU(t - (to + a))]} y(t) = 5£-1 { ~-m } = sin(t - l')OU(t - 1') = -sin tOU(t - 1').
s +1
_ I [e-S(t,-a) e-S(t,+a)] _ (e= _ e-=)
- 2a ~---s-- =e 2as . o O$t<l'. h .
sto
The graph of yet), which can be written as yet) = {"":sin t, t ~ 1" IS S own m
Therefore, Figure 8.12 Figure 8.12. Notice thaty(t) = 0 until the impulse is applied at t = 1'.
5£{o(t - to)} = lim 5£{oa(t - to)} = lim e-st,(e= - e-=).

If we interpret the problem in Example 9 as a spring-mass system with k = m, what
O'~o O'~o 2as
is the maximum displacement of the mass from equilibrium? Does the mass come to
Because the limit is of the indeterminant form 0/0, we use L'Hopital's rule to find rest?
that
5£{0(t - to)} = lim e-st,.(e= - e-=) = lim e-st,(se= + se-=)

a~O 2aS 0'--+0 2s
= e- Slo • 1 = e- sto .
EXERCISES 8.3
In Exercises 1-22, find the Laplace transform of the given

*11. -14Sin(t- 2n"l1.(t- 2;)
Find (a) 5£{ 8(t - I)}; (b) 5£{ o(t - 1')}; (e) 5£{ O(t)}. function.
1. -28"l1.(t - 3) 12. - 3 cosU - 2)"l1.(t - 2)
: lSolution (a) In this case to = I, so 5£{o(t - I)} = e- s. (b) With to = 'IT,. 2. -16"l1.(t - 6)
s *13 J(t) = {t, 0:5 t < 1 andJ(t) = J(t - 2) if
5£{o(t - 1')} = e- ". (e) Because to = 0, 5£{o(t)} = 5£{o(t- O)} = e-s(O) = I. *3. 3"l1.(t - 8) - "l1.(t - 4) . 0, 1:5 t < 2
t"" 2 (see Figure 8.13).
4. 2"l1.(t - 5) + 7"l1.(t - 4)
,'f',,·,I.' 5. "l1.(t - 2) - 7"l1.(t - 7) - 7"l1.(t - 6)

6. 4"l1.(t - 2) + 2"l1.(t - 7) - 4"l1.(t - 8)
Solve y"
, ] Solution
+ Y = o(t - 1') subject to yeO) = y' (0) = o.
We solve this initial-value problem by talGng the Laplace transfonn
*7. -42e,-4"l1.(t - 4)
8. 6e t - 2"l1.(t - 2)
9. 12 sinh(2 - t)"l1.(t - 2)
1
k: LL
1 2 3 4 5 t
of both sides of the differential equation. This yields 10. cosh(t - 2)"l1.(t - 2) Figure 8.13
458 Chapter 8 Introduction to the Laplace Transform 8.4 The Convolution Theorem 459
lOs t < 1 1 1 64. Show that .5e{g(t)OU(t - all = e-".5e{g(t + all· Show that
14. f(l) = {2'- I, 1st < 2 andf(l) = f(t - 2) *41.
(1 +e 3')(S2 + 16)
42.
S(S2 + 5)(1 + 2e~ (Hinl: Use the definition of the Laplace transform and
ift;,,: 2. the change of variable u = t - a.) G(s) = s(I - e "')"
I 1
0S I< I *43. 44.
O, S2(S2 + 4)(3e Zs - I) (s2 + 3)(4e 3s -1) 0 S I< a
I,
65. (Square Wave) If f(t) = { _ I, a S 1< 2a' 71. Use an integration device to assist in calculating the
15.f(t)= I, 1 st<2andf(t)=f(t-3)ift;":3
{ Laplace transform of each of the following periodic
2, 2 s t< 3 1 - e- as
(see Figure 8.14).
In Exercises 45-57, solve the initial-value problem. Graph the f(l + 2a) = f(I), show that F(s) = s(1 + e-"') = functions.
solution to each problem on an appropriate interval.
*45. y' + 3y = f(t), yeO) = 0,

1 tanh
s
as.
2
(a) f(l) = t
{~I, 0 ~ ~12J(t + 2) = f(t)
f(l) = {I,0, 0 S
t;": 2
t< 2 66. (Triangular Wave) Consider the function g(l) = (b) f(t) = {i _°1,S : : 11 < 2J(t + 2) = f(l)
46. y' + 5y = f(t), yeO) = 0, X, 0 S t<a 2 ' get + 2a) = get). Show that
(c) f(l) = {sint, 0 S t < 7T f(t + 27T) = f(t)
{ 2a -x, a=:; t< a
Figure 8.14 f(t) = {Sin m, 0 S I< 1 -smt, 7Tst<27r'
0, t? 1 G(s) = J.. tanh as. (d) Use the results obtained in (a)-(c) to solve the
S2 2 following initial-value problem for each function
1,OSt<1 *47. y" - 4y' + 3y = f(I), yeO) = 0, y'(O) = 1,
16. f(t) = I, I S 1<2 andf(t) = f(t - 3) (Hint: Use the square wave function and the relation- f(t) above.
{
3 - t, 2 S t<3 , {cos(.:!!..t), 0S t <1 ship g'(t) = f(l)·) d2X
ift;,,: 3. f(t) = 2 """ji2 + X = f(t)
*67. (Sawtooth Wave) Iff(t) = t/a, 0 S t < aJ(t + a) =
17. 8(t - 7T)
*19. 8(t - 1)
*21. sin tOU(t - 27T)
+ 8(t - 2)
+ 8(t -
IS. 8(t - 3)
20. 28(t - 3) - 8(t - 1)
7T/2)
4S. y"
f(t)
+ lly' + 30y = f(t), yeO) = -2, y'(O) = 0,
= { 2:
IOSt<1
I S 1< 2 andf(t) = f(1 - 2)
f(t), show that
1
F(s) = as2 - s(1 - e as)'
e- as 1 x(O) = dx(O) = 0
dt
Plot the solution in each case to compare the results.
22. cos 2tOU(t - 7T) + 8(t - 7T) if t? 2 68. (Rectified Sine Wave) Letf(t) = Isin(m/a)l. Show that 72. Solve y" + 2y' + Y = 8(t) + 1O,000OU(t - 27T) sub-
*49. x': +x = 0(1 - 7T) + I, x(O) = 0, x'(O) = 0 7Ta as ject to yeO) = 0, y'(O) = O. For what values of t
In Exercises 23-44, find the inverse Laplace transform of the
50. x" +x = oCt - 7T) + 8(1- 27T), x(O) = 0, x'(O) =0 F(s) = a2s' + -ri' coth 2' does yet) decrease? For what values of t does y(l)
given function. increase? Determine lim,_oo yet). Is there a relation-
'51. x" -I:' 3x' + 2x = oCt - 7T), x(O) = 0, x'(O) = 0 69. (Half-Rectified Sine Wave) Iff(t) = ship between this limit and the forcing function
-3 52. x" + 4x' + 13x = 0(1 - 7T), x(O) = 0, x' (0) = 0
23. se= sin(m/a), 0 S t < a f(t + 2a) = f(I), show that 8(1) + 1O,000GiL(I - 27T)? Will the limit be affected if
*53. x" + 4x' + 13x = 8(t - 7T) + sin t, x(O) = 0, { 0, as t < 2a ' the forcing function is changed to 1000(t) +
3s
*25. 2e - 3 + e 4'
3 x'(O) = 0 1O,000OU(t - 27T)?
se 4s
26. ---;F
54. x' + 4x' + 13x = 8(t - 7T) + 8(t - 27T), x(O) = 0, *73. Solve y" + 2y' + Y = 8(t) + 8(t - 27T) subject to
3e s - 4e 3s - 3 5 - 6e' - 3e Zs x'(O) = 0 yeO) = 0, y'(O) = O. What is the maximum value of
27. se 6' 2S. se 4s 70. (Stair Step) Consider the function g (t) =
55. x' + 9x = cos t + 8(t - 7T), x(O) = 0, x' (0) = 0 yet) and where does it occur? Determine lim,_oo yet).
-4s e- 3,r
x' + 9x = cos 3t + 8(t - 1), x(O) = 0, x'(O) = 0 L OU(I - Would this limit be affected if the forcing function is
*29. _e__
s- I 30.~ 56.
*57. x' + 3x' + 2x = 0(1 - I) + e-', x(O) = 0, x'(O) = 0 .-1 na). Sketch the graph of g (t) on 0 S t S 4.
changed to 8(1) + 8(1 - 27T) + 0(1 - 47T)?
31. e- 3,_s_ 32. e- 5,-_I__

s2 + 4 S2(S2 + 1)
In Exercises 58-63, solve the initial-value problem. Compare the
*33. -oc--,,-e-_5_,__ 1 - eS + se2s
34. s'(eZs _ 1)
solution to that of the corresponding homogeneous problem. ~ The Convolution Theorem
S2 - 7s + 10
2(1 - 2e Zs - e 4') -9
58. x' + 9x' = 8(t - 7T), x(O) = 0, x' (0) = 2 (" The Convolution Theorem ;- Integral and Integrodifferential Equations
35. s2e4'(1 _ e 4,) 36. e4S(s2 _ I) 59. x" + 9x' = 8(t - 7T), x(O) = 2, x' (0) = 0
60. x" + 9x' = oCt - 7T), x(O) = 0, x'(O) = 0 The Convolution Theorem
*37. s
e 4s(s2 - I) 3S. s3(l +e 3s) *61. x" + 6x' + lOx = 30(t - 7T), x(O) = 0, x'(O) = 0 In many cases, we are required to determine the inverse Laplace transform of a prod-
62. x" + 6x' + lOx = 38(t - 7T), x(O) = 2, x'(O) = 0 uct of two functions. Just as in integral calculus when the integral of the product of
40 s
39. s\1 +e 4s) . (1 + e Zs)(S2 + 9) 63. x" + 6x' + lOx = 38(t - 7T), x(O) = 0, x'(O) = 2 two functions did not produce the product of the integrals, neither does the inverse
460 Chapter 8 Introduction to the laplace Transform 8.4 The Convolution Theorem 461
Laplace transfonn of the product yield the product of the inverse Laplace transforms (why?), which can be written as
We had a preyjew of this when we found .
F(s)G(s) = fe-sl L'f(t - v)g(v) dv dt = ;e{L'f(t - v)g(v) dV} = ;e{(f* g)(t)}.
;e-I {F;S)} = ;e-I {+F(S)}
Therefore, ;e-I{F(s)G(s)} = ;e-I{;e{(f* g)(t)}} = (f* g)(t).
earlier in the chapter. We recall that;e-I{F(s)ls} = f' f(a.)da., wheref(t) = ;e-I{F(s)},
so we realize that 0
;e-1r;S)} * ;e-I H};e-I{F(S)} = f(t).
The Convolution theorem gives a relationship between the inverse Laplace trans.
o lifi"...••
Compute (f* g)(t) and (g * f)(t) iff(t) = e- I and get) = sin t. Verify the Convo-
fonn of the product of two functions, ;e-I{F(s)G(s)}, and the inverse Laplace trans. lution theorem for these functions.
fonn of each function, ;e-I{F(s)} and ;e-I{G(s)}.
: 1 Solution We use the definition and a table of integrals (or a computer algebra
. system) to obtain
Theorem 8.13 Convolution Theorem
(f* g)(t) = f'fU - v)g(v) dv = r' e-(t-v) sin v dv = e- I (' e V sin v dv
Suppose thatf(t) and get) are piecewise continuous on [0, (0) and both of expo-
a Jo Jo
.nential order b. Further, suppose that ;e{f(t)} = F(s) and ;e{g(t)} = G(s). Then,
= e- ' [;- (sin v - cos V)l = e-{~'(Sin t - cos t) - i(sin 0 - cos 0)]
;e-I{F(s)G(s)} = ;e-I{;e{(f* g)(t)}} = (f* g)(t) = f' f(t - v)g(v) dv.
o = 1. (sin t - cos t) + 1.e- 1
22'
Note that (f* g)(t) = f' f(t - v)g(v) dv is called the convolution integral.
o , With a computer algebra system, we find that
PROOF OF THE CONVOLUTION THEOREM

(g * f)(t) = {g(t - v)f(v) dv = f: sin (t - v)e V dv = i(sin t - cos t) + ie-I.
Now, according to the Convolution theorem, ;e{f(t)};e{g(t)} = ;e{(f* g)(t)}. In

We prove the Convolution theorem by computing the product F(s)G(s) with the
definition of the Laplace transfonn. This yields this example, we have
F(s)G(s) = fe-SXf(x) dx· fe-SVg(v) dv, F(s) = ;e{f(t)} = ;e{e- I} = _1_ and G(s) = ;e{g(t)} = ;e{sint} = -z_I_.
o 0
s+1 s +1
which can be written as the iterated integral Hence
F(s)G(s) = rr
o 0
e-s(x+vY(x)g(v) dx dv. ;e-I{F(s)G(s)} = ;e-I{_I_. _I_}
s + I SZ + I
Changing variables with x = t - v (t = x + v, dx = dt) then yields should equal (f* g)(t). We can compute
t =v F(s)G(s) = rr
o v
e-s'f(t - v)g(v) dt dv,
;e- t ! 1. SZ ~ I}
where the region of integration R is the unbounded triangular region shown in Fig- through the partial fraction expansion
R
ure 8.15. Recall from multivariable calculus that the order of integration can be in·
terchanged. Then we obtain _I_._I_=~+ Bs+C
Figure 8.15
F(s)G(s) = ro
f'e-s'f(t - v)g(v) dv dt
0
s+ I SZ
where A = C = 1/2 and B = -1/2. Therefore,

+ 1 s + 1 SZ + 1'
462 Chapter 8 Introduction to the Laplace Transform 8.4 The Convolution Theorem 463
I Solution We need to find h(t), so we first note that the integral in this equation
:£-1 {_I_ . _I_} _ ..!..:£-1{_1_ + -S + I} represents (h * g)(t) for g (t) = sin t. Therefore, if we apply the Laplace transform
s+ I S2
+ I - 2 s+ I S2 + I
to both sides of the equation, we obtain
=..!..:£-1{_1}_..!..:£-l{_S}+..!..:£-1{~} :£{h(t)} = :£{4t} + :£{h(t)}:£{sin(t)}
2 s+ I 2 + I 2 S2+ I S2
or
I _, I .
= -Ze - -Z cos t + -Z sm t, 4 I
H(s) = -;z + H(s) S2 + I'
which is the same result as that obtained for (f * g)(t).
lo where :£{h(t)} = H(s). Solving for H(s), we have
Notice that in Example I, (g * f)(t) = (1* g)(t). This is no coincidence. With a H(s) ( I- S2 ~ 1) = ~
simple change of variable, we can prove this relationship in general so that the convo-
lution integral is commutative. (See Exercise 29.) so
4(S2 + I) 4 4
H(s) = S4 = -;z + S4'
,'i',,'4iEi Then by computing the inverse Laplace transform, we find that
:£-1{~ ~} = + ~t3.
r
Use the Convolution theorem to find the Laplace transform of h(t) = + 4t
s· s 3
h(t) = cos(t - v) sin v dv.
o
Laplace transforms are also helpful in solving integrodifferential equations,
: I Solution Notice that h(t) = (1* g)(t), wheref(t) = cos t and g(t) = sin t. There- equations that involve a derivative as well as an integral of the unknown function.
. fore, by the Convolution theorem, :£{(I* g)(t)} = F(s)G(s). Hence
:£{h(t)} = :£{f(t)}:£{g(t)} = :£{cos t}:£{sin t} = (-2-S-)(-2-1_)

s+1 s+1 "ii,,','.'
s
=(s2+lf'
d +y +
Solve..1'.
dt
I' y(u) du = I subject
0
. to yeO) = O.
: ) Solution Because we must take the Laplace transform of both sides of this in-
i(:1': Find the Lap/ace transform of h(t) = I: e'-V sin v dv. . tegrodifferential equations, we tITst compute
t } Yes)
Integral and Integrodifferential Equations :£
{I y(u) du
0 = :£{(I * y)(t)} = :£{I}:£{y} = -s-·
The Convolution theorem is useful in solving numerous problems. In particular, this Hence,
theorem can be used to solve integral equations, which are equations that involve an
integral of the unknown function.
:£{ ~} + :£{ y} + :£{I: y(u) dU} = :£{ I}
,'i,,,,,.Ab sY(s) - yeO)

yes)
+ Yes) + -s- =;-
I
Use the Convolution theorem to solve the integral equation s2y(S) + sY(s) + Yes) = 1
h(t) = 4t + r
o
h(t - v) sin v dv. Yes) = S2 +~ + I.
8.5 Laplace Transform Methods for Solving Systems 465
y(t) Because
Yes) = -::---"--
1
24. h(l) = 2 - ro
hey) dv
35. Express the integrodifferential equation in Exercise 27
as an equivalent second-order initial-value problem
S2 + s + 1
*25. h(l) - 4e -2, = cos I - I: sin(t - v)h(v) dv
and solve this problem.
36. Express the integrodifferential equation in Exercise 28
0.2
0.1
yet) = -.L e-
v'3
1/2
sin v'3 t.
2 26. J(I) = 3 - ro
J(t - v)v dv
as an equivalent second-order initial-value problem
and solve this problem.
The graph of y is shown in Figure 8.16. Is the initial condition satisfied? *37. Calculate (a) ~-l{ (s _ 2)'~2 + 16J,
-0.1 In Exercises 27 -28, solve the given integrodifferential equation
~-t~}[;s - 2(/+ 4)Jl

using Laplace transforms.
Show that the integrodifferential equation in Example 4 is equivalent to the second_ (b)
order differential equation y" + y' + Y = 0 by differentiating the integrodifferential *27. d1: - f'
+ 40 y(v) dv = 13e 2<, yeO)
~-l{f.-[ 4(/+ 9) -
= 0
Figure 8.16
Graph of equation with respect to t. Substitute t = 0 into the integrodifferential equation to
4y
(e) 4j: 1)]l
yt
2
() = - e
V3
-112. V3
sm-t
2
find that y' (0) = 1. Solve the initial-value problem y" + y' + Y = 0, yeO) = 0,
y'(O) = 1. Is this problem equivalent to the initial-value problem in Example 4?
28. dt
dx f'
+ 16 ox(v) dv = sin 41, x(O) = 0 38. Verify each of the following:
29. Show that (J * g)(t) = (g • !l(I) by verifying that (a) (sin kt) • (cos kt) = f sin kt;
f'J(u)g(1 - u) du
o 0
= r g(u)J(t - u) dl. (Therefore,
(b) (sin kt) * (sin kt) = ;k sin kt - f cos kt;
EXERCISES 8.4 the convolution integral is commulalive.)
30. Show that the convolution integral is associative by (e) (cos kt) * (cos kt) = ..!.. sin kt + ..!.. cos kt.
proving that U. (g * h»(t) = (U* g) * h)(I). 2k 2
31. Show that the convolution integral satisfies the distrib- (d) Use these results to verify that
In Exercises 1-6, compute the convolution U' g)(t) using the
indicated pair of functions.
In Exercises 13-18, find the inverse Laplace transform of each utiveproperty (J* (g + h»(I) = (J* g)(I) + U* h)(I).
~-l{ (S2 :k2)2} = f sin kt,
function using the Convolution theorem and the results of Ex. 32. Show that for any constant «if) * g)(I) = k(J* g)(I).
~-l{ (S2 +k k')' }
2
ercise 1.
1. J(t) = I, get) = t 2 2. J(t) = e- 31, get) = 2 *33. Show that (sin I) * (cos kt) = (cos I) * Gsin kt). = ..!..sinkt- ..!..cos kt
2k 2 '
*3. J(t) = t, get) = e- I
4. J(t) = sin 2t, get) = e- I
13. _'_1_
S2(S + I)
14. _ _1_ _
s\s + 3)
34. Show that
tion 8.1.)
1-
112
* 1- 112 = 7T. (See Exercise 31, Sec- and ~-l{
(s
2+S2k 2)2} = ..!.. sin kt
2k
+ ..!.. cos kt.
2
5. J(t) = 2, get) = sin 4t
6. J(t) = sin 2t, get) = e- 71 *15. (S2 ~ 1)2 16.
(S2
1
+ 4)(s + I)
~ laplace Transform Methods for Solving Systems

17 1 18. 1
In Exercises 7-12, find the Laplace transform of h. • S4(s2 + 16) (S2 + 4)(s + 7)
7. h(t) = f' e-
o
v
dv
In Exercises 19-22, find the inverse Laplace transform of each In many cases, Laplace transforms can be used to solve initial-value problems that in-
8. h(t) = I: sin v dv
function using the Convolution theorem.
19. 1 20. 1
volve a system oflinear differential equations. This method is applied in much the same
way that it was in solving initial-value problems involving higher order differential
equations, except that a system of algebraic equations is obtained after taking the
*9. h(t) = I: (t - v)sin v dv
(S2 + 16)(s - 10) (s - 4)'(s2 + 100) Laplace transform of each equation. After solving for the Laplace transform of each
of the unknown functions, the inverse Laplace transform is used to find each unknown
22. s
10. h(t) = f'
o
v sin(t - v) dv
*21. (S2 + 100)2 (s - 1)8(S2 + 25) function in the solution of the system.
11. h(t) = f: ve' - V dv

In Exercises 23-26, solve the given integral equation using
Laplace transforms.
12. h(t) = f: eV(t - v)' dv 23. get) - t = - I: (t - v)g(v) dv

Solve X' = ( 0l O X 1) + ( cost t)
2 sin .
subject to X(O) = (2)
O'
466 Chapter 8 Introduction to the laplace Transform 8.5 laplace Transform Methods for Solving Systems 467
Previously we solved systems that involve higher order differential equations with
Let X(t) = (;~h)' Then we can rewrite this problem as differential operators. In many cases these systems can be solved with Laplace trans-
forms as well.
X' = y + sin t x(O) = 2 (0) = 0
{ y' = X + 2 cos t' ,y .
Taking the Laplace transform of both sides of each equation yields the system "f',,"'M,
sX(s) - x(O) = Yes) + - 21--
s + I
,
,
Solve { , +
x
x:
= 3X' - y' - 2x + y, x(O) = 0, x'(O) = 0, yeO) = -1.
y= 2x -y
2s '
+ +1
1sY(s) - yeO) = Xes) S2 : J Solution
.
We begin by taking the Laplace transform of both equations. For the
first equation, this yields
which is equivalent to
.:£{x"} = .:£{3x' - y' - 2x + y}
1
sX(s) - Yes) = - 2 -- +2 S2X(S) - sx(O) - x'(O) = 3(sX(s) - x(O» - (sY(s) - yeO»~ - 2X(s) + Yes)
s +1
2s . S2X(S) = 3sX(s) - sY(s) - I - 2X(s) + Yes)
xU)
30
1 -Xes) +sY(s) = -2--
s + 1
(S2 - 3s + 2)X(s) + (s - J)Y(s) = -1,
Multiplying the second equation by s and adding the two equations, we have
and for the second equation,
20
2s 2 2S2 + 3
(S2 - I)Y(s) = - - +- -- .:£{x' + y'} = .:£{2x - y}
10 S2 + 1 S2 + 1
2
sX(s) - x(O) + sY(s) - yeO) = 2X(s) - Yes)
-.J~""-
Y(s) = 2s2+1 +_2_= 4s +3
-3 -2 2 3 t
(S2 + I)(s2 - 1) S2 - 1 (S2 + I)(s2 - 1) (s - 2)X(s) + (s + I)Y(s) = -1.
// -10
/y(t)
= _._7_. _ ._._7_ + 1 After factoring the coefficient of X(s) in the first equation, we have the system of
(a) 4(s - 1) 4(s + 1) 2(S2 + 1)' equations
Taking the inverse Laplace transform then yields

(S - 2)(s - 1)X(s) + (s - I)Y(s) = -I
15 Y
{ (s - 2)X(s) + (s + J)Y(s) = -1.
10 y(t) = -i e' - -i e-' + t sin t. Multiplying the second equation by the factor -(s - 1) yields
At this point, we can solve the system for X(s) and compute x(t) with the inverse
-(s - I)(s - 2)X(s) - (s - I)(s + I)Y(s) = s - 1.
o+-f-t..--<-+--+- Laplace transform. However, we can find x(t) more easily by substituting yet) Adding this result to the first equation then eliminates Xes), so we can solve for Yes)
10 15 20 x
into the second differential equation y' = x + 2 cos t. Because y' (t) = *e' + ie -, + with
-5
t cos t and x = y' - 2 cos t, we have
[(s - 1) - (s - I)(s + I)]Y(s) = s - 2
-10
-15 x(t) = y'(t) - 2 cos t = -ie' + -ie-' + t cos t - 2 cos t -s(s - I)Y(s) = s - 2
s-2
:1» 7, 7 _, 3 Yes) = - s(s - 1)"
Figure 8.17 (a) Graph of
="4 e +"4 e - 2" cos t.
xCI) andYCI) for -3 :5 1:5 3 Using the partial fraction expansion
(h) Graph of {;i~~ for In Figure 8.17, we graph x(t), yet), and {;rh' What is the orientation of t~h? s - 2
Yes) = - s(s - 1) = - -;
(A + s -B)I
-3:51:53 L
8.5 Laplace Transform Methods for Solving Systems 469
shows that A = 2 andB = -1, so
Yes) = _(ls __s -1_) 1

and yet) = .:e-l{_(~ - s ~ I)} =
In a similar manner, we eliminate Y(s) from the original system of equations by
-2 + e' . - EXERCISES 8.5
In Exercises 1-26, use Laplace transforms to solve each initial- Ix' - y = 0

multiplying the first equation by the factor (s + 1) and the second equation by X(t) . *15. V + x = J(t), where
- (s - 1). This yields value problem. Graph x(t), yet), and { yet)" Descnbe other
x(O) = 0, yeO) = 0
methods that could be used to solve the system.
(s - 2)(s - I)(s + 1)X(s) + (s - I)(s + I)Y(s) = -(s + I) J(t) = {Sin t, if 0 ,; t < 7T
{ -(s - 2)(s - 1)X(s) - (s - I)(s + I)Y(s) = s - l. X' - + 3y = 0
2x 0, if t 2: 7T
1. ,x(O) = 0, y (0) = 4
Adding these equations results in
{
y' + 9x + 4y = 0 1 { x' + 7x + 4y = J(t)
6. y' + 6x - 3y = 0, where
X' + 9x - 2y = 0
(s - 2)(s - I)[s + I - 1]X(s) = -2 2. {
+ lOx - 3y = 0
, x(O) = -2, yeO) = 0 x(O) = l,y(O) = 0
s(s - 2)(s - 1)X(s) = -2

y'
5x - 5y = 0
X' - J( t) = {I,0, ift
if 0,; t < 1
1
2:
*3. { y' + 4x + 3y = 0,x(O) = 2, yeO) = 0
-2 X' - 2x + 4y =
0
Xes) = s(s - 2)(s - I)' + 2x - 4y = 0
X' 17. { y' _ x - 2y = J(t), where
4. , x(O) = 0, yeO) = 6
With the partial fraction expansion
{ y' + 2x - 2y = 0
x(O) = -1,y(O) = 0
X' - 5x + 4y - 2z = 0 I, if 0 oS t < 1
-2 ABC *5 y' + 2x + 2y + 2z = 0, J(t) = -1, if!,; t<2
Xes) = s(s - 2)(s - 1) = ~ + s - 2 +~, {
0, if t 2: 2
we find that A = -I, B = -I, and C = 2. Therefore,

·
(
X(0)=0.y(;;:~,:(~)=15 X' - 2x + 7y = J(t)
+ 5x -
4y + 2z = 0 18. { y' _ x + 2y = 0, where
I I 2.
Xes) = -~ - s _ 2 +~; so x(t) =
;e-l{ I
--; - s - 2
I 2 }
+-;-=-y
X'
6 y' - 6x - 2y - 2z = 0,
· z' - 5x - 7y + 6z = 0
x(O) = O,y(O) =-1
( t, if 0 ,; t < I d
x(O) = 2, yeO) = 0, z(O) = 2 J(t) = { 2 - t, if 1 ,; t <2 an
= - I - e 21 + 2e'.
X' - X - 3y = e 4 ' J(t) = J(t - 2) if t 2
{;rk
2:
+y
{;~~. What is the orientation of { y' _ 5x
*7. = 0 ,x(O) = 0, yeO) = 0
In Figure 8.18 we graph x(t), yet), and X' + 2x + 3y = 0
y' _ x + 6y = J(t), where
8 {x'y' ++ 4x5x ++ 4y3y = e"0 (0) = 0 Y (0) = I
· = X ,
*19. {
y x(O) = I, yeO) = 0
2 0.5 9 {x' + 2x - 3y = 0 (0) = 1 (0) = 0 O, if 0 ,; t < 1
· y' - 2x + Y = 0' x ,y J(t) = 1, if 1 ,; t < 2 and
{2, if2'; t < 3
-2 -1.5 -1'""-0.5 0.5 x
X' - 3x - 6y = 0 _ _ _
10. { y' _ x + 2y = 0' x(O) - 0, yeO) - I J(t) = J(t - 3) ift
2 t
::0.;; 2: 3
\. 11. { y'
X' -
+
5x - 4y = e - I _ _
2x + 4y = e2" x(O) - -1, yeO) - I
o{x' + 5x - 7y J(t)
2. y' + 3x - 5y = 0, where
=
-3
-4
xCtl
-l~tl·
-2.0
12.
[
y'
X' +6x+y=t
+ 2x + 7y = e 4 '
x(O) = 0, yeO) = 2
x(O) = l,y(O) =-1
t, if 0 ,; t < 1
J(t)= 1, if 1 ,; t<2 and
{
X' + 2x - 4y = cos 2t
3-t, if2,;t<3
Cal Cbl
*13. y' + 5x - 2y = sin 2t J(t) = J(t - 3) if t 2: 3
{
~;~
Figure 8.18 (a) Graph of x(t) and yet) for -2 ,; t,; 2 x(O) = O,y(O) =-1
- 2: = a
(b) Graph of {;i:~ for - 2 ,; t ,; 2 X' - 4x + 5y = e-
I
*21.
-2
d2y dx
x(O) = 2, x'(O) = 1,
' yeO) = 1, y'(O) = 2
14. y' - 5x + 2y = sin 2t ( dJl+y-Tt=cost
{
x(O) = -2, yeO) = 0
4i() Chaflter 8 Introduction to the Lafllace Transform 8.6 Applications Using Laplace Transforms 471
27. Use Laplace transfonns to solve the system termine values of a so that the output voltage, Q(t)/C (the voltage across the ca-
22 { - 2x" - x = 0 x(O) = 1, x' (0) = 1,
, 2y" - 2y + x' = e" yeO) = O,y'(O) = 1 I pacitor), matches the input voltage, E(t).
X' - y = e-
-2x" - 2y' = 0 x(O) = 2, x'(O) = 1, { y' + 5x + 2y = sin 3t'
23. { -2y" + 2y - 6x' = sin t' yeO) = -2, y'(O) = 0 'I Solution This circuit is modeled with
Graph the solution for different initial conditions.
24 {2 d 2xldt 2 2x - 2y = 1
- x(O) = -2, x'(O) = -1,
28. Use Laplace transfonns to solve
dQ 1
R-+-Q=E(t)
• d yld/ 2 + Y = sin t' yeO) = -1, y'(O) = 0
2
dt C
*25
•
{-x' - 2x + 2y' = e' x(O) = 1, x'(O) = -1,
-2y" - 2y = t' yeO) = -1,y'(O) = -2
{ -I'
2x" - 5y' = 0
- 3y - x' = sin t because L = O. Therefore, we solve the IVP,
26 { -2x + 2y - y' = / x(O) = -2, x'(O) = 1, subject to the conditions x(O) = 1, x'(O) = 0, yeO) " dQ
· -I' + y - x + 2x' = sin t' yeO) = -1, y'(O) = 2 1, and y' (0) = 1. Graph the solution. dt + Q = E(I), Q(O) = o.
The Laplace transform of E(t) is
as
1 La 1 [e-sIJa I - e-
9:{E(I)} = e- sl dt = - --=--'---0;".-
1 - e 2as 0 I - e- 2as --:::; 0 - s(l - e 2as)
~ Applications Using Laplace Transforms 1 - e- as
(' L-R-C Circuits Revisited (' Delay Differential Equations s(l - e -as)(1 + e as) s(l +e as)
L (' Couflled Spring-Mass Systems r The Double Pendulum
Taking the Laplace transform of each side of the differential equation then gives us
Laplace transforms are useful in solving the applications that were discussed in earlier I
R sections. However, because this method is most useful in alleviating the difficulties as- s9:{Q(t)} - Q(O) + 9:{Q(t)} = s(l + e-as)'
(a) sociated with problems involving piecewise defined functions, impulse functions, and
periodic functions, we focus most of oflT attention on solving spring-mass systems, so that
L-R-C circuit problems, and population problems that include functions of this type.
9:{Q(t)} = s(s + 1)(1 +e as)'
L L-R-C Circuits Revisited
Laplace transforms can be used to solve circuit problems (see Figure 8.19), which were Writing the power series expansion of -II-as gives us
+e
introduced earlier. Recall that the initial-value problems that model L-R-C and L-R
R
(b)
circuits are
~ + RI = E(t),
d2Q dQ I
I
~i!Jute 8.19 (a) L-R-C Circuit L7 +RTt+ C Q =E(t) L Thus
(b) L"R Circuit
dQ
Q(O) = Qo, J(O) = Tt (0) = Jo
and
! J(O) = Jo
1_[1 - e- as + e- 2as - e- 3as + ... J
n
9:{Q(t)} = __
R s(s + 1)
where L, R, and C represent the inductance, resistance, and capacitance, respectively. = __1_ _ ~ + e-
2as
_ e-
3as
+ ....
Q(t) is the charge of the capacitor and dQldt = J(t), where J(t) is the current. E(t) is s(s + 1) s(s + 1) s(s + 1) s(s + 1)
the voltage supply.
+ 1m
LJ'
Because 9:- 1 {lI[s(s = 1 - e- t ,
Q(t) = (l - e- I ) - (l - e-(I-a)OU(t - a) + (l - e- CI - 2a )OU(1 - Za)

""""'nD - (l - e -(1-3a) )OU(t - 3a) + ....
Consider an RC circuit in Figure 8.20 with the property R = C =!. If
Figure 8.21 shows graphs of the solution when a = 1, a = 10, a = ZO, and a = 50,
E(t) = g~' a 0:5~ ~~: and E(t + Za) = E(t), find the charge Q(t) if Q(O) = O. De- along with the input function, E(t). We notice that the output voltage, Q(t)/C, moves
8.6 Applications Using laplace Transforms 473
~loser t~ E(t) as a increases. Notice that if t is measured in milliseconds, then the

mput Wlth ~ = 1 has frequency 500 hertz (cycles per second); this corresponds to
a = 50, whIch has frequency 10 hertz. Therefore, low-frequency square waves
"i'"I,I.,
least altered by the circuit. are A tank contains 100 gal of water. A salt solution with concentration 2 Ib/gal flows
\... into the tank at a rate of 4 gal/min, and the well-stirred mixture flows out of the
tank at the same rate. If the salt concentration of the solution flowing out of the tank
equals the average concentration 1 min earlier, determine the amount of salt in the
Q tank at any time t.
Q
Solution Ifwe lety(t) represent the amount of salt in the tank at any time t, then
0.8 0.8 we determine the net change in the amount of salt in the tank with the differential
0.6 0.6
equation
0.4 0.4 dy = (2~)(4 g~l) _ (y(t - I) ~)(4~).
0.2 dt gal mm 100 gal mm
0.2
Therefore, because there is no salt in the tank initially, we solve the problem
4 t 20 30 40 t
dy 1
a=l a= !O dt = 8 - "2s y (t - 1), yet) = 0, -1 :5 t:5 o.
Q
In the exercises, we have you verify that X{ yet - I)} = e-sy(s). Using the Laplace
transform, we find
0.8 0.8
0.6 0.6 sY(s) - yeO) =!s - .l...e-sy(s)
25
0.4
(s + ;5 e-s)y(s)
0.4
0.2 0.2 = ~
20 40 80 t
8 8
60 50 100 150 200 t Y(s) = - - - -
a=20 a=50 s(s + f,e- S )
Figure 8.21
With the power series expansion 1/(1 + x) = 1- x +x2 - x3 + ... = I
n=O
(-x)",
we have
8 ~ ( 1 )n 8 ~ - (-1)" e- ns ~ (-1)" e-
ns
Yi(s) = -2 "
L _ _ e- s = -., "
L - n -- n = 8"
L -- n -n+"·
-
S n=O 25s s- n=O 25 S n=O 25 S -
Delay Differential Equations Therefore,

200
In some applications, there is a delay (or shift) in the argument of the dependent vari-
ISO ~ (-1)" (1- nr+ I
able. A differential equation that involves this type of shift is called a delay equation. y(t) = 8 ~o ~ (n + I)! rJU.(t - n).
100
Consider a mixture problem from Section 3.3. In this case, however, we assume that
there is a delay in the effect that a salt solution flowing into the tank has on the salt SO In Figure 8.22, we graph the solution [0, 150]. It appears that limH~ y(t) = 200.
concentration flowing out of the tank. To include this assumption in the mathematical We may ask how this solution differs from that of the problem that does not include
model of the situation, we assume that the concentration of the salt solution flowing 20 40 60 80 100120140 t a delay,
out of the tank equals the salt concentration in the tank at an earlier time. We illustrate dz 1
Figure 8.22
a problem of this type in the following example. dt = 8 - "2s z, z(O) = O.
.,
8.G Applications Using Laplace Transforms 475
To find x(t) and yet), we must solve the initial-value problem

y Solving this problem as a first-order linear equation, we find that z(t) '"
~:~ =
l
-200e- tIZ5 + 200. Certainly, these two functions have the same limiting value, but
2.5 how do they differ? In Figure 8.23 we graph yet) - z(t). The largest difference -5x + 2y
2 between the two functions is approximately 3, which occurs when t = 24. Another d 2y , x(O) = 0, x'(O) = 1, yeO) = 1, y'(O) = O.
1.5 observation is that yet) - z(t) ~ o. dt 2 = 2x - 2y
1
0.5 Taking the Laplace transform of both sides of each equation, we have
20 40 60 80 100120140 t
Coupled Spring-Mass Systems S2X(S) - sx(O) - x'(O) = -5X(s) + 2Y(s)
{sZY(s) - sy(O) - y'(O) = 2X(s) - 2Y(s) ,
Figure 8.23
The displacement of a mass attached to the end of a spring was modeled with a second_
which is simplified to obtain the system
order linear differential equation with constant coefficients in Chapter 5. Similarly, if
a second spring and mass are attached to the end of the first mass as shown in Figure (S2 + 5)X(s) - 2Y(s) = I
8.24, the model becomes that of a system of second-order equations. To more precisely { -2X(s) + (S2 + 2)Y(s) = s·
state the problem, let masses ml and mz be attached to the ends of springs SI and S2
Solving for Xes), we have
s, having spring constants kl and k z respectively. Spring S2 is then attached to the base
of mass mi. Suppose that x(t) and yet) represent the vertical displacement from the s2+2s+2 S2 + 2s + 2 S2 + 2s + 2
eqUilibrium position of springs S, and Sz, respectively. Because spring Sz undergoes Xes) = (S2 + 5)(S2 + 2) - 4 = S4 + tsz + 6 = (S2 + 1)(sz + 6)·
x=o
x both elongation and compression when the system is in motion (due to the spring Si Then, with the partial fraction expansion
S2 and the mass mz), according to Hooke's law, Sz exerts the force kz(y - x) on m2 and
y=o SI exerts the force -k,x on mi. Therefore, the force acting on mass m, is the sum Xes) = As + B + c~ + D,
-k,x + k 2(y - x) and the force acting on m2 is -k,(y - x). (In Figure 8.25 we show s2+1 s-+6
the forces acting on the two masses where up is the negative direction and down is we find that A = 2/5, B = 1/5, C = -2/5, and D = 4/5, so
positive.) Using Newton's second law (F = mal with each mass, we have the system
2s + I 2(2 - s)
Figure 8.24 A coupled Xes) = 5(S2 + 1) + 5(S2 + 6)·
spring-mass system
Taking the inverse Laplace transform then yields
I.
x(t) = - sm t + -2 2 ,r;:6+ 4
cos t - - cos V ot
. ,r;:6
, r;: sm v ot.
5 5 5 5v6
The initial displacement and velocity of the two masses ml and m2 are given by Instead of solving the system to find Yes), we use the differential eq1lllti pn
x(O), x'(O), yeO), and y'(O), respectively. Because this system involves second-order d 2xJdt2 = -5x + 2y to find yet). Because y = 1(x" + 5x), and
equations. Laplace transforms can be used to solve problems of this type. Recall the
following property of the Laplace transform: ;£{f"(t)} = s2F(s) - sf(O) - f'(O), where x'(t) = t cos t - + sin t + 2~ sin V6t + i cos V6t
F(s) is the Laplace transform off(t). This property is of great use in solving this prob·
lem because both equations involve second derivatives. We will see that the method is and
similar to that used in solving initial-value problems involving a single equation. With x"(t) = -51.sm t - 52 cos t +""5
12 _ r;:6 4V6., r;:6
cos vot - -5- sm vot,
Figure 8.25 Force diagram for systems, however, we end up solving a system of algebraic equations after taking the
a coupled spring-mass system Laplace transform of each equation. li 1. 2 1 2 , r;: 4V6., r;: .
y(t) = 2 \ -5 sm t - 5 cos t + ""5 cos v 6t - -5- sm v 6t + sm t +
V6t + ~ V6t)
o "f''''''.Consider the spring-mass system with m, = m2 = I, kl = 3, and k z = 2. Find the = 52.sm t + 54 cos t - 15
V6.V6 1 V6
sm 6t + 5 cos 6t.
2 cos t - 2 cos sin
position functions x(t) and yet) if x(O) = 0, x' (0) = I, yeO) = I, and y' (0) = o.
476 Chapter 8 Int,oduction to the Laplace Transform 477
8.6 Applications Using Laplace Transforms
In Fi~e 8.26(a), we graph xU) and y\t) simultaneously. Note that the initial point The Double Pendulum
of yet) IS (0, 1), :,hereas that of xU) IS (0, 0). Of course, these functions can be
graphed par~etncally m the xy-plane as shown in Figure 8.26(b). Notice that this In a method similar to that of the simple pendulum in Chapter 5 and that of the cou-
phase. plane IS dlffere~t from those discussed in previous sections. One of the rea. pled spring system, the motion of a double pendulum (see Figure 8.29) is modeled by
~ons IS that the equations in the system of differential equations are second-ord the following system of equations using the approximation sin 8 = 8 for small dis-
mstead offirst-order. Finally, in Figure 8.27, we illustrate the motion of the SPring~ placements:
mass system by graphing the springs for several values of t. em, + m2)ii8'{ + m2i,i28~ + (m, + m2)i,g8, = 0
{ m2i~8;; + m2i,i28~ + m2izg82 = 0'
where 8, represents the displacement of the upper pendulum and 8, that of the lower
(a)
pendulum. Also, m, and m2 represent the mass attached to the upper and lower pen-
dulums, respectively, and the length of each is given by i, and i 2 .
y Figure 8.29 A double
pendulum
o "ii,,','.'
Suppose that m, = 3, m2 = I, and each pendulum has length 16. If g = 32, deter-
-1 mine 8,(t) and 82 U) if 8,(0) = 1, 8,(0) = 0, 82 (0) = 0, and 8,(0) = -1.
(a) (b) (e) (d) (e) (f) (g) (h) (i) : I Solution In this case, the system is
Figure 8.27 (al t = 0 l 4(16i8'{ + 1628~ + 4(16)(32)8, = 0

(b) I =l (el t = I (d) t = (el t = 2 (f) t = ~ (g) t = 3
(b) 1= i (i) t = 4. { 162 8, + 16211~ + (16)(32)82 = 0'
(b)
l....
which can be simplified to obtain
Figure 8.26
(a)-(b)
In ~xample 3, what is the maximum displ,,!cement of each spring? Does one of the 48'{ + II~ + 811, = 0
{ 8'{ + II;; + 28, = O·
sprzngs always attain its maximum displacement before the other?
. If external forces FI(t) and F 2 (t) are applied to the masses, the system of equa- If we let .:t:{lW)} = Xes) and .:t:{82 (t)} = Yes), we have
lions becomes 4[S2X(S) - sll,(O) - 11;(0)] + [S2y(S) - SIl2(0) - 8~(0)] + 8X(s) = 0
{ [S2X(S) - sll,(O) - 11,(0)] + [S2y(S) - sll,(O) - 11,(0)] + 2Y(s) = 0
ml ~;: = -klx + k,(y - x) + FI(t)
or
SI
1 d~
m2 dt 2 = -k2(y - x) + F 2 (t)
which is again solved through the method of Laplace transforms. We investigate the
,
{
4(S2 + 2)X(s) + s2y(S) = 4s -
S2X(S)
Solving this system for Xes), we obtain
+ (s2 + 2)Y(s) = s -
1
r
x=o effects of these external forcing functions in Exercise 90. 3

3s 3 + 8s - 2
x Xi _ 3s + 8s - 2
The previous situation can be modified to include a third spring with spring con- (s) - 3s4 + 16s 2 + 16 (3s 2 + 4)(S2 + 4)'
stant k3 between the base of the mass m2 and a lower support as shown in Figure 8.28.
y=o T?e motion of the spring-mass system is affected by the third spring. Using the tech- With the partial fraction expansion
mques of the earlier case, this model becomes
S2
Xes) = As + B + Cs +D
3s 2 + 4 S2 +4'
we find that
Figure 8.28
Xes) = 1 2s - 1 + 1- 2~ + 1 .
4 3s 2 + 4 4 s- + 4
478 Chapter 8 Introduction to the Laplace Transform 8.6 Applications Using Laplace Transforms 479
Then,
1 2t v3 2t 1 1 .
O,(t) ="2 cos v3 - -8- sin v3 +"2 cos 2t +"8 sm 2t.
Differentiating, we have
0'I (t) = - -1 . 2t 1 2t . 2
- sm - - - - cos - - - sm t + -1 cos 2t
v3 v34 v3 4
(a) (b) (e) (d)
and
0,"(t) = - -
2 cos -2t- + -I-.sm -2t- - 2 cos 2t - -I sm
. 2t.
3 v3 2v3 v3 2
Using the differential equation 40? + O~ + 80, = 0 yields O~ = -40? - 89,.
Therefore,
O~(t) = -40M - 80,(t) = , ~ sin ,2~ - i cos, 2~ + sin 2t + 4 cos 2t.

v3·., v3 3 v3
Integrating, we have (e) (I) (g) (h)
0' ( )
2 t = -2I cos 2t
V3 - 2.
v3 2t
sm v3 - 1 cos 2t + 2'
2 sm 2t + c,.
Applying the initial condition 0;(0) = -I yields 0;(0) = -112 - 112 + c, = _I,
so c, = O. Again, by integrating, we obtain
V3. 2t 2t I.
O2 (t) = - - - sm , ;;: + co.s , ;;: - - sm 2t - cos 2t + Co.
4 v3 . v3 4 -
(i)
Then, because O2 (0) = 0, O2(0) =1- 1+ C2 = 0, so C2 = 0, which indicates that
v3. 2t 2t I. 2 Figure 8.31 (a) I = 0 (b) I = ~ (e) I = ~ (d) I =¥ (e) 1=5 (f) 1=11 (g) I = .if
O2 (t) = -4 sm V3 + cos v3 - "4 sm 2t - cos t. (b) I = ¥ (i) I = 10
These two functions are graphed together in Figure 8.30(a) and parametrically in Does Ihe system in Example 4 come to rest? Which pendulum experiences a greater
Figure 8.30(b) to show the solution in the phase plane. We also show the motion of displacement from equilibrium?
the double pendulum for several values of t in Figure 8.31.
EXERCISES 8.6
,
20 t -2
-I 2. Suppose that we consider a circuit with a capacitor
1. Suppose that we consider a circuit with a capacitor
Ii C, a resistor R, and a voltage supply E(I) = C, a resistor R, and a voltage supply E(t) =
8,(t) IOO, O:S; 1< 1 If L = 0 find Q(I) and 1(1) if 100 sin I, O:S; 1< 7T. If L = 0, find Q(I) and 1(1)
Figure 8.30 {0, I"" I ' {0, I"" 7T
(a) (b) (a)-(b) Q(O) = 0, C = 50- 1 farads, and R = 50 n. if Q(O) = 0, C = 10- 2 farads, and R = 100 n.
480
l'
..~
Chapter 8 Introduction to the laplace Transform 8.6 Applications Using laplace Transforms 481
*3. Consider the circuit with no capacitor, R = 100 n, and 16. Consider the circuit in Exercise 13 with E(I) '" In Exercises 20 and 21, use the given initial conditions to de- e-" 0:5 1< I
L = 100 H if E( ) = {50, 0:5 I < 1 d 1200(1 - 1). Find the current if J(O) = 100. tennine the displacement of the object of mass m attached to a
36. m = I, k = 4, C = 5,/(1) = {0, I?:. I
I 0, 1:5 I < 2 an
spring with spring constant k. COS 71'1, 0:5 I < 1
E(I + 2) = E(I). Find the current J(I) if J(O) = O. 17. Consider the R-C circuit in Figure 8.32 in which
20. m = 4, k = 16, x(O) = 1, x'(O) = 0 *37. m = I, k = 4, C = 5,/(1) = {0, I?:. 1
4. Consider the circuit with no capacitor, R = 100 n, and R = 2.5 X 10 n,
4
C = 10- 5
farad, and E(I) '"
O, 0:5 1<1
21. m = 1, k = 9, x(O) = 3, x'(O) = -2 3S. Show that the initial-value problem x' + x = 0(1),
L = 100 H if E(I) = {IOO sin I, 0:5 1<71' and
0, 71':5 1<271' x(O) = x' (0) = 0 is equivalent to x' + x = 0, x(O) =
{ 1, 1:5 I < 2 volts. Determine the output voltage, In Exercises 22-25, determine the displacement of the ~bject 0, x'(O) = 1. (Hinl: Solve each equation.)
E(I + 271') = E(I). Find the current J(I) if J(O) = O. 0, I?:. 2
of mass m attached to a spring with spring constant k If tbe
5. Consider the circuit with no capacitor, R = 100 n, and i(I)R (the voltage across the resistor), by solving the
damping is given by c dxldl. Use the initial conditions x(O) = In Exercises 39-41, assume that the mass is released with zero
appropriate integral equation. (Nole: This equation
L = IOOHifE(I) = {IOOI, 0:5 1< 1 and I, x' (0) = 0 in each case. initial velocity from its equilibrium position.
0, 1:5 1<2 cannot be solved as a differential equation because E(I)
E(I + 2).= E(I). Find the current J(I) if 1(0) = O. is discontinuous at I = 1 and I = 2.) 22. m = I, k = 6, c = 5 23. m = 1/2, k = 2, c = 2 39. Suppose that an object with mass m = I is attached
6. Consider the circuit with no capacitor, R = 100 n, and 24. m = 1, k = 13, c = 4 *25. m = 1, k = 4, C = 5 to the end ofa spring with spring constant 16. If there
c is no damping and the spring is subjected to the forc-
L = 100HifE(I) = {IOOI, 0:5 1< 1 and .-_-1JL
100(2 - I), 1:5 I < 2
E(I + 2) = E(I). Find the current J(I) if J(O) = O. I 1\
In Exercises 26-29, determine the displacement of the object
of mass m attached to a spring with spring constant k if the
damping is given by c dxldl and there is an external force f(l)·
ing functionf(t) = sin I, determine the motion of the
spring if at I = I, the spring is supplied with an up-
ward shock of 4 units.
*7. Solve the L-R-C circuit equation that was derived be-
In each case, use the initial conditions x(O) = x'(O) = O.
~
fore Example I for J(I) if L = 1 henry, R = 6 ohms, 40. An object of mass m = 1 is attached to a spring
t
C = farad, E(I) = 100 volts, and J(O) = O. R 26. m = 1, k = 6, c = 5,/(1) = ~ sin I
with k = 13 and is subjected to damping equivalent to
4 dxldl. Find the motion of the mass if the spring is
S. Solve the L-R-C circuit equation for J(I) if L = 1 henry, 27. m = 1/2, k = 2, c = 2,f(l) = le-' supplied with a downward shock of 1 unit at I = 2.
t
R = 6 ohms, C = farad, E(I) = 100 sin I volts, and
2S. m = 1, k = 13, c = 4,/(r) = e- 2'
1(0) = O. *41. An object of mass m = 1 is attached to a spring
9. Solve the L-R-C circuit equation for 1(1) using the pa-
*29. m = 1, k = 4, c = 5,f(t) = e-4< + 2e-' with k = 13 and is subjected to damping equivalent to
4 dxldl. Find the motion iff(l) = 8(1 - 1) + 8(1 - 3).
rameter values L = C = R = 1 and
Figure 8.32 In Exercises 30-37, determine the displacement of the object 42. (population Problems Revisited) Let X(I) represent
E(I) = {I, 0:5 I :5 71' 1(0) = 0 of mass m attached to a spring with spring constant k if the
0, I> 71" • the population of a certain country in which the rate
is. (Spring-Mass Systems ReviSited) Just as with sys- damping is given by c dxldt and there is a piecewise defined of population increase depends on the growth rate of
10. Solve the L-R-C circuit equation for 1(1) using the pa- tems, we may use Laplace transforms to solve the IVP external force f(I). 10 each case, use the initial conditions the country as well as the rate at which people are be-
rameter values L = C = R = I and x(O) = x'(O) = O.
~:~ + c :
ing added to or subtracted frorn the population because
E(I) = {cos I,
0,
0:5 I :5 71' J(O) = O.
I> 71' '
*11. Consider a circuit withL = I,R = 4, E(I) = 0(1 - 1),
and no capacitor. Determine the current J(I) if
(a) 1(0) = 0 and (b) 1(0) = 1.
12. Solve the initial-value problems in Exercise II using
! m
x(O) = a,
dx
+ /a
dt (0) =
= f(l)
(3
which models a spring-mass system involving one

'
spring. Suppose that m = 1 slug, k = 1 Ib/ft, and that

30. m = 1, k = 6, c = 5,/(1) = {0,
31. m = 1, k = 6, c = 5, f(l) =
sin 21 0:5 I < 71'/2
I;'" 71'/2
COS 71'1, 0:5 I < I

{0, I?:. 1
e- Z ' 0:51<1
32. m = 1/2, k = 2, c = 2,/(1) = {0, ~?:. 1
of inunigratioD, emigration, or both. Consider the IVP
x' + /a = 1000(1 + a sin I), x(O) = Xo, where a is a
constant. (a) Show that
X(I) = 1000 (1 - e- kt )
k
E(I) = 0(1 - I) + 8(1 - 2). there is no resistance due to damping. Suppose also 1000a kt k')-ld
+ 1 + k? (e- - cos I + SID I + xoe .
13. Consider an L-R circuit in which L = 1 and R = 1,
that the object is released from its equilibrium posi-
tion with zero initial velocity and the object is sub- *33. m = 1/2, k = 2, c = 2,/(1) =
. {e- 1,
2
' 0:51<1
1:5 1<2 (b) If k = 3, graph x(l) over 0 :5 I :5 10 for a = 0.2,
I, 0:5 I :5 1 . .
and E()I = { 1, I > 1 . Fmd the current If jected to the external force 0, I?:. 2 0.4, 0.6, 0.8. Describe how the value of ex affects the
1(0) = O. 34. m = I, k = 13, = 4,
solution.
f(l) = {cos I, ". 0 :5 I < 71'/2 C
14. Consider the circuit in Exercise 13 with a voltage 0, 1- 71'12 e- 2t cos 3t, O::S t < 7r
f(l) = { 0, I?:. 71' In Exercises 43-49, solve the initial-value problem using
I, 0:5 I :5 1 . d th . Show that ~{f(l)} = _,_s_ + e-'~12-2_I_.
source E( I) = { 0, I > 1 . Fm e current If Laplace transforms. Interpret each as a popUlation problem. Is
s-+I s +1 35. m = 1, k = 13, C = 4,
J(O) = O. the population bounded?
e-" cos 31, 0:5 1<71'
19. ShowthatX(s) = ~ + e-,m2--'-!--)2,SOthat 43. x' + 5x = 500(2 - sin I), x(O) = 10,000
*15. Consider the circuit in Exercise 13 with E(I) = (s + 1)- (s + 1 f(l) = e- 2 ' sin 31, 71':5 1<271'
{
0(1 - 2) + 0(1 - 6). Find the current if 1(0) = O. X(I) = ir sin I + i( -cos I - (I - 71'/2)sin 1)'lL(1 - 7d2). 0, I?:. 271' 44. x' + 5x = 500(2 + cos I), x(O) = 5000
482 Chapter 8 Introduction to the laplace Transform 8.& Applications Using laplace Transforms 483
+ 5x =
*45. x'
46. x' + 5x =
500(2 - cos I), x(O) = 5000
500(2 - sin I), x(O) = 5000
initial-value probl:m dx~dt = f(t) - kx, x(O) = xo, and
compare the solutIOn With that of Exercise 55.
1 hr, then one half of the initial dosage remains
after I hr. Therefore, we solve xoe-a(l) = xo/2 for
is Vz =' G)' (c) Show that two linearly independent
47. x' - 2x = 500(1 + sin t), x(O) = 5000 57. (Dr~g Dosage Pro~lem) The drug dosage problems a, which yields a = In 2. Similarly, we solve solutions corresponding W, = ±5V3 are
conSidered m ExerCises 54-56 may be thought of as yoe -b(4) = yol2 to find that b = (In 2)/4. Describe
48. x' - 2x = 500(cos I + I), x(O) = 5000 COS 5V3t ) and ( sin 5V3t ). those corre-
a two-~ (compartment) problem. Suppose that We what happens to the drug concentration in each ( -2 cos 5V3t -2 sin 5V3t '
*49. Suppose that the emigration function is
f(t) = {5000(2 - sin t), O:s t < 5

admmlster the drug once every 4 hours, where the dru
:oves from the gastrointestinal (GI) tract (tankA) in!
case. Assume that Co = 2.
(e) Suppose that the half-life of the drug in the GI
sponding to Wz = ±5 are V~~s5~t) and (2s~~n5~t ).
0, I;;e: 5 . e mculatory system (tank B). Let x(t) represent th tract is 30 min and that in the circulatory system (c) Form a general solution to the system by taking a
Solve x' - x = f(t), x(O) = 10,000. Determine

amount of drug in the GI tract at time t hours, yet) b: it is 4 hr. How does this change affect the drug
linear combination of the four solutions. (d) If the ini-
tial conditions are x(O) = 0, y (0) = 4, x' (0) = 0, and
the amount of drug in the circulatory system, and C(I) concentration in the circulatory system?
limr_oof(t). y' (0) = 0, determine the solution to the initial-value
be the dosage function. If a and b are the rates at whi h (f) Suppose that the half-life of the drug in the cir-
50. Suppose that the emigration function is the drug is consumed in the GI tract and the circul~_ culatory system in part (d) is 5 hr. How does this
problem.
f(t) = {5000(1 + cos I), O:s t < 10 tory system, respectively, and if initially neither the G[ change affect the drug concentration in the circu- 60. (Forced System: Method of Undetermined Coeffi-
0, t;;e: 10 . Solve
tract nor the circulatory system contain the drug cients) Suppose that the system can be written as
latory system?
x' - x = f(I), x(O) = 5000. Determine limr_oof(t) if model this situation with the system ' We x" = Ax + f, where f = F 0 cos wt. Show that if a par-
it exists.
51. If the emigration function is the periodic function
f( t) = {5000(1 + cos t), O:s I < I
: = e(t) - ax, x(O) = 0
(g) How does adding a buffer to the drug to increase
the half-life of the drug in the GI affect the drug
concentration in tbe circulatory system?
ticular solution bas the form xp(t) = c cos wt, tben c
satisfies the system (A + wZI)c = -Fo. If w 5,
5V3, find a particular solution of the system in the
*
0, l:s t < 2
solve x' - x = f(t), x(O) = 5000.
,J(I + 2) = f(I),
Suppose that
1 dy
dt = ax - by, yeO) = 0
58. (The Stiffness Matrix) (a) Sbow tbat we may write
the system
0
previous system if Fo = (5 0). (Note: The constants
will depend on w.)
~;~ =
52. If the emigration function is the periodic function
-k,x + k,(y -
r
x)
= g~OO?:S-1 ~n;),
ml 61. (Delay Equations) According to the definition,
f(l) O:s 1< \ f(1 + 2) = f(t), O:s t < 1/2
°
Co
e(l) = { 0,' I12:s 1<4 ' e(t - 4) = e(I). dZy :£{ yet - to)} = e-Sry(t - to) dt. If yeO) = for
solve x' - x = f(t), x(O) = 5000.
53. Suppose that a bacteria popUlation satisfies the dif-
(a) Show that for 0 :s t :s 24 (over one day), 1 m2 dl 2 = -kz(y - x)
- to :s t :s 0, use the change of variable v = I - to to
ferential equation dxldl = x + 2008(1 - 2) with
x(O) = 100. What is the popUlation at I = 5? .c(t) = I
6
eo["U(t - 4n) - "U(t - 4n - 112)] aod
in matrix form Mx" = Kx , where M = (nll0 °)
m2
sbow that :£{y(t - to)} = e-r.,y(s) .
. 11=0 In Exercises 62-67, solve tbe coupled spring-mass system

54. Suppose that a patient receives glucose through an IV e is the mass matrix and K = (-(klk~ kz) ~~J is modeled by
tube at a constant rate of e grams per minute. If at the :£{e(t)} = I6
....!!.[e- 4ns - e-(4n+ll2lj.
11=0 s the stiffness matrix_ (b) Show tbat we can write Mx"
same time the glucose is metabolized and removed
from the bloodstr.eam at a rate that is proportional to (b) Use the Laplace transform to solve dxldt = = Kx as x" = Ax, where A = M-IK. (Why does
c(t) - ax, x(O) = O. (Note: M- I exist?) (c) Assume that x(t) = ve'" is a solution
the amount of glucose present in the bloodstream the
of x" = Ax. Sbow that x(t) = Veal satisfies the system
rate at which the amount of glucose changes is ~od
eled by dxldt = e - kx, where X(I) is the amount of
Xes) = f __e_+o_[e-
.70S(S a)
4ns _ e-(4n+II2)S])
.
if and only if if = A, wbere A is an eigenValue of
using the indicated parameter values and initial conditions.
A and v is an associated eigenvector. (d) The eigen-
glucose in the bloodstream at time I and k is a con· 62. m, = 2, mz = 2, k, = 6, kz = 4, x(O) = 0,
stant. [fx(O) = xo, use Laplace transforms to findx(I).
Does x( I) approach a limit as t .... oo?
(c) Use the Laplace transform and x(t) to solve
dyldl = ax - by, yeO) = O. (Note:
values of A are typically negative real numbers.
Show that two solutions of the system are X,(t) = x'(O) = -1,y(O) = O,y'(O) = °
nl, = 2, nlz = 2, k, = 6, kz = 4, x(O) = 0, x'(O) = 0,
±.....E£...[(1. __ 1_)e-
Z
v cos wt and xz(t) = v sin wi, wbere if = A = - W 63.
*55. Suppose that in Exercise 54 the patient receives glu-
cose at a variable rate e(l + sin I). Therefore, the rate
Yes) =
n~Os +b s s +a
4ns or a = -:±wi. yeO) = 1, y'(0) = °
59. Let m, = 2, nl2 = 112, k, = 75, and kz = 25 in 64. nl,= 2, mz = 1, k, = 4, k2 = 2, x(O) = 0, x'(O) = 0,
at which the amount of glucose changes is modeled
by dxldl = e(l + sin t) - kx.Ifx(O) = xo, use Laplace
_(1.s __s +1_)e-(4n+l/2><].)
a
the previous system. (a) Show that the eigenvalues yeO) = O,y'(O) =-1
transforms to find x(t). How does the solution of this _ (-50
of A = M 'K = 50
25/2)
-50 are A, = -75 and *65. m, = 2, mz = I, k, = 4, kz = 2, x(O) = I, x'(O) = 0,
(d) If the half-life of the drug in the GI tract is I hr
problem differ from that found in Exercise 54? yeO) = 0, y'(O) = 0
and that in the circulatory system is 4 hr, then A2 = -25. Then, aT = A, = -wt
implies that W, =
66. m, = 2, nl2 = 2, k, = 3, k z = 2, x(O) = 0, x'(O) = 1,
56. If the person in Exercise 54 receives glucose periodi-
cally according to the function
grapb tbe amount of drug in the GI tract and the ±5V3; ,,~= Az = -oJ implies that = ±5. "'2 yeO) = 0, y'(O) = 0
circulatory system over [0, 24]. (Recall that we (b) Show that an eigenvector corresponding to AI =
e, O:s t < 1
f( I ) = { 0, l:s t < 2' f(1 + 2) = f(t), solve the discussed half-life with regard to exponential de·
cay in Chapter 3.) We note that if the half-life is
.
-75 IS v, = (I) .
-2; that correspondmg to Az = -25
67. m, = 2, mz = 2, k, = 3, k2
yeO) = 0, y'(O) = 0
= 2, x(O) = I, x'(O) = 0,
In Exercises 68-71, solve the coupled spring-mass system where L is the length of each pendulum, g is th ±q2i, where q, and q2 are real numbers. What does occur? (e) Find lim,_oo I(t) (the steady-state current).
modeled with gravitational constant, and ~ = giL. Use the metho~ this tell you about the solutions to this system? Does (d) What is the frequency of the steady-state current?
~f Laplace. transforms to solve this system if the this agree with the solution obtained with Laplace
m, ~~~ = -k,x + k2(y - x) + F,(t), lllltial conditions are x(O) = a, x' (0) = b, yeO) ~ c
~~=~ ,
transforms?
90. Solve the problem of the forced coupled spring-mass
system with m, = m2 = 1, k, = 3, and k2 = 2 if
86. In (a)-(d), determine the motion of the object of mass
1 d-y the forcing functions are F,(I) = 1 and F 2 (1) = sin t

m2 dl 2 = -k2(y - x) + F 2(t) 81. Solve the initial-value problem m attached to a spring with spring constant k if the and the initial conditions are x(O) = 0, x'(O) = 1,
damping is given by c dxldl and there is a piecewise yeO) = l,y'(O) = O. Graph the solution parametrically
mXu + m~x = -k(x - y)
defined periodic external force f(I). Use the initial
where m, = 2, m2 = I, k, = 4, k2 = 2, x(O) = 1, x'(O) = 0, as well as simultaneously. How does the motion
my" + m~y = -key - x)
yeO) = 0, y'(O) = 0 using the given forcing functions. conditions x(O) = x' (0) = 0 in each case. differ from that of Example 3? What eventually hap-
x(O) = -I, x'(O) = 0
[ (a) m = 1, k = 6, c = 5, pens to this system? Will the objects eventually come
68. F,(I) = 1, F 2(1) = 1 yeO) = 1, y'(O) = O.
69. F,(I) = 1, F 2 (1) = sin I
70. F, (I) = cos I, F2 (t) = sin I
What is the maximum displacement of X(I) and yet)? f(l) = g: ~::~ ;,J(I + 2) = f(t)
to rest?
91. Consider the three-spring problem shown in Figure
82. Solve the initial·value problem (b) m = 1, k = 6, c = 5, 8.25 with m, = m2 = 1 and k, = k2 = k, = 1. Deter-
{~1' 0 ~ ~ ~ ~ 2,J(1 + 2) = f(t) mine x(t) and y(l) if x(O) = 0, x'(O) = -1, yeO) = 0,
*71. F,(I) = cos 21, F 2 (t) = 0 mxu + m~x = -k(x - y)
f(t) =
{ my" + m~y = -key - x)' and y' (0) = 1. Graph the solution both simultaneously
In Exercises 72-79, solve the coupled pendulum problem us- (e) m = 1, k = 13, c = 4, and parametrically. When does the object attached to
ing the parameter values m, = 3 slugs, m2 = I slug, £, = £2 = xeD) = 0, x'(O) = 1, yeO) = 0, y'(O) = O. VerifY that the top spring first pass through its equilibrium posi·
16 ft, and g = 32 ftls 2 and the indicated initial conditions. the initial conditions are satisfied. f(l) = {osin I, ~ $<t 2< ",J(I + 2,,) = f(t) tion? When does the object attached to the second
, 7T' _ t 7T
83. The physical situation shown in Figure 8.34 is mod. spring first pass through its equilibrium position?
72. 0,(0) = I, 0;(0) = 0, O2(0) = 0, 0;(0) = 1 (d) m = 1, k = 13, c = 4,
eled by the system of differential equations
I 0 $ t< 1 92. Solve Exercise 91 with the forcing functions F,(t) = 1
73. 0, (0) = 0, 0;(0) = 0, 11,(0) = 0, 0;(0) = -1 f(l) = { ":1, 1 $ t < 2,J(t + 2) = f(t)
74. 0,(0) = 1, 0;(0) = 1, O2(0) = 0, 0;(0) = 0 m ~;~ + 3kx - Icy = 0
and F 2 (t) = cos t. Graph the solution both simultane-
ously and parametrically. When does the object at-
(e) Use a graphing device to investigate the solutions
*75. 0,(0) = 1, 0;(0) = 0, O2(0) = 0, 0;(0) = 0 d2 • tached to the top spring first pass through its equilib-
[ 2m--? + 3.:y - kx = 0 obtained in (a)-(d). How does the forcing func-
rium position? When does the object attached to the
76. 0,(0) = 0,0,(0) = 1, 11,(0) = -1, O2(0) = 0 dl tion f(l) affect the subsequent motion of the ob-
second spring first pass through its equilibrium posi-
77. 0,(0) = 0, 0; (0) = 1, O2 (0) = 0, 0,(0) = 0 If x(O) = 0, x'(O) = I, yeO) = 0, and y'(O) = 0, fmd ject? Does the effect of damping eventually cause
tion? How does the motion differ from that in Prob-
x and y. Compare these results to those found if the object to come to rest?
78. 0,(0) = 0, 0;(0) = 0, O2 (0) = -1, 0,(0) = -I lem 91?
X(O) = 0, x'(O) = 0, yeO) = 0, and y'(O) = 1. Does 87. Consider the L-R-C circuit in which R = 100 n,
*79. 0,(0) = 0, 0;(0) = 0, 8,(0) = -I, 0;(0) = 0 this model resemble one that Was introduced earlier in L = 0.1 henry, C = 10-', and E(t) = 155 sin 3771. *93. Consider the double pendulum system with m, =
the section? (Notice that the frequency of the voltage source is 3 slugs, m2 = I slug, and £1 = £2 = 16 ft. If g =
80. Consider the physical situation of two pendulums cou-
377/2" = 60 Hz = 60 cycles/sec.) 32 ftls2, determine 01(1) and 02(t) if 0,(0) = 1,
pled with a spring, as shown in Figure 8.33. The mo-
0; (0) = 0, O2 (0) = 0, O2(0) = -1. Graph the solution
tion of this pendulum-spring system is approximated (a) Find the current I(t). (b) What is the maximum
parametrically and simultaneously.
by solving the second-order system value of the current and where does it occur? (e) Find
lim,_oo 1(1) (the steady-state current). (d) What is the 94. Suppose that in the double pendulum m1 = 3 slugs,
mxu + mw5x = -k(x - y)
frequency of the steady-state current? m2 = 1 slug, and each pendulum has length 4 ft. If
{ my" + m~y = -key - x)'
88. Suppose that the L-R-C circuit in Exercise 87 has a g = 32 ftls2, determine 01(t) and l1,(t) if 0,(0) = 1,
Figure 8.34 0; (0) = 0, O2(0) = 0, 0,(0) = -1. Graph the solution
ISS cos 3771, 0 $ t < 1
voltage source E(I) = { 0, I 2: 1 . parametrically and simultaneously. How does the
m, ~;~= -k,x + k2 (y - x) (a) Find the current 1(1). (b) What is the maximum length of each pendulum affect the motion as com-
pared to that of Problem 93?
value of the current and where does it occur? (e) Find
84. Write the system d 2y as a
1 m2 dl 2 = -k,y - k2(y - x)
system of four first-order equations with the substitu-

lim,_oo I(t) (the steady-state current). (d) What is the
frequency of the steady-state current?
89. Suppose that the L-R-C circuit in Exercise 87 has
95. Suppose that m 1 = m2 = 1 slug, and each pendulum
has length 4 ft in the double-pendulum system. If
g = 32 ftls2, determine 0,(1) and 82 (t) if 0,(0) = 1,
tions x= xI,xI = X2, Y = Yl,Yr = Y2' . . {ISS, 0 < I < 1 0; (0) = 0, O2 (0) = 0, O2(0) = -1. Graph the solution
the pen odic voltage source E(t) = 0, 1 $ t < 2 ' parametrically and simultaneously. How does the mass
85. Show that the eigenvalues ofthe 4 X 4 coefficient mao
trix .in the system of equations obtained in Exercise 84 E(I + 2) = E(I). (a) Find the current 1(1). (b) What is of the first object affect the motion as compared to
Figure 8.33 are two complex conjugate pairs of the form :!:q,i and the maximum value of the current and where does it that of Problem 94?
---------------------------------------------------------
96. Consider the physical situation shown in Figure 8.35, (c) Use the conditiony(t) = yeO) for -1:5 t:5 0 to _1~Y(::.;,t),,"""",_f-+-+-+ yet)
, t
where the uniform bar has mass mlo a centroidal mo- simplifY the expression in (b). Substitute this re-
ment of inertia 1, and is supported by two springs each sult into the expression in (a) and solve for Yes) -5
with spring constant k. A mass m, is attached at the to find that
center of gravity (point G) of the bar by another spring -10
with spring constant k. Using the coordinates qlo q2, -eS + 2sy(0) - 2eSsy(0) - eSs'y(O)
and q" as shown in Figure 8.35, we frod that the mo- Yes) = -2s2 + eSs' -15
tion is modeled with , t
Solve this system if ml = 2, m, = m, = 1, kl = 112, k2 = I
(d) Use partial fractions to find that (a) (b)
and k, = 2, using the initial conditions XI (0) = x,(O) = x,(O) ,:
dx l dx2 dx, 2(1 + 2sy(O» I + 2sy(0) + s2y(0)
0, dI(O) = 2, dI(O) = 1, and dI(O) = 2. When does each (a) y(l) if yeO) = -I, (b) y(l) if yeO) = 0
Yes) = s'( -2 + eSs) + s'
mass first pass through its equilibrium position?
(e) Find .:e- I { I + 2sy(~; + s2y(0) l 99. The gestation period for whales is about two years.
State the (logistic) differential equation that should be
used to model the situation in which the decline in
s, population is proportional to the amount present and
+ 2sY (O»}
x=o (f) Find .:e- I { 2(I
,
s(-2+es)
S
.. .
by rewntmg tbls ex- the increase is due to the amount present two years
S2 ago.
pression as
y=o 100. In Figure 8.37, tbe grapb of the approximate solution
to the model in Exercise 99 (assuming that}L = v and
S, 2(1 + 2sy(O» 2(1 + 2sy(O» _ _.,..,..-::::-
y(l) = 0.8, -2 :5 1< 0) is shown. How does this re-
s'( -2 + eSs) s'es I - 2s Ie S
z= 0 sult compare to the solution to the logistic equation
= 2(1 + }S;(O»
s e
i: e-,)"
n=O
(2s- 1 without including the two-year delay?
2(1 + i: 2<+:
1.4~
Figure 8.36
= 2sy(O»
n=O S
1.2
Figure 8.35 (a) yet) if yeO) = -I . 98. Consider the problem
1
(b) y(l) ify(O) = 0 let) - 2y(t - I) = I 0.8
0.6
subject to the condition
(a) Determine the motion of the system if k = 2, (g) Show that 02468 1
e
ml = 2, m, = 1, 1 = I, = I and the initial condi- yet) = yeO) for -1:5 t:5 0
tions are ql(O) = 0, q;(O) = 1, q2(0) = 0, q,(O) = 0,
q,(O) = 0, and q,(O) = o. What is the maximum dis-
Laplace transforms can be used to solve equations of
this type.
.:e- 1{2 + 1 sn+4
n e- ns
+ 2n + 2 y(0) e-=}
S,,+3
= Figure 8.37
placement of qlo q2, and q,? (b) If the initial condi- 2n + 1(1 )"+'
(a) Take the Laplace transform of both sides of + 6y(0) + 101. The SIR model, which models the spread of a disease,
tions are ql(O) = 0, q;(O) = 0, q2(0) = 0, q,(O) = 0, - n "!L(t - n)[1
q,(O) = I, and q,(O) = 0, determine the motion of the
the equation to obtain (n + 3)! is discussed in Chapter 6 (Differential Equations at
Work, Project A). 10 this model, S(I) represents the
system. What is the maximum displacement of qlo Q2, (2y(0) - I)n] so that
and q,? (e) How do the changes in the initial condi- sY(s) - yeO) - 2[e-"y(t - I) dt = J,. number in the population susceptible to the disease,
tions affect the maximum displacement of each com-
o s
.:e- 1{2(l + 2SY (0»} = let) those infected with the disease, and R(t) those re-
ponent of the system? (b) Evaluate (e-"y(t-I) dl by letting u=I-I
s'(-2 + eSs) covered from the disease. In the standard SIR model,
97. Consider the system of three springs shown in Figure

8.36, where springs S 10 S2, and S, have spring con-
to obtain I [2n+1 (I - n)"+2 "!L(t -
~
n~O (n + 3)!
n)(t + 6y(0) +
S' = -ASI
J' = ASl- yI,
{
stants k" k2' and k" respectively, and have objects of
mass mlo m" and m" respectively, attached to them.
(2y(0) - I)nl R' = yI
In this case, summing the forces acting on each mass where A and y are positive constants; we assume that
and applying Newton's second law of motion yields (b) Findy(t). (See the following figure.) the members of the recovered class remain in that
_ _...-.I_IIIio......__ _____ _
488 Chapter 8 Introduction to the Laplace Transform Chapter 8 Review Exercises 489
class. However, suppose that the recovered do not re- Laplace transform of a periodic function
R(t) 1Section 8.61
ceive permanent immunity. Instead, after one unit of
time, those who have recovered lose immunity and ;E{f(l)} = -I I
- e
-,T iT e-'Y(I) dl
0
move into the susceptible class. How should you mod- L-R-C circuits in which the voltage is defined by a
ify the SIR model to include this delay? Unit impulse function piecewise defined function
102. A typical solution to the standard SIR model is shown f/..I - 10) = 0, t 1= 10 Delay Differential Equations
in Figure 8.38(a) and that for the SIR model which in-
cludes the delay in Figure 8.38(b). How do these so-
L 8(t- Io)dl=1
+~
Coupled Spring-Mass System: Two Springs

lutions differ? ;e{5(1- lo)} = e-"o d2X
Figure 8.38 (a)-(b) m'-2 = -k,x + k2(y - x) + F,(I)
dl
~ection 8.41 1 d 2y
m2di'£" = -k2(y - x) + F 2(1)
Coupled Spring-Mass System: Three Springs

Convolution theorem
d2X
;e-'{F(s)G(s)} = X-'{X{(f* g)(I)}} m'-2 = -k,x + k 2(y - x) + F,(I)
dl
r
= (j*g)(I)
=
o
1(1 - v)g (v) dv.
1 d 2y
m2 dl 2 = -k,y - k2(y - x) + F 2(1)
The Double Pendulum

Concepts & Formulas
Integral equation
em, + m2)fr8, + m2f,f28, + (m, + m2)f,g8, = °
Integrodifferential equation { m2f~8, + m2f,fo8, + mzf,g8z = °
1 Section 8.1 1
1Section 8.21
Definition of the Laplace transform Lapl~ce transform of the first derivative CHAPTER 8 REVIEW EXERCISES
XU(I)) = fe-'Y(I) dl. XU'(I)) = SXU(I)) - 1(0)
Laplace transform of higher derivatives

Linearity of the Laplace transform Xu(n)(I)) = SnXU(I)} - s"-'/(O) - ... - 1("-')(0)
X{al(l) + bg(l)) = aXU(I)) + bX{g(l)) In Exercises 1-4, find the Laplace transform of each function 15. 5(1 - 37T/2) 16. 7'll.(1 - 7)
Derivatives of the Laplace transform using the definition. 17. 6'll.(I - 7) - 4'll.(1- 4) 18. 36e -4>'ll.(1 - 2)
Laplace transform of sin kt and cos kl
X{I"/(I)) = (-1)" ~: (s) 1./(1)=1-1 19. -42e 5''ll.(1 - 1)
Exponential order
2. 1(1) = Ie -4' 20. 5 sin(1 - 5)'ll.(1 - 5)
°
A functionlis of exponential order (of order b) if there are
numbers b, C > 0, and T> such that
Solving initial-value problems with the Laplace transform
*3·/(1) =g: 051< 5,
12: 5
21. 12ou(t - 2)
1/(1)1 5 Ceb'
4./(1) = {10', 051< I,
22·/(1) °
= {~I, ~ ~ ~ ~ 2,f(1) =1(1 - 2) if 12: 2
forl>T. 1Section 8.31 12: I
Jump discontinuity
Unit step function In Exercises 5-26, find the Laplace transform of each function.
*23. 1(1) °
= {12' -t, 5 II < I
:::; t<2
,f(I) = 1(1 - 2) if I 2: 2
Piecewise continuous
'll.(1 _ a) = {O,1, °5 I< a 5. 15 +5 6. 2 sinh 41
24. f(l) = sin 7TI if °5 1< 1,f(I) = 1(1 - 1) if 12: I
t~a
Shifting property 7. te 2t 8. I' *25 f() = {Sin 7TI, if 05 I < I f(l) =f(1 - 2) if 12: 2
XU(I - a)'ll.(I- a)) = e-"'F(s) . I 0, 151<2 '
X{eaY(I)) = F(s - a) X-'{e-"'F(s)) = 1(1- a)'ll.(I- a) 9. t 3 e t 10. 2e' sin 51
1'051<2
Inverse Laplace transform Periodic function 11. I cos 31 12. I sin 21 26. f(l) = { 2, 2 5 I< 4 ,f(t) = f(1 - 6) ift 2: 6
1(1) = X-'{F(s)) 1(1 + T) = 1(1) 13. e -5, cos 31 14. 5(t - 27T) 6-1,451<6
~---------------------------------------- - - . ..
490 Chapter 8 Introduction to the laplace Transform Chapter 8 Review Exercises 491
In Exercises 27-36, find the inverse Laplace transform of each ries expansion for tan -1(lls). Use this expansion to using the given parameter values and functions. Assume that cigarette and abv !;.tW(VI, I)e -b(x-v,) tlx is deposited
the units of henry, olnn, farad, and volt are used, respectively, between x and x + tlx. Therefore, the total weight de-
function.
show that "e-1{tan-1(lls)} = si~ I. (Hinl: Use the posited by absorption in time I between x and x + tlx
for L, R, C, and E(I).
-2s
-10
27. S2 - 25 28. (s2 + 1)2
Maclaurin series expansion of sin x.)
57. L = 114, R = 1/2, C = 4/9, E(I) = 100,10 = ° is
*29. 3(40 - S6) 30. 2s

2
- 7s + 20 50. Use X{lnJ(I)} = (-1)" :,: F(s) in the form (n '" I) 58. L = 114, R = 3/2, C = 4/9, E(I) = lOO sin 51, 10 = ° abv Fa W(vu, u)e -b(x-vu) du tlx = w(x, I) tlx.
S6 S(S2 - 2s + 10) The total weight is then w(x, I) tlx + W(x, 0) tlx, so
f(l) = _+X-I{:' F(S)} to compute In Exercises 59 and 60, interpret the initial-value problem as a
_S2 - ISs - 52 that
*31.
33.
S(S2 +
7 + 6e 4'
------;F
lOs + 26)
32. -;;
se
34. _3e
6 s
-
(a) X-I {In : ~ ;} and (b) X-1{ln ~~: :}
51. Use the. shifting property to evaluate (a) X{ cosh t};
population problem. In each case, determine if the population
approaches a limit as 1--> "'.
59. x' - 2x = 1000(1 - 1), x(O) = 10,000
W(x, I) = W(x, 0) + abve- bx ro
W(vu, u)e
bvu
du.'
s-6
35 8s 36 -18
(b) X{sinh I}; (e) X{ cos I cosh I}; (d) X{sin I cosh I};
(e) X{cos I sinh I}; (f) X{sin I sinh I}.
, _ {IOOI,
60. x - 2x - 100, I 2: 1
°
s; 1< 1 ) _ 0000
,x(O - 1 ,
(a) Substitute x = vi to show that
. e 3'(s2 + 1) . (I - e 3')(S2 + 1) J(I) = W(VI, O)e

bv
, + abv Fa J(u) du,
52. Compute X{e"'ln} two ways by using: (a) the shifting 61. Suppose that the vertical displacement of a horizontal
In Exercises 37 and 38, compute the convolution (f' g)(I)
(translation) property; (b) the formula for the deriva. beam is modeled by the boundary-value problem whereJ(I) = W(vl,l)e bv,.
tives of the Laplace transform.
using the given pair of functions. d 4y (b) Solve the integral equation in (a) if W(VI, 0) =
dx 4 = W, yeO) = 0, y'(O) = 0, y(l) = 0, y(l) = 0, Woo
37. J(I) = 12, g(l) = e- 3' In Exercises 53 and 54, solve the initial-value problem mX' +
38. J(I) = cos I, g(l) = sin I ex' + kx = J(I), x(O) = a, x'(O) = b to determine the motion of where W is the constant load that is uniformly dis- In Exercises 64-68, solve each system.
an object attached to the end of a spring using the given pa- tributed along the beam. Use the two conditions given
In Exercises 39-48, solve the initial-value problem. rameter values. °
at x = with the method of Laplace transforms to
64.
Y" = -x - 2y, x(O) = 0, x'(O) = 1,
x" = -2x - 4y
39. y" + 6y' + lOy = 0, yeO) = 0, y'(O) = I 53. rn = 4, c = 0, k = 1, a = 0, b = -\.
obtain a solution that involves the arbitrary constants
A = y"(O) and B = y"'(O). Then apply the conditions
{
yeO) = 0, y'(O) = °
40. y" - 4y' + 5y = 0, yeO) = I, y'(O) = ° J(I) = {cos 21, 0 S; I < 7T
at x = I to find A and B. Y' -
3y' = -x' - 2y + x, x(O) = 0, x'(O) = 1,
41. y' + Sty' - lOy = 2, yeO) = 1, y'(O) = ° 0, t ~ 7T

62. Suppose that the load in Problem 61 is not constant.
65. y' + x' = 2y - x
{
= =
42. y' + 12y' + 32y = J(I), yeO) = 0, y'(O) = -I,
54. m = I, c = 112, k = 145116, a = 0, b = 0,
J(I) ~ 0(1 - 7T) " . b ()
Instead, It IS gIVen y ff"x =
{1O,
0,
° s; x < 112
1/2 s; x < 1 .
yeO) 0, y'(O) -I
x" = -3x + y + cos 41, x(O) = 0, x'(O) = 0,
J(I) = if °s; 1< 1
{I,2-1,lfIS;I<2 Solve 66. y" = 2x - 2y
{
In Exercises 55 and 56, solve the L-R-C series circuit modeled yeO) = O,y'(O) = 0
by d 4y
andJ(I) = J(I - 2) if 12: 2 d2Q
L7+ R
dQ I
Tt c
+ Q =E(I)
dx 4 = W(x),y(O) = O,y'(O) = O,y(I) = O,y'(I) = 0
*67.
x'=x-2y
y' =
3y + J(I) ,J(I) =
{I0' 0<1<2
1;;2
+ 6y' + to find the displacement of the beam. {
° '
1
*43. y" 8y = J(I), yeO) = 1, y'(O) = 0, x(O) = O,y(O) =
J(I) = {O,
I,
°
if s; I
if I 2: 1
< I dQ
Q(O) = Qo,I(O) = dr(O) = 10 63. (Filtration Through a Burning Cigarette) Suppose
that a cigarette has original length X and that the X'=-X+-&+J(I) _{-I,OS;I<I
44. x" + 3x' + 2x = 0(1 - 7T) + 0(1 - 27T), x(O) = 0, using the given parameter values and functions. (Assume that tip burns at a rate v so that the cigarette has length 68. y' = _lx + -4 ,J(I) - 1, I s; 1< 2
°
*45. x"
x'(O) =
+ 9x = cos I + 0(1 - 7T), x(O) = 0, x'(O) = °
the units of henry, olnn, farad, and volt are used, respectively,
for L, R, C, and E(I).)
X - vi at time I. Next let W(x, I) represent the weight
per unit length of the cigarette at position x and time
1 x(O)
4
= 0, yeO) =
4
°
0, 12:2
46. g(l) = 5 + 1- r
o
(I - v)g(v) dv
55. L = I, R = 0, C = 10- 4 ,
E(I) = {220,
0, 12:2
°
S; I < 2 Q = 1 =
,0 0
°
I, where x is measured from the burning end. There-
fore, the weight at the burning tip is W(vl, I). Assume
that during steady inhalation, a constant fraction a of
69. Solve the spring-mass system with two springs, as
shown in Figure 8.21, using ml = m2 = kl = 6,
k2 = 4, and the initial conditions x(0) = 0, x'(O) = 2,
any component of the tobacco that is burned at the tip
*47. g(l) = sin I + Fa g(1 - v)e- dv 56. L = 4, R = 80, C = 0.04,
V
yeO) = 0, andy'(O) = 0. (Assume no external forcing
E(I) = {O, ° I 1
S; <
is drawn through the cigarette and that the weight de-
functions.)
-j' y(v) dv = I + e-', yeO) = °
posited per unit length is bwoe -bx (where Wo is the
50, I S; 1<2'
48. dy
dl 0
49. Use the Maclaurin series expansion tan-Ix =

E(I + 2) = E(I), Qo = 10 = ° weight at x = 0). During the time period !;.t, a length
v !;.t is burned so that av tll W( vi, I) passes through the
70. Solve the spring-mass system with two springs, as
shown in Figure 8.21, if the forcing functions FI(I) =
3 S 7
x X x In Exercises 57 and 58, solve the L-R-C series circuit for let)
X - ""3 + '5 + T + ... to find the Maclaurin se-
by interpreting the problem as an integrodifferential equation • Applications in Undergraduate Mathematics in Engineering. Mathematical Association of America, Macmillan. New York (1967), pp. 153-157.
492 Chapter 8 Introduction to the Laplace Transform Differential Equations at Work 493 •
2 cos I and F 2 (1) = 0 are included with the parame- *72. Solve the double pendulum problem with m, ~ 3 where CI is a constant. In an attempt to use a convolution, we multiply by e -sy dy and
ters m, = m2 = I, k, = 3, k2 = 2 and initial condi- m2 = 1, C, = C2 = 16 ft, g = 32 fils2, and the initi~ integrate. Therefore,
tions x(O) = 0, x'(O) = 0, yeO) = 0, and y'(O) = O. conditions 8, (0) = 0, 8; (0) = 0, 82 (0) = -1, 82(0) ~
What physical phenomenon occurs in y (1)7 -\.
71. Solve the spring-mass system with three springs (as *73. Solve the double pendulum problem with m, ~ 3
I
shown in Figure 8.25) if the forcing functions F,(I) = m2 = 1, C, = f2 = 16 ft, g = 32 fils2, and the initi~
2 cos t and F 2 (t) = 0 are included with the parame- conditions 8,(0) = 0, 8; (0) = -1, 82 (0) = 0, 82(0) ~
ters m, = m 2 = 1, k, = I, k2 = I, k, = 1 and initial 1. By the Convolution theorem, this simplifies to
conditionsx(O) = O,x'(O) = O,y(O) = 0, andy'(O) = O.
I
Then, because ::e {t -112} = V:;;;, we have
E. =.£Ls
::e{ </>} .,;--; or
.: ' " Differential Equations at Work: Applying the inverse Laplace transform with ::e- I {s-1I2} = t- 1I2/V:;;:then yields
y
A. The Tautochrone
y From rest, a particle slides down a frictionless curve under the force of gravity as il-
lustrated in Figure 8.39. What must the shape of the curve be for the time of descent
to be independent of the starting position of the particle?
The shape of the curve is found through the use of the Laplace transform. (This
problem was originally solved in 1673 by Christian Huygens, a Dutch mathematician,
Y
-/.._
'V-v:;;:y
CI -1I2_ky-1/2
Recall that </>(y) = dO"ldy represents arc length. Hence

-
1 + (dxldy)2, so substitution into the previous equation gives
)1 + (tY = ky-'/2 or
.
1+ ( -dx)2
dy
=-.
k2
y
</>(y) = dO"ldy =
I
I
as he studied the mathematics associated with pendulum clocks.) Suppose that the par-
ticle starts at height y and that its speed is. v when it is at a height of z. If m is the mass Solving for dxldy then yields dxldy = Y(k2Iy) - 1, which can be integrated with the
of the particle and g is the acceleration of gravity, then the speed can be found by equat- substitution y = e sin2 O.
(0, 0) x ing the kinetic and potential energies of the particle with
I. Find x( 0) using this substitution.
Figure 8.39
"21 mv 2 = mg(y - z), 2. Use the identity sin2 0 = (1 - cos 20)12 to obtain a similar formula for yeO).
I
3. Use a device capable of parametric plotting to graph the curve (x(O),y(O)) on -7T/2
which can be written as ,,; 0 ,,; 0 for k = 1, 2, 3. How does the value of k affect the curve?
Figure 8.40 The principal sys-
tem and the vibration absorber 4. Find the time required for an object located at (x(-7T/2), Y(-7T/2)) to move along
v= v'2gVy="";. attached to the principal system the curve to the point (x(O), yeO)). Does this result depend on the value of O?
To avoid confusion with the s that is usually used in the Laplace transform of func-
tions, let 0" denote the arc length along the curve from its lowest point to the particle.
Therefore, the time required for the descent is
B. Vibration Absorbers
I
Time = L"'Y) -dO" = LY - - dz = L -<p(z)
1 dO" Y
1 dz
o v ovdz oV Vibration absorbers can be used to virtually eliminate vibration in systems in which it
where </>(y) = dO"ldy, which means that
cause the time is constant and v = v'2g
tt) is the value of dO"/dy at y
y - z, we have
= z. Now, be- is particularly undesirable and to reduce excessive amplitudes of vibration in others. A
typical type of vibration absorber consists of a spring-mass system constructed so that
its natural frequency is easily varied. This system is then attached to the principal sys-
y </>(s) tem that is to have its amplitude of vibration reduced, and the frequency of the absorber
Lo _vyr -- -z dz = c" system is then adjusted until the desired result is achieved. (See Figure 8.40.)
I
494 Chapter 8 Introduction to the Laplace Transform Differential Equations at Work 495
If the frequency w of the disturbing force Fo sin wi is near or equal to the natu. where
ral frequency Wn = Ykl/ml of the system, the amplitude of vibration of the system -3
couId become very large due to resonance. If the absorber spring-mass system, made
6
up of components with k2 and m2, is attached to the principal system, the amplitude of
the mass can be reduced to almost zero if the natural frequency of the absorber is ad-
justed until it equals that of the disturbing force
-3
I
II
and
These types of absorbers are designed to have little damping and are "tuned" by vary.
ing k2, m2, or both. This problem is modeled by the system of differential equations
We can then write the system as I
o1 0
x~
O)(X")
:~3 ( -33-3
I
(
o
4
0
0
1 X3
+
I 0-3
6
I
(a) Solve the system for XI and X2'
(b) When is the amplitude of XI equal to zero? What does this represent?
The mass ratio m2/m I is an important parameter in the design of the absorber. To
see the effect on the response of the system, transform these parameters to the nondi-
(a) Solve the system (~ ~ ~)(:~) + n~~3 (-~ ~~ -~)(~)
the initial conditions XI(O) = 0, x;(O) = 1, X2(0) = 0, x;(O) = -~, X3(0) = 0,
= m subject to
I
II
mensional form
X3(0) = 1.*
(b) Determine the period of Xi> X2, and X3 for c = 0.0001, 0.01, 0.1, 1, and 2, where
and
EI
c= m,e3'
(e) If Al is the amplitude of Xi> express 4/(Fo/k,) as a function of W/W22'
(e) Graph Xi> X2, and X3 over several periods for c = 0.0001, 0.01, 0.1, 1, and 2.
(d) Plot the absolute value of AI/(Fo/kl ) for p- = 0.2.
(e) When does AI/(Fo/k l ) become infinite? Do these values correspond to resonance?
(d) Illustrate the motion of the components of the airplane under these conditions.
I
(f) In designing the vibration absorber, what frequencies should the absorber not be
"tuned" in order to avoid resonance?
I
D. Free Vibration of a Three-Story Building
If you have ever gone to the top of a tall building, such as the Sears Tower, World Trade I
C. Airplane Wing
A small airplane is modeled using three lumped masses as shown in Figure 8.41. We
assume in this simplified model of the airplane that the wings are uniform beams of
Center, or Empire State Building on a windy day you may have been acutely aware of
the sway of the building. In fact, all buildings sway, or vibrate, naturally. Usually, we
are only aware of the sway of a building when we are in a very tall building or in a
building during an event such as an earthquake. In some tall buildings, like the John
I
e
I
length and stiffness factor EI where E and I are constants. (E depends on the mate- Hancock Building in Boston, the sway of the building during high winds is reduced by
rial from which the beam is made and I depends on the shape and size of the beam.) installing a tuned mass damper at the top of the building that oscillates at the same fre-
We also assume that m, = m3 and n72 = 4mi> where m, and n73 represent the mass of quency as the building but out of phase. We will investigate the sway of a three-story
each wing and n72 represents the mass of the body of the airplane. building and then try to determine how we would investigate the sway of a building
We find the displacements Xi> X2, and X3 by solving the system The tuned mass damper in the with more stories.
John Hancock Building, Boston
Figure 8.41 A simple model

(COUrtesy of }.IrIS Systems Corp.)
• M. L. James, G. M. Smith, J. C. Wolford. and P. W. Whaley, Vibration of Mechanical and Structural

I
of an airplane Systems. Harper and Row, New York (1989), pp. 346-349.
II
----_.-._-------- "'"
496 Chapter 8 Introduction to the Laplace Transform Differential Equations at Work 497
We make two assumptions to solve this problem. First, we assume that the mass where mj, m2, and m3 represent the mass of the building on the first, second, and third
distribution of the building can be represented by the lumped masses at the different levels, and kj, k2 , and k3 , corresponding to the spring constants, represent the total stiff-
levels. Second, we assume that the girders ofthe structure are infinitely rigid compared ness of the columns supporting a given floor. *
I
with the supporting columns. With these assumptions, we can determine the motion of
the building by interpreting the columns as springs in parallel. I. Show that this system can be written in matrix form as
Assume that the coordinates Xl> X2, and X3 as well as the velocities and acceler_
ations are positive to the right. If we assume that X3 > X2 > Xl> the forces that the
columns exert on the masses are shown in Figure 8.42.
I
In applying Newton's second law of motion, recall that we assumed that accel_
eration is in the positive direction. Therefore, we sum forces in the same direction
as the acceleration positively, and others negatively. With this configuration, Newton's
2
The matrix (kl_:2k2 k 2-: k 3 _Ok3) is called the stiffness matrix of the system.
•
I
second law on each of the three masses yields the following system of differential
equations: o -k3 k3
2. Find a general solution to the system. What can you conclude from your results?
Suppose that m2 = 2ml and m3 = 3ml. Then we can write the system in the
form
I
1 0 O)(x'l) (kl + k2 - k2
(oo 02 03 x~X3 + n!-- -0k2 1
k2 + k3
-k3
which we write as 3. Find exact and numerical solutions to the system subject to the initial conditions
II
Xl (0) = 0, x; (0) = 114, X2(0) = 0, X2(0) = -112, X3(0) = 0, and x,(O) = 1 if
kl = 3, k2 = 2, and k3 = I
for ml = 1, 10, 100, and 1000.
4. Determine the period of x" X2, and X3 for m, = 1, 10, 100, and 1000.
5. Graph Xj, X2, and X3 over several periods for ml = I, 10, 100, and 1000.
6. Illustrate the motion of the building under these conditions.
7. How does the system change if we consider a 5-story building? a 50-story build-
ing? a 1DO-story building?
I
m3 m, I E. Control Systemst
D m2
k,
X2
m2
k 3(x,-x:i)
k,(x,-x:i)
m2
I
Consider the spring-mass system or circuit described by the IVP
d 2x
a dt 2
dx
+ b dt + ex = JU), x(O) = x'(O) = O.
I
D ml
k2
X,
ml
k2
k2(X2 -Xl)
k'(X2 -Xl)
-- ml
Many times the constant coefficients a, b, and c are unknown, and it is the job of the
engineer to determine the response (solution) to the inputJU).
k1Xl
* M. L. James, G. M. Smith, J. C. Wolford, and P. W Whaley, Vibration oj Mechanical and Structural
Systems with Microcomputer ApplicatiOns. Harper & Row, New York (1989), pp. 282-286. Robert K.
Vierck, Vibration Analysis. Second Edition, Harper Collins, New York (1979), pp. 266-290.
t Joseph J. DiStephano, Allen R. Stubberud, and Ivan J. Williams, Feedback and Control Systems,
Figure 8.42 Diagram used to model the sway of a three-story building Schaum's Outline Series, New York (1967).
I
I
9
1. Show that Xes) = F(s)/(as 2 + bs + e). Let pes) = 1/(as 2 + bs + e). This function
is known as the transfer function of the system andp(t) = .;:e-l{p(s)} is called the
weight function of the system.
2. Use the Convolution theorem to show that the solution of Xes) = P(s)F(s) is
x(t) = f' p(v)J(t - v) dv, -called Duhamel's principle for the system.
o
3. Find the response to the spring-mass system
~:; + 2 : + 2x = J(t), x(O) = x'(O) = O.
4. Using the fact that .;:e{8(t)} = 1, show that pes) = .;:e{8(t)}/(as 2 + bs + e). There- Eigenvalue Problems
fore, the solution to
d 2x dx
a"""df2 + b dt + ex = 8(t), x(O) = x' (0) = 0
and Fourier Series
is pet) = .;:e-l{p(S)}, called the uni~ impulse response.
5. Let h(t) be the solution to a d 2x/dt 2 + b dx/dt + ex = OU(t). Show thatH(s) = P(s)/s
so that by the Convolution theorem, h(t) = f'P(V) dv. Therefore, h'(t) = pet).
o
Us~
Duhamel's principle to show that
x(t) = I: h'(v)J(t - v) dv.
6. Suppose that we differentiate the equation L d 2Q/dt 2+ R dQ/dt + ~l/C)Q = E(t)

with respect to t. Use the relationship.dQ/dt = I to obtain L d 2 I/dt + R dI/dt +
I
(lIC)I = E'(t). If we do not know the values of L, R, and C, we may select the n previous chapters we have seen that many physical situations can be
voltage source E(t) = t to determine the response, as in 5. In this case, show that modeled by either ordinary differential equations or systems of ordinary
differential equations. However, to understand the motion of a string at a
I(t) = i\'(V)E'(t - v) dv = fh'(v) dv, so we can use an ammeter to measure the
o 0 particular location and at a particular time, the temperature in a thin wire at a
response, h(t), when E' (t) = 1. Suppose that the readings on the ammeter given as particular location and a particular time, or the electrostatic potential at a point
the ordered pairs (t, h(t» are: (0, 3.), (0.299, 3.07), (0.598, 2.22), (0.898, 0.896),
on a plate, we must solve the partial differential equations listed in Table 9.1 be-
(1.2, -0.337), (1.5, -0.972), (1.8, -0.753), (2.09, 0.232), (2.39, 1.58), (2.69, 2.74),
(2.99,3.16), (3.29, 2.57), (3.59, 1.), (3.89, -1.14), (4.19, -3.23), (4.49, -4.65), cause each of these quantities depends on (at least) two variables.
(4.79, -4.96), (5.09, -4.06), (5.39, -2.23), (5.68, -0.0306), (5.98, 1.89), (6.28,
3.), (6.58, 3.07), (6.88, 2.22), (7.18, 0.896), (7.48, -0.337), (7.78, -0.972), (8.08, TABLE 9.1
-0.753), (8.38, 0.232), (8.68, 1.58), (8.98,2.74), (9.28, 3.16), (9.57, 2.57), (9.87, Wave equation
1.), (10.2, -1.14), (10.5, -3.23), (10.8, -4.65), (11.1, -4.96), (11.4, -4.06), (11.7,
-2.23), (12, -0.0306), (12.3, 1.89). Heat equation
Fit this data with a trigonometric function to determine h(t) to use h' (t) to fmd I(t) Laplace's equation U;u+Uyy=O
if
1 .
E(t) = 10 sm lOt. In Chapter 10, we introduce a particular method for solving these partial dif-
ferential equations (as well as others). To carry out this method, however, we
introduce the necessary tools in this chapter. We begin with a discussion of
boundary-value problems and their solutions.
499
500 Chapter 9 Eigenvalue Problems and Fourier Series 9.1 Boundary-Value Problems, Eigenvalue Problems, Sturm-Liouville Problems 501
Boundary-Value Problems, Eigenvalue Problems, I Solution Using the general solution obtained in the previous example, we have
Sturm-liouville Problems y(x) = Cl cos X + C2 sin x. As before, y'(O) = C2 = 0, so y(x) = Cl cos x. However,
"*
because y' (7C) = - Cl sin 7T = 0 1, the boundary conditions cannot be satisfied
r Boundary-Value Problems r Eigenvalue Problems with any choice of Cl' Therefore, there is no solution to the boundary-value
r Sturm-Liouville Problems problem.
L
Boundary-Value Problems
If theboundary conditions in Example 2 are y'(O) = 0 and y(7C) = 0, then how many
Boundary-value problems are solved in much the same way as initial-value problems solutions does the problem have?
except that the value of the function and its derivatives are given at two values of th~
As indicated in the general form of a boundary-value problem, the boundary
independent variable instead of one. A general form of a second-order (two-point)
conditions in these problems can involve the function and its derivative. However, this
boundary-value problem is
modification to the problem does not affect the method of solution, as shown in
a2(x)y" + al(x)y' + ao(x)y = f(x), a <X< b Example 3.
kly(a) + k2y'(a) = a, hly(b) + h2y'(b) = {3,
where kb k2, a, hb h 2, and {3 are constants and at least one of kb k2 and at least one
of hb h2 is not zero. Note that if a = /3 = 0, then we say the problem has homoge-
"Ei,,',',.
y"-y=O O<x<l
neous boundary conditions. We also consider boundary-value problems that include + 3y(0) = 0, y'(I)
Solve { y'(O) + y(l) = 1 .
a parameter in the differential equation. We solve these problems, called eigenvalue
problems, to investigate several useful properties associated with their solutions.
: I Solution The characteristic equation is r2 - I = 0, with roots r = :!:: 1. Hence
a general solution is y(x) = cle r + C2e-r, with derivative y'(x) = cle x - C2e-r.
'if 1,,','.' Applying y' (0) + 3y(0) = 0 yields y' (0) + 3y(0) = Cl - C2 + 3(Cl + C2) =
4Cl + 2C2 = O. Next
Y" + Y = 0, 0 < x < 7C
Solve { y'(O) = 0, Y'(7C) = 0 .' y'(l) + y(l) = cle l - C2e-1 + Ctel + C2e-l = 2c 1e = 1.
so Cl = lI(2e) and C2 = -lie. Thus the boundary-value problem has the unique
: J Solution Because the characteristic equation is r2 + 1 = 0 with roots r = :!:. i, a solution
general solution isy(x) = Cl cos x + C2 sin x, wherey'(x) = -Cl sin x + C2 cosx.
Applying the boundary conditions, we have y'(O) = C2 = O. Then y(x) = Cl cos x.
With this solution, we have Y'(7C) = -Cl sin 7C = 0 for any value of Ct. Therefore,
there are infinitely many solutions, y(x) = Cl cos x, of the boundary-value problem,
depending on the choice of Cl' When a second-order homogeneous dijferential equation with constant coefficients is
accompanied by a pair of homogeneous initial conditions (y(xo) = y'(xo) = 0), then the
initial-value problem has the unique solution y = O. Is the same true of a boundary-
From the result of Example I, we notice a difference between initial-value prob- value problem with homogeneous boundary conditions? (Hint: See Example 1.)
lems and boundary-value problems. This difference is that an initial-value problem (that
meets the hypotheses of the Existence and Uniqueness theorem) has a unique solution, Eigenvalue Problems
whereas a boundary-value problem may have more than one solution (or may have no
We now consider eigenvalue problems, boundary-value problems that include a pa-
solution).
rameter in the differential equation. Values of the parameter for which the boundary-
value problem has a nontrivial solution are called eigenvalues ofthe problem. For each
"E'"I"N, eigenvalue, the nontrivial solution that satisfies the problem is called the correspon-
ding eigenfunction. (Notice that if a value of the parameter leads to the trivial solu-
y" + Y = 0, 0 < x < 7C tion, then the value is not considered an eigenvalue of the problem.) We show how the
Solve { y'(O) = 0, Y'(7C) = I . eigenvalues and eigenfunctions are found in the following example.
502 Chapter 9 Eigenvalue Problems and Fourier Series 9.1 Boundary.Value Problems, Eigenvalue Problems, Sturm-Liouville Problems 503
"f'"',,,*, A = An = k/ = (n;t n = 1,2, ...

Solve y" + Ay = 0, 0 < x < p, subject to yeO) = 0 and yep) = o.
Notice that the subscript n is used to indicate that the parameter depends on the
value of n. (Notice also that we omit n = 0, because the value k = 0 was consid-
i 'I Solution We recall that the solution of this equation is determined by solving the
ered in Case 1.) For each eigenvalue, the corresponding eigenfunction is obtained
characteristic equation r2 + A = O. Of course, the values of r depend on the value
of the parameter A. Hence we consider the three cases below. by substitution into y(x) = C2 sin kx. Because C2 is arbitrary, let C2 = l. Therefore,
the eigenvalue An = (n7rlp)2, n = 1, 2, ... has the corresponding eigenfunction
Sin( n;). n=
Case I: A= 0
y(x) = Yn(x) = 1, 2, ...
In this case, the characteristic equation is r2 = 0, so the repeated roots are
r] = r2 = O. This indicates that a general solution is Notice that we did not consider negative values of n. This is due to the fact that
sine -n7TXlp) = -sin(n7TXlp). Because the negative sign can be taken account for in
y(x) = c]x + C2'
the constant, we do not obtain additional eigenvalues or eigenfunctions by using
Applying the boundary conditiony(O) = 0 yieldsy(O) = c](O) + C2 = 0, so C2 = O. n = -1, -2, ....
Then yep) = CIP = 0, so C] = O. Therefore, y(x) = 0 in this case. Because we ob.
tain the trivial solution, A = 0 is not' an eigenvalue.
To better understand the results of the previous example, consider the boundary-value
Case II: A< 0 problem
To represent A as a negative value, we let A = _k2 < O. Then the characteristic
equation is r2 - e
= 0, so r] = k and r2 = -k. A general solution is, therefore,
kx
y(x) = c]e + C2e-kx,
Using the eigenvalues and eigenfunctions obtained in Example 4, we know that this
or y(x) = c] cosh kx + C2 sinh kx. Substituting the boundary condition yeO) = 0
problem has the solution y(x) = sin(57TXlp), whereas problems like
yields yeO) = c] + C2 = 0, so C2 = -Ct. Because yep) = 0 indicates that yep) =
c]ekp + C2e-kp = 0, substitution gives us the equation yep) = c]ekp - c]e-kp =
c](ekp - e-kp) = O. Notice that ekp - e-kp = 0 only if ekp = e-kp, which can only
57r)2
y"- (P y=O Y" = 0
and
occur when k = 0 or p = O. If k = 0, then A = -k2 = -0 2 = 0, which contradicts [yeO) = 0, yep) = 0 {yeO) = 0, yep) =0
*
the assumption that A < O. We also assumed that p > 0, so ekp - e -kp O. Hence
have no nontrivial solution because A = - (5 7rlp? < 0 and A = 0, respectively, in the
yep) = c](ekp - e-kp) = 0 implies that c] = 0, so C2 = -c] = 0 as well. Because
A < 0 leads to the trivial solution, y(x) = 0, there are no negative eigenValues. differential equation, and the boundary conditions are the same as those considered in
Example 4.
Case III: A> 0 We will find the eigenValues and eigenfunctions in the previous example quite
useful in future sections. The following eigenvalue problem will be useful as well.
To represent A as a positive value, we let A = e
> O. Then we have the character·
istic equation r2 + e
= 0 with complex conjugate roots r].2 = ±k:i. Thus a general
solution is
y(x) = c] cos kx + C2 sin kx.

"F'"I,'.,
Solve y" + Ay = 0, 0 < x < p, subject to y' (0) = 0 and y' (p) = o.
Because yeO) = Ct cos k(0) + C2 sin k(0) = c], the boundary condition yeO) = 0
indicates that c] = O. Hence y(x) = C2 sin kx. Application of yep) = 0 yields i I Solution Notice that the only difference between this problem and that in the
yep) = C2 sin kp = 0, so either C2 = 0 or sin kp = O. Selecting C2 = 0 leads to the previous example appears in the boundary conditions. Again, the characteristic equa-
trivial solution, which we want to avoid, so we determine the values of k that sat· tion is r2 + A = 0, so we must consider the three cases A = 0, A < 0, and A > O.
isfy sin kp = O. Because sin n7r = 0 for integer values of n, sin kp = 0 if kp = n7T, Note that a general solution in each case is the same as that obtained in the previ-
n = 1, 2, .... Solving for k, we have k = kn = n7rlp, so the eigenvalues are ous example. However, the final results may differ due to the boundary conditions.
Case I: A=O Case I: A= 0
In this case y(x) = ClX + C2, so y'(X) = Cl. Therefore, y'(O) = Cl = 0, so The repeated roots of the characteristic equation are r = (-2 ± Vo)/2 = -I,
y(x) = C2. Notice that this constant function satisfies y'(p) = 0 for all values of c so a genera! solution is
If we let C2 = I, then A = 0 is an eigenvalue with corresponding eigenfunctio~ y(x) = cle- x + C2xe-x.
y(x) = Yo(x) = I.
Substitution of yeO) = 0 yields yeO) = Cl = O. Hence y(x) = C2xe -x. Applying
Case II: A< 0 y(2) = 0 implies that y(2) = 2c2e -2 = 0, so C2 = o. Therefore, the assumption that
A = 0 leads to the trivial solution, and thus A = 0 is not an eigenvalue.
If A = -e
< 0, then y(x) = cle ia + C2e-ia and y'(x) = clke ia - c2ke-Ia
Theny'(O) = clk - C2k = 0, so Cl = C2. Therefore, y'(p) = clkekp - clke-kp = o· Case II: A>O
which is not possible unless Cl = 0 because k,* 0 andp'*O. Thus Cl = C2 = 0, s~
Let A = k 2 > o. Then
y(x) = 0, and there are no negative eignvalues.
r=
-2 + \I4k2 = -2±2k = -I +k
Case III: A> 0 2 2 - ,
By letting e
,1.= > 0, y(x) = Cl cos kx + C2 sin kx and y'(x) = so a genera! solution is
-clksinkx + c2kcos kx. Hence,.Y'(O) = C2k = 0, so C2 = O. Consequently, y(x) = cleC-l+k)X + C2eC-l-k)x.
y'(p) = -clk sin kp = 0, which is satisfied if kp = n1T, n = 1,2, .... Therefore
the eigenvalues are ' Theny(O) = Cl + C2 = 0, so C2 = -Cl. Also,
= cle 2C - l + k) + c2e2C-l-k) = cle- 2(e 2k - e- 2k ) = 0, = C2 = O.

(~1Tr n = 1,2, ....
y(2) so Cl
A = An =
(Why?) Because we obtain the trivial solution, there are no positive eigenValues.
Note that we found C2 = 0 in y(x) = Cl cos kx + C2 sin kx, so the corresponding
Case III: A< 0
eigenfunctions are
Let A = -k2 < O. Hence
y(x) = Yn(x) = cos n7TX, n = 1, 2, ... ,
.p.
r=
-2 ±-v=4k2 -2±2ki=_I+k·
2 - ~,
if we let Cl = I. 2
Summarizing our results, y" + Ay = 0, 0 < x < p, subject to y' (0) = 0, and and a general solution is
y'(p) = 0 has the following eigenvalues and corresponding eigenfunctions: y(x) = cle -x cos kx + C2e -x sin kx.
0, n = 0 x
Then yeO) = Cl = 0, so y(x) = cze- sin kx. Applying y(2) = 0, we have y(2) =
An =
[ (p-).
n1T 2
n = 1,2, ...
and Yn(x) = [1 :;'0 n = 1,2, ... ·
C2e -2 sin 2k = O. Therefore, to avoid the trivial solution, we must require that
cos
p 2k = n1T, n = I, 2, .... Hence k = k n = n1T/2, n = 1,2, ... , and the eigenvalues
are
A = An = _k2 = - ( 2n1T)2 ' n = I, 2, ...
By letting Cz = I, the eigenfunctions are

Solve y" + 2y' - (A - I)y = 0 subject to yeO) = 0 andy(2) = o.
-x . n7TX
y ( x ) = Yn(x) = e sm -2-·
~I Solution In this case, the characteristic equation is r2 + 2r - (A - I) = 0, with
roots
r= -2±v'4+4(A-I) = -2±"V'4A Sturm-Liouville Problems

2 2' Because of the importance of eigenvalue problems, we express these problems in the
so again we must consider the three cases A = 0, ,1.< 0, and A > O. general form
az(x)y"(x) + al(X)Y'(X) + [aO(x) + A]y(x) = 0, a < x < b,

Theorem 9.1 Linear Independence and Orthogonality of Eigenfunctions
"*
where az(x) 0 on [a, b], and the boundary conditions at the endpoints x = a and
If Ym(x) and Yn(x) are eigenfunctions of the regular Sturm-Liouville problem,
x = b can be written as
"*
kly(a) + kzy'(a) = 0 and hty(b) + hzy'(b) = 0 nality condition r
where m n, then Ym(x) and Yn(x) are linearly independent and the orthogo-
a
s(x)Ym(x)Yn(x) dx = 0 (m "* n) holds.
for the constants kj, kz, hj, and hz, where at least one of kj, kz and one of hj, hz is not
zero. This equation can be rewritten by letting
Notice that the limits of integration on the orthogonality condition are obtained
fa,(x) dx ao(x) p(x) from the interval, a < x < b, on which the eigenvalue problem is defined.
p(x) = e a,(x) , q(x) = - (
az x
) p(x), and sex) = az(x)' Because we integrate the product ofthe eigenfunctions with the function sex) in
the orthogonality condition, we call sex) the weighting function.
By making this change, we obtain the equivalent equation
lif'",,'8:1
! (p(X) ~) :t (q(x) + As(x»y = 0,
Consider the eigenvalue problem y" + Ay = 0, subject to yeO) = 0 and yep) = 0
which we solved in Example 4. Verify that the eigenfunctionsYI(x) = sin(=lp) and
which is called a Sturm-Liouville equation, which along with boundary conditions
Y2(X) = sin(2m:lp) are linearly independent. Also, verify the orthogonality condition.
is called a Sturm-Liouville problem. This particular form of the equation is known
as the self-adjoint form and is of particular interest because of the relationship of the
function sex) and the solutions of the problem. :] Solution We can verifythatYt(x) = sin(=lp) andyz(x) = sin(2=lp) are linearly
. independent by finding the Wronskian.
sm-. = . 2=
sm--
p p
"E''''''O. -'1Tc o s
P
=-
P
2'1T
-cos--
p
2=
p
Place the equation xZy" + 2xy' + Ay = 0 in self-adjoint form.
= -
2'1T sin -= cos -
2=- - -'1T cos -=. 2=
sm - -.
p p p p p p
:JI Solution In this case, az(x) = xZ, aleX) = lx, and ao(x) = O. Hence
Notice that if x = p12,
ao(x) 2'1T '1T '1T '1T. 2'1T
p(x) =
a, (x)
e f a,(x) dx =e J2xJx
,
dx = e Zlnx = xZ, x > 0, q(x) = -C) p(x) = 0, and
az x
W(YI , yz) = - p sin -2 cos '1T - -p cos -2 sm '1T = - P
- "* O.
Because W(Yl> yz) is not identically zero, the two functions are linearly independ-
ent. (We could have also determined that the two are linearly independent, because
so the self-adjoint form of the equation is they are not scalar multiples of one another.) The equatio~ y" + ~~ = is in ?
self-adjoint form with sex) = I. (Why?) Hence the orthogonahty conditlOn IS
dx
Y
dx )
d ( x zd + Ay = O. ro
Ym(x)Yn(x) dx = O. We verify this with YI(X) and yzCx):
We see that our result is correct by differentiating. fof' YI(X)Y2(X) dx = fP0 sin -= 2=
sin - - dx = fP2 sm
' Z -= =
cos - .1••
u.<
P PoP P
Solutions of Sturm-Liouville problems have several interesting properties, two

= 2p [sin3
3'1T
=]P
P 0
= O.
of which are included in Theorem 9.1.
"\
-
EXERCISES 9.1 (e) y" + 2y' - (A - I)y = 0, yeO) = 0, y(2) = 0, In Exercises 37-38, solve the eigenvalue problem with non-
constant coefficients.
y(x) = Yn(X) = e- x sin n;", n = 1,2, ...
37. x 2y" - xy' + Ay = 0, y(I) = 0, y(e) = 0
35. Show that the eigenvalues and eigenfunctions of 2
38. x 2y" + xy' + Ay = 0, y(l) = 0, y(e ) = 0
In Exercises 1-14, solve the given boundary-value problem. In fy"+Ay=O, -p<x<p
28. y" + y' + Ay = 0, yeO) = 0, y(l) = 0
each case, state the number of solutions. lY(-p) = yep), y'(-p) = y'(p) are
29. y" - 4y' + 2Ay = 0, yeO) = 0, y(l) = 0 In Exercises 39-42, solve the nonhomogeneous boundary-value
1. y" = O,y(O) = 0,y(2) = 0 AO = 0, Yo(x) = I problem if a solution exists.
30. Orthogonality conditions can be established using
2. y" = 0, yeO) = 0, y'(I) = -2 other methods than those discussed previously. Use { An = ( n7T)2 Yn(x)
p , = an cos pn 7 +TbnXsm
. n7TX.
p 39. y" = I, yeO) = 0, y(2) = 0
*3. y" = 0, y'(O) = 0, y'(I) = I Legendre's equation in self-adjoint form 40. y" - y = x,y(O) = l,y(l) = 0
4. y" = 0, y'(O) = 0, y'(I) = 0
(The eigenfunctions of this eigenvalue problem will
d 2 *41. y" + Y = sin 2x, yeO) = 0, y(7T) = 0
5. y" - 4y = 0, yeO) = 0, y(l) = e
dx [(1 - x )y'] + n(n + I)y = 0 be very useful later in this chapter and in Chapter 10.)
36. Consider the general form of a homogeneous bound- 42. y" + Y = cos x, yeO) = 0, y(7T) = 0
6. y" + y' - 2y = 0, yeO) = 0, y(2) = 0 and the fact that two Legendre polynomials must sat. ary-value problem 43. Consider the Sturm-Liouville (eigenvalue) problem
isfy the relationships
*7. y" + 9y = 0, y'(0) = 0, Y(7T) = 0
8. y" + 4y = O,y(O) = O,y(7T) = 0
a,(x)y"(x) + al(x)y'(x) + ao(x)y(x) = 0, a < x < b, v+ Ay= 0
. :!x [(I - X2)p~(x)] + n(n + I)Pn(x) = 0 and kly(a) + kzy'(a) = 0 and hly(b) + h,y'(b) = O. )YeO) = 0, y(l) = O'
9. y" +Y = 0, yeO) = 0, y'(f) = 0
It can be shown that if YI(X) and y,(x) are linearly in-
Use an integration device to verify the orthogonality
d condition for any two eigenfunctions YI(x) = sin m7TX
+ m(m + l)Pm(x) = dependent solutions of a2(x)y"(x) + al(x)y'(x) +
10. y" + Y = 0, yeO) = 2, y(7T) = -2
dx [(I - X2)p;"(x)] 0
ao(x)y(x) = 0 such that
and Y2(X) = sin n7TX, m n. *
*11. y" + 4.,1'y = 0, yeO) + y'(O) = 0, y(l) = 27T
12. y" + 36y = O,y(O) = O,y(7T) - Y'(7T) = 0
to establish the orthogonality condition for the Le-
gendre polynomials. [Hint: MUltiply the first rela- klyl(a) + k2y~(a) kly,(a) + k2y~(a) I* 0, 44. In Section 4.8, we discussed Legendre's equation
dldx [(I - x 2 )dyldx] + n(n + I)y = 0 on the interval
13. y" + y = 0, yeO) - y'(O) = 0, y(7T) + y'(7T) = 0 tionship by P m(x) and the second by Pn(x), subtract IhlYI(b) + hZYI(b) hlY,(b) + hzYz(b) [-I, I], which has as solutions the Legendre poly-
the expressions that result, and then integrate over the then the only solution to the boundary-value problem nomials Po(x) = I, P,(x) = x, Pzcx) = 1(3x 2 - I),
14. y" - 9y = 0, yeO) = 0, y(l) + y'(I) = I
interval (-I, I)]. is y(x) = O. If P 3 (x) = j(Sx 3 - 3x), P 4 (x) = !(3sx4 - 30x 2 + 3),
Determine the eigenvalues and eigenfunctions of the following

klYI(a) + k,y',(a) k,y,(a) + k,Y2(a) = 0 I Ps(x) = h63x' - 70x 3 + ISx), . . . . Verify the or-
Let (J, g) deI10te the inner product of I(x) and g(x) with re- Ih,y,(b) + h 2 Yr(b) h,yz(b) + h 2y,(b) , thogonality condition for the set of eigenfunctions. (In
eigenvalue problems. Exercise 30, we proved this condition.)
spect to the weight function s(x) on the interval (a, b), in other then nontrivial solutions exist. Use this theorem with
15. y" + Ay = O,y'(O) = O,y'(1) = 0 the indicated solutions to determine if the following {Po(x), PI(X), P 2(x), P 3 (x), P 4 (x), P,(x)}.
words, (J, g) = fS(X)/(x)g(x) dx. In Exercises 31-33, verify
boundary-value problems have only the trivial solu-
16. y" + Ay = 0, y'(O) = 0, y(1) =0 the indicated property of the inner prodnct. *45. Consider the eigenvalue problemy" + Ay = 0, yeO) =
tion.
*17. y" + Ay = 0, yeO) = 0, y'(1) = 0 0, y(1) + y'(I) = O.
31. (J, g) = (g,J) (a) y" + Y = 0, yeO) = 0, y(7T) = 0, y,(x) = sin x,
18. y" + y' + Ay = 0, yeO) = 0, y(l) = 0 (a) Show that A = 0 is not an eigenValue.
yz(x) = cos x
19. y" + y' + Ay = O,y(O) = O,y'(I) = 0 32. (J, g + h) = (J, g) + (J, h) (b) Show that there are no negative eigenvalues.
33. (cf, g) = c(J, g), where c is any real constant. (b) y" + y = 0, yeO) = 0, y(f) = 0, YI(X) = sin x,
20. 2y" + 2y' + Ay = 0, yeO) = 0, y(l) = 0 (e) Show that the positive eigenValues A = k 2 satisfy
*21. 2y" + 2y' + Ay = 0, yeO) = 0, y'(I) = 0 34. The norm of the eigenfunction Yn(x), with respect to Y2(X) = cos x the relationship k = -tan k.
22. y" + 2y' + (I - A)y = O,y(O) = 0,y(2) = 0 the weighting function sex), is I Yn(x)11 = V (Yn,Yn) = (e) y" + 2y' + (.,1' + I)y = 0, yeO) = 0, yeS) = 0, (d) Approximate the first three eigenvalues. Notice
YI(X) = e- x sin 7TX, yz(x) = e- x cos 7TX that for larger values of k, the eigenvalues are ap-
23. y"- 4y' + 2Ay = 0, yeO) = 0, y(l) = 0
fS(X)[Yn(X)]Z dx. Calculate the norm of each of the (d) x 2y" + xy' + Y = O,y(l) = O,y(e = O,YI(X) = proximately the vertical asymptotes of y = tan k,
IT
)
24. y" - 4y' + 2Ay = 0, yeO) = 0, y'(I) = 0
sin(In x), y,(x) = cos(ln x) so An = [(2n - 1)7T/21'.
following eigenfunctions.
In Exercises 2S-29, verify the orthogonality condition that must (a) y" + Ay = 0, yeO) = 0, yep) = 0, Yn(X) =
be satisfied by the eigenfunctions of the given exercise. . n7TX
smp,n= 1,2, ...
25. y" + Ay = 0, y'(O) = 0, y'(1) = 0
(b) y" + Ay = 0, y'(O) = 0, y'(p) = 0, Yo(X) = I,
26. y" + Ay = 0, y'(O) = 0, y(l) = 0
Yn(x) = cos n7TX, n = 1,2, ...
*27. y" + Ay = 0, yeO) = 0, y'(1) = 0 p
510 Chapter 9 Eigenvalue Problems and Fourier Series 9.2 Fourier Sine Series and Cosine Series 511
§I Fourier Sine Series and Cosine Series which was described by Euler in a 1748 paper. Independent of Euler, Daniel Bernoulli
was able to show that for t = 0, the displacement of a vibrating string of length C can
One of the uses of the Taylor and Maclaurin series discussed in calculus is to approx. be described by
imate functions by taking a finite number of terms from the series to obtain Taylor and
Maclaurin polynomials. For example, the Maclaurin series for f(x) = cos x is
_ 1 2 1 4 I 6 _ ~ (-1)" 2n
f(x) - 1 - 2ix + 4T x - 6T x + ... - n~O (2n)! x , Subsequently, Euler asked if any functionf(x) can be written in the form
so the first four Maclaurin polynomials for f(x) = cos x are Po(x) = 1, P2(X) '=
2
I - t,x , P4(X) = I - t,x 2 + ;/,X4, and P6(X) = 1 - t,x 2 + ;/;x4 - t,x 6. In Figure 9.1,
we graphf(x) = cos x along with the approximating polynomial. Notice that as we in.
f(x) = t An sin ( n;x),
where the coefficients, Am are to be determined. * In fact, using the orthogonality re-
crease the number of terms in the polynomial, the graph of the polynomial becomes
lations developed in Section 9.1, we wi\l see that we can compute the coefficients, Am
closer to that of f(x) = cos x. Therefore, we improve the accuracy of the approxima.
in the same manner as they were computed by Clairaut, Euler, and, subsequently, Fourier.
tion by increasing the number of terms.
Taylor and Maclaurin series are called power series because they involve powers
In Chapter 10, we will see that the displacement of a vibrating string of length
C at time t can be described by of one variable, x. However, we can extend the idea of series by considering other func-
tions in our series, such as the sine functions in the series developed by Euler. In gen-
eral, suppose that
f(x) = I
n=O
CnYn(X),
where {y n(x)} is a set of eigenfunctions of a particular eigenvalue problem and the cn's
y are constants that must be determined. As with power series, we can (and Will) use
these eigenfunction expansions (or series) to approximate functions.
y" + Ay = 0 . .
Recall the eigenvalue problem { yeO) = 0, yep) = 0' which was solved m
-2 Example 4 in Section 9.1. The eigenvalues of this problem are A = An = (n7T/p)2, n =
-3 1,2, ... , with corresponding eigenfunctions Yn(x) = sin(nm:/p), n = 1,2, .... There-
-1 -4 fore, for some functions f we can find coefficients b m so that
(a) (b) ~ . nm:

f(x) = L bn sm-.
n~l P
y
A series of this form is called a Fourier sine series. To make use of these series, we
must determine the coefficients b n • We accomplish this by taking advantage of the or-
thogonality properties of eigenfunctions. Because the differential equation y" + Ay =
x
o is in self-adjoint form, we have that sex) = 1. (See Section 9.1.) Therefore, the or-
thogonality condition is
(c)
-1
(d)
-1
i" . P
o
. mm: dx
n7TX sm--
sm--
P
= 0 ( m -""
r n) .
'" Jesper Lutzen, ''The Solution of Partial Differential Equations," Studies ~n the Hist,0rJ! of Mathem.atics,
ed. Esther R. Phillips, MAA Studies in Mathematics. Volume 26, Mathemattcal AsSOCiatIOn of Amenca,
(1987). pp. 242-277.
512 Chapter 9 Eigenvalue Problems and Fourier Series 513
9.2 Fourier Sine Series and Cosine Series
To use this condition, mUltiply both sides of f(x) = Ibn sin(wrcx/p) by the eigen_
n=1
= 1.[_
71"
x cos nx]"" + 1.
n 071"0
cos nx
n 1"" ax = _ 2 cos n71"
n
+ 1.
71"
[Sin 2nx
n
r
0
function sin(mm:/p) and sex) = 1. Then integrate each side of the equation from
x = 0 to x = p (because the boundary conditions of the corresponding eigenvalue prob- = _ 2 cos n71" + ~(sin n71" - sin 0)
lem are given at these two values of x). This yields n 7m
p = _2 cos n71" +~(O)= _2cosn71".

. m7IX ax = ~ b sm
IP L . nm: . -
mm:
- ax.
Iof(x) sm -p- 0 n~1
n - - sm
P P
n 7m n
Notice that cos n71" = (-1)", because cos 71" = -1, cos 271" = 1, {;OS 371" = -1, ....
Assuming that term-by-term integration is allowed on the right-hand side of the equa-
tion, we have Hence
2(-1)" 2(-lr 1
(f(X) sin mm:
o p
ax = i (b sin~
n~1 0
n
P
sin mm:
P
ax = i bn(Sin~
n~1 0 P
sin m7IX dx.
p
bn = - - - n - = n
and the Fourier sine series is

Recall that the eigenfunctions ynCx) = sin(nm:/p), n = 1, 2, ... are orthogonal, so
I: sin(nm:/p) sin(mm:/p) ax = 0, if m.* n. On the other hand, if m = n, f(x) = I
~
"=0
b n sm -
. n7IX
7T
= I
n=l
~ 2(-1)"+1
n sin nx
F . nm: Sln--
Io sm--
p
. mm: ax = IF sm
Pop
. 2-
nm: P
- ax = I -1 ( 1 - cos--
02
2nm:)
p
ax = 2 sin x - sin 2x + t sin 3x - +sin 4x + ...
= 1. [x - ~ sin 2n7IXr = f!.... Notice that we can use a finite number of terms of the series to obtain a trigono-
metric polynomial. In Figure 9.2, we graphf(x) = x on 0 ,,;:: x ,,;:: 71" along With one
2 27m P 0 2
Therefore, each term in the sum
y
~ (P. n11X . m7TX
L bnj, sm - - sm - - dx
n~1
equals zero except when m = n. Hence

0 P P 2.5
2
1.5
/'
/7~.
2.5
2
1.5
/
/
/--::",
\
p 1 I
pnm: = p
2' / "-\ 0.5
Io f(x) sin bn • 0.5 l
2 2.5 3 x
0.5 1 1.5
so the Fourier sine series coefficients are given by
Ca) Cb)
P
bn = -2I f(x) sm
• nm:
- - ax, n = 1, 2, ... , y y
Pop
3 /"'>
where we assume thatfis integrable on (O,p). 2.5
2 t 2.5
2
o 'ifi,,"'.'
(
1.5 / 1.5
~/ 1 /'7
1
0.5 0.5 ~
~
0.5 1.5 2 2.5 3 x 0.5 1
Find the Fourier sine series for f(x) = x, 0 < x < 71". 1
Ce) Cd)
: • Solution In this case, p = 71", so
Figure 9_2 (a) Pl(X) = 2 sin x, (b) P2(X) = 2 sin x - sin 2x, (e) p,(x) = 2 sin x -
bn
2
= -7To 1"" nm:
f(x) sin - -
7T
ax = - 1""2
"lTo
x sin nx ax (integration by parts) sin 2x
2
+"3 sin 3x, (d) P4(X) = 2 sin x - sin 2x
2 .
+ "3 I. 4
8m 3x - 2" 8m x
j
"
~'
514 Chapter 9 Eigenvalue Problems and Fourier Series 515
y y 2 4 2_~
3.5 b6 -- b2(3) = -(4) = -317' b10 = b2(5) = -10
17 (4) - 5 17 , ... ,
3
2.5
/x
• I
\
3.5
3
2.5 /f:
/ ' ' ' \•
617
2
.- "" 2 r
b2 (2n-l) = (2n -4 1)7T ,n = 1,2, ... ,
1.5 1.5 /'/
1
.- 1 ,// so we have the series
0.5 0.5
0.5 1 1.5 2 2.5 3 x
",'"
0.5 1 1.5 2 2.5 3 x

II
lex) = ~1
~
(2n - 1)17 sm
4 . 2(2n - 1)= =
2
4f
n~1 (2n - 1)17
sin(2n - 1)=
(a) (b)
Figure 9.3 (a) 10 terms, (b) 20 terms =.i. sin = + ~ sin 3= + -54 sin 57T.X + ....
7T 37T 17
In Figure 9.4, we graph/ex) with severa! polynomials. Notice that wi~ a larg~ num-
of the polynonlials obtained from this series. As we increase the number of tenus ber of terms the approximation is quite good at values of x where / IS contmuous.
used in approximating f, we improve the accuracy. Because many more terms are
needed to yield a desirable approximation, we graph/ex) = x with the polynomials
y y
using 10 and 20 terms, respectively, ill Figure 9.3. Notice from the graphs that none
of the polynonlials attain the value ofJ( 17) = 17 at x = 17. This is due to the fact that \\
at x = 17, each of the polynomials yields a value of O. Hence our approximation can 0.5
only be reliable on the interval 0 S; x < 17. In general, however, we are only assured
of accuracy (at points of continuity off) on the open interval 0 < x < 17.
1\ 1.5 ;2 x
-0.5
0.5 tl 1.5 ~ x
-I
\ ,...,. /
v v
(a) (b)
"£'",,,- y y
1 /'0- ~0v'\
Fm d the Founer . ~or /()
" sme senes x = {I,
-1, 0 S; x<
I s; x s;I 2' 0.5 . \
0+---0'.5---;---I'.5--~2~x
:, Solution Because / is defined on 0 s; x s; 2, P = 2. Hence
-0.5 \ I
-I \V " - ~ '0 \.1
2
bn = 2: L/(X) sin MTX dx = Ll sin n= dx + (2(_1) sin ~ dx
(e) (d)
2 0 2 0 2 2 )1
2c o sn=]1
= [- -nl7 - - + [2
2 0
- c o sn=]2
nl7
--
2 1
y
. u \
y
0.5
\
= (- n:)(Cos n; -I) + (n:)(Cos nl7- cos n;) o+---~--+---~--~ o -1----+--+---l-~T-:x
0.5 1.5 ~ x I O5 1 1.5
-0.5 \ I -0.5 . \
2 -1 '-" ~ "'J -1 v" ~
= (n 17)( -2 cos n 17
2
+ cos nl7 + I).
?
(e) (I)
Calculating a few of the b;s, we find b 1 = 1. . 0 = 0, b2 = .l . 4 =.i., b3 = 4 sm
. m, (b) pCx) = -:;
4 sm
. m 4.
17 17 7T Figure 9.4 (a) pCx) = -:; + 37T sm 3 m, ()
C p
(x)=
2 2
3 17 . 0 = 0, b4 = 417 . 0 = 0, .... As we can see, most ofthe coefficients are zero. ..±.'Ti' sin m + ....±.. sin 3m + -54 sin 5m, (d) p(x) = .;. sin m + ... + 7~ sin 7m,
3'Ti' 7T
In fact, only those bn's where n is an odd multiple of 2 yield a nonzero value. For
example, (e) pCx) =.;. 1TX + ... + 1;'Ti'sin 151TX, (f) pCx) = ~ sin m + ... + 1:'Ti' sin 19m
516 Chapter 9 Eigenvalue Problems and Fourier Series
9.2 Fourier Sine Series and Cosine Series 517
The behavior of the series near points of discontinuity in that the approximation
overshoots the function is called the Gibbs phenomenon. The approximation Con_ ao = lIP
P 0
f(x) dx.
tinues to "miss" the function even though more and more terms from the series are
used. If m > 0, then we note that by the orthogonality property,
In Example 2, what is the value of each Fourier polynomial at x = I?

ro
an cos(n7TXlp) cos(m7TXlp) dx = 0 if m "* n. We also note that
Another important eigenvalue problem that has useful eigenfunctions is

o
1 m7TX
J(P '2ao cos p dx = 0 and
11' 0
2 n7TX P
cos - p dx = -2'
y" + Ay = 0
Hence
{ y'(O) = 0, y'(p) = O' 1' n7TX P
As we found in Example 5 in Section 9.1, the eigenvalues and eigenfunctions are 1f(x)
o
cos -
p
dx = 0 + an . 2'
a, n= 0 I, n= 0 Solving for am we have
A =
n { (p),
n7r 2
n = 1,2, ... and Yn(x) = !!!!!.
{ cos p ,
_
n - 1, 2, ...
.
a = -211' f(x) cos -
n poP
n7TX
- dx, n = 1, 2, ....
Therefore, for many functions f(x), we'can find a series expansion of the form
Notice that this formula also works for n = 0 because cos(O . 7TXla) = cos 0 = 1.
1
f(x) = -2 ao +~
/ an n7TX
cos - - .
n~l P
We call this expression a Fourier cosine series, where in the first term (associated with
Ao = 0), the constant iao is written in this form for convenience in finding the fonnula
,if·""".1
Find the Fourier cosine series for f(x) = x, 0 <x< 7r.
for the coefficients am n = 0, I, 2, .... We find these coefficients in a manner simi-
lar to that followed to find the coefficients in the Fourier sine series. Notice that in this I Solution In this case, p = 7r. Hence
case, the orthogonality condition is
n TTX m ir.x
2 (rr 2 [x2]rr
I'P
o
cos - - cos - - dx = 0 (m
p p
"* n). ao = -; J x dx = -;
o
'2 = 0 7r
and
We use this condition by multiplying both sides of the series expansion by cos mm/p
and integrating from x = 0 to x = p. This yields 2I rr n= 2 (rr x cos nx dx
a
n
= -7To x cos - - dx
'iT
= -J,
1To
(integration by parts)
L
J: -r
1' m7TX 11' I m7TX 11' n7TX m7TX
+
J:
00
1 f(x) cos - - dx = -2 ao cos - - dx an cos - - cos - - dx.

o Pop 0 n~l P P = ~W si: nx sinn nx dx} = ~[CO~2nx
Assuming that term-by-term integration is allowed,
= - 4 (cos n7r - 1) = -4[(-1)" - 1].
1' m7TX
1 f(x) cos - - dx =
11' 1 m7TX
-2 ao cos - - dx +L 00 11' n7TX m7TX
an cos - - cos - - dx. 7I7l 7I7l
o Pop n~l 0 P P
If m = 0, then this expression reduces to Notice that for even values of n, ( - 1)n - I - 0 . Therefore , an = 0 if n is even. On
the other hand, if n is odd, (-1)" - 1 = -2. Hence
1' 11' 1 ~ 11' an cos -
n7TX 4
1o f(x) dx = 0 2 ao dx + n=l
L
0 P
- dx,
where r
o
cos(n7TXlp) dx = 0 for n;:: 1 and
12 1' 1 1
o ao dx = 2 pao· Therefore, so the Fourier cosine series off(;<) = x on 0 < x < 7r is
I: f(x) dx = + pao, so f(x) = !!. - -

2
4
L00
7r n~l
1
(2n - 1)
2 cos(2n - l)x.
518 Charter 9 Eigenvalue Problems and Fourier Series 519
{~:
In Figure 9.5·, we plot the function with several teans of the series. Compare these O<X<7T
16. f(x) = y(x, t) = LAn sin(n=I€)cos(mrtl€). If the initial
results to those obtained when approximating this function with a sine series. Wbi h 7T<X < 27T n=O
series yields the better approximation with the fewer number of teans?
y y
C
*17·f(;c)= { I:
X 0<x<2
2<x<4
shape of the string is y(x, 0) = f(;C), then show that
An = 2/€ r
o
f(x) sin(n7lX/€) dx.
x,O<X<1
/ 18. f(x) = { 2 _ x, 1 < x < 2
/ 30. (a) Use the formula derived in Exercise 29 to find the
2.5
2
1.5 / ,/,/,/
2.5
2
1.5
/./
/'
19. f(x) = e- X , 0 <x< 1
displacement of the string iff(x) = x(€ - x), 0 < x <
€. (b) What is the displacement iff(x) = ~ sin 2x, 0 <
1
0.5 / ."
x
1
0.5 ./
/ 20. f(x) = e- + eX, 0 < x < 1
X
In Exercises 21-23, use trigonometric identities to obtain the

x < €. (Hint: Integration is not necessary.)
31. Find the Fourier cosine series and the Fourier sine se-
0.5 1 1.5 2 2.5 3 0.5 1 1.5 2 2.5 3 x ries for the following functions. In each case, graph
Fourier cosine series for each of the following functions. the function as well as approximations using a finite
(a) (b)
21. f(x) = cos 2 x number of terms from the series. Which series con-
y Y verges to the function more rapidly (that is, requires
22. f(;c) = sin2 2x the fewer number of terms to approxinTate the func-
2.5 2.5 *23. f(x) = cos' x (Hint: Expand cos x = j(e Lx + e -Lx) and tion)?
2 2 simplify with Euler's formula.)
1.5 (a) f(;c) = x', 0 <x< I
1.5
I I (b) f(x) = x 2 (1 - x), 0 <x< 1
In Exercises 24-25, use trigonometric identities to obtain the
0.5 0.5
x
Fourier sine series for each of the following functions. (c) f(;c) = X4, 0 < X < 1
0.5 I 1.5 2 2.5 3 0.5 I 1.5 2 2.5 3 x
24. f(x) = 4 sin x cos x (d)f(;c) = x 2 (1 - x 2 ), 0 <x < 1
(c) (d)
25. f(x) = sin' x (Hint: See Exercise 23 with sin x = 32. In Example 2, we found the Fourier sine series for
. 7T 4 7T 4 4 (1I2i)(e~ - e- iX ).)
FIgure 9.5 (a) p(x) ~ 2 - -:;;: cos x, (b) p(x) ~ 2 - -:;;: cos x - 97T cos 3x, (c) p(x) = I 0 :S x < I
p f(x) = {..: 1, I :S x < 2 to be
n7TX m7TX
7T 4 4, 4 7T 4 4
2 - -:;;: cos x - 97T cos ox - 257T cos 5x, (d) p(x) ~ 2 - -:;;: cos x - ... - 497T cos 7x 26. Prove that (a)
Lo an cos -p- cos -p- dx = 0 if
00 4
m * n, (b) Lo p
2 n= p
cos - - dx = -2'
f(x) = L
n~l
(2n _ 1)7T sin(2n - 1)=
p
27. (Boundary-Value Problems) Fourier sine series can 4.
= -:; 8m TTX
4sm
+ 37T '3 +4
1TX
' 57TX
517 8m + ....
EXERCISES 9.2 be used to solve boundary·value problems of the form
ay" + by' + cy = f(x), 0 < x < p,y(O) = O,y(p) = 0 Substitution of x = 1/2 into both sides of this equa-
by assuming that the solution is y(x) = tion yields
Determine the Fourier cosine series for the following functions

on the indicated interval. 8. f(x) = {x, 0 < x < I
Lb
n=l
n sin(n=lp). (Notice that each term in the
(1) =-:;;:51n
f2'
4 . 7T 4 . 37T 4 . 57T
2 + 37T S'"T+ 57T S'"T+'"
series satisfies the boundary conditions.) Then, be-
2 -x, 1 <x< 2
i. f(x) = x, 0 <x< I x
cause the particular solution must also satisfy the or I='!_"±"+"±"+'"
9. f(;c) = e- , 0<x< 1 boundary conditions, we must find the Fourier sine se·
7T 37T 57T
z. f(x) = x + I, 0 <x < I + eX, 0 < x <
*3. f(x) = x 2 • 0 < X< I
10. f(x) = e -x 1 ries forf(x) on (O,p). Both series are then substituted
into the differential equation so that the coefficients = ~( l - t + t -.. -)
4. f(x) = x{l - x), 0 <x< I In Exercises 11-20, determine the Fourier sine series. Cn can be found. Use this technique to solve y" + Y =
Therefore, 1 - 1/3 + 115 - ... = 7T/4. How many
x, yeO) = 0, y( 7T) = O.
5. f(x) = I, 0 < x < 7T 11. f(x) = x, 0 < x < 1 terms of this alternating series are needed to approx-
28. If the boundary conditions in Exercise 27 are y' (0) =
6. f(x) = {a,1, 0< x < 7T
7T<x<27T
12. f(x) = x
2
+ 1,0 < x < I
O,y'(p) = 0, then what type of series should be used?
imate 7T/4 to within an error of 1O- 5 ? What is this ap-
proximation? (Hint: Recall alternating series from cal·
*13. f(x) = x < X< I
, 0 Solve y" + Y = x, y'(0) = 0, Y'(7T) = O. culus.)
*7. f(x) = {xl', 0<x <2 14. f(x) = x(1 - x), 0<x < I 29. Euler described the displacement of a string of length
2<x<4 15. f(x) = 1, 0 < x < 7T We discuss convergence of series more in Section 10.2.
€ at position x and time t with the series solution
520 Chapter 9 Eigenvalue Problems and Fourier Series 9.3 Fourier Series 521
~ Fourier Series F m7TX fP 1 m7TX

f-p I(x) cos - - dx =
p -P
zao cos - - dx
P
(" Fourier Series (" Even, Odd, and Periodic Extensions
~ ( an cos -
Fourier Series +
f p
-P n~l
L...,
n7TX m7TX
- cos -
p
. -
- + b n sm n7TX m7TX)
- cos - - dx
p p p
Consider the eigenvalue problem that was given in Exercise 35 in Section 9.1:
Y" + Ay = 0, -p < x < P
= fP
-P
.!.ao cos!!!:!E. dx +
2 a n~l
f (fP
-P
an cos n7TX cos m7TX dx
P P
{ y(-p) = yep), y'( -p) = y'(p)"
This problem has eigenvalues

+ f p
-P
n7TX
an sm - . m7TX)
- cos -
P
- dx .
P
0, n= ° If m = 0, then using the integrals in Table 9.1 we notice that

An =
[ (p),
n7r 2
n = 1,2, ...
f F
-p
I(x) cos °. p
7TX dx = r -P
I(x) dx,
and eigenfunctions
and all of the integrals that we are summing have the value zero. Thus this expression
simplifies to
Yn(x) = [1,an cos--+
n::' . n7TX
bnsln--, n=1,2, ... ' F fP 1
p p f_p I(x) dx = _P zao dx
so we can consider a series made up of these functions. Hence we write p 1
f (an f
-P
I(x) dx = Zao[2p]
I(x) = tao +
n~l
cos n7TX + bn sin n7TX),
P P
which is called a Fourier series. As was the case with Fourier sine and Fourier cosine
series, we must determine the coeffiCients ao,' an (n = 1,2, ... ), and bn (n = 1,2, ... ).
aD =.!.
p
r
-P
I(x) dx.
Otherwise, if m '1= 0, then only one of the integrals on the right-hand side of the
Because we use a method similar to those used to find the coefficients in Section 9.2,
we give the value of several integrals in Table 9.2. Note that most of these integrals
were encountered in Section 9.2. r
expression yields a value other than zero, and this occurs with
-P
cos(m7TXlp) cos(n7TX/p) dx = p if m = n. Hence
TABLE 9.2 p n7TX
rr
p
n1TX
r m1TX. nm
f
-P
I~) cos - - dx = p . an
p
-p
cos--dx = D
p -p
sin nmc dx
P
= D
f p
-p
cos - - sm --dx = D
p P
an = -
1 fP I~)
n7TX_
cos - - dx, n - 1, 2, ...
p -P P
f mmc nmc
- dx = {D, m *" n {D, *" n
F
-p
cos -
p
- cos -
p p, m = n f F
-p
• mmc . nmc
SID--sm--
p p
dx =
p,
m
m= n
We fmd b n (n = 1, 2, ... ) by mUltiplying the series by sin(m7TX/p) and integrating
from x = -p to x = p. This yields
m7TX
f
OP •
I(x) sm - - dx =
We begin by finding aD and an (n = 1, 2, ... ). Multiplying both sides of -P P
I(x) = tao + f (an

n~l
cos n7TX + bn sin n7TX)
P P
F -ao
f-P 1
2
. m7TX
sm--dx+
P
~ (ancos--sm--+bnsm
fF L...,
-P n~l
n7TX. m7TX
p p
. n7TX sm
p
. !!!:!E.)
p
dx
by cos(m 7TX/p) and integrating from x = -p to x =p (because of the boundary condi- m7TX
f I(x)
p .
tions) yields sm - - dx =
-p p
522 Chapter 9 Eigenvalue Problems and Fourier Series 9.3 Fourier Series 5::p
~ (f'P 1 fO 1 f2
fOP
-p
1 . mnx
"2ao sm - - dx + L,
p n~' -p
nnx . m77X
an cos - - sm - - dx
P P
1 J2
an ="2 -2 f(x) COS
nnx
-2- dx ="2 _/1) cos -2-
nnx nnx
dx +"2 (2) cos -2- dx
0
+ fp
-p
• nnx . mnx )
an sm - - sm - - dx ,
P P = Hn 2
17 sin n;x t t· +
2
2 [n 17 sin n;x ]: = o.
where we assume that tennwise integration is pennitted. Again, with the integrals in 1 f2 . n77X I fO . nnx 1 L2 . nnx
Table 9.2, we note that only one of the integrals on the right-hand side is not zero. In bn ="2 -2 f(x) sm -2- dx ="2 _2(1) sm -2- dx +"2 0 (2) sm -2- dx
this case, we use
_1.. [~ cos _l. 2 [~ cos
f
p
-p
mnx
• -
sm . -
- sm
p
- dx = ~'
nnx
p ,
*
m n
m= n
=
2 n17
n77X]O
2 -2 2 n17
nnx]2
2 0
to obtain
nnx
f
p
f(x) sin - dx = P . bn
-p p = ...L(1 - cos n17) =...L(l - (-1)").
n7f' n'7T
bn = -I f'P "f()
- X sm n77X
. - - dx, n = 1,2, .... Therefore,
P -p P
3 ~ 1 . nnx
f(x)=-+ / (l-(-I)")-sm-
2 ,;7, n17 2
Definition 9.1 Fourier Series
= 1. + 1. sin 77X + ~ sin 3nx + ~ sin 5nx + ...
Suppose thatfis defined on -p < x < p. Then the Fourier series forfis 2 17 2 317 2 517 2
f(x) = tao i +
n~'
(an 'cos nnx + bn sin nnx),
P P
We graph f(x) with several tenns of the series in Figure 9.6. If we extend f
over more periods, then the approximation by the Fourier series carries over to these
intervals. I
where
ao = I
-
fP f(x) dx
p -p y y
2.5 2.5
1 fOP nnx
an = - f(x) cos - - dx, n = 1, 2, ... 2
p -p P 1.5 , 1.5
~-"-~? -'c-~
bn = 1.. fOP f() . nnx dx, n -- 1, 2, ...
X sm
p -p p 0.5 0.5
-2 -1 o 2 x -2 -1 o 2 x
o I'F'd,,'a,
Find the Fourier series off(x) = {I, -2:s: x< 0
o :s: x:S: 2 ' wheref(x + 4) = f(x).
(a)
2.5
y
(b)
2.5
y
F'19ure.
96 ()
2, 2 r-~~---.-.-c 2
a p ()
="23 + n7r
x
2 sm
. T
= + n7r
2 sm
. -2-
3=
1.5 1.5
:"'1. Solution In this case, p = 2. Hence ~"-~--'-,o4
-'--_____ 4 (b)p(x) = 1. + ~sin = + ... + ~ sin 7=
2 n7r 2 n7r 2
IJ2 If IL2
0
1 1
ao ="2 _J(x) dx ="2 _2(1) dx +"2 0 (2) dx = "2[xl~2 +"2' 2[x16
0.5 0.5
(e) p(x) = 1.2 + ~
n7r
sin =2 + ... + ~
n7r
sin 11=
2
-2 -1 o 2 x -2 -1 o 2 x
(d)p(x) = 1. + ~sin = + ... + ~ sin 15=
= 1 +2= 3, (e) (d) 2 n7r 2 n7r 2
In Example 1, what is the value of each Fourier polynomial p at x = O? How does if n is even. Putting this infonnation together, we can write the coefficients as
this value relate to the value off to the left and to the right of x = O?
2
a2n = (2n _ 1)(1 + 2n)71" n = 1,2, ...
"f'"I,'., Similarly,
· d the F
Fm ' senes
ouner . 0 . -1"; x:::;:<X
x ={O,
ff()
Sill 11X,
° ° $
l' wheref(x + 2) =f(x) . 1 JI f
bn = -1 (x)'sm nm dx = II.
-I
. nm dx ,
sm m sm
0
:J Solution In this case, p = 1, so so ifn = 1, then
ao =..L
1
fl f(x)
-1
dx = fO
-1
(0) dx + II sin m dx = _[cos m]1
0 7T 0
b l = II sin2 m dx = II..L(1 - cos 2m) dx =
o 0 2
..L(x
2
_ sin271'2m)1 = ..L.2
0
1 2
= --:;;:(cos 71' - cos 0) =-:;;:. If n * 1, then we use sin A sin B = 4[cos(A - B) - cos(A + B)]. Hence
II [cos(I - n)m - cos(1 + n)m]
an = + flex) cos nm dx = (1(0) cos nm dx + f sin m cos nm dx
1
bn = '2 0
_ ..L[sin(1 - n)m _ sin(l + n)m]1 _ _

dx
= f o
sin m cos nm dx.
- 2 (I - n)71' (1 + n)71' 0 - 0, n - 2, 3, ...
Therefore, we write the Fourier series as

The value of this integral depends on the value of n. If n = 1, then we have
*
I
al = I0 sin m cos m dx = '21 II0 sin 2m dx = -'21 [cos
If n 1, then we use the identity sin A cos B = Hsin(A - B) + sin(A + B)] to
2m]1
~ 0 = 0. I
~
~
j
1 1. 2 ~ 1
f(x) = -:;;: + '2 sm m - -:;;: ~I (2n _ 1)(1 + 2n) cos 2nm.
We graph f along with several approximations using this series in Figure 9.7.
obtain '
an = +f [sin(1 - n)m + sin(1 + n)m] dx i

a y y
= _..L[COS(1 - n)m + cos(1 + n)m]1

2 (1 - n)71' (1 + n)71' 0 0.8 0.8
= -'2
1 {[ cos(1 - n)71' cos(l + n)71']
(1 - n)71' + (1 + n)71' -
[II]}
(1 - n)71' + (1 + n)71' .
0.6
0.4
0.6
0.4
0.2
Notice that ifn is odd, then both (I - n) and (1 + n) are even. Hence, cos(1 - n)m - '--../
-1 -0.5 0 0.5 1 x -1- -0.5 0.5
= cos(1 + n)m = I, so
(a) (b)
an = -+{[ (1 _1 n)71' + (1 +1 n)71' ] - [(1 ~ n)71' + (1 +1 n)71' ]} = ° y Y
if n is odd. On the other hand, if n is even, (1 - n) and (1 + n) are odd. Therefore,

cos(l - n)m = cos(1 + n)m = -1, so
0.8
Figure 9.7 (a) p(x) = ~+ t 2
sin m: - 3'T/' cos 2m:
an = -'2 1{[ -1 -1] [ I I ] }

(1 - n)71' + (1 + n)71' - (1 - n)71' + (1 + n)71'
0.4
0.2
(b) p(x) = 1. + 1. sin m: -
7T 2
... - _2_ cos 4m:
157T
(e) p(x) = 1. + 1. sin m: - ... - _2_ cos 6m:
7T 2 357T
= __1_ + __1 _ = -0:-_-,-2;-,........,_ 2 -1 -0.5 0 0.5 1 x -1 -0.5 0 0.5
(1 - n)71' (1 + n)71' (1 - n)(l + n)71' (n - 1)(1 + n)'T/' (c)
(d) p(x) = 1. + 1. sin m: - ... - _2_ cos 8=
(d) 7T 2 637T
9.3 Fourier Series 527
Theorem 9.2 can be used to determine the sum of many series that cannot be
Theorem 9.2, proved by German mathematician Gustav Peter Lejeune Dirichl
found with the techniques introduced in calculus.
(1805-1859) in an 1829 paper, tells us that the Fourier series for any function c et
verges to the function except at points of discontinuity. on-
Theorem 9.2 Convergence of Fourier Series

"ii"I,'.'
Suppose that f and f' are piecewise continuous functions on - p < x < p. Then
Use the Fourier series for f(x) = U: -2
o <:~ Xx <:$ 02' f(x + 4) = f(x) to find the
the ~ourier series for f on -p < x < p converges to f(;c) at every x where f is value of
continuous. At x = Xo, where f is discontinuous, the Fourier series converges to I I I (_1)"+1 ~ (-1)"+1
the average 1--+---+'" + - - - + ... =
3 5 7 2n - I
L, -2n-- -
.=1 I .
f(xo +) + f(;co -)
2 : I Solution Recall that we found the Fourier series to be
= limx_xo+ f(x) = limx _ xo - f(x).
wheref(xo +) andf(xo -)
f(x) = 1. +
2
I (I -(-Ir).l... sin~
.=1 nTT 2
A proof of this theorem can be found in more advanced textbooks such as Werner Ro- = 1. + 1. sin = +.2.. sin 3= + .2.. sin 5= + ....
gosinski, Fourier Series, Second Edition, Chelsea, New York (1959), or Ruel Churchill 2 TT 2 3TT 2 5TT 2
and James Brown, Fourier Series and Boundary Vtllue Problems, Third Edition,
Notice thatfis continuous at x = I, so the Fourier series for f converges to f(1) at
McGraw-Hill, New York (1978).
x = 1. Hence
3 2. TT(1) 2 . 3TT(I) 2 . 5TT(I)
f(l) ="2 + -; sm-- + 3TT sm-2- + 5TT sm-2- + ...
2
Iff()
x = {I,
2, 0 :$ xx <
-2:$ 0 .
:$ 2 andf(x + 4)'= f(4), then determine the numerical value 2 =1.+1._.2..+.2..+ ...
2 TT 3TT 5TT
to which the Fourier series for f converges at each of the following values.
(a) x = -I; (b) x = I; (e) x = 0; (d) x = 3; (e) x = 2. t=~(I-t+t+ ... )
y :J Solution We begin by sketching the graph of two periods off in Figure 9.8. TT I I
2.5 (a) f is continuous at x = -I, so the Fourier series converges to f( -I) = I at x = "4=1-3"+"5+···.
-I ; (b) Similarly, because f is continuous at x = 1, the Fourier series converges to
Therefore, the sum of the series is TT14. Notice that other values of x (such as any
1.5 f(1) = 2 at x = I; (c)fis discontinuous at x = 0, so the Fourier series converges to
integer multiple of 1/2) where fis continuous can be used to obtain the same result.
the average
0.5
-2 0 4 6 x Even, Odd, and Periodic Extensions

(d) fis continuous at x = 3, so the Fourier series converges to f(3) = I at x = 3; So far we have assumed that the functionfwas defined on the interval -p < x < p.
Figure 9.8 (e) Becausefis discontinuous at x = 2, the Fourier series converges to Unfortunately, this is not always the case. Sometimes we must take a function that is
x=p x
f(2 +) + f(T) = I +2 = ~ defined on the interval 0 < x < p and represent it in terms of trigonometric functions.
2 2 2' Three ways of accomplishing this task are to extendfto obtain (a) an even function
on -p <x<p; (b) an odd function on -p <x<p; and (c) aperiodic function on
We can test these results by observing the sequence of graphs that were generated
_p < x < p. Using the function shown in Figure 9.9, we illustrate these three situa-
in Example 1.
Figure 9.9 tions in Figure 9.10.
y y
The odd extensionfodd off is an odd function, so
ao = 1. IF /odd(X) ax = 0,
p -P
x
X=p
an = -1
p
IF
-P
fodd(X) cos - -
nme
P
ax = °(n = 1, 2, ... ), and
bn = -
1 IF /odix) sin -
nme 2 (P
ax = -P J, f(x) sin -nme ax (n = 1, 2, ... ).
Ca) Cb)
p -P P op
Ce)
The periodic extensionJ;, has period p. Because half of the period is p/2,
Figure 9.10 (a) Even extension, (b) odd extension, (e) periodic extension
ao = 1. (P f(x)
p Jo
ax,
We notice some interesting properties associated with the Fourier series in each an = -2 iF 2nme
f(x) cos - - ax (n = 1,2, ... ), and
of these three cases by noting the properties of even and odd functions. Iff is an even PoP
function and g is an odd function, then the product fg is an odd function. Similarly, if
fis an even function and g is an even function, thenfg is an even function, and iffis
2
bn = -
P
Pop
i • 2nme
f{x) sm - - ax (n = 1,2, ... ).
an odd function and g is an odd functiiih, then fg is an even function. The proofs of
these three properties (which are summarized in Table 9.3) are left as exercises.
TABLE 9.3
o "fi"i,'.j°
Let f(x) = x on < x < 1. Find the Fourier series for (a) the even extension off;
f g fg
(b) the odd extension off; (c) the periodic extension off
Even Odd Odd
y : J Solution (a) The graph of the even extension offis shown in Figure 9.11. Here
Even Even Even
p = 1, so
Odd Odd Even
(1 [X2]1
ao = 2 Jo x ax = 2 2"" 0 = 1,
Recall from integral calculus that iffis odd on -p < x <p, then r f{x) dx =
=2
l
io x ax = -,---,(cos
2
0, and if g is even on -p < x <p, then r -P

g(x) ax = 2 (g(x)
0
are useful in determining the coefficients in the Fourier series for the even, odd, and
-P
ax. These properties

-1 -0.5 0.5 1 x
an
= +,[(-1)" -
cos nme
1] (n
n 7T
n7T -
= 1,2, ... ), and

1)
n7T
periodic extensions of a function, because cos(n=/p) and sin(nm:lp) are even and odd
periodic functions, respectively, on -p < x < p.
Figure 9.11
bn = ° (/I = 1, 2, ... ).
Because an =
as
°if n is even, we can represent the coefficients with odd subscripts
Definition 9.2 Even, Odd, and Periodic Extensions
4
The even extensionfeven offis an even function. Therefore,
ao = -1
p -P
P
If<:ven(x) ax= -2 J(P f(x)
P 0
ax, Therefore, the Fourier cosine series is
I IF n= nme
= -2 J,(Pf(x) cos -
ax ax 1 4
I (2n -
00
an = - f<:ven(x) cos - - - (n = 1, 2, ... ), and feven(x) = - - )2 2 cos(2n - 1)=.
p P PoP
I n~1
-P 2 1 7T
bn = -1
P
p -P
• nme
P
°
f<:ven(x) sm - - ax = (n = 1, 2, ... ). We graph the even extension with several terms of the Fourier cosine series in Fig-
ure 9.12.
,,J
y y y (e) The periodic extension (shown in Figure 9.15) has periodp = 1. Thus
ao = 2 fx dx = 2[x22]~ = 1,
cos 2n7T + 2n7T sin 2n7T
I
2 dx
an = 2 L X cos
0 n'TTX = 21 n2,,?
= 0 (n = 1,2, ... ),
-I -0.5 0 0.5 -1 -0.5 0 0.5 1 x and

Ca) Cb)
I 1 -sin 2n7T + 2n1T cos 2WTr 2n7T
1 4 1 4 4
Figure 9.15 bn =2 L
0 X sin 2n'TTX dx = -2 n2,,? =- 2n2,,?
Figure 9.12 (a) p(x) = "2 - 71 cos mc, (b) p(x) = "2 - 71 cos mc - 971 cos 3mc.
1
= -- (n = 1,2, ... ).
n7T
Hence the Fourier series for the periodic extension is
y (b) The graph of the odd extension is shown in Figure 9.l3. For the odd extension 00
1
Jodd, we note that ao = 0, an = 0 (n ;;; 1, 2, ... ), and j, (x) = - -
p 2
I -1
n=l n7T
sin 2n'TTX.
I WTrx 2 2(-1)"+1
0.5
bn =2 L 0 X sin -a- dx = - n7T cos n7T = n7T (n = 1, 2, ... ). We graph the periodic extension with several terms of the Fourier series in Figure
9.16.
-1 -0.5 0.5 x Hence the Fourier sine series is
-0.5 2(-1)"+1
I.
00
y
iodix) = n sin mrx.
-1 n=1 7T
Figure 9.13
We graph the odd extension along with .several terms of the Fourier sine series in
Figure 9.14. ~
'0.6
\
\
: .' 04 I
i . . \
. 0.2.
y y \/
-1 -0.5 0 0.5 1 x -1 -0.5 0 0.5 1 x
Ca) Cb)
0.5
Figure 9.16 (a) PV:) = 1.2 - 1.
1T
sin 2mc - ... - .J...
51T
sin 10mc
0.5 -1 -0.5
(b) p(x) = 1.2 - 1.
1T
sin 2mc - ... - _1_ sin 20mc
101T
-0.5
-1
The even extension oj the function in Example 5 (a) is the same as what periodic
(a) Cb) Junction on -1 <x < 1? The odd extension oJthefunction in Example 5(b) is the
same as what periodic function on -1 < x < 1? Is the series in Example 5 (c) a
F'19ure.1
9 4 (3
) p () 2. 1.
X = -:;;: sm mc - -:;;: sm 2=
2 sm
+ ... + 51T . 5mc Fourier sine series, a Fourier cosine series, or neither?
(b) p(x) = 1.
1T
sin mc - 1. sin 2 mc + . . . + .J... sin 10mc
1T 51T
EXERCISES 9.3
- In Exercises 34-37, find the value to which the Fourier series

converges at the indicated values of x. In each case, sketch the
periodic ftmction.
I, -I :5x<O
34. f(;<) = { -2, 0:5 x:5 1,J(x + 2) = f(x). (a) x = 0,
In Exercises 46-49, (a) sketch the periodic extension of the
function from the given exercise; (b) rmd the Fourier series of
the periodic extension.
46. f(x) = x
2
, ° :5 x < I
In Exercises 1-10, determine the Fourier series for the indicated 24. Suppose that f is an even ftmction and g is an odd
periodic function. In each case, sketch the periodic ftmction. ftmction. Prove that fg is an odd ftmction.
a, 0:5 x < I
(b) x = -I, (e) x = 1, (d) x = 112, (e) x = -112. 47. f()
x = {2, 1:5 x < 2
1. f(x) = x, -I :5 x < I,J(x + 2) = f(x) 25. Verify the integrals in Table 9.2. -x -7T<X<O
35. f(x) = { 7T,' 0 :5 x ~ 7T ,J(x + 2TT) = f(x). (a) x = 48. f(x) = I - x, ° :5 x < 7T
2. f(x) = x 2 , -7T:5 X < 7T,J(X + 27T) = f(x)
*3. f(x) = x', -7T:5 x < 7T,J(X + 27T) = f(x)
26. Show that the Fourier series coefficients of the con.
stant function f(x) = k, where k is a constant, are 0, (b) x = -7T,(e)x = 7T,(d)x= -7T/4,(e)x = 57T/4.
*49. f(x) = X(7T - x), ° :5 x < 7T
aD = 2k and an = bn = 0, n = 1, 2, . . . . (Hint:
50. Show that iff and g are periodic ftmctions of period
= x(I - x), -1 :5 x < I,J(x + 2) = f(x)
1~', :~'.'.:~!§
4. f(x) T, then the sum (f + g) is a periodic ftmction of
Represent f as the periodic ftmction f(x) = Ie,
-2, -2:5 x < 0 -p :5 x < p,J(x + 2p) = f(x).) period T.
5. f(x) = {0, 0:5 X < 2 ,J(x + 4) = f(x)
27. Show that the Fourier series coefficients of the sum 36. f(x) = I" +>,) - 1(" 51. Iff and g are periodic functions of period T, and Tz,
o -7T<X<O
(f + g) are the sums of the corresponding Fourier then is the sum (f + g) necessarily a periodic ftmc-
6. f(x) = {1: 0:5; < 7T ,J(x + 2TT) = f(x)
series coefficients off and g. 0, 2:::: x::: 1T
tion? If not, then what conditions on T, and T2 cause
X -1 :5x<O (f + g) to be periodic?
*7. f(x) = { 1: 0:5 x < I ,J(x + 2) =f(x) 28. Use the conclusion of Exercises 26 and 27 to rmd (a) x = 0, (b) x = -7T/2, (e) x = 7T/2, (d) x = 37T/4,
52. (Fourier Series-Complex Form) (a) Use Euler'S
2 2 (e) x = -7T.
8. f(x) = cos z 2x the Fourier series of (a) f(x) = {3 , 0- :5 x < 0 formula e iO = cos 8 + i sin 8 to show that cos nx =
, :Sx:S2' 1 He inx + e- inx ) and sin nx = ~(ei1U - e- 1nx ). (b) Write
9. f(x) = eX, -I :5 x < 1,J(;< + 2) = f(;<) where f(x + 4) = f(x). (Hint: See Example I.); 2, -1:5 x <-2
the Fourier series
o -1 <x<O (b)f(x) = x + 2, -I :5 x < I,J(x + 2) =f(x). (Hint:
1 1 I
10. f(x) = {:~x, 0 ~ x :5 1,J(x + 2) = f(x) See Example 5(b).); (c) f(x) = Ixl - 1, -1 :5 x < I, x, I
~
f(x + 2) = f(x). (Hint: See Example 5(a).) *37. f(x) = -2 :5 x < 2 ,J(x + 2) = f(x), f(x) = 2ao + (an cos nx + bn sin nx)
1
n~'
1
In Exercises 11-20, determine if the given periodic ftmction is 29. Show that the Fourier series coefficients of the func- -I, 2:5X:51 of the functionf of period 27T as
even, odd, or neither. tion cf, where c is a constant, are c times the corre-
1 I
sponding Fourier series coefficients off (a) x = 0, (b)x= -2' (e)x= 2' (d)x= -1, f(x) = Co +I (cne i= + Cne~i=).
11. f(x) = x', -I :5 x < 1,J(x + 2) = f(x) n~'
30. Use the conclusion of Exercises 26, 27, and 28 to (e)x = 1.
12. f(x) = 1 + 4x 2 , -1 :5 x < 1,J(x + 2) = f(x) (e) Show that
· d ..f fi() {2 -2:5x<0
*13. f(x) = x 2 - x, -7T:5 x < 7T,J(X + 27T) = f(x)
-I -7T<X<O
fm the Founer senes 0 (a) x = 4:
wheref(x + 4) =f(x); (b)f(x) = -lxi, -I :5x<l,
0:5 x :5 2 '
In Exercises 38-41, (a) sketch the even extension of the given Cn
I f~·
= -2 f(x)e~'= and
I f~
1:" = -2
-~
.
f(x)e'= dx.
Dmction; (b) find the Fourier series of the even extension. 7T -n 1T
14. f(;<) = { 2, , 0 :5 x ~ 7T ,J(x + 2TT) = f(x) f(x + 2) = f(;<); (e) f(x) = I -lxi, -I :5x< I,
f(x + 2) = f(x). 38. f(x) = x 2 , 0 :5 x < 1 (d) Using the notation C~n = em show that the series
15. f(x) = {-cos x, -7T:5 x < 0 f(x + 27T) = f(x)
cos x, O.s x < 7T ' 39. f(x) = {a,2, 0:5 x < I
can be written in the complex formf(x) = I cne 'nx ,
In Exercises 31-32, use the Fourier series 1:5 x < 2
16. f(x) = Isin xl, -7T :5 x < 7T,J(X + 27T) = f(x) where
<
*17. f(x) = xlxl, -2 :5 x < 2,J(x + 4) = f(x)
18. f(x) = e~l, -I :5 x < I,J(x + 2) = f(x)
pZ
3
4pz ~ (-1)"
- + ---z L. --z-
7T n=l n
cos - -
n7TX
P
40. f(x) = I - x, 0 :5 x
*41. f(;<) = X(7T - x), 0 :5 x
7T
< 7T
1 f~ f( x)e ~Inx dx , n --
en -- -2
7T ~~
° +1 , +2
,- - ,'"
2
19. f(x) = x 2
, 0 :5 X < 1,J(x + 1) = f(x) forf(x) = x , -p :5 x < p,J(x + 2p) = f(x) to verify the given
infinite sum. In Exercises 42-45, (a) sketch the odd extension of the ftmc- In Exercises 53-54, rmd the complex form of the Fourier se-
o f( ) - {I, 0:5 x < 1, f( 2) _ tion from the given exercise; (b) rmd the Fourier series of the
2. x - -1, 1 :5x<2' x+ -f(x)
21. Show that iffand g are even ftmctions, thenf + g is

31. 1 + t + t I~ +
1
+ 2 5 + ... = :2 odd extension.
42. f(x) = x 2 , 0 :5 x < I

ries for the given ftmction.
53. f(x) = x, -7T:5 x < 7T,f(X + 27T) = f(x)
an even function.
22. Show that iff and g are odd ftmctions, then f + g is
32. 1 - i t-+
1
116 + 2 5 + ... = rz 43. f(x) = {~'._, 0:5 x < 1
1:5 x < 2
54. f(x) = {
o
I: -7T<X<O
0:5 ~ < 7T ,f(x + 27T) = f(x)
an odd ftmction. 33. Use the results of Exercises 31 and 32 to show that
23. Suppose that f is an even ftmction and g is an even I I ~ 44. f(x) = I - x, °:5 x < 7T 55. (The Weierstrass M- Test) If ') Mn is a convergent
~l
ftmction. Prove that fg is an even ftmction.
1+
9 +25"+ ... =8·
*45. f(x) = X(7T - x), ° :5 x < 7T series of positive constants and lan(x)1 :5 Mn on the
534 Chapter 9 Eigenvalue Problems and Fourier Series 9.4 Generalized Fourier Series 535
Fourier series for f(x) = (7T - x)/2 on 0 < x < 27T is +T 63. In Example 4 we showed that I - 113 + liS + ... =
interval I of x, then L an (x) is uniformly convergent bn = (lip) I
S
S
f(x) sin(n=lp) dx, n = 1,2, ....

n=l L (sin nx)ln. (b) Verify thatf(x) satisfies the condi_
7T/4. Experimentally determine the number terms of
on 1. Show that i: ..;. sin nx is uniformly convergent

n=l Find the Fourier series for the given periodic function: the series that are needed so that the first five digits
of the approximation of 7T/4 agree. (Suggestion: Use
r
n=1 n tions for term-by-term integration. (e) Show that (a) f(x) = 2 - x, I S x < 3,J(x + 2) = f(x)
on -00 <x< co. a graphing utility to graph two periods of the periodic
f(x) dx = b(27T - b)/4. (d) Integrate term-by-term (b) f(x) = 2x - 7T. 7T S X < 27T,J(X + 7T) = f(x) extension of the functions in Exercises 1-10. Graph
56. (a) Show that the Fourier series for f(x) = x2 , -I S o
x < 1,J(x + 2) = f(x), is the Fourier series for f(x). (e) Replace b in the ex- o -112 S x < 1/2 an approximation for each function using several
pression in part (e) by x to show that
(e) f(x) = {I: 1/2 S x < 3/2 ,J(x + 2) = f(x) terms of the series. Also, carry out similar graphing
I ~ 4(-1)" techniques with Exercises 38-49.)
-3 +L -2-'- cos n=. ~ I ~ cos n7T (d) f(x) =
X
I:
-I S x < 1/2
lI2 S x < 2 ,J(x + 3) = f(x)
n=l n 7t" X(27T - x)/4 = L- 2" - L- - - 2 - ' {
n=l n n=l n
(b) Use the Weierstrass M-test to show that this series
converges uniformly for -00 < x < 00. (e) Graph (f) The sum of the series L
n~1
IIn2 represents the con-
several periods of this function simultaneously
stant aol2 in the Fourier series for ;,(27T - x)14 on
with a selected number of terms of its Fourier se-
o < x < 27T. Determine the value of L lIn 2 by f'md-
ries to observe the convergence.
ing ao. (g) What is the value of
.-1
~ Generalized
57. (Term-by-Term Differentiation) If f(x) is periodic,
Fourier Series
continuous, and piecewise smooth (that is,f' (x) exists
except perhaps at a finite number of values of x and -
I Lh X(27T - x)
- - - - cos nx?
7T 0 4
f' (x) is piecewise continuous), then the differentiated In addition to the trigonometric eigenfunctions that were used to form the Fourier se-
Fourier sreies off(x) converges tof'(x) at every value (h) What is the value of ries in Sections 9.2 and 9.3, the eigenfunctions of other eigenvalue problems can be
of x where f"(x) exists. (a) Differentiate the function I i2~ X(27T - x) . used to form what we call generalized Fourier series. We find that these series will
in Exercise 56 to obtain g (x) = 2x, -I S x < I, - ----smnx? TABLE 9.4
7T 0 4 assist in solving problems in applied mathematics that involve physical phenomena that
g(x + 2) = g (x). Compare the direct calculation of
the Fourier series of g with that obtained by term-by-
term differentiation of the series in Exercise 56. Are
60. (a) Calculate
o
r
f(x) dx for the function [(x) =
(7T - x)12 in Exercise 59. (b) Use term-by-term in-
j
'I
n
I 2.4048
cannot be modeled with trigonometric functions.
Recall Bessel's equation of order zero
they the same? (b) Is the convergence of the Fourier x 2y" + xy' + A2x 2y = 0,
series of g uniform on -00 < x < oo? (Note: Graph tegrati~n of L 2 5.5201
several terms of the Fourier series together with g. Is
the convergence the same as it was in Exercise 56? If L~f(X)
.-1 (sin nx)ln to approximate the value of
dx. How many terms are needed so that the

8.6537
with linearly independent solutions Jo(Ax) and Yo(Ax). If we require that the solutions
I
of this differential equation satisfy the boundary conditions y(o)1 < 00 (meaning that
the graph exhibits the Gibbs phenomenon, the con- the solution is bounded at x = 0) and yep) = 0, then we can find the eigenvalues of
11.7915
vergence is not uniform.) (e) Can we differentiate the difference between the exact and approximate values the boundary-value problem
terms of the Fourier series for g to obtain that of is less than 0.01? 14.9309
hex) = 2, -I S x < I, h(x + 2) = hex)? Why or why 61. Find the Fourier series for each of the periodic func- X + xy' + A2x 2y
2y" = 0, O<x<p
not? (d) What conciusion(s) do you draw concerning tions. In each case, graph !be function and an ap- {
ly(O)1 < 00, yep) = 0
the relationship between uniform convergence of the proximation using several terms of the series.
Fourier series of a function and the ability to differ-
A general solution of Bessel's equation is y = cjJo(Ax) + c2Yo(Ax). Because Iy(O)1 <
(a) f(x) = x 3 , -7T S X < 7T,J(X + 27T) = f(x) y 00,we must choose C2 = 0 because x!iW+ Yo(Ax) = -00. Hence yep) = cjJo(Ap) = O.
entiate the Fourier series term-by-term to obtain the
Fourier series of the derivative of the function? What
(b) f(x) = x 2 (1 - x), -7T S X < 7T, f(x + 271') = 1 Just as we did with the eigenvalue problems solved earlier in Section 9.1, we cannot
~
f(x)
property must the periodic function f possess so that 0.8 choose Cj = 0, so we must select A so that Jo(Ap) = O. Unfortunately, the values of x,
the convergence by its Fourier series is uniform? '(e) f(x) = X4, -7T S X < 7T,f(X + 27T) = f(x) 0.6 where Jo(Ax) = 0, are not as easily expressed as they are with trigonometric functions.
58. Compare the rate of convergence of the Fourier series (d)f(x)=x(l-x 3 ), -7TSX<7T, f(x+27T)= 0.4 From our study of Bessel functions in Section 4.8 we know that this function intersects
of f and g in Exercises 56 and 57. Which converges f(x) I., 0.2
0
the x-axis in infinitely many places. In this case, we let an represent the nth zero of
more quickly and why? (Hint: Consider the Fourier 62. (Fourier Series on Any Interval) If f is a periodic x the Bessel function of order zero, J o, where n = 1, 2, .... These values are found in
series coefficients.) function of period T = 2p defined on [s, s + Tj, then
its Fourier series coefficients are an =
I -0.2
-0.4 tables in many applied mathematics and physics books and can be approximated through
the use of a computer algebra system as well. We list a few of them in Table 9.4 and
I
59. (Term-by-Term Integration) If f(x) is periodic and
'+T
piecewise continuous, then the Fourier series for f(x)
can be integrated term-by-term. (a) Show that the
(lip),
I f(x) cos(n=lp) dx, n = 0, I, 2, ... , and
Figure 9.17 Jo(x)
show the graph of J o in Figure 9.17. By observing the graph, do the values in the table
correspond to the roots of J o?
9.4 Generalized Fourier Series 537
Therefore, in trying to find the eigenvalues, we must solve Jo(Ap) = O. From our
cises 4.8. Because A = An = an/p, n = 1,2, ... , the function Jo(anx/p) = Jo(Anx) sat-
definition of an> this equation is satisfied when Ap = an> n = I, 2, .... Hence the
isfies Bessel's equation of order zero given by
eigenvalues are
A = An = an, n = 1,2, ... ~ [x ~ Jo(AnX)] + A/xJo(Anx) = O.

p
Multiplying by the factor
and the corresponding eigenfunctions are
y(x) = Yn(x) = Jo(Anx) p ,

= J o( anx) n = 1,2, ....
we can write the expression as
As with the trigonometric eigenfunctions that we found in Section 9.2 and 9.3,
Jo( anx/p) can be used to build an eigenfunction series expansion of the form d [d
dx x dx Jo(An x ) ]2 + An22 dxd [Jo(An x)]2 -- 0.
X
f(x) = i
n~1
enJo(anx),
P
Integrating this expression from x = 0 to x = p, we have
where we use the orthogonality properties of J o( anx/p) to find the coefficients en.
We determine the orthogonality condition by placing Bessel's equation of order
zero in the self-adjoint form Because AnP = (an/p)p = an> we make the following substitutions
Because the weighting function is sex) = x, the orthogonality condition is Then J o(an) = 0, because an is the nth zero of Jo. Also, with n = 0, the identity
indicates that Jo(a n) = -J1(a n). Therefore,

Multiplyingf(x) = L
n=l
enJo(anx/p) by s(x)JoCAmx) = xJo(amx/p) and integrating from
X = 0 to x = p yields
fo Xf(X)Jo(amx)dx= f i enXJo(anX)Jo(amX)dx
p P 0 n~1P
i enfxJo(anx) Jo(amx) dx.

=
n~1 0 P P
where the value of J 1(a n ) can be found in many applied mathematics and physics text-
books (or through the use of a computer algebra system). Therefore, the series coeffi-
However, by the orthogonality condition, each of the integrals on the right-hand side cients are found with
of the equation equals zero except when m = n. Therefore,
en = 2 2 2 Jf' xf(x)Jo(anx)
- - dx, n = 1, 2, ....
p [J1(an )] 0 p
"ii"I,'.'
Find the Bessel-Fourier series for I(x) = 1 - x2 on 0 < x < I.
The value of the integral in the denominator can be found through the use of sev-
eral of the identities associated with the Bessel functions that were verified in Exer- : I Solution In this case, p = I, so
Cn = [J1(an)f
2 r1
Jo x(I - X2)JO(an x) dx
y y
0.8 0.8
= [Jl(!n)]2 {fXJo(anX)dx- fx3Jo(anX)dx}
0.6 0.6
Using the fonnula dldx [xnJn(x)] = -xnJn_1(x) with n = 1, 0.4 0.4

0.2 0.2
1 x 0.2 0.4 0.6 0.8 1 x
Note that the factor I/an is due to the chain rule for differentiating the argument of Ca) Cb)
J I(anx). We use integration by parts with u = x 2 and dv = xJo( anx) to evaluate _ 4J2(al)Jo(alx) (b) ( _ 4J2(al)JO(a,x) + 4J,(a2)JO(a2 x )
Figure 9.18 (a) p(x) - a,2(JI(a,)]2' P x) - a,2[J,(a,)]2 a22[J'("2)f
fx3Jo(anX) dx. As in the first integral, we obtain v = - ~n JI(anx). Then, because
du = 2x dx, we have
0.6 Theorem 9.3 Convergence of Bessel-Fourier Series
Suppose that/and/' are piecewise continuous functions on 0 < x < p. Then the
Bessel-Fourier series for I on 0 < x < p converges to I(x) at every x, where I is
continuous. At x = Xo, where I is discontinuous, the Bessel-Fourier series con-
verges to the average
Thus the coefficients are
2
Cn = [J (a )]2
r x(1 -
l
Jo x2)Jo(anx) dx
Ii
I n Figure 9.19 JI(x)
= [JI(!n)f {(xJo(anx) dx - f 3
x J o(a nx) dx}
Series involving the eigenfunctions of other eigenvalue problems can be fonned
as well. We illustrate this procedure in the following example.
a~2 J 2(an))]
o "fi,,'·'·'
= [JI(!n)f [:: JI(a n) - (:: Nan) -
4J2(a n)
a 2 [Jl(a )f' n = 1,2, ... In Example 2 in Section 9.1, we found that the problem y" + 2y' - (A - I)y = 0,
n n
subject to yeO) = 0 and y(2) = 0, has eigenvalues A = An = -k2 = -(n7T/2f,
so that the Bessel-Fourier series is n = 1,2, ... and eigenfunctions y(x) = Yn(x) = e -x sin(n1lX)/2. Use these eigen-
functions to approximate I(x) = e -x on [0, 2].
TABLE 9.5 y
0.6 : ] Solution To approximate f, we need the orthogonality condition for these eigen-
n ],(':'n) ],(an )
In Figure 9.18 we graph I along with several terms of the series. Notice that the functions. We obtain this condition by placing the differential equation in self-
0.4
1 0.5192 0.4318 polynomial with two tenns yields an accurate approximation off When using three adjoint fonn using the fonnulas given in Section 9.1. In the general equation,
f2dx
or four tenns, the graphs are practically indistinguishable. We list the values of J I (an) 0.2 a2(x) = 1, al(x) = 2, and ao(x) = O. Therefore, p(x) = e = e2x and sex) =
2 -0.3403 -0.1233
and J 2 (an ) in Table 9.5, as well as show the graphs of these functions in Figure 9.19 P(X)la2(X) = e 2x, so the equation is
3 0.2715 0.0627 and Figure 9.20, respectively.
-0.2 dx dY ] - (A - I)e 2xy = O.
d [ e2x dx
4 -0.2325
r
-0.0394
5 0.2065 0.0277 As was the case with Fourier series, we can make a statement about the conver-
gence of the Bessel-Fourier series. Figure 9.20 J 2 (x)
This means that the orthogonality condition. s(x)Ym(x)Ym(x) dx = 0 (m * n), is
jo2e 2x -x
e
.
sm -2-e
m= -x .
sm -2- dx
n=
=
j2.sm -2-
0
m= . n=
sm -2-dx = 0 (m"* n).
EXERCISES 9.4
We use this condition to detennine the coefficients in the eigenfunction expansion

~ ~
f(x) = I
n=1
enYn(x) = I
n=l
ene- X sin n;x. For each eigenvalue problem, verify the given eigenfunctions.
Then use the eigenfunctions to obtain the generalized Fourier
(Fourier-Legendre Series) In Exercises 10-20, we
investigate the use of Legendre polynomials in eigen-
series for each of the indicated functions f(x). function expansions.
MUltiplying both sides of the equation by y",(x) = e -x sin(m=/2) and sex) = e2x
and then integrating from x = 0 to x = 2 yields 1. y" + Ay = 0,0 < x < I, yeO) = 0, y'(1) = 0; y.(x) = 10. The Legendre polynomials satisfy the Rodrigues for-
jo2f(x)e 2x
e- X sin m;x dx =
j2 I~ ene- X sin n;x e2x e -x sin m= dx
. (Zn - 1)17X
sm Z ' n = I, Z, ... ; f(x) = I, f(x) = x, mula, p.(x) = z:n! J:. 2
[(x - 1)"]. Use this for-
0 n~l 2 f(x) =x 2 mula and integrate by parts n times to show that

2
jo f( x )e x .
sm m= dx = L
-2- "j
~ 2
.
n= sm
en sm -2- .' -2-
m= dx.
2. y" + Ay = 0, o<x<
(Zn - 1)17X
l,y'(O) = O,y(l) = O;y.(x) = I_,' 2
p.2(X) dx = -Z--I'
n n = 0, 1,2, ....
+
n=1 0 cos Z ' n = I, Z, ... ;I(x) = l,f(x) = x, 11. Show that the coefficients of the Fourier-Legendre
Each integral in the sum on the right-hand side of the equation is zero, except when f(x) = x 2 seriesf(x) = L c.p.(x) for the functionfdefined on
m = n. In this case, f sinZCn=/2) dx' = 1. Therefore, *3. y" + Zy' + (I - A)y = 0, 0 < x < I, yeO)
y(1) = 0; y.(x) = e- x sin n7TX, n = I, Z, ... ;
= 0,
(-I, I) are c. =
.~O
Zn + 1 ('
-z-L,
.
f(x)P.(x) dx. (Hznt: Use
en = f f(x)e X sin n;x dx.

I, 0 <x < liZ
f(x) = l,f(x) = {0, IIZ:5 x :5 I'
nomials, r-,
the orthogonality condition of the Legendre poly-
p.(x)P",(x) dx = 0, m * n, and the result

Forf(x) = e-X, 4. y" + y' + Ay = 0, 0 <x < 3, yeO) = 0, y(3) = 0;
2 , of Exercise 10.)
en = j0 eXe
-x .
sm -2- dx
n1TX
= Lsm -2- dx
- . nm y.(x) = e- x / 2 sm n;x, n = I, Z, ... ;f(x) = I,
f(x) = {I, 0< x < 3/Z

12. Find the coefficients c. for n = 0, I, Z, and 3 for the
I -1 :5x<O
0:5 x < 0: I.
= [_.2.. n11"
CQS.
.
n=]2 =
2 G
_.2.. (cos 1)
n11"
n11" -
5. y"
-I,
+ Zy' + (A +
3/Z:5x:53'
I)y = 0, 0 < x < 5, yeO) = 0,
Fourier-Legendre series forf(x) = {
13. Find the coefficients c. for n = 0, I, 2, and 3 for the

n17X o -I <=x<O
Because cos n 11" is equivalent to (-1)", we can write the eigenfunction expansion
{h,
yeS) = 0; y.(x) = e -x sin 5 ' n = I, 2, . Fourier-Legendre series forf(x) = 0;;x < I .
offas
I, 0<x<3
14. Iffis an even function, then all powers of x infmust
f(x) = f -.2..(( -1)" - l)e-
n~l n11"
X
sin n=
2
f(x) = l,f(x) = { 0,
6. y" + 4y' + (4 + 9A)y = 0,

3:5 x:5 5'
0 < x < 2, yeO) = 0,

be even nonnegative integers. Are the functions P 2.(x),
n = 0, 1, . . . , even? Therefore, if f is an even
n17X function, then the coefficients (with odd subscripts)
= -±-e-x sin = + --±-e-x sin 3= + --±-e-x sin 5= + ... y(Z) = 0; y.(x) = e -Ox sin -Z-, n = I, 2, ... ;
C2n+ 1 = O. Show that the other coefficients are given
11" 2 311" 2 511" 2
In Figure 9.21, we graph the approximations obtained with the fIrst three nonzero
f(x) = l,f(x) = { -I,
I, O<x<1
1:5 x :5 Z· by C2n = (4n + I) f
o
f(x)P 2.(x) dx.
terms of the series. *7. Solve the eigenvalue problem y" + y' + (I - A)y = 15. Find the first three nonzero coefficients in the Fourier-
0, yeO) = 0, y(l) = o.
Use the eigenvalues and eigen- Legendre series for f(x) = lxi, -1 < x < I.
functions obtained to form an eigenfunction expan- 16. If f is an odd function, then all powers of x in f
sion of an integrable functiong(x) on the interval 0 < must be odd nonnegative integers. Are the functions
o,!~,Y._ o.!~.:-\.
1 x < 1. (Hint: The eigenfunctions are orthogonal with P 2.+ I (X), n = 0,1, ... , odd? Therefore, iffis an odd
respect to the weighting function sex) = eX, which is function, then the coefficients (with even sUbscripts)
0.8
... ' obtained by placing the differential equation in self-
r
0.6 0.6 /' \. 0.6 i
\
' C2n = O. Show that the other coefficients are given by
0.4 0.4 I .~ -'. adjoint form.)
0.4' .. "" C2n+1 = (4n + 3) f(X)P2n+I(X) dx.
0.2 0.2 ' '-". 0.2 ,'~ ~" 8. Use the eigenfunction expansion obtained in Exercise o
-', 7 to approximate f(x) = x on the interval 0 < x < 1.
2 x
17. Find the first three nonzero coefficients in the Fourier-
0.5 l.5 0.5 1 l.5 2 x 0.5 1 1.5 2 x
9. Use the eigenfunction expansion obtained in Exercise -I -I <= x< 0
Figure 9.21 (a) Legendre series for f(x) = { I, , 0 :5 x ~ 1 .
(b) (e) 7 to approximate f(x) = x 2 on the interval 0 < x < 1
542 Chapter 9 Eigenvalue Problems and Fourier Series Chapter 9 Review Exercises 543
18. Term-by-term differentiation was discussed in Section ! Section 9.4!

9.3 with Fourier series. Can a Fourier-Legendre se-
23. The function I is called a generating @ection 9.2!
VI-2xu+u z
ries be differentiated term-by-term? Consider the
function for the Legendre polynomials because Fourier Sine Series Generalized Fourier Series
functions in Exercises 15 and 17.
19. One of the more common uses of Legendre polyno-
I
~=======;;Z =
~
L: Pn(x)un. [(x) = i: bn sin!!:'E.., where bn = l fJ(X) sin!!:'E.. dx
[(x) = L: cnYn(x), where the Yn(x) are eigenfunctions of an
mials is in solving Laplace's equation (a partial dif- VI - 2xu +U n~O n~' p PoP
n=l
eigenValue problem and the coefficients Cn are found with the
ferential equation) in spherical coordinates. In this
Verify this relationship by using the binomial eXpan- Fourier Cosine Series orthogonality condition of the eigenfunctions.
case, solutions appear in the form P nCcos If). Show that
. fi 1. Z
t;;e coefficients in the series of the formJ(1f) = slOn or VI _ w With w = 2xu - u . Expand the I ~ nm: Bessel-Fourier Series
J(x) = Iao + n~' an cos p'
L: c"Pn(cos 0) are found with
n-O
Cn = expression that results and collect like terms in the J(x) = i: CnJo( anx ), where
p LJ(x) cos P P
powers ofu. P n=l
2n +1 L~ where an = 2 0
nm: dx, n = 0, 1,2, ...
-2- 0 J(O)Pn(cos 0) sin 0 dO. 24. In Exercise II, the coefficients of the Fourier-
20. Use the formula in Exercise 19 to find the first three Legendre series J(x) :: L:
n-O
c"Pn(x) for the functionJ
coefficients in the series J( 8) = L: cnPn(cos 0) if defined on ( - I, 1) are found to be C = [section 9.31
J(O) = 0, °< 0< 71'.
n~O
2n + 1
-'-2-
I' . n
_ l(x)Pn(x) dx. Fmd the first ten coefficients
Fourier Series
21. Find the first ten coefficients of the Bessel-Fourier
series for (a)J(8) = 0, o:s: O:S: 71', (b)J(O) = .
for the functIOns (a) J(x) =
{II'-x,
-I:s: x<°
O:S:x<I' [(x) = tao + i: (an cos nm: + bnsin nm:), where
O, o:s: 0 < 71'/2
{ 71'/2 - 0, 71'/2:s: o:s: 71" In each case, graph J X', -I:s: x < ° n~' p p
simultaneously with the Bessel-Fourier polynomial

(b) J(x) = {x, O:s: x < I . In each case, graph J an = -1 fa J(x) cos -
nm:
- dx, n = 0, 1,2, ... , and
simultaneously with the Bessel-Fourier polynomial P -a P
using the first ten coefficients.
22. Verify the Rodrigues formula, Pn(x) =
using the first ten coefficients.
bn = -
I fP J(x) sm -
. n71X dx, n = I, 2, ...
p -p P
2!n! ;:n [(x' - I)"], for n = 0, I, ... , 5.
CHAPTER 9 SUMMARY CHAPTER 9 REVIEW EXERCISES
~ Concepts & Formulas

and yell = O. (a) Solve the eigenvalue problem
! Section 9.11 Self-Adjoint Form of the Sturm-Liouville Problem

1. The shape of a quickly turning jump rope can be de-
termined by solving an eigenValue problem. Suppose y" + Ay = °, ° .
{yeO) = 0, yell =
0<x <L .
(b) What are the elgen-
'
that the rope oflength L and constant linear density p
dx dY )
d (p(x) dx + (q(x) + As(x»y = 0, values? (c) For what values of w is y a function other
is whirled with a constant angular speed w. Assume
Eigenvalue Problems where that the rope whirls about the x-axis so that every point thany = O? (These values are called the critical speeds
Values of the parameter that satisfy the differential equation _ fa,(x) dx _ ao(x) p(x) on the rope moves in a circle about the x-axis. Lety(x) of angular rotation.)
are called eigenvalues of the problem, and for each eigen- p(x) - e a,(x) , q(x) - a,eX) p(x), and sex) = az(x), denote the displacement of the rope from the axis
value the nontrivial solution Y that satisfies the problem is of rotation. By considering all of the forces acting 2. The beam problem that was discussed in Section 5.4
called the corresponding eigenfunction. Linear Independence and Orthogonality of Eigenfunctions on the rope, we find that Y satisfies the equation can be interpreted as an eigenvalue problem. Let y(x)
If Ym(x) and Yn(x) are eigenfunctions of the regular Sturm- 1)/' + pwZy = 0, where T is the constant tension on represent the displacement from the equilibrium posi-
Sturm-Liouville Problems *
Liouville problems, where m n, then Ym(x) and Yn(x) all points along the rope. If we let A = pwzlT, tion of the beam, the x-axis. Then the situation of the
displacement of a beam that is SUbjected to a com-
az(x)y"(x) + a,(x)y'(x) + [ao(x) + A1Y(x) = 0, a < x < b, are linearly independent, and the orthogonality condition then this equation is y" + Ay = O. Because the ends
with boundary conditions a,y(a) + azy'(a) = aD, f3,y(b)
f3d(b) = O.
+
f s(x)Ym(x)Yn(x) dx = 0 (m * n) holds. of the rope are held at a constant height along the
axis, we also have the boundary conditions yeO) = °
pressive force is modeled by the differential equation
°
(dz/dxz)[Ely"l - Py" = 0, < x < L, where E is the
544 Chapter 9 Eigenvalue Problems and Fourier Series Differential Equations at Work 545
modulus of the beam material, f is the moment of in-

ertia of the beam's cross-section, and P is the magni- 7. f(x) = {I,-I, °<7T'<x:S
x :s 7T'
27T'
_ 2. 2. ~
.:S-,
f(x) - 2 + 7T'
sin(2n - I)x
2n - I .
f(x + 47T') = f(x). Determine the numerical value to
which the Fourier series off converges at (a) x = 0;
tude of the compressive force. If E, f, and P are con-
= {a, °< x :s 7T'
(b) x = -7T'/2; (c) x = 7T'; (d) x = 27T'; (e) x = 57T'/4.
stant, then let A = PIEf, so that the differential
equation becomes y(4) - Ay = 0. As we recall, the
8. f(x)
1, 2'1T
7T<X::;
Use this result to verify that
(-1)"+' I I I 7T'
*21. Using the trigonometric identities sin' x = i sin x -
beam can be subjected to a wide variety of boundary
°< :s
00
k,~=1-3+5-7+'''=4'
conditions as listed in Table 9.6, where each endpoint *9. f(x) = {x'
2 - x,
x I
I <x:S 2 4'I sm
. ,3 I 3 d .
3x and cos x = 4' cos x + 4' cos x, etermllle
has a pair of conditions. Investigate the solution of this -I, -27T':Sx:S-7T'
eigenvalue problem for each of the following cases. 10. f(x) = °< x2 :s I
{I0, - x,I <X:S 20. Consider the functionf(x) = x, -7T' < X :s 7T' ,
the Fourier series for f(x) = sin' x andf(x) = cos' x.
(a) Fixed-end conditions at x = ° and x = a

{
I, 7T'<x:S 27T'
(b) Free end atx = °
and x = a
(c) Simple support at x = °and x = a
In Exercises 11-16, compute the Fourier series for each peri-
odic function.
(d) Sliding clamped end at x = °and x= a
I, -7T':S x < -7T'/2
(e) Fixed end at x = 0; simple support at x = a 11. f(x) = 0, -7T'/2:S x < 7T'/2,f(x + 27T') = f(x)
{
(f) Free end at x = 0; sliding clamped end at x = a I, 7T'/2:s x< 7T'
Because of the relationship A = An = PIEf, we have
-I, -7T':S x < ° .~ Differential Equations at Work I I
P = P n = A,.EI. Therefore, we can interpret P, as the
largest load that the beam can withstand before it
12. f(x} = { I, o:s x < 7T' ,f(x + 27T') = f(x)
A_ Signal Processing I
buckles. Determine P, for each of the cases given pre-
viously.
*13. f(x) = { I:
X -7T:$X<O
o:s x < 7T' ,f(x + 27T') = f(x)
Suppose that we have a signal that is made up of oscillatory functions as well as a com-
X -1T':::::;;X<O
TABLE 9.6
14. f(·<) = { 0: o:s x < 7T' ,f(x + 27T') = f(x) ponent known as noise. The goal of signal processing is to find the useful part of the
signal while filtering out the noise. Let us suppose that the signal is
I. -I :sx<o
Fixed End y=O y' = ° 15. f(x) F { I _ x, o:s x < l,f(x + 2) = f(x)
JCt) = Ao + I
10
an cos(2mrt) + I
10
bn sin(2n1Tt) + noise(t)
Free End y" = ° yilt = 0 .. ' {x+ I, -I :Sx<o ,f(x + 2) = f(x)
16. f(x) = x, o:s x < I
n=l n=1
Simple Support y=O y"= ° 17. Verify that the Fourier series forf(x) =x 2 , -7T'<X<
Then, because of the orthogonality properties of the functions cos(2n1Tt) and sin(2n1Tt)
on the interval [0, 1], we can determine A o, an, and b., (n = I, 2, ... , 10) in the se-
Sliding Clamped End y' = ° y'" = 0 1T is ries expansion of the filtered portion of the signal with the calculation
_ -rr
f(x) - 3 +4 L
n=1
00 (-I)"
--2-
n
cos nx. fo
1 10
J(x)k(t, x) dx, where k(t, x) = 2 mi;,cos 2m1Tt cos 2m7TX + :&,
10
sin 2m1Tt sin 2m7TX.
I
In Exercises 3-6, find the Fourier sine series of each of the fol-
lowing functions. Use this series to find the sum of the following infi-
nite series by choosing a suitable value of x: (Note: The same results are obtained by calculating the integrals Ao = f J(x) dx, I
{I,-1, °<7T<x:::27T °
I
-r?
3. f(x) = :s 7T'
x
(a) k, (-1)"+'
00
-n-2-
I I I
= I - ? + 32 - 42 + ... = 1z an = J' J(x) cos 2m7T dx, and bn = f J(x) sin 2m7TX dx individually.)
4. f(x) = {a,1, °< x :s 7T' (b) f l...2 = I + l... + l...2 + l... + ... = -rr6 1.
° °
What is the frequency of cos 2m7TX and sin 2m7TX?
I
TT<X:$ 27T n~l n 22 3 2 4
2t, t < 1/4
*5. f(x) = {x' °<
x :s I
2-x,l<x:S2
18. Use the two series in (a) and (b) in the previous prob-
lem to show that 2. LetJ(t) = get) +
1 .
10 sm(3 °1Tt) - 4' 1
whereg(t) =
1 - 2t, 1/4 < t < 112
2t _ 1, 112 < t < 3/4' Graph
[
6. f(x) = {I0, - x,I <x:s
0< x :s I
2
f __1__ = I +l...+J..+
F', (2n - Ii 3 52
... = -rr.
8 2
2 - 2t, t> 3/4
the signal on [0, 1].
II
10 10
In Exercises 7-10, compute the Fourier cosine series of each

19. Show that the Fourier series for f(x) =
O, -7T'<X<O .
3. Filter the signal to obtain Ao + I
n=l
an cos 2n1Tt +I
n=l
bn sin 2n1Tt. Graph this func-
of the functions in the indicated exercise. { I, o:s x :s 7T' ,f(x + 27T') = f(x) , ,s tion on [0, 1].
~.
~
~
;;;
546 Chapter 9 Eigenvalue Problems and Fourier Series Differential Equations at Work 547
4. Describe the differences between the original signal and that found in 2. 3. Use the values of A and B in 2 to express A cos nt + B sin nt as 0' cos(nt - </»,
5. Take the difference between the original signal and that found in 2. What does thi where 0' = VA 2 + B2, cos</> = Ala, and sin </> = BIO'.
function represent? Graph this function on [0, I]. s 4. Write out the first ten terms of xp(t).
6. What terms in an expansion of the formf(t) = AD + ~an cos 2nm
should be used so that we filter out frequencies greater than four?
+ ~bn sin 2n7T/ 5. What is the natural frequency of the spring-mass system if damping is excluded
from the ODE? Which term in the series representation of xit) has the largest co- I
II
7. Use the number of terms determined in 6 to filter out frequencies greater than four efficient (amplitude)? Plot xp(t) over [0, 27T]. What is the approximate amplitude of
+ ~/40). Graph the original signal, the filtered signal,
xp(t) over most of this interval? How does this relate to the coefficients in xp(t)?
if the signal is f(t) = Sin( t
and the noise on [0, I].
c. Approximations with Fourier Series
B. Forced Damped Spring-Mass System Suppose that we find the Fourier series for the functionf(x) on the interval -p < x < I
Recall the forced spring-mass system
mx" + ex' + kx = fCt)

p, and suppose that we use the fust k terms of this Fourier series to approximate f If
we define the error in the approximation to be
[tao + ± + bn sin !:!!!!..)],

I
I
E = f(x) - (an cos!:!!!!..
that was first presented in Chapter 5. H~re we take another approach to solving this n=1 P p
problem. Suppose that m = 1 kg, c = 0.05 kg/s, k = 50 kg/s 2 , and then the size (or magnitude) of the error is
°
< t < 7T/4
f(t) =
2/7T'
0, 7T/4 < t < 77T/4 J(t + 27T) = f(t).
\\E\\ = [[f(x)f dx - p[±a02 + n l(a/
t + bn2)J
I
(
2/7T, 77T/4 < t < 27T
Recall from Example I in Section 9.3 that the Fourier series for
I. Classify f as even, odd, or neither, and use this information to show that the Fourier
series for f is
f(x) = {I,
2,
-2:0; x <
0:0; x:O; 2
° is
I +~ 4'. n7T f(x) = -3 +~

/. ( I - (I)n) - I sm n7TX
. -
27T ~1 n"? sm'4 cos nt.
j 2 ,;';;"'1
-
n7T 2
- or
I
Therefore, the ODE becomes
i f(x) =
3
2' + ~1
~ 2 . (2n - 1)7TX
(2n - 1)7T sm 2 .
I
I
"° x + 50x = - I + ~
x + .OS' 4 sm
L, ----,
27T n=1 n7r
. -n7T cos nt.
4 (a) Compute \\E\\ for n = 10, 20, 30, 40, 50. (b) What happens to the value of \\EII as
2. We now use the Method of Undetermined Coefficients to find a particular solution, 1 n increases?
(c) Because \\E\\-> °as n -> "', we have Parseval's equality,
I
xp(t), of this equation. For the constant term, 1I(27T), show that the corresponding
p I ~
portion of xP(t) is 1/(100 7T). Then for the terms n~ sin n; cos nt, we assume that 1. J
p -p
[f(x)), dx = -a02
2
+ I
n=1
(a n2 + b/).
I
the particular solution has the form A cos nt + B sin nt. Substitute into the ODE to
show that .~ Use Parseval's equality to determine the sum of the series
:~ ~ 4
A = ~2 sin n7T
n7T 4
50 - n
2
(50 - n2 )2 + (0.05f n2
and I
n=1 (2n - I) 2,,?'
(How do we know that this series converges?) Verify your response by approximating
B = _4_ sin n7T 0.05n
n7T 2 4 (50 - n 2 )2 + (0.05f n2 .
the sum with
1000 4 I
Does xP(t) correspond to the steady-state or transient solution to the ODE? )' (2n - 1)2,,?'
,;';;"'1
I
I
I
~
~
lxi, -2 :s x < 2,f(x + 4) = f(x), is
10
(d) The Fourier series for f(x) =
f(x) = 1 - ~ i
"rz n~l
1 2 cos (2n - 1)=.
(2n - 1) 2
Use Parseval's equality to find the swn of
~ 1
I
n~l (2n - 1)4'
PartiaL DifferentiaL
Equations
'\
I
n the previous chapters, we saw that many physical and mathematical sit-
uations are described by ordinary differential equations. We also saw that
others are described by partial differential equations. For example, previ-
ously we noted that the time-independent Schrodinger equation in spherical co-
ordinates is given by the partial differential equation
Cl (r-Cl\ft)
2 2
1 -
- Cl (.
- + -1- - - Cl\ft) +-----+-(E-V)\ft=O.
srnO- 1
2
Cl \ft 2/L
2
r2 Clr Clr r2 sin 0 ClO ClO r2 sin 0 Clef? h
In this chapter, we will investigate one way to solve some partial differential equa-
tions, the method of separation of variables, which was mostly developed by
Fourier in his study of the heat equation. Many other methods have been de-
veloped to solve many partial differential equations, both analytically and nu-
merically. Solving partial differential equations, when possible, can be a difficult
problem. In fact, understanding the behavior of the solutions of most partial dif-
ferential equations can be a difficult problem, as it is with most ordinary differ-
ential equations.
549
550 Chapter 10 Partial Differential Equations 10.1 Introduction to Partial Differential Equations and Separation of Variables 551
Introduction to Partial Differential Equations and Is u(x, y) = In(x 2 + y2) also a solution of u"" + Uyy = O?
Separation of Variables We notice that the solutions to Laplace's equation differ in form. This is unlike
solutions to homogeneous linear ordinary differential equations, in which we found that
(" Separation of Variables (" Modeling with Partial Differential Equations
solutions were similar in form. In this chapter we find a solution to a partial differen-
tial equation that models a physical situation by using the boundary or initial condi-
We begin our study of partial differential equations with an introduction to some of the
tions associated with the physical situation. For example, if we require that the (time-
terminology associated with the topic. A linear second-order partial differential
independent) temperature u(x, y) at each point in a rectangular region centered at
equation (PDE) in the two independent variables x and y has the form
(0, 0) obtained by solving U xx + uyy = 0 be bounded at the origin, then the solution
A(x, y)u"" + B(x, y)uxy + C(x, y)uyy + D(x, y)ux + E(x, y)uy + F(x, y)u = G(x, y), could not be u(x, y) = In(x2 + y2), because lim In(~ + i) = -co.
(x.y)-->(O.O)
where the solution is u(x, y). If G(x, y) = 0 for all x and y, then we say that the equa- If the solution to Laplace's equation must satisfy the boundary conditions u(O, y) =
tion is homogelleous. Otherwise, the equation is nonhomogeneous. u( 71", y) = 0, then which of the jUnctions u(x, y) = y2 - x 2 and u(x, y) = eY sin x
(that satisfy u"" + uyy = 0) must be eliminated?
Some of the techniques used in constructing solutions of homogeneous linear or-
dinary differential equations can be extended to the study of partial differential equa-
Classify the following partial differential equations: tions, as we see with Theorem 10.1.
(a) u= + uyy = u; (b) Uxy + uUx = x.
: J Solution (a) This equation satisfies the form of the linear second-order partial Theorem 10.1 Superposition Principle
differential equation with A = C = 1, F = -I, and B = D = E = O. Because If U" U2, ... , Um are solutions to a linear homogeneous partial differential equa-
G(x, y) = 0, the equation is homogeneous. (b) This equation is nonlinear because tion in a region R, then
the coefficient of Ux is a function of u. It is also nonhomogeneous, because m
G(x,y) = x.
CjUj + C2U2 + ... + CmUm = I
k~j
c~'"
where c" C2, ••• , Cm are constants, is also a solution in R.

Definition 10.1 Solution of a Partial Differential Equation
A solution of a partial differential equation in some region R of the space of the
independent variables is a function that possesses all of the partial derivatives The Superposition principle will be used in solving partial differential equations
that are present in the partial differential equation in some region containing R throughout the rest of the chapter. In fact, we will find that equations can have an in-
and satisfies the partial differential equation everywhere in R. fmite set of solutions so that we construct another solution in the form of an infinite
series.
Separation of Variables
A method that can be used to solve linear partial differential equations is called sepa-
Show that u(x, y) = y2 - ~ and u(x, y) = eY sin x are solutions to Laplace's equa- ration of variables (or the product method), first used by the French mathematician
tion u"" + uyy = o. Joseph Fourier (1768-1830) in approximately 1807 when solving the partial differen-
tial equation that describes the temperature on a rectangular plate. * When the plate
:1 Solution For u(x, y) = y2 -~, uxCx, y) = -2x, Uy(x, y) = 2y, uxx(x, y) = -2, reaches its equilibrium point (independent of time), Fourier determined that the tem-
and uyy(x, y) = 2, so that we have Uxx + uyy = (- 2) + 2 = O. Similarly, for u(x, y) ~ perature vex, y) at the point with coordinates (x, y) satisfies the equation
eY sin x, we have Ux = eY cos x, uy = e Y cos x, Uxx = -eY sin x, and uyy = eY sin x.
Therefore, Uxx + uyy = (-eY sin x) + eY sin x = 0, so the equation is satisfied by '" Jesper Lutzen, "The Solution of Partial Differential Equations by Separation of Variables: A Historical
both functions. Survey," Studies in the History of Mathematics, :MAA Studies in Mathematics, Volume 26, Mathematical
Association of America (1987). pp. 242-277.
552 Chapter 10 Partial Differential Equations 10.1 Introduction to Partial Differential Equations and Separation of Variables 553
av a
2
aX2 +
2
ay2 =
V
o. "f'"I,'.'
Use separation of variables to find a solution of xUx = uy
Fourier assumed that the temperature vex, y) could be written as the product vex, y) '"
X(x)Y(y). Substituting vex, y) into the equation results in
:1 Solution If u(x, y) = X(x)Y(y), then Ux = X'Yand uy = XY~ The equation then
a2v a2v
-:;--0 + -:;z = X'(x)Y(y) + X(x)Y'(y) = 0,
becomes
ox' dy xX'Y=XY',
where X"= d 2X/d:c 2 and Y"= d 2Y/dy2, so that which can be written as the separated equation
X"(x) Y"(y) xX' Y'

X(x) = - Y(y) . X=y'
The only way that X'(x)/X(x) = - Y'(y)/Y(y) for all values of x and all values of y is Notice that the left-hand side of the equation is a function of x and the right-hand
for X"(x)/X(x) and - Y"(y)/Y(y) to equal the same constant, say side is a function of y. Hence the only way that this situation can be true is for XX'lX
and Y'/Y to both be constant. Therefore,
xX' Y'
X=y=k,
This assumption results in the two ordinary differential equations so we obtain the ordinary differential equations xX' - kX = 0 and Y' - kY = O. We
X"(x) - ,\.2X(X) = 0 and Y"(y) + ,\.2y(y) = o. find X in the manner that follows.
xX' - kX= 0
Each of these equations can be solved for X and Y (using the boundary and initial con-
ditions based on the physical situation), so that the product vex, y) = X(x)Y(y) can be xdX=kX
formed, which results in a solution of the partial differential equation. d:c
In general, the goal of the method of separation of variables is to transform the
partial differential equation into a system of ordinary differential equations, each of dX=5..d:c
which depends on only one of the functions in the product form of the solution. Sup-
X x
pose that the function u(x, y) is a solution of a partial differential equation in the in- InlXl = k lnlxl + c,
dependent variables x and y. In separating variables, we assume that u can be written
X(x) = ec'x k = C,x k
as the product of a function of x and a function of y. Hence
Similarly, we find
u(x, y) = X(x)Y(y),
Y' - kY= 0
and we substitute this product into the partial differential equation to determine X(x)
and Y(y). Of course, to substitute into the differential equation we must be able to dif- Y'=kY
ferentiate this product. However, this is accomplished by following the differentiation
dY =kY
rules of multivariable calculus. For example, using the following short-hand notation dy
for convenience, we have
dY = kdy
and U yy =XY': Y
+ C2
where X' represents ~ X(x) and Y' represents ~ Y(y). After these substitutions are
InlYI = Icy
Y(y) = eC'eky = C2 eky.
made, if the equation is separable we can obtain an ordinary differential equation for
X and an ordinary differential equation for Y. These two equations are then solved to Therefore, a solution is u(x' y) = X(x)Y(y) = (C,x k )(C2eky) = C3x k eky,
where k
findX(x) and Y(y). and C3 = C, C2 are arbitrary constants. (Notice that the first equation is a Cauchy-
554 Chapter 10 Partial Differential Equations 10_1 Introduction to Partial Differential Equations and Separation of Variables 555
Euler equation, so we could have used the techniques covered in Section 4.7 to where r is the constant of proportionality. By taking the limit as Ax ~ 0, the approx-
solve it. Similarly, we could have used a characteristic equation to solve the second imation becomes better and better. Also because x ~ x + Ax, the value of x = x used
equation.) in the approximation obtained with the MVT is more accurate. Hence the flow of heat
through the wire is modeled by the partial differential equation
Modeling with Partial Differential Equations

Consider a wire (or thin rod) of length p, which is made of a material that conducts where c2 = rid 8 is called the thermal diffusivity of the wire. We call this the one-
-fv....l___~(J--x heat. Assuming that the wire is insulated so that no heat is lost or gained through the dimensional heat equation, because there is only one dimension in space, the posi-
x;O x;p
I sides of the wire and that the thermal conductivitY and specific heat from which the tion x along the wire. We solve this equation through separation of variables in the next
wire is made are constant, our goal is to determine the temperature u(x, t) that repre- section.
Figure 10.1 Wire oflength p
sents the temperature in the wire x units from x = 0 at time t. (See Figure 10.1).
We develop a mathematical interpretation by considering the heat flow through
a small cross-section of the wire, as shown in Figure 10.2. To model the heat flow
through the cross-section, we must make several simplifying assumptions.
1. The wire has uniform density (i.e., it h.as a uniform cross-section).
EXERCISES 10.1
Figure 10.2 Small cross· 2. The heat flow across a small surface is directly proportional to the product of the
section of wire rate of change in the temperature across the surface and the area of the surface.
3. The temperature in an object is proportional to the quotient of the flow of heat into
the obj ect and its mass.
Use separation of variables to find a solution of the following In Exercises 17-22, show that the function satisfies the wave
(Note: Simplifying assumptions 2 and 3 are basic principles of heat conduction.) If
u(x, t) represents the temperature, then heat flow (the rate at which u changes with re-
l PartiaI1~i:::n:~y equations.
equation U;cx = Uri.
17. u(x, I) = (x' + 1)2
spect to x) is given by u'" Then if the area ofthe face of the cross-section is A, the heat
I
2. Ux = uy 18. u(x. I) = 4xl
flow into the cross-section is proportional to - uxCx, t)A. Similarly, the heat flow out of
the cross-section is proportional to '::'uxCx +:Ax, t)A. (Notice that the negative signs in *3. Ux - YUy = 0 *19. u(x, I) = sin 2x cos 21
these relationships are due to the fact that heat flows in the direction of decreasing tem- 4.xux -Yuy =0 20. u(x, I) = (cos I + sin I) sin x
perature.) Hence if we let H represent the amount of heat flow from left to right through 5. Ux - uy = u 21. u(x, I) = sin kx cos kt
the cross-section, then the net flow through the cross-section is 6.u;x+u y =u 22. uex, I) = sin = sin wi
All = k[uxCx + Ax, t)A - uxCx, t)A] = kA[ux(x + Ax, t) - ux(x, t)] *7. u yx = 0
In Exercises 23 -28, show that the function satisfies the heat
8.uyx-u=0 equation U;c.( = UfO
By the Mean Value Theorem (MVT), there is a value x = x on the interval (x, x + 6.x)
such that 9. xu"" +u= 0 23. u(x, I) = e-' sin x
uxx(x, t) =
uxCx + Ax, t) - uxCx, t)
Ax I 10. u"" = u - u,.
24. u(x, I) = e -4, sin 2x
; In Exercises 11-16, show that the function satisfies Laplace's

*25. u(x, I) = lOx + 1 + e -16, cos 4x
Therefore, ~ equation u'" + u"" = O. 26. u(x, t) = 100 + e -25, cos 5x
MI = kAuxxex, t)Ax. 1 11. u(x, y) = x 2 _ y2 27. u(x, I) = e -k'r cos kx
= 21 + x 2
If the density of the wire is 8, then the mass of the wire is 8A Ax. Then accord- I 12. u(x, y) = 3xy2 - x 3 28. u(x, t)
29. Detennine the value(s) of k so that u(x, t) =
ing to the third simplifying assumption, where the rate at which the temperature changes I *13. u(x, y) = e r siny
e-k'r cos x satisfies the heat equation u= = 16u,
. . ( r All 14. u(x, y) = e-
2y
cos 2x
IS gIVen by utCx, t), we have u, x, t) = SA Ax' so 30. Determine the value(s) of k so that u(x, t) =
15. u(x, y) = tan- I I. e -2, sin kx satisfies the heat equation u"" = ~ur.
rkAuxxex, t)Lll rk _ x
*31. Determine the value of c so that u(x, t) =
u,(x, t) = 8A Ax = B uxx(x, t), 16. u(x, y) = cos x sinh y cos ct sin 4x satisfies the wave equation Uxx = 4u tr•
~---------------------------------------------------
556 Chapter 10 Partial Differential Equations 10.2 The One-Dimensional Heat Equation 557
32. Determine the value of c so that u(x, I) = sin 31 sin ex 40. (a) Show that u(x, I) = e -4' sin 2:c satisfies u"" = u, (b) Substitute the result obtained in (a) into axlar = for r and s and substituting the resulting values
satisfies the wave equation Uxx ;:::: 9urt• (b) If 0 :5 x :5 7T, then what is the maximum vaIue of - 3xl, x(O, s) = s and solve for x. into the result obtained in (c).
33. Does the solution u(x,y) = eX siny of Laplace's equa- u if 1= 1,2, and 3? (c) Substitute the results obtained in (a) and (b) into (e) Use the fact that the initial condition u(x, 0) = x
aular = xl, u(O, s) = s and solve for u.
tion U xx + uyy = 0 satisfy the boundary conditions
u(x, 0) = u(x, 7T) = O?
( e) If u represents the temperature in a thin wire at
position x and time I, what is the initial tempera_
ture of the wire at x = 7T/2?
(d) Solve for u(x, I) by first solving the equation has parametrization {: =~
u=s
to graph the solution
34. Does the solution u(x, y) = e -2y cos 2:c of Laplace's I = r for0:5s:515.
equation u= + uyy = 0 satisfy the boundary condi- 41. (a) Show that u(x, I) = 100 + e- 16' sin 4x satisfies { x = se- 3r/2
tions u( -7T/4, y) = u( 7T/4, y) = O? Uxx = Uc-
*35. Does the solution u(x, I) = sin 2:c cos 21 of the (b) If 0 :5 x :5 7T, then what is the maximum value of
wave equation u= = u" satisfy the initial condition u if 1= 1,2, and 3?
u,(x, 0) = O? (c) Does the value ofu eventually become independ_
36. If u(x, I) = e -4' sin 2:c represents the temperature at
time I and position x of wire of length 7T, then find
ent of I? If so, if u represents temperature in a thin
wire at position x and time I, what is the eventual
110.21 The One-Dimensional Heat Equation
the temperature at the endpoints x = 0 and x = 7T (for temperature at every point in the wire? (' The Heat Equation with Homogeneous Boundary Conditions
all values of I). (' Nonhomogeneous Boundary Conditions (' Insulated Boundary
42. A partial differential equation of the form
37. Use separation of variables to show that the Schre-
dinger equation a(x, y, u)
au
ax + b(x, y, u)
au
ay = c(x, y, u) One of the more important differential equations is the heat equation,
2
1 -
- a (r-a'l')
- + -- a (.SIn
I -- a'l')
e-
2
r ar ar r2 sin e ae ae is called a first-order, quasi-linear partial differen-
In one spatial dimension, the solution of the heat equation represents the temperature
tial equation. In the case when c(x, y, u) = 0, the
1 2
+ - 2- ' - 2 - 2
a 'l' + ----,
-a 2J.L
(E - V)'l' = 0 equation is homogeneous; when a and b are inde- (at any position x and any time t) in a thin rod or wire of length p. Because the rate at
r sm e ¢ h pendent of u, the equation is almost linear; and when which heat flows through the rod depends on the material that makes up the rod, the
2
c(x, y, u) can be written in the form constant c , which is related to the thermal diffusivity of the material, is included in
can be written as the three second-order ordinary dif-
the heat equation. Several different situations can be considered when determining the
ferential equations (see Section 6.4): c(x, y, u) = d(x, y)u + sex, y) temperature in the rod. The ends of the wire can be held at a constant temperature, the
d 2 CP _ 2
ends may be insulated, or there can be a combination of these situations.
Azimuthal equation d¢2 - -me g the equation is linear.
Quasi-linear partial differential equations can
Radial equation frequently be solved using the Method of character- The Heat Equation with Homogeneous Boundary
~~(r2
r2 dr
dR)
dr
+ 2J.L2 (E _ V _ ~ e(e +
h 2J.L r2
1») R = 0 istics. In addition, many mathematical software pack-
ages are capahle of graphing solutions to quasi-linear
Conditions
partial differential equations. The ftrst problem that we investigate is the situation in which the temperature at the
Angular equation
For example, to solve the equation ends of the rod is constantly kept at zero and the initial temperature distribution in the
1
sin
d
e de ( .
sm
dE»
ede [ m/ ]
+ e(e + 1) - sin2 e E> = O. rod is represented as the given function/(x). Hence the fixed-end zero temperature is
- 3xtux + u, = xl, u(x, 0) = x
given in the boundary conditions
38. Become familiar with a three-dimensional graphing using the Method of characteristics, we first write the
u(O, t) = u(p, I) = 0,
device by plotting the solution obtained in Example 2 characteristic system:
using various values for the arbitrary constants. and the initial temperature distribution is given in
39. (a) Show that u(x, I) = sin(x + I) satisfies the partial ~~ = -3xl, x(O, s) = s u(x, 0) = lex).
differential equation Un = U u-
al = I, 1(0, s) = 0
ar Because the temperature is zero at the endpoints, we say that the problem has homo~
(b) If 0 :5 x :5 7T, what is the maximum value of u if
geneous boundary conditions (which are important in finding a solution with sepa-
1= 1,2, and3?
(e) Ifu represents the displacement ofa string at po-
sition x and time I, what is the initial displacement
1auar = xl,
(a) Solve al/ar = 1,1(0, s) = 0

u(O, s) = s ration of variables). We call problems of this type initial boundary value problems
(IBVP), because they include initial as well as boundary conditions. We summarize
the problem as follows.
of the string at x = 7T/4?
2
Ut = C U"", 0 < X < p, I> 0
un(x, I) = Xn(x)TnC/) = bn sin ~
p
e-c'At,
U(O, I) = 0, U(p, I) = 0, I> O.
[
U(X, 0) = f(x), 0 < x <p where we have replaced the constant A by one that depends on n. To find the value of
We solve this problem through separation of variables by assuming that b m we must apply the initial condition u(x, 0) = f(x). Because
u(x, I) = X(x)T(t). un(x, 0) = bn sin n7TX e -c'A(O) = bn sin n7TX

p p
Substitution into the differential equation yields
is satisfied only by functions of the form sin(7TX!p), sin(27TXp), sin(37TX!p), ... (which
T' X" in general is not the case), we use the Principle of superposition to state that
c 2 T=X=-A,
~ ~
where - A is the separation constant. (Note that we selected this constant to obtain an u(x, I) = I un(x, I) = I bn sin ~ e -?At
eigenvalue problem that was solved in Section 9.1.) Separating the variables, we have n=I n=l P
the two equations . is also a solution of the problem. (Note that this solution satisfies the heat equation as
well as the boundary conditions.) Then when we apply the initial condition u(x, 0) =
T' + c 2 AT= 0 and X"+ AX= O. f(x), we find that
Now that we have successfully separatecfthe variables, we turn our attention to the ho-
mogeneous boundary conditions. In terms of the functions X(x) and T(/), these bound-
ary conditions become
u(x, 0) = i:
n~1
bn sin !!!E... e -?AeO) =
P
i:
n~1
b n sin ~ = f(x).
P
u(O, I) = X(O)T(t) = 0 and u(p, I) = X(p )T(t) = O. Therefore, bn represents the Fourier sine series coefficients for f(x), which are given by
In each case, we must avoid setting T(t) = 0 for all I, because if this were the case, our
solution would be the trivial solution u(x, I) = X(x)T(t) = O. Therefore, we have the
bn =l1 f()
Pop
. ~ d:x, n - 1, 2, ...
P
X Sln
boundary conditions
X(O) = 0 and X(P) = 0,
liii".".,
so we solve the eigenvalue probleIll
Ut = u"'" 0 < x < 1, t > 0
X"+ AX= 0, 0 <x O.
{ X(O) {
= 0, X(P) = 0 . u(x, 0) = 50, 0 < x < 1
Because we solved this problem in Section 9.1, we refer to the results that we found
there. The eigenvalues of this problem are : 1 Solution In this case, c = 1, p = 1, andf(x) = 50. Hence
A = An = (~'Trt n = 1,2, ... u(x, t) = I

n=l
b n sin n7TXe- lV ,
with corresponding eigenfunctions where
X(x) = xncx) = sin~, n = 1, 2, ... 211 50 Sln

bn = -1 . n7TX d:x = -100 [cos
-- n7TXJ1 100
- = -- - (cos n'Tr - 1)
p o n7T a n7T
Similarly, we solve T' + ? AT = 0, which is a first-order equation with characteristic = -~ [(-1)" - 1], n = 1,2, ...
equation r + c2 A = O. Because r = -c2 A, a general solution is n'Tr
T(/) = TnC/) = Ae -c' lV, and A = An = (n'Tr)2. Therefore, because bn = 0 if n is even, we write u(x, t) as
where A is an arbitrary constant and A = An = (n'Tr/p)2, n = 1, 2, .... Because X(x)
and T(t) both depend on n, the solution u(x, I) = X(x)T(t) does as well. Hence,
u(x, t) = i:
n~1
200
(2n - 1)'Tr
sin(2n - 1)7TXe-e2n-1)'".'t.
In Figure 10.3, we sketch the graph of u at various times. Notice what happens to we can choose the boundary conditions for S to be the nonhomogeneous conditions
the temperature as t increases.
S(O) = To and
In Example 1, what does the temperature at each point in the wire eventually be- and the boundary conditions for vex, t) to be the homogeneous conditions
come because of the zero fixed-end boundary conditions?
v(O, t) = 0 and v(p, t) = O.
Nonhomogeneous Boundary Conditions Of course, we have failed to include the initial temperature. Applying this condition,
y
we have u(x, 0) = vex, 0) + Sex) = fex), so the initial condition for v is
1=0 The ability to apply the method of separation of variables depends on the presence of
50 1= 0.025
homogeneous boundary conditions, as we saw in the previous problem. However, with vex, 0) = fex) - Sex).
4°tf~
the heat equation, the temperature at the endpoints may not be held constantly at zero.
30 ___----- Instead, consider the case when the temperature at the left-hand endpoint is To and at Therefore, we have two problems, one for v with homogeneous boundary conditions
:::.:'.
,/ t=O.1 ---....., and one for S that has nonhomogeneous boundary conditions. These two problems are
the right-hand endpoint it is T]. Mathematically, we state these nonhomogeneous
~: ./
boundary conditions as stated as follows.
u(O, t) = To ~d u(p, t) = Th V, = c2 v"", 0 < X < p, t > 0

S"= 0, 0 < x 0
so we would be faced with solving the problem [
Figure 10.3 Graph of u at vex, 0) = f(x) - Sex), 0 < x < p.
2
various times U' = c uxx , 0 < X < p, t> 0
Because S is needed to determine v, we begin by finding S. A general solution of
u(O, t) = To, u(p, t) = Th t> O.
[ S" = 0 is Sex) = c] + C2X. Then S(O) = c] = To and S(p) = To + C2P = Tlo so C2 =
u(x, 0) = f(x), 0 < x < P (T] - To)/p. Hence
In this case, we must modify the problem to introduce homogeneous boundary condi-
tions. We do this by using the physical observance that as t -7 00, the temperature in Sex) = To + (T] ; To )x.
the wire does not depend on t. Hence'
We are now able to fmd vex, t) by solving the heat equation with homogeneous bound-
lim u(x, t) = Sex), ary conditions for v. Because we solved this problem at the beginning of this section,
HOO
we do not need to go through the separation of variables procedure. Instead, we use

where we call Sex) the steady-state temperature. Therefore, we let the formula that we derived there using the initial temperature vex, 0) = f(x) - Sex).
u(x, t) = vex, t) + Sex), Therefore,
where vex, t) is called the transient temperature. We use these two functions to
obtain two problems that we can solve. To substitute u(x, t) into the heat equation
vex, t) = I
n=1
b
n
. n7T.X -?Ai
smpe ,
u, = c2 uxx , we calculate the derivatives
where
u,ex, t) = v,(x, t) +0 and uxx(x, t) = vxx(x, t) + S"(x).
Hence substitution into the heat equation yields
vex, 0) =I Cn sin n7T.X = f(x) - Sex).
n=] P
2
u, = c uxx This means that bn represents the Fourier sine series coefficients for the functionf(x)
Vt = c2 vxx + C 2S", - Sex) given by
bn = -2 JP (f(x) - Sex)) sm
. -n7T.X dx, n = 1, 2, ...
so we have the two equations v, = c2 v"" and S" = O. We then consider the boundary
conditions. Because PoP
u(O, t) = yeO, t) + S(O) = To and u(p, t) = yep, t) + S(p) = Tlo The solution to the original problem is, therefore, u(x, t) = vex, t) + Sex).
" I
c2uxx , 0 < X < p, t > 0

'if'"I"., UI =
uxCO, t) = 0, uxCP, t) = 0, t> 0
u/ = uxx , 0 < X < I, t> 0 [
u(x, 0) = f(x), 0 < x < p.
Solve u(O, t) = 10, u(1, t) = 60, t> O.
{ u(x, 0) = 10, 0 < x < I
Notice that the boundary conditions are homogeneous, so we can use separation of
variables to find u(x, t) = X(x)T(t) (i.e., there is no need to introduce the steady-state
: I Solution In this case, c = I,p = I, To = 10, Tl = 60, andf(x) = 10. Therefore and transient temperatures when the boundary conditions are homogeneous). By fol-
the steady-state solution is ' lowing the steps taken in the solution of the problem with homogeneous boundary con-
ditions, we obtain the ordinary differential equations
Sex) = To + (Tl ~ To)x= 10 + (60 ~ 10)x = 10 + 50x. and X"+ AX= O.
Then the initial transient temperature is However, when we consider the boundary conditions
vex, 0) = 10 - (10 + 50x) = -50x, uxCO, t) = X'(O)T(t) = 0 and ux(p, t) = X'(p)T(t) = 0,

we wish to avoid letting T(t) = 0 for all t (which leads to the trivial solution), so we
so that the series coefficients in the solution vex, t) =
.-
I
n=1
bn sin n7TXe- n'."z/ are have the homogeneous boundary conditions
2 {l {l 100 cos n1T X'(O) = 0 and X'(p) = O.

bn = T 10 (- 50x) sin n7TX dx = -100 10 x sin n7TX dx = n1T
Therefore, we solve the problem
100( -1)"
n1T ' n = 1,2, ... , X" + AX = 0, 0 < x <P
{ X(O) = 0, X(p) = 0
so the transient temperature is
to find X(x). This problem was also solved in Section 9.1, so we will not repeat the
vex, t) =In=1
bn sin n7TX e-c'A1
P . n=l
=I _1_O_O-'-(----'I)'--n Sin(n7TX)e- n'''''''.
n1T
steps at this time. We simply recall that the eigenvalues and corresponding eigenfunc-
tions of this problem are
Therefore, 0, n= 0 1, n = 0
10+-~---+--4---+-----
0.2 0.4 0.6 0.8 x
u(x, t) = vex, t) + Sex) = I ~
n=1
100(-1)" .
n1T
, ,
sm (n7TX)e- n .". ,+ 10 + 50x.
An =
[ (p),
n1T 2
n = 1,2, ...
and Xix) = n7TX
[ cos P ,
_?
n - 1, . . . , ...
.
Figure 10.4 Graph of u at We graph u(x, t) for several values of t in Figure 10.4. Notice what happens to the Next, we solve the equation T' + c2 AT = 0 for the An given previously. First, for A =
various times temperature as t increases. Ao = 0, we have the equation T' = 0, which has the solution T(t) = Ao, where Ao is a
constant. Therefore, when A = Ao = 0, the solution is the product
In Example 2, find E.~ u(x, t) both from the formula and from Figure 10.4. uo(x, t) = Xo(x)To(t) = Ao.
Then for A = An = n7TX/p, we solve T' + c 2 AT = 0, which has as a general solution
Insulated Boundary
T(t) = Tit) = ane -c'At. For these eigenvalues, we have the solution
Another important situation concerning the flow of heat flow in a wire involves insu-
lated ends. In this case, heat is not allowed to escape from the ends of the wire. Math- un(x, t) = Xix)Tn(t) = an cos n7TX e-c'Ar.
p
ematically, we express these boundary conditions as
Therefore, by the Principle of superposition, the solution is
uxCO, t) = 0 and uxCP, t) = 0,
because the rate at which the heat changes along the x-axis at the endpoints x = 0 and
x = p is zero. Therefore, if we want to determine the temperature in a wire of length
p with insulated ends, we solve the initial boundary value problem Application of the initial temperature yields
EXERCISES 10.2
u(x, 0) = Ao + L~ an cos -
nm:
=J(x),
n=1 p
which is the Fourier cosine series for J(x), where the coefficient Ao is equivalent to
arJ2 in the original Fourier series given in Section 9.2. Therefore, In Exercises 1-6, solve the following problems (heat equation 11. {Uxx=u" O<x<l,t>O
with homogeneous boundary conditions). u(O, I) - ux(O, t) = To, u(l, t) = 0, I>°
Ao = tao = t; (J(x) dx = ~ (J(X) dx and
1.
uX> = u" 0< x < I, t>
u(O, t) = 0, u(l, t) = 0, t >
°° 12. {Uxx = °
u" 0< x < I, t >
u(O, t) = To, u(l, I) + ux(l, I) = h t> °
an = -2 i" nm:
J(x) cos - - dx, n = 1, 2, ....
{
u(x, 0) = 100x, 0< x 
PoP °u(l,< < I, t> ° u(O, t) To, ux(l, t) 0,
= = I>
°< xt >< °
2UX> = u" x
2. u(O, t) = 0, t) = 0,
t) Tlo t> °
{ 14. {Uxx-u=u" O<x<I,I>O
u(x, 0) = x(1 - x), I u(O, To, I) = u(l, =
"fi"i,'. t U:c;r = Ul , 0 ~X < 1, t >0 15. {Uxx - (u - T) u" °< < t> °
u(O, = To, u(l, = h t> °
= x I,
*3. ( u(O, t) = 0, u(l, t) = 0, t> ° I) I)
U, = uxx , 0 < X < 7T, 1 > 0
Solve uxCO, I) = 0, uxC 7T, I) = 0, 1 >'0.
u(x, 0) = x(1 - x), <xO
{
u(x, 0) = X, 0 < X < 7T
4.
uxx = u" °
< x < 7T, 1 >
u(O, I) = 0, u( 7T, I) = 0, 1 >
°° u(O, To, u(l, t) To,
I) = = I>
{
u(x, 0) = sin 2x, < x < 7T ° In Exercises 17-20, solve the problems with insulated ends.
: I Solution In this case, p = 7T and c = 1. Because we need the Fourier cosine se-
ries coefficients for J(x) = x, we refer back to Example 3 in Section 9.2. There, we °
= u" < x < 7T, 1 >
UXX °° 17.
u, = u"'" °
< x < 7T, t >
uxCO, t) = 0, UxC7T, t) = 0, I>
° °
5. u(O, t) = 0, u( 7T, t) = 0, t > {
found that {
u(x, 0) = sin 2x + sin 8x, <x < ° 7T
u(x, 0) = I, O<X<7T
U XX= u" °u(2,< x t)< > t°> °

2, t
18.
u, = u"'" O<x<7T,I>O
°
uxCO, I) = 0, ux ( 7T, I) = 0, I>
6. u(O, t) = 0, = 0, {
u(x, 0) = cos 2x - 4 cos 3x, ° <x<
t= 0.025
t=0.05
and with integration by parts
(
u(x,
0)
=
{X,O:SX ° °
7T, 1
7T
1.8 *19. uxCO, I) = 0,

1.7 an 2
= -'lToi'" X cos -nm: dx = -7rn22 [(-1)n - 1], n = 1, 2, ... In Exercises 7-10, solve the problem with nonhomogeneous { 2
7T, I) = 0,
u(x, 0) = 4 sin x, x
I>
°< <
7T
1.6
1.5
f--__--,f-_..:.t.::=-"O"'.4_ 7T boundary conditions.
u= = u" °
< x < 1, t> °° 20.
U, = u"'"
ux(O, I) = 0,
°u<x( < > °> °
7T, I)
x 7T,
= 0,
t
1
1.4
Therefore, the solution is
7. u(O, I) = 0, u(l, t) = 10, I>
{
°
{
u(x, 0) = x(1 - x), °< < x 7T
1.3
~O+:.5-+-j-jr.- +---::-f-.s--<3--::
5 2 2
u(x, I) = !!. -
2
I
n=1
4 2
(2n - 1) 7T
cos«2n - 1 )x)e -(Z.-I)'".",
u(x, 0) = 0, <x °t ° In Exercises 21-23, use separation of variables to solve the
8. u(O, t) = 10, u(l, I) = 10, > initial boundary value problem. (Hinl: The eigenvalue problems
Figure 10.5 Graph of u at where we have used the fact that an = 0 if n is even. The graph of u is shown in {
u(x, 0) = 0, <x ° ° UI = U"'" °
< x < I, I>
21. u(O, I) = To, uxCI, t) = 0, I>
° °
*9.
{
u(O, t) = 20, u(l, t) = 10, t>
u(x, 0) = 0, <x< I ° {
u(x, 0) = I(x), <x° UI = U"'" 0< x < I, I> ° °
10.
u= = u"
u(O, I) = 0, u(l, t) = 10, I> ° 22. uxCo, I) = 0, u(l, I) = To,
°< x 
{
u(x, 0) = lOx, <x 0
.In Exercises 11-16, determine the steady-state solution to the
boundary-value problem.
*23.
{
u( - 7T, I) = u( 7T, I), ux ( - 7T, I) = ux ( 7T, t),
u(x, 0) = I(x), -7T < x < 7T
1> °
566 Chapter 10 Partial Differential Equations 10.3 The One-Dimensional Wave Equation 567
24. Solve the heat equation in Exercise 21 withf(x) = To.

25. Solve the heat equation in Exercise 21 with f(x) =
32. Consider the problem
110.31 The One-Dimensional Wave Equation
Ut=C uxx
2
+e- r ,0<X<I,t>0
2To· u(O, I) = 0, u(\, I) = 0, 1> 0 , The one-dimensional wave equation is important in solving an interesting problem.
26. Solve the heat equation in Exercise 22 with f(x) = [ Suppose that we pluck a string (e.g., a guitar or violin string) of length p and constant
u(x, 0) = f(x), 0 <x< 1
x(l - x) + To· mass density that is fixed at each end. A question that we might ask is, "What is the
where the term e -r represents the loss of heat through
27. Solve the heat equation in Exercise 22 withf(x) = To. position or displacement of the string at a particular instance of time?" We answer this
radioactive decay of the wire. Use a procedure simi.
28. Solve the heat equation Exercise 23 with f(x) = To. question by modeling the physical situation with a differential equation, namely the
lar to that described in Exercise 3i to find a solution
*29. Solve the heat equation in Exercise 23 withf(x) = Ixl. of this problem.
wave equation in one spatial variable. We will not go through this derivation, because
we did it with the heat equation in Section 10.2, but we point out that it is based on
30. Sturm-Liouville theory was motivated by the Swiss
In Exercises 33-38, use technology to assist in calculating the determining the forces that act on a small segment of the string and applying Newton's
mathematician Jacques Charles Francois Sturm's
(1803-1855) study of heat conduction in an inhomo- coefficients in the solution of each of the following problems. second law of motion. The partial differential equation that is found is
geneous bar of length L with an unequally polished Also, use a graphing utility to plot an approximation of the so.
surface,· which showed that the partial differential iution using several terms in the series soiution. Observe the
equation that describes the temperature at time t at the behavior of the solution which describes the temperature in a which is called the (one-dimensional) wave equation. In this equation c2 = Tip, where
point x is thin rod of length \. T is the tension of the string and p is the constant mass of the string per unit length.
~, The solution u(x, t) represents the displacement of the string from the x-axis at time t,
g ~~ = :xH~) -iu, 33. u(O, I)
{
O<X< I, I> 0
UX;;'= Ut,
= O. u(l,
I) = 0, 1 > 0
u(x,0)=x(l-x 3 ).0<x<i
x=O
Figure 10.6 String

x-p
as shown in Figure 10.6. To determine u we must describe the boundary and initial
conditions that model the physical situation. At the ends of the string, the displacement
where g, k, and ( are functions of x. If the surround- displacement from the x-axis is fixed at zero, so we use the homogeneous boundary conditions,
ing temperature is constant, then the endpoints of the uxx= Ut, 0 <x< 1,1>0
bar satisfY 34. u(O, t) = 0, u(\, I) = 0, I> 0 lI(O, t) = 0 and u(p, t) = 0 for t:2: O.
{
u(x. 0) = eX, 0 < x 0 point of the string at t = O. If the initial displacement is given by f(x) and the initial
1 au
k ax + Hu = 0
for some positive constants hand H.

when x = L
*35. u(O, I) = 50, u(l, I) = 100, I> 0
{
. u(x, 0) = x(i - x 3 ), 0 < X 0
UXX
velocity by g(x), then we have the initial conditions,
u(x, 0) = fC;<) and U,(x, 0) = g(x) for o sxsp.
36. u(O, I) = 10, u(\, I) = 100, I> 0 Therefore, we determine the displacement of the string with the initial boundary value
(a) Assume that u(x, t) = V(x)e -r< and u(x, 0) = f(x). {
u(x, 0) = e- r , O<x<1 problem
(b) Show that (k(x) V'(x))' + (rg(x) - (x)) Vex) = 0
= ut , 0 < X < i, I> 0
U XX 2
and C UXX = Uti, 0 < X < p, t> 0
37. ux(O, I) = 0, ur(\, I) = 0, I> 0
k(0)V'(0) - hV(O) = 0
{
u(x, 0) = er sin m:, 0 < x < 1 u(O, t) = 0, u(p, t) = 0, t:2: 0
[
{ k(L)V'(L) + HV(L) = o· = Ut, O<x< 1, I> 0
UXX
u(x, 0) = f(x), u,(x, 0) = g(x), 0s x S p
31. Consider the initial boundary value problem given by 38. ur(O, I) = 0, ux(i, I) = 0, I> 0 Notice that the wave equation requires two initial conditions, whereas the heat equa-
{
U(x, 0) = e-r sin 2m:, O<x<1 tion only needed one. This is because there is a second derivative with respect to t in
Ut = Uxx + x, 0 < x < 1, t> 0
*39. Solve the wave equation, and there is only one derivative with respect to t in the heat equa-
u(O, t) = To, u(l, I) = T" 1 > O.
[ Ut = Uxx , 0 < x < i, I> 0 tion. Note also that for consistency we must require thatf(O) = f(p) = 0 and g(O) =
u(x, 0) = f(x), 0<x0 g(p) = O.
(a) Let u(x, t) = vex, I) + Sex). Show that S satisfies [ This problem is solved through separation of variables by assuming that u(x, t)
u(x, 0) = f(x), 0 < x < 1,
S" = -x, S(O) = To, S(l) = Tl and find S. = X(x)T(t). Substitution into the wave equation yields
where the second boundary condition indicates that
(b) Find v. heat at the end, x = 1, radiates freeiy into the air of
(e) Find u iff(x) = To(l - x) - x 3/6. temperature zero.
>II Jesper Lutzen, "The Solution of Partial Differential Equations," Studies in the History of Mathematics, ed. Ester R. Phillips, MAA Studies in
Mathematics, Volume 26, Mathematical Association of America (1987), pp. 242-277. so we obtain the two second-order ordinary differential equations
X"+AX=O and T"+ e2 AT= O. Then
At this point, we solve the equation that involves the homogeneous boundary con- u,(x, 0) = L~ bn
en7T.
- - SID - -
n7TX
= g(x),
ditions. As was the case with the heat equation, the boundary conditions in tenus of n~' p P
u(x, t) = X(x)T(t) are u(O, t) = X(O)T(t) = 0 and u(p, t) = X(p)T(t) = 0, so we have
so b n en7T
p represents the Fourier sine series coefficient for g(x), which means that
X(O) =0 and X(p) = O.
Therefore, we detenuine X(x) by solving the eigenvalue problem
b = -p- -
2 fP n7TX . 2
g(x) SID - - dx = - -
f" g(x) SID. -n7TX- dx, n = 1,2, ....
n en7T P o P en7T 0 p
X" + AX= 0, 0 <x <p,
{X(O) = 0, X(p) = 0
which we encountered when solving the heat equation and solved in Section 9.1. The
eigenvalues of this problem are "f'"I".' UXX = u", 0 < x < I, t> 0
A = An = (np7TY, n = 1,2, ... , Solve u(O, t) = 0, u(l, t) = 0, t> 0
{
u(x, 0) = x(1 - x), u,(x, 0) = 0,
with corresponding eigenfunctions
X(x) = Xn(x) = sin n7TX, n = 1, 2, .... : ") Solution In this case, e = P = I, 1(X) = x(1 - x), and g(x) = O. From this in-
p formation, we compute
Next we solve the equation T" + e 2 AT = 0 using the eigenvalues given previously. From
our experience with second-order equations, a general solution is an = -2
10
f'
x(l - x) . n7TX dx
SID = -~
n7r
4 cos n7T + 3
4
n7r
3
Tn(t) = an cos eVA;.t + bn sin eVA;.t = an cos enm + bn sin enm,

= n3~ (l - (-In, n = 1,2, ... , .
p p
where the coefficients an and b n musfbe detennined. Putting this information together,
where integration by parts must be used twice. (Note: Of course, a table of integrals
we obtain
or computer algebra system could be used as well.) Because g(x) = 0, the coeffi-
enm . enm) . n7TX cients bn = 0 for all n. Using the fact that an = 0 when n is even and an = 8/(n 3 ,,;»
unCx, t) = ( an cos - - + b n SID - - SID - - ,
P P P if n is odd, then the solution is
so by the Principle of superposition, we have 8
u(x, t) = 2:
00
3,,;> cos(2n - l)m sin(2n - 1)7TX.

u(x, t) = ~ ( an cos -
L enm . -
- + bn SID enm) . n7TX
- SID - - .
n~' (2n - I)
n~' p p P The graph of u(x, t) for various values of t is shown in Figure 10.7.
Applying the initial displacement yields
u(x, 0) = I
n~'
an sin n7TX = I(x),
p
l'f',,·,I.,
so an is the Fourier sine series coefficient for I(x), which is given by UXX = Uti, 0 < X < I, t> 0
Solve u(O, t) = 0, u(l, t) = 0, t 2: 0
an = -
2 fP
I(x) SID -
n7TX .
dx, n = 1, 2, .... {
u(x, 0) = 0, u,(x, 0) = sin 27TX, 0:5 x :5 I
PoP
To determine b", we must use the initial velocity. Therefore, we compute : 1 Solution Here the string is initially horizontal and an initial velocity that depends
. on the position x is given. With I(x) = 0, we know that an = 0 for all n. Then
u(x t) = 2:
00
-a -en7T.
( enm
-SID-- +b en7T enm). n7TX
- - c o s - - SID--. using the initial velocity function g(x) = sin 27TX, we note that
" n~' n p p n p p p
u
0.3 0.3 0.2 0.2
0.3
0.15 0.15
0.2 0.2 0.1 0.1
0.1 0.05 0.05
-/-----r-----.---r--1r--'I--
-0.1 0.2 0.4 0.6 0.8 1 x -0.1 0.2 0.4 0.6 0.8 -0.1 0.2 0.4 0.6 0.8 1 x -0.05 0.2 0.4 0.6 0.8 1 x _0.05-f--+-M-+---+----,.f- 1 x
-0.2 -0.1 -0.1
-0.2 -0.2 -0.15 -0.15
-0.3 -0.3 -0.3 -0.2 -0.2
u
0.3 0.2 0.2 0.2
0.3
l
0.15 0.15 0.15
0.2 0.2 0.2 0.1 0.1 0.1
0.1 0.1
I I I I I
0.1
.j 0.05 0.05 0.05
-0.1 0.2 0.4 0.6 0.8 1 x -0.1 0.2 0.4 0.6 0.8 1 x -0.1 1 x l -0.05 1
x
-0.05 0.2 0.4 0.6 0.8 1 x -0.05 1 x
-0.2 -0.2 -0.2
J -0.1 -0.1 -0.1
,
(1' -0.15 -0.15 -0.15
-0.3 -0.3 -0.3 -0.2 -0.2 -0.2
u u
~"
0.3 0.2 0.2 0.2
0.3 0.3 0.15 0.15 0.15
0.2 0.2 0.2 0.1 0.1 0.1
0.1 0.1 0.1 0.05 0.05 0.05
I I I I
1 x 1 x 0.2 0.4 0.6 0.8 x
-0.1 -0.05 -0.05
-0.1 -0.1
-0.2 -0.15 -0.15
-0.3 -0.2 -0.2
Figure 10.7 String displacement at various times Figure 10.8 String displacement at various times
ul~, 0) = I bn n7r sin n71X = sin 271X. as nodes and represent positions where there is no motion in the string. We graph
n=l sin = n7TX in Figure 10.9 to illustrate the nodes. Notice that the frequency Wn increases
Then without the use of integration, b 2 • 27r = 1. Therefore, b2 = 1/(27r) and as c is increased. Because c2 = Tip, the larger the tension T on the string, the higher
*
b n = 0 for n 2. This yields the solution the frequency of the motion. Therefore, a higher pitch results.
1. . D' Alembert's Solution

u (x, t) = 27r sm 2m sm 271X,
The wave equation was first derived and solved by the French physicist and mathe-
because all other terms have zero coefficient. We graph this solution in Figure 10.8 matician Jean Ie Rand d'Alemberl (1717-1783) in 1746. He hoped that he would be
for several values of t to observe the motion of the string.
able to use a similar method to solve many other problems, but this did not prove to
be the case. Improvements to his solution were subsequently made by Euler.
Describe the motion of the string at x = 1/2 in Example 2. An interesting version of the wave equation is to consider a string of infmite
length. Therefore, the boundary conditions are no longer of importance. Instead, we
For each n, the function un(x, t), called the nth normal mode (or standing wave) of the
simply work with the wave equation with the initial displacement and velocity func-
string, represents harmonic motion with frequency
tions. To solve the problem
An cn7rip cn
Wn = 27T = ~ = 2p' c2 uxx = U'" - 00 < x < 00, t > 0
{ u(x, 0) = f(x), u,(x, 0) = g(x)
The function U,(X, t) is the fundamental mode of vibration, and the frequency W, is
called the fundamental frequency. The values of x such that sin(n7Txlp) = 0 are known we use the change of variables
J
10.3 The One-Dimensional Wave Equation 573
572 Chapter 10 Partial Differential Equations
y y where F is an anti derivative off and G is an arbitrary function of s. Returning to our

original variables gives us
0.5 0.5
U(x, t) = F(x + ct) + G(x - ct).
1 x
-0.5 -0.5 The functions F and G are determined by the initial conditions, which indicate that
-1 -1 U(x, 0) = F(x) + G(x) = f(x)

ut(;<, 0) = cF'(x) - cG'(x) = g(x).
1
0.5 0.5
We can rewrite the second equation by integrating to obtain
x F'(x) - G'(x) = g(x)

-0.5 -0.5 c
-1
n=3
-1
n=4 F(x) - G(x) = -1 LX g(v) dv.
c 0
Figure 10.9 Graphs of y = sin mTX for n =..t, 2, 3, 4 Therefore, we solve the system
F(x) + G(x) = f(x)
F(x) - G(x) = -I LX g(v) dv.

r= x + ct c 0
s=x-ct. Adding these equations yields

Using the Chain rule, we compute the derivatives U= and u" in terms of the variables
rand s. F(x) = t~(X) + ~ I: g(v) dV]
and subtracting gives us

U= = (u r + us)rrx + (u r + us)ssx =
Ut = urrt + UsS t = CUr - CUs = c(ur - Us)
Urr + 2u rs + Uss
G(;<) = t~(X) - ~ I: g(v) dV]'
Therefore,
U" = c[(u r - us)rrt + (u r - us)ss,] = c2[u rr - 2urs + uss ]·
Substitution into the wave equation yields F(x + ct) =
Ir
"2f(;<
I (x+ct
+ ct) + -; Jo g(v) dv
]
and
1 r I (X-ct
G(x - ct) = "2f(x - ct) - -; J
]
g(v) dv ,
c2[Urr + 2urs + uss ] = c 2[Urr - 2urs + uss] o
4C2Urs = 0 so the solution is
U rs = 0
u(x, t) =
1
"2
rf(x + ct) + -;1 J(x+ct g(v) dv] +"2I rf(x - 1 (X-ct
ct) - -; J g(v) dv
]
o o
The partial differential equation U rs = 0 can be solved by first integrating with respect
to s to obtain 1 1 lx+ct
= -2 [f(x + ct) + f(;< - ct)] + -2
C x-ct
g(v) dv.
Ur = f(r),
where f(r) is an arbitrary function of r. Then integrating with respect to r, we have 1 r+ ct g(v) dv obtained in D'Alembert's solution?
How is the term -2
c Jx - ct
u(r, s) = F(r) + G(s),
574 Chapter 10 Partial Differential Equations 10.3 The One·Dimensional Wave Equation 575
o '.fi,,·,',..
y Solve {U xx = u'" - 00 < x < 00, t> 0
u(x, 0) = 2/(1 + x 2 ), u,(x, 0) = o·
2
1.5 : I Solution Because c = 1,J(x) = 1/(1 + x 2 ), and g(x) = 0, we have the solution -20 -10 t= 13 10 -20 -10 t= 14 10 -20 -10 ,= 15 10
u(x, t) = 1.. [f(x + t) + f(x - t)] = 1.. [ 2 + 2 ] Figure 10.11 continued String displacement at various times
2 2 1 + (x + t? 1 + (x - t?
O.
1 1 D' Alembert's solution is sometimes referred to as the traveling wave solution be-
-20 -10 0 10 1 + (x + tf + 1 + (x - t)2 cause of the behavior of its graph. The waves appear to move in opposite directions
The gra~h of the initial position is shown in Figure 10.10. In Figure 10.11 we plot along the x-axis as t increases, as we can see in Figure 10.11.
Figure 10.10 Initial string
displacement the solutlOn for various times to illustrate the motion of the string of infinite length.
1.2
1.0
o "F'"I".' Uxx = u lt , -00 < X < 00, t> 0
0.8 Solve { u(x, 0) = 0, ut(x, 0) = -4xe -x".
JSolution Using c = 1 and the initial conditions,f(x) = 0 and g(x) = -4xe-x',

10 20 x -20 -10 10 20 x
we have
t= I t=2
u(x, t) = -2
I f.x+t 2
-4ue- u du = [e- U
2
];::::
..,
= e-(x+'t - e-(x-t) .
2
x-,
We graph this solution for t = 0, I, 2, ... , 5 in Figure 10.13. In this problem,
the string is initially horizontal. Notice that traveling waves are produced by the ini-
tial velocity function as they were in the previous example. However, because
10 -20 10 20 x -20 Figure 10.12 Initial velocity g(x) = -4xe-x" is negative for x < 0 (as shown in Figure 10.12), a negative dis-
t=4
function placement results to the left of x = O. In the exercises, we use D' Alembert's solu-
-10 10 -20 -10 10 0.5

t=7
j
t=8 1=9
~--~--~~~~~--
i -5 10 x -10 -5 10 x
-10 -5 10 x -10
I -0.5
-1
I
10 -20 -10 10 20 x E= 1 1=2
f= 10 t= 11 t= 12 E=O
I Figure 10.11 String displacement at various times Figure 10.13 Displacement at various times
II J
576 10.3 The One-Dimensional Wave Equation 577
Chapter 10 Partial Differential Equations
u A change of variables can be used to solve some partial differ- urr + U ss = O. How does this compare to Laplace's
ential equations in a manner similar to that used in D' Alem- equation?
bert's solution. Use the indicated change of variables to trans· *27. Show that the solution of the problem
form the equation into an equation involving r and s.
l
uxx = u", 0 < x < I, I> 0
11. uyY = U;cn r = x + y, s = x - y
u(O, I) = u(l, I) = 0, 12: 0
-10 10 x -10 10 x -10 lOx 12. Un - 2uxy - 3uyy = 0, r = x - y, s = y + 3x u(x, 0) = sin =,
u,(x, 0) = 0, 0:5 x :5 I
*13. uyx - Un = 0, r = X + Y. s = y
obtained using D' Alembert's solution is the same as
14. uyy - 2uxy + u= = 0, r = x + y, s = y
that found using separation of variables.
28, Show that the solution of the problem
The partial differential equation Au= + Buxy + CUyy + Dux +
Euy + Fu = G(x, y) is classified in the following manner:
l
u= = UtI> 0 <X< 7r, I> 0
(a) byperbolic if B2 - 4AC> 0; (b) parabolic if B2 - 4AC = u(O, I) = u(7r, I) = 0, 12:0
Figure 10.13 continued Displacement at various times
0; (c) elliptic if B2 - 4AC < O. Classify each of the following
equations as hyperbolic, parabolic, or elliptic.
u(x, 0) = 0, ut(x, 0) = sin x, 0:5 x :5 7r
tion to solve several wave equation problems involving a string with fixed-end obtained using D' Alembert's solution is the same as
15. u" - 2uX)'+ uyy + Ux = 0
boundary conditions. This is possible wpen the solution automatically satisfies these that found using separation of variables.
16. 4u" + 4uxy + uyy + 2uy = 0
boundary conditions, as we do not consider boundary conditions when deriving 29, Solve
*17. u" - 4uxy + 2uyy = 0
D'Alembert's solution.
l
18. 3u= - 4uxy - uyy + u = 0 U"" = u", 0<x< 7r, t >0
19. u= + Uxy + uyy + Ux = 0
u(O, I) = u(7r, t) = 0, t 2: 0
u(x, 0) c= ksin x, ut(x, 0) = -k sin x,
20. u= - 2uxy + 2uyy + Ux - u = 0
21. Show that Un + 2uX)' - 8u". = 0 is hyperbolic. Use using D'Alembert's solution.
the change of variables r = y - 4x, s = y + 2x to 30, Consider the string problem with an endpoint that
show that this hyperbolic equation can be written in oscillates parallel to the u-axis (vertical axis). If the
EXERCISES 10.3
equivalent form u" = O. Solve this equation to find a string has length p and c = I in the wave equation,
general solution of the equation. then an initial boundary value problem that models
22. Show that the equation u= = uyy is hyperbolic. Notice this situation is
that this is the wave equation with the variables x and
<X 0
l
Solve the following problems using the formula derived earlier. Un = UtI> 0 < X < I, t> 0 uxx = Uti> 0 7r,
y (instead of x and I) and constant c = 1. Using the
In each case, plot an approximate solution using several terms 6.
{
=
u(O, I) =
0, u(l, I) 0, I 2: 0 same change of variables as was used with D' Alem- u(O, I) = 0, u( 7r, I) = A sin I, 12:0
of the series to observe the motion of the string. u(x, 0) = x(l - x), ut(x, 0) = x(l - x), 0:5 x :5 1 u(x, 0) = 0, 0:5 x :5
bert's solution, show that the wave equation is equiv- 7r,
u= = u", O<X< I, I> 0 alent to u" = O.

Un = u", 0 < X < 7r, I > 0 where A is a small number. Assume that the solution
{
= =
1. u(O, I) 0, u(l, I) 0, I 2: 0 *7. u(O, I) = 0, u(7r, t) = 0, 12:0 23. Show that the heat equation (in the variables x andy), of this problem is u(x, I) = F(x + I) + G(x - I). Use
u(x, 0) = sin =,
ut(x, 0) = 0, 0:5 x :5 I {
u(x, 0) = sin 2x, u,(x, 0) = 0, 0:5 x :5 7r Uyy = U;n is parabolic. the boundary and initial conditions to frod u(x, I)
Un = u,,' 0 < X < I, I> 0 24. Show that the equation 4u= + 12uxy + 9uyy = 0, is within an arbitrary constant. [Hint: Try F(t) =
Un = UtI> 0 < X < 7r, I > 0
2.
{
u(O, I) = 0, u(l, I) = 0, 12: 0 parabolic. Use the change of variables r = 3x - 2y, BI sin I.]
u(x,O) = sin =, u,(x, 0) = sin 2=, 0:5 x :5 I 8. u(O, I) = 0, u(7r, I) = 0, 12:0
{ s = Y to show that this equation is equivalent to 31. Use separation of variables to solve the initial bound-
u(x, 0) = 0, u,(x, 0) = sin 4x, 0:5 x :5 7r
u= = 0 < X < I, I> 0
Uti> u~ = O. How does this equation compare to the heat
ary value problem
*3. u(O, I) = 0, u(l, I) = 0, 12:0 u= = u", 0 < X < 2, I >0 equation? Solve the equation U rr = 0 to obtain a
l
{
l
2
u(x, 0) = 0, ut(x, 0) = sin 3=, 0:5 x:5 I u(O, I) = 0, u(2, I) = 0, I 2: 0 general solution to 4u= + 12uxy + 9uyy = 0 in the C u= = Uti, 0 < X < p, I > 0
Un = UtI> 0 < X < I, I> 0 variables x and y. u.(O, I) = 0, ux(p, I) = 0, t 2: 0
9. u(x 0) = {x, 0:5 x < I
4. u(O, I) = 0, u(l, I) = 0, 12: 0 • 2-x,l::5xS2 25. Show that Laplace's equation, u"" + uyy = 0, is u(x, 0) = f(x), ut(x, 0) = g(x), 0:5 x :5 P
{
u(x, 0) = x(1 - x), ut(x, 0) = 0, 0:5 x :5 I ut(x, 0) = 0, 0 :5 x :5 2 elliptic.
Find a general formula for the unknown coefficients.
Un = u", 0 < x < I, I> 0 u= = u", O<x<7r,I>O 26. Show that Un + 2uX)' + 17u>y = 0 is elliptic. Use the
32, Solve the initial boundary value problem in Exercise
5. u(O, I) = 0, u(l, I) = 0, 12:0 10. u(O, I) = 0, u(7r, t) = 0, 12:0 change of variables r = x - y, s = 4x to show that
{ 31 if P = 7r,f(x) = 1 + cos 2x, and g(x) = o.
u(x, 0) = x(1 - x), u,(x, 0) = sin =, 0:5 x :5 I {
=
u(x, 0) =
0, u,(x, 0) sin x, 0:5 x :5 7r the equation u"" + 2u"" + 17uvy = 0 is equivalent to
10.4 Problems in Two Dimensions: Laplace's Equation 579
110.41 Problems in Two Dimensions: Laplace's Equation so substitution into Laplace's equation yields
(" Laplace's Equation (" Steady-State Temperature X"Y+XY"= 0

XI! f"
Laplace's Equation -x= --y= -A,
Laplace's equation, often called the potential equation, is given in rectangular COor- where - A is the separation constant. Therefore, we have the ordinary differential
dinates by equations
U"" + u yy = 0, X"+ AX= 0 and y"-AY=O.
and is one of the most useful partial differential equations in that it arises in many Notice that the boundary conditions along the lines x = 0 and x = a are homogeneous.
fields of study. These include fluid flows as well as electrostatic and gravitational po- In fact, because u(O, y) = X(O)Y(y) = 0 and u(a, y) = X(a)Y(y) = 0, we have X(O) =
tential. Because the potential does not depend on time, no initial condition is required, o andX(a) = O. Therefore, we first solve the eigenvalue problem
so we are faced with solving a pure boundary-value problem when working with
Laplace's equation. The boundary conditions can be stated in different forms. If the X" + AX= 0
value of the solution is given around the boundary of the region, then the boundary- {X(O) = O,X(a) = 0'
value problem is called a Dirichlet probl~m, whereas if the normal derivative of the
which was solved with a = p in Section 9.1. There we found the eigenvalues and cor-
solution is given around the boundary, the"problem is known as a Neumann problem.
responding eigenfunctions to be An = (n'TT/ai, n = I, 2, ... and ~nCx) = sin(nwa),
We now investigate the solutions to Laplace's equation in a rectangular region by, first,
n = 1,2, .... (Notice that the problem is defined on 0 < x < a mstead of 0 < x <
stating and solving the general form of the Dirichlet problem:
p.) We then solve the equation Y" - A2y = 0 using these eigenvalues. From our expe-
Un + uyy = 0, 0 < X < a, 0 < Y < b rience with second-order equations, we know that Yn(Y) = ane vA;;; + bne-vA;;;, which
u(x, 0) = !ICx), u(x, b) = hex), 0 < x < a , can be written in terms of the hyperbolic trigonometric function as
[
u(O, y) = gt(Y), u(a, y) = g2(Y), 0 < Y :;;y + En sinh V AnY = An cosh -a- + En SInh -a-'
where the boundary conditions are shown in Figure 10.14.
This boundary-value problem is solved.through separation of variables. We be- Then using the homogeneous boundary condition u(x, 0) = X(x)Y(O) = 0, which indi-
gin by considering the problem cates that YeO) = 0, we have
Un + uyy = 0, 0 < X < a, 0 < Y < b YeO) = yncO) = An cosh 0 + En sinh 0 = An = 0,
u(x, 0) = 0, u(x, b) = 1(X), 0 < x < a.
[
u(O, y) = 0, u(a, y) = 0, 0 < Y < b so An = 0 for all n. Therefore, YnCy) = En sinh Ay, so that a solution is
In this case, we assume that n'Tl)' n'TTX

un(x, y) = En sinh -a- sin -a-'
u(x, y) = X(x)Y(y),
By the Principle of superposition, we have
n'Tl)' . n'TTX
u(x, y) = I
00
n=1
En sinh -a- sm -a-'
y
where the coefficients are determined with the boundary condition u(x, b) = I(x). Sub-
U(X, b) = [,(x)
y=b stitution into the solution yields
u(O,y) = u(a,y) = n'TTb n'TTX
g,(y) g2(Y) u(x, b) = I
00
En sinh -a- sin ---;;- = I(x),

n=1
x
u(x, 0) = Nx) 1 where E sinh n'TTb represents the Fourier sine series coefficients given by
x;:a Figure 10.14 Boundary equation on the region n a
.j
580 Chapter 10 Partial Differential Equations 10.4 Problems in Two Dimensions: Laplace's Equation 581
. n'7Tb 2 La f(x) sm. -n7TX dx I Solution Assume that u(x, y) = X(x)Y(y). Notice that this ~~oblem ~ffers from
Bn sinh - = -
a a 0 a the previous problem in that the homogeneous boundary condinons are m tenns of
Bn = 2
a sinh n'7Tb
r0
f(x) sin n7TX dx
a
the variable y. Hence, when we separate variables, we use a different constant of
separation. This yields
a X"Y+XY"= 0
o "fi""'.' uxx + uyy = 0, 0 < x < 1, 0 <y < 2

X"
X = -y=A,
Y"
so we have the ordinary differential equations X" - AX = 0 and Y" + AY = o.

Solve u(x, 0) = 0, u(x, 2) = x(l - x), 0 < x < 1. Therefore with the homogeneous boundary conditions u(x, 0) = X(x)Y(O) = 0 and
{
u(O, y) = 0, u(l, y) = 0, 0 < Y < 2 u(x, 1) = X(x)Y(I) = 0, we have YeO) = 0 and Y(I) = O. Recall that the eigenvalue
problem
:. Solution In this case, a = 1, b = 2, andf(x) = x(I - x). Therefore,
Y"+ AY= 0
B n -_ 22
--:----nh
SI n7r 0
L' ( ). cos n7r
x 1 - x sm n7TX. dx = - .- - - -~
SInh 2n7r n 7r
1 (4 4)
+ -n 3,,?-
{ YeO) = 0, Y(I) = 0
.;
has the eigenvalues An = (n7r/I)2 = n2 ,,?, n = 1, 2, . . . and solutions Yn(Y) =

n
3,,?
S
~ 2n7T (l - (-I)"), n = 1,2, ... , sin n7l)i, n = 1,2, .... We then solve the equation X" - AX =.0 usmg these eIgen-
values to obtain Xn(Y) = ane n= + bne- n=, which can be wntten m tenns of hy-
so the solution is
perbolic trigonometric functions as
_ ~ B·nh . ~ 8 sinh((2n - I)1l)') sin((2n - I)m:)
u (x, y ) - L n SI n7l)i sm n7TX = L 3 3 X(x) = Xn(x) = An sinh n7TX + Bn sinh n7r(7r - x).
n~1 n~1 (2n - 1) 7r sinh(2(2n - 1)7r) .
Figure 10.15 The graph of u In Figure 10.15, we plot u(x, y) using the first five tenns of the series solution. Now, because the boundary conditions on the boundaries x = 0 and x = 7r are non-
l. homogeneous, we use the Principle of superposition to obtain
In Example 1, what happens to the value of u away from y = 2?

u(x, y) = L (An sinh n7TX + Bn sinh n7r( 7r - x» sin n7l)i
Any version of Laplace's equation on a rectangular region can be solved through n=l
separation of variables as long as we have a pair of homogeneous boundary conditions Therefore,

in the same variable. (Note that a general fonn of this problem is presented in Exer- ~
cise 33.) u(O, y) = L Bn sinh n"? sin n7l)i = sin 21l)',

Notice that in Example 1, we could have used the two linearly independent so- n=l
lutions,
so B2 sinh 2"? = 1 or B2 = lIsinh 2"? and Bn = 0 for n * 2. Similarly,
sinh n1l)' inh n7r(b - y)
and
a sa' u( 7r, y) = L An sinh n"? sin n1l)' = 4,
n=1
of Y" - ,\?Y = 0 to form a general solution. (See Exercise 33). Doing so makes ap-
plying the boundary conditions more efficient, as we see in Example 2. which indicates that An sinh n"? represents the Fourier sine series coefficients for
the constant function 4, which are given by
",',,','Mj A sinh n"? = -2

n 1
L' 4 .
0
sm n1l)' dy = -8 [cos
-- n1l)']'
-
n7T 0
u= + uyy = 0, 0 < x < 7r, 0 < Y < 1
Solve u(x, 0) = 0, u(x, 1) = 0, 0 < x < 71"
{
u(O, y) = sin 21l)', u( 7r, y) = 4, 0 < Y < 1
•
= ~![(-I)" -I], n = 1,2, ....
the steady-state temperature satisfies Laplace's equation,

S;cx + Syy = 0,
because there is no dependence on t. Also, we describe the temperature around the
boundary of the rectangular region as we did in the Dirichlet problem, which gives us
a similar problem to solve to find Sex, y).
Find the steady-state temperature in the rectangle bounded by x = 0, x = 7T, Y = 0,

and y = I if the edges y = 0 and y = I are insulated and the temperatures along the
Figure 10.16 The graph of u Figure 10.17 A different view of the
graph ofu edges x = 0 and x = 7T are given by S(O, y) = 100 and S( 7T, y) = cos 1l)'.
: .1 Solution To describe the insulated edges mathematically, we recall that insula-

. tion stops the flow of heat over those edges. Because flow across these edges is in
From this formula, we see that An = 0 if n is even. Therefore, we express these co-
efficients as the y direction, we have
and Six, I) = 0 for O<X<7T.
16
A 2n - 1 = (2n _ 1)7T sinh((2n _ I)~) , n = 1,2, ... Therefore, we solve the boundary-value problem
so that the solution is Sxx + Syy = 0, 0 < x < 7T, 0 < Y < I
u(x, y) = Sinhl2~ sinh 27T( 7T - x) sin 21l)' ISix, 0) = 0, S/';X, I) = 0, 0 < x < 7T
S(O,y) = 100, S(7T,y) = cos 1l)', 0 <y < I
Assuming that Sex, y) = X(x)Y(y), we perform separation of variables. Notice that
•
+ f
n";;"'l
16
(2n - 1)7T sinh((2n - I)~)
sinh n7TX sin mry. in this case, the homogeneous boundary conditions are associated with Y because
Sy(x, 0) = X(x)Y'(O) = 0 and Six, 1) = X(x)Y'(I) = 0 lead to the conditions
Two different views ofu are shown in Figure 10.16 and Figure 10.17. Y'(O) = 0 and Y'(1) = O.
Based on our experience with eigenvalue problems, we choose the constant of sep-
The other type of boundary conditions associated with Laplace's equation in- aration so that we obtain the equations
volves the derivative on the boundary (the Neumann problem) or a mixture of the X"- AX= 0 and Y"+ AY= 0
Dirichlet and Neumann problems. We consider these other types of problems in the
when separating variables. Taking advantage of the homogeneous boundary condi-
foIIowing context.
tions, we first solve the problem
Y" + AY = 0, 0 < Y < I
Steady-State Temperature
{ Y'(O) = 0, Y'(I) = 0
One application of Laplace's equation is the steady-state solution of the heat equation
in a rectangular region. Recall that as t -? "', the temperature no longer depends on t. with eigenvalues and corresponding eigenfunctions
Therefore, if the temperature in a rectangular region at the point (x, y) at time t is given 0, n = 0
by u(x, y, t), then
~n;, u(x, y, t) = Sex, y),
A =
n (( T)
n7T 2 ?
= n-rr>, n = I, 2, ...
and
where sex, y) is the steady-state temperature. Because the heat equation in two spatial 1 n= 0
dimensions, x and y, is given by
Yn(Y) = ( cos n7 = cos n1l)', n = 1,2, ...
r c - - - - - - - - - - - - - - - - - - ----
584 Chapter 1 0 Partial Differential Equations

Using these eigenvalues, we solve X" - AX = O. If A = ,1.0 = 0, then we have the

solution EXERCISES 10.4
X(x) = Xo(x) = c, + C2X.
2
Similarly, if A = An = n ,,?, n = 1,2, ... , we find that
In Exercises 1-8, the problems involve three homogeneous and U XX+ u",= 0, O<x<"',O<y<b
X(x) = xncx) = An cosh n7TX + Bn sinh n7TX. one nonhomogeneous boundary condition. Use the methods dis- u(x, 0) = 0 and u(x, b) = 0, 0 < x < ",
cussed in the examples to solve the boundary-value problems. u(x, y) is bounded as x ---> ",
Combining these results, we find with the Principle of superposition that
Sex, y) = Xo(x)Yo(y) +I Xn(x)Yn(Y) = c, + C2X

Un + u", = 0, 0 < X < I, 0 < Y < 2
1. u(x, 0) = 0, u(x, 2) = 0, 0 < X < 1
! u(O, y) = fey), 0 <Y < b
(a) Using separation of variables with u(x, y) =
u(O, y) = 0, u(I, y) = 1 + y, 0 < y < 2
{
n=l
X(x)Y(y), show that Y satisfies the eigenvalue
~
UXX + u", = 0, 0 < X < 7T, 0 < Y < 7T problem
+I (An cosh n7TX + Bn sinh n7TX) cos n'llJ'. 2. u(x, 0) = 0, u(x, 7T) = 0, 0 < X < 7T
Y + AY = 0, 0 < Y < b
N
n~' {
u(O, y) = 0, U(7T, y) = 10, 0 <y < 7T
{ YeO) = 0, Y(b) = 0
Applying the condition S(O, y) = 100 yields U XX + u", = 0, 0 < X < 1, 0 < Y < 2
*3. u(x, 0) = 0, u(x, 2) = 0, 0 < X < 1 and find the eigenvalues and eigenfunctions of
S(O, y) = c, + Z An cos mry = 100, {
u(O,y) =y, u(I,y) =0, 0<y<2 this problem.
n=l (b) Show that X.(x) = a.e·=Ib + b.e-·=Ib. Find a
Un+ u", = 0, 0 < X < 7T, 0 < Y < 27T
so c,and An are the Fourier cosine series coefficients for f(x) = 100, which are 4. u(x, 0) = 0, u(x, 27T) = 0, 0 < X < 7T restriction on one of the constants so thatX.(x) is
found in this case by matching the coefficients on each side of the equation (or by {
u(O,y) = y2, U(7T,y) = 0, 0 <y < 27T bounded as x .... "'.
carrying out the appropriate integration). In this case, c, = 100 and An = 0 for all (e) Fonn the series solution u(x, y) =
u=+uyy =0,0<X<I,0<y<2 ~ m1)l
n. For S( 71", y) = cos 77)1, we have
5. u(x, 0) = 1, u(x, 2) = 0, 0 < X < 1
{
I B.e -.=lb sin -b- and find B. using the
n=1
u(O,y) = 0, u(I,y) = 0, 0 <y<2
S( 71", y) = 100 + C271" + I Bn sinh n"? cos n77)l = cos 77)1.
u=+ uyy = 0, O<x<I,O<y<I
boundary condition u(O, y) = fey).
n=1
6. u(x, 0) = sin 2=, u(x, 1) = 0, 0 < x < 1 In Exercises 12-14, use the solution found in Exercise 11 to
Therefore, 100 + C271" = 0, which means that C2 = -100171". Then by comparing { u(O, y) = 0, u(I, y) = 0, 0 < Y < 1 solve each boundary-value problem using the given parameter
the coefficients of the cosine series and the function cos 77)1 with n = I, we find value and boundary condition.
that B, sinh ,,? = I, so B, = (llsinh ,,?) and Bn = 0 for n;:;': 2. The steady-state Un + u", = 0, 0 < x < 1, 0 < Y < 2
temperature in the region is then given by
*7. u(x, 0) = 0, u(x, 2)
{
= sin 0<X< 1 =, 12. b = 7T, u(O, y) = 100
u(0,y)=0,u(I,y)=0,0<y<2 13. b = I, u(O,y) = To
100 1 . u=+uyy =O,O<X<I,O<y<7T 14. b = 2, u(O, y) = Y
Sex, y) = 100 - -;-x + sinh"? Sinh TTX cos 77)1.
8. u(x, 0) = 0, u(x, 7T) = sin 2x, 0 < x < 1
{ 15. Use a method similar to that in Exercise II to solve
u(O, y) = 0, u(1, y) = 0, 0 < y < 7T
In Figure 10.18, we graph sex, y) over the region.
u= + u", = 0, O<x<a,O<y<'"
In Exercises 9-10, solve each problem using separation ofvari- u(O, y) = 0, u(a, y) = 0, 0 < Y < ",
u(x, y) is bounded as y .... ",
abies. (Note that the method of separation of variables must be
used due to the boundary conditions.)
U XX + u", = 0, 0 < X < 0<Y < 1
7T,
! u(x, 0) = f(x), 0 <x <a
9. uxCO, y) = 0, ux ( 7T, y) = 0, 0 < Y < 1 In Exercises 16-18, use the results of Exercise. 15 to find the
{
u(x, 0) = 4 cos x + 1, u(x. 1) = 0, 0 < x < 7T solution of each boundary-value problem using the given pa-
rameter value and boundary condition.
u= + u", = 0, 0 < x < 1, 0 < Y < 1
10. u(O,y) =y, u{l,y) = 0, O<y<I 16. a = 7T,f(X) = 1
Figure 10.18 The steady state of the region { uy(x, 0) = 0, uy(x, 1) = 0, 0 < x < 1
17. a = 2,f(x) = x
11. (Steady-State Temperature in a Semi-Infinite Strip)
Consider the boundary-value problem
18. a = I,f(x) ="3I sm
.
2=
586 Chapter 10
19. Find the steady-state temperature in the rectangle

Partial Differential Equations
c2(u= + u",,) = u 0 < X <

ft , <Y< b
a, 0
1
,~
29. a = 1, b = 2,J(x, y) = xy(1 - x)(1 -
10.4 Problems in Two Dimensions: Laplace's Equation
y), g(x, y) =0 Find a general formula for each of the unknown co-
587
bounded by x = 0, x = 1, y = 0, and y = Tr if the efficients An and B n ·

u(x, 0, t) = 0, u(x, b, t) = 0, <x <a
0 30. a = b = 7T,J(X, y) = 0, g(x, y) = xy
edges x = 0 and x = 1 are insulated and the temper- (c) In a marmer similar to that in (b), solve
u(O, y, t) = 0, u(a, y, t) = 0, 0 <Y < b 31. Determine the steady-state solution of the two-
ature along the edges y = 0 and y = Tr are given w= + w"" = 0, 0 < x < a, 0 < Y < b
u(x, y, 0) = J(x, y). 0 < x < a, 0 < y < b dimensional heat equation
by Sex, 0) = cos 7TX and Sex, Tr) = O. (Assume that
ulx,y, 0) = g(x,y), 0 < x < a, 0 <y < b, w(x, 0) = 0, w(x, b) = 0, 0 < x < a
c = I). k(U= + u",,) = u, 0 < x < a, 0 < Y < b (
which determines the displacement of a rectangu_ u(x, 0, t) = /J(x), u(x, b, t) = f,(x), 0 < x
<a w(O, y) = hey), w(a, y) = key), 0 < Y < b.
20. Find the steady-state temperature in the square
bounded by x = 0, x = Tr, Y = 0, and y = Tr if the lar membrane (a rectangular drumhead). (a) Let u(O, y, t) = gl(X), u(a, y, t) = g2(X), 0 < y < b Find a general formula for each unlmown coefficient.
u(x, y, t) = Sex, y)T(t), and show that S satisfies the [
edges y = 0 and y = Tr are insulated and the temper- u(x, y, 0) = F(x, y), 0 < x < a, 0 < Y < b.
atures along the edges x = 0 and x = Tr are given partial differential equation S= + S"" + v2S = 0 In Problems 34-42, use the results of the Exercise 33 to solve
by S(O, y) = 0 and S( Tr, y) = cos 2y. (Assume that (caUed the Helmholtz equation), where (-v2) is the 32. Use the results of Exercise 31 to find the steady-state
each problem. Use technology to assist in determining the co-
c = 1). constant of separation. Also, T satisfies T" + A2r = temperature in the rectangular plate R: 0 < x < 1,
efficients in the solutions and graph the solutions (with a finite
0, where A = cv. (b) LetS(x,y) = X(x)Y(y). Show that o < Y < 2 if the temperature is fixed at zero around number of terms of series solutions when necessary), and no-
2!. (Double Fourier Series) The series X and Y satisfy X"+~X=O and Y"+p2 y =0, each edge of the region except u(x, 0, t) = sin 7TX.
tice how the boundary conditions affect the solution.
~ . mry where ~ + p2 = y2. (e) Show that the four homo-
J(x, y) = L L Bmn sm -m7TX
00
a
•
sm -b- geneous boundary conditions can be expressed as

33. To solve
u= + UJoY = 0, 0 < x < 1, 0 < Y < 1
m=l rr=l
X(O) = X(a) = 0 and YeO) = Y(b) = O. Use these u= + UJoY = 0, 0 < x < a, 0 < Y < b 34. u(x, 0) = 10, u(x, 1) = 0, 0 < x < 1
is a double Fourier series. (The function J can be bOUI1aary conditions to solve the eigenvalue problems u(x, 0) = J(x), u(x, b) = g(x), 0 < x < a,
{ u(O,y) = O,u(l,y) = 10, O<y<1
written as such a series over a region R: 0 :::; x ::::: Q, for X and Y. Combine these results to show that (
u(O, y) = hey), u(a, y) = key), 0 <Y < b UXX + u"" = 0, 0 < x < Tr, 0 < Y < 1
o oS Y oS b ifJ, Ix, /y,f=, and J"" are continuous in the 35. u(x, 0) = cos x, u(x, 1) = 0, 0 < x < Tr
region R.) (a) To find the coefficients B m", we assume that u(x, y) = vex, y) + w(x, y). (a) Show {
u(O,y) = l,u(Tr,y) =0, O<y<1
let km(y) = ~l Bmn sin n;: and show that the series
m = 1, 2, ... , n = 1, 2, .... (d) Show that A =
that if v and w satisfy the problems
u= + UY)' = 0, 0 < x < Tr, 0 < Y < 1
~ nt7rX
v= + VY)' =
0, 0 < x < a, 0 < Y < b 36. u(x, 0) = 0, u(x, 1) = x(l - x), 0 < x < 7T
becomesJ(x, y) = ~ km(y) sin - - . (b) For a fixed Amn = CTr V(m/a)' + (nIb)2 and that the solutions vex, 0) = J(x), vex, b) = g(x), 0 < x < a and { U(O,y) = l,u(r.,y)=O, O<y< 1
m=l a of Tn + A2T = 0 are Tmn(t) = Bmn cos Amnl + (
y, J can be considered a function of x only, so show v(O, y) = 0, v(a, y) = 0, 0 <Y< b u= + UJoY = 0, 0 < x < 7T, 0 < Y < I
Cmn sin Amnt. (e) Combine the results of (e) and (d)
that km(y) = a
2 (a m7TX
J J(x, y) sin -a- dx. (c) For the
o
. to show .that
w= + w"" = 0, 0 < x < a, 0 < Y < b
*37. u(x, 0) = 0, u(x, 1) = x(l - x), 0 < x
{ u(O,y) = I, U(7T,y) = 10, 0 <y < 1
< Tr
function of y, km(y) = i:
n=l
Bmn sin n;: , show that
umn(x, y, t) =
(B mn cos Amnt +C '
mn sm Am"t
) sm
. -a-
m7TX . n71)'
Sm -b-
(
w(x, 0) = 0, w(x, b) = 0, 0 < x < a
w(O, y) = hey), w(a, y) = key), 0 <Y< b
,
38.
u= + Uyy = 0, O<x<I,O<y<1
u(x, 0) = x(l - x), u(x, 1) = 0, 0 < x <1
2 (b nny { u(O, y) = 0, u(l, y) = y, 0 < Y < I
Bmn = bJ km(y) sin -b- dy. (d) Combine the re- then u(x, y) = vex, y) + w(x, y) is a solution of the
o are solutions to the wave equation in rectangular co·
original equation. u=+ uJoY= 0, O<x<I,O<y<1
sults of (b) and (c) to show that ordinates. (f) Show that the coefficients Bmn are found
with the formula in Exercise 21. (g) Show that (b) Show that the solution of 39. u(x, 0) = 0, u(x, 1) = x 2, 0 < X < 1
4 (b (a . m7TX . nny { u(O,y) =y, u(l,y) = 1 -y, O<y< 1
Bmn = ab J J J(x, y) sm -a- sm -b- dx dy, 4 (b (a . m7TX • nny v= + v"" = 0, 0 < x < a, 0 < Y < b
o o
Cmn = abAmn J J g(x, y) Sm -a- sm -b- dx dy, vex, 0) = J(x), vex, b) = g(x), 0 < x < a u= + uyy = 0, 0 < x < 1,0 < Y < 1
m = 1,2, ... , n = 1,2, .... o o
( 40. u(x, 0) = sin 7TX, u(x, I) = x, 0 < x < 1
m = 1,2, ... , n = 1,2, ....
v(O, y) = 0, v(a, y) = 0, 0<Y <b { u(O, y) = y, u(l, y) = 1 - y, 0 <Y< 1
In Exercises 22-24, use the formula derived in Exercise 21 to is
u=+ u",,= 0, 0<x<I,0<y<2
lmd the double Fourier series for each function over R. In Exercises 26-30, use the formula derived in Exercise 25 to 2
*41. u(x, 0) = 0, u(x, 2) = x(1 - x ), 0 < <1
v(x,y) = X
solve the wave equation in R using the given parameter values {
u(O, y) = y, u(1. y) = 0, 0 < Y < 2
22. J(x, y) = I, R: 0 oS x oS 1, 0 oS Y oS 1 and initial conditions. ~ ( an cosh -
L
nl1y .
- + bn smh nny). n7TX
- - sm - - ,
n=\ a a a U= + uyy = 0, 0 < x < 1, 0 <y < 2
23. J(x, y) = x + y, R: 0 oS X oS 1, 0 oS Y oS Tr
26. a = b = Tr,J(x, y) = ~ sin 2x sin 2y, g(x, y) = 0 42. u(x, 0) = 0, u(x, 2) = x(l - x 2 ), 0 < X < 1
which can be written as { u(O,y) = y, u(l,y) = sin ny/2, 0 <y < 2
24. J(x, y) = xy, R: 0 oS:c oS Tr, 0 oS Y oS Tr
25. (Two-Dimensional Wave Equation in Rectangular 27. a = b = Tr,J(x, y) = 0, g(x, y) = ± sin x sin 2y
v(x,y) =
Coordinates) Solve the initial boundary value

problem 28. a = 1, b= 2,J(x,y) = O,g(x,y) = 1
f (An sirth nnya + Bn sirth WIT(b - y) )sin n7TX.
n7 1 Q a
In Problems 43-46, find the potential in the square R: 0 oS
x oS Tr, 0 oS Y oS Tr by solving (by hand) the following bound-
Chapter 10 Partial Differential Equations 10.5 Two-Dimensional Problems in a Circular Region 589
588
ary-value problem using the given boundary conditions. In each 43. f(x) sin x Also, we want our solution to be bounded at r = 0, so another boundary condition is
case, graph the equipotential curves along which u(x, y) is con- 44. f(x) = sin 3x lu(O, (1)1 < 00 for
stant.
u= + u>y = 0, 0 < x < "IT, 0 < Y < "IT

*45. f(x) = sin x - isin 2x
I
Finally, we can specify the value of the solution around the boundary of the circle. This
is given by
u(x, 0) = 0, u(x, "IT) = f(x), 0 < x < "IT
[ u(O, y) = 0, u("IT, y) = 0, 0 < y < "IT 46. f(x) = sin 2x +I sin 5x ~
u(p, (1) = f(l1) for
1
'! Therefore, we solve the following boundary-value problem to solve Laplace's equation
(the Dirichlet problem) in a circular region of radius p.
:2
[
u~ +.;ur + u&& = 0,
110.51 Two-Dimensional Problems in a Circular Region
, Laplace's Equation in a Circular Region ,The Wave Equation in a !
1
u(r, -7/") = u(r, "IT), uoCr,
lu(O, (1)1 < 00,
0 < r < p, -7/" < 11 < "IT
-7/") = uoCr, 7/"), 0 < r
u(p, (1) = f(I1), -"IT < 11 < 7/".
<p
Circular Region
.
)
I
Using separation of variables, we assume that u(r, (1) = R(r)H( (1). Then, substitution
In some situations, the region on which.we solve a boundary-value problem or an into Laplace's equation yields
initial boundary value problem is not rectaD.gular in shape. For example, we usually do
not have rectangular-shaped drumheads, and the heating elements on top of a stove are R"H + 1. R'H + RH"= 0
r
not square. Instead, these objects are typically circular in shape, so we find the use of
polar coordinates convenient. In this section, we discuss problems of this type by pre- R"H + 1. R'H = -RH"
senting two important problems solved on circular regions, Laplace's equation, which r
is related to the steady-state temperature, and the wave equation, which is used to find rR" + R' -H"
---=--=A
the displacement of a drumhead. rR H
Laplace's Equation in a Circular Region Therefore, we have the ordinary differential equations
In calculus, we found that polar coordinates are useful in solving many problems. The H"+ ill = 0 and r2R" + rR' - AR = O.
same can be said for solving boundary-value problems in a circular region. With the Notice that the boundary conditions
change of variables
u(r, -"IT) = u(r, 7/") and uoCr, -"IT) = uoCr, 7T)
fx = r cos a
1Y=rsmO imply that
we transform Laplace's equation in rectangular coordinates, Uxx + u yy = 0, to polar co- R(r)H( - 7T) = R(r)H( 7/") and R(r)H'( -7T) = R(r)H'( 7T),
ordinates so that
I I
u~ + -; U r + --;'1 u&& = 0, 0 < r < p, -"IT < 11 < 7/". H(-7T) = H(7/") and H'(-7T) = H'(7/").
This means that we begin by solving the eigenvalue problem
(We leave this transformation as an exercise.) Recall that for the solution of Laplace's
equation in a rectangular region, we had to specify a boundary condition on each of H" + ill = 0 - 7T < 11 < "IT
the four boundaries of the rectangle. However, in the case of a circle, there are not four { H(-7T) = H(7/"), H'(-7T) = H'(7T),
sides, so we must alter the boundary conditions. Because in polar coordinates the points
which was solved in Exercise 35 in Section 9.1 and used to establish the Fourier se-
(r, 7/") and (r, -"IT) are equivalent for the same value of r, we want our solution and its
ries. The eigenvalues and corresponding eigenfunctions of this problem are
derivative with respect to 11 to match at these points (so that the solution is smooth).
Therefore, two of the boundary conditions are n=O I n= 0
and Hn( (1) = ' b . 1 2
{ an cos nl1 + n sm nu,
Q
u(r, -"IT) = u(r, 7/") for 0< r < p. n = 1,2, ... n= , , ...
and
590 Chapter 10 Partial Differential Equations 10.5 Two-Dimensional Problems in a Circular Region 591
Because r2R" + rR' - A2R = 0 is a Cauchy-Euler equation, we assume that 1 JOT IOldO=-
1 JOT OdO=-
1 [02]OT 'IT
R(r) = rm. Ao=- - =-,
27r -OT 7r 0 7r 2 0 2
and with integration by parts,
rm[m(m - 1) + m - A] = 0
1 JOT
An = --n 101 cos nO dO = 1
----;;=J JOT 0 cos nO dO = (- 2
I 2 ) [cos n'IT - 1]
7r2 -OT 7r2 n 'IT n
Therefore, 0
= ('IT2n~ln2 )[(-1)" - 1], n:2: 1.

m = :tA.
-2
If Ao = 0, then Ro(r) = Cl + C2 In r. However, because we require that the solution Notice that A2" = 0, n :2: I, and A2n - 1 = _2 2" 2 2' n:2: 1, so the solution
.. (2n - I)
be bounded near r = 0 and limr-. o' In r = - co, we must choose C2 = O. Therefore is
Ro(r) = Cl' On the other hand, if An = n2, n = 1,2, ... , then Rn(r) = c3r" + c4r-.: eo 2n- 1
Similarly, because limr-.o+ r -n = co, we must let C4 = 0, so Rn(r) = C3rn. By the Prin- u(r, 0) =!!:. -
2
L
n~l 'IT2
2n_ 2
r
(2n -
--1-
2 cos(2n - 1)0.
)
ciple of superposition, we have the solution
In Figure 10.19, we graph this solution over the circular region.
u(r, 0) = Ao + L
n~l
r"CAn cos nO + Bn sin nO),
where Ao = Clo An = C3am and Bn = C3b". We find these coefficients by applying the
boundary condition u(p, 0) = f(O). This yields
u(p, 0) = Ao + L pn (An cos nO + Bn sin nO) = f(O),

n=l
Figure 10.19 Graph of the potential
so Ao, Am and Bn are related to the Fourier series coefficients in the following way: over the circular region
Ao = 2~ rOT f(O) dO
In an earlier example using rectangular coordinates, we discussed how we asso-
ciate the solution to Laplace's equation to the steady-state temperature. Of course, the
An = -1" JOT f(O) cos nO dO (n = 1,2, ... )
'ITP -OT same can be done with polar coordinates. We illustrate this approach in Example 2.
I JOT
B" = -" f(O) sin nO dO (n = 1,2, ... )
'ITP -OT "f'"I"8j
What is the potential at the center of the region? What does this value represent? Find the steady-state temperature in the circular region of radius p = 10, if
u(IO, 0) = 70.
"f'"I".'1 1 Solution Instead of introducing a new variable to represent the steady-state tem-
perature as we did in Section lOA, let u(r, 0) be this temperature. Common sense may
1 1 tell us that the temperature at each point in the circular region eventually attains the
U rr + -;: Ur + r2 Uee = 0, 0< r < 2, -'IT < 6< 'IT
Solve value of the temperature around the boundary. The formula for u above verifies
u(r, -'IT) = u(r, 'IT), ue(r, -'IT) = ue(r, 'IT), 0 < r < 2.
1
lu(O, 0)1 < co, u(2, 0) = 101, -'IT < 0 < 'IT. this because Ao = 1/(2'IT) rOT 70dO = 70, An = l/('ITIO") rOT 70cosnOdO= 0,
and B" = 1/( 'ITI0") rOT 70 sin nO dO = O. Therefore, u(r, 0) = 70.
J Solution Notice thatf(O) = 101 is an even function on -'IT < 0 < 7r. Therefore,
Bn = 0 for n :2: 1,
j
~----------------------------------------
The Wave Equation in a Circular Region As with other problems, we are able to use separation of variables to find u by as-
suming that u(r, t) = R(r)T(t). Substitution into the wave equation yields
One of the more interesting problems involving two spatial dimensions (x andy) is the
wave equation, 2
c (R"T + ~ RT) = RT"
rR"+R' =..J::....=-1C
This is due to the fact that the solution to this problem represents something with which rR c2 T '
we are all familiar, the displacement of a drumhead. Because most drumheads are cir-
where -!?- is the separation constant. Separating the variables, we have the ordinary
cular in shape, we investigate the solution of the wave equation in a circular region.
differential equations
Therefore, we transform the wave equation into polar coordinates. In the previous sec-
tion on Laplace's equation, we found that and T"+ c 2 !?-T= o.
We recognize the equation r2R" + rR' + !?-r2R = 0 as Bessel's equation of order zero,
which has solution
Then the wave equation in polar coordinates becomes
where J o and Yo are the Bessel functions of order zero of the first and second kind, re-
spectively. In terms of R, we express the boundary condition lu(O, t)1 < 00 as
If we assume that the displacement of the drumhead from the xy-plane at time t is the
same at equal distances from the origin, then we say that the solution u is radiaUy
IR(o)1 < 00.
symmetric. (In other words, the value of u does not depend on the angle e.) There- Therefore, because limr... o+ YO(A.T) = -00, we must choose C2 = O. Then applying the
fore, u •• = 0, so the wave equation can be expressed in terms of r and t only as other boundary condition R(p) = 0, we have
c u" + .;u,)
2
( = U".
To avoid the trivial solution with Cj = 0, we find k satisfying

Of course, to find u(r, I) we need the appropriate boundary and initial conditions. Be-
cause the circular boundary of the drumhead must be fixed so that it doesn't move, we kp = am
say that
where an is the nth zero of J o, which was discussed in Section 4.8. Because k depends
u(p, I) = 0 for t> o. on n, we write
Then as we had in Laplace's equation on a circular region, we require that the solution
u be bounded near the origin. Therefore, we have the condition
lu(O, 1)1 < 00 for I> O. The solution of T" + c 2 !?-T = 0 is
The initial displacement and initial velocity functions are given as functions of r as Tn(t) = An cos cknt + Bn sin cknl,
u(r, 0) = f(r) and u,(r, 0) = g(r) for 0< r< p. so with the Principle of superposition, we have
Therefore, the initial boundary value problem to find the displacement u of a circular
drumhead (of radius p) is given by
u(r, t) = I (An cos cknl + Bn sin cknt)Jo(knr),
n=l
+~
where the coefficients An and Bn are found by applying the initial displacement and
2
C ( U" u,) = u", 0 < < r p, t >0
velocity functions. With u(r, 0) = I AnJo(knr) = f(r) and the othogonality conditions
(
u(p, t) = 0, lu(O, 1)1 < 00, I> 0 n=l
of the Bessel functions, we find that
u(r, 0) = f(r), u,(r, 0) = g(r), 0 < r < p.
1 10.5 Two-Dimensional Problems in a Circular Region 595
A
n
r
=~o~
I:
rf(r)JO(knr) dr
_ _ __
r [Jo(knr)f dr
-()]2
[
J 1 an
2 fP rf(r)Jo(knr) dr, n =
0
1, 2, .... ! I
I~
0.800
0.400
O'800~
0.600
0.400
0.200
0.400
0.300
0.200
0.100
,=0 ,=0.25 ,=0.5

Similarly, because
~
_0'500~ -0.200
-0,400 -O.400~
u,(r, t) = I (-cknAn sin cknt + cknBn cos cknt)Jo(knr),
-0.100
-0.150
-0.200
~
-0,600
-0.800
-0.800~
11=1
we have ,=0.75 ,= 1.0 ,= 1.25

~
u,(r, 0) = I
n=1
cknBnJo(knr) = g(r).
_0'200~
-0.400
-0,600
-0.800
Therefore,
,= 1.5
I: rg(r)Jo(knr) dr
2' fP rg(r)Jo(knr) dr, n = Figure 10.20 Circular drum at various values of t
Bn =
ckn r
o
r[Jo(knr)f dr
Ck n [J 1 ( an )]
2
0
1, 2, ....
o lifi,,','I*"
EXERCISES 10.5
l
(u + +U r
2
C rr r) = UII, 0 < < I, t> 0
Solve u(l, t) = 0, Iu(O, t) I < 00, t> 0
u(r, 0) = 1 - r2, ut(r, 0) = 0, 0 <r<1 In Exercises 1-10, solve Laplace's equation in a circular region In Exercises 11-16, solve the wave equation in a circular re-
of radius I given the boundary condition u(l, 0). gion of radius p = I that is radially symmetric using the initial
: J Solution In this case, p = 1,/(r) = I - r2, and g(r) = O. Therefore, Bn = 0 for 1. u(l, 0) = cos 0
condition functions given as follows. (If possible, graph the so-
lution with a computer algebra system for various times to view
n2: I and
2. u(l, 0) = sin 20 the motion of the drumhead.)
An = [J (a )f
2 f1 2 4J2(an)
r(1 - r )Jo(anr) dr = a/[J (a )]2' n 2: 1,
*3. u(l, fJ) = cos' 0
11. fer) = Jo(",r), g(r) = °
1 n
0
1 n 4. u(l, 0) = sin' 20
5. u(l, 0) = 0'
12. fer) = 0.1 J o( 0I3r), g(r) = °
where the previous integral was calculated with integration by parts with the *13. fer) = 0, g(r) = Jo(",r)
formula !
[xnJn(.x)] = x nJ n_ 1(x). (See Exercise 20.) Then the displacement is
6. u(l, 0) = 0 sin 0
*7. u(l, 0) = {1_'I,-1T< 0<0

I
14. fer) = 0, g(r) = '4 Jo(",r)
0<0<1T I I
15. fer) = '8 J o("2r), g(r) = '4 Jo( 0I2r)
u(r, t) = L
~ 4J2 (an )
2
n=1 an [J1(a n)]
2 (cos cant)Jo(anr). _{a,100, °<
8. u(l, 0) -
-1T<0<0
0 < 1T 16. fer) = J o(OI,r), g(r) = J o(OI,r)
We show the graph of u for several values of t in Figure 10.20 (using nine terms
from the series solution). To actually watch the drumhead move, an animation de-
9 (I 0) =
. u,
{a,0, -1T < 0 <
0<0<1T
° In Exercises 17-20, verifY the integrals involving Bessel func-
tions by using the identity ~ [x"Jn(x)] x"In_,(x) and inte-
vice of a computer algebra system can be used. (Note: The coefficients also may be
found numerically by using a computer algebra system.)
10. u(l, 0) = {-O, -1T< 0<
1T,0<0<1T
° gration by parts when necessary.
=
i, j
II
26. Solve the boundary-value problem in Exercise 24 if

f(O) = 1000(7T - 0).
27. (Laplace's Equation in Spberical Coordinates) In
spherical coordinates, Laplace's equation is
1! (Hint: r -\
f-I' p/(x) dx = 2n
2
+
10.5
P m(x)p"(x) dx = 0, m
Two-Dimensional Problems in a Circular Region
'* n, and
I' n = 0, 1,2, .... )
32. (Thermal Explosions: Approximating the Solution
of a Nonlinear Partial Differential Equation) Sup-
pose that a reaction produces heat throughout a cylin-
der of radius a at a rate proportional to the tempera-
597
2 1 ture, Qe -P(T Therefore, the situation is modeled with

the nonlinear partial differential equation
I [a ( 2aU) I a (. au) I a u] (e) On the other hand, if we use the boundary condi-
p2 ap p ap + sin a<l> sm a<l> + sin2 ae> = 0. tion lim u(r, <1» = 0, which is used when finding
As we have discussed in other situations, the steady.
"....,~
the temperature outside of the sphere, then form pc at

aT = k(aar2T+ 1.raT)
2
are
+ Q -PIT
,
state temperature satisfies Laplace's equation. There. the solution in this case. Find the unknown coef-
fore, in a sphere we consider this form of the equa. ficient. where T(r, t) is the temperature in the cylinder at ra-
tion. Suppose that the temperature on the surface of dius r and time t, p is the density, and c is the specific
In Exercises 21-22, use the integrals in Exercises 17-20 to a sphere of radius R is f( <1». Because the boundary In Exercises 28-30, fmd the first three nonzero terms of the heat of the solid. Suppose that the temperature on the
solve the wave equation in a circular region ofradius r = I us- condition is independent of 0, so is u. Therefore, steady-state temperature inside the sphere of radius I using the surface of the cylinder is T(a, t) = To. Therefore, we
ing the given initial conditions. assume that T does not differ much from To.
a 2 ulae> = 0, and Laplace's equation becomes given boundary condition.
21. fer) = I - r2, g(r) = ° 1.... (p2 au) + _1_.1.. (Sin 2!:!..) = °
28. u(I,<I» = 100 (a) Show that
22. fer) = 0, g(r) = I - r2 ap ap sin a<l> a</> . 29. u(I,<I» = cos 2<1> T T- To
23. Verify the change of variables of Laplace's equation 30. u(I,</» = cos 2 - 2 cos + I
-=1+--
To To
(a)"t.et u(p, </» = R(p)F(</». Show that R satisfies
by using the Chain rule to show that if x = r cos 0 and 31. (Laplace's Equation in Cylindrical Coordinates)
p2R" + 2pR' - kR = 0, where kis the constant of so that
y = r sin 0, then Suppose that the surface of the cylinder x 2 + y2 = a2
separation.
(b) Show that if k = n(n + I), then the solutions
°
between the planes z = and z = b is insulated and 1.="'!"(I+ T-To)-I.
and
of p2R" + 2pR' - kR =
Rn(r) = I/(r"+I).
°
are R"(r) = r" and
that there is a heat source on the top and bottom sur-
faces. We can find the steady-state temperature by
T To
Therefore, we use the appr~ximation
To
solving
y x (c) Show that F satisfies the equation I I T- To
= -; U r + ,2
uy U(J.
I d(. dF)
I -
- a (au)
r- +-=0
a u
2
r<a O<z d<l> sm d</> + n(n + I)F = 0.
and Qe -PIT = Qe -PIToeP(T-To)ITi (Why?).
To solve this equation, let w = cos <1>. Show that Tr
au (a, z) = 0, Iu(O, z) I < "', °
<z = -sin (dldw). Use this change of vari· pc
u(r, 0) = fer), u(r, b) = g(r), r <a
able to obtain the differential equation these variables to obtain the equation
(A similar formula can be found for u"'.)
(Notice that the boundary conditions do not depend
24. Consider Laplace's equation in a semicircular region 2 d 2F dF on 0, so neither does u.) 1. au = a2U + 1. au + qePu
(1 - w ) dwZ - 2w dw + n(n + I)F = 0, at ar2 r ar '
R bounded by the graphs of y = ~ and the K
(a) Let u(r, z) = R(r)Z(z). Show that R satisfies the
x~ax.is. Legendre s equation (Section 4.8), which has so· boundary-value problem where K = k/ pc.
(a) If the potential u is zero on the lower boundary of lutions FnCw) = p"(w) = p"(cos <1», n = 0, I, ....
(rR ')' + A2rR = 0, R'(a) = 0, IR(O)I < "', (c) The steady-state temperature u(r) will satisfy
R: - I :S x :S I and satisfies f( 0) on the semicir- (d) Consider the solutions obtained in (b). If u is
cular boundary, show that the solution satisfies the bounded as p .... 0+, which must be assumed when with solutions R"(r) = Jo(A"r). Use the relation- ° = d u2
2
+ 1. du + qeP".
boundary-value problem fmding the temperature inside the sphere, then
ship ! [Jo(x)] = -JI(x) to find A".
dr r dr
u"' + ~ u, + 7 u•• = 0, 0< r < I, 0< 0 < 7T

R"(r) = 1/(r" + 1) cannot be included in the solu·
tion. (Why?). In this case, the solutions are (b) Show that Z satisfies the equation Z" - A2Z = ° (Why?) To solve this nonlinear ordinary differen-
tial equation with the boundary condition u(a) = °
u(r, 0) = u(r, 7T) = 0, °< r< I
u"(r, <1» = a"r"P"(cos <1», so so that (Why?), let v = r du/dr. Show that the differential
l u(l, U) = feU), °
lu(o, 0)1 < "', < 0 < 7T
(b) Solve this problem, finding a general formula for
u(r, <1» = I
"~I
a"r"P"(cos </». u(r, z) = Ao + Bo= + I Jo(A"r)
"-I
equation becomes
1. dv = _qePu.
Show that r dr
all coefficients.
25. Solve the boundary-value problem in Exercise 24 if 2n + I lW . (d) Differentiate 1. dv = _qeP u with respect to r to
a" = --zR'" 0 f(<I»p"(cos </» sm </> d</>. fmd that r dr
f(O) = To· Find the unknown coefficients.
d (I dV)
dr-;:dr=-;:Zdr
pv dv
or
(c) fer) = r(l - r), g(r) = r HI> + p.2H= 0
f" rfer, 6)Jm(Am"r)sin(mO)r dr dO
2
.!:...(r dv _ 2V) = Ip d(v )
drdr 2 dr
(Wave Equation with Dependence on 8) The problem that de-
pends on the angle e is more complicated to solve. Due to the has solutions
{H(-7T) = H(7T), H'(-7T) = H'(7T) b
mn
1T r
= _0_0_ _-:-:_ _ _ _ _ _ _--,
[Jm(Amnr)]2r dr
presence of u •• we must include two more boundary conditions
I n= °+ 2
Lo "L0
to solve the initial boundary value problem. So that the Solu- P
(e) Integrate Hn(x) = { a: cos nO bn sin nO' g(r, O)Jo(Ao"r)r dr dO
tion is a smooth function, we require the "artificial" boundary
Aon =
.!:...(r dv _ zv) = Ip d(~) conditions
drdr Z dr
u(r, 1T, t) = u(r, -1T, t) and n=
2 {a, °
f" r
which correspond to the eigenvalues P.n = = I 2 .
°<
2
with respect to r to show that n, n " ...
ueCr, 1T, t) = u.(r, -1T, t) for r O.
The corresponding solutions ofr 2R" + rR' + (r2A2 - p.2)R = g(r, 6)Jm(Am"r)cos(mO)r dr dO
dv v(pv + 4)
r-=---- Therefore, we solve the problem 0, which we recognize as Bessel's equation of order n, are Amn = ....::0-0~--L-:P;;-----2- - - - ,
dr Z RnCr) = cnJ"(Ar) + d"Y"(Ar). Because IR(O)I < "', d" = O. Be- 1TAmnc 0 [Jm(Am"r)] r dr
c(u", + ;u, + r'z u••) u"
2 cause R(p) = O,J"(Ap) = 0, so Am" = am"lp, where a mn denotes
J'' r
using yeO) = 0. =
the nth zero of the Bessel function Jm(x). Then the solution is
(f) Use separation of variables to solve the first- u(p, e, e,
t) = 0, lu(O, 1)1 < "', -1T < e< 1T, I> 0 g(r, 0) Jm(Am"r)sin(mO)r dr dO
Bmn=....::o-'-o--~~-------
order equation obtained in (e) and simplify the
resuIt to yield v/(pv + 4) = - Cr2, where C is the
constant of integration. (Note: Because v de-
u(r, 1T, t) = u(r, -1T, I),
uk, 1T, 'ij = uk, -1T, I),
u(r, 0, 0) = fer, 0),
°
< r < p, I> 0
and
1TAmnc r
o
[Jm(Am"r»)'r dr
creases as r increases and v increases as r de- u,(r, 0,0) = g(r, 0), 0< r < p, -1T< 0 < 1T,
creases, v = r duldr < 0. Why?) Use technology to solve the wave equation in a circular region
and the displacement of the drumhead that is not radially sym- of radius p = I, which depends on 0 using the initial condition
(g) Show that u = (-Zip) In(l + pCr') + B, where metric. Assuming that u(r, 0, I) = R(r)H(O)T(I) and using sep- m." functions f and g given as follows. Use the formula derived pre-
B is an arbitrary constant. Use the boundary
°
condition u = at r = a to find B in terms of C.
aration of variables yields viously and plot an approximation of the solution for various
times to view the motion of the drumhead. (Assume: c = 1.)
Find that C satisfies the relationship 8C =
q(l + pCa 2 )2 by substituting the solution into the
2
c ( R"HT + ;R 'HT + r12 RH'T) = RHr
m,"
35. fer, 0) = sin 21Tr sin ZO, g(r, 8) = °
differential equation in (c). Let pCa2 = x and 36. fer, 0) = 0, g(r, 0) = (r - 1)sin 0
Then division by RHT gives us
8/(qpa 2 ) = a to obtain the quadratic equation
*37. fer, 0) = sin 21Tr sin 20, g(r, e) = (r - I)sin 0
X2 + (Z - a)x + 1 = 0. Show that if qpa S Z,
2 m."
+I +.l.. H")
then the temperature approaches a finite value.
c2(R"
R
R'
r R r2 H
= T" = -A2
T ' where 38. fer, 0) = cos ;-, g(r, 0) = °
Otherwise, an explosion results.
where - A2 is the constant of separation. Separating variables,
f" r fer, 6) Jo(Ao"r)r dr dO
39. fer, 0) = 0, g(r, 0) = (r - I) cos ""2
1T0
33. Solve Laplace's equation in a circular region of radius
1 given the boundary condition u(l, e). Graph the so-
lution using five terms of the series solution.
we obtain
and R" + 1...!t. + 1- H" = _}..?

R r R r2 H
aO
n
21T r
= ...:o::.......::o~:;;-_ _ _ _ _~
[Jo(Ao"r)]2r dr, 1Tr 1T0

40. fer, 0) = cos 2' g(r, 8) = (r - 1) cos ""2
(a) u(l, 6) =
(h) u(l, e) =
eO cos e
eO sin e
Separating variables in the second equation yields f"r fer, 0) Jm(Amnr)cos(m6)r dr dO
= ....::0-'-0---:r::-----2-----·
*41. fer, 8) = (r - 1)0, g(r, 0) = °
R" R' -H" a mn p 42. fer, 0) = 0, g(r, 0) =3
(c) u(l, 6) = eO(l + eO) r2R +r + r2A2 = ~= p.2, 1T L [Jm(Am"r)] r dr,
R o 43. fer, 0) = (r - 1)0, g(r, 0) = 3
34. Solve the wave equation in a circular region of radius
where p.2 is the constant of separation. Therefore, we have two
r = I that is radially symmetric using the initial con- more ordinary differential equations
dition functions that follow. Use the formula derived
previously and plot an approximation of the solution
for various times to view the motion of the drumhead.
H"+ p.2H= ° and
(a) fer) = r(l - r), g(r) = ° The boundary conditions in terms of Hand R become R(P) =
0, IR(O)I < "', H(-1T) = H(1T), and H'(-1T) = H'(1T). Recall
(h) fer) = r(l - r), g(r) =I that the problem
600
CHAPTER 10 SUMMARY
Chapter 10 Partial Differential Equations
1
.!
Show that the solution is

Chapter 10 Review Exercises
9. Use separation of variables with the steps (a)-(c) to

solve Laplace's equation (in spherical coordinates) in-
601
u(r, U> = aD + L>-n(an cos nO + b n sin nO) side the sphere of radius a (0 < 7T)
Concepts & formulas n=1
and find integral formulas for the coefficients. subject to the boundary condition u(e, </» = f( </»,
0<</><7T.
6. Use the results of the previous problem to solve
1Section 10.11 1Section 10.31 (a) Let u(p, </» = R(p)F(</». Separation of variables
l
u",+1. u,+-\-u ••
r r
=o, r>I,-7T<O<7T leads to the equations
Linear-Second-Order Partial Differential Equation

The One-Dimensional Wave Equation P'R" + 2pR' - A2R = 0
u(l, 0) = 10, -7T< 0< 7T
Au"" + Bury + Cu", + Dux + Euy + Fu = G(x, y) c2 u;a = UtI and
lu(r, u>1 < "', r -7 '"
sin </>F" + cos </>F' + A2 sin </>F = 0,
Separation of Variables *7. Consider Laplace's equation in a disk,
Substitute u(x, y) = X(x)Y(y) into the partial differential 1Section 10.41 where the separation constant is A2 •
l
equation and separate the variables to obtain two ordinary U" + 1. Ur + -\- u •• = 0, 0 < r < p, -7T < 0 < 7T (b) Substitute x = cos </> into
differential equations in the variables x and y. Laplace's Equation r r
u(p, U> =f(U>, -7T < 0 < 7T ' sin </>F" + cos </>F' + A2 sin </>F = 0
U;u+ uy'y= 0
lu(r, u>1 < "', r -7 0 and solve by using Legendre's equation with
1Section 10.21 with solution A2 = n(n + I), n = 0, 1,2, ....
The One-Dimensional Heat Equation 1Section 10.51 u(r, 0) = Ao + I rn(An cos nO + Bn sin nO). Evalu-
(c) Solve the Cauchy-Euler equation,
Ur= c 2 u;o:, where u(x, t) is the temperature at position x and n-l p2R" + 2pR' - A2R = 0,
timet. Laplace's Equation in Polar Coordinates 1 f~
ate u(O, 0) with the formula AD = 27T _~ f(O) dO. with A2 = n(n + I), n = 0,1,2, ... '.
Steady-State Temperature, Sex) U r ,. + 1- u,. + -\- U(JO = 0 Interpret this result in terms of the average value of a
r r (d) Express u as a series, and determine the unlmown
E..
~ u(x, t) = Sex)
Wave Equation in Polar Coordinates
function. coefficients using ute, </» = f(</», 0 < </> < 7T
8. In spherical coordinates (See Figure 10.21), Laplace's with the orthogonality condition of the Legendre
Transient Temperature, vex, t)
u(x, t) = vex, t) + Sex) 2(
CUr,.
~
1r ,r1)
+ -U,. + -YUO(J = Ute equation is polynomials.
10. Laplace's equation can be interpreted as the steady-
J....{~(p' au) + _ 1 ..i...(Sin </>~)
p' ap ap sin </> a</> a</>
state temperature. Using the formula derived in the
previous problem, f'md the steady-state temperature
1 a u}
2
+ sin2 </> ae> = O.

in a sphere of radius 1 if the initial heat distribution
on the surface of the sphere is (a)f(O) = 100,
CHAPTER 10 REVIEW EXERCISES IOO, 0 < 0 < 7T/2
Show that if u does not depend on 0, then this equa- 0< 0< 7r, (b)f(O) = { (0, 7T/2< 0< 7T .
tion becomes
11. In cylindrical coordinates (see Figure 10.22), La-
2 1 cot</>
u pp + - up + -, u.,., + - - 2- U., = O. place's equation is
1. Determine the value of e so that each of the fol- 4. The polynomial p(x, y) = A + Bx + Cy + Dx 2 + P P P
lowing functions satisfies the wave equation e2 uxx = E;ry + Fy2 satisfies the noohomogeneous equation Urr + .lur + -\-U(J9 + Uz:: = O.
r r
r + x 2; (b) u(x, t) = cos 4t sin x;
u". (a) u(x, t) =
(c) u(x, t) = sin t sin 16x.
Uxx + u", + U x = 10 for certain values of the constants
A, B, C, D, E, and F Find the restrictions on these con-
stants. Find a polynomial that satisfies the boundary
.P(x,y. z)
/'i
:
If u does not depend on 0, then this equation becomes
1
2. Determine the value of e so that each of the follow- /'
Ur,. + -;: u,. + u= = O.
(}3i~~
conditionsp(x, 0) = 0 andp(x, b) = O.
ing functions satisfies; the heat equation u, = e2 u=
(a) u(x, t) = e-' sin x; (b) u(x, t) = e- 16 n', cos x; 5. Consider Laplace's equation on the exterior of a disk: Use separation of variables to solve
(c) u(x, t) = e- 4' cos 9x.
l
u"' + 1. u, + -\- u•• = 0,
r r
r> p, -7T < 0 < 7T u", + ~ Ur + u= = 0, 0 < r < a, 0 < z < b
*3. Show that each of the following functions satisfies
Laplace's equation. (a) u(x, y) =;ry; (b) u(x, y) = u(p, 0) = f(O), -7T < 0 < 7T uta, z) = 0, 0<z< b
e 3x cos 3y; (c) u(x, y) = In(x2 + y2). [
lu(r, 0)1 < "', r -7 '" Figure 10.21 A point in spherical coordinates u(r, 0) = 0, u(r, b) = To, 0 < r < a.
J
(Hint: Use the separation constant k = _A2. The so-

lution involves the Bessel function of the first kind of
13. Suppose that w(x, y) and v(z, t) satisfy the equa-
tions w"" + Wyy = 0 and v= = J, VII' Show that
1
l
~
j
;\
2. Show that the transformation of the partial differential equation is

2
603
order zero. c d U - s 2U = -sm

---;J;2 . x,
u(x, y, Z, t) = w(x, y) . v(z, t) satisfies the wave equa-
Un + uyy + Uzz = ~ Uft.
P(~Y.')
tion in three dimensions
C with solution U(x, s) = Uh(x, s) + Up(x, s). Show that Uh(x, s) = C,e-
s
+ C2e s
14. The vibration of a cylinder can be found by Solving
1 .
d.,"!
the wave equation in a cylinder and Up(x, s) = S2 + I sm x.
u" + 1.. u, + u= =
r
J,
c
U/t, 0 < r < a, 0 < Z < b, t > 0 3. Apply the boundary conditions to show that C, = C2 = 0, so that U(x, s) =
1 .
~_~ __________:-=---...J u(r, 0, t) = 0, u(r, b, t) = 0, 0 < r < a, t > 0 S2 + 1 sm x.
u(a, z, t) = 0, 0 < z < b, t > 0
[u(O, z, t)[ < "', 0 < z < b, t > 0
4. Determine u(x, t) = ::e-'{U(x, s)} = sin x ::e-'L2 ~ I}'
Figure 10_22 A point in cylindrical coordinates u(r, z, 0) = fer, z), 0 < r < a, 0 < z < b The Laplace transform is particularly useful when the initial boundary value prob-
12. Use the results of the previous problem to tmd the u,(r, z, 0) = g(r, z), 0 < r < a, 0 < z < b lem involves the interval 0 < x < 00. For example, suppose that a long string is eX7
steady-state temperature in a cylinder of radius 5 and Use separation of variables to find u(r, z, t). (Hint: tended along the nonnegative x-axis. 0 :s x < 00. It is clamped at x = 0, and it slides
height 10 ifu(5, z) = 0, 0 < z < 10, and u(r, 0) = 0, The solution involves the Bessel function of the first along a frictionless vertical support at the right end. The string falls under its own
u(r, 10) = 50, 0 < r < 5. kind of order zero in the variable z.) weight. The initial boundary value problem that describes this situation is
e2~:~ ~;,
l
-g= O<x<oo,t>O
~', , , • Differential Equations at Work: u(O, t) = 0, lim

X-7 OO
~u
aX
(x, t) = 0, t> 0
I
A. Laplace Transforms u(x, 0) = 0,

au (x, 0) =
at 0, 0<x< 00,
The Laplace transform can be used to solve partial differential equations. For exam- where g is the gravitational constant.
ple, for the function of two variables, u(x, t), the Laplace transform is defined by
r
5. Show that the Laplace transform of the partial differential equation is
2
::e{u(x, t)} = e-stu(x, t) dt = U(x, s). d U g ,
e2 ---;J;2 - -; = s- U - su(x, 0) - u,(x, 0),
In addition, we have the following transforms of the partial derivatives 2 2

and apply the initial conditions to obtain d 2U/dx 2 - s2/e = gl(e s).
::e{ ~~ } = sU(x, s) - u(x, 0)
6. Determine the Laplace transform of the boundary conditions by verifying that
and ::eg; } = S2U(X, s) - su(x, 0) - u,(x, 0),
U(O, s) = ::e{u(O, t)} = ::e{0} = 0 and
which mirror the properties of ordinary derivatives discussed in Chapter 8.
1. Use the definition of ::e{u(x, t)} to verify that ::e{a 2ulax 2} = d 2U!dx 2. (Hint: As-
lim dU (x, s) = ::e lim
dx
x--+oo
f
lx--+oo
~u
aX
(x, t)} = O.
sume that we may interchange integration and differentiation.)
Consider the initial boundary value problem d2U S2 ~
7. Find a general solution of dx2 -? = e2s'
1
~:~ = ~;, O<X<7T,t>O 8. Apply the boundary conditions to determine C, and C2 ·
9. Compute u(x, t) = ::e- 1 {U(x, s)}. What is the piecewise defmition ofu(x, t)7
u(O, t) = 0, u(l, t) = 0, t> 0
10. Graph the solution using e = 1 ft and g = 32 fils 2 for several values of t. What is
u(x, 0) = 0, ~~ (x, 0) = sin x, 0<x< 11" lim u(x, t)?
HOO
1, Differential Equations at Work 605
B. Waves in a Steel Rod With the change of variables

Piston flow
_ n x liL
Just as we considered the displacement of a plucked string in Section 10.3, we may in- n =-, X= -L,PeB =-E' and
no =
vestigate the waves fonned when stress is applied to a steel rod of length L. In this
Turbulent flow where E z is the axial dispersion coefficient, no is the initial concentration, L is the pipe
case, E, Young's modulus, is given by E = 0-1 e, where CT is the lateral stress applied to
the rod and e is the axial strain. Also, P is the density of steel. The partial differential I
===;;;;;!.== it; (O.82)u max length, PeB is the Pe'clet number (or Bodenstein number), x is the axial direction, and
equation that describes the situation is a fonn of the wave equation, t is the nominal holding time, we can rewrite the differential equation as
a
2
u 2
u a Viscous flow
it; (O.5)u m", a2n _ PeB an _ an = o.
E ax2 = PaT'
ax2 ax a</>
where u(x, I) is the longitudinal displacement of a cross-section of the rod located x Figure 10.23 Distribution in-
units from x = O. (Nole: We have assumed that E and P are homogeneous and do not side round pipes of velocities in If the initial conditions are
depend on x.) Longitudinal waves propagate through the rod at a velocity of Co, called fluids exhibiting different types
of flow </> = 0, X > 0, n = 1
the bar velocity, given by Co = VE/P.
For steel, this value is Co = 5.1 x 103 mls. For { </> = 0, X < 0, n = 0
air, this value is much larger (250 mls). Suppose that the rod is compressed at one end.
and the boundary conditions are
1. Assuming that the solution of the panial differential equation is of the fonn
u(x, I) = I (an cos wnl + bn sin Wnl)Un(X),

X ..., 0+ , aanX - PeBn- = 0 ,
n=l
( an
use the free end conditions X -41, ax = 0,
au
ax (0, t)
au
= ax (L, I) = 0 then use separation of variables to find n. •
to detennine Un(x).
2. What are the natural frequencies, wn ? D. Numerical Methods for Solving Partial Differential Equations
3. If the steel rod is 0.25 m long; what is Wn and what is the frequency (in Hertz)?
4. Graph Un(x) for n = 1,2,3,4. When U.(x) > 0, particles move from left to right, Here we discuss methods used to approximate the solution to partial differential equa-
and when U.(x) < 0, particles move from right to left. Indicate particle movement tions by replacing partial derivatives with finite difference approximations. We begin
with arrows on the four graphs. with a discussion of the classification of second-order partial differential equations and
then present numerical methods for approximating the solution to equations in each
category.
C. Media Sterilization
Classification of Second-Order Partial Differential Equations
In designing a continuous media sterilizer in which the raw medium passes through
round pipes, we must note that not all portions of the medium spend the same length A second-order partial differential equation of the fonn
of time in the holding section of the sterilizer. This is because the mean velocity li of
the fluid is a function of the radial distribution of fluid velocities occurring across the
Au", + 2Buxy + CUyy = F(u" uy , u, x, y)
pipe. The mean velocity of a viscous-fluid flow through a pipe is one half the maxi- is elliptic if AC - B2 > 0; parabolic if AC - B2 = 0; and hyperbolic if AC - B2 <
mum velocity found at the axis of the pipe (radial distribution of velocities is para- O. For example, Laplace's equation, u'" + Uyy = 0, is elliptic. With the independent vari-
bolic). In the turbulent condition, the mean velocity is 82% of the maximum value. ables x and I, the heat equation, u, = c 2u"-,, is parabolic and the wave equation, u" =
These situations are illustrated in Figure 10.23. 2
c u"-,, is hyperbolic.
The material balance is then given by the partial differential equation
a2n _ - an _ an
E z ax2
oft Source: S. Aiba, A. E. Hwnphrey, N. F. Millis, Biochemical Engineering, Second Edition, Academic
u ax - al' Press. New York (l973), pp. 258-267.
Numerical Solution to Elliptic Equations [Noles: The mesh size is 2/3, and (0, 0) is the lower left-hand corner. Find u at the
boundary points UOb U02, UIO, U20, Un, U23, U310 and U32 with the functions fey) = 0,
Consider the boundary-value problem that solves Laplace's equation over a rectangu_ X, 0 <x < I
lar region g(y) = y(2 - y), 0 < Y < 2; and hex) = 0, 0 < x < 2; k(x) = { 2 - x, I OS; x < 2'
UI2 "22
UXX + uyy = 0, 0 < x < a, 0 < Y < b (See Figure 10.24,) Solving
U(O, y) = fey), u(a, y) = g(y), 0 < Y < b
[
Ul! U21
Ui+l,j + U",j+l + Uj_I,} + Ui,j-l - 4uij = 0
u(x, 0) = hex), u(x, b) = k(x), 0 < x < a.
for uij, we find that
A problem of this type is called a Dirichlet problem because the value of U is given
around the boundary of the region in the boundary conditions. Figure 10.24 Values of u at uij = 4'1 (U'+I,j + U',j+1 + u'_I.j + U,.j_I)'
Two central differences for the function u(x, y) are u(x + h, y) - 2u(x, y) + the mesh points
u(x - h, y) and u(x, y + h) - 2u(x, y) + u(x, Y - h). Therefore, the finite difference Start by using the boundary conditions to define the values of u on the boundary.
approximations for the second-order partial derivatives U xx and uyy are Graph the solution.]
1 2. Repeat the method using n = 4.
Uxx = 11 [u(x + h, y) - 2u(x, y) + u(x - h, y)]
and Numerical Solution to Parabolic Equations

1 .',
uyy = 11 [u(x, y + h) - 2u(x, y) + u(x, Y - h)]. We develop this approximation method using an initial boundary value problem thaI
involves the heat equation,
Adding these equations, we obtain an approximation to the Laplacian Ut 0 < x < p, I> 0
= e 2uxx ,
1 u(O, I) = Tio u(p, 0) = T2 ,
I > O.
Uxx + uyy = h2 [u(x + h, y) + u(x, Y + h) + u(x - h, y) + u(x, Y - h) - 4u(x, y)], [
u(x, 0) = f(x), 0 < x < p
called a five-point approximation. This means that Laplace's equation can be ap- Substituting the central difference approximation
proximated by the difference equation
1
Uxx = 2' [u(x + h, y) - 2u(x, y) + u(x - h, y)]
u(x + h, y) + u(x, Y + h) + u(x - h; y) + u(x, Y - h) - 4u(x, y) = O. h
Using the notation u(x, y) = uij, u(x + h, y) = U'+I.j, u(x, Y + h) = u'.j+1o u(x - h, y) and the forward difference approximation
= u,_I.j, and u(x, y - h) = U',j_ 10 we obtain the equation
u, = k1 [u(x, I + k) - u(x, I)]
Ui+l,i + Ui,j+l + Ui-1.j + Ui,j-l - 4uij = O.
To make this topic more understandable, suppose that we take the rectangular re- into the heat equation u, = e2uxx , we obtain
gion and divide it by using lines parallel to the x- and y-axes. These lines are called
2 I
grid lines. If these parallel lines are h units apart, then we say that the mesh size is h. ~2 [u(x + h, I) - 2u(x, t) + u(x - h, I)] = k [u(x, 1+ k) - u(x, I)].
The points of intersection of these lines are called mesh points. The goal of this nu-
merical method is to use the equation given above to approximate u at the mesh points. With A = e2k/h 2, ui} = u(x, t), U'+I,j = u(x + h, I), Ui-I,j = u(x - h, t), and U'.j+l =
\. Use the five-point approximation u(x, I + k), this approximation to the heat equation becomes
Ui+lJ· + Ui,J+l + Uj-l,j + Ui,j-l - 4uij = 0, U',j+1 = AUi+I,j + (1 - 2A)uij + AUi-l,i
with n = 3 to approximate the solution to the boundary value problem Suppose that we want to approximate the solution over the rectangular region O. os; x os;
p, 0 os; lOS; T, where T is a finite value of time. To partition this rectangular reg~on, w.e
Uxx + Uyy = 0, 0 < x < 2, 0 < Y < 2 use a partition size of length h = pin on the x-axis and k = Tim on the I-aXIS. This
u(O, y) = 0, u(2, y) = y(2 - y), 0 <Y< 2 means that the grid lines perpendicular to the x-axis are x, = ih (i = 1, 2, ... , n) ~nd
X, 0 <x < 1 those perpendicular to the I-axis are 0 = jk, (j = 1, 2, ... , m). The procedure begms ,
u(x, 0) = 0, 0 < x < 2; u(x, 2) = { 2
1 2 - x, los; x < by defining the boundary conditions
1
~
'i
(j + l)st UOj = u(O, t) = TI and Unj = u(p, t) = T2U = 1, 2, ... , n) I Ut = (1/4)uxx> 0 < x < 2, 0 < t < 3/10
Ui,j+l
time line 'i
and the initial condition u(O, t) = 0, u(2, 0) = 0, 0:$ t:S 3/10
jth
time line
UiO = u(x, 0) =I(x). ru(x, 0) = sin TrX, O:S x :$ 2
using the Crank-Nicholson method with n = 8 and m = 30.

After these values are deImed, we use
Figure 10.25 Values of u at Ui.j+1 = AUi+I.j + (1 - 2A)uij + AUi-l.j Numerical Solution to Hyperbolic Equations
the mesh points Consider the initial boundary value problem
to approximate the solution. We depict this calculation in Figure 10.25.
3. Approximate the solution to e2Uxx = UtI> 0 < X < p, t > 0
u(O, t) = 0, u(p, t) = 0, t> 0
Ut = uxx, 0 < x < 1,0 < t < 112 [
u(O, t) = 0, u(1, 0) = 0,0:$ t < 112
u(x, 0) =I(x), ut(x, 0) = g(x, 0), 0:$ x :$ p,
r u(x, 0) = sin TrX, O:s x:$ 1, which has a unique solution if I and g have continuous second derivatives on the in-
tervalO < x < p and1(0) = I(p) = 0, In this numerical method, we use the central dif-
using n = 5 and m = 50. (Notes: In this case, e = 1, P = 1, T = 0.5, n = 5, m = ference approximations
50, h = 115, k = 1/100, and A = 114.
1
Uxx = };'i [u(x + h, t) - 2u(x, t) + u(x - h, t)l
Crank-Nicholson Method
A popular method for solving the heat equation is the Crank-Nicholson method, an and
implicit finite difference method developed in 1947 by 1. Crank and P. Nicholson. This
1
me~od requires that we solve a system of equations at each level of the method to ap- Uti = ~ [u(x, t + k) - 2u(x, t) + u(x, t - k)j.
proxunate the value of U at the mesh points. This method differs from the previous
method in that Uxx is approximated with the average of two central difference quotients, Therefore, e 2 uxx = u" becomes
one which is evaluated at t and the other at t +: k, given by
=.!. [ u(x + h, t) - + u(x - h~ t)

2
~2 [u(x + h, t) - 2u(x, t) + u(x
1
- h, t)l = ~ [u(x, t + k) - 2u(x, t) + u(x, t - k)]. I
U
xx 2
2u(x, t)
h2 I
I
With the substitution A = ek/h, we have
+ u(x + h, t + k) - 2u(x, t + k) + u(x - h, t + k) ] Ui,j+1 = A2Ui+I.j + 2(1 - A2)uij + A2Ui_I.j = Ui,j-I
h2 •
I
for i = 1, 2, . . . , n - 1 and j = 1, 2, . . . , m - 1. In this method, we use n
With this substitution, we have gridlines perpendicular to the x-axis of a partition size h = pin and m gridlines
~ [ u(x + h, t) - 2u(x, t) + u(x - h, t) perpendicular to the t-axis of a partition size k = Tim. (Note: m and n are positive
integers.) Therefore, the gridlines are located at Xi = ih (i = 0, 1, 2, ... , n) and 0 =
2 h2
jk (j = 0, 1,2, ... , m). The recursive equation
u(x + h, t + k) - 2u(x, t + k) + u(x - h, t + k) ] 1
I
Ui,j+1 = A2u,+l.j
(j+ l)sl
+ h2 = k [u(x, t + k) - u(x, t)], U~j+ I time line
+ 2(1 - A2)Uij + A2ui_I.} - Ui.j-I
j th makes it possible for us to calculate Ui.j+ I on the (j + l)st time line from the values
which when using the notation introduced earlier and the values A = e 2 k/h 2 , a = UiJ time line
Ui-I,j, uij' and Ui+I.) from the jth time line and the value of Ui,j-I fro~,the (j - l.lst
2(1 + lIA), and f3 = 2(1 - lIA), becomes U,_I,j Ui+ 1,1'
time line. (See Figure 10.26.) In the process, we use the boundary conditIOns to defme
UOj = u(O, jk) = 0 and un} = u(p, jk) = 0, as well as the initial condition to define
1-----1 (j - l)sl
UiO = U(Xi' 0) =I(x;). We also make use of the initial velocity. Whenj = 0, we have
for j = 0, 1, ... , m - 1 and i = 1,2, ... , n - 1.
h lime line
the equation I
4. Approximate the solution to
Figure 10.26 Values of u at
the mesh points I
I
I
~
To calculate Ui.-I> we use u,(x, 0) = g(x), because at t = 0,

Answers to Selected Exercises
U(Xi, k) - U(Xi, - k)
g(x;) = U,(Xi, 0) = 2k
so that U(Xi, - k) = U(xi, k) - 2kg(x;) or
Ui.-I = Uil - 2kg(Xi).

Therefore, we obtain the equation when} = 0 by substituting Ui.-I = Uil - 2kg(Xi) into
CHAPTER 1
Uil = A?Ui+I.O + 2(1 - A2)UiO + A2Ui_I.O - Ui.-I.
Exercises 1.1 29 . 2x + 2y dy = 0 =; dy = -::.. 02 + y' = 16 =; y = ±4;
This gives us dx dx y'
1. (a) ordinary; (b) second-order; (e) linear
(0, ±4)
3. (a) partial; (e) linear
5. (a) ordinary; (b) first-order; (e) nonlinear
5. Approximate the solution to the initial boundary value problem 7. (a) partial; (e) linear
9. (a) ordinary; (b) second-order; (e) nonlinear
4U= = Utt , 0 < X < I, 0 < t < I 11. (a) partial; (e) nonlinear
u(O, t) = 0, u(I, t) = 0, 0 s; t s; 1
[
u(x, 0) = sin 7TX, u,(x, 0) = 0, 0 S; x s; I, 13. (a) ordinary; (b) first-order; and (e) linear in
with n = 5 and m = 20. y( ~ = 2x - y); nonlinear in x(: = 2x ~ y)

15. (a) ordinary; (b) first-order; (e) nonlinear in
y(~
y
= 2x; y), nonlinear inx(: = 2x - y) dy dy 3x 2 + 2xy
31. 3x 2 + 2xy + x 2 dx = 0 =; dx = ---x-2- ;
17. ~ + 2y = (-2e-2<) + 2e-2< = 0 13 + Y = 100 =;y = 99; (I, 99)
19. ~ +Y = ( -e -x + I x+ I x) +
sin cos
I I . ) .
( e-
x
- 2" cos x + 2" sm x = sm x
2
d y
21. dx2
+ 9 dy
dx
= 81Be- 9x + 9(-9Be- 9x ) = 0
23. : = (- A + ±-t) sin t +

-20 -10
-10
10 20 x
(B + ~ - 1+ ±) cos t;
ddt x (-A + li4 - I) cos t +
2
2
= y dy
33. -+-Inx+cosy-xsmydx=O=;
x dx
. dy
(-B - ~ + 1) sin t
. dy Y
(In x - x smy) dx = -cosy - X =;
3
d2
dy _ 2B 2< _ 2C -2<.2 = 4Be2<
25. dx - e e , dx2
+ 4Ce-2<·' -ddxy3 = d
Y
cosy+-
2:. = . x ; y In I + cos Y = 0 =;
8Be2< - 8Ce-2< dx xsmy-Inx
6
27. x2(30Ax4 + 42Bx') - 12x(6Ax' + 7Bx ) + cosy= 0 =;y= (2n ~ 1)'1T, n = 0, ±I, ±2, ...
42(Ax 6 + BX7) = 0
A-I
A-2 Answers to Selected Exercises Answers to Selected Exercises A-3
57. y(x) = -sin(~) + 2 77. Exercises 1.2

10 1. no
2. yes
1 l.!i 2 2.5 3 x
-5
-10
3. yes
4. no
4x' - 5y + 4xy2
79 . (a) dy
dx
=
-5x + 4x'y + 4y' 5.
0.5 (b)
I
35. y(x) = -I COS(X2) + C
37. y(x) = In lin xl +C
39. y(x) = -xe-x - e- x +C 59. vet) = mg + e- ct/m (_ mg + vo),lim vet) = mg -2
C C t-¥'» C
x - x2
41. y(x) = f (x + 1)(x2 + I) dx 61. first-order,. nonlinear
f [- ~ 1+ ~ I] dx
-2
= x x2
63. x(O) = 3; : = -12 sin 41 + 9 cos 41, : (0) = 9
= -In Ix + t[ + tan - I x + C (partial fractions) (e) (1,0.21979), (I, 1.20684)
(d)(-1.l0797, -0.319), (-0.00649774, -0.319)
43. y(x) = f (4 - X
2 )'12 dx = {(4 - x 2 )'/2 + 65. Ut = 16ke- '6t
cos 4x, U= = 16e- 16t
cos 4x; U(7T, 0) = 2;
81. Y = e-
xl2
( ;;4 cos 3x + ;2 sin 3X) +
7.
limH~
3; ~ + 6 sin- ' (f) + C (trig. substitution;
67. m = I, m
u(x, t) = 3
=; 2 eXn. (
37 cos 3x +
-"'104 111.)
208 sm 3x
x=2sin6)
45. y(x) = 2e- a 69. y(x) C" e- X + Ce-a
47. y(x) = te4x -te-'x .dljJ(~)

71. Because --;;;- = LAn7T cos (nm:)
L and ~ =
2
d 1jJ(x) 2.5
49. y(x) = "3I sm

.
3x
-A: 2
n"
2
2
sin (n;x).
2
-2.5
-5
-7.5
1 2
2
h An n" . (nm:)
-f -? + fe a
h d 1jJ(x) -10
-2m ~+ U(x)I/I(x) = 2m ~sm L =
51. y(x) =
53. y(x) = 8x 6 - 7x 7
EA sin (n;x). 9. (a), (b) !~ y(x) = 7T/2; (e), (d) !~ y(x) = -7T/2
h 2n 2 n"
1 Therefore, E = 2mL 2 •
55. y(x) = X4 - Ix2 + 2x + I
2x x 2 dy dy y2 + 2xy x2
+Y -
'l~ :
73. 1 y2 dx = 0 => dx = --x-2- ; Y = C - x
so y = 0 is singular
75.
0.2 0.4 0.6 0.8 1 t
4 x
3 4 5 .x
11. y = 1 + Yo - e-X,!~ (yo + tan-I x) = Yo + 7T/2
J
A-5
A-4 Answers to Selected Exercises Answers to Selected Exercises
4x
26. 13. y' = 3Ae 3x + 4Be4x and y" = 9Ae 3x + 16Be so
fx'= y 3x 4x
19. y" - 7y' + 12y = 9Ae 3x + 16Be 4x - 21Ae - 28Be
lY'= -16x + sin ( + 12Ae 3x + 12Be4x + 2 = 2
y 15. (y cos xy + sin x) dx + x cos xy dy = 0 :}
,~-:6'_~ __ , dy sin x + ycosxy
___ -4. dx = x cosxy
dY ). dy
// cosxyxdx+Y
( +slDx=O:}dx=
-6-'~-4--2 _··'2'--4 .6-...f
_.. --.". --- '-2' - .. .-"/ -"- sin x + y cosxy
x cos xy
13. (a) f~ X(I) and E.'!! y(l) do not exist; (b) E.'!! X(I) and ,-4 ... '_ ::4··-·--·---
E..'!! y(l) do not exist; (e) f.!..'!! x(t) = E;!;,y(l) = 0 ,'~-, --.. ------~ 17. Y = _x 2 cos X + 2 cos x + 2x sin x + C
Chapter 1 Review Exercises 19. Y = !.. v?-=! + lln(x + v?-=!) + C
1. (a) ordinary; (b) first-order; (e) linear 2 2
3. (a) ordinary; (b) second-order; (e) linear
21. Y = '4x + cos 2x
5. (a) partial; (e) nonlinear
'
25. (a) lim x(J) = 0307392, lim y(l) = 0384615
t-+oc ." (-+00
7. Ix = 2(e<
d 1
+ sin x - cos x) so 23. Y = -+ 3
cos x + +
dy _ Y _ sin x = ..!.(e X + sin x - cos x) -
dx 2
..!.(eX - cos x - sin x) - sin x = 0
M
0.8 0,.
2 0.'
t:: ~4x"
0.4
~ d~ 0.3
15. 0.• 9. - = -3e 3«cos 6x + 3 sin 6x) and - 2 = 0,2
dx dx 0.1
0,4 - ge 3x (7 COS 6x + sin 6x) SO 27t 3rt 47t x
3x
y" - 6y' + 45y = -ge (7 COS 6x + sin 6x) +
0.2
18e 3X(COS 6x + 3 sin 6x) +
45e 3x(cos 6x - sin 6x) 25. du = -70ke- kt ; u(O) = 100; lim u(l) = 30
=0 dt t-'l>ct:>
0.2 0.4 0.6 0.8

2y 27. ut(x, I) = -"t-ke--.'kt sin m + 4"t-ke- 4 -.'kt sin 2m
o dy = 5Ax' _ 3B,; d , = 20Ax3 + 12B,
l·dx xdx x uxx(x, I) = -"t-ke --.'kt sin m + 4"t-ke -4-.'kt sin 2m
2 2 2[2 cos(1n x) + sin(1n x)] +
11. x y" + 3xy' + 2y = x . x3 u(l, 0) = 0; f.!..'!! u(x, I) = 0
(b) The species with population yet) becomes extinct. 2xy -2xy
3x' -2 cos(ln x) + 2. cos(1n x) - sin(1n x) = 0
x2
29. uxx(x, y) = (x 2 + y2)2; uyy(x, y) = (x' + y2)'
fX'= Y X
17. lY'= -13x-4y'
0,8
CHAPTER 2
0 .•
0.4
Exercises 2.1 5 . 5y - _9_
7y7
+ 6x + X4 = C
0.2
1. y' =i x3 +C 7. cosh 4y = 2 sinh 3x + C
3. J..=~+C 9. _ _
2_ + lx 3 + !x 312 = C
0.2 0.4 0.6 0.8 I x y. 7X7 3y 312 3 3
J
11. -3 cos x = 4 sin y +C

43. y(x)=~
x-2
6
5x 4'
13. """6 - sm x = -"92.sm 9y 2
+ "7 cos 7y + C 47. (c)
15. 20 cosh y = -t sinh 6x - %- cosh 4x + C

10
49. (b)
51. y = 0, stable
17. lOx - fe- 3X
= eY - 2y4 +C 53. y = 0, unstable; and y = 2, stable
19. - 3 cos x + teas 3x = i sin 4y - 4 sin y +C 63. x~ - 9 sin-I (x/3) + eY(cosy + siny) = C
10 x i I
2 ......:..-:......~..!......-..!...-...:...-
21. 2y + 2. + sin 2x + 2x = C
23. t tan' y =
Y
t cos 2x
4
+C
37. ty' +Y = eX - 1 so Y = -1127=1 - I
-~--+----1-- .....
1 2 3 4:.x
25. -t cos(x') = 2 sin vY + C -2

6 x -I
27 I I. 4 +C 55. Y = I, stable; and y = 4, unstable

. l-siny ="4 sm x 65. y = eC-(cJx)
-8
-10
31. y(x) = iX4 + 4 4
3 .
-_-:::--_-:::--~,~---
I
y(x) 39. -(lny)'
2
= x ."+ -21 so y = e .v2x+1
r.:--7"7 2 •
3 4 5 6:C
57. N(O) = No => N(t) = Noe- kt; D = (In 10)/k = Exercises 2.2
2.30259k- 1
1. Y = lx, + Cx- I
3
58. (b) C = e""o (1 - No!::..)
-4 -,
No kd 3. y = eX: - x-leX + Cx- 1
5. y = In(l + x') + C + x' 1n(1 + x') + ex'
33. y(x) = Inlsec x + tan xl + 2 Ao(Bo - Yo) (K(Bo - Ao)t)
Ao - Yo exp V' - Bo 7. y = v?"=l(2v?"=l + C)
10 Y 59. (b)y(t) = ~::c..."':"::'_~-,-_-,--:-'-_
7.5 41. y(x) = tan- t x +I Bo - Yo
Ao - Yo exp
(K(Bo - Ao
V'
)!) - I
9. Y = -te-2x + te-2x tan x + C sec x
'.5 lino Y (t) = Ao;
11. y = _x 3 + 4x + (x' - 4)312 lnlx + V"?""=41 +
x=-1Cf2 C(x' - 4)312
(e) ~(0 = e-B"KliV,(yo - Ao + AoeB"Kllv,), lim yet) =
-1 -0,5
-2.5
-5
05 I Ij x
Ao; t---+«>
13. Y = tx4 + ix3 - 2; + cY 4x' - 9
-7.5
(d) The approximation improves for larger values of Bo.
-10
61. y= sin(C- cos x)
15. Y = x' + ~ + CY9x' + 49
J
A-8 Answers to Selected Exercises
17. y = (ex' 12x 2 - 2e x'/2 + C)e -%'12 = x2 - 2 + Ce -x'12 33. x = -2 + 4e' - t

31. x = vy => dx = v dy + Y dv => dx
dy
= v + y dv
dy
19. x = _y2 - 2y - 2 + CeY v + Y dv = le- v + v => ve v dv = 2dy => ve v _ eV =

dy v y
os
21. x = 3y2 + Cy 21nlyl + C
'
eXIY(~ - I) =
20
21nlyl +e
23. P = '!'t 4 + Ct 15
3 10
25. Y = 4e X (x + 1) 33.
y2 = .!.x2(1
2
+ x2) ory = x~
V2
-I -0.5 0.5 1 (,5 2 r
+k- + xe X)
lL
65. Y = e -x(1 eX
80
60
37 . y=e -x + Scosx+SSlnx+
2 I. Ce -2x
40
20
39. (a) -4 4 X
-0.5 0.5 I 15 2.:c
_ {2e- X
+ x-I, OSxS!.
43. Y - -x > 5 x
2e." x 1
-2
2.'
45. y = {eee-
~2x, 0 x
X, x>l 4
S S 1
35. Y = l(x 2 - 1)
2
1.5 Exercises 2.3
1 47. y(x) = Ce x
- ~5 cos 2x - .!.5 sin 2x
0,'
1. y2 = -2x - 2 + Ce x or y = V Ce x - 2x - 2
-I -05 0.' 1.5%
49. y(x) = Ce~x + xe-x 2cosx+2xsinx+ C
29. Y = x- l (1 + sin x) 3. y2 = x
51. y(x} = Ce
5x
;-1- 2~ 9 3,
-05
5. y
3/2_
- - 20 cos x + 20 Sin X + e3x
e
53. y(x) = Ce x12 + 4e x - 2 cos x + 4 sin x
= -lnlxl + 9 or y = ,J~
7. yes, n = I y3
37. -3 ~ 3x(9 -lnlxO'/3
55. y(x) = Ce- IOx + ~ex 9. no 3x
11 11. yes, n = 0
57. y(x) = Ce x + 2xe x 13. no 10
15. no
rcJ2 31t12 21t % 59. y(x) = Ce-x + cosx+ sin x +x-1
17. -lnl~ - II + 21nl~ - 11 = -Inlyl + e
x2 2 61. (a) yet) = -kr (1 - e kt ) so lim yet) = -kr; (b) yet) =
31. Y = 2(e X + 1) + eX + I H~
19. '!'X 2y 2 - xy3 = C

1 2
4 Y
I - -2 (e-' + cos t - sin t) so lim yet) does not exist;
,-- 21. y3 = 3x 3 + Cx 3
Inlxl 4 X
35 the concentration approaches a periodic state; (c)y(t) =

te-' so t~y(t) = 0,
23. y = -x lnlxl + ex
39. Y = ux => y4 dx + (x 4 - xy3) dx = 0 =>
25 5/6 ~ (UX)4 dx + (x 4 - XU 3x 3 )(u dx + x du) = 0
63. Y.:c = e-x(I - eX + xe X), Ysinx = 25. x = (3y _ 2x)2/3
1 u3 - 1
1.5
I -xC - I +e X cosx-e X ·) 27. Y = Cey/{4x) => x 4u dx + x'(1 - u3 ) du = 0 => -; dx = --u- du
-"'2e SlnX'YC05X=
I
0,5
t e - X (-1 + eX cosx + eX sin x), Yc' = te-X(-I +e
2x
)
29. xy - '!'y2 = C => lnlxl = '3u3 - Inlul + C
-2 -1
,2
J
A-l0 Answers to Selected Exercises
l' Answers to Selected Exercises A-ll
I 11. Y =x 3 + C 37. xy2 + cos 2x + Y = 2

~ Inlxl = 3(Y/X)3 -Inly/xl +C
13. xy2 = C
I
y(l) = 2 ~ 0 = 3' 8- In 2 +C ~ C = 15. x 2 + xy3 + 4y = C
I
3(3 In 2 - 8) = -1.07352 17. x 2y + 13 y 3 = C
(
19. x sin2 y =C '
-2
21. 1= c+ cx- 1
+x- 1
y -4
si~ x, 0 =C
1."t=
63. (a) Si'(x) = 23. y2 COS(X2)
7
25. x + y sin(xy) = C
-2
27. (3 + x) sin(x + y) = C 39. !x(x, y) = _1_2 - y2 ~ f(x, y) = tan- 1 x _ xy2 +
I+x
-4 1.5
29. y= -x1nlxl + Cx g(y) ~ hex, y) = -2xy + g'(y)
1.25
tan-I x _xy2 = C; tan- 1 0 - 0.0 2 = C~ C= 0;
31. x 2y2
1
0.75 =I tan- 1 X-xy2 = 0
0.5
41. y2 = 2x 2 Inlxl 0.25 .'~
~~
+
02.5 5 7.5 1012.5 IS 17.5 ~ 2
43. (d)
(b) - ,
r'
45. (c)
49. (e)y = _2_x 312

3\13 -6 -4 -2_
-2,
1
6 x
-) -1
47. xy = C
-2
2
51. f(x) = x - I; g(x) = x - x -3
general solution: (xc - y) - I = c 2 ~ Y = ex - I
49. p.(y) = exp(J ;dY ) = exp(lnlyl) = y, y > 0
- C
x3+ I
+ c - c2 33. y = - x 2 - 1 y2 dx + (2xy - y'eY) dy = 0 ~ !x(x, y) = y' ~
(e) ~lny + Si(~) = -In 2 + Si(t) f(x, y) = xy2 + g(y) ~ hex, y) = 2xy + g'(y) ~
singular solution: 1x [xy' - y - I] = 1x [(y')2 - y'] ~ g'(y) = -y 2 eY ~ g(y) = -y'eY + 2ye Y - 2e Y ~
xy" + y' - y' = 2y'y" - y" xy2 _ y'eY + 2yeY - 2eY = C
x+1 1 2
(x - 2y'+ I)y"= 0 ~ y' = -2-~ Y ="4x -+ 51. _2'. + 2x + 1 y' = C
x -3 -2 -I
-2
I 2 3' x 2
2 4
53. f(x) = I - 2x; g(x) = x- 2 ; general solution:

-4
if 53.
y
C~-I
x
2
1 - 2(xc - y) = c- 2 ~Y = t(c- 2 + 2eX - I); singular
35. xe Y - x y = 0 57. e'Y - x + Y + sin(xy) = C
solution: y = t(3X 2/3 - I) Exercises 2-4

1. exact
57. Y = (Ce- P - 2)(p + 1) + (2p + I) 3. exact
59. z = H-t + ce 2V )
5. exact
7. not exact
-2
61. x 413 + y4/3 = C 9. exact
1
~ Answers to Selected Exercises A-13
.~ af I 27. Ixl < a

+y 13. f(x,y) = vy2 - I and ay (x, y) = I(y2 - 1)-ll2.
59. x sin(cy) cos(cx) = I
3 I I 3)
( graph for c = -2, -2' -I, -2,0, 2' I, 2' 2 Y
2y = Yy2 _ I ; unique solution guaranteed for (a)
29. general solution: t sin2 x - sin y cos x + t 2
sin y = C;
.. "f 7/" 37/"

solution IS not unIque 1 Xo = 2 or 2'
only.
l 15. (0, (0)

17. (0, (0)
19. (-00, I)
4 6 21. (-2,2)
( 23. x> 0, Y = sin x - cos x, -co

x
< X < +00
Exercises 2.6
46 X _6_4_2246.(-6-4 4 6
-6 -4 -2
-2 1. 47.3742,63.2572 13. 185.34, 206.981
-4 3. 1.8857, 2.0984715. 15. 1.95547, 1.95609
-6 5. 79.8458, 123.048 17. 90.6405, 90.6927
7. 1.95109, 1.95388 19. 2.43501,2.43514
y
9. 83.6491, 88.6035 21. 216.582,216.992
J 6~ ~ 11. 2.37754,2.41897 23. 1.95629, 1.95629

25-27. (a)y(t) = e-',y(l) = 1/e = 0.367879
2 4 6 x h = 0.1 h = 0.05 h = 0.025
25. (Euler's) y(l) = 0.348678 y(l) = 0.358486 y(l) = 0.363232
26. (Improved Euler's) y(l) = 0.368541 y(l) = 0.368039 y(l) = 0.367918
27. (Fourth-order Runge-Kutta) y(l) = 0.36788 y(l) = 0.414831 y(l) = 0.429069
29. y(0.5) = 0.566144, 1.12971, 1.68832,2.23992,2.78297
7. Yes. y = e 2 (x 31Z -1)/3

Exercises 2.5
5. y5=e5x +C
1. (a) yes; (b) no; (c) no . af
9. Yes.f(x, y) = sm y - cos x and ay (x, y) = cos yare 7. l(ln y)2 + le -2< = C
f liS af ( ) I -415' 2 2
3. In this case,. (x, y) = y , so ay x, y = Sy lS not
continuous in a region containing (7/", 0).
continuous at (0, 0); uniqueness is not guaranteed. 0.5 9. J....eY cos 6y + .§....e Y sin 6y =
37 37
_J....
20
cos lOx
Solutions: y = (~x)'14, Y = O. 11. Y = sec x => y' = sec x tan x = Y tan x andy(O) =
af 7/" 7/"
1234567 X
I
sec 0 = I; ay (x, y) = x is continuous on -2 < x < 2 +Scos4x+C
5. Because f(x, y) = 2 y Iyl =

_,r;-; {2vY,
2V=;;,
y 2:
y
0
< 0' 7/" 7/"
~ {y-~ y > O . .
Y < 0 lS not contmuous at
andf(x, y) = sec x is continuous on -2 < x < 2 so
1. y' = tx6 + C
11. _lx 2 + xy
2
+ ly2 =
2
C
ay (x, y) = _( _y)-lI2, the largest interval on which the solution is valid is 13. y2 +X = C
(0, 0). Therefore, the hypotheses of the Existence and 7/"
-2<x<T
7/" 3. sinhy - t cosh x = C 15. Y = C(y2 - 4x)512
Uniqueness Theorem are not satisfied.
Answers to Selected Exercises A-15
17. t x 3y - Cos x - siny = C 27. Y = ex + 2 In c; Sing: y = 2 In( ~2) - 2, x < 0

CHAPTER 3
19. x 2 In Y + 2y = C 8 _ 4 _
21. Y = 1 + Ce-x'12 29. x = "3 P + Cp 2, Y = "3P2 + 2Cp 1
cos x + x sin x + C Exercises 3_1 so that y =

200, an students do not theoretically learn of
23. Y = 31. sin(x -y) +y = 7r
x 1. 102400, 1 = 16.23 days the rumor.
33. x sin y - y sin x = 0 3. 1 = 2.71 days 25. y(l) = I + 49ge-"; y(20) = I; quickly
25. y3 = 2.e x
4
+ Ce -3r
35. Y In x + x In y = 0 5. y(l) = 99.93 g; y(500) = 70.71 g
7. y(24) = 1036.8 g, 1 = 22.81 hours
27. lim
m-¥"
SO(I + l.)mt
m
is found with I:Hopital's rule. Because
37. 39.
9. 1 = 12.92 hours, y(15) = 8I Yo lim In(1 +
,"-)om
l.)m'
m
= lim ml In (I +
m-+O!l
1.)
m
=
Yn Yn Yn Yn -klm 2
(Improved (Runge-Kutta (Improved (Runge-Kutta 13. 9.72 days
Yn Yn
n Xn (Euler's) Euler's) of order 4) n Xn (Euler's) Euler's) of order 4) 15. y(100) = 0.96yo In(1 +1.) 1+1.
lim m = lim _ _m_ = lim _kt_ = kt
17. t,oo' = 3561.13 years old, tfo,," = 4222.81 years old, 1 1 k '
0 1 1 1 1 0 0 I I I m-;.a> m--¥oO m-+o:>
tfo,," - !too, ='661.68 years. No. mt tm2 1+-;
( k)- = (k)- =
1 1.05 1 1.00559 1.00678 1 0.05 I 1.00125 1.00125 21 1,000,000 (
I . y= I + ge-ll(oo; Y 25) = 124,856; lim So I + - So lim I + - Soe"'.
, m-+cr> m m-)co m
2 1.1 1.01118 1.0181 1.01921 2 0.1 1.0025 1.00501 1.00501 !~y(t) = 1,000,000
, 29. (e) yet) = _2_, lim y(l) = 0;
3 1.15 1.02608 1.03383 1.03486 3 0.15 1.0075 1.01129 1.01129 200 1 + ere t-+«l
1 23. Y = ( 3 )'; y(2) = 191; because there is no t
4 1.2 ! 1.04368 1.052 1.05296 4 0.2 1.01503 1.02012 1.02013 1+199- (d)y(l) = -- 63 '" limy(l) = 0
i i I 199 e t~OO
5 1.25 1.06345
I
1.07219 1.07309 5 0.25 1.02511 1.03156 I 1.03157
6 1.3 1.08505 1.0941 1.09494 6 0.3 1.03779 1.04564 1.04565 31. (a) y = 0, unstable; y = 2, semi-stable; (b) y =
I I 0, semi-stable; y = 1 unstable; (e) y = 0, semi-
7 1.35 1.10823 1.11752 I 1.11831 7 0:35 .1.0531 1.06242 1.06245 stable; y = I unstable; (d) Y = - 3, unstable;
8 1.4 1.13281 1.14228 1.14303 8 0.4 1.07112 1.08198 1.08201 y = 3, stable; y = 0, semi-stable
I
9 1.45 1.15866 1.16825 1.16896 9 0.45 1.09189 1.10437 1.10441
1
10 1.5 1.18565 1.19533 1.19601 10 0.5 1.11548 1.12966 1.12971

5 x
11 1.55 1.21368 1.22343 1.22407 11 0.55 1.14194 1.15789 1.15795 -I
12 1.6 1.24268 1.25247 1.25307 12 0.6 1.17132 1.1891 1.18918 (a) (b)
13 1.65 1.27256 1.28237 1.28295 13 0.65 1.20364 1.22329 1.2234
14 1.7 1.30328 1.31309 1.31364 14 0.7 1.23889 1.26043 1.26056
15 1.75 1.33478 1.34456 1.34509 15 0.75 1.27702 1.30042 1.30058
16 1.8 1.36699 1.37675 1.37726 16 0.8 1.31791 1.34309 1.34329
17 1.85 1.3999 1.40961 1.4101 17 0.85 1.36139 1.38818 1.38842

I 5 x 6 8 r
(e)
18 1.9 1.43344 1.44311 1.44357 18 0.9 1.40717 1.43532 1.43561
I -2
19 1.95 1.46759 1.4772 1.47764 19 0.95 1.45488 1.48403 1.48438
20 2. 1.50232 1.51186 1.51229 20 1. 1.50399 1.53369 1.5341

Cd)
J
33. YIIIOO 1 1120' Exercises 3.2 15. Because the object reaches its max. height when
= 100 _ 9ge d100,Y1120 = 20 _ 1ge
= 0 or I = ~
7'LD
1. 1= 55.85 min v = - gl + Vo and the air resistance is
70
76
I I 74
g
YIIlO = 10 _ ge r/l0,Y1I2 = 2 _ e-tI2,YI = 1 3. 12.59 min
72
ignored, the object hits the ground when I = 2vo.
5. 1= -2.45 hr, 12:30 P.M. g
6. Therefore, the velocity at this time is
7. T(5) = 76.3°F
= -gC;o) + Vo = Vo.
5 10 15 20 r
0.' v(2;0)
0.6 9. 75°F
11. 1= 4.7 min (e) u{l) = ~(333 + 37,,1 + (27 - 2.1)e- tl4
0.4 9 + 7r 17. c = 5
0,2
13. u{l) ~ -52 {-8.1 - (2.1 + 27)e- tl4 - 45 cos ('171/12) - IS". sin ('171112» 19. The velocity of the parachutist after the parachute is
+ '1T
,0LD () 817e81+13 th r ..
9 ..
10 " opened IS gIven by V I = 17eSI _ 13; e ImItmg
35. (b) I = 2958 + 3'1TSin(~) + 9 cos(~) - 72) 7. velocity is lim vet) = 8.
70
76 HOO
200 15. u{l) = 9 ~S.1( -14,,1+ .1e- + 3'1TSin(~)1/4 74

72 21. (b)~~ = ~~ ~ = v ~>e) ~~ v = v~ - 2
2gR
100
-100
1000 2000 300 + 9 cos(~) - 126) 5 10 - 15 20 r 23. g = 32 ftls2 = 0.006 mils 2; Vo = V2iR =
-200 \/2(.165)(0.006)(1080) = 1.46 mils
17. If R, = R2 , the volume remains constant, so V{I) = Vo. 23. Ud = 69.8273
If RI > R 2 , V increases. If RI < R 2 , V decreases. 25. Q(I) = EoC + e -tl(RC)( - EoC + Qo);
~
300
(e) h = 0: P{I) = - - - -
h
I + 2~;7 e -3(/100 '
="2:1 P{I) = dV
gal
dt = 4 -
gal
19. RI = 4 min' Rz = 3 min
3 = I, V{O) = 200 => V(I) = 1+ 200

70
71
70.5
69.5
",
1(1) = - RICe-,/(RC)(-EoC + Qo)
27. vet) = 12(4 - \13) - 12(4 - \I3)e- I13 ;

x(t) = 12(4 - \I3)t + 36(4 - \I3)e- l13 - 36(4 - \13)
-371 - 841v7 + {841v7 - 317)e'V7/IOO dy =
dl
(2~)(4~)
gal min
- (-Y-)(3~) =
I + 200 min
5 10 15 20 1
29. v(l) = 32, :~'!! (32 + Ce -I) = 32
50 -3S00 - 1447v7 + (1447v7 - 3S00)e'V7/IOO; , 3y
Exercises 3.3 31. v(l) = -gm/c, HOC
lim (-gm/c + Ce- cllm ) = -gm/c
h = I: P{I) = 8 - 1+ 200;Y(0) = 10
-265 - 1841\15 + (1841\15 - 26S)e d(20Vs) dy + ~ = 8 => J.L = eJ3/(I+200)dl = e 3In (t+200) =
1. vet) = 32 - 32e- l; v(2) = 27.67 ftls
33. c = i: V(I) = 64e- 1/2(-1 + e '12 );
50 -2S00 - 1447\15 + (1447\15 - 2S00)e '/(20Vs) ; dl 1+200
(I + 200)3
3. vet) = i Ii+ e- 64I ; v(l) = O.S ftls c = I: v(t) = 32e -I( - I + e'l;
c = 2: v(l) = 16e-21(-1 + e 21 )
d
h =3...: P{I) = dI[(1 + 200),y] = 8(1 + 200)' => (I + 200)'y = 5. V(I) = 4 - 68e -SI; v(O) = 0 when
-~8 In (-.L)
2
50~C=I12
,
-53 - 947\13 + {947\13 - 53)e v'JIlOO 2{1 + 200)4 + C => y = 2(1 + 200) + C(t + 200)-3 I =
17 = 0 .354 s
10 = 400 + C· 200- 3 => C = -390' 200 3 =>
150 _ ISOO - 1447\13 + (1447\13 - 1500)e ,v'J/IOO ; 40
y = 21 + 400 - 390.200 3(1 + 200)-3 7. {~ = 32 - v, v{O) = O} => v(l) = 32 - 32e- 1 30 C'" I
-1947 + 1894e'1l00 20
h = 2: P{I) = 100 _ 1947 + 947e'/IOO 21. (a) u(t) = e -1/4( -6 + 76e '/4 )
{~ = vet), 5(0) = o} => set) = 321 + 32e -I - 32; lO c=2
h = 2.: P{I) = 50 -10600 + 43411. ~~.i:==------- 5(4) = 96.S9 ft < 300 => 203.41 ft above the ground 0.5 I 1.5 2 2.5 3 l
O:CO··c
4 - 100000 + 14471' 60
50 9. vet) = 49 _ 49 e -I, lim vet) = 49 35. (numerical solutions)
5 5 5
+ tan- I-
1-+ 00
5 P{I) = -50 ( -3 40
h = -: + tan ( - I 1447))
- 30
2 200 500 20 n. V(I) = -9800 + 9900e-111000; v(l) = 0 => t = 10.152 s;
20~
10 set) = -9800t + 9900000(-e-tlIOOO + I);
0.6 0.6
10 15 20 t 5(10.152) = S06.76 m
15 0.4 0.4
'0 tl4 13. {~~ = -g, v(O) = vo} => vet) = -gt + Vo; 0.2 0.2
5 (b)U(I) = 9: ,,1(639 + 71,,1 + (81 - .1)e-
lO 20 30 40 50 60 - 90 cos (mIl2) - 30". sin (m/12»
{~ = vet), s(O) = so} => s(l) = -iglZ + Vol + So 0.2 0.4 0.6 0,8 1 t 0.2 0,4 0.6 0.8 1 r
J
37. approximately 300.772 sec 230000 39089

vet) = 12 => I = -~ In 41000 = 17.23 s
dy 164000
CHAPTER 4
dt = v,y(O) = 0 =>y =~I+
1
1. y = 0: unstable; y = "2: stable Exercises 4.1
37720000000 ( -637t/230000 _ 1) 45. y = C si~ t, C arbitrary
3. y = 0: unstable; y = 4: stable 405769 e 1. W(S) = I * 0 for any t; lin. indep. t
W(S) = 2e-'Ot * 0 for any t; lin. indep.
H=y(17.23) = 104.17798
3.
5. y = P 03,14; t = 4 !n 5 = S.86 days e - 16; r = e + 2; W(S) = 3e- 6t * 0 for any t; lin. indep.
0.2
1 3 23. (a) 4r2 = 32 cos 2 2 sec
5.
1)'11700 (c) r =
e
-6 cos 2 '2 + 6
0.15
D.'
7. y = Yo ("2 ; y(SO) = 0.9798yo (97.98% of Yo) 7. y' = Clef - cze- 1, y"= Clef + cze- r
0.05
9. y'= e-t(-c, + C2 - C21);y"= e-t(c, - 2C2 + C21)
1000 In 9 29. y = ~ +k -0.05
9. y = (1)' ; Y = 7S0 => t = In3 = 2 days 2 11. y' = - 2c, sin 2t + 2C2 cos 21;
-0,1
1+3
3 y" = -4c, cos 2t - 4C2 sin 21 -0.15
O 13. y = e- t - 2e 2t -0.2
11. T(t) = 90 - So(tr ; T(30) = 90 - 50(t)"2 = n.3°F
15. y = e 4t(1 - I)
13. ~= k(T- 325), T(O) = 100, T(45) = ISO 17. y = cos t + sin t + x sin t Exercises 4.2
1. Y = cle- 2t + cze- 6t
T(t) = Ce'" + 325; T(O) = 100 => C = -225 => T(t) = 19. (d) I~:: ~:: til = cosh2 t:'" sinh 2
t= I
-225e'" + 325 3. y = 4t
+ C2ef
45k
T(45) = ISO => -225e + 325 = 150 => 25. Y2(t) = e 2t
cle-
e
45k_ 175 _ 7 k_(7)1145
-225-'9=>e-'9 27. Yl(t) = te 2t
5. y = c, cos 41 + C2 sin 4t
29. Y2(t) = sin 21 7. y = c, cos v0t + C2 sin v0t

T(t) = -225(~r5 + 325 => T(-60) = 1O.43°F
31. xy = k
31. Yl(t) = I 9. y = cle
3r
cos 4t + C2e3t sin 4t
15. vet) = 4 - 4e- s,; v(3) = 4 ftls; s(t) = 4t + t e - St - t; 1 11. y = c,e-'t cos 3t + C2e-3t sin 3t
33. yil) = t In I
s(3) = 11.5 ft 13. Y = cle
3117
+ cze- r
35. a = 2, b = 5
3r
!~ = -9.8 - v, v(O) = 40 => vet) = _ 9; + 2~9 e- t; 15. y = CI e + c zte 3t
17. 41. Y = c, + c2 tan(c3 + C4 In I) is a solution of the equation
if 17. y(l) = -21 + 2le t/3
V(I) = O=> t= -In~= III
. s
299 <a) y = c, (note that if C2 = 0 or C4 = 0, Y is a constant
ds
dl = v, s(O) = 0 => S(I) = -5 t
99
-
299 _
-5-e t
299
+ -5-;
function); or
+ C2 + C2 In t). The Principle of

20~
15
s( -In ;:9) = 18.11 ft

33. Yes. The orthogonal trajectories ofy2 - 2cx = c 2 satisfy
dy __ y
(b) y = - t tan(C,
Superposition does not hold .
10
5
• ~. One solution of this
dv I 2 8(1ge" + II) dx x:!: V x 2 + y2
19. dt = 32 -"2 v ,v(O) = 30 => v(t) = 1ge5t _ 11 ; 43. y = cos 4t 0.5 1 1.5 2 t
homogeneous equation can be written as y2 + 2Cx = t
lim vet) = 8 fils
H~ C2. Now, replace C by -c. 19. y(l) = 14e 3t - Ile 4t
~
dv 82 637
21. m = 230 kg => dt = ill - 230000 v, v(O) = 0; 35. Equipotential lines: x 2 + y2 = c; orthogonal traj.: y = kx 2
= 164000 (I _ -637t/230000)
37. (e) +x2 + xy - +y2 = k, and _+x2 + xy + ~/ = k2
,
1.5
V ()
I 637 e 0.5 0.1 0.2 0.4 0.5 t
-5
-0.5 3 5 6 r -10
-, -15
,J
21. yet) = eSI 31. y= I + 2t 47. yet) = 0.0546098YoeO.471263 ' +

43. yet) = ±((-a - b)e-" + (3a + b)e-' ); (a) No local yoe-O.2076S41(O.0546098 cos(1.12621t) +
'2~
1 0.855013 sin(1.l2621t)
extrema if --3 b < a :s -b or -b :S a < -lb'
'0
3 '
8
6 (b) a 2: max {-b, -tb}, a * -tb; 4'
Exercises 4.5
1.S={t,l}
4 -4
2 (e) a:S min {-b, -tb}, a" -tb 3. SI= {e 21 },S,= {I}
0.1 0.2 0.3 0.4 0.5 t
5. SI = {e-I}, S, = {t4, t', t 2, t, I}
23, yet) = cos lOt + sin lOt Exercises 4.3
7. S, = {sin 21, cos 21}, S2 = {e- 41 }
1. W(S) = 0 for all I; lin. dep. 33. Y = e ' + e -tl2( -cos(V3tI2) - V3 sin(V3tI2)) 9. S, = {cos 3t, sin 3t}, S2 = {cos 2t, sin 2t}
*
3. W(S) = 16e 51 0 for any t; lin. indep. 35. Y = -te- 21 + te" 11. S, = {e- I cos 2t, e- I sin 2t}, 8 2 = {I}
05 5. W(S) = - 37e -., * 0 for any t; lin. indep. 37. y = 8t' + 4t + 8 21
-D,S
7. W(S) = !080e-'21 * 0 for any I; lin. indep. 41. (a) 3D(t) = 3; (b) (1 - D)(t) = I - 1; (e) d(t 2 ) = 2;
13. y/-t) = Ae
(d) (d + 1)(cos t) = 0 15. yp(t) = Ate"

-, 17. W(S) = 4725
17. yit) = At2 + BI + C
19. W(S) = 2'S8e 51 45. (a)y(t) = 212 (-2e-" + 2 cos V21 + 8V2 sin V2t)
25, yet) = e -2t(l + 5t) 21. (a) W( {f(I), t J(t))) = (I(I)i; 19. yp(t) = A cos 21 + B sin 2t
(b) W({f(t), tJ(t), t'J(t))) = 2(1(1))'; 21. yp(t) = AI cos 2t + Bt sin 2t + Ct + D
2
1.4
(e) W({f(t), tJ(t), ... ,t' J(t))) = 12(1(1))4, 23. yp(t) = At 6 + Bt S + CI 4 + Dt' + Et
1.2
W( {f(t), t J(t), ... , 14 J(t))) = 288(1(t))', ...
0.8 25 . Y = c,e
t
+ C2e -21 + "21
0.6
Exercises 4.4
0.4
27. Y = c,e" + C2e-2' - 41 + I
0.2 1. yet) = c,e- 21 + C2te-21 + c,t'e- 2t + C4e21, 4,
0.5 I 1.5 2 2.5 3 t /4)'+ 4y'" -:' 16y' - 16y = 0 29. Y = c,e- I cos 5t + C2e-1 sin 5t - 13t + 1
3. yet) = c, + c,t + c, cos 3t + C4 sin 31, 4, /4) + 9y" = 0 (b)y(t) = -ie21(3(-2 + 7t) cos V31 + 31. Y = c,e -1/4 + c,e -t/2 + 5t 2 - 60t + 280
27, Y(I) = e -21(2 cos 41 + sin 41)
5. y(l) = e -"((c, + C2t) cos 4t + (c, + C4t) sin 4t) +
C5e-sl + C6e-I13, 6, 3y(6) + 52/S) + 455y(4) +
V3(-7 + 2t) sin V3t)
33. y = Cl + C2e21 - 4 sin 3t + t cos 31
2336y'" + 7105y"+ 11,500y' + 3125y = 0 '2 35. y = Cle" + C2e-" - cos 3t - 3t sin 3t
10
7. no: W(S) = 0 37. y = c,e- 2' + czte- 2, - 4t cos 2t + 3 cos 21-
9. yes: 4/4 ) - y'" - 8y" - 14y' + 4y = 0 4t 2 sin 2t + 4t sin 21
15 2 2.5 3 r
11. Y = Cl + c,e SI + c,le sI 39. Y = Cle-I + C2esl - 36t'e SI + 18t'e 51 - 6te"
13. y = Cle-" + e-"(c2 cos t + c, sin t)
2t
0.2 0.4 0.6 0.8 I t 41. y = c,e-" + C2e-zI + c,e- I + te-" + ~te-' - tlZe-1
cle- + cze + C3 e7t
t
(1 +I - C) esc I 15. Y =
n· y (1=
) t- C 17. Y = Sf
Cle- ! + cze
2t
+ C3 te2t (e) yet) = 0.00028893Ie-'6.9961 + 43. y = cle 4r + cze- cos t + C3e-St sin t
Sr
+e t
e°.4334." ( -1.00029 cos(0.523649t) -
t[(5c, - c,) cos 5t - (5c, + C2) sin 51] 19. Y = Clef + C2e-6t + C3te-6t 45. y = Cl e 2l+ C2te2t + C3e3t cos t + C4e31 sin t + e-1
. 3~. y = (c, cos 5t + C2 sin 5t) sin t 1.06119 sin(0.5236491)
21. y = Cl + ezf + C3 e -31 + C4e3t 47. yet) = -4 + 2e- I + 2e'
_:~12'4S61
(5c, + 4C2) cos 4t - (5c, + 4c,) sin 4t
35, y = (c, cos 4t + c, sin 4t)1 tan 4t 23. y = c,e- 2t + c,e" + c, cos 2t + C4 sin 21
49. yet) = l_ 2e-" + 2e'
25. y = ere -31 + C2 e2t + C3e3r + C4 e4t 3
37, (a)y(t) = crt + c,t 2l3 ; (b)Y(I) = crt + c21 In t
41. (a) yet) = Ce - I sin 21; (b) no solution; 27. Y = cle- + cze 21 + C3e-4r + C4te-4t
t -,
-2
-4 51. yet) = t+ e -41 + 4te -41

cre-Zt + c z te- + C3e( cos t + C4ef sin t
(e)y(l) = e- I cos 2t 2r
29. Y =
69. (a) yet) = -1~2 e- 2' - 2e" + 2.40716e,·05455, I

74. (a) Y(I) = ( I4s - 145(1 + e
2
19. y = c,e-' + C2e' + (± + il)e-' + (±I - ±)e' or
+ e -0.0272772'(0.6004616
cos(3.027521) 12 - ce -<12)'
- 0.446426 sin(3.027521») ( I4s sIn 3I y = c,e-' + C2e' + lle-' + lie'
1.5 2 2
21. y = c,e 2' + C21e2, - e" In I
~
.
40~ Cte-3t + C2te-3t + te- In t

05 3t
23. Y =
,0
0.5 I 1.5 2.5 3 x 20 25. y = c,e-" + c,le-' - e-" cos (e')
-<1.s
±
: s " '
-, )'0:-1/25 10
-, 27. y = e'(c, + c21 + t(t2 + 2)'Vl='? + sino, I)
O.S 1 1.5 2 t -4
-1.5
29. y = ±e"(c, + c21 + (I' - I) tan-' 1- t In(t' + I)

(b)y(l) = ~- 0.010325ge-4.07594'
55. y = 5e' - 7e" + 2
6te ' - 2t 2e 2' + 31 + tt t
2
- + 0.308775e- 0.90 '589' + 0.20545ge o.27'954,
(b) not possible
Exercises 4.6
31. y = (CI - 21 - 4 In Isec(1/2)1) cos(1/2)
(C2 + 21 - 4 Inlcsc(1/2)1) sin(t/2)
+
- 0 .00390835e4.95557, + II
5
1. y = cle u + C2e5 , - ie" 33. Y = c, + C2 cos 21 + c, .
SIn
I
2t - 4'1 cos 21-
-, -,
-I
,,
-2 ,'
1 Isec
gin 2t ISIn + sin 2tl
' 21 +gln~
1 II
-2
-, 35. y = c,e' + C21e' + c,t 2e' + ±12 e' In t
-4
-s 7. y = CI cos 41
.' 41 - 4'1
+ C2 SIn 1 cos 4t +
71. (a) a, = 0, a2 = -5, a, = 0, ao = 4; (b) a, = a2 = a, = 39. y = -cos t 1n(1 + sin t) + cos I In(cos t) +
116 sin 41 In Isin 4tl
57. Y = ~t'
6
_lt 2
2
+ t + le-"
5 - 5I 0, ao = -I; (c) a, = 0, a2 = 2, a, = 0, ao = I; (d) If 2sint+2cost
a, = 6, a2 = 2; a, = 0, ao = I, theny(4) + 2y"+ y =
24(cos I - sin I) *- cos I => not possible. 9. Y =cle-' cos 7t + c,e-' sin 71 - lle-' cos 7t + x=rc/21
- 7 I
73. (graphs for c = 0, 1,2,2.5, 3, 3.5, 4, 5, and 6)
:9 e -, sin 7t Inlsin 711
2.5 I
~
0.3
0.2
~
c==.I.
11. y = CI e' cos 5x + c,e' sin 5x +
c=Q
51(2~ Inlcos 5tl- f) +

0.5
0.1 '
e' cos
24681012 -0.1 4 6 I 2.'
-0.5
-,
-0.2
-0.3
'
e' sin 51( ;5 In Isin 5tl + f)
13. y = cle" cos 51 + C2e" sin 51 +
~
';2'S ~';'
~
59. (c)
0.6 2
0." '='5. 51C~
41. y = _l +...L cos 41 +...L sin 4t + e-2< -
2 25 50
e-"
8:i I
0.4
0.2
-
e" cos sin 5t - ;5 In Isec 51 + tan 5(1) -
~,;s .'="
61. (b)
-0.2 6 8 t 1 12 r -0.2 2 ~
63. (b) no; (d) yet) = -0.4 -1 -0.4 ;5 e" sin 51 cos 5t
-0.6 -2 -0.6 -w4
a -tb/a k
c, - hC2e + ht; 15. Y = cle cos x +
6r
C2e6t sin t + te 6t sin x +
';4
~
k 2 eO' cos I Inlcos II
(e) yet) = c, + C2t + 2a t 0.3 0.15 0.1
~ ~ ~
, 17. y = cle" + C2e-" - 118 + liS e" 1n(1 + eO"~) - -,
67. (a) w = 2; (b)y(l) = -0.1 r -0.05 2 t
~ ~ ~ . -4
c, cos 2t + C2 sin 21 - i t cos 2t
-0.3 -0.15 -0.1
...Le -" In(l + e")
IS
J
3 3 2 1 2 . B > 1 => r12 < 0 => y = CIX" + C2X"l ~ 0 as x ........JI 00;
43 Y = -l!e'
• 13
+ ~te' + J.. e -2' sin 3t
78 18
7. y = c,x cos(3 In x) + C2 x sin(3 In x)
9. y = c,x 2 + C2X-'
37. Y = 2"x - 2"x cos(ln x) + 2"x sm(in x) '*
B < 1 ~ r':2 = a ± if3, f3 0, a> 0 ~ y =
xa(c, cos(f3In x) + C2 sin(f3 In x)) unbounded as x'" Co.
11. Y = c, sin(2 In x) + C2 cos(2 In x)
40 63. (a)y =
13. Y = C,X -2 sin(3 In x) + C2X -2 cos(3 In x)
'0 -;5X-2(-53 + 28 cos(5lnx) - 35 sin(51nx))
15. y = c,x- to + C2X-7 + c,x- l
17. y = c,x + C1X 2 + c,x- l (b) (0.241468, 11.3898) (min), (0.377358, 46.6735)
-4
19. y = c,x + C2X sin(ln x) + c,x cos(in x) (max), (0.848419, 0.922604) (min), (1.32588,
3.78057) (max)
_1..4 t 4 t' - 3t 2 - 6t - 6 + 6e'
21. y = 9!' (9c,x' + 9C1X' in x + 1) -,
45 . Y = -
23. Y = ..!,(1 - 5c,x 2 sin(ln x) + 5C1X2 cos(ln x))

5x
39 Y = _1_ + 22 v;: _ 2...
. 5x 2 15 3x
1
25. Y = -2"x + C,X 2 + C2 X -,
27. Y = 2 + c, sin(2 In x) + C2 cos(2 In x)
I ,', 1.25 1.5 1.75 x
29. Y = 25x 4 (25c, + 25c2X6 + 25c,x 6 In x + x)
-l.S , <
-2
31. Y = tx'l3 + tx2 Exercises 4.8
55. y=c,t-'+c2t-'lnt+lnt-2 1. x = 0, R 2: I
c, sin(2 In t) + C2 cos(2 In t) + i t
~lc2)x2+
56. y = -3
25
3. y=C,+C2X+(-15C,+
2
57. y = c,t 6 + C2t-' + 2... - 1.. In t

18 3
,5
, 41 Y =J..x'
. 25
+ 24
25
v;: -!V;:inx
5 ( -55c,
91) x' + (455
+ "6C2 671) x 4 + ...
- 4 c , + 24"c2
05
-05 5. Y = c, + C2X + (c, + tC2 + t)X2 +

-,
33. Y = cos(2 In x)
(tc, + tC2)x' + (t + ±c, + ;4 C2)X4 + ...
-,
~
.
1 , 3 4 9 s
7. Y = c, + C2 X - '4C2X + 16 c,x - 80 C2 X + .. ,
-1
, '2 4 5 6'
61. (a) y = c,t 2 cos t + C2t2 sin t + t; (b) all solutions; -, 1 4
9. y= 1 +"3x +
1 8
42 x +
1
1386 x
'2 + _1_ x '6 + ...
83160
(e) no 43. W({x", x"}) = (r2 - rtlx"+"-'
cos 2t sin 2t 1 47. (a) approaches zero as x -> 0+ C2 2 5C2 3 j
63. (a) y = c'V! + C2 V! + V!

I -'0
35. y= 44 x
8 3 2
+TIx-'4x 11. Y = c, + C2(X - 1) - 8(x - 1) + '96(x - 1) + ..
(b) approaches zero as x -> 00
(b) no solution (e) bounded as x -> 0+ and as x -> 00
cos 2t I '5 51. Y = c] + C2X2 + c,x- 2

(ely = - V! + V!
05
53. y = c] + C2X2 + C3X -2 + 2 In x 15. (a) y = ao' + 2aoa,x + (at' + 2aOa2)x2 +
55. y = c]x-] + C2X-2 (2aoa, + 2a,a,)x' + ... ;
Exercises 4.7
-0.' 1, 1 4 1, 7 6
1. Y = c,v;: + C2X'12 -, 57. Y = -sin(2 In(-x)) (b)y=x+2"x -8 x -8 x -12o x +
3. Y = c,x + C2X4 -t.5 59. B = I ~ Y = c, cos(ln x) + C2 sin(ln x) unbounded as _1_I_ x 7 + ...
-2 1680
5. Y = c, v;: cos(2 In x) + C2 v;: sin(2 In x) X~ 00;
J
1
[I Answers to Selected Exercises A-27
4x
17. x = 0, irregular; x = - I, regular 47. (a) (b) W(y"Y2) = Ce- J4dx = Ce- ;
~
19 Y = C'X7/2(1 +.lx + ...!...x 2 + _1_x 3 + ... )
, 1. Lin. indep.
W(e- 2x cos 3x, e- 2x sin 3x) = 3e- 4x
. 3 22 286 0'
O. 3. Lin. indep.
(e) W(y" Y2) = Ce- J4dx = Ce- 4x ;
+ C2 (1 _lx + ..l.x 2 - ..l.x 3 + ... ) 04
0.2
W(e- 2x , xe-2x) = e- 4x
5 10 10 5. Lin. indep.
21. y=
-02 1 2 3 4. 5 6 x (d) W(y" Y2) = Ce- JOdx = C; W(cos 3x, sin 3x) =3
-04
Jl. yil) =~ 57. (a) W({Yl,Y2)) =
lit; (b) W({y"Yz)) = 15;
c,X(1 _.lx + ...!...x 2 _ _ 1_x 3 + _1_x 4 _ " .) I
3 24 360 8640 (b)y = yeO) + y'(O)x + iY(0)x 3 + I~Y'(0)X4 + 13. Y = cle- 4tl3 + cze,{2
(e) W({y"Yz)) = I
+ C2 X
(2
-, In X X - 3"x
1 1
+ 24 x
3 4- 1,
360 x + ...
)
58. (a) Y = tex-, - e-'-'
1 (0.
!Soy )x +
1
504 Y (O)x
, 7
+ ... 15. y = c,e-' + C2e-2'
+ C2 X -, ( - 2 425 4
lx + "9 x - 288 x
3
17. y = t/2
+ cze 21
-
(e)
cle (b) A = 0 => Y = 0; A < 0 => Y = 0; A > 0 => An =
157, ) 19. y = cle-
tl4
+ C2eftS (nTr/p)2 => Yn = sin(n77X/p), n = 1,2, ...
+ 21600 x + ...
Cl + cze- + C3e-r/2 (e) A = 0 => Y = 1; A < 0 => Y = 0; A> 0 => An =
t
21. y =
,.
7 + l60
77 x 2 + 384
77 3 23. y = e -2v' (c, sin I + cos I) (nTr/p)Z => Yn = cos(n77X/p), n = I, 2, ...
23. Y = c,x 7/4 ( 1 + gX x + ...
) C2
1 (d) A = 0 => y = 0; A> 0 => Y = 0; A < 0 => An =

+ C2 X -514 In x (15, 105 4
8 x +"64x + ... ) 25. Y = c, + C2e
_" 21
+ 1s - 51
2
+"31,
1 -(nTr/2)2 => Yn = e- x sin(n77X/2), n = 1,2,
Z
13 x 3 + ... )
15 2 -2 -, -, . 12 2 3. 2 59. Y = c,x + CZx'
+ C2 X -514(12 + 15x+""4x 27. Y = c,e cos 21 + C2e sm 21 - 17 cos 1+ 17 sm I
61. Y = c,x- 1 + C2X-lIZ,y = -x-' + lx- lI2
25. Y = c,x + C2X-7 (d)
29. Y = c, + C2e
2,
-
1 cos 4I - 20
10 1sm
. 4I 63. Y = x-'[c, sin(4lnx) + Cz cos(4Inx)]
29. (a)y=c,(I-x 2 - i x4 - 3~x·+ "')+C2X; 65. Y = c,x' + C2x5 + x
"2 3t • 2 3 -"It
(b) Y = c, (1 - 3x 2 + t x4 + ... ) + C2 (x - tx 3)
31. Y = c,e cos 1+ C2 e sm 1+ 29 e -
33. Y = cle- 4t + cze- 3t - 6te- 4t 3t 2e- 4t t 3e- 4t

67. Y = c, + + (fc, -
C2X CZ)XZ + (tc2 - c, + i)x'
+ (t c , - t
- -
31. F(1, b, b; x) = 1 + x + x 2 + x3 + ... = _I_ t 4t + 1 2 -2t C2 )x'

I-x 35. y = c,e -
-"It
+ C21e -2t -1- c,e 4t + 36"e 121 e
. 1
37. (a) y = C')1I4(X) + C2Y1I4(X); 49. (e) L,(x) =1- x, L 2(x) = "2(x 2 - 4x + 2), 37. y = c,e-' cos 21 + C2e-' sin 21 + c,le-' cos 21 + (...!... - .lc, + ~C2)X' + ...
20 2 120
(b) y = c,),(3x) + c2Y,(3x) 1 3 +
L3(X) = "6(-x + 9x 2 - 18x + 6), +
-, . 2
C4 /e sm I
4
625 - ill 81 1 2
+ 1s 1 -'13( 1- 5I x+ 20I 1 X3 1
39. y,(x) = F(I, 0, I; x) = I; second lin. indep. soln.: 69. Y = C,X X2 + 60 + 960 X
4
Y2(X) = x + Inlx - 11
45. (b)y = 1 - x + .lx' - ...!...x 4 + _I_x' + _1_x·_

I 4
L4(x) = 24 (x - 16x 3 + 7lx2 - 96x + 24),
1
39. Y = -te-" te- +
2
' + 33~00x' + ... ) + C2(1 + -fix + 3~8x2
6 18 360 225 L,(x) = 12O(-x' + 25x 4 - 200x' + 600x 2 - 41. y = cos 51 I 3 1 x4 + 1 x' + ... )
359 7 + 15708 X + 1256640 144513600
226800 X + ... ; 600x + 120), _ _.l2. -4, +.l2. '+ 1. ,
I 43. Y - 25 e 25 e 5e 7( 1 2 1 4 1 • __1 _x ,
+ 450x 4 - 71. Y = c,x 1 + gX + l60 x + 5760 x +
(e) , L.(x) = 720 (x· - 36x' 2400x' + 322560
~ hL
1 .
, 2.5
2
5400x 2 - 4320x + 720), 45. Y = "2 lsml + ... ) + C2X lnx( -1800x· - 225x' - ~ x'O
0.'
0 .•
0.4
,
1.5 I 7 4
L7(X) = 5040 (-x + 49x· - 88lx' + 7350x - 47. y = c, cos 1+ C2 sin I + sin I Inlsin 11- x cos I _ 2... x 12 _ _ 5_x14 + ... ) + C2X(-86400 + 21600x 2
0.2 0.' 16 896
-0.2
-0.4
I -2 3 " 5 6 x I 4 5 6 x
1
2
29400x' + 52920x - 35280x + 5040), 49. Y = ~e 4
• - e 4• + I e4• _ 5400x4 + 1125 X' + Elx '0 + ... )
L,(x) = 40320 (x' - 64x 7 + 1568x· - 18816x' + 8 32
1 1 3
t;
Graph of nwnerical approximate 2 2 x
Graph of numerical approximate 51. y= -'4ex+xex+"2x eXlnx-'4x e
4
solution solution together 'Wllh the grnph of 117600x - 376320x 3 + 564480x 2 - 322560x + 73. Y = C1F( -t(l3 + V209), -t(13 - V209). -x)
the polynomial approximate solution
40320) 53. y = expG(r + 2C2)) cos(C, + I)
y= l-x+t x J_-fgx4+ 3~Ox'
(f) 1 56. (a) W(y" Y2) = Ce - J'dx = Ce -'x;
+ C2x2l'F( -i(35 + 3V209), -i(35 - 3V209).
+ ~5 x 6
2;::00x
7
-
+ 15~~60 Xi + 38~~~!OO xQ 51. (b) ( [Pn(x)f dx = 2n ~I

4
Wi(e- 4X e'" = e- • 4x eXI = e- 3x
,} _4e- eX
I + 4e-'x = 5e- 3x ~.
3'
-x)
J
23. = 32, 0 < C < 32

CHAPTER 5 25.
C
C = 10/13
d2 d 4 5 6 t
27. m dt 2 (u + v) + cdi(u + v) + k(u + v) = 0 =:0
Exercises 5. t
1. m = 4 slugs, k = 9 Ib/ft; released 1 ft above eq. with
31. ± Y199/15 = 0.910586 d 2u du d 2v dv
m"d(i" + cdt + ku = 0 and m dt 2 + cdt + kv = 0;
-0.6
-0.8
zero init. vel. x(O) = u(O) + yeO) = a =:0 u(O) =a and yeO) = 0; -I
Exercises 5.2
3. m =
114 slugs, k =
16 Ib/ft; released 0.75 ft (8 in.)
dx (0) = du (0) + dv (0) = 13 =:0 du (0) = 0 and
below eq. with an upward initial vel. of 2 ft/sec 1. m = 1, C = 4, k = 3; released from equilibrium with an dt dt dt dt
upward inil. vel. of 4 ft/sec Exercises 5.3
5. x(t) = 5 cos(t - q,), q, = -cos-I(3/5) = 0.93 rads; dv (0) = (3
3. m = 1/4, C = 2, k = 1; released 6 inches above equi!.
per. = 27T, amp = 5
WIth a downward init. vel. of 1ft/sec
dt 1. x ()
t = -72 cos 4t + 2I sm
. 2
4t + 7 cos 3t, w = 4
29. x(t) = cle- PI + C2te-P', p = C/2m; x(O) = a, x'(O) =
'1/65 cos(4t -
7. x(t) = -4- q,), q, = cos- I( - 8)
'1/65 =
5. x(l) = v;o e- 2t cos(3t - q,), q, = cos-I(3rv'iO) =
0=:0 x(t) = ae -P'(I + pt); x(t) = 0 =:0 t = -lip < 0
3. x(t) = W
35 _ r;;
cos(4 v 3t) +
4
47 cos t
3.02 rads; per. = 7T/2; amp = '1/65/4
0.32 rads, Q.P.: 27T/3, 1= 0.63 35. C = 2V6: x = te- N6 ; C = 4V6: x =
9. x(t) = V; cos(3t - q,), q, = -cos- I( ~) = W2 ~ 9 (cos 3t - cos wt),
7 . x (I ) v29 -, cos(5t - q,), q, = cos- I (5/v29)
= ~e - r= = _I-e -Vz(3+3V3)'(_1 + e6 'v,). C = V6: x = 5. x(t) = 2
6\/2 '
-7T/4 rads; per. = 27T/3; amp = \/213 (
11. x(t) = cos 8t, max. dis. = 1 ft when -8 sin 8t = 0 or

0.38 rads, Q.P.: 27T15, t = 0.39 J.\/2e-'v'3i2 sin
3 \/2
t) (_3_ 3tsin3t, w=3
resonance occurs if w = 3
8t = n7T =:0 t = n7T/8, n = 0, 1,2, ... 9• X
(t) -- I -5, + 2e
-2e I -3,; overdamped; does not pass
4 1
+
13. x(t) = -1/4 cos 8t; t = 7T/16 sec; x(5) = 0.167 ft; X(I) =
1 . 8 7T
8sm t;I="8 sec
through equilibrium; x ( t t))
In( = 0.093
7. x(t) = -e- '(2 0 cos 3t + 1;0 sin 3t)
I ( .
402cost-smt)
11. x(t) = - 3e -, + 2e -'/2; overdamped; x(l) = 0 =:0 t =
15. As k increases, the frequency at which the spring-mass
system passes through equilibrium increases. 2 In (1) = 0.811; max. dis. = I at t = 0 9. x(t) = t cos t - t ± t f;
cos 2t; sin decreases
17. b = VlO23/16 13. x(t) =" 4e -4' +. I 4te -4,; critically damped; does not pass
19. m = 2 slugs through equilibrium; max. dis. = 4 at t = 0
37. The object does not come into contact with the floor in
11. x(t) = e -d2 (2 cos ¥- ¥) - 4 sin 2 cos t + 1I sin t;
21. w Yol- + {3'-lw2 15. x(t) = -5e- s, - 24te- S ,; critically damped; does not
d 2y -rrr2p pass through equilibrium; max. dis. = 5 at t = 0
any of the cases. trans'. e -1/2 (2 cos 24 2 2!. -
sin ' steady-state'. 2!.).
23. "d(i" + --;;;-Y
= 0
~
17. C = 2 -2 cos t + 11 sin t
25. y(l) = 1.61 cos 3.51 - 0.856379 sin 3.51; max.
displacement = Y(1.61)2 + (0.856379), = 1.8236 ft
19. x(t) = -le e -4, + -6,; does not pass through _o.osr, 34'
o~
I
13. (a)x(t) = a cos( ~t) -
equilibrium; max. dis. = ~ at t = 0 -0.1
27. a = 1: x = cos 21; a = 4: x = 4 cos 21; " = -2: -0.15 -0.5 1 2 3 4 t
Fw 1m. ( fk) + k-mw
F .
t
x = - 2 cos 21; (3 = 1: x = sin 2t; 13 = 4: x = 2 sin 2t; 21. x(t) = e-
2t
( -3 cos(4/¥-t) -
-0.2
-I
k_mw2yTsm y-;;;t 2 smwt
[13 -~]
x'(0)=-1 x'(O)=0
13 = -2: x = -sin 2t
l[fsi+ (b)x(t) = Gi sin ( V;;
k- mw' Yl:
lIt) +
~~
fit)) orx(t) =
4~a=43"~
0.2 0.4 F .
2~ cos(4 f i t - k- mw2 sm wt
0.15 0.3
x q,). where q, = 0.1 0.2
, 0=-2
cos
-I -2v'lT
--7- = 2.81646; x(t) = 0 =:0
0,05
1 3 4 t
0.1
1 3 4 r
(c)x(t) = a cos( Ift) +
-2
-4
a=l -1
-2 x'(O) =t x'{O) = 2
[(3 - k- mw2
FW] YT fk) + k-Fmw' smwt
1m sm. (y;;t "
_ I 1s((2n + 1)7T) .
1- 4y lT 2 +q"nanymtegeror
= 2: x = e-'(-I + 2t); C = v8: x =
29. t YI3712 = 1.37941, /2 vl99 = 1.17556
t = 0.739471, 1.26899, 1.7985, 2.32802, 2.85753 ...
39. C
(-2 + 1/\/2)e-(I+v'2J, + (1 - 1/\/2)e(I-v'2J,
15
.
x(t) = {I - cos t,
-2 cos t,
0 '" t '" 7T
t> 7T
~L
17. x(l) = 5000 I Qo . I 5000
27. x(l) = 89733775561 (cos(2995811150) - cos(20vI51)) 15. Q(I) = Qo cos '\/LC; [(I) = - '\/LC sm '\/LC; max. 4000 60000
I-Sinl,O:;;I:;;1 50000
3000 40000
-2 sin 1 cos I + (2 cos 1 - 1) sin I + 2000 30000
2 - I, 1 <I:;; 2 1.10-7 Q _- Q.
0,
m [_
ax. -
Qo
'\/LC 20000
1000 10000
•
( 5·]0"..8
(sin 2 - 2 sin 1) cos I + 4 cos 1 sin 2
.!.2 sin I' I> 2 2 4
(a)
6 8 10.t 2 4
(c:)
6 8 10 x
_5.10-8 r
= 0: x = f = 1: x = f(2
2000~
_1.10-7
21. b I sin I; b sin I + I sin I)
1500
1000
Exercises 5.4 500
2 4 6 8 10 x
1. Q( I) = J.l - J.l cos 41 (d)
8 8
1 6 5x, 1 0 0 x I"
3. Q(I) =!.. - J... sin 161 25. (a) sex) = 36000 x - '36 + -9- ; (b) no so utlOn,
4 64
1 6 5,
(e) s(x) = 36000 x - 9'x + 150x; (d) sex) =
100
23. (a) X(I) = """39(cos(191/10) - cos 21);
5. Q(I) = 8
1
'fI e-(12512)t sin(25~ I} max.: 0.0225 at I = 21. (a)s(x) = 3~ox6 - 1;5 x' + 5~0 x 2; (b)s(x) = 1 6 5,+25 x
0.0147 36000 x - nX 6
100 _1_ x 6 _ 500 x' + 1250x 2 . (e) sex) = _1_ x 6 -
(b) X(I) = 41(cos 21 - cos(211/10)) 360 9 ' 360
::~~~L
7
·
Q(I) =
e
-125tI2(_~
12502813 cos
(25v7 )
2 I
_
500 I 6 175, 500
(a) -9-x' + 750x 2; (d) sex) = 360 x - -9- x + -3- X2
26554371 . (25v7)) 12185
312570325v7 sm - 2 - 1 + 12502813 cos I + 400
10 200
tL=i~ ~~
62526875 . .
12502813 sm I. steady-state charge: i~ Q(I) = 7000 2 4 6 8 10 x 2 4 6 8 10 x
~ggg
(a) (e)
1500
12185 . 62526875. 1000 4000
12502813 cos I + 12502813 sm I; steady-state current: '000
500' 2000 12
12185. 62526875 lOOO 10
- 12502813 sm I + 12502813 cos 1 2 4 6 8 10 x 2 4 6 8 10 x
25. x(l) = Ca) (b)
I - sin I + a cos I + b sin I, 0:;; I:;; 1

+ .!.3 x 2 J... '. _ 1 4
lL-. tL=i
9
-I - sin I + 2sin(I-1) + 2 + acosl + b sin I, 1 <I:;; 2 · () () = _1_ x 4
a sx 300
_
15 x ,(b)s(X) -30x + 2 4 6 8 lOX
20000 1500 (d)
{ -sin I + 2 sin(1 - 1) - sin(1 - 2) + a cos I + b sin I, I > 1
.!.Q.x 2 _lx'· (e) sex) = .!.x 4 + .!QQ.x 2 - 20 x' 15000
10000 . 1000
3 3' 3 3 3
~ Pv
5000 500 Exercises 5.5
1.5 1.5
I
0.5
I
0.5
11 · ()
a s () 1 4 10
x = 300 x +"""3 x -
1x, ;
TS 2 4
(e)
6 8 10 x 2 4
(d)
6 8 10 x
1. (a) 8(1) = ;0 cos 41; (b) 8(1) = 210 cos 41 + ±
sin 41;
-0.5
-I
-1.5
I 2, 3 4 5 ( -05
-I
1 2 3 5 6 t
(b) sex) = J... X 4 + .!QQ. x _lx'·
30 3 3'
(e) eel) = ;0 cos 41 - ± sin 41; max. displacement:
23. (a)s(x) = 4800001T- sin( ~} (b) no solution;

-1.5
4 (a) 1/20; (b) and (e): \/26/20 = 0.255
1 4 1000 20,.
~ e -tV'! sin 31 +
(e) s (x) = '3x + -3-x -"""3x ; slDlple support leads
3. (a) 8(1) = ;0 e -tV? cos 31;
I~
to larger max. displacement. (e)s(x) = -801T- 4 (-600= - 300,",x + ,",x'-
0.75
0.5
0.25 13 I 6 10000 125, 6000 sin( ~));
.
(b) 8(t) =
(v7
60 + '3 1) e
-tv:, . 1 31
sm 31 + 20 e -tV? cos;
t
· (a) sex) = 360 x + - 9 -x - -9- x ;
-0.25 12 45 6
~))
v7 1) e -tv:, sm
(e) 8(1) = ( 60-'3 . 3 1 -tV? cos 31
-0.5 () = _1_ x 6 + 1250 _ ~ x ,
(d) sx (d)s(x) = 2401T- 4 ( -100= + =' + 2000 sin( 1+ 20e
360 3 x 18
~i
,I
Answers to Selected Exe rClses
.
! Answers to Selected Exe rClses
. A-33
9. T = 21TV'iJ9i, =
2~ ~r (': J, }fd+2
11. T _ 2.83 sec 25.
- 21TYSm "" 3.14 sec
-2rv
IS
13. 21TVLI9.8 = I
17. b _ 4L
2
=> L = 0.248 m
g > 0, overclam d 2
I
05
-as ., 4468 6 t
i9 clamped; b2 - 4Lg < 0,pe

• (e) yes; (d) no
unde - 4Lg -- 0, critically
; b clam
r ped -2
-I
-15
'
-IS
-2
2~ :~~ ,(,: , _;~fd\.2

21. (a) e(l) = 2 sin I' (b)
(d) O(t) = -sin '. 8(1) = 2 cos I; (e) 0(1) = - 1.5
(f) 8(1) = I, <.e) 8(1) = -2 sin 2 cos I; I
, cos 1- sm I; (g) 0(1) = _I; cosI+sinl

05
4 68
_2~_IS
_0.S2468 t
::lAf litpv, -22~

-I '
-1.5
-2
8
-I
-2
~_~~ 2
6 B t =~l
U
4 6 t _~;
~ 2 4 6 B t
-2 -1.5
23'08'~' ~ ~ :1~ L1PtF ~.2~hP

8
06 04
02 06
=g.~ ~~
04 02
4 6 8 I =g: 2 4 8'
_05 05 I
=;;~'"' ~'~' .~2~,

-i'J
-~~ =~!
-08 24 B t -0.5 2 6 0.5
-2 8 B t 2 6 B t
-06
-05 2 4 6 8 '
05
025
-~o~
2
246
8 I
t
26.
~!~ (\ Il~
~fV'"' ~l ' , · · ,.~ , -13
Ilfvh
-::r2
4 O'S~ 2~
-2 6 8 '
-ii~l
-I 8 ' -05 4 t
1.5I
0,5
~!fV. ~21~8
-:!,kP\'2
;1~, " ~~~;'5~824
-I 4
-15
-2
6 t
-1!-2
-OS 2 4
6 8
05
'-05 -15 2 4 6 8 t
_~!~2
4 ~1tA- ~12M7'
68'
~-0.5 -1
2 8,05
-1
-05 2 8 t
-1:!-2 6 8 '-05
-1:! 2 4 0.5
-2
6 8 t
-1:!
-0.5 2
6'
~:!~ 11~ ~2lP'

~-~:ilr:2146B'~24
2 -13-2 6 8 '-0.5
-1:l 2 4
6 8 t
__________________________________________________JJ ___ J•
_

13. sex) -_ -3-
250 2 125 3
x - -9- x + 12 X
1 5 1
- 360 x
6 3
X(I) = - 10 cos I - I~ sin I - 2c,e 2, +
15
1. X(I) =
1
31 . 1 71" 71"
cos 81; max. dis. = 3; I = 16; 8"
3. x(t) = --3 e- 2, sin 31, lim X(I) = 0; quasiper. = 2371";

15. sex) _- -3-
1250 2 250 3
x - -9- x
I I
+ 12x - 360 x
5
6
27. 1.
y(l) = "2 sm + (3 C2 4)
(3C3 - 2c2)e -, - 2c31e-'
C3 e
-, + 3 C3 1-'
e
5.
C
50
26 -22
8 315)
-15
I~)
I - 30
1-->00
( -8
17. 6(1) = cos 8t; max. dis. = I; 1 = 1~ (
max. dis. = Ht tan-
1
f)1 = 0.144; 1= f
19. 6(1) = e -8, + 81e -8,; motion is not periodic.
2
z(t) = 5" cos t
3 sm
+ 10 . I
2t
+ c,e + C2e -( + C3 1-'
e
7. -30
-8
-18
30 -2
1
5. x(1) = 4 - 41 cos 2I; maximum
. d'Isplacement IS
. "2
1 and X(I) = -20 + 20e' - 20le' -5
r- 29. { y(l) = -30 + 30e' - 20te' ( 22 -13)

71" 21. 6(t) = 0.2168 cos 3.71 + 0.0471622 sin 3.71 9. AB = 22 -24 -17 and
occurs when I = "2' -31 -4
Z 37
- 1 I 3 cos I, 0 oS I oS I
I - 3 cos
7. x () 31 cos 2 t; beats,. -3
+2 sm"2
. I -
(-,: 9
I - 23. y(l) = (-3 + 2 cos I + 2 sin l)cosl- 35 16
2'
- 25 25 -10,
9. Q(I) -109 -10g e
3 250 -10, . 3
cos I - 327 e sm I;
(2 cos I - 2 sin 1 + sin 2) sin I, 1>1 -20
BA=
0
20
-9
20
-8 -':1
-28
1(1) = ~e-l0'
3
sin 31; lim Q(t) =~; lim 1(1)
109 ,__
= 0
25. X(I) = 25 cos 251 CVS ) -40
-27 -11 -22
22
11 11
1-->00
11 . 27. X(I) = A siIiewnl + B cos wnl + f; 05

-60
-80
11. AB= (-1
-2
-11) ( -2
-28 andBA= -16
-3)
-27
11. Q(I) = 500 - 500 cos 1001; 1(1) ="""5 sm 1001; -3 -2.5 -2 -1.5 -[ -0.5 0.5 ;r
lim Q(I) and f--+fXJ

[~OC
lim 1(1) do not exist x(l) = (xo - f) cos wnl + f 35. k = -4 13. AB =
( -I
9
-12
22
-4)
-3 and
37. c *4 -10 -36

-4
-9
~~ ~3Xl + Yl ; (b) (~~: ~Xl + 2Y1 + cos 3t

( :
( -I
BA = -22 25 11)
-6
39. (a) YI = Y2 YI = Y2 -6 14 -12
CHAPTER 6 Y2 = -x, - 2Yl yz = -2x, + Yl 15. [A[ = 17
41. 17. [A[ =0
Exercises 6.1 19. [A= 10
X' = Y
(-: ~)
15. { y' = -16x+lsinl
1. {X(I) = tlz + C lo y(l) = sin I + Cz} 21.
3. {x(l)
7. x"
= C lo y(l) =
5. {x,(I) = _e- 3" X2(t) = I}
+ 4x'
-tC,e-7'
C2e- z ,}
- 21x = 0, {X(I) = C,e- 7'
+ cz e 3'}
+ C2e 3',y(l) =
17. {~:::
;;" = x - 3y - 6z - 3w
t+
o 23.
(-~
I
-I
0 ~~)
24. A, = -5, VI = (-:) A2 = -3, V2 = (-~)
9. x" + x = 0, {X(I) = C, cos I + C2 sin I, y(l) =
21.
X(I) = c,e 3' + C2e - 4 , 25. A, = -5, V, = G} A2 = -4, V2 = G)
Exercises 6.2
-t(C, + C2) cos I + t(C, - C2 ) sin t} (
Y ()
I = 35 - 5
"2c,e
3,
+ cze
-4,
1. (-~ ~) 27. Al = A2 = -3; V, = (~) V2 = G)

11. x" + x = I, {X(I) = C, cos I + C2 sin t + I,y(t) =
-C, sin I + C2 cos t} xCI) = ~- le' + c,e- 3
' + 2lcze- 3'
23 9 4 -5 21) 29. A, = A2 = -5; V = (~)
~y _ 4x
. I 3. ( 1-3
13. {;:: ( y(l) = 3 + cze- 3
'
9. linearly independent
25
11. linearly independent 20

200
13. linearly independent 15
,SO
15. yes 10 ,00
so
17. yes
19. no 10 15 20 25 A"
4
2e ' X(t) = -8 sin 41
21. X(I) = ( 3e 4t
37. { yet) = 8 cos 41
23. X(I) = (2 cos 21 + 2 sin 21 2 cos 2t - 2 sin 21)(C') X(I) = 2c, cos 41 + 2C2 sin 41 ,0 \0
-Sin 2t -cos 2t C2 ' 25. { yet) = -c, sin 4t + C2 cos 41
Jx = -8 cos 2t
lY = 2 cos 2t + 2 sin 2t
37. sin t
.L.... cos t - t sin t) , ( sin t COsI+lsinl)
+c Exe,.cises 6.4
-10
(
cos t sin t + t cos t - cos t sint-tcost
1. XCt) = c,G)e
2t
+ C2 (:)e-' _10 -'0
x(r)
(~)e-4' + C2(~)el' e 2' - 2e

3t
3. X(I) = c\
X(t) =
31
39. y(t) = 2e' - 2e 21 + 2e
{ 21 3t
z(t) = - 2e + 2e
X(I) = c,e' + e'[(-c2 - fC3)cos 21
-10) (2:;: 6,) e l ( -cos 21 + sin 2t))
31. + (fc 2 - C3 ) sin 21]
v, = ( ~~' ,1.2.3 = 3 ± 6i, V2.3 = ~5 7. X(t) = c, (

l
e' cos 2t yet) = e t ( -C3 cos 21 + C2 sin 21)
z(l) = e'(c2 cos 2t + C3 sin 2t)
2'
e ( -cos 21 - sin 2t)) -2
45. (a) ,1.,.2 = I ± ki; (b) ,1.\,2 = k ± , + C2 ( e t sin 2t X(t) = 4e 2' - 4e'
33. { y (t) = 4e 2'
Exe,.cises 6.3
i. (;:) = G _;)(;)
10 4 x(t)
41. a and b are linearly independent so I:~ ::I=

a,b2 - a2b, 0, Thus, *
W(ea'(a cos {3t - b sin {3t), ea'(a cos f3t + b sin {3t))
3. (;:) = (~ _~)(;) ,I eat(a, cos {3t - b, sin {3t) ea'(a, cos {3t + b, sin {3t)1
= eat(al cos {3t - b2 sin {3t) e a'(a2 cos {3t + b2 sin {31)
l
5. C:)=(: ~)C)+(~) 6 r
2a
= (a,b 2 - a1 b,)e ' sin(2{3t) 0 *
43. (e)
7. (;:) = (_~ ~)(;) + (2 s~n J

45. (a)
X(t) = 8e 4t
35. { y(l) = 16te 4 ' 47. (d)
1
4 c, - 13cz
_ (13 12 + 13c3
3 ) e -3t + (9 12 3 ) cos 21 + (13c,
6 18 + 13c3
24) SIn
. 21 _2e- 4t
x - 13c, + 13c2 -13c3 -13c2 ~ +..1!..... 4t _It + 1!te4t
17. X(t) = ':"'3e- 4t 80 1369 e 4 37
( 2e- 4t
51. _ (4
y - -13 c, + 13 - 13 e + 13 c, + 13cz + 13c3 cos 21 + "26c,
12 cz 3c3) -3t (4 1 3) ( 1 8
17) .
-13cz + 26 C3 SIn 21 43 59 4t 3 8 4t
- 200 - 1369 e + lOt + 37 te
_ (4 c 12 3 ) -3t ( 4 12 16) (6 8 2) .
z - -13 , + 13c2 - 13c3 e + 13c, -13cz + 13c3 cos 21 + -13c, -13c2 + 13c3 SIn 21 _~ + .!.ll _ tZ 91 48 4t I 4 4t
16 4 400 + 1369 e - 20 t - 37 te
_~+..!2.t_IZ
3~'1.S~ '~'
2
I Y O.S
, ,
Y
32 8
.!.l_lt
16 4
X(I) = t - I + e- t
25. { yet) = 0
2~
-1 4 r -05 2 3 5 6 r 2 3 5 r
-2 -, , .l" -1
-3 -1.5 l
1:1' 15
, .«t)
53. (a)
0.5~ 0.5
o y(r)
o 0.5 1 1.5 2 2.5 3;t 0.511.52253 1
x(t) = -"32 e -t - 2I e t +"6e

I 5t
27.
-4 -2
-2
() - 7 -4t
X I -16e
43
+ 85 cos 1+ 85 SIn I + 19 .
c,e
4t
+ 7c2e
-2t
19.
1yet) = te-t + ±et + 112 eSt
1
7. 10 12
-4 - _...!.
Y (I) - 16 e
-4t.2. +.1....
+ 85 cos I 85 SIn I + c, e
4,
+ C2e
-2t 10
(b) lX(I) = (i~ Xo - 177 YO) e 2t + C37 Xo + 177 yo)e- 11t/3 ; 9. f X(I)= e-6t(~2 cos 41 - c, sin 41 - ±)
t -_( -uxo
Y () 6 + UYo 3) e 2t + (6UXo + U14) Yo e
-l1t/3 1 y (I) = e- (c, 6t
cos 41 + Cz sin 41 + ±)
x(t) = - t +! _let + c,e 5t + C2e-t + C3le-t
-4 -2
-2
-4
2 4 6 8 lOX
E,~ X(I) =0 if Yo = 2xo; }.!,~ y(l) = 0 if 5 4 L'
-± t 3 I t 3
j
5 -t -t
e [c
3t ="5 -
_ Yo. . _ _ {X(I) = 0 cos 21 - sin 21) + yet) '4e + c,e t + (cz - 2c 3)e + C3te
e2t(7
-- cos 3t + - I . 3t)
I
x(l) = 2( 21.
Xo - 2' (c) only ,fxo - Yo - 0 => y(l) = 0 x(t) = -20 +
c, ( - t cos 21 + ± sin 21)]

tI 12
z(t) = '4e - "'5 +t+ 5)
c,e 5t + ( - 2cz + '4C3 e -t -
29. 13 13 39
SIn
yet) = 30 + e Zt (13
-13 4 cos 3t + 39
31 .)
SIn 3t
2c3te-t
Exercises 6.5 11. + te3t(1O - cos 21 - 12/ cos 21 + 3 sin 21 +
150 40
41 sin 21)
X(I)=C,(~}-7t+cz@e-5t+ml+(~) e3 1c, cos 21 + C2 sin 21] + 4t
e-2t(t cos t- ±sin t)
(-.
1. Y(I) = _7e 20
3 -7, -5t 6 173 te3t(-3 + 41 cos 2t - sin 21) 23.

x I) - - c,e + cze + 35 1 + 1225 X(I)= _2e 4t e-Zt(-tcosl+±sint)
1
or 0.5
(
_ -7t 1 I 4t
y(l) - c,e + 71- 49 x = ( sin I cos I)(C,) + (I cos I - Inlcos Iisin I) e
-toO 50 '00'
-20
13. (t) -cos I sin I C2 t sin I + Inlcos tlcos t -40

-50
X(I) = II -...!. - C2 e - 6t X(t) = c, sin I - C2 cos t -

2 sin t Inlsin tl + e-
Zt
(± cos t + t sin I) )
-± t) (~:) +
3. 6 36 31. (b)
( _ I 2
y(l) -"6 1
I 1
-12 1
+ 54 + c,e -6t
+ C21e
-6t
15.
( y(t) = c, cos I + C2 SIn t - 2 +
. 2 sin t Inlcos t + II
e- (
Zt
cos t - t sin 33. (a)
2 cos I Inlcos I + II - 2 cos t Inlsin tl e- 2t sin t C3 35. (c)
Answers to Selected Exercises A,41
A-40 Answers to Selected Exercises I
I
Exercises 6.6 17. (A - 2)(A + 1) = A2 - A - 2; x" -x' - 2x= 0;

X(t) = _1_ e - s'(453 - 796e 7t + 343e s,
1372
+ 84te 7 , - 294t'e 7') 0.5
1. A, = -11, A, = 1; saddle point, unstable {;: : ix + y; saddle point, unstable
37. (a) 2. A, = -4, A, = 6; saddle point, unstable
y(t) = __I_ e - s'(-151 - 1564e7 ' + 343e s, 19. (A + 3)(A + 3) = A2 + 6A + 9; x' + 6x' + 9x = 0;
5'
1372
- 28te 7 2 7
588t e
-0.5
-,
-0.5
-, 3. A, = A2 = 2, v, = (~) deficient node, unstable {x:y =
= y 9 6 ; deficient node, asymptotically stable pr
-x-y ,
\ ' - ')
4. A, = 2, A, = 10; improper node, unstable X' = -3x

{
y' = _ 3y; star node, asymptotically stable
5. A, = -12, A, = -8; improper node, asymptotically
(b) (curve in gray is the graph of the solution to the
nonhomogeneous system)
stable 21. (e) A, = A2 = 1; v, = (~) deficient node, unstable
o.s~
6. A, = -6, A, = -3; improper node, asymptotically stable
0.6
0.4
02 x 0.5
y
Nonhomogeneous
syStem 7. A1 = A, = -1, v, = (~). v, = G} star node,
-0.2 2 68 lOy' asymptotically stable
8. A, = A2 = 3, v, = G)' v, = (~) stat node, unstable

x(t) = - 8\ e -2'(179 - 260e 3 ' + 60t - 90re 3 ,
9. A, = 2 + 5{V2, A2 = 2 - 5{\/Z; spiral point, unstable
+ 90t' + 135t'e 3' + 63 13 ) 10. A, = -5 + i, A, = -5 - i; spiral point, asymptotically
(b) I 23. (-4. -2). A, = -3, A, = -I; improper node.
yet) = -e-"(346 - 346e 3' + 84t - 180te 3' stable
162 asymptotically stable
\ + 126t' + 189t 2 e 3' + 126t 3 ) 11. A, = iV41, A2 = -iV41; center, stable
41. (a) (solution to homogeneous system on left; solution to
nonhomogeneous system on right; x in black; y in 12. A, = A2 = -2; v, = (-~) deficient node,
'0 gray)
10 asymptotically stable
(0)
'~::~'2
-'0 13. (e). lim X(t) =
t-+r:.o el
_::+--0. 2.EOf.
-20
-40
=-+-:--:+:---+-;' O. 0.8"
-100
x(t) = - 92~8 e -'( - 320 - 2720t + e'( -8928 cos 4t -60 x y
(e) + 2312t cos 4t + 6958 sin 4t - 4624t sin 4t»

yet) = - 46~4 e-'(176 - 8161 + e'(-4800 cos 41 (b) (curve in gray is the graph of the solution to the
nonhomogeneous system)
\ + 2312t cos 4t - 2642 sin 4t»
Nonhomogeneous y Exercises 6.7
system
I
,~,
500
15. A, = 0, A2 =·2;y = -2x +C 1. (0, 0), saddle point, unstable; (I, 0), inconclusive-
center or spiral point
3. (0, 0), saddle point, unstable; (I. -I), spiral point,
_, x
unstable
Homogeneous
-2 y system 5. (0, 0), inconclusive-center or spiral point
-2
-500 / 7. (0, 0), node or spiral point, unstable
9. (0. 0), improper node, asymptotically stable; (0, 4),
39. (a) (solution to homogeneous system on left; solution to -1000 saddle point, unstable; (2, 0), saddle point, unstable;
nonhomogeneous system on right; x in black; y in
(3. I), spiral point, unstable
gray)
11. (-4,4), saddle point, unstable; (1, -1), saddle point, Exe,.cises 6_8
x(t) = -cos 3t cosh 2t - 31 cosh 2t SIn
.
3t
unstable 1. xCI) = 8.64479, y(l) = 8.29263
13. (2, -I): Al = -6, ,1.2 = I, saddle; (8, -I): Al = 6, ,1.2 =
I, unstable node
3. xCI) = -7.2362, y(1) = -1.5998
26.
+ cos 3t sinh 2t + t sin 3t sinh 2t
= -0.I13115,y(I) = -1.06576
~;
5. xCI) 1
15. (0,0): Al = - V2, ,1.2 = V2, saddle; (1, I): Al = yet) = 13 (1 3 cos 3t cosh 2t + cosh 2t sin 3t
7. x(l) = -326.204, y(I) = 608
-1 + i, ,1.2 = - I - i, stable spiral
17. (0,2): Al = 2, ,1.2 = 4, unstable node; (0, -2):
9. xCI) = 0.164251, y(1) = -0.504587 I cos 3
- 13 ' 2t - 39
t Sinh l1"SIn 3t SI nh 2t)
23. (e)
Al = -2, ,1.2 = -4, stable node 11. x(1) = 1.52319,y(I) = 0.419975
25. (d)
(2,0): Al = -2\1'2, ,1.2 = 2\1'2, saddle; (-2, 0):
Al = -2\1'2, ,1.2 = 2\1'2, saddle 27. (c) 13. {x::
y - -
y 2x 3; Euler's method yields
- y
19. (1, I): Al = 2, ,1.2 = -2, saddle; (-I, -I):
29. (a) x(t) =
c 2
C 1e-', yet) = --+e- 2 ' +
x(O) = 0, yeO) = -3
Al = 1 + iV3, ,1.2 = I - iV3, unstable spiral xCI) = -0.723913, y(I) = 0.40179; exact solution is
x(t) = 3e- 2' - 3e-' so x(1) = 3e- 2 (I - e) =
C1C2 )
(~:, 0):
X2
21. (0,0): Al = alo ,1.2 = -a2, saddle; Al = -alo C2e' ( ory= -""3+-x-; -0.697632.
,1.2 = -(a2- C~~I). stable node (b) unstable;

(e) x(t) = C l e-', yet) = C2 e', saddle
15. {x:y == y- 9x ; Euler's method yieldS
x(O) = 0, yeO) = 3
( _~,
C2
aIe2 + Q2bt):
CIC2 31. (d) (J.L = -I, -i, -t, -~, O,~, t, i, I) x(1) = 0.346313, y(I) = -4.49743; exact solution is
x(t) = sin 3t so x(l) = sin 3 = 0.14112.
X' =y 1 ; Euler's method yields

17.,
{y = -(2 ( 16x + ty)
X(t) = e -'(2 + 2t)
x(l) = O,y(I) = 4 27. { yet) = -2te-'
x(2) = 0.354942, y(2) = -2.90834; exact solution is
x(t) = sine4 In t) so x(2) = sine4 In 2) = 0.360687.
19. (See 13) Runge-Kutta yields xCI) = -0.697621,
y(I) = 0.291602 0.75
0.5
21. (See 15) Runge-Kutta yields xCI) = 0.141307,
0.25
y(I) = -2.96975
05
23. (See 17) Runge-Kutta yields xCI) = 0.360845, -0.25
y(I) = -1.86541 -0.5
25.
X(t) =
1
tee' - e- 2')
-0.75
-1
{ yet) = 3(e' + 2e- 2')
28. (a)
2.5
1.5
__6 _ 4 _ 2 2 4 : S X
C2
-4
0.5
-6
0.5 1 1.5 2.5 x
(b)(ii)
3. (-3, -2), f~), C~)} CHAPTER 7
S.
(1, 2), {C~), (~)} Exercises 7.1
Q(t) = -te- I + 45e- f + 45 sin 1- 45 cos I
Q(t) = 3vS e- 5t13 Sin(vS t) (b) 12 (1) = e- f
1. (a) 5 3 { 1,(1) = te- f - 45e- f - 45 sin I + 45 cos I
SI Sin(~ t) + e- SI/3 cos(~ t)
7.
(2, 3, 4), {(=i). (-:). (-i)} 1 l(t) = _vSe- /3
Q(t) = 90 - i e - f/4 - 266 e- f
3 3
(iv) Q(t) = Ie-I + 7vS e- SI/3 Sin(vS t) - 11. (a) 12(1) = e -1/4 ;
{(I) en ~ ~i) en ~ ~i)}

3 15 3
+ .
9. {-4,-2_3z}, 0,
I I
Z , -z (b) t e- S
l/3 cos(~ t) 1 1,(1) = -90 + ie-fl' +
3
266 e- f
3
+ 45 sin I - ~e-tl4 + 1;9 e- f
l
o Q(I) = -45 cos I
l(t) = -Ie-' - 2vS e- SI/3 Sin(vS t) +
61 3 3 3 (b) 12(1) = e -tl4
11. X = (_e6t
~e-SI/3 cos(~ t)
4elt,)
+ ~e -f/4 - 1;9 e- f
t
e e Cz 13(1) = 45 cos 1- 45 sin I
{xy== 3e"
l
13. 2e- r + ell
3. (a) {Q(t) = 1O-6e -I cos(Yzt)
13.
~=y
d
-+'.
' A - _3z, undamped
(v) .!'.!'. = -9x
Q(t) = !:..e -II' + 4Yz e - I sin(V2t) - dl
15. X= ("
sm--cos-
2 ,
2 sm 2+
"cos 2 tCOS
')
2 (:')
9 9
l
. 1
sm '
1999991 - I
9000000 e cos
(Yz )
t ~ =y
2 (b) 15. d ' A, = -I, A2 = -9; overdamped
8 -1/2 1999991 - I ' ( ' ;;::2)
.!'.!'. = -9x - lOy
l()
17. X'= (-4
. 5
-I) X
-.2 '
, I = 'ge - 4500000Yz e sm v 1.t
t e - I cos(Yzt)
- dl
l
dx =y
-6
X = e-31(Sin '2t - 2 cos 21 -cos 2t
cos 2t - 2 sin 2t c,
t,) Q(I) = 1O-6e -"(1 + I)
17. :1
.!'.!'.
, A, = -5 + 5i,
-5 sin 2t S. (a) { 1,(1) = 1O-61e -2f . ~ dt
= -50x - lOy
(c)
19. X (-2= cos 4t 2 sin 4t)(C,)
sm 41 cos 41 c,
1(1) = -1O-6 e -21(1 + 2t) ~
1,(1) = _1O-6 e -21(1 + t)
A2 = - 5 - 5i; underdamped
l
~ - ~e-21 - 451e- 2f ~ =y
l
X(t) = e -f(2 cos 3t + sin 31) Q(t) =
(b) 2 2 ~ 19. d ' A, = -5, A2 = -5;
21. { y(t) = 3e-f(cos 3t + 3 sin 3t) .!'.!'. = -25x - lOy
2
I (I) = 452
_ 45 e- 21 - 451e- 21
2
dl
-6
23. X = e'f(: 1; t)(:~) l(t) = 90e -21(1 + t) ~ l z(t) = ~ e -21(3 + e2t + 2t) critically damped
21 (13) fx(t) = cos 31 .
Q(t) = 1O-6e - I - j. 1O-6t 2e -I
t • jy(t) = -3 sin 31'
l
x(t) = c,e- f + 9c2e'f + 7. (a) 12(t) = IO- 6te- I
lX(I) = te- f - i e - 9f
e7t - 181e'f
(d) No. The solution in (v) of part (b) is not tangent to { 13(t) = j. 1O-6 2 -f
vectors in the direction field. That in part (c) is. If
the system were linear, these two solutions would
25.
y(t) = c,e- f + C2e'f + te" - 2te'f
t e
Q(t) = 1O-6e - I + te- I -j. 1O-612 e - I
(I5)
y(t) = _.2. e -f + .2. e -9f
coincide. (b) 12(t) = IO- 6Ie- I 8 8
X= 1- Cte-t + 2c2eSf { 13(t) = -te- f + j. 1O-6 2 -I
t e
Chapter 6 Review Exercises 27. { y = -1 + c,e- f + C2 eSf 23. A = - 2 c ± 2 ~ V c' - 4 lan, overdamped if c' >
Q(t) = 90 - le- f - 90e- f m
4 lan; critically damped if c' = 4 lan; underdamped if
1. (-4, II), f~)' G)} 29.
X =
{y =
cos t - t sin t
sin t + t cos t
9. (a) I,(t) = e- f
{ 13(t) = -90 + te -f + 90e- f
c' < 4 lan.
d
1
25. lim Q(t) does not exist; lim 1(1) does not exist. Exercises 7.2
1-+0lJ (-+00
X(t»)
25 4,
-3 e + 3'e
5 -21) x(t) = __I_[bV,(l - e-P~V,+v,y(V,VV) +
1.
X(t)=f-t e -'
3 I _'
.
):,n;! (x(t), y(I» =
(32' 23) 19. X(t) = (
yet)
=
( 25e 41 _ 5e-21
41.
v, + V2
a(V, + V2e-P~v,+vvl(V,V')1
I yet) ='2 + '2 e ' 21 ~ V [aV2(1 - e-PI(V,+V,)/(V,Vv ) +
1
Y(I) = V,
_ (x(t») _ (15e - lOe -') 2
21. X(t) - yet) - lOe-1 b(V2 + V,e-P'(V,+V,)/(V,VV )]
3.
X(t) = f4 - 8 fe-
3d2
'
.
hn;! (x(t), yet»~ =
(83" 3'4) . ~+~~ .
Ca)hmxCt)=-V
~+~~
V' h m y ( I ) =+-VV
67 e -,0, + 171 71 + 663 -31 '_00 +
I y(t)=3'+3'e-3<12 H -7 Toe 70 e
(b) VI > V2, VI = V2
1 2 l--+'X' 1 2
[x(t) = I + 3e-'S'/4 . _
5. )y(I) = 4 _ 3e-IS'/4',~ (x(t),Y(I» - (1, 4)
23. X(t) = (;~~) =
5 -101+ 171 71 221 -3,
e Toe -Toe
(e) bV, - aV2 > 0, aV2 - bV, > 0, no
43. The problem can be solved as a system or the system
z(t)
27. ~ Q(I) does not exist; E,n;! 12 (1) does not exist. 52 e -,01 + 171 71 + 663 -31 can be written as a second-order ODE, where x" =
7. (x(t) = 10 - lOe- d5 , yet) = 10 - lOe-'/5 - 2te- d5 }, 7 Toe 70 e (a, - a2)x' + bty' (derivative of first eqn.) or y' =
)i,n;! x(l) = )~ Y (t) = 10, x(t) x(l) = z(l) = 18752.89
....!...(x" - (a, - a2)x'). Substitution of y' and y =
bl
9. {X(t) = 4e- d5 ,Y(I) = 4e- d5 + %te-'/S},
;, Cx' - (al - a2)x) into the second equation and
)i,n;! X(I) = )i,n;! Y(I) = 0, X(I) simplification yields x' - [(al - a2) + (b l +
11. (a) (x(1) = 300 - 300e- d20 , y(l) = 150 + 150e-'/IO-
25. X(I) = (;i%) = [(bl - b 2)(al - a2) - a2b,]x = O.
- b2 )]x'
z(t)
300e-'120}, (b) )i,n;! x(l) = 300, )i,n;!y(t) = 150 We consider the characteristics of this equation (same as
-2
the eigenvalues of the system).
13. {X(t) = 3~5 _ \2; e-3<125 _ 1:5 e-'/25,y(t) = 2~0 + Z(I) = 7095.86 Periodic if al + b l = a2 + b 2 and a,b l - 2a2bl -
a,b 2 + a2b2 > O. For example, if al = a2 = -I, b l =
29. E,n;! Q(I) = 91. ).!,n;! 12(1) = 0, )i,n;! 1,(t) = -43.5 ~e-3<125 - ~e-'/25}.
lim X(I) = ~ X(I) = xoe -al b 2 = I,xo = 1, and Yo = 0, then (x(t) = cos I,y(t) =
6 2 ' I-->~ 6 ' 29. yet) = yoe -a' + axote -a' ; -sin I}.
{
I· ( 200 )
I) = -3-; x(t = yet) at approx. 1= 29.1 mm. . z(t) = Xo + Yo + Zo - (Xo + yo)e -at - axote -at
Q(,) ,:,n;!y Exponential decay if (b 2 - b l ) > (a, - a2) and
"----- lim x(t)
(--'1>00
= 0, lim yet) = 0, lim z(t)
1-+00 1-+00
= Xo + Yo + Zo (b l - b 2)(al - a2) - a2b, 2: O. For example, if al = -I,
dx = 2y - 2x + 1 a2 = 0, b l = b 2 = 1, Xo = I, Yo = 0, then (x(t) = e -',
to 15 20 1 15. system:
dl
d X(I) = 7e- 6' y(l) = OJ.
I E.1'.=-3y+x+1
dt 33.
47 _, 42 -61
y(l) = 5 e - 5 e ; ~Z(I) = 16
Exponential growth if Cal - a2) > (b 2 - b l ) (will have
at least one positive eigenvalue). For example, if al = 2,
X(t) = t e -3<12«3e' - I)xo + 2(e' - I)yo)

{x ()
I
_ 5
- 4 -]2e
5
23
lim x(t) = -4, '->,00

i3 e -'}, £-400
-41
3
+
8 -I
3'e , y(l) _- 43 + ]2e
lim yet) = -4; max. x = 4, y = I
23 -4,
+
l z(t) = 16 - 47 e- I
5
+ 2 e - 61
5
a2 = 0, b, = b 2 = I, Xo = I, Yo = 0, then (x(l) = e 2t,
yet) = OJ.
31. I ; stable X(I) = e- 41 Exercises 7.3

I
j!: ~;~ ~O
yet) = -4e-3'/2(3(e' - l)xo + 2(e' - 3)yo) 35. y(t) = 3e- 21 - 2e- 41 ; lim z(t) = 3
{ z(l) = 3 - 3e- 21 + e-41 H~ 1. -a Inlyl + by- clnlxl + dx= C
node
I:
3. Ca) (2k, k) stable spiral; (b) C2k, k) center
17. system: X(t) = 2

dl 10 10 37. y(I)=2-e-' ; 5. = y , (hr, 0), k = 0, ±I, ±2, ... ,
{
dz=...L_...E... z(l) = I + I + e- I 2= -sin x
dt 10 10 Ji,~ x(t) = 2, J~ y (t) = 2, J~ z(t) = <Xl
dl
(vertical position of pendulum)
X(I) = 100 - 100e -dlO
i Y(I) = 100 - 100e -1110
z(t) = 100 - 100e- II10
lim X(I) = lim Y(I) = lim z(t) = 100

t-+oo (-+00 t-we
-
-
IOle -<1,0
10te-d,o - t I2e -'/1O
39.
X(t) = 10 - ,2e- '
y(l) = 10 - 8e-' - 2te-'
{ zCt) = lOt - 8 + lOe-' + 2te- 1
ii,~ x(t) = 10, i~ y(t) = 10,

;
ii,n;! z(l) = 00
7.
I
~ = y . dy = dyldt = -k x.
dy = -k2; dx
dt
(0, 0) center
dxldt y' 2Y
2
2
~
+ 2
2
k x 2 + C-
'
27. Use a numerical solver with initial guess Xo = 1 and 5. (a) (X(I) = e-' + le-', y(l) = -le-'} (critically
.~)
Ii :~gMI +~;(xO'YO)=
du th ..
Notice that dx = v m(x I, so e equation IS graph the solution. Observe graph to see that Xo = 2. damped);
+ (b) (X(I) = cos 21, y(l) = -2 sin 21} (undamped)

1
9. !!..[dU dx] V (x) = 0
dl dx dl ~ .
dl - x' (R - xi
. du dudx
R(M I + -vM:Mz), 0) = ( VM,RVM, , 0); saddle NotIce also that dt = dx dt' so we have
- 15 5 -, () _ 15
( M,-M, - vM; 2
!!..[dU] + V (x) = 0 or d x + V (x) = 0 I -""2 -"2 e ,y I -""2
7. {x () + "25 e -,}
Y dldl~ dl'~'
11. If dx = y - F(x) = Y - IX feu) du and dd = -g(x), 3'
dl 0 I
2
d x dy dx f( dx h' h . 19. paths: circles; equilibrium point: center; agree 3
then dJ2 = dt - f(x) dt = -g(x) - x) dt' w IC IS 9. (x(t) = 15e 3' + 5e', y(t) = 15e ' - 5e'}
d'x dx 21. (a) center if x = 1; saddle if x = -1; (b) saddle if x = 0;

equivalent to dJ2 + f(x) dt + g (x) = O. center if x = 1, x = -1.
13. In each case, no limit cycle in the xy-plane: x(t) = _le- s, - ~e-w + ..Q2.e3'
23.\ :~
3 51 17
j
2
(a) fx(x. y) + gy(x, y) = 3x + I + x' =
= y FR ; equilibrium points: (e, 0); 11 () - J..6 e -5' _ 64 -14, +..Q2. 3,
4x' + 1 > 0; . y I - 51 e 34 e
E.!'. = --;:- sgn(y)
(b)fx~"C, y) + gy(x. y) = -1 - 1 = -2 < 0; dl '1
"(I) =
"
_l6 e -5, + 208 -14' +..Q2. 3,
51 e 34 e
. dt
de > 0, then we mtegrate
. I dt
dO dO
d (dO)
dt = - F'R' WIth
(e) fx(x, y) + gy(x, y) = y' + 8 > o. If
\
~~ =y 1 I (dO)'
respect to e to obtain the parabolas "2 dt = - FRe
15. d ,; y = 0 -3
¥e=a-u=ku- + C, ~~ > O. Similar calculations follow for ~~ < O. 12. J(x, y) = ( g 0 I );
-TcOSX -b
, I + \I'T="4ka
~b);
anda-u-l;,,-=O=>u= 2k
25. (a) ~(X2 + y' + Z2) = 2xx' + 2yy' + 2zz' and J(n7T, 0) = ( g 0
We choose u =
1- \I'T="4ka
2k
..
because It IS closer to Chapter 7 Review Exercises -T cos n-rr
substitute for Xl, y', and :;', where x' = y and y' = -x
1+ \I'T="4ka . . on x 2 + y' + z, = 1. (b) z(l) = C" so z(l) = Zoo (e) Use
lQ() - 1 (3 -,m -,) n is odd: A =
-b ± VI7+4iii
2 (A2 < 0 < A,J real;
u = 0 than u = 2k ' and we are consldermg the change of variables in Exercise 10. I - 500000"2e - e
1. (a)
. . (1 + \I'T="4ka 0) 'C) I ( -',/3 + e -') -b±~
~x' -=-~e):
a small change m the orbIt. J 2k ' = "I = 500000 -e" n is even: A = 2 (complex conjugate
26. (a) (-1.44225, -0.44225); J(x,y) = ('
_~ I) =>;.,2 + VI -
( -vl-4ka 4ka = 0; center. eigenvalues of J( -1.44225, -0.44225); Al.' =
pair or two negative real)
0
17. Differentiating tm(X)( :r + Vex) = E with respect to

-0.0977925 ± 0.431302i; spiral point, asymptotically
stable
Q()
- - 4500001 -, - 3 -,
I - 500000 e Ie
+ 9000003
1000000 e
-'d3
1
(b)
'( ) = 3000001 -, + 3 -, _ 3000001 -2,/3 13. (0, 0); A, = a > 0, A2 = -c < 0, saddle point, unstable;
I yields
1 (dx)3
"2 m '(x) dt
dx d'x
+ m(x) dt dJ2 + V'(x)
dx
dt =
" I 500000 e Ie 1000000 e
(t, 0), A, = -a < 0, A2 = -c + at = d(t - ~) > 0,
dx [lm'(x)(dx )' + m(x) d'~ + vex)] = 0, so (~, ~ ),
Q(I) = 100~000 e-' cos I
dl 2 dI dl" saddle point, unstable;
)2 + m(x) d'~ + Vex) = 3. (a) _ I _,.; V -4aed 2 + 4e'dk + e2 k 2

lm'(x)(dx
2 dI
~ [tm'(x)(:r + m(x) ~:~ + Vex)] =

dl
0
0
-,
1fz( I) - 1000000 e sm I Al.2 =
-ck ±
where 4de( -ad + ek) < 0, so

2d
2 -ek - V -4aed' + 4e'dk + e2 k 2 < 0 if -4aed 2 +

1.. m'(x) (dx)2 +~ d x + V'(x) = 0 Q( 1) -- 60 + 60 e -,. 59999999_,
sm I - 1000000 e cos
2~dl dl' ~
1 4e'dk + e2 k 2 > 0 (improper node) and AI" has negative
real part if -4aed' + 4e2 dk + e2 k 2 < 0 (spiral point);
!!..[~
dl
dx]
dl
+ V'(x) = O.
~
(b)
1[ () _ 60'
2 I -
59999999 -,.
- 1000000 e sm I -
60 -,
e cos I asymptotically stable
CHAPTER 8 87. e f cos 7t
89. e
71
cos t
5. y = Z cos 2t
Exercises 8_1 6s ab(b 2 - a 2 ) s(b 2 - a 2 )

43. (S2 + 1)2
91. (a) (S2 + a2)(s2 + b2)' (S2 + a2)(s2 + b2);
~~
1.
2
45. 14s
(S2 - 49)2 (b) b2 ~ a2 [.; sin at - i sin bt]'
3.~ I
.l[_l_+_l_]= s-7 b 2 _ a 2 [cos at - cos btl -3
47. 2 s - 6 s- 8 (s - 7)2 - 1
4
5. S2 +4 s+Z
49. (s + 2i + 16 93. (a)1~4~s=-I,(b)}i..~s~1 =3, = ll.e- 3f + 11. e 3/ _ .!Q. e4/
s-5 2 5S 3 7. Y 42 6 7
51. (s _ 5)2 + 49 (e) lim _s__ = +00, (d) lim -2-- = +00. The
S--I'''''' 4s + 10 S + 1 $-)00:>
9 . _1_(e-,-,"2 + s) + __8_ =
2
S2 + 1 53 -128 8s inverse Laplace transform does not exist for any of these
. (S2 + 16)' s2 + 16 (S2 + 16)2 functions.
11. 1,(e- 3S (2s
s-
+ 1) + s - 1) s
13.
2e- lOs
- -s- + -
1
55. (s2 + 9)2
1[S+ 4k2 + -;1] = s(S2
S2 + 2e
95. C = t and T = 0.44
s 59. "2 S2 + 4k2)
k
15. S2 + k2 V2(S+l)
61. (a) -Z- V2(S-l)
S2 + 1 : (b) -2- S2 + 1 ; Exercises 8.2
21. (s _ 1)2
s- 1
+9 1 (V3s- 1)
(e)"2~;(d)"27+1
1 (s + V3) 1. Y = te-8f - fe-3f 9. y = -4e -, - cos 2t - Z sin 2t
23. 5 , v:;;: y
(s + 1)' + 25 63. (a) ~{t-LI2} =,...;-;;;; (b) ~{t1l2} = zs 37, 0.4

6
0.2
Z 64. (b)f(t) = e b ' sin t, b > 0, is of exponential order b, but
25. (s + liZ)'
lim (e b' sin t)(e -b,) = lim sin t does not exist. -0.4 20.20.40.60 It
t-).<Xl (---tOO .2
-18
27. s - 3 67. t
s 69. te 2t
29. S2 + Z5 lt 2 e- 6t
71.
S2 + 5s + 25 Z
31. S(s2 + 25) 73. 4te 3 ' 11. y = 1;6 (2 cos 2t + 154e
6
cos 2t + 3 sin 2t-
-16
33. s - 2
75. cos 4t 3. Y = -te-5' - 4e 2t
456e 6
' sin Zt)
'
2 -2t + 7 e 7t
77. "ge "g
5040
35. ---;s 4 -0.5 0.5 1.5 t [.5
79. .l + .le '
2 2 2
37. (s + 3)' + lest
.5
81. le6t 0.5 t
120 2 2 -1.5 -I -0.5
-0.5
39. (s + 4)6 _le- + l..e Zt 6t -1
83. -1.5
s2 - 9 8 8
-8 -2
41. (S2 + 9i 85. e' sin t
39. t - (t - 4)"lL(t - 4) + (t - 8)"lL(t - 8) - 49. x(t) = sin(t - 7T)"lL(t - 7T) + I - cos t

13. Y = - 47
70 e
-31 1"
-We
53
- 238 e -
41
no
1
cos t + (e) X{J,(x)} = _ ,,-:-:
v s- + l(s +
_ ,,-:-: ,
v s- + 1)
(t - 12)"lL(t - 12) + (t - 16)"lL(t - 16) + ...
13 .
no smt X{J2 Cx)} = ----r=~-'I~-;===-,: 41. t[sin 4t - sin(4t - 12)"lLCt - 3) + sin(4t - 24)"lL(t -
v?+'l(s + W+J)2 '
6) - sin(4t - 36)"lL(t - 9) + ... J
X{J3 Cx)} = ----r=~_I'-----==~
W+J(s + W+J)3'
-I -0.5 0.5 It 1. 1 )
X{J4(X)} = ----r=~-'I'-----==~ 43. ( "8 sm 2t - '4t +
-2
v?+'lCs + W+J)4 ' 1 1 )m.
-4 3 ( '2 - '4t +"8I slD(2t
.
- 4) '!L(t - 2) +
X{J,(x)} = ----r=~_I~r==~
-6 v?+'l(s + W+J)'
9( 1 - t t + t sin(2t - 8»)"lL(t - 4) +
51. x(t) = "lL(t - 7T)(e~-1 - e2~-21)
Exercises 8.3 3
27 ( '2 - 1
'4t +"81. )m.
slD(2t - 12) '!L(t - 6) + ...
-28
15. Y = 2t'
LG
1.~ 0.25
17. y= 2t2 0.2
3 - e 4s
19. Y = 2t 2 3.
~ 0.15
0.1
12 cos 3t
25. Y = e -21(13 39 + 61t )sm
+ (37 . 3) 1
t + 13 -7 - te s + e 5s
o:L
5. 0.05
0.3
0.25
-42e- 4 , I 2 3 4 5 6
7. s- I
O'''[L'
0.2
t
0.\5
0.1
0.6 9. -12 0,05
0.4 e"(s2 - 1) I 2 3 4 1 53. xCt) = -1.e 2r.-21"lLCt - 7T) sin 3t + 2... sin t -
0.2 3 40
-14
I I e -21' 3 I -21 3
2 4 6 7 10 I
11. e2=/'(:,2 + 1) 40 cos t - Uo sm t + 40 e cos t
27. (a) y = J,,(x) => " = yeO) = JiO) and f3 = y'(0) = eS -s -'1 47. Y = [ _ _7T_ et- L + ~e31-3
13. s2e'(l - e ")
4 + 7T 2 36 + 7T-
~
1
+ 7T2~~~ + 7T2)
z-(Jr'(O) - J,,+,(O». O.IS
e' + 2 + (4 cos(f(t - 1») 0.1
15. s(e" + e' + 1)
+~2~3:7T: 7T2) sin(f(t -
0.05
17. e-= + (4 1») ]"lL(t - 1)
p. 0 2 4 6 7 9 [ 2 3 5 1
19. e- s + e- 2s 2 -0,05
8 + 7T e' + 48 + ~ e3t
o o 0 0 0 0 0 0 0 -'= -0.1
" 21
•
~+e-~/2
S2 + 1
2(4 + 7T 2 )
2
2(36 + ~)
f3 0 In 0 0 0 0 0 0 0 0 + 48 - 47T cos (.:!!..t)
23. - 3"lL(t - 7T) (4 + 7T 2)(36 + 7T2) 2
+ 1. cos t _1. cos 3t
7T2~~~ + 7T2) Sin( ft)
25. 2"lL(t - 1) - 3"lL(t - 4) 55. x(t) = _1."lL(t - 7T) sin 3t
(b) p. = 1 => Yes) = c, _ ~ + C2, P. = 2 => Yes) = - (4 + 3 8 8
27. -3"lL(t - 6) + 3"lL(t - 5) - 4"lL(t - 3)
2'OU
vs 2 + 1 ,
2s' +1 S(4s2 + 3) 29. e - 4 "lL(t - 4)
~
CI _ ~ + c,s, p. = 3 => Yes) = CI _ ~ + 31. cos(2t - 6)"lL(t - 3) 0.4
V S2 +1 V s2 +1
8s 4 + 8s 2 + 1 33. +(e,,-2' - e 21 - 10 )"lL(t - 5)
200 0.2
c2(4s 2 + 1), p. = 4 => Yes) = c, _~ + 150
vs 2 + 1 100 1 ., 3 6 t
c2s(2s 2 + 1), p. = 5 => Yes) = 35. 2[ -t + (4 - 2t)"lL(t - 2) + (12 - 2t)"lL(t - 6) + '0
-0.2
s(l6s 4 + 20S2 + 5) (20 - 2t)"lL(t - 10) + (28 - 2t)"lL(t - 14) + ... J -0.4
c, v?+'l + c2(16s 4 + 12s2 + 1) 37. cosh(t - 4)"lL(t - 4)
05 I 1.5 2r
d
57. x(t) = -"U(t - l)e 2- Z

+ "U(t - + e- 2'
te- 1 - e- t
, I)e ' -' + 67. yet) = -te-' + te-"'lL(t) + (10,000 - 10,OOOe h -' +
20,OOO7Te'~-' - lO,OOOte2~-')"U(t - 27T); ~y(t) = 40
yet)
9.
x(t) = t t e' + e - 4'
0,4
0.3
10,000; no, the limit is not affected if the forcing
function is changed to 1000(1) + 10,OOO"U(t - 27T).
zo
l'
1 yet) = le' _le- 4 '
5 5
0.2 Exercises 8.4

0,1
1. it3 -40
x(r)
0.5 1.5 2 t
3. t - 1 + e-' -1
x(l) = 4e' sin 21 + 2e' cos 2t
3
5. -32"t+ I t 3 +T28sm
3 ' 4t 3. { y(l) = -4e' sin 2t
4 -2
I 10
200 -3
61. x(t) = 3"U(1 - 7T)e3~-3' sin(t - 7T) 7.
s(s + 1)
100
63. x(t) = 2e -3, sin t + 3"U(1 - 7T)e3~-3' sin(t - 7T) I
9.
S'(S2 + I)
-10
eS-l eS-l
65. (a) see' + I); (b) sl(e' + I);
e'TT'S + 1
H.
S2(S -
1
I)
-100
x(t) = - (0 e-' - t e2' - 325 e 4 ' - ~6 e- 3f
(c) (S2 + I)(e"'" - I)

(d), (a) x(t) = (I - cos t)"U(t) + (-2 + 2 cos(t - I))
13. t- 1 + e-'
15.
I
"2 t
,
sm t
-10 -200
n.
1 y
(I) =.1..
10 e
"+1. -, + J... e + ..!1.e-3'
5e 70 35
4f
"U(t - I) + (2 - 2 cos(t - 2))"U(t - 2) + X(I) = -2!..e' + 8e 6' + 2. e -3,

2 2
(-2 + 2 cos(t - 3))"U(t - 3) + ... 1 3 I I. 4 5. yet) = - 3e' - 2e 6' + 5e -3,
14 Y(I)
17. %t - 256 t + 1024 sm I ( 12
Z(I) = 15e'
M
10
0,6
I 10, I 5 '4
0.4
o.z
19. li6e. -1i6'cos 41 - 232 sm I
300 1.5 2 I
4 1
2~0 I cos 101 + 20 00
-0.2 1 2 3 1 200
-0.4 21. - sin 101 -5
100
'::(1)
23. g(l) = sin I
(d), (b) x(t) = (t - sin t)"U(I) + (2 - 2t + 2 sin(t - I)) -14 -12 -10 -8 --6 -4 -2 -10
5 _ r;c 2Vz, _ r;c 10_" 0,2 0,8 l'
"U(t - I) + (-4 + 2t - 2 sin(t - 2))"U 25. h(l) = "3 cos( v 21) - -3- sm( v 21) + Te -100
(t- 2) + ... yet)
-200
27. Y(I) = (J...10
15 + 1.t4)e
4
2'
x(t) = -61 cos 2t + 6I cos 41 +6
I, 5, 4
sm 21 - 6 sm I
0.6
d 2y
"l
0.4
0,' 35. ""dt2 - 4di +
dy
4y = 2t 3e '
Z
+ 3t 2e", yeO) = 0, y'(O) = 0 13.
1 y(l) = -41 cos 21 - 43 cos 4t + 6I sm
'2 7 . 41
t -12 sm
It-
39. (a) ( 5 I )e" +
25 I cos 4t -
25 3,
100 sm 4'I,
, I
(d), (c) x(t) = "2( -t cos t + sin t)"U(t) + «t - 7T) cos t -
(b) I~ t 7 - tt
5
+ 3
15t - 45t + ~ Vz sin(tVz); 25
sin t)"U(t - 7T) + «t - 2 7T) cos t + sin t)"U

(t- 27T) + ... (c) -11.12
6
+ 121
n + J...
~
cos 3t - 2. cos t
2
20
15
-2 -1 2 x
1.5 10
UiD,D,
r 2 4 6 8 10 12
Exercises 8.5
X(t) = e- 7' - e"
1. { yet) = 3e- 7' + e 5 ' l'
-2
d
1 15 I 13 2< 3 -2' 5. 1(1) = (1- 1 + e-') - (t - 1)"lL(1 - 1) +

x(t) = t( -7T cos 1+ 7T cos t "lL(7T - I) - X(I) = 2 e -', + le- 3' _ X(I) = TO cos 1+"8 - 41 + 40 e -TOe (I - 3 + e -C'-2))"lL(1 - 2) - (I - 3)"lL(t - 3) +
15.
1
I cos I "lL(7T - I) + sin I "lL(7T - I))
y(t) = t sin 1(7T - 7T"U(7T - I) + I "lL(7T - I))
2 2
[-f - f e -'('-3) + e- 3U - 3)]"lL(t - 3) -
2~
1 10 4 20
3 3
y(l) =...L sin I _ l - .!....e2' _ _ e- 2'
5
(t - 5 + e- U- 4 ))"lL(t - 4) - (I - 5)"lL(1- 5) +
7. I(t) = 100te-"
...
[1.. + .2.. e -'U-2) - 1.. e -3« -2)]"lL(t - 2)- 9. 1(1) = tY3e-'/2 sin(,? t) -
5 10 2 60
-10
2 + .2.. e -'U-I) - 1.. e - 3U-I)]"lL(1 - 1) 40
[ 5 10 2 20
tY3e-'I2+.-I2 sin(,? (I - 7T))"lL(t - 7T)
+ ...
19. 11. (a) 1(1) = e- 4'+"'lL(t - I);
y(t) = 2 e -" + 1.. e -3, -
-20
0.' 3 '
(b) /(1) = e-" + e-'r+"'lL(1 - 1)
2 2
13. 1(1) = I - 1 + e-' - (I - 2 + e-'+I)"lL(I- 1)
[.±..
15
_le-'U-3) + 1.. e - 3U-3)]"lL(1 -
-2 -40
1 3)
yet)
05 15 5 3
~<"-+--~M+--+r
-60
15. 1(1) = e 2-'''lL(t - 2) + e 6 -'''lL(t - 6)
-2
-05
-I
-1.5
+ [.l..
15
_.2..10 e -'U-2) + 1..6 e -3U-2)]"lL(t - 2)
19. X(I) =t{{-ICOS I + sinl)"lL{I) - «7T-I)COSI+
+ [.l.. _.2.. e -'U-I) + 1.. e - 3U-I)]"lL(I_ I)
15
+ ;.,.
10 6 X(I) = - t - t e' + 2 sin 1- 3 cos I -
sin 1)"lL(t - 7T))
4v2 . _ r,;;: 29 _ r,;;:
25. -3- sm v21 + 6 cos v21 21. x(t) = 3 cos 3t - 3"2 sm
.
31
x(l) = -e 2 ' cos 21 + (-1.. - 1.. e2' sin 21

2 2 0.8 1 y(l) = -1..1
2
_l2 sin I - cos I 23. x(t) = e -2'(21 + 1)
+ teo< cos 21) + (l + e 2U - 1) sin 2(t - 1) -

0.6 25. x(t) = te-"(4e3' - 1)
15
e 2(<-1) cos 2(t - 1))"lL(1 - I) 27. X(I) = e- 2'(4 - 4e' + 2t + 2te')
10
+ (-1.. - 1.. e 2U-2) sin 2(1 - 2) +
2 2 29. x(t) = te -4r(l - e 3' - 31 + 6Ie")
te2('-2) cos 2(1 - 2) )"lL(t - 2)
-5 2
17. 31. X(I) = 12 4 [«7T - 6) cos 7TI-
y(l) = _1..2 e 2' sin 21 + (-1.. + 1.. e 2< sin 21 -10 36 + 137T + 7T
4 4 e- '(-12e + 18e'+2 - 3e 7T 2 + 27T2e'+2 +
3 3 3
-15 2
57Te 3' sin 7TI)"lL(1- 1)) - e- 3'(-12 + 18e' - 37T +
+ 1.. e 2' cos 21) + (1.. - 1..2 e 2(<-1) sin 2(1 - I) +
4 2 27T 2e' + e 3'(7T 2 - 6) cos 7TI- 57Te 3' sin 7TI)"lL(t)]
te2(<-1) cos 2(1 - 1))"lL(t - 1) 5 15 _ (1 _
27. x(t)=I-TlI+illv22sm-ztv22-
=. =) 33. x(l) = -te -2'«3e4 + e 2' - 2te 4 )"lL(1 - 2) -
+ (-1.. +
4
1.. e 2U-2) sin 2(t -
4
2) +
21. v 2sm
x(l) = I + 1 + 2 . v 221 + cos v212 ' I
- sm 9"5 cos _=) + 9"5 cos t,y(l) = Tl5 +
(I-ztv22 (-2 + e 2 + e 2' + 41 - 21e 2 - 212)"lL(1 - 1) - 2t 2 "lL(t))
{
±e2U - 2) cos 2(1 - 2) )"lL(t - 2) 6 35. x(t) = 118 e -1«3t sin 3t "lL(t) - (67T cos 3t - 31 cos 31 +
y(l) = I + v2 sin v21 - cos v21 + cos I
tv'22 Sin(ttv'22) + 1 1 cos(tlv'22) - %sin I
sin 3t)"lL(1 - 2 7T) + (37T cos 3t - 31 cos 31 + sin 31 +
37T sin 31 - 31 sin 3t)"lL(1 - 7T))
..rCr)
60
Exercises 8.6 1
60 1. Q(t) = 2 - 2e-' - 2(1 - el-')"lL(t - 1),I(t) = 37. x(t) = 3(7T4 + 17,1 + 16) [(3(,1- 4) cos 7TI-
'0
2e-' - 2(1 - el-')o(l- I) - 2e l -'''lL(t - I) e- 4t(15'l7'e 4t sin 7Tt + 16e 3t + 1 - 4e 4 + 7T 2 e3t + 1 -
'0
30 3. 1(1) = t(l - e-') - t(l - e-('-I))"lL(t - 1) + 47T2e 4 )). "lL(t - I) - e- 4 '(3e"(7T 2 - 4) cos 7TI + 16e"
20 -2 + 7T 2e 3' - 47T 2 - 157Te4 ' sin 7TI- 4)"lL(t)]
10
60 25 1
-4
-6 -2
t(l - e- C'-2))"lL(t - 2) - t(l - e- U- 3))"lL(t - 3) +
39. X(I) = /5 (sin t - ± sin 41 - 15 sin(4 - 4t)"lL(t - I))
(d)Ur:
41. x(t) = te -21(e" sin(3t - 9)OU(t - 3) + 0.8 1.5

7
x(l) = 18 cos t + ;~ cos 21 + 112 I sin 2t
89. See Instructor's Resource Manual and Student Resource
Manual.
e' sin(31 - 3)OU(t - I» 0.6
1.25
I 71. ( 7 7 I. 2
0.4 0.75
yet) = 9" cos t - 9" cos 2t -12t Sill I
127150 -SI 250 1250 . x(l) = -tv'3 sin v'3t
43. x(l) = 200 + --13-e + 13 cos t - 1"3" Sill t, 0.2
05
1.;;;3· 2 1 1 ' 2 91.
0.25 0,(1) = -gVj Sill v'3 + gSIll t ( y(l) = +v'3 sin v'3t
bounded
~-
5 ill ~ W 25 S ill ~ W 25'
' 73. ( e,(t) = -1..v'3 sin 1.. sin 21
45. x(t) = 200 + 63 6;0 e -SI 12;0 cos I - 25 0 sin I; 4 v'3 4
(e)~
-
1 13 1.5
bounded 1.25 O'(t) = -t cos ~ + -t cos 21
I
75. 2t
47. x(l) = -250 + 5350e zI - 200 sin 1 - 100 cos t, 0,75 ( 0z(l) = cos v'3 - cos 2t
unbounded 0.5
0.25
49. x(t) = -10000 + 17500e ' + 2500 cos t + 2500 sin 1 + 5 10 15 20 25 t ° ,
(t) = v'3 sin
4 v'3
+ 1.
4
~sin 2t
(10000 - 10000e ' - s + 2500e' - s cos 5 -
2500 cos 5 cos(t - 5) + 2500e ' - s sin 5 -
2500 sin 5 cos(l - 5) - 2500 cos 5 sin(1 - 5) +
(f)t:.
1.75
\.5 .'
77.
1
• ;;;
v3 . 2t I. 2
O,(t) = ""2 Sill v'3 - 2" Sill t
0,(1) =t[4 cos 2t+ 4 cos(~t) + sin2t-
2500 sin 5 sin(1 - 5»OU(1 - 5) 1.25
0.75
, .
79.
0,(1) = -± cos ~ + ±
cos 2t
v'3 sin( ~I)]
51. x(l) = -5000{ I - te'
±C-2 + 2e t - 1 + e l 1
+ -t(cos
cos 1-
I - sin I)
cos t - e t - 1
+
sin 1 +
o.s
025
1° Z
(t) = -1.. cos ~ - 1. cos 21
2 v'3 2
93.
02(1) = -cos 2t + cos( ~ I) - ±sin 21 -
- . 5 10 15 20 25 I
((\/2k+k;;')
Vm ,.
v'3.(2)
v'3 1
S111
sin t)OU(1 - I) + (I _le'-'
2
+ 1..(cos(l -
2
2) -
59. (d) X(I) = cos 51 - cos 5v'3/, y(/) = 81.
x(t) = -cos 4
sin(t - 2» )OU(I - 2) + ... }
2(cos 51 + cos 5v'3/)
. 2 2
1 y(t) = cos ( t~)
Vm "
x(t) = 5 cos' I - 5 cos V61 max. distance = 1 for x and y.
53. X(I) = e ' - Z(lOOe 2 + 200OU(1 - 2» => x(5) =
e'(200 + 100e z ) = 18,858 63.
( Y(I) =
4
5 cos I + 5 cos
I.
v 61
r;
83. x(t) has frequency w" where w,-, = 9k kVU
2m -~;
, 9k kVU
55. x(/) = _I_,e-"'(-c + ce'" + ck - ck' + ck'e'" + Y(/) has frequency Wo, where Wo- = 2m +~
k+k X(I) =tcos I + tcos 21
kxo + k'xo - cke'" cos I + ck 2 e'" sin I)
65. 2 2 = 292175000 -100(5+Zv6)1 _
±~[(I - ( Y(/) ="3 cos I -"3 cos 2t 87. ( a) Q()

t 159587072641 e
*
57. (b)x(/) = e- a (I-4n)OU(1 - 4n)-
n-Q a 7387691623 -100(5+2v6)1 +
(1 - e~(1-4n-'I2)OU(1 - 4n - -t)] x(l) = cos v'31 + cos t --5z 1276696581128V6 e

95. See Student Resource Manual.
(e) y(/) = n~o co[ t(l - e- (I- 4n mu(t - 4n)-

b 67. 2 2
( y(l) = -5 cos v'3t + 5 cos v'2
I 292175000
159587072641 e
400V6t-'OO(5+2v6)1 +
_1_(e- a (I-4n) + e- b (I-4n)OU(1 - 4n)] _ 7387691623 400V6t-'OO(5+2v6),

b-a ( I I . 2t+ 7 cos 2t+jgSillt
XI)=4"+jgSill I. + 1276696581128V6 e
12
nto co [ tCI - b
e- (Hn-lIZ)OU(1 - 4n - -t)- 1.. cos t - 1.. t cos
6 6
I
584350000 204799950 .
159587072641 cos 3771 - 159587072641 Sill 377t
_1_(e- O(I-4n-ll2) + 69. y(l) = 1.. - ...!.. sin 2t - ..2. cos 21 + ± sin t + (e) 1(1) =
584350000
b-a 4 18 12 9 159587072641 cos 377t-
e -b Ct -4 n-II2)OU(1 - 4n - -t)] 1.. cos t - 1.. t cos t 204799950 . 7 -1.5

3 3 159587072641 Sill 3 71
97.
{
Xl(t) = -2 sin 1
X2(t) = sin 1
45. x(t) = t cos 1 - t cos 31 - + sin(3t)OU(1 - 1T) CHAPTER 9
X3(t) = 2 sin t
47. g (t) = 1 + sin I - cos I
Exercises 9.1 L'e- 4X (e 2x sin m=)(e" sin n=) dx =
1
51. (3)_S_ (b) 1. y = 0
s' - I
S3
s' - 1
S' + 2 3. no solution
II
D
sin m7TX sin m1TX dx =
(e) + 4 (d) Sin(m - n)= sin(m + n)=]l = 0

S4 S4 + 4
S' - 2 2s [ 2(m - n) 2(m + n) D •
-I (f)
(e) S4 + 4 S4 + 4 37. An =1+ n'''?-, n = 1,2, ... , Yn(x) = X sin(n1T In x)
7. y = 0
53. x(t) = - 2 sin f + 115 eos(2t)OU(1 - 1T) - 9. y = C sin x 39. y(x) = t x ' - x

/5 sin(f)ou(t - 1T) - /5 eos(2t) + 115 eos(f)
11. y = 21T cos 2= - sin 2=
41. y(x) = C sinx - + sin 2x
1. 1.. -?- 3.
IOboo - IOboo
13. y=O
45. AI = (2.02876i = 4.11586, A, = (4.91318)' = 24.1393,
5(s4 + 24)
55. Q(I) = -220,( eos(1001 - 200») 15
•
A - {O,2 n = 0
,,- n 71'2, n = 1,2, ... A3 = (7.97867)' = 63.6591
5. 7. (s - 2)-2
OU(I - 2) +
11 11
500 - 500 eos(1001) y (x) =
n
{I,cosn = 0
n7TX, n = 1,2, ...
S' - 9 Exercises 9.2
9. 6(s - 1)-4 11.
(s' + 9)' 57. /(1) = 100Vze-' sin(2VzI) _ [(2n - 1)1T]' _ . (2n - 1)=
17. A" - 2 ' y,,(x) - sm 2 ' 1. aD = 1. an = +,[(-1)"
n-1T
- I], n;': 1;
s+5 e- 3 7T'sf2 59. x(t) = 10000e" + 100e"-"'lL(t - 1)
13. 15. n = 1,2, ... 1 ~ 4
s' + lOs + 34 1(X) = -2 - L
n=l
(? _ 1)2 , eos(2n - 1)= =
_n 1T
2(3 - 2e'S) -42e'-s 61. y(x) = ~(X4 - 2x 3 + x') 19. An = ±(1 + k/). where tan k" = 2k", n = I, 2,
17.
se 7s
19.
s- 5
24 1..2 - ~ cos = - ...,'!- cos 3= - ...
Yn(x) = e -xl' sin(knx) 1T- 3-.".'
2 4 4 65. {X = ~1 . 3
e' 2 4
21. s3 e Zs + s2e'L<; + se'2s Y= -2 + 2e' - 2 e2' 21. An
1
= 2(1 + k/), where tan k" = 2k", n = I, 2, ... , 3. aD = 3' an = n2 1T' (-1)", n;': l;J(x) =
1 ~ 4(-1)" I 4
2s
e -2e +l s
eS-l Yn(x) = e -x12 sin(k"x) - +L -,-,- cos = - - ---:;- cos +
-f + e' - fe"~ +
n1TX TTX
23. s'eZs(1 _ e-Zs) = s'(eS + 1) 3 ,,~I n 1T 3
x= 71"
s
Tre + 7T 1Te s 23. A" = t(4 + n'''?-), n = 1,2, ... , y,,(x) = e2x sin n= _4_ cos 2= + ...
(+ +
25.
eS(s'
27. -2 sinh 5t
29. t' - 3
+ 1T2)(1 - e Zs) (e S - 1)(s' + 1T2) 67.
j y = -"31 + "31 e " -
2
(1
ie"-6 - te'-' )OU(t - 2)
-"3 + "31 e " _)

6 OU(t - 2)
25. sex) = I, Ym(X) = cos m=, Yn(X) = cos n=: if m
Iol cos m1TX cos n 1TXdx =
* n, 5. aD
7. ao =
2'1T 2
= 2, an = 0,
2, an = +,(-4
n
n ;,: 1,f(x) =1
+ 4cos n7r + n7T' sin n7T),
71' 2 2
31. e -5t cos t - 2 69. f X(I) = i(2Vz sin(VzI) + V3 sin( Jd) Sin(m - n)=
[ 2(m - n) +
sin(m + n)=]1
2(m + n) D = o.
n;': 1,f(x) =
~ 2 (
1 + ~I n21T2 -4
n 7 1 ' . n1T)
+ 4 cos 2 n=
+ n71' sm 2 cos -4-
33. 7OU(t - 7) + 6OU(t - 3)
35. 8 cos(t - 3)OU(t - 3)
lY(I) = t( -2Vz sin(VzI) + 4V3 Sin( Jd) 27. sex) =
. (2m - 1)=
1, Ym(X) = sm 2 . y,,(x) =
9. aD = -2(e- 1 - 1), a" = _,_;_ _ [(_I),,+l e-1 + 1],
. (2n - 1)= . n-71' + 1
I- cos V3t + Isin I)
lf
1 sm 2 :lfm*n,
37. 2 7 e- 3 '(9I'e" - 6te" + 2e" - 2) X(I) = t(eos n;': l,f(x) = (1 - e- I ) +
71. r l . (2m - 1)= . (2n - 1)= _ ~ 2
)' _,_,__ [(_I),,+l e -1 + 1] cos n7TX =
39. yet) = e -" sin I yet) = t ( -cos I + cos V31 +I sin t) JD Sln 2 SID 2 dx-
;;-'1 n-71" + 1
41. yet) = 6t' +1 Sin(2m - 2n)= _ sin(2m + 2n + 2)=]1 = 0 I
2(1 + e- I )
43. y(t) = -e -4, + 2e-" 73. x(l) = H -3 sin 21 - V3 sin ( Jd) [ 2(2m - 2n) 2(2111 + 2n) D •
= ef - 4dx = e -4\ SeX) = e -4" Ym(X)

(1-e- )+
2(1- e- I
71"+1
)
cos=+
+ (t + t e4
-
4
' - ±e'-")OU(t - 1) lY(I) = ±(3 sin 2t - V3 sin( Jd) 29. p(x)
e2x sin mrrx, ynCx) = e2x sin n77X; if m "*
=
n,
2Z'jT2 +1 cos 27TX + ...
.. ~
2
11. b. = _(-I)"+I;f(x) = I
n7T
00 2
-(-1)'+1 sin n= =
n= 1 n7T
sine series for I(x), so An = %f I(x) sin( n;x) dx, b. = ...!...(2(-I)"+1 + I), n 2: 1;
n7T
43. b n = n~[ -cos n7T + cos n; 1 n 2: I;
1.
1T
sin = _.1... sin 2= +.1... sin 3= + ...
21T 31T
31. cosine series: (a) ao = 1/2,
2
I
4 ,,=1 n 7T'
I
I(x) = - + I 22'(1 - (-I)") cos n= +
00 [
f ~[-cos
.~ln7T
n7T + cos n7T] sin!!]!!. = .±- sin = +
2 27T 2
a. = 4 4 (6 + 3( -2 + n2.".2) cos n1T);
13. b. = 2(-1)"[----f, _...!...] __ 4_
3 3
n1T ...!...(2(-1)"+1 + 1) sin n=] = l +.1, cos = +
n7T 4 7T
--±- sin 3= + ~ sin 5= + .. ,
37T 2 57T 2
I(x) =
n 7T' n7T'
f [2(-I)'[-+-, -...!...] - --f-,] sin

n 7T
n= =
(b) ao = 116, a. = -i---.(
n 7T
(-6 + n 2.".2) cos n7T)'
-6 -
' 1. sin = - ...!... sin 2= + ' , .
n=1 n 7T n1T n 7T 7T 27T 45. b = +[1 - cos n7T], n 2: I;
(c) ao = 2/5, a. = -i---.(4( -6 + n27T 2) cos n7T)' • n 7T
2(:; +~) sin = +2( -2~) sin 2= + (d) ao = 4/15
n 7T '
9. ao = (e - e
-I (-I)'
); an = 1 + n 27T 2 (e - e
-I 1
), n 2: ; 00 4
I-,-[I-(-I)"]sinnx=
2(--=i..
3 1T
+ ...!...) sin 3= + ...
3
31T 3 a. = :;-4. (-12 + n27T 2 ) cos n7T
n7T
b = (-I)"+l n7T (e _ e- I) n 2: l'/(x) = .!.(e - e- I )
n 1 + n l 7T 2 " 2
+
n=1
f
n 7T
8, sin (2n - I)x = ~ sin x +
(-I)" .~I (2n - I) 7T 7T
15. b. = .1...(1 - (-I)"), I(x) =
n1T
sine series: (a) b n = -;-23 (-6 + n27T 2 ) cos n7T;
n7T
I00 [
--2-2 (e - e- I ) COS n= +
n=1 1 + n 7T ' 3x+ 5387T SID
8 sm ' 5X+
3'7T •••
00 4 . -4 (-I)"+l n7T _, ] I -I
~I (2n - I)1T sID(2n - 1)= =
(b) b n = n 3 7T 3 (I + 2 cos n7T); 1 + n27T2 (e - e I) sm n= = 2'(e - e )-
(c) b. = +(24 - (24 - l2n 27T 2 + n4 7T 4 ) cos n7T)

I I 47. ao = 2, a. = 0, n 2: I; b. = -.1...(1 + (-I)'), n 2: I;
e-e- cos=- (e-e- ) sin=+ n7T
.±- sin x + --±- sin 3x + --±- sin 5x + ... nn 1 + 7T
2
1 + 7T 2 2
1T 31T 51T 2 2 2 2 00
(d) b. = -4(12 + n 7T + (5n 7T - (2) cos n7T) e - e- I 2(e - e- I )7T , I(x) = I + I - - ( I + (-1)") sin n= = 1 -
n5 7T s
n~:2 (n1T (cos n; -
,,=1 n7T
17. b. = cosn1T) - 4 sin n;), n 2: 1 1 + 227T2 cos 2= + 1 + 227T 2 sm 2= + ' ' .
--±- sin 2= + ~ sin 4= + . "
I(x) = Exercises 9.3 11. odd;f(-x) = (-4 = -x 3 = -/(x) 27T 47T
n~:2 (n1T (cos n21T - cos n1T) - 4 sin n21T) sin n;x
1. ao = 0; an = 0, n 2: 1; b" = -~ cos n7T =
13. neither;f( -x) = (-xi - (-x) = x 2 + X "" -/(x) Or
n7T I(x)
19. b = ~(I - e-I(-I)') I(x) = 15. odd;f( -x) = -/(x)
n
f ~(I
1 + n 2 7T 2
- e-I(-I)") sin
,
n= =
.1...(_1)"+1 it2:I;f(x)=f .1...(-I)"+l sinn==
n7T' ~l n7T' X2 x:>
17. odd;f(x) = { _~2, ; < o'/(-x) = -/(x)
° b. = 0,
71'2
n 2: I;
~ 1 71'2
n=11+ n
2
7T- 1.
7T
sin:"" - .!.sin 2= +.1... sin 3= +'"
1T 37T
I(x) = -
6
- I
.~I n
2'cos 2nx = - - cos 2x-
6
21T(I + e- I ) , 41T(1 - e- I ) , 19. neither
lcos4x-lcos9x-" .
I + 1T2 SID = + 1 + 221T2 SID 2= + 3. ao = 0; an = 0, n ~ 1; b" = ( 712 - -n-
27T2)
cos n7T = 31. hint: let x = p, 4 9
61T(I+e- l ) ,
1 + 32.".2 SID 3= + ... (~; - 2;2)(_1)", n 2: 1;f(x) =
33. hint: add the series in Exercises 31 and 32,
51. If TI = nT2' n is an integer.
7T 7T 37T
35. (a) 2; (b) 7T; (e) 7T; (d) "4; (e) 4" + e-i.rr(1 + in7T)
2
21. cos x = I + ICos 2x f (I;n -
/1=1 n
2
27T )(_1)' sin nx = (27T 2 - (2) sin x +
3 I I 1
53. c. =
_einrr(1 - in7T)
27m2 ,en =
3 3 1 2 37. (a) 0; (b) 4; (e) -4; (d) '2; (e) '2 ei.rr(1 _ in7T) + e -'.rr(1 + in7T)
23. cos x = "4 cos x + "4 cos 3x ( 2-
3 7T sm 2x + (27T
2)' -3--9 4),sm 3x+··· 2=2
,3 3, 1, 5. ao = - 2; a. = 0, n 2: 1;
39. ao = 2, a. = -~ sin n 7T, n 2: l;f(x) = I +
~( - -
2 cos n7T sm
, nx + -2-
2. . nx
)
25 . SID x=4sIDx-4Stn3x n1T 2 1(;<) = L sm n7T sm
b. =.1...(1 - (-1)"), n 2: 1; f (-~ sin n7T) cos n= = I-'±- cos = +
n=l n n 7T
27. x = I
00 2(-1)"+1 ,
SID nx on °< x < 1T; b. =
n7T
f
"~l n1T 2 2 7T 2
55. I
sin nx "" 2';
1 L~ 2'I convergent p-senes
'2 (p = > I) .
n=1
_2(-1)"+1 n:>2
n I(x) = -1 +
.~I
4
(2n - I)7T
sin (2n - 1)=
2
--±- cos 3= + ...
37T 2 I--2
n
-
n ,,=1 n
n(1 - n2 )' - = -1 +.±-sin = +--±-sin 3= 7T2
2 7T 2 00 4(-l)n+1
7T 2 37T 2 41. ao = 3' a. = --;;>[(-1)" + I], n 2: l;f(x) = + "6 57. (a) I ----;;;;;- sin n=; (b) no; (e) no, the FS for g is
n1T ,,=1
29. y(x, 0) = .tl A. sin( n;x) cos ( e' 0) = + ~sin 5= +,., 00 2 not uniformly convergent.
57T 2 I - 2'[(-1)" - I] cos nx = I + 4 cos x +
~I A. sin(n;':) =/(x);~1 A. sin(n;x) is the Fourier 7. ao = 2';1 a. = 1
n27T 2 (1 - (-1)"), n 2: 1;
,,=1
~cos
2
n
3x + ... 59. ao = 27T
2 i
0
2rr
X(27T - x) 1T2 2
- - 4 - - dx = 3' so aol2 = 7T /6.
3
d
61. (a) an = 0, n ;;: 0; b n = ..!. I~ x' sin nx dx = 5. p(x) = e2.r, sex) = e2.r; Cn = z
21. (a) Co = 7T 2/4, C1 = -3, C2 = 5-rr /16, c, = -28/9, c. = ~
-2( -6 + n2-rr2)
n3
-rr 0
cos n71', n ~ 1
ff e2xf(x)( e -x sin m;) dx; (a) cn = 81-rr 2/256, Cs = -704/225, C6 = 3257T zII024, C7 =
-7681245, Cs = 20825-rr zI65536, c. = -311296/99225;
7 • a 0 = O·, an = n7T
sin !!!!..
2
2n7T 5 9. a = I· a = _4_[2 cos!!!!.. - 1 - cos n7T]

(b) Co = 0, C1 = 37T/4, Cz = -15/8, c, = 77T/24, C4 = o ,n n2 7T 2 2
1 I~ 2 2-rr2 nZ7TZ + 25 [e cos n7T - I] =
(b) ao = -; 0 x (l - x) dx = -3-' an = -45/32, Cs = 3197T/480, C6 = -637/384, C7 =
2n1T 5 n
1597T/448, Cs = -289/192, c. = 1012517T/161280 1 a0 = 1·,n
1. a = -~
n-rr sin n-rr
2
1 I~ 4 nZ7TZ + 25 [e (-I) - I]; (b) cn =
- xZ(1 - x) cos nx dx = - , cos n-rr, n ;;: 1
3~7T - 3~-rr) -
n
13 • a 0 = 1 - ~.
-rr 0 a = __I_[cos n-rr - I]· b =
nZ-rrz2+ 25 [e'( n-rr cos 5 sin n-rr] 2' n
bn = ..!. L~ xZ(l - x) sin nx dx =
n27T ' n
7T 0 3 - nZ-rrz
2( -6 + n2 7T z ) 7. An = --4--' n = I, 2, ... ; Cn =
-..!..[(7T + I) cos n7T - I]
. n7TX 1 2 n7T
1. (a) Yn(x) = sm L' n = , , ... ;
n3
1 I~
cos n7T, n ~ 1
27T4 1 I~
2 f e'g (x)(e -xl2 sin n=) dx
2
n 7T
2
7 'n=
(e) ao = -
71'0
X4 dx = - , an = -
5
X4 cos nx dx =
71'0
9. Cn = 2 f x
e x2(e -xl2 sin n=) dx =
(b) An = I, 2, ... ;
17. (a)letx= 0; (b) x = 7T
n4
8( -6 +n 2
4
7T 2 )
cos n7T, n 2: 1; bn = 0, n 2: 1
o
4(-I)"n7T 2
(1 + 4n27TZ)' (-112n -rr' + 34 + 32n
4.
7T )-
(e) Wn = n; Jij;, n = 1,2, .. . 19. let x = f
1 I~
(d)ao=- x(l-x')dx= - - a =
7T
2-rr
0 5 ' n
64n7r{4n2-rr'.~ 3)
(1 + 4nz7Tz)'
b =
3o n 1-[1. +
1Tnn
1.(_I)n _ .±n cos n-rr]
2 21. (a) b 1 =
3
4' b, =
I
-4' bn = 0, n "" I, 3; (b) a1 =
3
4'
..!.7T I~X(I
0
-8( -6 + n 7T z )
- x') cos nx dx =
2
11. f f(x)P m(x) dx = f rcnPn(x)Pm(x) dx. If m = n, 5• bn = _8- !!!!..
n2 71'2 sin 2 a, =
I
4' an = 0, n '" I, 3
2n ~
n4 cos wrr, n ::::: 1, bn
{ f(x)Pn(x) dx = :n1-;/(x) dx = I Cn·
..!. I~x(1
7T 0
.
- x') sin nx dx =
. I
2n
cos WlT n;;: I
' If m "" n, r-1
cnPn(x)Pm(x) dx = O. CHAPTER 10
63 • Al tematmg senes. 2n + I = 0.000005 when n = 13. (See Section 4.8 for Legendre polynomials)
99999.5. Requires n = 99999 terms. 1 I1 I 3 I1 Exercises 10. 1 31. Uxx = -16 cos ct sin 4x, 4u u = -4c 2 cos ct sin 4x;
Co ="2 0 2x(l) d:x = "2; C1 ="2 0 2x(x)dx = I;
C = ±2
Exercises 9.4 5 I1
Cz ="2
z (I
0 2x "2(3x - I) dx =
) 5
8; c, =
1. u(x, y) = Cek(x-y)
3. u(x, y) = CyV'"
33. yes
= - 2 sin 2x sin 21
1. (See Exercise 17, Section 9.1) Cn =
(2n - 1)= 4 I f2x (±(5X' - 3X») dx = o. Then,f(x) = 5. u(x, y) = Ce"'+(k-1)Y 35. yes, U,
oI
1
2 f(x)sm
•
2
dx; (a) C =
n
.
(2n - 1)7T' coPo(x) + C1P1(X) + C2P2(X) + c,P,(x) + .... 7. u(x, y) = G(x) + F(y) 39. (b) ~
ax = cos(x + I) = 0, x = 7T/2 - I, u(7T/2 - I, 1) =
(b) C
n
= 8
(2n - 1}'1T z
sin (2n - 1)7T = 8(-1)"+1 .
2 (2n - I)z-rrz' 15. Co = fX(l) dx = ±; Cz = 5 f X(±(3X2 - I)) dx = t;
9. u(x, y) =
11. uxx = 2, uyy = -2
exke -ylk
I; u(O, 2) = sin 2; u(O, 3) = sin 3; The maximum
displacement, luI. is 1 in each case; (e) u( -rr/4, I) =
13. = eX sin y, uy.y = -ex sin y sin(t + 7T/4), u( 1T14, 0) = sine -rr/4) = V2/2.
(e) C = 16 . (2n - 1)7T 32 2
+ 3») dx = -136
U::rc
C4 = 9 fx( t(35x' - 30x
n (2n - l)z7T 2 Sm 2 (2n _ I)'-rr' - 2xy -2xy 41. (b) 100; (e) 100
16(-1)"+1
(2n - 1)27T 2
32
(2n - I)'-rr' 17. C1 = 3 f x dx = t; c, f = 7 (±(5x
3
- 3X)) dx = -t; 15. u= = (x 2 + yZ)" uyy = (xz + y2)2
17. u;cx=2,u tt =2 Exercises 10.2
3. p(x) =e 2x
, sex) = e2.r, Cn = 2 f e2xf(x)(e -x sin n=) dx; Cs = 11 f( t(65x S
- 70x' + 15X») dx = ~~ 19.· u= = -4 sin 2x cos 21, Utt = -4 sin 2x cos 21
= -kz sin kx cos kt, u" = _k sin kx cos kt
1 -200
1. bn = 2 0 100 x sin n7TX dx = ---;;;:- cos n7T, u(x, I) =
I
f
2
21. Uxx
(a) Cn = ~(l - e cos n7T' = 19. If x = cos 0, then dx = -sin 0 dO. If x = -I, then 200(-1)"+1 sin n=e-n'-n',
nZ-rrz + I ) 0= 7T; ifx= I, 0= O. Then, Cn = 23. Uxx = -e -t sin
X, U f = -e -t sin x
n=l n7T
2n-rr 2n+lfo 25. u= = -16e- 16' cos 4x, u, = -16e- 16' cos 4x
nZ7Tz + 1 [I - e( -I)"]; (b) Cn = - 2 - J(O)Pn(x)(-sin odO) = 3. bn = 2 (' x(1 - x) sin n= dx = ~\ (cos n7T - I);
27. U;a: = _k 2e- 12t cos lex, Ut = _k 2e-!2t cos kx 10 n 7T
nZ-rr;+ I (n7T-e
112
n7Tcos n 7T +e 112 sin
2
n;) 2n + I L~
- 2 - 0 f(O)Pn(x) sin 0 dO.
29. Uxx = -e-I<', cos x, 16u, = -16k'e-l<', cos x; k = ±± u(x. t) = Lb n
2
sin n'TTXe -n rtJ2
n=l
1
r
A-66 Answers to Selected Exercises A-G7
5. b, = 1, b 8 = 1, b n = 0, n '" 2, 8; u(x, I) = sin 2xe- 4 , + x) . n11)'

sin 8xe- 64r 25. an = 2f' [2To - To] sin (2n - 1)= dx _
o 2
4To
- (2n - 1)71'
~ ( an cos ""2
9. u(x, I) = ~, n1l1 . n1l1) . n=
+ bn sm -2- sm -2-; 3. u(x, y) = L An Sinh n7r(12-
00
n~l
•
sm -2-; An =
II . J2 (2 - .
20 2 I' n11)' -4
7. Sex) = IOx, bn = n7r cos n7r, u(x, I) = Sex) + 27. u(x, I) = To an = 0
n1TX
x sm -2- dx + ,
n11X
x) sm -2- dx = '2 0 y sin -2- dy = n7r sinh n7r/2 cos n7r
I
n~'
b n sin n"lTXe _"l7ilc 29. Ao --If~
- x dx
7TO
71' an = -2f~ x cos nx dx =
= -,
2 7TO n27T2
8 . 2'
sm n7r. b -
n -
° ,n_
> 1
5. u(x, y) = L Bn sinh n7r(2 - y) sin n=;
9. Sex) = 20 - lOx, bn = 1Q.(cos n7r - 2) u(x I)

n'1T' "
=
?
n;7r(cos n7r- 1), b n = ° 11. Zlx = u,Yx + u~x = u,. + us; Uxx = Urr + 2urs + uss ; uy = 2
n=l
urTy + U~y = u,. - us; u yy = Ur,. - 2urs + u ss ; Urs = 0; Bn = n7r sinh 2n7r (cos n7r - 1)
Sex) +L b n sin nme -n'-n', 31. (a) Sex) = -ix3 + (T' - To + i)x + To; (b) vex, I) = u(x, y) = F(x + y) + G(x - y)
n=1
~ , 13. Ux = u,.; Un = u,.,.; Uxy = U,.r + U rs ; u(x, Y) = 7. u(x, y) = L Bn sinh n11)' sin n=; u(x, 2) =
11. Sex) = To (1 - x)
2
L Bn sin nme-n'-n'" Bn = 2f [I(x) - Sex)] sin n= dx' F(x + y) + G(y) n=l
n=l 0 '
L Bn sinh 2n7r sin n= = sin m => Bl sinh 271' = 1,
13. Sex) = To (c)Bn = 2 f[ -X(T' + i)] sin nm dx =
15. B' - 4AC = (-2)' - 4(1)(1)
17. s> - 4AC = (-4)' - 4(1)(2) = 8 > 0, hyperbolic
= 0, parabolic n~'
1
15. Sex) = T- ~(-Te-'
e-e
+ T- T, + Toe-')e X
+ ..!.~(6T+l) 19. B' - 4AC = (If - 4(1)(1) = -3 < 0, elliptic
Bl = sinh 271'; Bn = 0, n '" 1
3 n7T' 1
~(T
e-e
- eT + eTo - T,)e- X 21. B' - 4AC = (2)' - 4(I)( -8) = 36 > 0; u= = 9. u(x, y) = ao + boy + L (an cosh ny + b n sinh ny) cos nx;
16u,.,. - 16u'-$ + 4uss ; U::cy = -4u,-,. - 2urs + 2uss ; Uyy = n~'
17. Ao = 1, an = 0, n ;:,: 1; u(x, I) = 1 U rr + 2urs + u ss ; Uxx + 2uX)' - Uyy = - 36u'-$; uCr, s) = u(x, 0) = ao +L an cos nx = 1 + 4 cos x; ao = 1,
19. Ao = 2. a, = -2, an = 0, n '" 2; u(x, I) = F(r) + G(s); u(x, y) = F(y - 4x) + G(y + 2x) n=1
°
r
at = 4, an = 0, n ;::: 2; b n = 0, n ;:=: 1, b o = -1
2 - 2 cos 2xe- 4r 23. B' - 4AC = (Of - 4(0)(1) =
35. Sex) = 50 + SOx,
21. Sex) = To, {:(O~ t;== viO, t) = 0, t> ° , bn = 2 f [x(1 - x
3
) - (50 + SOx)] sin n= dx =
25. s> - 4AC = (Of - 4(1)(1) = -4 < ° 11. (a) Yn(Y) = sin n;:, An = (nb
7r
(b) an = 0;
o n11)'
vex, 0) = u(x, 0) - Sex) = J(x) - To 2( -24 - 50n4 7r 4 + 4(6 - 3n'7r' + 25n4 7r 4 » 27. u(x, t) = t[sin 11\x + I) + sin 11\x - t)] = sin = cos 111 (c) u(O, y) = L
00 •
Bn sm -b- = J(y) => Bn =

. (2n - l)m n571's cos nrr; n=l
vex, I) = X(x)T(t), Xn(x) = sm 2 •n =
29. u(x, I) = t[t sin(x + I) + t sin(x - I)] - b2Ib0 J(y) sin -b-
n11)'
dy
1,2, ... , An = [(2n; 1)7rJ. n = 1,2, ... ; 1 r +, 1 .
x
2" Jx - t 8" sm v
d 1.
8" sm x cos t - "8 sm x sm t
1..
~ . (2n - l)m
v= 13. Bn = 12 I'
0
2To
To sin n11)' dy = -;;:;;- (1 - cos n7r)
vex, t) = L an sm e -Apt,
.2 nt, C~1I1 +
f'
n=1 nm
(2n - l)m 31. u(x, I) = AI +B + (an cos
15. u(x, y) = L Bne -n",fa sin -a-'
00
Bn =
an = 2 0 [I(x) - Sex)] sm 2 dx, u(x, t) =
Sex) + vex, I) . cn1l1)
b n SID p
cos n= A = 1 11' p;
0 g(x)dx,B = P -2
a
I
0
a
n=l
J(x) sin -nm dx

a
P
22. Sex) = To, vex, I) = X(x)T(I), Xn(l) = cos (2n - 1)=
2 '
-1
poP
fP J(x) dx, an = -2I
0
J(x) cos -n= dx, b n =
P 2
17. Bn='2
I' . nm -4
xsm--dx= n7r cosn7r
I
0
P 2
n = 1,2, ... , An = [(2n; 1)7rJ. n = 1,2, ... ,
Exercises 10.3
2
cn7r 0 g(x) cos p
n=
dx
vex, I) = i
,,""1
a cos (2n - 1)= e- A., A
n 2 • 11
= [(2n - 1)7rJ2
2 ' 1. u(x, t) = cos 111 sin =
19. Sex, y) = ao + boy + L
n=l
(an sinh n7r( 71' - y) +
Exercises 10.4 b n sinh n11)') cos n=; Sex, 0) =

n -- I , 2, ... , an -- 2 f' 0 [I(x) - Sex)] cos (2n -2 1)= dx
3. u(x, t) = 1/(371') sin 3111 sin 3=
~
1. u(x, y) = L Bn sinh -2- sm -2-; Bn =
. n=. n11)' ao + I. an sinh n7T
2
cos n'1TX =
5. u(x, I) = 1171' sin 111 sin = +L an cos n1l1 sin n= n~'
n~'
23. u(x, I) = Ao +L (an cos nx + bn sin nx)e -2n", c

1 12 n11)' cos =, al = -.-1_" an = 0, n '" 1; Sex, 71') =
n=l _ -2(1 + cos n7r) n ' --- (1 +y)sin-dy= SInh 71'
where an - n 2 ,,? •n ~ 1 sinh n7r 0 2
1 l~ Il~
Ao = 271' _~ J(x) dx, an = -;; _~ J(x) cos nx dx, 2 b o7r +L bn sinh 7r'y cos n= = 0, b n = 0, n ;:,: 0;
1 2 - 6 cos n7r
b n = - Il~ J(x) sin nx dx
7. u(x, I) = L (an cos nt + bn sin nl) sin nx; a2 =
n=l
1,
n7T n71'
n~l
sinh 71'(71' - y)
To -71
an = 0, n * 2, b n = 0, n ;:,: 1; u(x, t) = cos 21 sin 2x sinhT U(x, y) = sinh 71'2 COS =
A-68 Answers to Selected Exerc;ses Answers to Selected Exercises A-69
23. B m" = (l~I) ff (x + y) sin m= sin n7l)l dx dy = n 2: 1, u(x, y) = I (b" sinh n7T(1 - y) sin n7TX + 24. u(r, 0) = ~ rUb" sin nO, b" = -;
I 2 f" f(O) sin nO dO, (e) Ao = - 1
27T
f"orr aZ(1 + aZ) dO , + 37T )/15,
= 7T-(5 2
n=l n=1 0
~[-4mn7T
m n-'1T
cos m7T + mn7T(1 + 7T) cos(m - n)7T
e" sinh n7TX sin n7l)l + d" sinh n7T(l - x) sin n7l)l) n::: 1
An = -1 f" aZ(1 + aZ) nO dO = cos
- 2mn 7T cos n7T + mn7T(1 + 7T) cos(m + n)7T]
2
41. a" = 'nh 2
SI
2
n7T
f'
0
x(1 - x
2
) sin n= dx =
25. b = -2 f" To -2To
sin nO dO = --(cos n7T - I), n 2: I
7T -"
4( -12 + n 2(1
2
+4 27T » cos n7T ,En -_ 0, n_
:> I
27. B m" = 0; A'2 = e7T J(~r + (~r = eYs, -12 csch 2n7T cos n7T b = 0
n 3 Tf , n , e"
= 0 d
, n
n 11'0 n-rr

n
1 (1) 1
I
sinh(n7T/2}
f2.
y sm(n7l)l12) dy =
28. u(r, t) = 100r
C'2 = A;;-"4 = 4eYs 0
1. (a) c = 1; (b) c = 4; (e) c = 16
I
4 I
-4 cos n7T csch(n7T/2), u(x, y) = 29. cos 2cf> = 2 cos' cf> - I; "3P2(cos cf» -"3; u(r, cf» =
29. B m " = (1)(2)
4 f2f' (xy)(1 -
0 0 x)
n7T 3. (a) u= = 0, uyy = 0; (b) u= = ge'x2 cos 3y, uyy =
4 2 I 2(y2 _ x ) 2(x' _ y2)
"3r P2(COS cf» - "3
X (1 - y) sin m= sin T dx dy
') (a" sinh n7l)l sin n= +
n~l
dn sinh n7T(l - x)/2 sin(n7l)l/2»
_ge'x cos 3y; (e) Un = (x 2 + /)2' Uyy = (x2 + y2/
1 frr = -7Tp1n frr

-8 .., 2 2 .,
= m'n'n" «2n-7T - 8) cos m7T cos n7T - (2n 7T- - 8) 43. u(x, y) = sinh y sin xlsinh 7T 31. R'(a) = ! [Jo(Ar)]I'~Q = -AJ,(Aa) = 0, Aa = 5. ao = -27T -7T
f(O)dO, a"
-'IT
f(O) cos nO dO,
X cos WTr + 8 cos m7T - 8); Cm " = 0; n, m ~ 1 sinh,y sin x sinh 2y sin 2x a,." (nth root of J,); An = a,."la f"
1 _"f(O) sin nO dO, n ~ I
bn = 7Tpn
2 f~
35. a" = -; 0 cos x sm nx dx
. =
2n(1 + cos n7T)
(n2 _ 1)7T ,b" = 0,
45. u(x, y) = sinh 7T - 2 sinh 27T
1 7. u(O, t) = Ao; u at the center of the disk equals the
33. (a) Ao = 2 7T1:" aZ cos 0 dO = -2, An = average value off around the boundary of the disk.
= 0, n ::: 1, d n = -2 f~ sin
en
7T 0
n."'( dx = Exercises 10.5
1. u(r, 0) = r cos 0
2.f"
7T orr
aZ cos 0 cos nO dO = 11. rR" + R' + A'rR = 0, R(a) = 0, Z" - A Z = 0, Z(O)
0, R(r) = c,Jo (AI') + C2Yo(Ar), C2 = (bounded), °
2
=
-2n(cos n~ - 1) 4
n2 7T ' u(x, y) = (47T - 4n 7T) cos n7T B = 0 n:> I Jo(Aa) = 0, A" = a"la, an = nth zero of Jo. Z(z) =
2
3. cos 0 = 21 + 21 cos 20, u(r, 0) = 2I + 21'
1 2
cos 20 7T(n - 1)'(n + I)' , " ,- c, cosh Az + C4 sinh Az, c, = 0, u(r, z) =
') (a" cosh ny sin nx + d" sinh n7T( 7T - x) sin n7l)l) 2To
~l
1 f~
= -31 7T 2 , An = -'IT1 f~ (b) Ao = 0, An = 0, Bn = 2.frr-rr aZ sin 0 cos nO dO = ni;:, an sin A"z Jo(An r), an = -si-nh-b-A--2"'2-J,'2"-(a-'A""-n)
37. b" =
2 csch n(2
-7T-s-~-n- f X (1
+ (-2 + n2(7T -
- x) sin nx dx =
1)7T) cos n7T)

5. Ao = ?_'IT
42 cos n7T, Bn = 0, n
n
_~
aZ dO
•
2::: 1
_~
aZ cos nO dO =
3
(8n - 8n ) cos n7T n:> I
(n - I)'(n + I)" -
7T
r rJo(An r ) dr
n
~-~~~--n"-7T~-~~-~'~=
2
----
f~'
sm nx dx =
-2(-1 + cos n7T) csch n7T,
7. Bn = --;2 frr0 sin nO dO = - n7T
2
(1 - cos n7T),
7T sinh n1T 0 n71' n 2: 1; An = 0, n 2::: 0
e" = . 2 2
'1T'sinh n7T 0
f'
sin n7l)l dy =
9. Ao 1
= 27T f" 0 dO = "4; A" = -;1 f~ 0
0
7T 0 cos nO dO =
-20( -1 + cos n7T) ,
n~ csch n7T-, n ~ 1 1
-,-(cos n7T - I); Bn = -1 frr 0 sin nO dO =
n-1T 7T 0
u(x, y) = ') (b n sinh ny sin nx + en sinh n7TX sin n7l)l +
~l _1 cos n7T, n 2: 1
dn sinh n7T( 7T ~ x) sin n7l)l) n
39. b" = 7T si! n7T

2
f
x sin n7TX dx =
11. u(r, t) = cos(ca,t)Jo(a,r)
13. u(r, t) = _1_ sin(ca,t)Jo(a,r)

-2(2 + (-2 + n2 7T 2» cos n7T ea,
wrr csch n'1T', ell
2 n7T
7T sinh f'0
. 2 csch n7T, d" =
(1 - y) sm n7l)l dy = n7T
15. u(r, t) = (t COS(ca2t) + l
4e a2 Sin(ea2t»)Jo(a2r)
2
7T sinh n7T
f"0 y sm n7l)l dy = -2 cos n7T
n7T csch wrr,
= __ 2_[J,(an) _ J 2(<xn)]. B = 0 :> 1
21. An [J,(a n)]2 a" a,>' n ,n-
Index
A Bonhoeffer-Van der Pol oscillator, 416 Curie, Marie, 94

Boundary-value problem (BVP), 182, Curie, Pierre, 94
Abel's formula, 146, 159, 229
189,500 Curves, orthogonal, 124
Adjoint, 303
Air resistance, 112
Airplane wing, 494 C 0
Airy's equation, 223, 444
Cauchy-Euler equation, 197 D'Alembert's solution, 571
Algae growth, 128 nonhomogeneous, 200 Damped
Amplitude, damped, 251 amplitude, 251
Center, 355
Analytic, 205 motion, 247
Chain, vibrating, 229
Angular equation, 236 Characteristic Dating works of art, 102
Annihilator, 167 equation, 147, 160, 306 Death, time of, 106
Antibiotic production, 132 complex roots, 149, 162 Decay, radioactive, 94
Archimedes' principle, 246 real and equal roots, 148, 161 Decrement, logarithmic, 256
Art, dating works of, 102 real and unequal roots, 147, 160 Deflection of a beam, 270
Associated linearized system, 362 polynomial, 303 Dependent, linearly, 137, 153,317
Asymptotically stable, 27, 357, 363 Characteristics, method of, 556 Destruction of microorganisms, 34
Azimuthal equation, 235 Chebyshev equation, 221 Determinant, 302
Chemical reactor, 422 Diabetes, testing for, 230
B Clairaut, Alexis, 53 Differential equation
Clairaut equation, 53 first-order, 4
Balance, harmonic, 280 Cofactor matrix, 303 Bernoulli,46
Beam, deflection of, 270 Competing species, 21, 419 exact, 55
Beats, 262 Complex conjugate, 150 homogeneous, 48
Becquere1, Henri, 94 Compound interest, 125 separable, 26
Bendixson's theorem, 414 Concentration of a drug, 39 linear,S, 36
Bernoulli equation, 46 Convolution nonlinear,S, 294
Bessel, Friedrich Wilhelm, 217 integral, 460 nth order, 153
Bessel equation, 217, 445, 535 theorem, 460 order, 4
Bessel-Fourier series, 537 Condition, initial, 3 ordinary, 2
Bessel functions Convergence of series solution, 208 partial, 2
of the first kind, 219 Cooling, Newton's law of, 104 second-order, 135
of the second kind, 219 Corresponding homogeneous equation, system of, 9
orthogonality of, 222 169 Differential
properties of, 222 Coulomb damping, 284 operator, 167
zeros of, 535, 538 Critically damped harmonic motion, total, 56
Bode plots, 289 248 Diffusion, through a membrane. 399
1-1
1-2 Index
Index 1-3
Dirac delta function, 455 nth-order equations, 155 mode, 570

Laplace transfonn of, 455, 456 second-order equations, 139 Laguerre's equation, 222, 223 dependent, 137, 153,317
set of solutions, 142, 158,318
Direction field, 16, 23 Explicit solution, 7 Laplace, Pierre de, 425 independent, 137, 153, 3 I 7, 507
Idealized unit impulse function, 455
Dirichlet problem, 578 Exponential growth and decay, 26, 94 Laplace's equation, 11,499, 578 Logarithmic decrement, 256
Identity matrix, 302
Disease, controlling spread of, 384 Exponential order, 429 G in circular region, 588 Logistic equation, II, 44, 97
Impedance, 273
Double Fourier series, 586 in cylindrical coordinates, 597 with harvesting, 89
Gamma function, 216 Implicit solution, 7
Drug concentration, 39 in spherical coordinates, 596 with predation, 9 I
Improper
Duhamel's principle, 498 General solutions of linear equations Laplace transfonn, 425 Lotka-Volterra equations, 409
F first-order, 36 integral, 426
of convolution, 459
nonhomogeneous, 170 node, stable, 348, 363
Factor, integrating, 36, 63 of Dirac delta function, 455 M
nth-order, 158 Improved Euler's method, 74
Falling-body problems, II I of derivatives, 439, 440
E second-order, 141 Impulse functions, 454
Fennentation, 85 existence of, 430 Malthus, Thomas, 26
Independent
Eigenfunction, 501 Field systems offirst-order equations, of integral equations, 462 Malthus model, 26, 94
319 linearly, 137, 153,317
Eigenlines, 349 direction, 16, 23 inverse, 432 Matrix, 299
Generalized Fourier series, 535 variable, 3
Eigenvalue, 305 slope, 14, 23 linearity, 428, 433 adjoint, 303
Generating function, 542 Indicial roots, 212
multiplicity, 309 Finance, applications to, 125 of periodic functions, 451 cofactor, 303
Gibbs phenomenon, 516 Initial
Eigenvalue problem, 501 First-order ordinary differential equa- properties of, 428, 432, 433, 435, derivative, 311
Gompertz equation, 122 boundary value problem, 557
Eigenvector, 305 tions 447,448,432 detenninant, 302
Green's function, 195 condition, 3
Electrical Bernoulli, 55 table of, 435, 436 eigenvalue, 305, 309
Growth and decay, 26, 94, 97 Initial-value problem (IVP), 3
networks, 267, 391, 393, 397, 470 exact, 55 with partial differential equations, eigenvector, 305
solving with Laplace transforms,
resonance, 273 existence and uniqueness theorem. 602 fundamental, 320
440
Envelope, 261 65, 67 Lasers, 33 identity, 302
H Integral equation, 462
Epidemic, 385 homogeneous, 42, 48 Law of mass action, 122 integral, 311
Integrating factor, 36, 63
Equation, characteristic, 149 linear, 36 Legendre's equation, 208, 444 inverse, 303, 305
Half-life, 94 Integrodifferential equations, 463
Equilibrium point, 348 numerical methods, 71, 74, 78, 80 Leibniz, Gottfried Wilhelm, 2 Jacobian, 362
Hannonic Interest, compound, 125
center, 355, 363 Ricatti, 151 Limit cycle, 411 multiplication, 300
balance, 280 Inverse Laplace transfonn, 432
saddle point, 349, 363 separable, 26 Limiting velocity, 113 transpose, 299
motion Inverse matrix, 303, 305
stable deficient node, 354, 363 systems of, 293 Line, phase, 27 Method
critically damped, 248 Irregular singular point, 210
stable improper node, 348, 363 Fishing, 123 Linear differential equation, 5, 22 of characteristics, 556
overdamped, 248
stable spiral, 355, 363 FitzHugh-Nagumo equation, 387 characteristic equation, 221 of Frobenius, 211
simple, 241
stable star node, 353, 363 Food chains, 420 existence and uniqueness theorem, of undetennined coefficients, 43,
underdarnped, 248
unstable deficient node, 354, 363 Forced motion, 257 67,155 173,340
Harvesting, 88, 103 Jacobian matrix, 362
unstable improper node, 349, 363 Fourier-Legendre series, 541 first-order, 36 Microorganisms, destruction of, 34
Heat equation, 499, 555, 557 Jump discontinuity, 430
unstable spiral, 355, 363 Fourier series, 520 fundamental set of solutions, 142, Mixture problems, 108, 400
derivation, 554
unst1ble star node, 353, 363 complex form, 533 158,318 Monochlorobenzene, production of,
Hennite's equation, 221
Equilibrium solution, 27 cosine series, 516 K general solution, 170 418
Hilbert number, 415
asymptotically stable, 27 double, 586 homogeneous, 42 Multiplicity of an eigenvalue, 309
Homogeneous Kidney dialysis, 129
semi-stable, 102 generalized series, 535 nonhomogeneous, 153
boundary conditions, 557 Kinetic reactions, 35
unstable, 27 nth order, 153
on any interval, 534 insulated boundary, 560 Kirchhoff's laws, 268, 389, 390
N
Error, 75 sine series, 511 particular solution, 168
differential equation, 48, 135, 153
Euler, Leonhard, 2 Fox population, 383 reduction of order, 143 Natural frequency, 259
corresponding, 169
Euler's method, 71, 369 Frequency L second-order, 135 Negative criterion, 414
linear system with constant coeffi-
improved, 74 fundamental, 570 series solutions, 205, 210 Neumann problem, 578
cients
Exact differential equation, 55 natural, 259
L-R circuits, 470 systems of, see Systems Newton, Isaac, 2
complex conjugate eigenvalues,
second-order, 205 L-R-C circuits, 267, 470 undetennined coefficients, 43,171, Newton's law
resonance, 266 325
solving, 59 four loops, 397 340 of cooling, 104
Frobenius, method of, 211 distinct real eigenvalues, 323
test for exactness, 56 one loop, 391 variation of parameters, 183, 191, of motion, III
Fundamental repeated eigenvalues, 330
Existence and uniqueness theorems three loops, 397 342 Node, 348, 353, 354, 363, 571
frequency, 570 Hooke's law, 238
first-order equations, 65, 67 two loops, 393 Linearization, 362 Nonhomogeneous
matrix, 320 Hypergeometric equation, 221
Lagrange equation, 54 Linearly boundary conditions, 560
-. ~- ;",:, ~
1-4 Index Index 1-5
~.-;:"":"'.-"
different.ial equation, 153 portrait, 17, 348 s Stability T Unstable

general solution;"170 Piecewise continuous, 430 asymptotically stable, 357, 363 deficient node, 354
Nonlinear . " - .. ;·.r· Point SIS model, 87 stable, 357, 363 Table of units, 112 equilibrium point, 27, 357
differeiiti~l equatio~j equilibri urn, 348 Saddle point, 349 unstable, 357, 363 Tautochrone, 492 improper node, 349
system, 294 ordinary, 205, 221 SChriidinger equation, 12 Stable Taylor method, 85 spiral,355
classification of equilibrium singular, 205, 210, 221 in spherical coordinates, 235, 549, deficient node, 354 Term-by-term star node, 353
points, 363 Pollution, 34 556 equilibrium point, asymptotically, differentiation, 534
Norm, 508 Population growth, 402 Second-order ordinary differential 27 integration, 534
Normal mode, 570 logistic equation, 11, 44, 97 equation, 135 improper node, 348, 363 Testing for diabetes, 230
self-adjoint form, 506 Testosterone production, 382 v
Numerical methods Malthus model, 26, 94 spiral, 355, 363
Euler's method, 71, 369 rabies, 383 Semi-stable equilibrium solution, 102 Standard form, 36, 185, 191 Threshold, popUlation model with a, Van der Pol equation, 221, 279, 295,
higher-order with Taylor series ex- with harvesting, 88, 89, 103 Separable differential equation, 26 Star node 100 410
pansion,84 with predation, 91 Separation of variables, partial differ- unstable, 353, 363 Time of death, 106 Variables, separation of, 26
improved Euler's method, 74 with a threshold, 100 ential equations, 551 stable, 353, 363 TorriceUi's law, 123 Variation of parameters
partial differential equations, 605 Potential equation, 578 Series solution Steady-state Total differential, 56 for linear differential equations,
Runge-Kutta, 78, 80, 371 Power series solution, 205, 208 about a regular singular point, 208 charge, 270 Trajectories 183, 191
Predator-prey, 408 convergence of, 208 solution, 28, 263 oblique, 124 for systems of equations, 342
Principle of superposition, 138,316, method of Frobenius, 210 temperature, 560, 582 orthogonal, 124 Vector, solution, 315
o 154 Shifting property of Laplace trans- in semi-infinite strip, 585 Transfer function, 498 Velocity, limiting, 113
Product method, 551 form, 432 Step function, unit, 446 Transient Verhulst equation, 97
Oblique trajectories, 124
Production of monochlorobenzene, Shoes, running, 85 Sturm-Liouville problem, 506 solution, 263 Verhulst, P. F., 98
Operator notation, 167, 297
418 Signal processing, 545 Superposition, principle of, 138, 154, temperature, 560 Vibrating chain, 229
Order, 4
Simple harmonic motion, 241 316,551 Transpose of a matrix. 299 Vibration absorber, 493
reduction of, 143 Traveling wave, 575
Simple modes of a vibrating chain, Systems of differential equations, 9 Vibrations, of three-story building, 495
Ordinary differential equation (ODE),
Q 229 Euler's method, 369 Trivial solution, 23 Volleyball, modeling motion of, 322
2,22
Sine integral function, 55 Laplace transform method, 465 Volterra's principle, 416
Ordinary point, 205 Quasiperiod, 251, 253
Singular point, 205 linear, 293
at infinity, 221
at infinity, 221
u
Orthogonal degenerate, 297
irregular, 210
curves, 124 R
regular, 210
fundamental matrix, 320
fundamental set of solutions,
Underdamped harmonic motion, 248
Undetermined coefficients
w
trajectories, 124
Rack-and-gear systems, 285 Slope field, 14, 23 318 for linear differential equations, 43. Wave equation, 499, 567
self-orthogonal, 124
Radial equation, 236 Solution general solution, 319 171 in circular region, 592
Orthogonality condition, 507
Radially symmetric, 592 equilibrium, 27 homogeneous, 293 for systems of differential equa- two-dimensional, 596
Overdamped harmonic motion, 248
Radioactive decay, 94 explicit, 7 nonhomogeneous, 293 tions, 340 Weight, 116
Rayleigh equation, 221, 297 fundamental set of, 142, 158, 318 undetenruned coefficients, 340 Unit Weighting function, 507
p Reactance, 273 general, 2, 23,141,158,170,319 variation of parameters, 342 impulse response, 498 Weirstrass M-test, 533
Reduction of order, 143 implicit, 7 nonlinear, 294 step function, 446 Wronskian, 140, 141, 156,317
Parameters, variation of, 183, 191,342 Regular singular point, 210 particular, 168 Runge-Kutta method, 371 Units, table of, 112 Abel's identity, 146, 159,229
Parseval's equality, 547 at infmity, 221 trivial, 23
Partial differential equation (PDE), 2 Resistance, 112 vector, 315
linear second-order, 550 Resonance, 260 Species, competing, 419
numerical methods, 605 electrical, 273 Spring constant, 238
quasi-linear, 556 frequency, 266 Spring-mass system, 239, 395, 474,
Particular solution, 168 Ricatti, Jacopo Francesco, 151 546
Pendulum equation, 6, 275 Ricatti equation, 151 Springs
Periodic function, 450, 451 Rodrigues' formula, 541 aging, 289
Phase Runge-Kutta method, 78, 80, 371 hard, 288
line, 27 Running shoes, 85 soft, 288

Abell Braselton Modern Differential Equations Second Edition

Cargado por

Información del documento

Derechos de autor

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Abell Braselton Modern Differential Equations Second Edition

Cargado por

Copyright:

ABELL

f sec" u du =n~ I tan u sec"-2 u + a 2 + b2 (a cos bu + b sin bu) + C

f esc u du = lnlcsc u - cot ul +C

f u" In u du = _u_- [en

f sec u tan u du = sec u +C f u cos u du = cos u + u sin u + C

f cos 2 I + 1- sin 2u + C .. sin(a - b)u

sin kt + kt cos kt eX=I+x+-+-+"'=)'-

8k 3s 2 {f(a) da b = In a - In b'' a , b > 0

cos kt cosh kl eX)' = (eXV n=O

We are combining the strengths of our

Our mission is to make learning

We are now Harcourt College Publishers.

HARCOURT COLLEGE PUBLISHERS

Publisher: Emily Barrosse

MODERN DIFFERENTIAL EQUATIONS, Second Edition

4 4.1 Second-Order Equations: An Introduction 135

9 Eigenvalue Problems and Fourier Series 499

10 Partial Differential Equations

Technology Differential Equations is written in an easy-to-read, yet mathematically precise, style.

Martha Abell and James Braselton

so that C = 5. Therefore, the solution to the IVP is y 2

:1 Solution (a) Differentiating, we obtain dy/dt = CI cos t - C2 sin t and d 2y/dt 2

dy ~ (c) To integrate y = J16 ~ x 2 dx, we use a trigonometric substitution. Letting

dy/dx is y instead of an expression involving only the independent variable x. Hence,

equation because it models the motion of a pendulum, is nonlinear because it in-

J u dv = uv - J v du, with u = x, dv = eX dx, du = dx, and v = eX. This gives

"E'"I,I.' Find C ify(l) = 2.

(a) y = J cos x dx = sin x + C. (See Figure 1.4.) = -

(b) To evaluate y = J~+ x2 1

Figure 1.5 Graph of

y(x) = x-I[A cos (2 In x)

salt in the tank? As I ..... 00, what happens to the amount

74. The current I(t) in an L-R circuit, which contains a

independent Schriidinger equation.

72. Use implicit differentiation to show that

dy = e-(v'hi2)' = e-1n2 = e1n2-' =..!..

::.:::.:::~:~-~o ,,~, O),~~oo~ f SffiUIM". (I.

first-order ODEs is equivalent to the second-order ODE d 2x/dt 2 + x = O. Often, how-

dt= -x -10: '5 -10 -5 10 x -10 10 x

By writing the second-order ODE as a system of fIrst-order ODEs, we investi-

-2 -l,5 -1-0.5 0.5 1 l,5 2

Figure 1.17 ii Figure 1.21 Figure 1.24

In Exercises 13-14, use the direction field of the given system d 2x

+ 4x' + 13x = e-'

2 *21. (a) Use a computer algebra system to solve dy/dx =

22. (a) Use a computer algebra system to grapb the slope

dition yeO) = 5. Does Iimx_oo y(x) appear to exist?

Explicit Solution Direction Field

0.4-· CHAPTER 1 REVIEW EXERCISES

= _1_[tan- (2Y- V2) + tan-I(2Y V2

Solve the initial-value problem dy/dx = e 2x + y , yeO) = o.

+ _1_ Inll + V2y + y21 + C = ~ x5 + x.

and the solution to

.;::. Atom in upper Atom in lower

66. (a) Assume that y(l) > 0 and r I

efp(l)dl d: + p(t)eJp(l)dly = q(t)eJp(l)dl.

12.21 First-Order linear Equations :t[eJP(I)dly] = q(t)eJp(l)dl.

so p(t) = - k and q(t) = - h are constant functions.

dy 1 Solve the initial-value problem ~ + 5x y 4

grating factor to be /L(t) = e JMr = e kt . Therefore, we obtain ~ [ektc] = ektD(t) so 20 to=4 20

fo t;dl = 24; the average dosage is

Figure 2.13 (a) k = 0.05 (b) k

= 0.10 (e) k = 0.15 (d) k = 0.20

(e-k(t-r,) - e- kt ) if t > to'

• 1. D. Murray. Mathematical Biology. Springet.Veriag (1990), pp. 645-649.

32_ ~dx + ----l=--

Solve the IVP. Determine lim yet). How does this