22 vistas

Título original: Linear Algebra (SM), Kuttler

Cargado por cantor2000

Linear Algebra (SM), Kuttler

Matemáticas

© All Rights Reserved

- Matrices and Determinants
- Cambridge sche
- Wikipedia: Matrix
- Ansys2Excite_UsersGuide.pdf
- Mechanical Engg 2014
- 19b_FP3_(new)_June_2009
- Study Notes on Matrix Algebra PrepLadder
- Intro to Continuum Mechanics - Raymond.pdf
- 100314 Ott 205 Exam Ssoln05o Ln
- Principal-component-analysis-applied-to-remote-sensing.pdf
- Rotor_Software_Manual_v1
- GATE Mathematics Questions All Branch By S K Mondal.pdf
- CE Engineering Mathematics
- The Determinant Function
- Plus Two Chapter wise important 3,6 and 10 marks
- Lab 5 solns
- invertible matrix theorem
- 1894270-AQA-MFP4-W-QP-JUN11.pdf
- Prac 1 Control
- tut1

Está en la página 1de 183

Exercises

Applications

F.12

Exercises

1.6

1. Let z = 5 + i9. Find z 1 .

1

(5 + i9)

5

106

9

106 i

50

62 + 5i, 5 i, 45 + 28i, and 53

37

53 i.

x4 + 16 = 0, Solution is: (1 i) 2, (1 + i) 2, (1 i) 2, (1 + i) 2.

4. Graph the complex cube roots of 8 in the complex plane. Do the same for the four

fourth roots of 16.

(1 i) 2, (1 + i) 2, (1 i) 2, (1 + i)

2. When you graph these, you will have three equally spaced points on the circle of

radius 2 for the cube roots and you will have four equally spaced points on the circle

of radius 2 for the fourth roots. Here are pictures which should result.

z = |z| .

z

If z = 0, let = 1. If z 6= 0, let =

|z|

n

6. De Moivres theorem says [r (cos t + i sin t)] = rn (cos nt + i sin nt) for n a positive

integer. Does this formula continue to hold for all integers, n, even negative integers?

Explain.

Yes, it holds for all integers. First of all, it clearly holds if n = 0. Suppose now that

n is a negative integer. Then n > 0 and so

n

1

1

n = n

r

(cos

(nt)

+ i sin (nt))

[r (cos t + i sin t)]

Exercises

=

rn

rn (cos (nt) + i sin (nt))

=

(cos (nt) i sin (nt))

(cos (nt) i sin (nt)) (cos (nt) + i sin (nt))

n

r (cos (nt) + i sin (nt))

7. You already know formulas for cos (x + y) and sin (x + y) and these were used to prove

De Moivres theorem. Now using De Moivres theorem, derive a formula for sin (5x)

and one for cos (5x).

sin (5x) = 5 cos4 x sin x 10 cos2 x sin3 x + sin5 x

8. If z and w are two complex numbers and the polar form of z involves the angle while

the polar form of w involves the angle , show that in the polar form for zw the angle

involved is + . Also, show that in the polar form of a complex number, z, r = |z| .

You have z = |z| (cos + i sin ) and w = |w| (cos + i sin ) . Then when you multiply

these, you get

=

=

|z| |w| (cos cos sin sin + i (cos sin + cos sin ))

(x + 2) x i 3 + 1

x 1i 3

10. Write x3 + 27 in the form (x + 3) x2 + ax + b where x2 + ax + b cannot be factored

any more using only real numbers.

x3 + 27 = (x + 3) x2 3x + 9

just the fourth roots of 16. Then to factor, this you get

x (1 i) 2

x (1 + i) 2

x (1 i) 2

x (1 + i) 2

12. Factor x4 + 16 as the product of two quadratic polynomials each of which cannot be

factored further without using complex numbers.

x4 + 16 = x2 2 2x + 4 x2 + 2 2x + 4 . You can use the information in the

preceding problem. Note that (x z) (x z) has real coefficients.

prove zw =Pzw and then show by induction that z1 zm =

m

m

z1 zm . Also verify that k=1 zk = k=1 zk . In words this says the conjugate of a

product equals the product of the conjugates and the conjugate of a sum equals the

sum of the conjugates.

(a + ib) (c + id) = ac bd + i (ad + bc) = (ac bd) i (ad + bc)

Exercises

(a ib) (c id) = ac bd i (ad + bc) which is the same thing. Thus it holds for

a product of two complex numbers. Now suppose you have that it is true for the

product of n complex numbers. Then

z1 zn+1 = z1 zn zn+1

and now, by induction this equals

z1 zn zn+1

As to sums, this is even easier.

n

X

(xj + iyj ) =

j=1

n

X

j=1

xj i

n

X

xj + i

j=1

n

X

yj =

j=1

n

X

j=1

xj iyj =

n

X

yj

j=1

n

X

(xj + iyj ).

j=1

14. Suppose p (x) = an xn + an1 xn1 + + a1 x + a0 where all the ak are real numbers.

Suppose also that p (z) = 0 for some z C. Show it follows that p (z) = 0 also.

You just use the above problem. If p (z) = 0, then you have

p (z) = 0 = an z n + an1 z n1 + + a1 z + a0

= an z n + an1 z n1 + + a1 z + a0

= an z n + an1 z n1 + + a1 z + a0

= an z n + an1 z n1 + + a1 z + a0

= p (z)

1 = i2 =

2

1 1 = (1) = 1 = 1.

This is clearly a remarkable result but is there something wrong with it? If so, what

is wrong?

16. De Moivres theorem is really a grand thing. I plan to use it now for rational exponents,

not just integers.

1 = 1(1/4) = (cos 2 + i sin 2)

1/4

does this tell you about De Moivres theorem? Is there a profound difference between

raising numbers to integer powers and raising numbers to non integer powers?

It doesnt work. This is because there are four fourth roots of 1.

17. Show that C cannot be considered an ordered field. Hint: Consider i2 = 1.

It is clear that 1 > 0 because 1 = 12 . (In general a2 > 0. This is clear if a > 0. If a < 0,

then adding a to both sides, yields that 0 < a. Also recall that a = (1) a and

2

2

2

that (1) = 1. Therefore, (a) = (1) a2 = a2 > 0. ) Now it follows that if C can

be ordered, then 1 > 0, but this is a problem because it implies that 0 > 1 = 12 > 0.

Exercises

18. Say a + ib < x + iy if a < x or if a = x, then b < y. This is called the lexicographic

order. Show that any two different complex numbers can be compared with this order.

What goes wrong in terms of the other requirements for an ordered field.

From the definition of this funny order, 0 < i and so if this were an order, you would

need to have 0 < i2 = 1. Now add 1 to both sides and obtain 0 > 1 = 12 > 0, a

contradiction.

19. With the order of Problem 18, consider for n N the complex number 1 n1 . Show

that with the lexicographic order just described, each of 1 im is an upper bound to

all these numbers. Therefore, this is a set which is bounded above but has no least

upper bound with respect to the lexicographic order on C.

This follows from the definition. 1 im > 1 1/n for each m. Therefore, if you

consider the numbers 1 n1 you have a nonempty set which has an upper bound but

no least upper bound.

F.13

Exercises

1.11

1. Give the complete solution to the system of equations, 3x y + 4z = 6, y + 8z = 0,

and 2x + y = 4.

x = 2 4t, y = 8t, z = t.

and 4x + z = 9.

x = y = 2, z = 1

zero and so 5x + 2y z = 5x 2y z which is equivalent to y = 0. Thus x and

z can equal anything. But when x = 1, z = 4, and y = 0 are plugged in to the

equations, it doesnt work. Why?

These are invalid row operations.

4. Give the complete solution to the system of equations, x+2y +6z = 5, 3x+2y +6z = 7

,4x + 5y + 15z = 7.

No solution.

x + 2y + 3z

4x + 5y + z

=

=

5, 3x + 2y + z = 7,

7, x + 3z = 5.

x = 2, y = 0, z = 1.

6. Give the complete solution of the system of equations,

x + 2y + 3z

4x + 5y + 5z

= 5, 3x + 2y + 2z = 7

= 7, x = 5

No solution.

Exercises

7. Give the complete solution of the system of equations

x + y + 3z

4x + 9y + z

2, 3x y + 5z = 6

8, x + 5y + 7z = 2

x = 2 2t, y = t, z = t.

8. Determine a such that there are infinitely many solutions and then find them. Next

determine a such that there are no solutions. Finally determine which values of a

correspond to a unique solution. The system of equations is

3za2 3a + x + y + 1 = 0

3x a y + z a2 + 4 5 = 0

za2 a 4x + 9y + 9 = 0

If a = 1, there are infinitely many solutions of the form x = 2 2t, y = t, z = t. If

a = 1, t then there are no solutions. If a is anything else, there is a unique solution.

2a

1

1

x=

,y =

,z =

a+1

a+1

a+1

y + z = 2, z + w = 0, y 4z 5w = 2, 2y + z w = 4

x = t, y = s + 2, z = s, w = s

10. Find all solutions to the following equations.

x+y+z

2, z + w = 0,

2x + 2y + z w

4, x + y 4z 5z = 2

F.14

Exercises

1.14

1. Verify all the properties 1.11-1.18.

2. Compute 5 (1, 2 + 3i, 3, 2) + 6 (2 i, 1, 2, 7) .

3. Draw a picture of the points in R2 which are determined by the following ordered

pairs.

(a)

(b)

(c)

(d)

(1, 2)

(2, 2)

(2, 3)

(2, 5)

5. Draw a picture of the points in R3 which are determined by the following ordered

triples. If you have trouble drawing this, describe it in words.

(a) (1, 2, 0)

(b) (2, 2, 1)

(c) (2, 3, 2)

Exercises

F.15

Exercises

1.17

1. Show that (a b) =

1

4

h

i

2

2

|a + b| |a b| .

2. Prove from the axioms of the inner product the parallelogram identity, |a + b| +

|a b|2 = 2 |a|2 + 2 |b|2 .

P

3. For a, b Rn , define a b nk=1 k ak bk where k > 0 for each k. Show this satisfies

the axioms of the inner product. What does the Cauchy Schwarz inequality say in

this case.

The product satisfies all axioms for the inner product so the Cauchy Schwarz inequality

holds.

4. In Problem 3 above, suppose you only know k 0. Does the Cauchy Schwarz inequality still hold? If so, prove it.

Yes, it does. You dont need the part which says that the only way a a = 0 is for

a = 0 in the argument for the Cauchy Schwarz inequality.

5. Let f, g be continuous functions and define

Z

f g

f (t) g (t)dt

show this satisfies the axioms of a inner product if you think of continuous functions

in the place of a vector in Fn . What does the Cauchy Schwarz inequality say in this

case?

The only part which is not obvious for the axioms is the one which says that if

Z

|f | = 0

6. Show that if f is a real valued continuous function,

Z

f (t) dt

Z

b

f (t) dt

a

!2

(b a)

1 dt

a

1/2

= (b a)

!1/2

Z

f (t) dt.

|f (t)| dt

2

|f (t)| dt

!1/2

!1/2

which yields the desired inequality when you square both sides.

Exercises

F.16

Exercises

2.2

1. In 2.1 - 2.8 describe A and 0.

2. Let A be an nn matrix. Show A equals the sum of a symmetric and a skew symmetric

matrix.

A=

A+AT

2

AAT

2

3. Show every skew symmetric matrix has all zeros down the main diagonal. The main

diagonal consists of every entry of the matrix which is of the form aii . It runs from

the upper left down to the lower right.

You know that Aij = Aji . Let j = i to conclude that Aii = Aii and so Aii = 0.

4. Using only the properties 2.1 - 2.8 show A is unique.

Suppose that B also works. Then

A = A + (A + B) = (A + A) + B = B

5. Using only the properties 2.1 - 2.8 show 0 is unique.

If 0 is another additive identity, then 0 = 0 + 0 = 0

6. Using only the properties 2.1 - 2.8 show 0A = 0. Here the 0 on the left is the scalar 0

and the 0 on the right is the zero for m n matrices.

0A = (0 + 0) A = 0A + 0A. Now add the additive inverse of 0A to both sides.

7. Using only the properties 2.1 - 2.8 and previous problems show (1) A = A.

A. Thus A = (1) A.

8. Prove 2.17.

T

(AB)ij (AB)ji =

Ajk Bki =

T T

Bik

Akj = B T AT

P

(Im A)ij k Iik Akj = Aij and so Im A = A.

ij

n

10. Let A and

y Rm . Show (Ax, y)Rm =

be a real m n matrix and let x R and

T

k

x,A y Rn where (, )Rk denotes the dot product in R .

P

P P

(Ax, y) = i (Ax)i yi = i k Aik xk yi

P

P

P P

x,AT y = k xk i AT ki yi = k i xk Aik yi , the same as above. Hence the two

are equal.

T

11. Use the result of Problem 10 to verify directly that (AB) = B T AT without making

any reference to subscripts.

(AB)T x, y (x, (AB) y) = AT x,By = B T AT x, y . Since this holds for every

x, y, you have for all y

T

(AB) x B T AT x, y

T

Exercises

12. Let x = (1, 1, 1) and y = (0, 1, 2) . Find xT y and xyT if possible.

1

0 1 2

xT y = 1 0 1 2 = 0 1 2

1

0 1

2

0

xyT = 1 1 1 1 = 1

2

13. Give

1

1

1

1

an example

1

1

1

1

1

1

1

1

1

14. Let A = 2

1

if possible.

1

0 0

=

1

0 0

1

0 0

=

1

0 0

1

1

1 3

1

1

2

1 , B =

0 . Find

, and C = 1 2

2 1 2

2

3 1 0

(a) AB

(b) BA

(c) AC

(d) CA

(e) CB

(f) BC

15. Consider the following digraph.

Write the matrix associated with this digraph and find the number of ways to go from

3 to 4 in three steps.

0 1 1 0

1 0 0 1

1 1 0 2

0 1 0 1

Exercises

0

1

1

0

1

0

1

1

1

0

0

0

3

0

1

1

= 3

4

2

1

1

5

2

5

3

2

0

1

1

4

5

8

3

16. Show that if A1 exists for an n n matrix, then it is unique. That is, if BA = I and

AB = I, then B = A1 .

From the given equations, multiply on the right by A1 . Then B = A1 .

17. Show (AB)

= B 1 A1 .

ABB 1 A1 = AIA1 = I

B 1 A1 AB = B 1 IB = I

Then by the definition of the inverse and its uniqueness, it follows that (AB)

and

(AB)1 = B 1 A1

exists

1

T

18. Show that if A is an invertible n n matrix, then so is AT and AT

= A1 .

T

T

AT A1 = A1 A = I

T

T

A1 AT = AA1 = I Then from the definition of the inverse and its uniqueness,

T

1

it follows that AT

exists and equals A1 .

19. Show that if A is an n n invertible matrix and x is a n 1 matrix such that Ax = b

for b an n 1 matrix, then x = A1 b.

Multiply both sides on the left by A1 .

2

0 1

1 0

=

1 0

0 1

21. Give an example of matrices, A, B such that neither A nor B equals zero and yet

AB = 0.

1 1

1 1

0 0

=

1 1

1 1

0 0

x1 x2 + 2x3

x1

2x3 + x1

in the form A x2 where A is an appropriate matrix.

22. Write

3x3

x3

3x4 + 3x2 + x1

x4

1 1 2 0

x1

x1 x2 + 2x3

1 0 2 0 x2

x1 + 2x3

0 0 3 0 x3 =

3x3

1 3 0 3

x4

x1 + 3x2 + 3x4

23. Give another example other than the one given in this section of two square matrices,

A and B such that AB 6= BA.

Almost anything works.

1 2

1 2

5 2

=

3 4

2 0

11 6

10

Exercises

1

2

2

0

1

3

2

4

7

2

10

4

24. Suppose A and B are square matrices of the same size. Which of the following are

correct?

2

(a) (A B) = A2 2AB + B 2

Note this.

2

(e) A2 B 2 = A (AB) B This is all right.

3

(g) (A + B) (A B) = A2 B 2 Not this.

(i) All of the above. They are all right.

1 1

25. Let A =

. Find all 2 2 matrices, B such that AB = 0.

3

3

1 1

x y

x z w y

=

3

3

z w

3x + 3z 3w + 3y

z w

, z, w arbitrary.

z

w

26. Prove that if A1 exists and Ax = 0 then x = 0.

Multiply on the left by A1 .

27. Let

Find

1

2

1

28. Let

Find

1

2

1

29. Let

1 2

A= 2 1

1 0

3

4 .

2

3

4 .

2

1

2 3

2 4 5

1 4 = 0

1 2

0 2

1 2 3

1 0

A= 2 3

1 0

1

0 3

2 0

3

3 4 = 0 13 32

0 2

1 0 1

1 2

A= 2 1

4 5

3

4 .

10

11

Exercises

Find

1

2

4

30. Let

Find

1

1

2

1

F.17

2 3

1 4 , row echelon form:

5 10

1 0 53

0 1 23 A has no inverse.

0 0 0

1

1

A=

2

1

2 0

1 2

1 3

2 1

2

0

2

2

1

1

1

1 21

2 0 2

2

2

1

1 2 0

12 25

2

= 3

1 0

1 3 2

0

1

1

9

2 1 2

2 43

4

4

Exercises

2.7

1. Show the map T : Rn Rm defined by T (x) = Ax where A is an m n matrix and

x is an m 1 column vector is a linear transformation.

This follows from matrix multiplication rules.

2. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /3.

1

cos (/3) sin (/3)

21 3

2

=

1

1

sin (/3) cos (/3)

2 3

2

3. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /4.

1

cos (/4) sin (/4)

2 21 2

2

=

1

1

sin (/4) cos (/4)

2 2

2 2

4. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /3.

1

1

cos (/3) sin (/3)

2

2 3

=

1

sin (/3) cos (/3)

21 3

2

5. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of 2/3.

2 cos (/3) 2 sin (/3)

1 3

=

2 sin (/3) 2 cos (/3)

3

1

6. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /12. Hint: Note that /12 = /3 /4.

12

Exercises

cos (/3) sin (/3)

cos (/4) sin (/4)

sin (/3) cos (/3)

sin (/4) cos (/4)

1

1

1

1

4 23

4 23 + 4 2

4 2

=

1

1

1

1

4 2 3 4 2

4 2 3+ 4 2

7. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of 2/3 and then reflects across the x axis.

1

1 0

cos (2/3) sin (2/3)

12 3

2

=

1

0 1

sin (2/3) cos (2/3)

21 3

2

8. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /3 and then reflects across the x axis.

1

12 3

1 0

cos (/3) sin (/3)

2

=

0 1

sin (/3) cos (/3)

21 3

12

9. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /4 and then reflects across the x axis.

1

1

1 0

cos (/4) sin (/4)

2

2

2

2

=

1

1

0 1

sin (/4) cos (/4)

2 2 2 2

10. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /6 and then reflects across the x axis followed by a reflection across the

y axis.

1

1

1 0

1 0

cos (/6) sin (/6)

2 3

2

=

0 1

0 1

sin (/6) cos (/6)

21

21 3

11. Find the matrix for the linear transformation which reflects every vector in R2 across

the x axis and then rotates every vector through an angle of /4.

1

1

cos (/4) sin (/4)

1 0

2

2

2

2

=

1

1

sin (/4) cos (/4)

0 1

2

2

2 2

12. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of /4 and next reflects every vector across the x axis. Compare with the

above problem.

1

1

1 0

cos (/4) sin (/4)

2

2

2

2

=

1

1

0 1

sin (/4) cos (/4)

2 2 2 2

13. Find the matrix for the linear transformation which reflects every vector in R2 across

the x axis and then rotates every vector through an angle of /6.

1

1

cos (/6) sin (/6)

1 0

2 3

2

=

1

sin (/6) cos (/6)

0 1

21 3

2

14. Find the matrix for the linear transformation which reflects every vector in R2 across

the y axis and then rotates every vector through an angle of /6.

1

cos (/6) sin (/6)

1 0

2 3 12

=

1

sin (/6) cos (/6)

0 1

12

2 3

13

Exercises

15. Find the matrix for the linear transformation which rotates every vector in R2 through

an angle of 5/12. Hint: Note that 5/12 = 2/3 /4.

=

cos (2/3) sin (2/3)

cos (/4) sin (/4)

sin (2/3) cos (2/3)

sin (/4) cos (/4)

1

1

1

1

3 4 2

4 23 4 2 4 2

1

1

1

1

4 2 3+ 4 2

4 2 3 4 2

16. Find the matrix for proju (v) where u = (1, 2, 3)T .

T ei =

ei (1,2,3)

14

(1, 2, 3)

1

2

3

1

2 4 6

14

3

6

9

17. Find the matrix for proju (v) where u = (1, 5, 3)T .

1

1

5

35

3

5 3

25 15

15 9

T

1 0

1

0 0

10

3 0

3

0

9

19. Give an example of a 2 2 matrix A which has all its entries nonzero and satisfies

A2 = A. Such a matrix is called idempotent.

You know it cant be invertible. So try this.

2 2

a a

a + ba a2 + ba

=

b b

b2 + ab b2 + ab

Let a2 + ab = a, b2 + ab = b. A solution which yields a nonzero matrix is

2

2

1 1

20. Let A be an m n matrix and let B be an n m matrix where n < m. Show that

AB cannot have an inverse.

This follows right away from Theorem 2.3.8. This theorem says there exists a vector

x 6= 0 such that Bx = 0. Therefore, ABx = 0 also and so AB cannot be invertible.

21. Find ker (A) for

1

0

A=

1

0

2

2

4

2

3

1

4

1

2

1

3

1

1

2

.

3

2

14

Exercises

Recall ker (A) is

1 2 3 2

0 2 1 1

1 4 4 3

0 2 1 1

1 0

1

0

2 0

. After row operations,

0

3 0

2 0

0

0

1

0

0

1

2

1

2

0

0

0

0

1

1

0

0

0

0

0

0

22. If A is a linear transformation, and Axp = b. Show that the general solution to the

equation Ax = b is of the form xp + y where y ker (A). By this I mean to show

that whenever Az = b there exists y ker (A) such that xp + y = z.

If Az = b, Then A (z xp ) = Az Axp = b b = 0 so there exists y such that

y ker (A) and xp + y = z.

23. Using Problem 21, find the general solution to the following linear system.

1

0

1

0

2t1 t2 + t3

21 t1 12 t2 t3

t1

t2

t3

2

2

4

2

4

7/2

0

0

0

3

1

4

1

2

1

3

1

x1

1

x2

2

x3 =

3

x4

2

x5

11

7

18

7

, ti F

24. Using Problem 21, find the general solution to the following linear system.

1

0

1

0

2t1 t2 + t3

21 t1 12 t2 t3

t1

t2

t3

2

2

4

2

1

7/2

0

0

0

3

1

4

1

2

1

3

1

x1

1

x2

2

x3 =

3

x4

2

x5

6

7

13

7

, ti F

25. Show that the function Tu defined by Tu (v) v proju (v) is also a linear transformation.

This is the sum of two linear transformations so it is obviously linear.

26. If u = (1, 2, 3)T and Tu

satisfies Au x = Tu (x).

1 0 0

1

0 1 0 1 2

14

0 0 1

3

13

2 3

14

4 6 = 17

3

6 9

14

17

5

7

37

3

14

37

5

14

15

Exercises

Show that there exists a basis for Im (T ) T (V )

{T v1 , , T vm }

and that in this situation,

{v1 , , vm }

is linearly independent.

Say

m

X

ci vi = 0

i=1

m

X

ci T vi = T 0 = 0

i=1

28. In the situation of Problem 27 where V is a subspace of Fn , show that there exists

{z1 , , zr } a basis for ker (T ) . (Recall Theorem 2.4.12. Since ker (T ) is a subspace,

it has a basis.) Now for an arbitrary T v T (V ) , explain why

T v = a1 T v1 + + am T vm

and why this implies

v (a1 v1 + + am vm ) ker (T ) .

Then explain why V = span (v1 , , vm , z1 , , zr ) .

ker (T ) is also a subspace so it has a basis {z1 , , zr } for some r n. Now let the

basis for T (V ) be as above. Then for v an arbitrary vector, there exist unique scalars

ai such that

T v = a1 T v1 + + am T vm

Then it follows that

T (v (a1 v1 + + am vm )) = T v (a1 T v1 + + am T vm ) = 0

Hence the vector v(a1 v1 + + am vm ) is in ker (T ) and so there exist unique scalars

bi such that

r

X

v (a1 v1 + + am vm ) =

bi zi

i=1

29. In the situation of the above problem, show {v1 , , vm , z1 , , zr } is a basis for V

and therefore, dim (V ) = dim (ker (T )) + dim (T (V ))

The claim that {v1 , , vm , z1 , , zr } is a basis will be complete if it is shown that

these vectors are linearly independent. Suppose then that

X

X

ai zi +

bj vj = 0.

i

Then do T to both sides. Then you have j bj T vj = 0 and so each bj = 0. Then you

use the linear independence to conclude that each ai = 0.

16

Exercises

from W to U where V, W, U are all subspaces of some Fp . Explain why

A (ker (BA)) ker (B) , ker (A) ker (BA) .

ker(BA)

ker(B)

ker(A)

A(ker(BA))

If x ker (BA) , then BAx = 0 and so Ax ker (B) . That is, BAx = 0. It follows

that

A (ker (BA)) ker (B)

The second inclusion is obvious because if x is sent to 0 by A, then B will send Ax

to 0.

31. Let {x1 , , xn } be a basis of ker (A) and let {Ay1 , , Aym } be a basis of A (ker (BA)).

Let z ker (BA) . Explain why

Az span {Ay1 , , Aym }

and why there exist scalars ai such that

A (z (a1 y1 + + am ym )) = 0

and why it follows z (a1 y1 + + am ym ) span {x1 , , xn }. Now explain why

ker (BA) span {x1 , , xn , y1 , , ym }

and so

dim (ker (BA)) dim (ker (B)) + dim (ker (A)) .

This important inequality is due to Sylvester. Show that equality holds if and only if

A(ker BA) = ker(B).

Let {Ax1 , , Axr } be a basis for A (ker (BA)) . Then let y ker (BA) . Thus It

follows that

Ay A (ker (BA)) .

Then there are scalars ai such that

Ay =

r

X

ai Axi

i=1

Hence y

that

Pr

i=1

ai xi ker (A) . Let {z1 , , zm } be a basis for ker (A) . This shows

dim ker (A) + dim ker (B)

To see this, say

X

X

aj zj +

bi xi = 0

j

17

Exercises

Then do A to both sides and obtain each bi = 0. Then it follows that each aj is also

zero because of the independence of the zj . Thus

dim ker (BA) = dim ker (A) + dim (A ker (BA))

32. Generalize the result of the previous problem to any finite product of linear mappings.

Qs

Ps

dim (ker i=1 Ai ) i=1 dim ker (Ai ). This follows by induction.

33. If W V for W, V two subspaces of Fn and if dim (W ) = dim (V ) , show W = V .

Let a basis for W be {w1 , , wr } Then if there exists v V \ W, you could add in v

to the basis and obtain a linearly independent set of vectors of V which implies that

the dimension of V is at least r + 1 contrary to assumption.

V = V1 Vm

(6.27)

v = v1 + + vm

where each vi Vi . This is called a direct sum. If this uniqueness condition does not

hold, then one writes

V = V1 + + Vm

and this symbol means all vectors of the form

v1 + + vm , vj Vj for each j.

Show this is equivalent to saying that if

0 = v1 + + vm , vj Vj for each j,

then each vj = 0. Next show that in the this situation, if i = ui1 , , uimi is a basis

for Vi , then { 1 , , m } is a basis for V .

span ( 1 , , m ) is given to equal V. It only remains to verify that { 1 , , m } is

linearly independent. Suppose vi Vi with

X

vi = 0

i

XX

i

cij uij = 0

X

cij uij = 0

j

and now, since these uij form a basis for Vi , it follows that each cij = 0 for each j for

each i. Thus { 1 , , m } is a basis.

18

Exercises

35. Suppose you have finitely many linear mappings L1 , L2 , , Lm which map V to V

where V is a subspace of Fn and suppose they commute. That is, Li Lj = Lj Li for all

i, j. Also suppose Lk is one to one on ker (Lj ) whenever j 6= k. Letting P denote the

product of these linear transformations, P = L1 L2 Lm , first show

ker (L1 ) + + ker (Lm ) ker (P )

Next show Lj : ker (Li ) ker (Li ) . Then show

ker (L1 ) + + ker (Lm ) = ker (L1 ) ker (Lm ) .

Using Sylvesters theorem, and the result of Problem 33, show

ker (P ) = ker (L1 ) ker (Lm )

Hint: By Sylvesters theorem and the above problem,

X

dim (ker (P ))

dim (ker (Li ))

i

First note that, since these operators

commute, it follows that Lk : ker Li ker Li .

P

Let vi ker (Li ) P

and consider i vi . It is obvious, since the linear transformations

commute that P ( i vi ) = 0. Thus

ker (L1 ) + + ker (Lm ) ker (P )

dim ker (P )

If vi ker (Li ) and

vi = 0, then apply

Y

i6=k

Li vk = 0

i6=k

Since each of these Li is one to one on ker (Lk ) , it follows that vk = 0. Thus

ker (L1 ) + + ker (Lm ) = ker (L1 ) ker (Lm )

P

Now it follows that a basis for ker (L1 ) + + ker (Lm ) has i dim ker (Li ) vectors in

it.

X

dim ker (P )

dim ker (Li )

i

dim ker (P )

ker (L1 ) ker (Lm ) = ker (P )

19

Exercises

36. Let M (Fn , Fn ) denote the set of all n n matrices having entries in F. With the usual

operations of matrix addition and scalar multiplications, explain why M (Fn , Fn ) can

2

be considered as Fn . Give a basis for M (Fn , Fn ) . If A M (Fn , Fn ) , explain why

there exists a monic polynomial of the form

k + ak k + + a1 + a0

such that

Ak + ak Ak + + a1 A + a0 I = 0

The minimal polynomial of A is the polynomial like the above, for which p (A) = 0

which has smallest degree. I will discuss the uniqueness of this polynomial later. Hint:

2

Consider the matrices I, A, A2 , , An . There are n2 + 1 of these matrices. Can they

be linearly independent? Now consider all polynomials and pick one of smallest degree

and then divide by the leading coefficient.

A basis for M (Fn , Fn ) is obviously the matricies Eij where Eij has a 1 in the ij th

place and zeros everywhere else. Thus the dimension of this vector space is n2 . It

2

follows that the list of matrices I, A, A2 , , An is linearly dependent. Hence there

exists a polynomial p () which has smallest possible degree such that p (A) = 0 by the

well ordering principle of the natural numbers. Then divide by the leading coefficient.

If you insist that its leading coefficient be 1, (monic) then the polynomial is unique

and it is called the minimal polynomial. It is unique thanks to the division algorithm,

because if q () is another one, then

q () = p () l () + r ()

where the degree of r () is less than the degree of p () or else equals 0. If it is not

zero, then r (A) = 0 and this would be a contradiction. Hence q () = p () l () where

l () must be monic. Since q () has smallest possible degree, this monic polynomial

can only be 1. Thus q () = p ().

37. Suppose the field of scalars is C and A is an n n matrix. From the preceding

problem, and the fundamental theorem of algebra, this minimal polynomial factors

( 1 )r1 ( 2 )r2 ( k )rk

where rj is the algebraic multiplicity of j . Thus

r

rk

(A 1 I) 1 (A 2 I) 2 (A k I)

=0

apply the result of Problem 35 to verify that

Cn = ker (L1 ) ker (Lk )

and that A : ker (Lj ) ker (Lj ). In this context, ker (Lj ) is called the generalized

eigenspace for j . You need to verify the conditions of the result of this problem hold.

r

Let Li = (A i I) i . Then obviously these commute since they are just polynomials

in A. Is Lk one to one on ker (Li )?

rk

(A k I)

rk

= (A i I + (i k ) I)

20

Exercises

rk

X

rk

j

r j

(A i I) (i k ) k

j

j=0

rk

X

rk

rk

= (i k ) I +

(A i I)j (i k )rk j

j

j=1

(A k I)

rk ri

rk ri

= (i k )

ri

I + g (A) (A i I)

where g (A) is some polynomial in A. Let vi ker (Li ) . Then suppose (A k I)rk vi =

0. Then

r r

r r

0 = (A k I) k i vi = (i k ) k i vi

and since i 6= k , this requires that

Q vi = 0. Thus Lk is one to one on ker (Li ) as

hoped. Therefore, since Cn = ker i Li , it follows from the above problems that

Cn = ker (L1 ) ker (Lk )

Note that there was nothing sacred about C all you needed for the above to hold is

that the minimal polynomial factors completely into a product of linear factors. In

other words, all the above works fine for Fn provided the minimal polynomial splits.

38. In the context of Problem 37, show there exists a nonzero vector x such that

(A j I) x = 0.

This is called an eigenvector and the j is called an eigenvalue. Hint: There must

exist a vector y such that

rj 1

(A 1 I) 1 (A 2 I) 2 (A j I)

(A k I)

rk

y = z 6= 0

r 1

r

r

r

(A j I) z = (A j I) (A 1 I) 1 (A 2 I) 2 (A j I) j (A k I) k y

= (A 1 I)r1 (A 2 I)r2 (A j I)rj (A k I)rk y = 0

39. Suppose Q (t) is an orthogonal matrix. This means Q (t) is a real n n matrix which

satisfies

T

Q (t) Q (t) = I

Suppose also the entries of Q (t) are differentiable. Show QT = QT Q QT .

This is just the product rule.

Hence

21

Exercises

40. Remember the Coriolis force was 2 vB where was a particular vector which

came from the matrix Q (t) as described above. Show that

Q (t) = i (t) j (t0 ) j (t) j (t0 ) k (t) j (t0 ) .

i (t) k (t0 ) j (t) k (t0 ) k (t) k (t0 )

There will be no Coriolis force exactly when = 0 which corresponds to Q (t) = 0.

When will Q (t) = 0?

Q (t) u = u1 e1 (t) + u2 e2 (t) + u3 e3 (t)

where

u u1 e1 (t0 ) + u2 e2 (t0 ) + u3 e3 (t0 )

Q (t) u =

X

j

u ej (t0 ) ej (t)

T

= er (t0 )

X

j

= er (t0 ) es (t)

which shows the desired result.

41. An illustration used in many beginning physics books is that of firing a rifle horizontally and dropping an identical bullet from the same height above the perfectly

flat ground followed by an assertion that the two bullets will hit the ground at exactly the same time. Is this true on the rotating earth assuming the experiment

takes place over a large perfectly flat field so the curvature of the earth is not an

issue? Explain. What other irregularities will occur? Recall the Coriolis acceleration

is 2 [(y cos ) i+ (x cos + z sin ) j (y sin ) k] where k points away from the

center of the earth, j points East, and i points South.

Obviously not. Because of the Coriolis force experienced by the fired bullet which is

not experienced by the dropped bullet, it will not be as simple as in the physics books.

For example, if the bullet is fired East, then y sin > 0 and will contribute to a force

acting on the bullet which has been fired which will cause it to hit the ground faster

than the one dropped. Of course at the North pole or the South pole, things should

be closer to what is expected in the physics books because there sin = 0. Also, if

you fire it North or South, there seems to be no extra force because y = 0.

F.18

Exercises

3.2

1. Find the determinants of the following matrices.

1 2 3

(a) 3 2 2 (The answer is 31.)

0 9 8

22

Exercises

4

(b) 1

3

1

1

(c)

4

1

3 2

7 8 (The answer is 375.)

9 3

2 3 2

3 2 3

, (The answer is 2.)

1 5 0

2 1 2

2. If A1 exist, what is the relationship between det (A) and det A1 . Explain your

answer.

1 = det AA1 = det (A) det A1 .

3. Let A be an n n matrix where n is odd. Suppose also that A is skew symmetric.

This means AT = A. Show that det(A) = 0.

det (A) = det AT = det (A) = det (I) det (A) = (1)n det (A) = det (A) .

4. Is it true that det (A + B) = det (A) + det (B)? If this is so, explain why it is so and

if it is not so, give a counter example.

Almost anything shows that this is not true.

1 0

1 0

det

+

=

0 1

0 1

1 0

1 0

det

+ det

=

0 1

0 1

0

2

5. Let A be an r r matrix and suppose there are r 1 rows (columns) such that all rows

(columns) are linear combinations of these r 1 rows (columns). Show det (A) = 0.

Without loss of generality, assume the last row is a linear combination of the first r 1

rows. Then the matrix is of the form

rT1

..

rTn1

Pn1

T

a

r

i=1 i i

Then from the linear property of determinants, the determinant equals

T

T

r1

r1

n1

.. n1

..

X

X

ai det . =

ai det . = 0

T

T

rn1

rn1

i=1

i=1

T

ri

0T

Where the first equal sign in the above is obtained by taking 1 times a the ith row

from the top and adding to the last row.

6. Show det (aA) = an det (A) where here A is an n n matrix and a is a scalar.

Each time you take out an a from a row, you multiply by a the determinant of the

matrix which remains. Since there are n rows, you do this n times, hence you get an .

23

Exercises

7. Suppose A is an upper triangular matrix. Show that A1 exists if and only if all

elements of the main diagonal are non zero. Is it true that A1 will also be upper

triangular? Explain. Is everything the same for lower triangular matrices?

This is obvious because the determinant of A is the product of these diagonal entries.

When you consider the usual process of finding the inverse, you get that A1 must be

upper triangular. Everything is similar for lower triangular matrices.

8. Let A and B be two n n matrices. A B (A is similar to B) means there exists an

invertible matrix S such that A = S 1 BS. Show that if A B, then B A. Show

also that A A and that if A B and B C, then A C.

This is easy except possibly for the last claim. Say A = P 1 BP and B = Q1 CQ.

Then

A = P 1 BP = A = P 1 Q1 CQP = (QP )1 C (QP )

9. In the context of Problem 8 show that if A B, then det (A) = det (B) .

det A =

=

det P 1 BP = det P 1 det (B) det (P )

det (B) det P 1 P = det (B) .

10. Let A be an n n matrix and let x be a nonzero vector such that Ax = x for some

scalar, . When this occurs, the vector, x is called an eigenvector and the scalar,

is called an eigenvalue. It turns out that not every number is an eigenvalue. Only

certain ones are. Why? Hint: Show that if Ax = x, then (I A) x = 0. Explain

why this shows that (I A) is not one to one and not onto. Now use Theorem 3.1.15

to argue det (I A) = 0. What sort of equation is this? How many solutions does it

have?

1

you could multiply on the left by it and then conclude that x = 0. Therefore, (I A)

is not one to one and not onto.

11. Suppose det (I A) = 0. Show using Theorem 3.1.15 there exists x 6= 0 such that

(I A) x = 0.

If that determinant equals 0 then the matrix I A has no inverse. It is not one

to one and so there exists x 6= 0 such that (I A) x = 0. Also recall the process for

finding the inverse.

a (t) b (t)

12. Let F (t) = det

. Verify

c (t) d (t)

F (t) = det

Now suppose

a (t) b (t)

c (t) d (t)

+ det

a (t) b (t)

c (t) d (t)

F (t) = det d (t) e (t) f (t) .

g (t) h (t) i (t)

24

Exercises

Use Laplace expansion and the first part to verify F (t) =

a (t) b (t) c (t)

det d (t) e (t) f (t) + det d (t) e (t) f (t)

g (t) h (t) i (t)

g (t) h (t) i (t)

+ det d (t) e (t) f (t) .

g (t) h (t) i (t)

Conjecture a general result valid for n n matrices and explain why it will be true.

Can a similar thing be done with the columns?

The way to see this holds in general is to use the usual proof for the product rule and

the theorem about the determinant and row operations.

a (t + h) b (t + h) c (t + h)

F (t + h) F (t) = det d (t + h) e (t + h) f (t + h)

g (t + h) h (t + h) i (t + h)

det d (t) e (t) f (t)

g (t) h (t) i (t)

a (t + h) b (t + h) c (t + h)

a (t)

b (t)

c (t)

det d (t + h) e (t + h) f (t + h) det d (t + h) e (t + h) f (t + h)

g (t + h) h (t + h) i (t + h)

g (t + h) h (t + h) i (t + h)

a (t)

b (t)

c (t)

a (t)

b (t)

c (t)

e (t)

f (t)

+ det d (t + h) e (t + h) f (t + h) det d (t)

g (t + h) h (t + h) i (t + h)

g (t + h) h (t + h) i (t + h)

a (t)

b (t)

c (t)

a (t) b (t) c (t)

e (t)

f (t) det d (t) e (t) f (t)

+ det d (t)

g (t + h) h (t + h) i (t + h)

g (t) h (t) i (t)

Now multiply by 1/h to obtain the following for the difference quotient

det

a(t+h)a(t)

h

d (t + h)

g (t + h)

b(t+h)b(t)

h

e (t + h)

h (t + h)

+ det

c(t+h)c(t)

h

f (t + h) +det

i (t + h)

F (t+h)F (t)

.

h

a (t)

b (t)

c (t)

d(t+h)d(t)

h

e(t+h)e(t)

h

f (t+h)f (t)

h

g (t + h)

a (t)

d (t)

b (t)

e (t)

c (t)

f (t)

g(t+h)g(t)

h

h(t+h)h(t)

h

i(t+h)i(t)

h

h (t + h)

i (t + h)

Now passing to a limit yields the desired formula. Obviously this holds for any size

determinant.

13. Use the formula for the inverse in terms of the cofactor matrix to find the inverse of

the matrix

t

e

0

0

.

et cos t

et sin t

A= 0

t

t

t

0 e cos t e sin t e cos t + et sin t

25

Exercises

et

0

0

1

0

et sin t

t

t

e cos t + e sin t

0

0

et (cos t + sin t) (sin t) et

et (cos t sin t) (cos t) et

0

et cos t

et cos t et sin t

et

0

=

0

the following n n block matrix

A 0

C=

.

D B

where the D is an m r matrix, and the 0 is a r m matrix. Letting Ik denote the

k k identity matrix, tell why

A 0

Ir 0

C=

.

D Im

0 B

Now explain why det (C) = det (A) det (B) . Hint: Part of this will require an explanation of why

A 0

det

= det (A) .

D Im

See Corollary 3.1.9.

The first follows right away from block multiplication. Now

A 0

Ir 0

det (C) = det

det

D Im

0 B

A 0

Ir 0

= det

det

= det (A) det (B)

0 Im

0 B

from expanding along the last m columns for the first one and along the first r columns

for the second.

15. Suppose Q is an orthogonal matrix. This means Q is a real nn matrix which satisfies

QQT = I

Find the possible values for det (Q).

2

You have to have det (Q) det QT = det (Q) = 1 and so det (Q) = 1.

16. Suppose Q (t) is an orthogonal matrix. This means Q (t) is a real n n matrix which

satisfies

Q (t) Q (t)T = I

Suppose Q (t) is continuous for t [a, b] , some interval. Also suppose det (Q (t)) = 1.

Show that it follows det (Q (t)) = 1 for all t [a, b].

You have from the given equation that det (Q (t)) is always either 1 or 1. Since Q (t)

is continuous, so is t det (Q (t)) and so if it starts off at 1, it cannot jump to 1

because this would violate the intermediate value theorem.

26

Exercises

F.19

Exercises

3.6

1. Let m < n and let A be an m n matrix. Show that A is not one to one. Hint:

Consider the n n matrix A1 which is of the form

A

A1

0

where the 0 denotes an (n m) n matrix of zeros. Thus det A1 = 0 and so A1 is

not one to one. Now observe that A1 x is the vector,

Ax

A1 x =

0

which equals zero if and only if Ax = 0.

The hint gives it away. You could simply consider a vector of the form

0

a

where a 6= 0.

2. Let v1 , , vn be vectors in Fn and let M (v1 , , vn ) denote the matrix whose ith

column equals vi . Define

d (v1 , , vn ) det (M (v1 , , vn )) .

Prove that d is linear in each variable, (multilinear), that

d (v1 , , vi , , vj , , vn ) = d (v1 , , vj , , vi , , vn ) ,

(6.28)

and

d (e1 , , en ) = 1

(6.29)

where here ej is the vector in Fn which has a zero in every position except the j th

position in which it has a one.

This follows from the properties of determinants which are discussed above.

3. Suppose f : Fn Fn F satisfies 6.28 and 6.29 and is linear in each variable.

Show that f = d.

Consider f (x1 , , xn ) . Then by the assumptions on f it equals

X

f (x1 , , xn ) =

xi1 xin f (ei1 ein )

i1 , ,in

i1 , ,in

i1 , ,in

i1 , ,in

x1i1 x1in sgn (i1 , , in ) d (e1 e1 )

x1i1 x1in d (ei1 ein ) = d (x1 , , xn )

27

Exercises

4. Show that if you replace a row (column) of an n n matrix A with itself added to

some multiple of another row (column) then the new matrix has the same determinant

as the original one.

This was done above.

5. Use the result of Problem 4 to evaluate by hand

1 2 3

6 3 2

det

5 2 2

3 4 6

1

6

det

5

3

2

3

2

4

3

2

2

6

2

3

=5

3

4

t

e

cos t

et sin t

et cos t

et

et

et

cos t

sin t

cos t

1

2

1

2

the determinant

2

3

.

3

4

1

sin t

cos t

sin t

1 t

2e

cos t + 12 sin t

sin t 12 cos t

0

sin t

cos t

sin t

cos t .

sin t

1 t

2e

1

1

2 sin t 2 cos t

1

2 cos t 12 sin t

7. Let Ly = y (n) + an1 (x) y (n1) + + a1 (x) y + a0 (x) y where the ai are given

continuous functions defined on a closed interval, (a, b) and y is some function which

has n derivatives so it makes sense to write Ly. Suppose Lyk = 0 for k = 1, 2, , n.

The Wronskian of these functions, yi is defined as

y1 (x)

yn (x)

y1 (x)

yn (x)

..

..

.

.

(n1)

y1

(x)

(n1)

yn

(x)

W (x) = det

y1 (x)

y1 (x)

..

.

(n)

y1 (x)

yn (x)

yn (x)

..

.

(n)

yn (x)

Now use the differential equation, Ly = 0 which is satisfied by each of these functions,

yi and properties of determinants presented above to verify that W + an1 (x) W = 0.

Give an explicit solution of this linear differential equation, Abels formula, and use

28

Exercises

your answer to verify that the Wronskian of these solutions to the equation, Ly = 0

either vanishes identically on (a, b) or never.

The last formula above follows because W (x) equals the sum of determinants of

matrices which have two equal rows except for the last one in the sum which is the

displayed expression. Now let

(n1)

mi (x) = an1 (x) yi

+ + a1 (x) yi + a0 (x) yi

(n)

Since each yi is a solution to Ly = 0, it follows that yi (t) = mi (t). Now from the

properties of determinants, being linear in each row,

y1 (x)

yn (x)

y1 (x)

yn (x)

..

..

.

.

(n1)

(n1)

y1

(x) yn

(x)

=

d A(x)

e

W (x) = 0

dx

and so W (x) = CeA(x) . Thus the Wronskian either vanishes for all x or for no x.

8. Two n n matrices, A and B, are similar if B = S 1 AS for some invertible n n

matrix S. Show that if two matrices are similar, they have the same characteristic

polynomials. The characteristic polynomial of A is det (I A) .

Say A = S 1 BS. Then

det (I A)

det I S 1 BS

det S 1 S S 1 BS

det S 1 (I B) S

det S 1 det (I B) det (S)

det S 1 S det (I B) = det (I B)

=

=

=

=

tn + an1 tn1 + + a1 t + a0

and that a0 6= 0. Find a formula A1 in terms of powers of the matrix A. Show that

n

A1 exists if and only if a0 6= 0.In fact a0 = (1) det (A).

From the Cayley Hamilton theorem,

An + an1 An1 + + a1 A + a0 I = 0

Also the characteristic polynomial is

det (tI A)

and the constant term is (1) det (A) . Thus a0 6= 0 if and only if det (A) 6= 0 if and

only if A1 has an inverse. Thus if A1 exists, it follows that

a0 I = An + an1 An1 + + a1 A = A An1 an1 An2 a1 I

29

Exercises

and also

a0 I = An1 an1 An2 a1 I A

1

An1 an1 An2 a1 I

a0

p (t) p (t )

is the characteristic polynomial of A + I. Then show that if det (A) = 0, it follows

that det (A + I) 6= 0 whenever || is sufficiently small.

p (t) det (tI A) . Hence p (t ) = det ((t ) I A) = det (tI (I + A)) and

by definition, this is the characteristic polynomial of I + A. Therefore, the constant

term of this is the constant term of p (t ). Say

p (t) = tn + an1 tn1 + + a1 t + a0

tn + an1 tn1 + + ak tk

Then

n1

p (t) = p (t ) = (t ) + an1 (t )

k

+ + ak (t )

then the constant term of this p (t) is ak (1) k + terms having raised to a higher

power. Hence for all small enough, this term will dominate the sum of all the others

and it follows that the constant term is nonzero for all || small enough.

11. In constitutive

modeling of the stress and strain tensors, one sometimes considers sums

P

of the form k=0 ak Ak where A is a 33 matrix. Show using the Cayley Hamilton

theorem that if such a thing makes any sense, you can always obtain it as a finite sum

having no more than n terms.

Say the characteristic polynomial is q (t) . Then if n 3,

tn = q (t) l (t) + r (t)

where the degree of r (t) is either less than 3 or it equals zero. Thus

An = q (A) l (A) + r (A) = r (A)

and so all the terms An for n 3 can be replaced with some r (A) where the degree of

r (t) is no more than 2. Thus, assuming there are no convergence issues, the infinite

P2

sum must be of the form k=0 bk Ak .

12. Recall you can find the determinant from expanding along the j th column.

X

det (A) =

Aij (cof (A))ij

i

Think of det (A) as a function of the entries, Aij . Explain why the ij th cofactor is

really just

det (A)

.

Aij

This follows from the formula for determinant. If you take the partial derivative, you

get (cof (A))ij .

30

Exercises

13. Let U be an open set in Rn and let g :U Rn be such that all the first partial

derivatives of all components of g exist and are continuous. Under these conditions

form the matrix Dg (x) given by

Dg (x)ij

gi (x)

gi,j (x)

xj

The best kept secret in calculus courses is that the linear transformation determined

by this matrix Dg (x) is called the derivative of g and is the correct generalization

of the concept of derivative of a function of one variable. Suppose the second partial

derivatives also exist and are continuous. Then show that

X

(cof (Dg))ij,j = 0.

j

X

Next differentiate with respect to xj and sum on j using the equality of mixed partial

derivatives. Assume det (Dg) 6= 0 to prove the identity in this special case. Then

explain why there exists a sequence k 0 such that for gk (x) g (x) + k x,

det (Dgk ) 6= 0 and so the identity holds for gk . Then take a limit to get the desired result in general. This is an extremely important identity which has surprising

implications.

First assume det (Dg) 6= 0. Then from the cofactor expansion, you get

X

gi,k cof (Dg)ij = jk det (Dg)

i

If j = k you get det (Dg) on the right but if j 6= k, then the left is the expansion of a

determinant which has two equal columns. Now differentiate both sides with respect

to j using the above problem. You have to use the chain rule.

X

X

gi,kj cof (Dg)ij +

gi,k cof (Dg)ij,j

i

= jk

X (det (Dg))

r,s

gr,s

gr,sj =

r,s

Next sum on j

X

X

X

gi,kj cof (Dg)ij +

gi,k cof (Dg)ij,j =

jk cof (Dg)rs gr,sj

i,j

i,j

r,s,j

X

X

X

gi,kj cof (Dg)ij +

gi,k cof (Dg)ij,j =

cof (Dg)rs gr,sk

i,j

r,s

i,j

i,j

Subtract the terms at the end from each side using equality of mixed partial derivatives.

This gives.

X

X

X

0=

gi,k cof (Dg)

=

gi,k

cof (Dg)

ij,j

i,j

ij,j

31

Exercises

Since Dg is assumed invertible, this requires

X

cof (Dg)ij,j = 0

j

In case det (Dg (x)) = 0 for some x, consider k , a sequence of numbers converging to

0 with the property that det (Dg (x) + k I) 6= 0. Then let gk (x) = g (x) + k x. From

the above, you have

X

X

cof (Dg)ij,j (x) = lim

cof (Dgk )ij,j (x) = 0

k

1

a0

a20

..

.

n1

a

0

an

0

1

a1

a21

..

.

1

an

a2n

..

.

an1

1

an1

an1

n

ann

Y

(aj ai )

0i<jn

By this is meant to take the product of all terms of the form (aj ai ) such that j > i.

Hint: Show it works if n = 1 so you are looking at

1 1

a0 a1

Then suppose it holds for n 1 and consider the

polynomial.

1

1

a0

a

1

2

a20

a

1

p (t) .

..

.

.

.

n1

n1

a

a

1

0

n

an

a

0

1

Explain why p (aj ) = 0 for i = 0, , n 1. Thus

p (t) = c

n1

Y

i=0

1

t

t2

..

.

tn1

tn

(t ai ) .

Of course c is the coefficient of tn . Find this coefficient from the above description of

p (t) and the induction hypothesis. Then plug in t = an and observe you have the

formula valid for n.

1 1

= (a1 a0 ) so the formula holds. Now

In the case of n = 1, you have

a0 a1

suppose it holds for n and consider p (t) as above. Then it is obvious that p (aj ) = 0

32

Exercises

for i = 0, , n 1 because you have the determinant of a matrix with two equal

columns. Thus

n1

Y

p (t) = c

(t ai ) .

i=0

because p (t) has degree n and so it has no more than n roots. So what is c? The

n

coefficient

it

Q of t is the Vandermonde determinant which is n n and by induction,

equals 0i<jn1 (aj ai ) . From the above product, the coefficient of tn is c. Thus

p (t) =

0i<jn1

(aj ai )

n1

Y

i=0

(t ai )

Y

p (an ) =

0i<jn1

0i<jn

F.20

(aj ai )

n1

Y

i=0

(an ai )

(aj ai )

Exercises

4.6

1. Let {u1 , , un } be vectors in Rn . The parallelepiped determined by these vectors

P (u1 , , un ) is defined as

( n

)

X

P (u1 , , un )

tk uk : tk [0, 1] for all k .

k=1

{Ax : x P (u1 , , un )}

is also a parallelepiped.

A typical thing in {Ax : x P (u1 , , un )} is

n

X

k=1

tk Auk : tk [0, 1]

2. In the context of Problem 1, draw P (e1 , e2 ) where e1 , e2 are the standard basis vectors

for R2 . Thus e1 = (1, 0) , e2 = (0, 1) . Now suppose

1 1

E=

0 1

where E is the elementary matrix which takes the third row and adds to the first.

Draw

{Ex : x P (e1 , e2 )} .

33

Exercises

In other words, draw the result of doing E to the vectors in P (e1 , e2 ). Next draw the

results of doing the other elementary matrices to P (e1 , e2 ).

For E the given elementary matrix, the result is as follows.

P (e1 , e2 )

E(P (e1 , e2 ))

1 0

In caseE =

, the given square P (e1 , e2 ) becomes a rectangle. If > 0,

0

the square is stretched by a factor of in the y direction. If < 0 the rectangle

is reflected across the x axis and

in addition is stretched by a factor of || in the y

0

direction. When E =

the result is similar only it features changes in the x

0 1

direction. These elementary matrices

the area unless || = 1 when the area

do change

0 1

is unchanged. The permutation

simply switches e1 and e2 so the result

1 0

appears to be a square just like the one you began with.

3. In the context of Problem 1, either draw or describe the result of doing elementary

matrices to P (e1 , e2 , e3 ). Describe geometrically the conclusion of Corollary 4.3.7.

It is the same sort of thing. The elementary matrix either switches the ei about or it

produces a shear or a magnification in one direction.

4. Consider a permutation of {1, 2, , n}. This is an ordered list of numbers taken from

this list with no repeats, {i1 , i2 , , in }. Define the permutation matrix P (i1 , i2 , , in )

as the matrix which is obtained from the identity matrix by placing the j th column

of I as the ith

j column of P (i1 , i2 , , in ) . What does this permutation matrix do to

T

the column vector (1, 2, , n) ?

T

5. Consider the 3 3 permutation matrices. List all of them and then determine the

dimension of their span. Recall that you can consider an m n matrix as something

in Fnm .

Here they are.

1

0

0

0

1

0

0 0

1 0 ,

0 1

1 0

0 0 ,

0 1

0 0

1 0

0 1

1 0

0 0

0 1

1

0 ,

0

0

1 ,

0

0

0

1

0

0

1

1 0

0 1

0 0

0 1

1 0

0 0

So what is the dimension of the span of these? One way to systematically accomplish

this is to unravel them and then use the row reduced echelon form. Unraveling these

yields the column vectors

34

Exercises

1

0

0

0

1

0

0

0

1

0

0

1

1

0

0

0

1

0

0

1

0

0

0

1

1

0

0

0

1

0

1

0

0

0

0

1

1

0

0

0

0

1

0

1

0

0

0

1

0

1

0

1

0

0

row reduced echelon form.

1 0 0 0 1 0

1

0 0 1 1 0 0

0

0 1 0 0 0 1

0

0 1 0 1 0 0

0

0 0 1 0 1 0

0

0 0 1 0 0 1

0

0 1 0 0 1 0

0

1 0 0 1 0 0

0

0

1

0

0

0

0

0

0

0

0

0

1

0

0

0

0

0

0

0

0

0

1

0

0

0

0

0

The dimension is 5.

0 1

0 1

0 1

0 1

1 1

0 0

0 0

0 0

0 0

1 2 0

(a)

0 1 7

This one is not.

1 0 0 0

(b) 0 0 1 2

0 0 0 0

This one is.

1 1 0 0 0 5

(c) 0 0 1 2 0 4

0 0 0 0 1 3

This one is.

7. Row reduce the following matrices to obtain the row reduced echelon form. List the

pivot columns in the original matrix.

1 2 0 3

1 0 0 3

(a) 2 1 2 2 , row echelon form: 0 1 0 0

1 1 0 3

0 0 1 2

1 2 3

1 0 0

2 1 2

0 1 0

(b)

3 0 0 , row echelon form: 0 0 1

3 2 1

0 0 0

1 0 0 0

1 2 1 3

(c) 3 2 1 0 , row echelon form: 0 1 12 0

3 2 1 1

0 0 0 1

Exercises

35

8. Find the rank and nullity of the following matrices. If the rank is r, identify r columns

in the original matrix which have the property that every other column may be

written as a linear combination of these.

0 1 0 2 1 2 2

0 1 0 2 0 1 1

0 3 2 12 1 6 8

0 0 1 3 0 1 2

(a)

0 1 1 5 0 2 3 , row echelon form: 0 0 0 0 1 1 1 Rank

0 2 1 7 0 3 4

0 0 0 0 0 0 0

equals 3 nullity equals 4. A basis is columns 2,3,5

0 1 0 2 0 1 0

0 1 0 2 0 1 0

0 3 2 6 0 5 4

0 0 1 0 0 1 2

(b)

0 1 1 2 0 2 2 , row echelon form: 0 0 0 0 0 0 0 Rank

0 2 1 4 0 3 2

0 0 0 0 0 0 0

is 2 nullity equals 5. A basis is columns 2,3.

0 1 0 2 1 1 2

0 1 0 2 0 1 0

0 3 2 6 1 5 1

0 0 1 0 0 1 0

(c)

0 1 1 2 0 2 1 , row echelon form: 0 0 0 0 1 0 0 Rank

0 2 1 4 0 3 1

0 0 0 0 0 0 1

is 4 and nullity is 3. A basis is columns 2,3,5,7.

9. Find the rank of the following matrices. If the rank is r, identify r columns in the

original matrix which have the property that every other column may be written

as a linear combination of these. Also find a basis for the row and column spaces of

the matrices.

1 2 0

1 0 0

3 2 1

0 1 0

(a)

2 1 0 , row echelon form: 0 0 1 . The rank is 3.

0 2 1

0 0 0

1 0 0

1 0 0

4 1 1

0 1 0

(b)

2 1 0 , row echelon form: 0 0 1 . The rank is 3

0 2 0

0 0 0

0 1 0 2 1 2 2

0 1 0 2 0 1 1

0 3 2 12 1 6 8

0 0 1 3 0 1 2

(c)

0 1 1 5 0 2 3 , row echelon form: 0 0 0 0 1 1 1 .

0 2 1 7 0 3 4

0 0 0 0 0 0 0

The rank is 3.

0 1 0 2 0 1 0

0 1 0 2 0 1 0

0 3 2 6 0 5 4

0 0 1 0 0 1 2

(d)

0 1 1 2 0 2 2 , row echelon form: 0 0 0 0 0 0 0 . The

0 2 1 4 0 3 2

0 0 0 0 0 0 0

rank is 2.

0 1 0 2 1 1 2

0 1 0 2 0 1 0

0 3 2 6 1 5 1

0 0 1 0 0 1 0

(e)

0 1 1 2 0 2 1 , row echelon form: 0 0 0 0 1 0 0 . The

0 2 1 4 0 3 1

0 0 0 0 0 0 1

rank is 4.

10. Suppose A is an m n matrix. Explain why the rank of A is always no larger than

min (m, n) .

36

Exercises

It is because you cannot have more than min (m, n) nonzero rows in the row reduced

echelon form. Recall that the number of pivot columns is the same as the number of

nonzero rows from the description of this row reduced echelon form.

11. Suppose A is an m n matrix in which m n. Suppose also that the rank of A equals

m. Show that A maps Fn onto Fm . Hint: The vectors e1 , , em occur as columns

in the row reduced echelon form for A.

It follows from the fact that e1 , , em occur as columns in row reduced echelon form

that the dimension of the column space of A is n and so, since this column space is

A (Rn ) , it follows that it equals Fm .

12. Suppose A is an m n matrix and that m > n. Show there exists b Fm such that

there is no solution to the equation

Ax = b.

Since m > n the dimension of the column space of A is no more than n and so the

columns of A cannot span Fm .

13. Suppose A is an m n matrix in which m n. Suppose also that the rank of A

equals n. Show that A is one to one. Hint: If not, there exists a vector, x 6= 0 such

that Ax = 0, and this implies at least one column of A is a linear combination of the

others. Show this would require the column rank to be less than n.

The hint gives it away. The matrix has at most n independent columns. If one is a

combination of the others, then its rank is less than n.

14. Explain why an n n matrix A is both one to one and onto if and only if its rank is

n.

If A is one to one, then its columns are linearly independent, hence a basis and so it

is also onto. Since the columns are independent, its rank is n. If its rank is n this just

says that the columns are independent.

15. Suppose A is an m n matrix and {w1 , , wk } is a linearly independent set of

vectors in A (Fn ) Fm . Suppose also that Azi wi . Show that {z1 , , zk } is also

linearly independent.

P

P

If i ci zi = 0, apply A to both sides to obtain i ci wi = 0. By assumption, each

ci = 0.

16. Let A, B be m n matrices. Show that rank (A + B) rank (A) + rank (B).

This is obvious from the following.

rank (A) + rank (B)

17. Suppose A is an m n matrix, m n and the columns of A are independent. Suppose also that {z1 , , zk } is a linearly independent set of vectors in Fn . Show that

{Az1 , , Azk } is linearly independent.

37

Exercises

to one. If

n

X

cj Azj = 0,

j=1

then A

P

n

j=1 cj zj

= 0 and so

Pn

j=1 cj zj

= 0 therefore each cj = 0.

dim (ker (AB)) dim (ker (A)) + dim (ker (B)) .

Hint: Consider the subspace, B (Fp ) ker (A) and suppose a basis for this subspace

is {w1 , , wk } . Now suppose {u1 , , ur } is a basis for ker (B) . Let {z1 , , zk }

be such that Bzi = wi and argue that

ker (AB) span (u1 , , ur , z1 , , zk ) .

Here is how you do this. Suppose ABx = 0. Then Bx ker (A) B (Fp ) and so

Pk

Bx = i=1 ai Bzi showing that

x

k

X

ai zi ker (B) .

i=1

Let {w1 , , wk } be a basis for B (Fp ) ker (A) and suppose {u1 , , ur } is a basis

for ker (B). Let Bzi = wi . Now suppose x ker (AB). Then Bx ker (A) B (Fp )

and so

k

k

X

X

Bx =

a i wi =

ai Bzi

i=1

so

i=1

k

X

i=1

Hence

x

k

X

ai zi ker (B)

ai zi =

i=1

r

X

b j uj

j=1

Note that a little more can be said. {u1 , , ur , z1 , , zk } is independent. Here is

why. Suppose

k

X

X

a i ui +

bj zj = 0

i

j=1

X

bj Bzj = 0

j

follows that in fact,

dim (ker (AB)) = k + r dim (ker (A)) + dim (ker (B))

and the remaining inequality is an equality if and only if B (Fp ) ker (A).

38

Exercises

19. Let m < n and let A be an m n matrix. Show that A is not one to one.

There are more columns than rows and at most m can be pivot columns so it follows

at least one column is a linear combination of the others hence A is not one too one.

20. Let A be an m n real matrix and let b Rm . Show there exists a solution, x to the

system

AT Ax = AT b

Next show that if x, x1 are two solutions, then Ax = Ax1 . Hint: First show that

T

AT A = AT A. Next show if x ker AT A , then Ax = 0. Finally apply the Fredholm alternative. Show AT b ker(AT A) . This will give existence of a solution.

T

T

AT A = AT AT

= AT A. Also if AT Ax = 0, then

AT Ax, x = |Ax|2 = 0

and so Ax = 0. Is AT b ker AT A ? Say AT Ax = 0. Then Ax = 0. Then

AT b, x = (b,Ax) = (b, 0) = 0

21. Show that in the context of Problem 20 that if x is the solution there, then |b Ax|

|b Ay| for every y. Thus Ax is the point of A (Rn ) which is closest to b of every

point in A (Rn ). This is a solution to the least squares problem.

|bAy|

=

=

=

=

|bAx+AxAy|

1

0

0 1

T

4

22. Here is a point in R : (1, 2, 3, 4) . Find the point in span

2 , 3

3

2

is closest to the given point.

This just says to find the least squares solution to the

1 0

1

0 1 x

2

=

2 3 y

3

3 2

4

1

0

2

3

T

0

1

0

1

3 2

2

3

0

1

0

1

x

=

2

3 y

2

3

T

0

1

2

1

3 3

2

4

which

equations

39

Exercises

14 12

12 14

x

y

19

19

, Solution is:

19

0 + 19

2

26

26

3

19

26

19

26

1

=

3

2

19

26

19

26

95

26

95

26

23. Here is a point in R4 : (1, 2, 3, 4) . Find the point on the plane described by x + 2y

4z + 4w = 0 which is closest to the given point.

2y + 4z 4w

y

where y, z, w are arbitary. Thus it is

The plane is of the form

z

w

the span of the vectors

2

4

4

1 0 0

0 , 1 , 0

0

0

1

and so you are looking for a least squares solution to the system

2 4 4

1

a

1 0 0

0 1 0

3

c

0 0 1

4

squares.

T

2 4 4

2 4 4

2 4 4

1

a

1 0 0 1 0 0

1 0 0 2

b =

0 1 0 0 1 0

0 1 0 3

c

0 0 1

0 0 1

0 0 1

4

56

5

8

8

a

0

37

8 17 16 b = 7 , Solution is: 147 The closest point is

37

112

8 16 17

c

0

37

then

28

2

4

4

37

147 0 112 0 56

56 1

37

37 0 + 37 1 + 37 0 = 147

37

112

0

0

1

37

28

Check:

2 4 4

37

56

37

147

37

112

37

=0

24. Suppose A, B are two invertible n n matrices. Show there exists a sequence of row

operations which when done to A yield B. Hint: Recall that every invertible matrix

is a product of elementary matrices.

Both A, B have row reduced echelon form equal to I. Therefore, they are row equivalent.

E1 Ep A = I, F1 Fr B = I

40

Exercises

and so

E1 Ep A = F1 Fr B

Now multiply on both sides by the inverses of the Fi to obtain a product of elementary

matrices which multiplied by A give B.

25. If A is invertible and n n and B is n p, show that AB has the same null space as

B and also the same rank as B.

From the above problem, AB and B have exactly the same row reduced echelon form.

Thus the result follows.

26. Here are two matrices in row reduced echelon form

1 0 1

1

A = 0 1 1 , B = 0

0

0 0 0

0 0

1 1

0 0

Does there exist a sequence of row operations which when done to A will yield B?

Explain.

No. The row reduced echelon form is unique.

27. Is it true that an upper triagular matrix has rank equal to the number of nonzero

entries down the main diagonal?

1 0 2 0

0 1 1 7

No.

0 0 0 1

0 0 0 0

28. Let {v1 , , vn1 } be vectors in Fn . Describe a systematic way to obtain a vector vn

which is perpendicular to each of these vectors. Hint: You might consider something

like this

e1

e2

en

v11

v12

v1n

det

..

..

..

.

.

.

v(n1)1

th

v(n1)2

v(n1)n

This is a good hint. You formally expand along the top row. It will work. If you

replace ei with vki so that the top row becomes.

vk1 vk2 vkn

then the determinant will be equal to 0 because it has two equal rows. However, this

determinant is also what you get when you take the dot product of the above formal

determinant having the ei on the top row with vk .

29. Let A be an m n matrix. Then ker (A) is a subspace of Fn . Is it true that every

subspace of Fn is the kernel or null space of some matrix? Prove or disprove.

Let M be a subspace of Fn . If it equals {0} , consider the matrix I. Otherwise, it has

a basis {m1 , , mk } . Consider the matrix

m1 mk 0

41

Exercises

30. Let A be an nn matrix and let P ij be the permutation matrix which switches the ith

and j th rows of the identity. Show that P ij AP ij produces a matrix which is similar

to A which switches the ith and j th entries on the main diagonal.

This is easy to see when you consider that P ij is its own inverse and that P ij multiplied

on the right switches the ith and j th columns. Thus you switch the columns and then

you switch the rows. This has the effect of switching Aii and Ajj . For example,

a b c d

1 0 0 0

a d c b

1 0 0 0

0 0 0 1 e f z h 0 0 0 1 n g h t

0 0 1 0 j k l m 0 0 1 0 = j m l k

e h z f

0 1 0 0

n t h g

0 1 0 0

More formally, the iith entry of P ij AP ij is

X ij

ij

Pis Asp Ppi

= Pijij Ajj Pjiij = Aij

s,p

31. Recall the procedure for finding the inverse of a matrix on Page 48. It was shown that

the procedure, when it works, finds the inverse of the matrix. Show that whenever

the matrix has an inverse, the procedure works.

If A has an inverse, then it is one to one. Hence the columns are independent. Therefore, they are each pivot columns. Therefore, the row reduced echelon form of A is I.

This is what was needed for the procedure to work.

F.21

Exercises

5.8

1.

2.

3.

4.

5.

1 2 0

1 0 0

1 2

Find a LU factorization of 2 1 3 . = 2 1 0 0 3

1 2 3

1 0 1

0 0

1 2 3 2

1

0

0

1

1

0 0

Find a LU factorization of 1 3 2 1 . = 1

5 0 1 3

5 10 1

0

1 2 1

1 0 0

1 0

Find a P LU factorization of 1 2 2 . = 0 0 1 2 1

2 1 1

0 1 0

1 0

1 2 1 2 1

Find a P LU factorization of 2 4 2 4 1 .

1 2 1 3 2

1 0 0

1 2 1 2 1

1 0 0

= 0 1 0 2 1 0 0 0 0 0 1

0 0 1

1 0 1

0 0 0 1 1

1 2 1

1 2 2

Find a P LU factorization of

2 4 1 .

3 2 1

1 0 0 0

1 0 0 0

1 2

1

0 0 0 1 3 1 0 0 0 4 2

=

0 0 1 0 2 0 1 0 0 0 1

0 1 0 0

1 0 1 1

0 0

0

0

3

3

3

2

1 1

24 17

0

1 2

1

0 0 3 1

1

0 0

1

2

1

0

42

Exercises

6. Is there only one LU factorization for a given matrix? Hint: Consider the equation

0 1

1 0

0 1

=

.

0 1

1 1

0 0

Apparently, LU factorization

0 1

1 0

0

=

0 1

1 1

0

0 1

1 0

0

=

0 1

0 1

0

7. Here is a matrix and

1

A= 1

0

is not unique.

1

0

1

1

an LU factorization

2 5 0

1

1 4 9 = 1

1 2 5

0

of it.

0 0

1

1 0 0

1 1

0

1

Ax = 2

3

2

5

0

1 1 9

0

1 14

1 0 0

a

1

1

1 1 0 b = 2 , Solution is: 1

0 1 1

c

3

4

24t4 9

x

1 2

5

0

1

y

23t4 5

1 , Solution is:

0 1 1 9

z =

4 14t4

0 0

1 14

4

w

t4

1 2 1

3 2 1

1 0 2

1 2 1

3 2 1

1 0 2

1

11 11

3

11 11

1

11 11

11

0

0

1 2 1

3 0 1

1 0 2

, t4 R

13

2 11 61 2

66

5

66 2 11 61 2

1

2

33 2 11

3 2

4

6

11

11 11

11

6

2

2 11

11 2 11

11

0

1

1

43

Exercises

1 2 1 0

3 0 1 1

1 0 2 1

1

1

10

11

0

11 11

11

3

3

1

= 11

11 110

1011 310 2 5

1

1

11 110 10 11 10 2 5

11

2

6

4

11

11

11

11

11

11

11

2

1

2

0

11 10 11

22 10 11 55 10

11

1

1

0

0

2

5

2

5

2

5

10. If you had a QR factorization, A = QR, describe how you could use it to solve the

equation Ax = b. Note, this is not how people usually solve systems of equations.

You could first solve Qy = b to obtain y = QT b. Then you would solve Rx = y which

would be easy because R is upper triangular.

11. If Q is an orthogonal matrix, show the columns are an orthonormal set. That is show

that for

Q = q1 qn

it follows that qi qj = ij . Also show that any orthonormal set of vectors is linearly

independent.

Pn

Say i=1 ci qi = 0. Then take the inner product of both sides with qk to obtain

X

ci ij = 0

12. Show you cant expect uniqueness for QR factorizations. Consider

0 0 0

0 0 1

0 0 1

and verify this equals

and also

1 2

2

1

2 2

1

0

0 0

1

0

2 0 0

2

0 0

0 12 2

1 0

0 1

0 0

0

0

0 0

1

0

0 0

0 1 .

0 1

follow there is only one QR factorization?

0

1

0

0 0

0 0

2

1

1 2 0

0 0 0 = 0 0

2

2

2

1

1

0 0

0 0 0

2 2 0 2 2

1 0 0

0 0 0

0 0 0

0 1 0 0 0 1 = 0 0 1

0 0 1

0 0 1

0 0 1

2

0

0

0

1

1

44

Exercises

Now if A1 exists and A = QR = Q R where R, R have positive entries down

1

the main diagonal, then you get R (R )

= QT Q U, where U is an orthogonal

matrix. Thus you have an upper triangular matrix which has positive entries down

the diagonal equal to an orthogonal matrix. Obviously, this requires that the upper

triangular matrix is the identity. Hence R = R and Q = Q . Why is this so obvious?

Consider the following case of a 33 which is orthogonal.

a b c

0 d e

0 0 f

Then this matrix

a b

0 d

0 0

a 0

b d

c e

2

c

a 0 0

a + b2 + c2 ce + bd cf

e

b d 0

ce + bd

d2 + e2 f e

=

f

c e f

cf

fe

f2

2

0

a b c

a

ab

ac

0 0 d e = ab b2 + d2

de + bc

f

0 0 f

ac de + bc c2 + f 2 + e2

Thus both of these on the right end are the identity and so b = c = 0, a = 1 since

the diagonal entries are all positive. Then also d = 1 and c, e = 0 and f = 1. Now it

follows that all the off diagonal entries must equal 0 since otherwise the magnitude of

a column would not equal 1.

A = a1 an

Form a QR factorization for A.

a1

an

q1

qn

r11

0

..

.

r12

r22

..

.

r1n

r2n

rnn

Prove that every subspace of Rn has an orthonormal basis. The procedure just described is similar to the Gram Schmidt procedure which will be presented later.

From the way we multiply matrices,

ak span (q1 , , qk )

Hence span (a1 , , ak ) span (q1 , , qk ). In addition to this, each rii > 0 because

the ai are a linearly independent set and if this were not so, then the triangular matrix

on the right would fail to be invertible which would be a contradiction since you could

express it as the product of two invertible matrices. Multiplying on the right by its

inverse, you can get a similar expression for the qi and see that span (q1 , , qk )

span (a1 , , ak ) . Thus

span (a1 , , ak ) = span (q1 , , qk )

45

Exercises

If you have a subspace, let {a1 , , ak } be a basis and then extend to a basis of

Rn , {a1 , , an }. Then form the above QR factorization and you will have that

{q1 , , qk } is an orthonormal basis for the subspace.

14. Suppose Qn Rn converges to an orthogonal matrix Q where Qn is orthogonal and Rn

is upper triangular having all positive entries on the diagonal. Show that then Qn

converges to Q and Rn converges to the identity.

Let

Q = (q1 , , qn ) , Qk = qk1 , , qkn

k

where the q are the columns. Also denote by rij

the ij th entry of Rk . Thus

It follows

Qk Rk = qk1 , , qkn

k

r11

..

k

rnn

k k

r11

q1 q1

and so

k k

k

r11

= r11

q1 1

Therefore,

qk1 q1 .

k k

k k

r12

q1 + r22

q2 q2

k

lim r12

= lim q2 qk1 = (q2 q1 ) = 0.

k

Therefore,

k k

lim r22

q2 = q2

k

k

and since r22

> 0, it follows, as in the first part that r22

1. Hence

lim qk2 = q2 .

k

lim rij

=0

k

lim rjj

= 1, lim qkj = qj .

Thus Rk I and Qk Q.

46

Exercises

F.22

Exercises

6.6

1. Maximize and minimize z = x1 2x2 + x3 subject to the constraints x1 + x2 + x3

10, x1 + x2 + x3 2, and x1 + 2x2 + x3 7 if possible. All variables are nonnegative.

The constraints lead to the augmented

1 1 1

1 1 1

1 2 1

The obvious solution is not feasible.

1 1

0 0

0 1

matrix

1

0

0

0 0 10

1 0 2

0 1 7

Do a row operation.

1 1 0 0 10

0 1 1 0 8

0 1 0 1 3

An obvious solution is still not feasible. Do another operation couple of row operations.

1 1 1 0 1 0 2

0 0 0 1 1 0 8

0 1 0 0 1 1 5

tableau.

1

0

0

1

1 1 0 1 0 0 2

0 0 1 1 0 0 8

1 0 0 1 1 0 5

2 1 0 0 0 1 0

1 1

0 0

0 1

0 2

1

0

0

0

0

1

0

0

1

1

1

1

0

0

1

0

0

0

0

1

2

8

5

0

First lets work on minimizing this. There is a +2. The ratios are then 5, 2 so the pivot

is the 1 on the top of the second column. The next tableau is

1 1 1 0 1 0 0 2

0 0 0 1 1 0 0 8

1 0 1 0 2 1 0 3

2 0 2 0 1 0 1 4

There is a 1 on the bottom. The ratios of interest for that column are 3/2, 8, and so

the pivot is the 2 in that column. Then the next tableau is

1

1

1

7

1

0 0

0

2

2

2

2

1

13

1

0

1 0 12 0

2

2

2

1 0 1 0 2 1 0

3

23 0 23 0 0 12 1 11

2

Now you stop because there are no more positive numbers to the left of 1 on the

bottom row. The minimum is 11/2 and it occurs when x1 = x3 = x6 = 0 and

x2 = 7/2, x4 = 13/2, x6 = 11/2.

47

Exercises

Next consider maximization. The simplex tableau was

1 1 1 0 1 0 0

0 0 0 1 1 0 0

0 1 0 0 1 1 0

1 2 1 0 0 0 1

2

8

5

0

This time you work on getting rid of the negative entries. Consider the 1 in the first

column. There is only one ratio to consider so 1 is the pivot.

1 1 1 0 1 0 0 2

0 0 0 1 1 0 0 8

0 1 0 0 1 1 0 5

0 3 0 0 1 0 1 2

There remains a 1. The ratios are 5 and 8 so the next pivot is the 1 in the third row

and column 5.

1 2 1 0 0 1 0 7

0 1 0 1 0 1 0 3

0 1 0 0 1 1 0 5

0 4 0 0 0 1 1 7

0, x3 = 0, x4 = 3, x5 = 5, x6 = 0.

2. Maximize and minimize the following if possible. All variables are nonnegative.

x1 + 2x2 + x3 7

an augmented matrix for the constraints is

1 1 1 1 0 0 10

1 1 1 0 1 0 1

1 2 1 0 0 1 7

The obvious solution is not feasible. Do some row operations.

0 0 0 1 1 0 9

1 1 1 0 1 0 1

1 0 1 0 2 1 5

Now the obvious solution is feasible.

0 0 0

1 1 1

1 0 1

1 2 0

First preserve the simple columns.

0 0 0

1 1 1

1 0 1

3 0 2

1 1 0 0 9

0 1 0 0 1

0 2 1 0 5

0 0 0 1 0

1

0

0

0

1

1

2

2

0

0

1

0

0

0

0

1

9

1

5

2

48

Exercises

Lets try to maximize first. Begin with the first

Use it.

0 0 0 1 1 0

1 1 1 0 1 0

0 1 0 0 1 1

0 3 1 0 1 0

so the new pivot is the 1 on the third row.

0 1 0 1 0

1 2 1 0 0

0 1 0 0 1

0 4 1 0 0

0 9

0 1

0 6

1 1

1

1

1

1

Next consider the minimum.

0 0 0 1 1 0

1 1 1 0 1 0

1 0 1 0 2 1

3 0 2 0 2 0

There is a positive 2 in the bottom row left of 1.

2.

1

0 21 1 0 12

2

1

1 1 1 0 0

2

2

2

1 0 1 0 2 1

2 0 1 0 0 1

0

0

0

1

= 0.

3

7

6

7

9

1

5

2

0

0

0

1

0 13

2

0 72

0 5

1 7

and x1 + x2 + x3 7

This time, lets use artificial variables to find an initial simplex tableau. Thus you

add in an artificial variable and then do a minimization procedure.

1

1

1

0

1

1

1

0

1

3

1

0

1

0

0

0

1 1 1

1 1 3

1 1 1

1 1 3

Now use the third column.

2 2

3

1

2

3

0

1

2

3

0

3

0

0

0

1

0

0

0 0

0 1

1 0

0 1

0

0

0

1

8

1

7

0

1 0 0 0 0 8

0 1 0 1 0 1

0 0 1 0 0 7

0 1 0 0 1 1

1 13 0

0 1 0

0 13 1

0 0 0

31

1

31

1

0

0

0

1

23

3

20

3

49

Exercises

It follows that a basic solution is feasible if

2 2

0 1 13 0

3

3

1 1 3 0 1 0

2

2

0 0 13 1

3

3

2

2

0

3

3

1 1 3

2

2

0

3

3

1 2 3

1 13 0

0 1 0

0 13 1

0 0 0

23

3

20

3

0

0

0

1

23

3

20

3

2

2

0 1 13 0 0 23

3

3

3

1 1 3 0 1 0 0 1

2

2

0 0 13 1 0 20

3

3

3

2 1 0 0 1 0 1 1

Lets do minimization first.

0

1

0

3

0 2 1 1 0 0 7

1 3 0 1 0 0 1

0 2 0 1 1 0 6

0 3 0 2 0 1 2

in the bottom row left of the 1. Work with that

0 0 0 1 0 1

1 1 1 0 0 1

0 0 2 0 1 1

3 0 1 0 0 2

column.

0

1

0

7

0

6

1 14

0

0 0 1 0 1 0

1

1

1 1 0 0 1 0

7

2

2 0 0 1 3 0 20

4 1 0 0 0 3 1 21

Next consider the maximum. The simplex tableau was

2

2

0 1 13 0 0 23

3

3

3

1 1 3 0 1 0 0 1

2

0 0 13 1 0 20

3

3

3

2 1 0 0 1 0 1 1

Use the first column.

0

1

0

0

0

1

0

3

2

3

2

6

1

0

0

0

1

1

1

1

0

0

1

0

0

0

0

1

7

1

6

1

50

Exercises

There is still a negative on the bottom row to the left of 1.

0 0 0 1 0 1 0 1

1 1 1 0 0 1 0 7

0 0 2 0 1 1 0 6

0 3 4 0 0 1 1 7

There are no more negatives on the bottom row left of 1 so stop. The maximum

is 7 and it occurs when x1 = 7, x2 = 0, x3 = 0.

x1 + 2x2 + x3 7.

The augmented matrix for the constraints is

1 1 1 1 0 0 10

1 1 1 0 1 0 1

1 2 1 0 0 1 7

The basic solution is not feasible because of that 1. Lets do a row operation to

change this. I used the 1 in the second column as a pivot and zeroed out what

was above and below it. Now it seems that the basic solution is feasible.

2 0 2 1 1 0 11

1 1 1 0 1 0 1

1 0 1 0 2 1 5

Assemble the simplex tableau.

2

0

2 1 1

1

1

1 0 1

1 0 1 0 2

2 1 0 0 0

0

0

1

0

0

0

0

1

11

1

5

0

2

1

1

1

0 2 1 1 0 0 11

1 1 0 1 0 0 1

0 1 0 2 1 0 5

0 1 0 1 0 1 1

Now lets maximize.

2

1

1

1

Lets do minimization first. Work with the third column because there is a positive

entry on the bottom.

0 2 0 1 1 0 0 9

1

1 1 0 1 0 0 1

0

1 0 0 1 1 0 6

2 1 0 0 0 0 1 0

0

1

0

0

2

1

1

1

1 1 0

0 1 0

0 2 1

0 1 0

0 11

0 1

0 5

1 1

51

Exercises

Lets begin with the first column.

0 2

1 1

0 1

0 1

There is still a 2 to the left of

0 3

1 2

0 1

0 3

0

1

0

2

1

0

0

0

1

1

1

2

0

0

1

0

0

0

0

1

0

1

0

2

1

0

0

0

9

1

6

2

3

7

6

14

0 1 0

0 1 0

1 1 0

0 2 1

7, x2 = x3 = 0.

x1 + 2x2 + x3 7.

The augmented matrix for the constraints is

1 1 1 1 0 0 10

1 1 1 0 1 0 1

1 2 1 0 0 1 7

Of course the obvious or basic solution is not feasible. Do a row operation involving a pivot in the second row to try and fix this.

2 0 2 1 1 0 11

1 1 1 0 1 0 1

1 0 1 0 2 1 5

Now all is well. Begin to assemble the

2

0

2

1

1

1

1 0 1

1 2 0

simplex tableau.

1 1

0 1

0 2

0 0

0

0

1

0

0

0

0

1

11

1

5

0

2 0 2 1 1 0 0 11

1 1 1 0 1 0 0 1

1 0 1 0 2 1 0 5

1 0 2 0 2 0 1 2

Next lets maximize. There is only one negative

3

0 23 1 0 12

2

1 1 1 0 0 1

2

2

2

1 0 1 0 2 1

0 0 1 0 0 1

0 27

2

0 72

0 5

1 7

52

Exercises

Next lets find the minimum.

2

1

1

1

0

1

0

1

0

1

0

0

2

1

1

2

1 1 0

0 1 0

0 2 1

0 2 0

0 11

0 1

0 5

1 2

1

0

0

0

0

0

0

1

has a 2.

2

1

1

2

0

1

0

0

1

1

1

0

0

0

1

0

9

1

6

0

1.

two contradict but show they contradict using the simplex algorithm.

You can do this by using artificial variables, x5 .

1 1 1 0 1

1 2 0 1 0

0 0 0 0 1

Thus

0 12

0 5

1 0

1 1 1 0 1 0 12

1 2 0 1 0 0 5

1 1 1 0 0 1 12

Next start with the 1 in the first column.

0 1 1 1 1

1 2

0

1 0

0 1 1 1 0

0 7

0 5

1 7

Thus the minimum value of z = x5 is 7 but, for there to be a feasible solution, you

would need to have this minimum value be 0.

4. Find a solution to the following inequalities for x, y 0 if it is possible to do so. If it

is not possible, prove it is not possible.

(a)

6x + 3y 4

8x + 4y 5

Use an artificial variable. Let x1 = x, x2

artificial

variable x5 . Then minimize

x5 as

6 3 1 0 1 0 4

8 4 0 1 0 0 5

0 0 0 0 1 1 0

Keep the simple columns.

6 3 1 0

8 4 0 1

6 3 1 0

described earlier.

1

0

0

0 4

0 5

1 4

53

Exercises

Now proceed to minimize.

0 0

8 4

0 0

1 43

0

1

1 43

these inequalities with x1 , x2 0.

6x1 + 4x3 11

(b) 5x1 + 4x2 + 4x3 8

6x1 + 6x2 + 5x3 11

The augmented matrix is

6 0 4

5 4 4

6 6 5

1

0

0

1

4

0

0

1

5

1

4

1

0

0

0 0 11

1 0 8

0 1 11

It is not clear whether there is a solution which has all variables nonnegative.

However, if you do a row operation using 5 in the first column as a pivot, you get

0 24

45 1 65 0 75

5

5

4

4 0 1 0 8

6

1

0

0 65 1 75

5

5

6x1 + 4x3 11

(c) 5x1 + 4x2 + 4x3 9

6x1 + 6x2 + 5x3 9

The augmented matrix is

6 0 4 1 0 0 11

5 4 4 0 1 0 9

6 6 5 0 0 1 9

Lets include an artificial variable

6 0 4

5 4 4

6 6 5

0 0 0

Preserving the simple columns,

6 0 4

5 4 4

6 6 5

5 4 4

Use the first column.

0 6 1

0 1 1

6

6 6

5

0 1 61

1 0 0 0 0 11

0 1 0 1 0 9

0 0 1 0 0 9

0 0 0 1 1 0

0

1

0

0

0

0

0

1

11

9

9

9

0 1

1 65

0

1

1 65

0

1

0

0

0

0

0

1

1 0 0

0 1 0

0 0 1

0 1 0

1

0

0

0

2

3

2

9

3

2

It appears that the minimum value for x7 is 3/2 and so there will be no solution

to these inequalities for which all the variables are nonnegative.

54

Exercises

(d)

x1 x2 + x3 2

x1 + 2x2 4

3x1 + 2x3 7

The augmented matrix is

1

1

3

1 1 1

2 0 0

0 2 0

0 0

1 0

0 1

2

4

7

Lets add in an artificial variable and set things up to minimize this artificial

variable.

1 1 1 1 0 0 0 0 2

1 2 0 0 1 0 1 0 4

3 0 2 0 0 1 0 0 7

0 0 0 0 0 0 1 1 0

Then

1 1 1

1 2 0

3 0 2

1 2 0

3

2

3

0

0

2

0

0

1

0

2

0

1 0 0

0 1 0

0 0 1

0 1 0

1

0

0

0

21

1

0

0

0

0

1

0

0

0

0

1

2

4

7

4

0

1 0

0 0

1 1

4

4

7

0

0

1

0

0

1

2

It appears the minimum value of x7 is 0 and so this will mean there is a solution

when x2 = 2, x3 = 0, x1 = 0.

(e)

6x1 3x2 + 5x3 2

5x1 2x2 + 4x3 5

The augmented matrix is

5

6

5

5

6

5

0

2 4 1

3 5 0

2 4 0

0 0

1 0

0 1

1

2

5

2 4 1 0 0 0 0 1

3 5 0 1 0 1 0 2

2 4 0 0 1 0 0 5

0 0 0 0 0 1 1 0

5 2 4 1 0 0

6 3 5 0 1 0

5 2 4 0 0 1

6 3 5 0 1 0

0

1

0

0

0

0

0

1

1

2

5

2

55

Exercises

work with the first column.

5 2

0 3

5

0 0

0 53

4

1

5

0

1

5

5

2

1 1

2

4

0

0

41 21

1

56

1

56

4

0

0

0

0

1

0

1

1

45

1

45

0

0

1

0

0

1

0

1

0

0

1

0

0

1

0

0

0

1

0

0

0

0

0

1

0

0

0

1

1

4

5

4

4

5

1

3

4

4

3

4

minimum value of x7 is not 0.

5. Minimize z = x1 + x2 subject to x1 + x2 2, x1 + 3x2 20, x1 + x2 18. Change

to a maximization problem and solve as follows: Let yi = M xi . Formulate in terms

of y1 , y2 .

You could find the maximum of 2M x1 x2 for the given constraints and this would

happen when x1 + x2 is as small as possible. Thus you would maximize y1 + y2 subject

to the constraints

M y1 + M y2

M y1 + 3 (M y2 )

M y1 + M y2

20

18

2M 2

y1 + y2

4M 20 y1 + 3y2

2M 18 y1 + y2

You could simply regard M as large enough that yi 0 and use the techniques just

developed. The augmented matrix for the constraints is then

1 1 1 0

0

2M 2

1 3 0 1 0 4M 20

1 1 0 0 1 2M 18

Here M is large. Use the 3 as a pivot to zero out above and below it.

2

3

1

2

3

0

3

0

row and pick the 2/3 as a pivot.

0 0

0 3

2

0

3

1 13

0

0 1 0

0 13 1

+ 14

3

4M 20

2

34

3M 3

2

3M

1 0

0 23

1

0

3

1

3

2

16

3M 3

2

34

3M 3

56

Exercises

Now it appears that the basic solution is feasible provided M

the simplex tableau.

0

0 1 0

1 0

16

3

3

0

3

0

0

3M

3

2

2

2

1

2

34

0

0

1

0

M

3

3

3

3

1 1 0 0

0 1

0

0 0

0 3

2

0

3

0 0

1 0

0 23

0 31

0 0

0

3

0

2

1 0

1 1

16

3M 3

2

34

3M 3

2M 18

There is a negative number to the left of the 1 on the bottom row and we want to

maximize so work with this column. Assume M is very large. Then the pivot should

be the top entry in this column.

0 0 1

0 1 0

16

0 3 3 3 0 0 3M 27

2

2

2

0 1

0 0 23 M + 14

3

3

3

0 0 1

0 0 1 2M 2

It follows that the maximum of y1 +y2 is 2M 2 and it happens when y1 = M +7, y2 =

M 9, y3 = 0. Thus the minimum of x1 + x2 is

M y1 + M y2 = M (M + 7) + M (M 9) = 2

F.23

Exercises

7.3

1. If A is the matrix of a linear transformation which rotates all vectors in R2 through

30 , explain why A cannot have any real eigenvalues.

Because the vectors which result are not parallel to the vector you begin with.

2. If A is an n n matrix and c is a nonzero constant, compare the eigenvalues of A and

cA.

c

3. If A is an invertible n n matrix, compare the eigenvalues of A and A1 . More

generally, for m an arbitrary integer, compare the eigenvalues of A and Am .

1 and m .

4. Let A, B be invertible n n matrices which commute. That is, AB = BA. Suppose

x is an eigenvector of B. Show that then Ax must also be an eigenvector for B.

Say Bx = x. Then

B (Ax) = A (Bx) = Ax = (Ax)

57

Exercises

larger than 1. Show that if is an eigenvalue of A then || equals either 0 or 1.

Let x be the eigenvector. Then Am x = m x,Am x = Ax = x and so

m =

Hence if 6= 0, then

m1 = 1

and so || = 1.

6. Show that if Ax = x and Ay = y, then whenever a, b are scalars,

A (ax + by) = (ax + by) .

Does this imply that ax + by is an eigenvector? Explain.

The formula is obvious from properties of matrix multiplications. However, this vector

might not be an eigenvector because it might equal 0 and

7. Find the eigenvalues and eigenvectors of

1

1

1

the matrix

1 7

0 4 .

1 5

Determine

whether

defective.

the matrix is

1 1 7

3

2

1 0 4 , eigenvectors:

1

1

1,

2. This is a defective matrix.

1 1 5

1

1

3 7 19

2 1 8 .

2 3 10

Determine

whether

3 7 19

1

2

3

2 1 8 , eigenvectors: 1 1, 2 2, 1 3

1

1

2 3 10

1

9. Find the eigenvalues and eigenvectors of the matrix

7 12 30

3 7 15 .

3 6 14

Determine

whetherthe matrix is defective.

7 12 30

5

2

2

3 7 15 , eigenvectors: 1 , 0 1, 1 2

3 6 14

0

1

1

has a 2 dimensional eigenspace.

58

Exercises

7

8

2

the matrix

2 0

1 0 .

4 6

defective.

7 2 0

21

0

8 1 0 , eigenvectors: 1 3, 0 6

2 4 6

1

1

3 2

0 5

0 2

matrix

1

1 .

4

defective.

0

3 2 1

1

1

0 5

1 , eigenvectors: 0 , 12 3, 1 6

0

0 2

4

1

1

This matrix is not defective.

6 8 23

4 5 16

3 4 12

Determine whether the matrix is defective.

5 2 5

12 3 10 .

12 4 11

Determine

5 2

12 3

12 4

defective.

5

5

31

6

10 , eigenvectors: 1 , 0 1

11

0

1

This matrix is defective. In this case, there is only one eigenvalue, 1 of multiplicity

3 but the dimension of the eigenspace is only 2.

14. Find the eigenvalues and eigenvectors of the matrix

20 9 18

6 5 6 .

30 14 27

Determine whether the matrix is defective.

20 9 18

6

5 6 , eigenvectors:

30 14 27

59

Exercises

1

9

2

13

1, 1 2, 3 3

13

1

1

1

3

4

1

4

Not defective.

1

26 17

4

4

4 .

9 18

9

Determine whether the matrix is defective.

3 1

11 3

8

0

matrix

2

9 .

6

3

3 1 2

4

11 3 9 , eigenvectors: 1 0 This one is defective.

4

8

0 6

1

2

1 2

11 2 9 .

8

0 7

Determine whether the matrix is defective.

3

2

1 2

4

11 2 9 , eigenvectors: 1 1

4

8

0 7

1

This is defective.

2 1 1

2 3 2 .

2 2 1

2 1 1

1

1

2

2 3 2 , eigenvectors: 1 , 0 1, 1 2

2 2 1

0

1

1

This is non defective.

4 2 2

0 2 2 .

2 0

2

60

Exercises

4 2 2

0 2 2 , eigenvectors:

2 0

2

i

i

1 4, i 2 2i, i 2 + 2i

1

1

1

9

6 3

0

6

0 .

3 6 9

1

9

6 3

2

1

0

6

0 , eigenvectors: 1 , 0 6, 0 12

1

1

0

3 6 9

This is nondefective.

4 2 2

21. Find the complex eigenvalues and eigenvectors of the matrix 0 2 2 . De2 0

2

termine whether the matrix is defective.

4 2 2

0 2 2 , eigenvectors:

2 0

2

i

i

1 4, i 2 2i, i 2 + 2i

1

1

1

4 2

0

22. Find the complex eigenvalues and eigenvectors of the matrix 2 4 0 .

2 2 2

Determine whether the matrix is defective.

4 2

0

2 4 0 , eigenvectors:

2 2 2

1

1

0

0 , 1 2, 1 6

1

0

1

This is not defective.

1

1 6

23. Find the complex eigenvalues and eigenvectors of the matrix 7 5 6 .

1 7

2

Determine whether the matrix is defective.

1

1 6

7 5 6 , eigenvectors:

1 7

2

i

i

1 6, i 2 6i, i 2 + 6i

1

1

1

61

Exercises

This is not defective.

4 2 0

24. Find the complex eigenvalues and eigenvectors of the matrix 2 4 0 . Deter2 2 6

mine whether the matrix is defective.

4 2 0

2 4 0 , eigenvectors:

2 2 6

1

0

1

0 6, i 4 2i, i 4 + 2i

1

1

1

This is not defective.

1

0

0

0

0

0

c

2

a 0

1 b

0 2

0 0

Find values of a, b, c for which the matrix is defective and values of a, b, c for which it

is nondefective.

First consider the eigenvalue = 1

0 a

0 0

0 0

0 0

0

b

1

0

1

b

0

0

0

c

1

0

0

0

0

Then you have ax2 = 0, bx3 = 0. If neither a nor b = 0 then = 1 would be a defective

eigenvalue and the matrix would be defective. If a = 0, then the dimension of the

eigenspace is clearly 2 and so the matrix would be nondefective. If b = 0 but a 6= 0,

then you would have a defective matrix because the eigenspace would have dimension

less than 2. If c 6= 0, then the matrix is defective. If c = 0 and a = 0, then it is non

defective. Basically, if a, c 6= 0, then the matrix is defective.

26. Here is a matrix.

a

0

0

0

1

c

where a, b, c are numbers. Show this is sometimes defective depending on the choice

of a, b, c. What is an easy case which will ensure it is not defective?

An easy case which will ensure that it is not defective is for a, b, c to be distinct. If

you have any repeats, then this will be defective.

27. Suppose A is an n n matrix consisting entirely of real entries but a + ib is a complex

eigenvalue having the eigenvector, x + iy. Here x and y are real vectors. Show that

then a ib is also an eigenvalue with the eigenvector, x iy. Hint: You should

remember that the conjugate of a product of complex numbers equals the product of

the conjugates. Here a + ib is a complex number whose conjugate equals a ib.

A (x + iy) = (a + ib) (x + iy) . Now just take complex conjugates of both sides.

62

Exercises

28. Recall an n n matrix is said to be symmetric if it has all real entries and if A = AT .

Show the eigenvalues of a real symmetric matrix are real and for each eigenvalue, it

has a real eigenvector.

Let be an eigenvalue which corresponds to x 6= 0. Then xT = xT AT . Then

xT x

= xT AT x

= xT A

x = xT x

= . We have A

which shows that

x =

x and Ax = x so

A (

x + x) = (

x + x)

Hence it has a real eigenvector.

29. Recall an n n matrix is said to be skew symmetric if it has all real entries and if

A = AT . Show that any nonzero eigenvalues must be of the form ib where i2 = 1.

In words, the eigenvalues are either 0 or pure imaginary.

Let A be skew symmetric. Then if x is an eigenvector for ,

xT x

= xT AT x

= xT A

x = xT x

and so = .

30. Is it possible for a nonzero matrix to have only 0 as an eigenvalue?

12 16

Sure. Try this one.

9

12

31. Show that the eigenvalues and eigenvectors of a real matrix occur in conjugate pairs.

This follows from the observation that if Ax = x, then Ax = x

32. Suppose A is an n n matrix having all real eigenvalues which are distinct. Show

there exists S such that S 1 AS = D, a diagonal matrix. If

1

0

..

D=

.

0

define eD by

and define

eD

e1

0

..

.

en

eA SeD S 1 .

Next show that if A is as just described, so is tA where t is a real number and the

eigenvalues of At are tk . If you differentiate a matrix of functions entry by entry so

that for the ij th entry of A (t) you get aij (t) where aij (t) is the ij th entry of A (t) ,

show

d At

e

= AeAt

dt

Next show det eAt 6= 0. This is called the matrix exponential. Note I have only

defined it for the case where the eigenvalues of A are real, but the same procedure will

work even for complex eigenvalues. All you have to do is to define what is meant by

63

Exercises

ea+ib .

of the form

t

e 1

0

1

..

S

S

.

0

etn

And so

d At

e

=

dt

= S

0

1

..

0

..

1

.

= AetA

0

..

1 et1

1

S

n etn

t1

0

e

n

0

1

S S

..

.

etn

et1

0

..

etn

eigenvalues are 31 , 1, and 12 listed according to multiplicity.

1

1

2

1

1 , 1 , 1 , 1 , 1 , 1

2

2

3

1

0

1

7

12

41

1

6

1

S

1

S

14

7

12

16

1

6

61

2

3

. The

5

31 31

3

7

1

1

The eigenvalues are 1, 2, and 1. What is the physical interpreta3

6

6

1

1

7

3

6

6

tion of the repeated eigenvalue?

There are apparently two directions in which there is no stretching.

1 1

2

2

2

1 , 0 1, 1 2

0

1

1

3 1 1

1 2 0 The eigenvalues are 1, 4, and 2.

1 0 2

64

Exercises

The eigenvalues are given and so there are oscillatory solutions of the form

1

1 (a cos (t) + b sin (t)) ,

1

0

1 c sin

2t + d cos

2t

1

2

1 (e cos (2t) + f sin (2t))

1

b1 , , bn

and the rows of A are

aT1 , , aTn .

Show the columns of AB are

Ab1 Abn

and the rows of AB are

aT1 B aTn B.

The ith entry of the j th column of AB is aTi bj . Therefore, these

columns are as

indicated. Also the ith row of AB will be aTi b1 aTi bn . Thus this row is

just aTi B.

37. Let M be an n n matrix. Then define the adjoint of M , denoted by M to be the

transpose of the conjugate of M. For example,

2

i

2 1i

=

.

1+i 3

i

3

A matrix M, is self adjoint if M = M. Show the eigenvalues of a self adjoint matrix

are all real.

2

|x| = x x = (x) x = (M x) x = x M x

= x M x = x x = |x|

Hence = .

38. Let M be an n n matrix and suppose x1 , , xn are n eigenvectors which form a

linearly independent set. Form the matrix S by making the columns these vectors.

Show that S 1 exists and that S 1 M S is a diagonal matrix (one having zeros everywhere except on the main diagonal) having the eigenvalues of M on the main diagonal.

When this can be done the matrix is said to be diagonalizable.

65

Exercises

Since the vectors are linearly independent, the matrix S has an inverse. Denoting this

inverse by

T

w1

..

1

S = .

wnT

wiT xj = ij .

Therefore,

w1T

S 1 M S = S 1 (M x1 , , M xn ) = ... (1 x1 , , n xn )

wnT

1

..

0

.

39. Show that a n n matrix M is diagonalizable if and only if Fn has a basis of eigenvectors. Hint: The first part is done in Problem 38. It only remains to show that if the

matrix can be diagonalized by some matrix S giving D = S 1 M S for D a diagonal

matrix, then it has a basis of eigenvectors. Try using the columns of the matrix S.

The formula says that

M S = SD

Letting xk denote the k th column of S, it follows from the way we multiply matrices

that

M xk = k xk

where k is the k th diagonal entry on D.

40. Let

and let

1

3

A=

0

B=

2

4

1

0 1

1 1

2 1

2

0

2

, A21 = 0 1 and A22 = (3) .

0

1 2 2

0 1

6 5

AB = 3 4 0 1 1 = 4 7

0 1 3

2 1

7 4

1

3

2

4

, A12 =

66

Exercises

Now consider doing it by block multiplication. This leads to

1 2

0 1

2

2

1

+

3 4

0

1 1

0 1

0 1

+3 2 1

1 1

6 5

6 5

4 7 = 4 7

=

7 4

7 4

41. Suppose A, B are n n matrices and is a nonzero eigenvalue of AB. Show that then

it is also an eigenvalue of BA. Hint: Use the definition of what it means for to be

an eigenvalue. That is,

ABx = x

where x 6= 0. Maybe you should multiply both sides by B.

If Bx = 0, then the left side of the given equation would be 0 and the right wouldnt

be.

42. Using the above problem show that if A, B are n n matrices, it is not possible that

AB BA = aI for any a 6= 0. Hint: First show that if A is a matrix, then the

eigenvalues of A aI are a where is an eigenvalue of A.

{1 , , n } are the eigenvalues of AB, then the eigenvalues of AB aI = BA are

{1 a, , n a} which cannot be the same set, contrary to the above problem.

0

1

C =

0

..

.

0

..

a0

a1

..

.

an1

matrix for the given polynomial.

I C is of the form

..

.

..

.

1

+ an1

( + an1 ) det

a0

a1

..

.

..

.

..

.

1

67

Exercises

0

1

1

+1 det

det

..

.

..

.

1

a0

a1

a2

..

.

an2

form

..

.

a0

a1

a2

..

.

..

.

1

+ (an2 )

n1 + (an2 ) n2 + + a1 + a0

and so, the sum of the two is of the form

n1 ( + an1 ) + n1 + (an2 ) n2 + + a1 + a0

= n + an1 n1 + an2 n2 + + a1 + a0

It is routine to verify that when n = 2 this determinant gives the right polynomial.

44. A discreet dynamical system is of the form

x (k + 1) = Ax (k) , x (0) = x0

where A is an n n matrix and x (k) is a vector in Rn . Show first that

x (k) = Ak x0

for all k 1. If A is nondefective so that it has a basis of eigenvectors, {v1 , , vn }

where

Avj = j vj

you can write the initial condition x0 in a unique way as a linear combination of these

eigenvectors. Thus

n

X

x0 =

aj vj

j=1

x (k) =

n

X

aj Ak vj =

j=1

n

X

aj kj vj

j=1

which gives a formula for x (k) , the solution of the dynamical system.

The first formula is obvious from induction. Thus

Ak x0 =

n

X

j=1

aj Ak vj =

n

X

aj kj vj

j=1

68

Exercises

suppose the characteristic polynomial of A is

det (I A) = n + an1 n1 + + a1 + a0

Explain why

An + an1 An1 + + a1 A + a0 I v = 0

If A is nondefective, give a very easy proof of the Cayley Hamilton theorem based on

this. Recall this theorem says A satisfies its characteristic equation,

An + an1 An1 + + a1 A + a0 I = 0.

Let v be any vector in a basis of eigenvectors of A. Since A is nondefective, such a

basis exists. Then

An + an1 An1 + + a1 A + a0 I v = n + an1 n1 + + a1 + a0 v = 0

holds for vectors in a basis, it holds for any v and so

An + an1 An1 + + a1 A + a0 I = 0

46. Suppose an n n nondefective matrix A has only 1 and 1 as eigenvalues. Find A12 .

P

Let x be some vector. Let x = j aj vj where vj is an eigenvector. Then

A12 x =

n

X

aj 12

j vj =

j=1

n

X

aj vj = x

j=1

47. Suppose the characteristic polynomial of an n n matrix A is 1 n . Find Amn where

m is an integer. Hint: Note first that A is nondefective. Why?

The eigenvalues are distinct because they are the nth roots of 1. Hence from the above

formula, if x is a given vector with

x=

n

X

aj vj

j=1

then

nm

nm

x=A

n

X

aj vj =

j=1

so Anm = I.

n

X

j=1

nm

aj A

vj =

n

X

aj vj = x

j=1

48. Sometimes sequences come in terms of a recursion formula. An example is the Fibonacci sequence.

x0 = 1 = x1 , xn+1 = xn + xn1

Show this can be considered as a discreet dynamical system as follows.

xn+1

1 1

xn

x1

1

=

,

=

xn

1 0

xn1

x0

1

69

Exercises

Now use the technique of Problem 44 to find a formula for xn .

What are the eigenvalues and eigenvectors of this matrix?

1 1

, eigenvectors:

1 0

1 1

1

1

1

1

1 1

2 2 5

2 5+ 2

5,

5+

1

1

2 2

2

2

Now also

1

1

5

2 10

1

2

1

2

5

1

1

5+

10

2

xn+1

xn

1

2

5+

1

1

2

1

1

n 1 1

1

1 1

1

2 2 5

5

1

2 10

2 2

n 1

1

1

1

1

1

2 5+ 2

+

5+

5+

1

10

2

2

2

In particular,

xn =

n

n

1

1 1

1

1

1

1

1

5 +

5+

5+

2 10

2 2

10

2

2

2

det (I A) = n + an1 n1 + + a1 + a0

n

The characteristic polynomial equals det (I A) . To get the constant term, you plug

in = 0 and obtain det (A) = (1)n det (A).

F.24

Exercises

7.10

1. Explain why it is typically impossible to compute the upper triangular matrix whose

existence is guaranteed by Schurs theorem.

To get it, you must be able to get the eigenvalues and this is typically not possible.

2. Now recall the QR factorization of Theorem 5.7.5 on Page 133. The QR algorithm

is a technique which does compute the upper triangular matrix in Schurs theorem.

There is much more to the QR algorithm than will be presented here. In fact, what

I am about to show you is not the way it is done in practice. One first obtains what

is called a Hessenburg matrix for which the algorithm will work better. However,

the idea is as follows. Start with A an n n matrix having real eigenvalues. Form

A = QR where Q is orthogonal and R is upper triangular. (Right triangular.) This

70

Exercises

can be done using the technique of Theorem 5.7.5 using Householder matrices. Next

take A1 RQ. Show that A = QA1 QT . In other words these two matrices, A, A1 are

similar. Explain why they have the same eigenvalues. Continue by letting A1 play the

role of A. Thus the algorithm is of the form An = QRn and An+1 = Rn+1 Q. Explain

why A = Qn An QTn for some Qn orthogonal. Thus An is a sequence of matrices each

similar to A. The remarkable thing is that often these matrices converge to an upper

triangular matrix T and A = QT QT for some orthogonal matrix, the limit of the Qn

where the limit means the entries converge. Then the process computes the upper

triangular Schur form of the matrix A. Thus the eigenvalues of A appear on the

diagonal of T. You will see approximately what these are as the process continues.

Suppose you have found An . Then An = Qn Rn and then

An+1 Rn Qn = QTn An Qn = QTn QTn1 An1 Qn1 Qn

Now the product of orthogonal matrices is orthogonal and so this shows by induction

that each An is orthogonally similar to A.

3. Try the QR algorithm on

1 2

6

6

which has eigenvalues 3 and 2. I suggest you use a computer algebra system to do the

computations.

1

6

1 2

37

37 37

37

37 38

37

37

=

6

6

1

6

6

37 37

0

37 37

37 37

191

1

38

6

37 37 37 37

260

37 37 37

37

37

A1 =

=

36

6

6

6

1

37

0

37

37

37

37 37

37 37

191

260

37

37

=

36

6

37

37

191

1

36

1348

1021

37

1021

37 777 371021 37 777 37

37 37 1021 37 777

36

191

6

0

37 777 37 1021

37 777 37 1021

1021 37 1021

3959

7952

1021

1021

A2 =

216

1146

1021

1021

3959

7952

1021

1021

=

0.211 56 1146

1021

0.998 51

5. 447 9 102

3. 883 3 7. 715 7

5. 447 9 102

0.998 51

0

1. 545 1

3. 883 3 7. 715 7

0.998 51

5. 447 9 102

A3 =

=

0

1. 545 1

5. 447 9 102

0.998 51

3. 457 2

7. 915 8

8. 417 6 102 1. 542 8

You can keep on going. You see that it appears to be converging to an upper triangular

matrix having 3, 1 down the diagonal.

0

2

1

0

Show that the algorithm cannot converge for this example. Hint: Try a few iterations

of the algorithm.

Exercises

71

1

0 1

2 0

=

0

1 0

0 1

2 0

0 1

0 2

A1 =

=

0 1

1 0

1 0

0 2

0 1

1 0

=

1 0

1 0

0 2

1 0

0 1

0 1

A2 =

=

. Now it is back to where you started.

0 2

1 0

2 0

0 1

0 2

Thus the algorithm merely bounces between the two matrices

and

2 0

1 0

and so it cant possibly converge.

0 1

0 2

5. Show the two matrices A

and B

are similar; that is

4 0

2 0

there exists a matrix S such that A = S 1 BS but there is no orthogonal matrix

Q such that QT BQ = A. Show the QR algorithm does converge for the matrix B

although it fails to do so for A.

0

2

Both matrices have distinct eigenvalues 2 and so they are both similar to a diagonal

matrix having these eigenvalues on the diagonal. Therefore, the matrices are indeed

similar. Lets try the QR algorithm on the second.

0 2

0 1

2 0

=

2 0

1 0

0 2

2 0

0 1

0 2

A1 =

=

and so the algorithm keeps on returning

0 2

1 0

2 0

the same matrix in the case of B. Now consider the matrix A.

0 1

0 1

4 0

=

4 0

1 0

0 1

4 0

0 1

0 4

A1 =

=

0 1

1 0

1 0

0 4

0 1

1 0

=

1 0

1 0

0 4

1 0

0 1

0 1

A2 =

=

0 4

1 0

4 0

0 1

0 4

In this case, the algorithm bounces between

and

.

4 0

1 0

6. Let F be an m n matrix. Show that F F has all real eigenvalues and furthermore,

they are all nonnegative.

The eigenvalues are real because the matrix F F is Hermitian. If is one of them

with eigenvector x,

(x, x) = (F F x, x) = (F x,F x) 0

7. If A is a real n n matrix and is a complex eigenvalue of A having eigenvector

z + iw, show that w 6= 0.

A (z + iw) = (a + ib) (z + iw)

72

Exercises

Now subtract to obtain A (2iw) = 2iaw + 2ibz. Now if w = 0, then you would have

0 =ibz and this requires b = 0 since otherwise, the eigenvector was equal to 0 which

it isnt. Hence was not complex after all.

8. Suppose A = QT DQ where Q is an orthogonal matrix and all the matrices are real.

Also D is a diagonal matrix. Show that A must be symmetric.

AT = QT DT Q = QT DQ = A.

9. Suppose A is an n n matrix and there exists a unitary matrix U such that

A = U DU

where D is a diagonal matrix. Explain why A must be normal.

A A = U D U U DU = U D DU

AA = U DU U D U = U DD U. But D D = DD and both are equal to the

diagonal matrix which has the squares of the absolute values of the diagonal entries

of D down the diagonal.

A = U DU where D

is a real diagonal matrix. Therefore,

det (A) = det U 1 det (U ) det (D) = det (D) R.

11. Show that every unitary matrix preserves distance. That is, if U is unitary,

|U x| = |x| .

|U x|2 = (U x, U x) = (U U x, x) = (x, x) = |x|2 .

12. Show that if a matrix does preserve distances, then it must be unitary.

Let U preserve distances. Then

2

2

|U (x + y)| = (U x + U y,U x + U y)

2

= |U x| + |U y| + 2 Re (U x, U y)

= |x|2 + |y|2 + 2 Re (U x, U y)

Therefore,

Re (U x, U y) Re (x, y)

Re (U U x x, y)

= 0

= 0

y. Then you get

Re U U x x,

y = Re (U U x x, y) = |(U U x x, y)| = 0

and since y is arbitrary,

(U U x x, y) = 0

73

Exercises

13. Show that a complex normal matrix A is unitary if and only if its eigenvalues have

magnitude equal to 1.

Since A is normal, it follows that there is a unitary matrix U such that

AU = U D

where D is a diagonal matrix. Then from the above problems, A is unitary if and only

if it preserves distances. Let the columns of U be the orthonormal set {u1 , , un } .

Then we have Auk = k uk . For A to preserve distances, you must have |k | = 1.

Conversely, if |k | = 1 for all k, then {k uk } is also an orthonormal set and so A

preserves distances and is therefore normal.

14. Suppose A is an n n matrix which is diagonally dominant. Recall this means

X

|aij | < |aii |

j6=i

This follows from Gerschgorins theorem. The condition implies that 0 is not an

eigenvalue and so the matrix is one to one and hence invertible.

15. Give some disks in the complex plane whose union contains all the eigenvalues of the

matrix

1 + 2i 4 2

0

i 3

5

6 7

B (1 + 2i, 6) , B (i, 3) , B (7, 11)

16. Show a square matrix is invertible if and only if it has no zero eigenvalues.

To say the matrix has no zero eigenvalues is to say that it is one to one because it

maps no nonzero vector to 0.

17. Using Schurs theorem, show the trace of an n n matrix equals the sum of the

eigenvalues and the determinant of an n n matrix is the product of the eigenvalues.

Let A be an n n matrix. By Schur, there exists a unitary matrix U and an upper

triangular matrix T such that

A = U T U

Then det (A) = det (T ) = the product of the diagonal entries of T. These are

eigenvalues of T and both A and T have the same eigenvalues because they have

same characteristic polynomial. Thus the determinant of A equals the product of

eigenvalues. Since A and T are similar, they have the same trace also. However,

trace of T is just the sum of the eigenvalues of A.

the

the

the

the

nP

n matrix having eigenvalues

P complex

2

n

2

{i } listed according to multiplicity, then i,j |Aij | i=1 |i | . Show that equality

holds if and only if A is normal.

Then

X

|Aij |2 = trace (AA ) = trace (U T U U T U )

i,j

74

Exercises

= trace (U T T U ) = trace (T T )

n

X

i=1

|i |

Equality holds if and only if there are no nonzero off diagonal terms. This happens if

and only if T is a diagonal matrix. But if this is so, then A must be normal because

U AU = D, a diagonal matrix and this implies that A is normal.

19. Here is a matrix.

1234

6

5

3

0

654

9

123

98

123 10, 000 11

56

78

98

400

I know this matrix has an inverse before doing any computations. How do I know?

Gerschgorins theorem shows that there are no zero eigenvalues and so the matrix is

invertible.

20. Show the critical points of the following function are

1

(0, 3, 0) , (2, 3, 0) , and 1, 3,

3

f (x, y, z) = 23 x4 + 6x3 6x2 + zx2 2zx 2y 2 12y 18 32 z 2 .

0= 23 x4 + 6x3 6x2 + zx2 2zx 2y 2 12y 18 32 z 2 .

2xz 2z 12x + 18x2 6x3 = 0

4y 12 = 0

x2 2x 3z = 0

to these systems of equations.

18x2 + 36x + 2z 12 0 2x 2

0

4

0

Hessian is

2x 2

0

3

(0, 3, 0) ,

12 0

0

4

2

0

2

0

3

2

0

3

The eigenvalues are all negative and so this point is a local maximum.

(2, 3, 0)

12 0

0

4

2

0

1, 3, 31

16

0

0

3

0 4 0

0

0 3

75

Exercises

21. Here is a function of three variables.

f (x, y, z) = 13x2 + 2xy + 8xz + 13y 2 + 8yz + 10z 2

change the variables so that in the new variables there are no mixed terms, terms

involving xy, yz etc. Two eigenvalues are 12 and 18.

The function is

x

13 1 4

x

f (x, y, z) = y 1 13 4 y

z

4 4 10

z

13 1 4

1 13 4 , eigenvectors:

10

4 1 4

2

1

1

1 6, 1 12, 1 18

2

1

0

1

Then in terms of the new variables, the quadratic form is 6x2 + 12y 2 + 18z 2. Note

how I only needed to find the eigenvalues. This changes in the next problems.

f (x, y, z) = 2x2 4x + 2 + 9yx 9y 3zx + 3z + 5y 2 9zy 7z 2

change the variables so that in the new variables there are no mixed terms, terms

involving xy, yz etc.

The function is f (x, y, z) =

T

x

x

2

9/2 3/2

x

y 9/2

5

9/2 y + 4 9 3 y + 2

z

z

3/2 9/2 7

z

2

9/2 3/2

9/2

5

9/2 , eigenvectors:

7

3/2 9/2

0

2

3 1, 1 17 , 3 19

2

2

1

1

1

35

0

71 14

7

3

1

3

35 10

10 14

14

1

3

1

35

10

14

35

10

14

1

0

0

1

0

0

, A = Q 0 17 0 QT Thus you need to have

QT AQ = 0 17

0

2

2

19

19

0

0

0

0

2

2

the new variables be

1

T

35

0

71 14

x

x

7

3 35 1 10 3 14 y = y

35

10

14

1

3

1

z

z

35 35

10 10

14 14

Then in terms of the new variables, the quadratic form is

76

Exercises

T

1

0

0

x

x

y 0 17 0 y +

2

19

z

0

0

z

2

1

35

0

17 14

x

7

3

1

3

y + 2

4 9 3 35

35

10

14

10

14

1

3

1

z

35 35

10 10

14 14

19 2

19

2

=(x )2 + 72 5 7x 17

2 7z + 2

2 (y ) + 2 (z ) + 7

f (x, y, z) = x2 + 2xy + 2xz y 2 + 2yz z 2 + x

change the variables so that in the new variables there are no mixed terms, terms

involving xy, yz etc.

The function is

T

x

1

y 1

z

1

1 1

1

1 1 1

1

1 1

1

1

x

x

1 1 y + 1 0 0 y

1 1

z

z

1

1

1

, eigenvectors: 1 1, 1 , 1 2

0

2

1

1

1

1

3 3 2 2 6 6

Appropriate orthogonal matrix: Q = 31 3 21 2 16 6 .

1

1

0

3 3

3 6

New variables,

x

y =

z

1

3 3

1

3 3

1

3 3

12 2

1

2 2

0

61 6

x

61 6 y

1

z

3 6

T

x

1 0

0

x

y 0 2 0 y

z

0 0 2

z

1

1

x

3 3 2 2 6 6

1

+ 1 0 0 13 3

61 6 y

2 2

1

1

z

0

3 3

3 6

24. Show the critical points of the function,

are points of the form,

(x, y, z) = t, 2t2 + 6t, t2 3t

for t R and classify them as local minima, local maxima or saddle points.

77

Exercises

2yx2 6yx 4zx2 12zx + y 2 + 2yz = 2x2 6x + 2y + 2z

4x2 12x + 2y

2x2 6x + 2y + 2z = 0 Let x = t. Then y = 2t2 + 6t. Now place these into the

4x2 12x + 2y = 0

middle equation and you then find that 2t2 6t + 2 2t2 + 6t + 2z = 0, Solution is:

z = t2 3t. This also works in the top equation.

6 2t2 + 6t 12 t2 3t 4t 2t2 + 6t 8t t2 3t = 0

4y 8z 4x 6 8x 12

4x 6

2

2

8x 12

2

0

Plug in the critical points.

0

4t 6 8t 12

4t 6

You need to consider the sign of the eigenvalues.

2

2

8t 12

2

0

4t 6 8t 12

= 80t2 + 240t 3 + 22 + 184. Thus you

2

2

det 4t 6

8t 12

2

3 22 80t2 240t 184 = 0

2 2 184 + 80t2 + 240t = 0

184 + 80t2 + 240t has a minimum value when t = 23 and at this point, the value of

2

this function is 184 + 80 23 + 240 32 = 4. Therefore, from the quadratic formula,

there is a positive eigenvalue and a negative eigenvalue for any value of t. Thus this

function always has a direction in which it appears to have a local min. and a direction

in which it appears to have a local max. for each of the critical points.

1 4

1

x 4x3 + 8x2 3zx2 + 12zx + 2y 2 + 4y + 2 + z 2 .

2

2

are (0, 1, 0) , (4, 1, 0) , and (2, 1, 12) and classify them as local minima, local

maxima or saddle points.

6x2 24x 6z + 16 0 12 6x

0

4

0

Hessian:

12 6x

0

1

f (x, y, z) =

(0, 1, 0)

16 0

0 4

12 0

12

0

1

This is a saddle point because it has a negative and two positive eigenvalues.

(4, 1, 0)

22 0

0

4

12 0

12

0

1

78

Exercises

This is also a saddle point because it has a negative and two positive eigenvalues.

(2, 1, 12)

64 0 0

0 4 0

0 0 1

26. Let f (x, y) = 3x4 24x2 + 48 yx2 + 4y. Find and classify the critical points using

the second derivative test.

12x3 2xy 48x

4

2

2

3x 24x + 48 yx + 4y =

4 x2

Critical points: (2, 0) , (2, 0) .

4

(2, 0)

36x2 2y 48 2x

2x

0

96 4

4 0

Clearly it has eigenvalues of different sign because the determinant is negative. Therefore, this is a saddle point. (2, 0) works out the same way. To have some fun, graph

this function of two variables and see if you can see the critical points are saddle points

from the picture.

27. Let f (x, y) = 3x4 5x2 + 2 y 2 x2 + y 2 . Find and classify the critical points using the

second derivative test.

12x3 2xy 2 10x

4

2

2 2

2

3x 5x + 2 y x + y =

2y 2x2 y

Critical points: (1, 1) , (1, 1) , (1, 1) , (1, 1) , (0, 0) , 61 5 6, 0 , 16 5 6, 0

36x2 2y 2 10 4xy

3x4 5x2 + 2 y 2 x2 + y 2 , Hessian is

4xy

2 2x2

(1, 1)

24 4

4 0

(1, 1) , saddle point also. (1, 1) saddle point (1, 1) saddle point.

61 5 6, 0

20 0

0 13

Local minimum. 61 5 6, 0 also a local minimum.

28. Let f (x, y) = 5x4 7x2 2 3y 2 x2 + 11y 2 4y 4 . Find and classify the critical points

using the second derivative test.

20x3 6xy 2 14x

5x4 7x2 2 3y 2 x2 + 11y 2 4y 4 =

6x2 y 16y 3 + 22y

1

1

7 10, 0 , 10

7 10, 0

Critical points (1, 1) , (1, 1) , (0, 0) , 10

5x4 7x2 2 3y 2 x2 + 11y 2 4y 4 ,

79

Exercises

Hessian is

(1, 1)

60x2 6y 2 14

12xy

12xy

6x2 48y 2 + 22

40 12

12 32

(1, 1)

40 12

12 32

(0, 0)

1

10 7 10, 0

14 0

0

22

28 0

0 89

5

1

This is a local minimum. So is 10 7 10, 0 .

29. Let f (x, y, z) = 2x4 3yx2 + 3x2 + 5x2 z + 3y 2 6y + 3 3zy + 3z + z 2 . Find and

classify the critical points using the second derivative test.

2x4 3yx2 + 3x2 + 5x2 z + 3y 2 6y + 3 3zy + 3z + z 2

= 3x2 + 6y 3z 6

5x2 3y + 2z + 3

Critical points: (0, 1, 0)

6x

6

3 At the point of interest this is

Hessian is

10x

3

2

54 0

0

0

6 3

0

3 2

It has two positive eigenvalues and one negative eigenvalue so this is a saddle point.

30. Let f (x, y, z) = 3yx2 3x2 x2 z y 2 + 2y 1 + 3zy 3z 3z 2 . Find and classify

the critical points using the second derivative test.

3yx2 3x2 x2 z y 2 + 2y 1 + 3zy 3z 3z 2

6xy 6x 2xz

= 3x2 2y + 3z + 2

x2 + 3y 6z 3

Critical points: (0, 1, 0)

80

Exercises

6y 2z 6

6x

Hessian is

2x

eigenvalues:

6x 2x

2

3

3

6

0

0

0

0

2

3

13 4, 13 4, 0

0

3

6

This has two directions in which it appears to have a local maximum. However, the

test fails because of the zero eigenvalue.

31. Let Q be orthogonal. Find the possible values of det (Q) .

1

32. Let U be unitary. Find the possible values of det (U ) .

U is unitary which means it preserves length. Therefore, if U x = x, it must be the

case that || = 1. Hence = ei for some . In other words, is a point on the unit

circle.

33. If a matrix is nonzero can it have only zero for eigenvalues?

1 1

Definitely yes. Consider this one.

1 1

34. A matrix A is called nilpotent if Ak = 0 for some positive integer k. Suppose A is a

nilpotent matrix. Show it has only 0 for an eigenvalue.

Say is an eigenvalue. Then for some x 6= 0,

Ax = x

Then k is an eigenvalue for Ak and so k = 0. Hence = 0.

35. If A is a nonzero nilpotent matrix, show it must be defective.

If it is not defective, then there exists S such that

S 1 AS = 0

But then A = 0. Hence, if A 6= 0 is nilpotent, then A is defective.

36. Suppose A is a nondefective n n matrix and its eigenvalues are all either 0 or 1.

Show A2 = A. Could you say anything interesting if the eigenvalues were all either

0,1,or 1? By DeMoivres theorem, an nth root of unity is of the form

2k

2k

cos

+ i sin

n

n

Could you generalize the sort of thing just described to get An = A? Hint: Since A

is nondefective, there exists S such that S 1 AS = D where D is a diagonal matrix.

S 1 AS = D and so S 1 A2 S = D2 = D = S 1 AS and so, in the first case, A2 = A.

If the eigenvaluse are 0,1,1, then all even powers of A are equal and all odd powers

are equal.

In the last case, you could have the eigenvalues of A equal to cos 2k

+ i sin 2k

.

n

n

Then

S 1 An S = I

and so An = I.

81

Exercises

37. This and the following problems will present most of a differential equations course.

To begin with, consider the scalar initial value problem

y = ay, y (t0 ) = y0

When a is real, show the unique solution to this problem is y = y0 ea(tt0 ) . Next

suppose

y = (a + ib) y, y (t0 ) = y0

(6.30)

where y (t) = u (t) + iv (t) . Show there exists a unique solution and it is

y (t) = y0 ea(tt0 ) (cos b (t t0 ) + i sin b (t t0 ))

e(a+ib)(tt0 ) y0 .

(6.31)

Next show that for a real or complex there exists a unique solution to the initial value

problem

y = ay + f, y (t0 ) = y0

and it is given by

y (t) = ea(tt0 ) y0 + eat

t0

Hint: For the first part write as y ay = 0 and multiply both sides by eat . Then

explain why you get

d at

e y (t) = 0, y (t0 ) = 0.

dt

Now you finish the argument. To show uniqueness in the second part, suppose

y = (a + ib) y, y (0) = 0

and verify this requires y (t) = 0. To do this, note

y = (a ib) y, y (0) = 0

and that

d

2

|y (t)|

dt

Thus from the first part |y (t)| = 0e2at = 0. Finally observe by a simple computation

that 6.30 is solved by 6.31. For the last part, write the equation as

y ay = f

and multiply both sides by eat and then integrate from t0 to t using the initial

condition.

38. Now consider A an n n matrix. By Schurs theorem there exists unitary Q such that

Q1 AQ = T

where T is upper triangular. Now consider the first order initial value problem

x = Ax, x (t0 ) = x0 .

82

Exercises

Show there exists a unique solution to this first order system. Hint: Let y = Q1 x

and so the system becomes

y = T y, y (t0 ) = Q1 x0

(6.32)

yn = tnn yn , yn (t0 ) = Q1 x0 n .

Then use the solution you get in this to get the solution to the initial value problem

which occurs one level up, namely

yn1

= t(n1)(n1) yn1 + t(n1)n yn , yn1 (t0 ) = Q1 x0 n1

Continue doing this to obtain a unique solution to 6.32.

39. Now suppose (t) is an n n matrix of the form

(t) =

x1 (t)

xn (t)

where

(6.33)

Explain why

(t) = A (t)

if and only if (t) is given in the form of 6.33. Also explain why if c Fn ,

y (t) (t) c

solves the equation

y (t) = Ay (t) .

This is because y (t) = (t) c = A (t) c = Ay (t) .

40. In the above problem, consider the question whether all solutions to

x = Ax

(6.34)

are obtained in the form (t) c for some choice of c Fn . In other words, is the

general solution to this equation (t) c for c Fn ? Prove the following theorem using

linear algebra.

Theorem F.24.1 Suppose (t) is an n n matrix which satisfies

(t) = A (t) .

1

exists for some t.

1

1

Furthermore, if (t) = A (t) , then either (t) exists for all t or (t) never

exists for any t.

(det ( (t)) is called the Wronskian and this theorem is sometimes called the Wronskian

alternative.)

Hint: Suppose first the general solution is of the form (t) c where c is an arbitrary

1

constant vector in Fn . You need to verify (t) exists for some t. In fact, show

83

Exercises

1

there exists c Fn such that there is no solution x to

c = (t0 ) x

By the existence part of Problem 38 there exists a solution to

x = Ax, x (t0 ) = c

but this cannot be in the form (t) c. Thus for every t, (t)1 exists. Next suppose

1

for some t0 , (t0 ) exists. Let z = Az and choose c such that

z (t0 ) = (t0 ) c

Then both z (t) , (t) c solve

x = Ax, x (t0 ) = z (t0 )

Apply uniqueness to conclude z = (t) c. Finally, consider that (t) c for c Fn

1

either is the general solution or it is not the general solution. If it is, then (t)

1

exists for all t. If it is not, then (t) cannot exist for any t from what was just

shown.

1

41. Let (t) = A (t) . Then (t) is called a fundamental matrix if (t)

t. Show there exists a unique solution to the equation

x = Ax + f , x (t0 ) = x0

(6.35)

1

x0 + (t)

(s)

f (s) ds

t0

Now these few problems have done virtually everything of significance in an entire undergraduate differential equations course, illustrating the superiority of linear algebra.

The above formula is called the variation of constants formula.

Hint: Uniquenss is easy. If x1 , x2 are two solutions then let u (t) = x1 (t) x2 (t) and

argue u = Au, u (t0 ) = 0. Then use Problem 38. To verify there exists a solution, you

could just differentiate the above formula using the fundamental theorem of calculus

and verify it works. Another way is to assume the solution in the form

x (t) = (t) c (t)

and find c (t) to make it all work out. This is called the method of variation of

parameters.

For the method of variation of parameters,

x (t) =

x (t) =

and so

c (t) =

Z

(s)

f (s) ds + k

t0

84

Exercises

Then the solution desired is

x (t) = (t) k + (t)

(s)

f (s) ds

t0

1

Z t

1

1

x (t) = (t) (t0 ) x0 + (t)

(s) f (s) ds

t0

Note that by approximating with Riemann sums and passing to a limit, it follows that

the constant (t) can be taken into the integral to write

Z t

1

x (t) = (t) (t0 ) x0 +

(t) (s)1 f (s) ds

t0

42. Show there exists a special such that (t) = A (t) , (0) = I, and suppose

1

(t) exists for all t. Show using uniqueness that

1

(t) = (t)

and that for all t, s R

(t + s) = (t) (s)

Explain why with this special , the solution to 6.35 can be written as

Z t

x (t) = (t t0 ) x0 +

(t s) f (s) ds.

t0

xj = Axj , xj (0) = ej .

Use uniqueness as required.

Just follow the hint. Such a (t) clearly exists. By the Wronskian alternative in one

1

of the above problems (t) exists for all t.

Consider A (t) (t) A = (t)

(t) =

=

=

A (t) (t) A

AA (t) A (t) A

A ( (t))

and also (0) = 0. By uniqueness given above, (t) = 0 for all t. Therefore,

( (t) (t))

and so

(t) (t) = C

a constant. However, when t = 0, C = I and so (t) (t) = I. Also fixing s,

(t) (t + s) (t) (s)

85

Exercises

(t)

= (t + s) (t) (s)

= A (t + s) A (t) (s)

= A (t)

and (0) = 0. Therefore, (t) = 0 for all t by uniqueness. It follows that for all t, s,

(t + s) = (t) (s)

Therefore, in the variation of constants formula, it reduces to

Z t

x (t) = (t t0 ) x0 +

(t s) f (s) ds

t0

One thing might not be clear. If (t) = 0, why is (t) a constant? This works for

scalar valued functions by a use of the mean value theorem. However, this is a matrix

valued function. Nevertheless this is true. You just consider the ij th entry which is

eTi (t) ej

Its derivative equals 0 and so it is constant.

43. You can see more on this problem and the next one in the latest version of Horn

and Johnson, [16]. Two n n matrices A, B are said to be congruent if there is an

invertible P such that

B = P AP

Let A be a Hermitian matrix. Thus it has all real eigenvalues. Let n+ be the number

of positive eigenvalues, n , the number of negative eigenvalues and n0 the number of

zero eigenvalues. For k a positive integer, let Ik denote the k k identity matrix and

Ok the k k zero matrix. Then the inertia matrix of A is the following block diagonal

n n matrix.

In+

In

On0

Show that A is congruent to its inertia matrix. Next show that congruence is an

equivalence relation. Finally, show that if two Hermitian matrices have the same

inertia matrix, then they must be congruent. Hint: First recall that there is a

unitary matrix, U such that

Dn+

Dn

U AU =

On0

having the positive eigenvalues of A, Dn being

matrix

defined similarly. Now let Dn denote the diagonal matrix which replaces each entry

of Dn with its absolute value. Consider the two diagonal matrices

1/2

Dn+

Dn 1/2

D = D =

In0

Now consider D U AU D.

86

Exercises

The hint gives it away. When you block multiply, the thing which results is the inertia

matrix. Congruance is obviously an equivalence relation. Consider the transitive law

if you have two Hermitian matrices which have the same inertia matrix, then since

congruance is an equivalence relation, it follows that the two matrices are congruant.

44. Show that if A, B are two congruent Hermitian matrices, then they have the same

inertia matrix. Hint: Let A = SBS where S is invertible. Show that A, B have the

same rank and this implies that they are each unitarily similar to a diagonal matrix

which has the same number of zero entries on the main diagonal. Therefore, letting

VA be the span of the eigenvectors associated with positive eigenvalues of A and VB

being defined similarly, it suffices to show that these have the same dimensions. Show

that (Ax, x) > 0 for all x VA . Next consider S VA . For x VA , explain why

1

(BS x,S x) =

S 1 A (S ) S x,S x

= S 1 Ax,S x = Ax, S 1 S x = (Ax, x) > 0

Next explain why this shows that S VA is a subspace of VB and so the dimension of VB

is at least as large as the dimension of VA . Hence there are at least as many positive

eigenvalues for B as there are for A. Switching A, B you can turn the inequality

around. Thus the two have the same inertia matrix.

The above manipulation is straight forward. Now an orthonormal basis exists for Fn

which consists of eigenvectors of B. Say you have {v1 , , vq , w1 , , wr , z1 , , zs }

where the vi correspond to the positive eigenvalues, the wi to the negative eigenvalues

and the zi to the eigenvalue 0. We know that B is positive on span (v1 , , vq ) . Could

it be positive on any larger subspace? Could it be positive on span (v1 , , vq , u)

where

u span (w1 , , wr , z1 , , zs )?

No, it couldnt because you could then consider (Bu, u) and it would end up being no

larger than 0. It follows that S VA VV = span (v1 , , vq ) and so the dimension

of S VA , which is the same as the dimension of VA which is the same as the number

of positive eigenvalues, counted according to multiplicity, is no larger than q. Now

reverse the argument letting A B. Hence the two have the same number of

positive eigenvalues. Since they have the same number of zero eigenvalues, this implies

they have the same inertia matrix.

45. Let A be an m n matrix. Then if you unraveled it, you could consider it as a vector

in Cnm . The Frobenius inner product on the vector space of m n matrices is defined

as

(A, B) trace (AB )

Show that this really does satisfy the axioms of an inner product space and that it

also amounts to nothing more than considering m n matrices as vectors in Cnm .

(A, B) =

Aij Bij =

i,j

(A, A) =

i,j

X

i,j

Aij Aij =

X

i,j

|Aij |

(aA + bB, C) = a (A, C) + b (B, C)

87

Exercises

and so this does satisfy the axioms of the inner product. Also, this inner product

coincides with the standard inner product on Cnm .

46. Consider the n n unitary matrices. Show that whenever U is such a matrix, it

follows that

|U |Cnn = n

Next explain why if {Uk } is any sequence of unitary matrices, there exists a subsequence {Ukm }

m=1 such that limm Ukm = U where U is unitary. Here the limit

takes place in the sense that the entries of Ukm converge to the corresponding entries

of U .

From the above and the fact that U is unitary,

2

2

are contained in a closed and bounded set, it follows that there is a subsequence which

converges to U . Why is U a unitary matrix? You have

I = Ukm Ukm U U

and so U is unitary.

47. Let A, B be two n n matrices. Denote by (A) the set of eigenvalues of A. Define

dist ( (A) , (B)) = max min {| | : (B)}

(A)

Explain why dist ( (A) , (B)) is small if and only if every eigenvalue of A is close

to some eigenvalue of B. Now prove the following theorem using the above problem

and Schurs theorem. This theorem says roughly that if A is close to B then the

eigenvalues of A are close to those of B in the sense that every eigenvalue of A is close

to an eigenvalue of B.

Theorem F.24.2 Suppose limk Ak = A. Then

lim dist ( (Ak ) , (A)) = 0

Uk Ak Uk = Tk ,

where Tk is upper triangular. I claim that there exists a subsequence {Tkm }

m=1 such

that

Tkm T

Where T is unitarily similar to A.

Using the above problem, there exists a subsequence km such that

lim Ukm = U

lim Tkm = lim Ukm Akm Ukm = U AU T

88

Exercises

Since Ukm Akm Ukm is upper triangular, it follows that so is T . This shows the claim.

Next suppose that the conclusion does not hold. Then there exists > 0 and a

subsequence, still called k such that

dist ( (Ak ) , (A))

It follows that for every choice of Uk such that Uk Ak Uk = Tk , a triangular matrix and

for every choice of unitary U such that U AU = T, a triangular matrix,

|Tk T |Cn2 dist ( (Ak ) , (A))

Recall from the proof of Schurs theorem, you change the order of the eigenvalues

on the diagonal by adjusting the unitary matrix U . No matter how you permute

the entries on the diagonals, the two are always far apart. Now pick Uk for each k,

such that Uk Ak Uk = Tk a triangular matrix. By the first part of the proof, there

exists a subsequence {km } such that Tkm T where T is unitarily similar to A. This

contradicts the above.

a b

48. Let A =

be a 2 2 matrix which is not a multiple of the identity. Show

c d

that A is similar to a 2 2 matrix which has one entry equal to 0. Hint: First note

that there exists a vector a such that Aa is not a multiple of a. Then consider

B=

Aa

1

a Aa

p

p

Let the vector be

such that this vector and A

are linearly independent.

q

q

Then by brute force you do the following computation.

1

cp + dq ap bq

a b

p ap + bq

q

p

c d

q cp + dq

D

1

0

(ad bc) bq 2 cp2 + apq dpq

cp2 bq 2 apq + dpq (a + d) bq 2 cp2 + apq dpq

D

and since the trace equals 0 the term in the bottom right is also equal to 0.

49. Let A be a complex n n matrix which has trace equal to 0. Show that A is similar

to a matrix which has all zeros on the main diagonal. Hint: Use Problem 30 on

Page 122 to argue that you can say that a given matrix is similar to one which has

the diagonal entries permuted in any order desired. Then use the above problem and

block multiplication to show that if the A has k nonzero entries, then it is similar to

a matrix which has k 1 nonzero entries. Finally, when A is similar to one which has

at most one nonzero entry, this one must also be zero because of the condition on the

trace.

Suppose A has k nonzero diagonal entries. First of all, we can assume k 2 because

the trace equals 0. Then from Problem 30 on Page 122 it is similar to a block matrix

of the form

P Q

R S

89

Exercises

where the diagonal entries are just re ordered and P is a 2 2 matrix which which is

not a multiple of the identity and has both diagonal entries nonzero. Otherwise, the

trace could not equal zero if all the diagonal entries were equal. Then from the above

problem, there exists L such that L1 P L has at least one zero on the diagonal.

1

1

L

0

P Q

L 0

L P L L1 Q

=

0

I

R S

0 I

RL

S

Thus this new matrix is similar to the one above and it has at least one fewer nonzero

entries on the main diagonal. Continue this process obtaining a sequence of similar

matrices, each having fewer nonzero diagonal entries than the preceding one till there

is at most one nonzero entry. It follows that this one must be zero also because the

trace equals 0.

50. An n n matrix X is a comutator if there are n n matrices A, B such that X =

AB BA. Show that the trace of any comutator is 0. Next show that if a complex

matrix X has trace equal to 0, then it is in fact a comutator. Hint: Use the above

problem to show that it suffices to consider X having all zero entries on the main

diagonal. Then define

1

0

(

Xij

2

ij if i 6= j

A=

, Bij =

.

..

0 if i = j

0

n

The first part is easy because the trace of a product is the same in either order.

Suppose then that X has zero trace. Then by the above problem, it is similar to a

matrix Y which has a zero diagonal. Suppose you can show that Y = AB BA. Then

if Y = S 1 XS, you could write

S 1 XS = AB BA,

X = S (AB BA) S 1 = SAS 1 SBS 1 SBS 1 SAS 1

Thus X will be a comutator. It suffices then to show that whenever a matrix has all

zero diagonal entries, it must be a comutator. Now use the definition of A, B in the

hint.

X

X

(AB BA)ij

Ais Bsj

Bir Arj

s

iBij Bij j =

iXij

Xij

j

= Xij

ij

ij

iBii Bii i = 0

Thus it yields Xij .

F.25

Exercises

8.4

90

Exercises

1

1

1

0

1. Let H denote span 2 , 4 , 3 , 1 . Find the dimension of H

0

0

1

1

and determine a basis.

1 1 1 0

1 0 0 1

2 4 3 1 , row echelon form: 0 1 0 0 . The first three vectors form

0 0 1 1

0 0 1 1

a basis and the dimension is 3.

2. Let M = u = (u1 , u2 , u3 , u4 ) R4 : u3 = u1 = 0 . Is M a subspace? Explain.

3. Let M = u = (u1 , u2 , u3 , u4 ) R4 : u3 u1 . Is M a subspace? Explain.

4. Let w R4 and let M = u = (u1 , u2 , u3 , u4 ) R4 : w u = 0 . Is M a subspace?

Explain.

5. Let M = u = (u1 , u2 , u3 , u4 ) R4 : ui 0 for each i = 1, 2, 3, 4 . Is M a subspace?

Explain.

NO. Multiply something by 1.

M = u = (u1 , u2 , u3 , u4 ) R4 : w u = 0 and w1 u = 0 .

Is M a subspace? Explain.

7. Let M = u = (u1 , u2 , u3 , u4 ) R4 : |u1 | 4 . Is M a subspace? Explain.

8. Let M = u = (u1 , u2 , u3 , u4 ) R4 : sin (u1 ) = 1 . Is M a subspace? Explain.

No. Let u1 = /2 and multiply by 2.

P

0 = i 0xi

2t + 3s

s t : s, t R .

t+s

Is this set of vectors a subspace of R3 ? If so, explain why, give a basis for the subspace

and find its dimension.

2

3

Yes. A basis is 1 , 1

1

1

91

Exercises

11. Consider the vectors of the form

2t + 3s + u

st

t+s

: s, t, u R .

Is this set of vectors a subspace of R4 ? If so, explain why, give a basis for the subspace

and find its dimension.

It is a subspace. It is spanned by

3

1

1

0

It will be

3 2

1 1

1 1

0 0

pendent.

a basis if it is a linearly

1

0

, row echelon form:

0

1

2

1

1 0

, ,

1 0

0

1

independent set.

1 0 0

0 1 0

2t + u + 1

t + 3u

t+s+v

: s, t, u, v R .

Is this set of vectors a subspace of R4 ? If so, explain why, give a basis for the subspace

and find its dimension.

Because of that 1 this is not a subspace.

13. Let V denote the set of functions defined on [0, 1]. Vector addition is defined as

(f + g) (x) f (x) + g (x) and scalar multiplication is defined as (f ) (x) (f (x)).

Verify V is a vector space. What is its dimension, finite or infinite?

Pick n points {x1 , , xn } . Then let ei (x) = 0 unless x = xi when it equals 1. Then

n

{ei }i=1 is linearly independent, this for any n.

14. Let V denote the set of polynomial functions defined on [0, 1]. Vector addition is

defined as (f + g) (x) f (x) + g (x) and scalar multiplication is defined as (f ) (x)

(f (x)). Verify V is a vector space. What is its dimension, finite or infinite?

This is clearly a subspace of the set of all functions

defined on

[0, 1] . Its dimension is

still infinite. For example, you can see that 1, x, x2 , , xn is linearly independent

for each n. You could use Theorem 8.2.9 to show this, for example.

15. Let V be the set of polynomials defined on R having degree no more than 4. Give a

basis for this vector space.

1, x, x2 , x3 , x4

92

Exercises

16. Let the vectors be of the form a + b 2 where a, b are rational numbers and let the

field of scalars be F = Q, the rational numbers. Show directly this is a vector space.

What is its dimension? What is a basis for this vector space?

A basis is 1, 2 . In fact it is actually a field.

1

2

a+b 2

= aab

2 2b2 . Since a, b are rational, it follows that the denominator is not

0. Thus every element has an inverse so it is actually a field. It is certainly a vector

space over the rational numbers.

17. Let V be a vector space with field of scalars F and suppose {v1 , , vn } is a basis for

V . Now let W also be a vector space with field of scalars F. Let L : {v1 , , vn }

W be a function such that Lvj = wj . Explain how L can be extended to a linear

transformation mapping V to W in a unique way.

Pn

Pn

L ( i=1 ci vi ) i=1 ci wi

18. If you have 5 vectors in F5 and the vectors are linearly independent, can it always be

concluded they span F5 ? Explain.

Yes. If not, you could add in a vector not in the span and obtain a linearly independent

set of 6 vectors which is not possible.

19. If you have 6 vectors in F5 , is it possible they are linearly independent? Explain.

No. There is a spanning set having 5 vectors and this would need to be as long as the

linearly independent set.

20. Suppose V, W are subspaces of Fn . Show V W defined to be all vectors which are in

both V and W is a subspace also.

If x, y V W then a + b V and is also in W because these are subspaces. Hence

V W is a subspace.

21. Suppose V and W both have dimension equal to 7 and they are subspaces of a vector

space of dimension 10. What are the possibilities for the dimension of V W ? Hint:

Remember that a linear independent set can be extended to form a basis.

See the next problem.

22. Suppose V has dimension p and W has dimension q and they are each contained in

a subspace, U which has dimension equal to n where n > max (p, q) . What are the

possibilities for the dimension of V W ? Hint: Remember that a linear independent

set can be extended to form a basis.

Let {x1 , , xk } be a basis for V W. Then there is a basis for V and W which are

respectively

{x1 , , xk , yk+1 , , yp } , {x1 , , xk , zk+1 , , zq }

It follows that you must have k + p k + q k n and so you must have

p+qnk

23. If b 6= 0, can the solution set of Ax = b be a plane through the origin? Explain.

No. It cant. It does not contain 0.

Exercises

93

24. Suppose a system of equations has fewer equations than variables and you have found

a solution to this system of equations. Is it possible that your solution is the only one?

Explain.

No. There must then be infinitely many solutions. If the system is Ax = b, then

there are infinitely many solutions to Ax = 0 and so the solutions to Ax = b are a

particular solution to Ax = b added to the solutions to Ax = 0 of which there are

infinitely many.

25. Suppose a system of linear equations has a 24 augmented matrix and the last column

is a pivot column. Could the system of linear equations be consistent? Explain.

No. This would lead to 0 = 1.

26. Suppose the coefficient matrix of a system of n equations with n variables has the

property that every column is a pivot column. Does it follow that the system of

equations must have a solution? If so, must the solution be unique? Explain.

Yes. It has a unique solution.

27. Suppose there is a unique solution to a system of linear equations. What must be true

of the pivot columns in the augmented matrix.

The last one must not be a pivot column and the ones to the left must each be pivot

columns.

28. State whether each of the following sets of data are possible for the matrix equation

Ax = b. If possible, describe the solution set. That is, tell whether there exists a

unique solution no solution or infinitely many solutions.

(a) A is a 5 6 matrix, rank (A) = 4 and rank (A|b) = 4. Hint: This says b is in

the span of four of the columns. Thus the columns are not independent.

Infinite solution set.

(b) A is a 3 4 matrix, rank (A) = 3 and rank (A|b) = 2.

This surely cant happen. If you add in another column, the rank does not get

smaller.

(c) A is a 4 2 matrix, rank (A) = 4 and rank (A|b) = 4. Hint: This says b is in

the span of the columns and the columns must be independent.

You cant have the rank equal 4 if you only have two columns.

(d) A is a 5 5 matrix, rank (A) = 4 and rank (A|b) = 5. Hint: This says b is not

in the span of the columns.

In this case, there is no solution to the system of equations represented by the

augmented matrix.

(e) A is a 4 2 matrix, rank (A) = 2 and rank (A|b) = 2.

In this case, there is a unique solution since the columns of A are independent.

29. Suppose A is an m n matrix in which m n. Suppose also that the rank of A equals

m. Show that A maps Fn onto Fm . Hint: The vectors e1 , , em occur as columns

in the row reduced echelon form for A.

This says that the columns of A have a subset of m vectors which are linearly independent. Therefore, this set of vectors is a basis for Fm . It follows that the span of

the columns is all of Fm . Thus A is onto.

94

Exercises

30. Suppose A is an m n matrix in which m n. Suppose also that the rank of A equals

n. Show that A is one to one. Hint: If not, there exists a vector, x such that Ax = 0,

and this implies at least one column of A is a linear combination of the others. Show

this would require the column rank to be less than n.

The columns are independent. Therefore, A is one to one.

31. Explain why an n n matrix A is both one to one and onto if and only if its rank is

n.

The rank is n is the same as saying the columns are independent which is the same as

saying A is one to one which is the same as saying the columns are a basis. Thus the

span of the columns of A is all of Fn and so A is onto. If A is onto, then the columns

must be linearly independent since otherwise the span of these columns would have

dimension less than n and so the dimension of Fn would be less than n.

32. If you have not done this problem already, here it is again. It is a very important

result. Suppose A is an m n matrix and B is an n p matrix. Show that

dim (ker (AB)) dim (ker (A)) + dim (ker (B)) .

Let {w1 , , wk } be a basis for B (Fp ) ker (A) and suppose {u1 , , ur } is a basis

for ker (B). Let Bzi = wi . Now suppose x ker (AB). Then Bx ker (A) B (Fp )

and so

k

k

X

X

Bx =

a i wi =

ai Bzi

i=1

so

i=1

k

X

i=1

Hence

x

k

X

ai zi ker (B)

ai zi =

i=1

r

X

b j uj

j=1

Note that a little more can be said. {u1 , , ur , z1 , , zk } is independent. Here is

why. Suppose

k

X

X

a i ui +

bj zj = 0

i

j=1

X

bj Bzj = 0

j

follows that in fact,

dim (ker (AB)) = k + r dim (ker (A)) + dim (ker (B))

and the remaining inequality is an equality if and only if B (Fp ) ker (A).

This is because B (Fp ) ker (A) ker (A) and these are equal exactly when they have

the same dimension.

95

Exercises

33. Recall that every positive integer can be factored into a product of primes in a unique

way. Show there must be infinitely many primes. Hint: Show that if you have any

finite set of primes and you multiply them and then add 1, the result cannot be

divisible by any of the primes in your finite set. This idea in the hint is due to Euclid

who lived about 300 B.C.

Consider p1 p2 pn + 1. Then if

p1 p2 pn + 1 = lpk

you would have

1

+m=l

pk

where m is an integer, the product of the other primes different than pk . This is a

contradiction. Hence there must be another prime to divide p1 p2 pn + 1 and so

there are infinitely many primes.

34. There are lots of fields. This will give an example of a finite field. Let Z denote the set

of integers. Thus Z = { , 3, 2, 1, 0, 1, 2, 3, }. Also let p be a prime number.

We will say that two integers, a, b are equivalent and write a b if a b is divisible

by p. Thus they are equivalent if a b = px for some integer x. First show that

a a. Next show that if a b then b a. Finally show that if a b and b c

then a c. For a an integer, denote by [a] the set of all integers which is equivalent

to a, the equivalence class of a. Show first that is suffices to consider only [a] for

a = 0, 1, 2, , p 1 and that for 0 a < b p 1, [a] 6= [b]. That is, [a] = [r] where

r {0, 1, 2, , p 1}. Thus there are exactly p of these equivalence classes. Hint:

Recall the Euclidean algorithm. For a > 0, a = mp + r where r < p. Next define the

following operations.

[a] + [b] [a + b]

[a] [b] [ab]

Show these operations are well defined. That is, if [a] = [a ] and [b] = [b ] , then

[a] + [b] = [a ] + [b ] with a similar conclusion holding for multiplication. Thus for

addition you need to verify [a + b] = [a + b ] and for multiplication you need to verify

[ab] = [a b ]. For example, if p = 5 you have [3] = [8] and [2] = [7] . Is [2 3] = [8 7]?

Is [2 + 3] = [8 + 7]? Clearly so in this example because when you subtract, the result

is divisible by 5. So why is this so in general? Now verify that {[0] , [1] , , [p 1]}

with these operations is a Field. This is called the integers modulo a prime and is

written Zp . Since there are infinitely many primes p, it follows there are infinitely

many of these finite fields. Hint: Most of the axioms are easy once you have shown

the operations are well defined. The only two which are tricky are the ones which

give the existence of the additive inverse and the multiplicative inverse. Of these, the

first is not hard. [x] = [x]. Since p is prime, there exist integers x, y such that

1 = px+ky and so 1ky = px which says 1 ky and so [1] = [ky] . Now you finish the

argument. What is the multiplicative identity in this collection of equivalence classes?

Of course you could now consider field extensions based on these fields.

The only substantive issue is why Zp is a field. Let [x] Zp where [x] 6= [0]. Thus x

is not a multiple of p. Then x and p are relatively prime. Hence from Theorem 1.9.3,

there exists a, b such that

1 = ap + bx

Then

[1 bx] = [ap] = 0

96

Exercises

and it follows that

[b] [x] = [1]

so [b] = [x]

0 1

0 0

0 0

0 1

1

=

0 0

1 0

1 0

0 0

0

0

1

Thus the identity is a comutator. Compare this with Problem 50 on Page 198.

1 0

1 0

This is easy. The left side equals

=

0 1

0 1

36. Suppose V is a vector space with field of scalars F. Let T L (V, W ) , the space of

linear transformations mapping V onto W where W is another vector space. Define

an equivalence relation on V as follows. v w means v w ker (T ) . Recall that

ker (T ) {v : T v = 0}. Show this is an equivalence relation. Now for [v] an equivalence class define T [v] T v. Show this is well defined. Also show that with the

operations

[v] + [w]

[v]

[v + w]

[v]

this set of equivalence classes, denoted by V / ker (T ) is a vector space. Show next that

T : V / ker (T ) W is one to one, linear, and onto. This new vector space, V / ker (T )

is called a quotient space. Show its dimension equals the difference between the

dimension of V and the dimension of ker (T ).

It is obviously an equivalence relation. Is the operation of addition well defined? If

[v] = [v ] , similar for w, w , is it the case that [v + w] = [v + w ]? Is

v + w (v + w ) ker (T )?

Of course so. It is just the sum of two things in ker (T ) . Similarly the operation of

scalar multiplication is well defined. Is T well defined? If v w, is it true that

T v =T w? Of course so.

T (v w) = 0

and so v ker (T ) . Hence [v] = [0] . Thus T is certainly one to one. If T is onto, it is

obvious that T is onto. Just pick w W . Then T v = w for some v. Then T [v] = w.

37. Let V be an n dimensional vector space and let W be a subspace. Generalize the

above problem to define and give properties of V /W . What is its dimension? What

is a basis?

Here you define an equivalence relation by v u if v u W. This is clearly

an equivalence relation. Define [v] + [u] [u + v] and [u] [u]. Then by a

repeat of the above, everything is well defined. It is clear that V /W , the set of

equivalence classes is a vector space with these operations as just defined. What is

its dimension? Let {w1 , , wr } be a basis for W . Now extend to get a basis for

V, {w1 , , wr , v1 , , vnr }. Then if

nr

X

ci [vi ] = [0] ,

i=1

97

Exercises

it follows that

Pnr

i=1

r

X

di wi +

i=1

nr

X

ci vi = 0

i=1

Now, since {w1 , , wr , v1 , , vnr } is a basis, it follows that all the di and ci are

equal to 0. Therefore, in particular the vectors [vi ] form a linearly independent set in

V /W. Do these vectors span V /W ? Suppose [v] V /W . Then

v=

r

X

di wi +

i=1

nr

X

ci vi

i=1

[v] =

nr

X

ci [vi ]

i=1

38. If F and G are two fields and F G, can you consider G as a vector space with field

of scalars F? Explain.

Yes. This is obvious.

39. Let A denote the algebraic numbers, those numbers which are roots of polynomials

having rational coefficients which are in R. Show A can be considered a vector space

with field of scalars Q. What is the dimension of this vector space, finite or infinite?

As in the above problem, since A is a field, it can be considered as a vector space

over Q. So what is the dimension? It is clearly infinite. To see this, consider xn 7.

By the rational

root theorem, this polynomial

is irreducible. Therefore, the minimal

polynomial for n 7 is xn 7 and so for = n 7, 1, , 2 , , n1 are linearly independent since otherwise, xn 7 would not be the minimal polynomial of and would

end up not being irreducible. Since n is arbitrary, this shows that the dimension of A

is infinite.

40. As mentioned, for distinct algebraic numbers i , the complex numbers {ei }ni=1 are

linearly independent over the field of scalars A where A denotes the algebraic numbers,

those which are roots of a polynomial having integer (rational) coefficients. What is

the dimension of the vector space C with field of scalars A, finite or infinite? If the

field of scalars were C instead of A, would this change? What if the field of scalars

were R?

How many distinct algebraic numbers can you get? Clearly as many as desired. Just

consider the nth roots of some complex number. If you have n distinct algebraic

n

numbers i , then from the Lindemann Weierstrass theorem mentioned above, {ei }i=1

is linearly independent. Of course if the field of scalars were C this would change

completely. Then the dimension of C over C is clearly 1.

41. Suppose F is a countable field and let A be the algebraic numbers, those numbers

which are roots of a polynomial having coefficients in F which are in G, some other

field containing F. Show A is also countable.

You consider each polynomial of degree n which has coefficients in the countable field

F. There are countably many of these. Each has finitely many roots in A. Therefore,

98

Exercises

there are countably many things in A which correspond to polynomials of degree n

The totality of A is obtained then from taking a countable union of countable sets,

and this is countable.

R (x) =

p (x)

, degree of p (x) < degree of denominator.

q1 (x) qm (x)

where the polynomials qi (x) are relatively prime and all the polynomials p (x) and

qi (x) have coefficients in a field of scalars F. Thus there exist polynomials ai (x)

having coefficients in F such that

1=

m

X

ai (x) qi (x)

i=1

Explain why

P

m

X

p (x) m

a (x) p (x)

i=1 ai (x) qi (x)

Qi

R (x) =

=

q1 (x) qm (x)

j6=i qj (x)

i=1

Pm

p (x) i=1 ai (x) qi (x)

p (x)

R (x) =

=

q1 (x) qm (x)

q1 (x) qm (x)

Then simplifying this yields

R (x) =

m

X

a (x) p (x)

Qi

j6=i qj (x)

i=1

Now continue doing this on each term in the above sum till finally you obtain an

expression of the form

m

X

bi (x)

i=1

qi (x)

Using the Euclidean algorithm for polynomials, explain why the above is of the form

M (x) +

m

X

ri (x)

i=1

qi (x)

where the degree of each ri (x) is less than the degree of qi (x) and M (x) is a polynomial.

You just note that

bi (x) = li (x) qi (x) + ri (x)

where the degree of ri (x) is less than the degree of qi (x) . Then replacing each bi (x)

by this expression, you obtain the above expression.

Now argue that M (x) = 0.

You have

M (x) +

m

X

ri (x)

i=1

qi (x)

p (x)

q1 (x) qm (x)

99

Exercises

Now multiply both sides by q1 (x) qm (x) . If M (x) is not zero, then you would have

the degree of the left is larger than the degree on the right in the equation which

results. Hence M (x) = 0.

From this explain why the usual partial fractions expansion of calculus must be true.

You can use the fact that every polynomial having real coefficients factors into a product of irreducible quadratic polynomials and linear polynomials having real coefficients.

This follows from the fundamental theorem of algebra in the appendix.

n

What if some qi (x) = ki (x) ? In this case, you can go further. You have the partial

fractions expansion described above and you have a term of the form

a (x)

n

n , degree of a (x) < degree k (x)

k (x)

n

X

ai (x)

i=1

k (x)

where the degree of ai (x) < degree of k (x). This can be proved by induction. If

n = 1, there is nothing to show. Suppose then it is true for n 1.

!

a (x)

1

a (x)

=

k (x)n

k (x) k (x)n1

n1

, then there is nothing to show.

Induction takes care of it. If this is not the case, you can write

!

!

n1

a (x)

1

a (x)

1

b (x) k (x)

+ r (x)

=

n =

n1

k (x) k (x)n1

k (x)

k (x)

k (x)

where the degree of r (x) is less than the degree of k (x)

n1

a (x)

b (x)

r (x)

1

+

n =

k (x) k (x)n1 k (x)

k (x)

The second term gives what is desired by induction. I claim that the degree of b (x)

n

must be less than the degree of k (x) . If not, you could multiply both sides by k (x)

and obtain

a (x) = b (x) k (x)n1 + r (x)

n

and the degree of the left is less than the degree of k (x) while the degree of the

n

right is greater than the degree of k (x) . Therefore, the desired decomposition has

been obtained. Now the usual procedure for partial fractions follows. If you have

a rational function p (x) /q (x) where the degree of q (x) is larger than the degree

of p (x) , there exist factors of q (x) into products of linear and irreducible over R

quadratic factors by using the fundamental theorem of algebra. Then you just apply

the above decomposition result.

43. Suppose {f1 , , fn } is an independent set of smooth functions defined on some interval (a, b). Now let A be an invertible n n matrix. Define new functions {g1 , , gn }

as follows.

g1

f1

..

.

. = A ..

gn

fn

100

Exercises

Is it the case that {g1 , , gn } is also independent? Explain why.

P

Suppose ni=1 ai gi = 0. Then

0=

X

i

ai

Aij fj =

It follows that

X

j

fj

Aij ai

Aij ai = 0

for each j. Therefore, since AT is invertible, it follows that each ai = 0. Hence the

functions gi are linearly independent.

F.26

Exercises

9.5

1. If A, B, and C are each n n matrices and ABC is invertible, why are each of A, B,

and C invertible?

This is because ABC is one to one.

2. Give an example of a 3 2 matrix with the property that the linear transformation

determined by this matrix is one to one but not onto.

1 0

0 1

0 0

4. Review problem: Suppose det (A I) = 0. Show using Theorem 3.1.15 there exists

x 6= 0 such that (A I) x = 0.

Since the matrix (A I) is not invertible, it follows that there exists x which it sends

to 0.

5. How does the minimal polynomial of an algebraic number relate to the minimal polynomial of a linear transformation? Can an algebraic number be thought of as a linear

transformation? How?

They are pretty much the same thing. If you have an algebraic number it can

be thought of as a linear transformation which maps A to A according to the rule

() 6= .

6. Recall the fact from algebra that if p () and q () are polynomials, then there exists

l () , a polynomial such that

q () = p () l () + r ()

where the degree of r () is less than the degree of p () or else r () = 0. With this in

mind, why must the minimal polynomial always divide the characteristic polynomial?

That is, why does there always exist a polynomial l () such that p () l () = q ()?

Can you give conditions which imply the minimal polynomial equals the characteristic

polynomial? Go ahead and use the Cayley Hamilton theorem.

101

Exercises

the above equation,

0 = r (A)

and this cannot happen unless r () = 0 because if not, p () was not the minimal

polynomial after all.

7. In the following examples, a linear transformation, T is given by specifying its action

on a basis . Find its matrix with respect to this basis.

1

1

1

1

1

(a) T

=2

+1

,T

=

2

2

1

1

1

1

1

The basis is

,

. Formally,

2

1

1

1

1

1

1

2

+1

,

=

,

A

2

1

1

2

1

1 1

1 1

=

A

5 1

2 1

1

1 1

1 1

2 0

A=

=

2 1

5 1

1 1

0

0

1

1

0

(b) T

=2

+1

,T

=

1

1

1

1

1

0

1

0

0

1

2

+1

,

=

,

A

1

1

1

1

1

2 1

A=

1 0

1

1

1

1

1

1

(c) T

=2

+1

,T

=1

0

2

0

2

0

2

1

1

1

1

1

1

2

+1

,1

=

,

A

2

0

0

2

0

2

1 1

A=

2 1

8. Let = {u1 , , un } be a basis for Fn and let T : Fn Fn be defined as follows.

!

n

n

X

X

T

a k uk =

a k b k uk

k=1

k=1

First show that T is a linear transformation. Next show that the matrix of T with

respect to this basis is [T ] =

b1

..

.

bn

Show that the above definition is equivalent to simply specifying T on the basis vectors

of by

T (uk ) = bk uk .

Formally,

(b1 u1 , b2 u2 , , bn un ) = (u1 , u2 , , un ) [T ]

102

Exercises

9. In the situation of the above problem, let = {e1 , , en } be the standard basis for

Fn where ek is the vector which has 1 in the k th entry and zeros elsewhere. Show that

[T ] =

1

u1 un [T ] u1 un

(6.36)

We have T ui = bi ui and so we know [T ] . The problem is to find [T ] where is the

good basis. Consider the diagram

Fn

Fn

Fn

[T ]

[T ]

Fn

Fn

Fn

1

u1 un

On the left side going down you get x x

x.

Going up on the

right, you need to have x u1 un x u1 un x. Then you need

[T ] =

u1

un

[T ]

u1

un

1

10. Generalize the above problem to the situation where T is given by specifying its

action on the vectors of a basis = {u1 , , un } as follows.

T uk =

n

X

ajk uj .

j=1

Letting A = (aij ) , verify that for = {e1 , , en } , 6.36 still holds and that [T ] = A.

This isnt really any different. As explained earlier, the matrix of T with respect to

the given basis is just A where it is as described. Thus A = [T ] and by a repeat of

the above argument,

1

[T ] = u1 un [T ] u1 un

11. Let P3 denote the set of real polynomials of degree no more than 3, defined on an

interval [a, b]. Show that P3 is a subspace of the vector space

of all functions defined

on this interval. Show that a basis for P3 is 1, x, x2 , x3 . Now let D denote the

differentiation operator which sends a function to its derivative. Show D is a linear

transformation which sends P3 to P3 . Find the matrix of this linear transformation

with respect to the given basis.

0, 1, 2x, 3x2 = 1, x, x2 , x3 A where A is the desired matrix. Then clearly

0 1 0 0

0 0 2 0

A=

0 0 0 3

0 0 0 0

12. Generalize the above problem to Pn , the space of polynomials of degree no more than

n with basis {1, x, , xn } .

The pattern seems pretty clear. A will be an (n + 1) (n + 1) matrix which has zeros

everywhere except on the super diagonal where you have, starting at the top and going

towards the bottom, the numbers in the following order. 1, 2, 3, , n.

103

Exercises

13. In the situation of the above problem, let the linear transformation be T = D2 + 1,

defined as T f = f + f. Find the matrix of this linear transformation with respect to

the given basis {1, x, , xn }.

First consider the operator D2 .

0 0 2 6x 12x2 =

x x2

0

0

A=

0

0

0

Then the matrix will be this one added

1 0

0 1

0 0

0 0

0 0

x3

x4

would have

0

0

0

0

0

2

0

0

0

0

0

6

0

0

0

0

0

12

0

0

2 0 0

0 6 0

1 0 12

0 1 0

0 0 1

14. In calculus, the following situation is encountered. There exists a vector valued function f :U Rm where U is an open subset of Rn . Such a function is said to have

a derivative or to be differentiable at x U if there exists a linear transformation

T : Rn Rm such that

lim

v0

|f (x + v) f (x) T v|

= 0.

|v|

First show that this linear transformation, if it exists, must be unique. Next show

that for = {e1 , , en } , , the standard basis, the k th column of [T ] is

f

(x) .

xk

Actually, the result of this problem is a well kept secret. People typically dont see

this in calculus. It is seen for the first time in advanced calculus if then.

Suppose that T also works. Then letting t > 0,

|f (x + tv) f (x) tT v|

|f (x + tv) f (x) tT v|

= lim

=0

t0

t0

t

t

lim

In particular,

t (T v T v)

=0

t0

t

which says that T v = T v and since v is arbitrary, this shows T = T .

lim

15. Recall that A is similar to B if there exists a matrix P such that A = P 1 BP. Show

that if A and B are similar, then they have the same determinant. Give an example

of two matrices which are not similar but have the same determinant.

These two are not similar but have the same determinant.

1 0

1 1

,

0 1

0 1

104

Exercises

You can see these are not similar by noticing that the second has an eigenspace of

dimension equal to 1 so it is not similar to any diagonal matrix which is what the first

one is.

16. Suppose A L (V, W ) where dim (V ) > dim (W ) . Show ker (A) 6= {0}. That is, show

there exist nonzero vectors v V such that Av = 0.

If not, then A would be one to one and so it would take a basis to a linearly independent

set which is impossible because W has smaller dimension than V .

17. A vector v is in the convex hull of a nonempty set S if there are finitely many vectors

of S, {v1 , , vm } and nonnegative scalars {t1 , , tm } such that

v=

m

X

tk vk ,

k=1

m

X

tk = 1.

k=1

Suppose now that S V,

Pm

a vector space of dimension n. Show that if v = k=1 tk vk is a vector in the convex

hull for m > n + 1, then there exist other scalars {tk } such that

v=

m1

X

tk vk .

k=1

Thus every vector in the convex hull of S can be obtained as a convex combination

of at most n + 1 points of S. This incredible result is in Rudin [23]. Hint: Consider

L : Rm V R defined by

!

m

m

X

X

L (a)

ak vk ,

ak

k=1

k=1

Explain why ker (L) 6= {0} . Next, letting a ker (L) \ {0} and R, note that

a ker (L) . Thus for all R,

v=

m

X

(tk + ak ) vk .

k=1

If any of the tk = 0, there is nothing to prove. Hence you can assume each of these

tk > 0. Suppose then that m > n + 1 and consider

!

m

m

X

X

L (a)

ak vk ,

ak

k=1

k=1

this mapping L cannot be one to one. Hence

Pmthere exists a 6= 0 such that La = 0.

Thus also a ker (L). By assumption, v = k=1 tk vk . Then for each R,

v=

m

X

(tk + ak ) vk

k=1

P

P

P

(tk + ak ) = 1 because

ak = 0. Also, for = 0,

because m

k=1 ak vk = 0. Now

each of the tk + ak > 0. Adjust so that each of these is still 0 but one vanishes.

Then you have gotten v as a convex combination of fewer than m vectors.

105

Exercises

18. For those who know about compactness, use Problem 17 to show that if S Rn and

S is compact, then so is its convex hull.

Suppose that the convex hull of S is not compact. Then nthere exists

o

Pn+1

v1k , v2k , , v(n+1)k k=1 , each in S and sk C s : i=1 si = 1, si [0, 1]

such that

(n+1

)

X

sik vik

i=1

k=1

denoted as k such that for each i, limk vik = vi . This is because S is given to be

compact. Also, taking yet another subsequence, you can assume that limk sik = si

because of the compactness of C. But this yields an obvious contradiction to the

assertion that the above has no convergent subsequence.

19. Suppose Ax = b has a solution. Explain why the solution is unique precisely when

Ax = 0 has only the trivial (zero) solution.

This is because the general solution is yp +y where Ayp = b and Ay = 0. Now A0 = 0

and so the solution is unique precisely when this is the only solution y to Ay = 0.

20. Let A be an n n matrix of elements of F. There are two cases. In the first case,

F contains a splitting field of pA () so that p () factors into a product of linear

polynomials having coefficients in F. It is the second case which of interest here where

pA () does not factor into linear factors having coefficients in F. Let G be a splitting

field of pA () and let qA () be the minimal polynomial of A with respect to the field

G. Explain why qA () must divide pA (). Now why must qA () factor completely

into linear factors?

In the field G

pA () = qA () l () + r ()

where r () either equals 0 or has degree less than the degree of qA (). Then it

follows that r (A) = 0 and so it must be the case that r () = 0. Hence qA ()

must divide pA (). Now G was the splitting field of pA () and this new polynomial

divides this one. Therefore, this new polynomial also must factor completely into

linear polynomials having coefficients in G.

pA () = qA () l ()

and all roots of pA () are in G so the same is true of the roots of qA () because these

are all roots of pA ().

21. In Lemma 9.2.2 verify that L is linear.

106

Exercises

L a

= L

n

X

i vi + b

i=1

n

X

n

X

(ai + b i ) vi

i=1

n

X

= a

i Lvi + b

i=1

n

X

i Lvi

i=1

n

X

i=1

F.27

(ai + b i ) Lvi

i=1

n

X

= aL

i vi

i=1

i vi

+ bL

n

X

i vi

i=1

Exercises

10.6

1. In the discussion of Nilpotent transformations, it was asserted that if two nn matrices

A, B are similar, then Ak is also similar to B k . Why is this so? If two matrices are

similar, why must they have the same rank?

Say A = S 1 BS. Then Ak = S 1 B k S and so these are also similar. You can write

S 1 BSS 1 BSS 1 BS S 1 BS

and note that the inside SS 1 cancels. Therefore, the result of the above is as claimed.

Hence Ak is similar to B k .

Now the dimension of A (Fn ) is the dimension of S 1 BS (Fn ) which is the same as the

dimension of S 1 B (Fn ) which is the same as the dimension of B (Fn ) because S, S 1

are one to one and onto. Hence they each take a basis to a basis.

2. If A, B are both invertible, then they are both row equivalent to the identity matrix.

Are they necessarily similar? Explain.

1 1

1 0

Obviously not. Consider

,

. These are both in Jordan form.

0 1

0 1

3. Suppose you have two nilpotent matrices A, B and Ak and B k both have the same

rank for all k. Does it follow that A, B are similar? What if it is not known that A, B

are nilpotent? Does it follow then?

Yes, this is so. You have A = S 1 JS and B = T 1 J T where J, J are both Jordan

canonical form. Then you have that J = SAS 1 and J = T BT 1 and so J k and J k

have the same rank for all k. Therefore, J = J because these matrices are nilpotent.

It follows that

SAS 1 = T BT 1

and consequently,

"

1

B = T 1 SAS 1 T = S 1 T

AS 1 T

so that A is similar to B. In case they are not nilpotent, the above problem gives a

counter example.

107

Exercises

4. When we say a polynomial equals zero, we mean that all the coefficients equal 0. If

we assign a different meaning to it which says that a polynomial

p () =

n

X

ak k = 0,

k=0

when the value of the polynomial equals zero whenever a particular value of F

is placed in the formula for p () , can the same conclusion be drawn? Is there any

difference in the two definitions for ordinary fields like Q? Hint: Consider Z2 , the

integers mod 2.

This sort of thing where it equals 0 if it sends everything to 0 works fine when we

think of polynomials as functions. However, consider 2 + where the coefficients

come from Z2 . It sends everything to 0 but we dont want to consider this polynomial

to be the zero polynomial because its coefficients are not all 0.

5. Let A L (V, V ) where V is a finite dimensional vector space with field of scalars F.

Let p () be the minimal polynomial and suppose () is any nonzero polynomial such

that (A) is not one to one and () has smallest possible degree such that (A) is

nonzero and not one to one. Show () must divide p ().

Say (A) x = 0 where x 6= 0. Consider all monic polynomials () such that (A) x = 0.

One such is the minimal polynomial p (). This condition is less severe than saying

that (A) = 0. Therefore, there are more polynomials considered. If you take the one

of least degree, its degree must be no larger than the degree of p () . By assumption

there is a polynomial () for which (A) 6= 0 but such that (A) x = 0. You can do

the usual thing.

p () = l () () + r ()

where r () = 0 or else it has smaller degree than () . Then

0 = l (A) (A) x + r (A) x

It follows that r (A) x = 0. If r () 6= 0 this would be a contradiction to the definition

of (). Therefore, () divides p () .

6. Let A L (V, V ) where V is a finite dimensional vector space with field of scalars F.

Let p () be the minimal polynomial and suppose () is an irreducible polynomial

with the property that (A) x = 0 for some specific x 6= 0. Show that () must

divide p () . Hint: First write

p () = () g () + r ()

where r () is either 0 or has degree smaller than the degree of (). If r () = 0 you

are done. Suppose it is not 0. Let () be the monic polynomial of smallest degree

with the property that (A) x = 0. Now use the Euclidean algorithm to divide ()

by () . Contradict the irreducibility of () .

Say (A) x = 0 where x 6= 0. Consider all monic polynomials () such that (A) x =

0. One such is the minimal polynomial p (). This condition is less severe than saying

that (A) = 0. Therefore, there are more polynomials considered. If you take the one

of least degree, its degree must be no larger than the degree of p () . You can do

the usual thing.

p () = g () () + r ()

108

Exercises

where r () = 0 or else it has smaller degree than () . Then

0 = g (A) (A) x + r (A) x

It follows that r (A) x = 0. If r () 6= 0 this would be a contradiction to the definition

of (). Therefore, () divides p () . Also

() = () l () + r ()

where the degree of r () is less than the degree of () or else r () = 0. If r () 6= 0,

then r (A) x = 0 and it would have degree too small. Hence r () = 0 and so ()

must divide () . But this requires that the two must be scalar multiples of each

other. Hence () must also divide p ().

det (I A) =

q

Y

j ()nj

j=1

where the j () are irreducible. Explain using Corollary 8.3.11 why the irreducible

factors of the minimal polynomial are j () and why the minimal polynomial is of

the form

q

Y

r

j () j

j=1

where rj nj . You can use the Cayley Hamilton theorem if you like.

This follows right away from the Cayley Hamilton theorem and that corollary. Since

the minimal polynomial divides the characteristic polynomial, it must be of the form

just mentioned.

8. Let

1 0

A= 0 0

0 1

0

1

0

You know that this satisfies a polynomial of degree no more than 3. Consider

2

3

1 0 0

1 0 0

1 0 0

1 0 0

0 1 0 , 0 0 1 , 0 0 1 , 0 0 1 These are

0 0 1

0 1 0

0 1 0

0 1 0

1 0 0

1 0 0

1 0

0

1 0 0

0 1 0 , 0 0 1 , 0 1 0 , 0 0 1

0 0 1

0 1 0

0 0 1

0 1 0

Thus it is desired to find a linear combination of these which equals 0 and out of all

of them to pick the one using as few of them as possible starting with the left side.

An easy way to do this is to use the row reduced echelon from. Write them as column

vectors and then do row operations.

109

Exercises

1

0

0

0

1

0

0

0

1

1

1

1

1 1

1

1

1

0

0

0

0 0

0

0

0

0

0

0

0 0

0

0

0

0

0

0

0 0

0

0

0

0 1 0 , 1 0 1 0 , row echelon form: 0

1 0

1

0 1 0

1

0

0

0

0

0 0

0

0

0

1

0 1

0 1

0 1

0

0 1 0

1 0 1 0

0

0

1

0

0

0

0

0

0

0

0

0

1

0

0

0

0

0

0

1

1

1

0

0

0

0

0

0

3 2 + 1

Does it work?

1

0

0

3

0 0

0 1

1 0

0 0

1 0

0 1 0 1

1 0

0 0

1

0

0

2

0

1 +

0

0

0 0 0

0 = 0 0 0

1

0 0 0

1

0

0

0

0

1

9. SupposeA is an n n matrix

and let v be a vector. Consider the A cyclic set of

vectors v, Av, , Am1 v where this is an independent set of vectors but Am v is

a linear combination of the preceding vectors in the list. Show how to obtain a monic

polynomial of smallest degree, m, v () such that

v (A) v = 0

Now let {w1 , , wn } be a basis and let () be the least common multiple of the

wk () . Explain why this must be the minimal polynomial of A. Give a reasonably

easy algorithm for computing v ().

You could simply form the cyclic set of vectors described, make them the columns of

a matrix and then do the row reduced echelon form to reveal linear relations. This is

how you can compute v (). Then, having found this, for each vector in a basis, and

letting () be the least common multiple, it follows that (A) v = 0 for each v in a

basis and so (A) = 0. Also, if (A) = 0, then () must be divisible by each of the

v () for v in the given basis. Here is why.

() = l () v () + r () , degree of r () < degree of v () or r () = 0

Then if r () 6= 0, you could contradict the fact that v () has smallest degree out of

all those () for which (A) v = 0. Therefore, () is divisible by each v () for

each v in the given basis. The polynomial of smallest such degree is by definition the

least common multiple.

10. Here is a matrix.

7 1

21 3

70

10

1

3

10

Using the process of Problem 9 find the minimal polynomial of this matrix. It turns

out the characteristic polynomial is 3 .

110

Exercises

1 0 0

1

7 1 1

1

0 0 0

1

0 1 0 0 21 3 3 0 0 0 0 0

0 0 1

0

70

10 10

0

0 0 0

0

1

7

0

1 7 0

1 0 0

0 21 0 , 0 21 0 , row echelon form: 0 1 0 It fol0

70

0

0 70 0

0 0 0

2

lows that e1 () , = 0 + 0. The same sort of thing will happen with e2 , e3 .

Therefore, the minimal polynomial is just 2 . Note that

7 1 1

0 0 0

21 3 3 = 0 0 0

70

10 10

0 0 0

11. Find the minimal polynomial for

1 2 3

A= 2 1 4

3 2 1

by the above technique. Is what you found also the characteristic polynomial?

First here

1

0

0

1 1

0 2

0 3

1

4

26

2 8 24

3

2

6

4 26

1 0 0 14

8 24 , row echelon form: 0 1 0

0

2 6

0 0 1

3

e1 () : 14 + 32 = 3

The minimal polynomial is then 3 32 +14. We know it cant have any larger degree

and so this must be it.

2

1 2 3

1 2 3

2 1 4 3 2 1 4

3 2 1

3 2 1

0

1 2 3

0 0

+14 2 1 4 = 0 0

3 2 1

0 0

0

0

0

This is also the characteristic equation. Note that no computation involving determinants was required.

12. Let A be an n n matrix with field of scalars C. Letting be an eigenvalue, show

the dimension of the eigenspace equals the number of Jordan blocks in the Jordan

canonical form which are associated with . Recall the eigenspace is ker (I A) .

Let the blocks associated with in the Jordan form be

J1 ()

J2 ()

..

.

Jr ()

111

Exercises

Thus this is an upper triangular matrix which has down the main diagonal. If you

take away I from it, you get a matrix of the form

N1 ()

N2 ()

..

.

Nr ()

where

0

Ni () =

0

..

..

1

0

How many columns of zeros will there be? Exactly r where r is the number of those

blocks. You get one at the beginning of each block. Hence there will be exactly r free

variables and so the dimension of the eigenspace will equal to the number of blocks.

13. For any n n matrix, why is the dimension of the eigenspace always less than or

equal to the algebraic multiplicity of the eigenvalue as a root of the characteristic

equation? Hint: Note the algebraic multiplicity is the size of the appropriate block

in the Jordan form.

This is obvious from the above problem. The dimension of the eigenspace equals the

number of blocks which must be no larger than the length of the diagonal.

14. Give an example of two nilpotent

minimal polynomial if possible.

2

0 1 0 0

0 0 0

0 0 0 0

0 0 0

0 0 0 0 = 0 0 0

0 0 0 0

0 0 0

2

0 1 0 0

0 0 0

0 0 0 0

0 0 0

0 0 0 1 = 0 0 0

0 0 0 0

0 0 0

These are two which work.

0

0

,

0

0

0

0

0

0

15. Use the existence of the Jordan canonical form for a linear transformation whose

minimal polynomial factors completely to give a proof of the Cayley Hamilton theorem

which is valid for any field of scalars. Hint: First assume the minimal polynomial

factors completely into linear factors. If this does not happen, consider a splitting field

of the minimal polynomial. Then consider the minimal polynomial with respect to

this larger field. How will the two minimal polynomials be related? Show the minimal

polynomial always divides the characteristic polynomial.

First suppose the minimal polynomial factors. Then the linear transformation has a

Jordan form.

J (1 )

..

J =

.

J (r )

112

Exercises

where each block is as described in the description of the Jordan form. Say J (i )

is ri ri in size. Then the characteristic polynomial is the determinant of I J a

matrix of the form

( 1 ) I + N (1 )

..

.

( r ) I + N (r )

where the blocks N (i ) have some 1s on the super diagonal and zeros elsewhere.

Thus the characteristic polynomial is the determinant of this matrix which is

q () =

r

Y

i=1

r

Y

i=1

J (1 i )

..

ri

( i )

ri

(J i I)

J (0)

..

.

J (r i )

form

r

J (1 i ) i

..

..

.

ri

J (r i )

Qr

ri

Therefore, the product i=1 (J i I) = 0 from block multiplication.

ri

is of the

Now in the general case where the minimal polynomial pF () does not necessarily

factor, consider a splitting field for the minimal polynomial. The new minimal polynomial divides the old one. Therefore, the new minimal polynomial with respect to

this splitting field G can be completely factored. Hence the Jordan form exists with

respect to this new splitting field and new minimal polynomial pG () . Then from the

above, it follows that pG (J) = 0. However, pG () divides pF () and so it is also the

case that pF (J) = 0. Since J is a matrix of T with respect to a suitable basis, it

follows that pF (T ) = 0 also, where T is the original linear transformation.

16. Here is a matrix. Find its Jordan canonical form by directly finding the eigenvectors

and generalized eigenvectors based on these to find a basis which will yield the Jordan

form. The eigenvalues are 1 and 2.

3 2 5 3

1 0 1 2

4 3 6 4

1 1 1 3

113

Exercises

Why is

3

1

4

1

2 5 3

4

1

0 1 2

1

, eigenvectors: 1, 0 2

3

1

3 6 4

1 1 3

1

0

3 2 5 3

1 0 0 0

x1

4

1 0 1 2 0 1 0 0 x2 1

4 3 6 4 0 0 1 0 x3 = 3

1 1 1 3

0 0 0 1

x4

1

4t4 4

4

x1

4 2 5 3

t4 + 1

1 1 1 2 x2 1

1

x4

1 1 1 2

t4

A generalized eigenvector is

3 2

1 0

4 3

1 1

5 2 5

1 2 1

4 3 4

1 1 1

5

1

6

1

3

2

4

1

3

1 0

0 1

2

2

0 0

4

3

0 0

x1

x2

x3 =

x4

A generalized eigenvector is

4

1

2

0

0

x1

x2

0

0 x3

1

x4

1

0

=

1

0

1

t3

1

0

, Solution is:

t3

1

0

1

0

0

1

0

4

4

1

0

1 1 0 1

3 2 1 0 ,

1

0

0

1

0

1

0

1

4 4 1

1 1 0

3 2 1

1 0 0

4 4 1

1 1 0

S=

3 2 1

1 0 0

0

1

0

1

0

1

0

1

Now to get the Jordan form you take S 1 AS, but of course we know the Jordan form

at this point. There will be two blocks, one for 1 and one for 2. Nevertheless, here it

114

Exercises

is

4

1

3

1

4

1

2

0

1

0

3

1

1

0 4

1

1

1

0

=

0

0

1

0

1

0

2

0

3

1

1

1

0

0

5

1

6

1

0

0

2

0

4

3

4 4 1

1 1 0

3 2 1

1 0 0

0

0

1

2

0

1

0

1

17. People like to consider the solutions of first order linear systems of equations which

are of the form

x (t) = Ax (t)

where here A is an n n matrix. From the theorem on the Jordan canonical form,

there exist S and S 1 such that A = SJS 1 where J is a Jordan form. Define

y (t) S 1 x (t) . Show y = Jy.

x = SJS 1 x

and so

S 1 x = J S 1 x , y = Jy

Now suppose (t) is an n n matrix whose columns are solutions of the above

differential equation. Thus

= A

Now let be defined by SS 1 = . Show

= J.

You have

= S 1 S

and so

S 1 S = S 1 AS = S 1 ASS 1 S

and so

det ( (t)) = Cetrace(A)t

This is called Abels formula and det ( (t)) is called the Wronskian.

It is easiest to consider = J and verify that det ( (t)) = Cetrace(J)t . This will do

it because J and A are similar and so are and .

det () =

n

X

det i

i=1

115

Exercises

where i has the ith row differentiated and the other rows of left unchanged. Thus,

letting i denote the ith row of , you have that the j th entry of i is given by

X

ij (t) =

Jik kj = i ij + ai (i+1)j

k

Thus

where ai either is 0 or 1. Then since the determinant of a matrix having two equal

rows equals 0, it follows that

(det ()) =

n

X

i=1

and so

det () = Cetrace(J)t

19. Let A be an n n matrix and let J be its Jordan canonical form. Recall J is a block

diagonal matrix having blocks Jk () down the diagonal. Each of these blocks is of

the form

1

0

..

Jk () =

..

. 1

0

1

0

D =

.

..

k1

0

Show that D1 Jk () D has the same form as Jk () but instead of ones down the

super diagonal, there is down the super diagonal. That is Jk () is replaced with

..

..

.

0

Now show that for A an n n matrix, it is similar to one which is just like the Jordan

canonical form except instead of the blocks having 1 down the super diagonal, it has

.

That D1 Jk () D is of the

1

0

0

0 1

0

0

0

2

0

0

0

right form is

0

0

0

0 0

0

3

1 0 0

1 0 0

0 0

1 0

0 1 0 0 2

0 0

0 0 0

the 4 4 case.

0

0

0

3

116

Exercises

1

0

0 1

=

0

0

0

0

0

0

2

0

0

0

0

0 0

3

0 0

0 0

0 0

0 0

0 0 0

J1

0

..

.

Jm

0

2

2

0

0

0

3

3

D1

0

..

.

0

Dm

and on the left by the inverse of this matrix. Using block mutiplication you get the

desired modification. Note that this shows that every matrix in n n matrix having

entries in C is similar to one which is close to a diagonal matrix.

20. Let A be in L (V, V ) and suppose that Ap x 6= 0 for some x 6= 0. Show that Ap ek 6= 0

for some ek {e1 , , en } , a basis for V . If you have a matrix which is nilpotent,

(Am = 0 for some m) will it always be possible to find its Jordan form? Describe how

to do it if this is the case. Hint: First explain why all the eigenvalues are 0. Then

consider the way the Jordan form for nilpotent transformations was constructed in the

above.

Yes, it is possible to compute the Jordan form. In fact, the proof of existence tells how

to do it. A nilpotent matrix has all eigenvalues equal to 0. This is because if Ax =x,

then Ak x = k x and so for large k, you get k x = 0. Finding the eigenvalues is always

the difficulty.

21. Suppose A is an n n matrix and that it has n distinct eigenvalues. How do the minimal polynomial and characteristic polynomials compare? Determine other conditions

based on the Jordan Canonical form which will cause the minimal and characteristic

polynomials to be different.

If the eigenvalues are distinct, then the two polynomials are obviously the same. More

generally, A has a Jordan form

J (1 )

..

.

J (r )

k being the distinct eigenvalues. Corresponding to one of these J (k ) you get the

term

( k )rk

117

Exercises

in the characteristic equation. If J (k ) is of the form

k 1

..

.

k

.

.

. 1

k

then J i I is of the form

J (1 i )

..

.

J (0)

..

.

J (r i )

where J (0) i = 0 and no smaller exponent will work. Therefore, the characteristic

polynomial and the minimal polynomial will coincide. However, if J (k ) is composed

of many blocks, the extreme case being when it is a diagonal matrix having k down

the diagonal, a smaller exponsne will work and so there will be a difference between

the characteristic and minimal polynomials.

For example, consider the minimal polynomial for the matrix

0 1 0 0

0 0 0 0

N =

0 0 0 1

0 0 0 0

of ones down the diagonal, then the minimal polynomial would also be 4 . Getting

difference between the two polynomials involves having lots of repeated eigenvalues

and lots of blocks for the Jordan form for each eigenvalue.

22. Suppose A is a 3 3 matrix and it has at least two distinct eigenvalues. Is it possible

that the minimal polynomial is different than the characteristic polynomial?

You might have distinct eigenvalues in which case

or you would have the following Jordan forms.

a 1 0

a

0 a 0 or 0

0 0 b

0

0

a

0

0

0

b

In the first case, the two polynomials are the same. The characteristic polynomial is

( a)2 ( b)

In the first case

a 1

0 a

0 0

0

a

0 0

b

0

0

a

0

0

a

0 0

a

0

1

a

0

0

b

0 0

b

0

0 0

b 0

0 b

118

Exercises

0

= 0

0

ab

0

0

0

0 6= 0

0

second case however,

a 0 0

a 0 0

a

0 a 0 0 a 0 0

0 0 b

0 0 a

0

0

= 0

0

0

a

0

0 0

0 0

0 0

0

b

0 0

b

0

0 0

b 0

0 b

23. If A is an n n matrix of entries from a field of scalars and if the minimal polynomial

of A splits over this field of scalars, does it follow that the characteristic polynomial

of A also splits? Explain why or why not.

If the minimal polynomial splits, then the matrix has a Jordan form with respect to

this field of scalars. Thus both the characteristic polynomial and minimal polynomial

are of the form

m

Y

r

( i ) i

i=1

24. In proving the uniqueness of the Jordan canonical form, it was asserted that if two

n n matrices A, B are similar, then they have the same minimal polynomial and also

that if this minimal polynomial is of the form

p () =

s

Y

ri

i ()

i=1

where the i () are irreducible, then ker (i (A)ri ) has the same dimension as ker (i (B)ri ) .

Why is this so? This was what was responsible for the blocks corresponding to an

eigenvalue being of the same size.

Since the two are similar, they come from the same linear transformation T . Therefore,

this is pretty obvious. Also, if B = S 1 AS, then

ri

r

r

ker (i (B) i ) = ker i S 1 AS

= ker S 1 (i (A) i ) S

r

25. Show that a given matrix is non defective (diagonalizable) if and only if the minimal

polynomial has no repeated roots.

If the matrix is non defective, then its Jordan form is

J (1 )

..

.

J (r )

119

Exercises

polynomial is

Y

( i ) .

i

Conversely, suppose the minimal polynomial is of this form. Also suppose that some

J (i ) say J (i ) is not diagonal. Then consider

Y

(J i I)

i

J1 (1 i )

..

.

Ji (0)

..

.

Jr (r i )

where for some i, Ji (0) is not the zero matrix. It follows that the above product

cannot be zero either. In fact, you could let x Fri be such that J (0) x = y 6= 0.

Then

0

0

.. ..

. .

Y

(J i I)

x = z , z 6= 0

. .

i

.. ..

0

0

because Ji (i k ) is one to one on Fri whenever k 6= i. It follows that each of the

J (k ) must be a diagonal matrix.

matrix using row operations. Next show that if p () and p () are relatively prime,

then p () has no repeated roots. With the above problem, explain how this gives a

way to determine whether a matrix is non defective.

You take the matrix A and consider the sequence

I, A, A2 , A3 , , An

There is some linear combination of these which equals 0. You just want to find the

one which involves the lowest exponents. Thus you string these out to make them

2

vectors in Fn and then make them the columns of a n2 n matrix. Find the row

reduced echelon form of this matrix. This will allow you to identify a linear relation

which gives the minimal polynomial.

If p () and p () are relatively prime, then you would need to have

Y

p () =

( i )

i

the i being distinct. Thus the minimal polynomial would have no repeated roots and

so the matrix is non defective by the above problem.

120

Exercises

27. In Theorem 10.3.4 show that the cyclic sets can be arranged in such a way that the

length of xi+1 divides the length of xi .

You make this condition a part of the induction hypothesis. Then at the end when

m1

you extend to (A)

(V) ker ((A)) , you note that, thanks to Lemma 10.3.2,

m1

each of the yi for yi ker (A)

has length equal to a multiple of d, the degree

m1

m1

of () while the x for x (A)

(V ) \ ker (A)

each have length equal to

d thanks to the assumption that () is irreducible over the field of scalars and each

of these is in ker ( (A)).

28. Show that if A is a complex n n matrix, then

a Jordan block. Note that

0 0 1

1 0

0

0 1 0 0 1 0

1 0 0

0 0

1

0

0 1

1 0 = 1

0 1

0 0

0

0

Such a backwards identity will work for any size Jordan block. Therefore, J is similar

to J T because you can use a similarity matrix which is block diagonal, each block

being appropriate for producing the transpose of the corresponding Jordan block.

Thus there is a matrix S such that

J = S 1 J T S

Now also J = P 1 AP for a suitable P. Therefore,

P 1 AP = S 1 P 1 AP

and so

A =

=

T

1

P S 1 P T AT P T

SP 1

1

1

1

PT

SP 1

AT P T

SP 1

29. Let A be a linear transformation defined on a finite dimensional vector space V . Let

the minimal polynomial be

q

Y

m

i () i

i=1

o

n

be the cyclic sets such that ivi , , ivri

is a basis for

1

1

i P P

i

mi

i

ker (i (A) ). Let v = i j vj . Now let q () be any polynomial and suppose that

and let

ivi , , ivri

q (A) v = 0

First suppose that V is a vector space with a cyclic basis x, Ax, , An1 x such

that An x is a linear combination of the vectors just listed. Let () be the monic

polynomial of degree n which results from this. Thus (A) x = 0. Then actually,

(A) = 0. This is because if v is arbitrary, then

v=

n1

X

ai Ai x.

i=0

121

Exercises

Then

(A) v =

n1

X

ai Ai (A) x = 0

i=0

q () = () l () + r ()

where r () either has smaller degree than () or r () = 0. Suppose that r () 6= 0.

Then

r (A) x = 0

and the degree of r () is less than n which contradicts the linear independence of

x, Ax, , An1 x

Now consider the problem of interest. Since each span ivi is invariant, it follows

j

P

i

i

i

that q (A) vj span vi . Therefore, 0 = i,j q (A) vj . By independence, each vector

j

in the sum equals 0 and so from the first part of the argument q (A) restricted to

ker (i (A)mi ) equals 0. Since the whole space is a direct sum of these, it follows that

q (A) = 0.

F.28

Exercises

10.9

1. Letting A be a complex n n matrix, in obtaining the rational canonical form, one

obtains the Cn as a direct sum of the form

span x1 span xr

the span of the previousvectors. Now apply the Gram Schmidt process to the ordered

basis x1 , x2 , , xr , the vectors in each xi listed according to increasing power

of A, thus obtaining an ordered basis (q1 , , qn ) . Letting Q be the unitary matrix

which has these vectors as columns, show that Q AQ equals a matrix B which satisfies

Bij = 0 if i j 2. Such a matrix is called an upper Hessinberg matrix and this shows

that every n n matrix is orthogonally similar to an upper Hessinberg matrix.

You note that qk is in the span of

x1

x2

z

}|

{ z

}|

{

x1 , , Am1 x1 , x2 , , Am2 x2 , , xl , Axl , , Aj xl

where there are k vectors in the above list. Since each span xi is A invariant, Aqk

is in the span of

x1

x2

z

}|

{ z

}|

{

x1 , , Am1 x1 , x2 , , Am2 x2 , , xl , Axl , , Aj xl , Aj+1 xl

span (q1 , , qk+1 )

122

Exercises

It follows that if p k + 2, then

qp Aqk = 0

Hessinberg.

2. In the argument for Theorem 10.2.4 it was shown that m (A) l (A) v = v whenever

r

r

v ker (k (A) k ) . Show that m (A) restricted to ker (k (A) k ) is the inverse of the

rk

linear transformation l (A) on ker (k (A) ) .

In this theorem you had polynomials m () , n () such that

rk

m () l () + n () k ()

Thus you also have

=1

=0

z }| {

r

m (A) l (A) x + n (A) k (A) k x = x

det (I A) =

q

Y

nj

j ()

j=1

where the j () are irreducible. Explain using Corollary 8.3.11 why the irreducible

factors of the minimal polynomial are j () and why the minimal polynomial is of

the form

q

Y

r

j () j

j=1

where rj nj . You can use the Cayley Hamilton theorem if you like.

From the Cayley Hamilton theorem, the minimal polynomial divides the characteristic

polynomial and so from the mentioned corollary, it follows that the minimal polynomial

has the form described above with rj nj .

4. Find the minimal polynomial for

1 2 3

A= 2 1 4

3 2 1

by the above technique with the field of scalars being the rational numbers. Is what

you found also the characteristic polynomial?

1 2 3

2 1 4 , characteristic polynomial: X 3 3X 2 + 14

3 2 1

It is irreducible over the rationals by the rational root theorem. Therefore, the above

must be the minimal polynomial as well.

123

Exercises

5. Show, using the rational root theorem, the minimal polynomial for A in the above

problem is irreducible with respect to Q. Letting the field of scalars be Q find the

rational canonical form and a similarity transformation which will produce it.

To find the rational canonical form, you need to look for cyclic sets. You have a single

irreducible polynomial, 3 3 + 14 = (). Also (A) sends every vector to 0.

Therefore, you can simply start with a vector. Take e1 .

1

1 2

0 , 2 1

3 2

0

3

1

1 2

4 0 , 2 1

1

0

3 2

3

1 2 3

1

2 1 4 0

3 2 1

0

2

3

1

4 0

1

0

26

1

1

4

0 , 2 , 8 , 24

6

0

3

2

The last is definitely a linear combination of the first three because the first three are

an independent set. Hence, you ought to have these as columns. Then you have

1 1 4

1 2 3

1

0 2 8 2 1 4 0

0 3 2

3 2 1

0

0

= 1

0

0 14

0

0

1

3

1 4

2 8

3 2

6. Find the rational canonical form for the matrix

1 2 1 1

2 3 0 2

1 3 2 4

1 2 1 2

This was just pulled out of the air. To find the rational canonical form, look for cyclic

sets.

1 0 0 0

1

1 2 1 1

1

1 2 1 1

1

0 1 0 0 0 2 3 0 2 0 2 3 0 2 0

0 0 1 0 0 , 1 3 2 4 0 , 1 3 2 4 0

0 0 0 1

0

1 2 1 2

0

1 2 1 2

0

1

2

1

1

2

3

3

2

3

1 1

1

0

0 2

2 4 0

1 2

0

1

2

,

1

1

2

3

3

2

1

0

2

1

4

1

1

0

2

4 0

2

0

124

Exercises

These reduce to

1

0

0

0

1

1

0 2

,

0 1

0

1

1

2

1

1

5

10

13

8

30

56

93

54

5

30

10 56

,

13 , 93

8

54

1

181

0

336

, row echelon form:

0

600

0

343

181

336

,

600

343

0

1

0

0

0

0

1

0

0 3

0 1

0 11

1

8

4 83 + 112 + + 3

Then

1

0

0

0

1

1 5 30

1 2

2 3

2 10 56

1 13 93 1 3

1 8 54

1 2

1 1

1 1

0 2

0 2

2 4 0 1

1 2

0 1

this. Also,

5 30

10 56

=

13 93

8 54

0

1

0

0

0

0

1

0

0

0

0

1

3

1

11

8

7. Let A : Q3 Q3 be linear. Suppose the minimal polynomial is ( 2) 2 + 2 + 7 .

Find the rational canonical form. Can you give generalizations of this rather simple

problem to other situations?

That second factor is irreducible over the rationals and so you would need the following

for the rational canonical form.

2 0 0

0 0 7

0 1 2

8. Find the rational canonical form with respect to the field of scalars equal to Q for the

matrix

0 0 1

A = 1 0 1

0 1 1

Then find the rational canonical form if the field of scalars equals C or Q + iQ.

First note that the minimal polynomial is 3 2 + 1 and a use of the rational

root theorem shows that this factors over the rationals as ( 1) 2 + 1 . Then you

need to locate cyclic sets starting with a vector in ker A2 + I . Lets find this first.

0 0

1 0

0 1

2

1

1 0

1 + 0 1

1

0 0

0

1

0 = 0

1

1

s t

s

t

1 1

0 0

1 1

125

Exercises

Starting with one of these vectors, lets find cyclic sets.

1

0

1 , 1

0

0

1

0 1

1

0 0

0 1 1 , 1 0

1 1

0

0 1

1

0

1

1 , 1 , 1

0

1

0

1

1

1 1

1

0

These of course will be dependent since otherwise we dont really have the minimal

polynomial. Thus the cycle can consist of only the first two. Thus we will have these

two as the first two columns and then we will have the eigenvector corresponding to

ker (A I)

1 0

1

ker 1 1 1

0

1

0

1

This eigenvector is just 0 . Then the rational canonical form of this matrix is

1

1

1 0 1

0 0

1 1 0 1 0

0

1 1

0 1

21

21

=

1

2

1

2

21

1

2

1

2

1

2

1

2

1

1 0 1

1 1 1 0

1

0

1 1

1

1 0 1

1 1 1 0

1

0

1 1

0

0

1

0 0

1 0

0 1

0 1

1 0

0 0

(

( i) and so the rational canonical form is just the diagonal matrix,

1) ( + i)

1 0 0

0 i 0

0 0 i

9. Let q () be a polynomial and C its companion matrix. Show the characteristic and

minimal polynomial of C are the same and both equal q ().

Let q () = a0 + a1 + + an1 n1 + n . Then its companion matrix is

0

0

a0

1 0

a1

C =

..

.

..

.

0

an1

Then you can show by induction that the characteristic polynomial of C is q (). See

Problem 43 on Page 66 It was also shown that q (C) = 0. (Cayley Hamilton theorem

for companion matrices.) Let p () be the minimal polynomial. Thus it has degree no

more than the degree of q () and thus must divide q (). Could it have smaller degree

126

Exercises

than q ()? No, it couldnt. To see this, consider, for example a companion matrix

for a monic polynomial of degree 4 denoted as A. The following gives the sequence

A0 , A, A2 , A3 . Note that these are linearly independent. Watch the ones to see this.

0 0 a

ad

1 0 0 0

0 0 0 a

0 1 0 0 1 0 0 b 0 0 b a + bd

0 0 1 0 0 1 0 c 1 0 c b + cd

0 1 d d2 + c

0 0 0 1

0 0 1 d

0 a

ad

a d2 +

c

2

0 b a + bd

b

d

+

c

+

ad

0 c b + cd a + c d2 + c + bd

1 d d2 + c b + d d2 + c + cd

This pattern will occur for any sized companion matrix. You start with ones down the

main diagonal and then the string of ones moves down one diagonal and then another

and so forth. After n 1 multiplications, you get a 1 in the lower left corner. Each

multiplication moves the ones into slots which had formerly been occupied with zeros.

That is why these are independent. Therefore, the degree of the minimal polynomial

is at least n and so the minimal and characteristic polynomials must coincide for any

companion matrix.

10. Use the existence of the rational canonical form to give a proof of the Cayley Hamilton

theorem valid for any field, even fields like the integers mod p for p a prime. The earlier

proof was fine for fields like Q or R where you could let but it is not clear the

same result holds in general.

The linear transformation has a matrix of the following form,

M1

..

M =

.

Mq

mk

Say the minimal polynomial is

p () =

q

Y

k ()

mk

k=1

C1k

..

.

Clkk

r

the above problem, the minimum polynomial for Cjk is just k () j , rj mk which

is also the characteristic polynomial of this matrix. The characteristic polynomial is

then

q

q Y

lk

Y

Y

q () =

det (I Mk ) =

det I Cjk

i=1 j=1

k=1

q Y

lk

Y

kj () =

k=1 j=1

q Y

lk

Y

k=1 j=1

k () j , rj mk

127

Exercises

It follows that

q (A) =

q Y

li

Y

k (A)rj

i=1 j=1

k rj

We know that k Cj

B1

..

B=

.

Bq

and () is a polynomial,

(B1 )

..

(B) =

.

(Bq )

Therefore, the above product for q (A) is the product of block diagonal matrices such

that every position on the diagonal has a zero matrix in at least one of the block

matrices in the product. This equals the zero matrix. Like this:

0 0 0

D 0 0

F 0 0

0 B 0 0 0 0 0 H 0

0 0 C

0 0 E

0 0 0

When you multiply the block matrices, it involves simply multiplying together the

blocks in the corresponding positions. Since at least one is zero, it follows the product

is the zero matrix.

11. Suppose you have two n n matrices A, B whose entries are in a field F and suppose G

is an extension of F. For example, you could have F = Q and G = C. Suppose A and

B are similar with respect to the field G. Can it be concluded that they are similar

with respect to the field F? Hint: Use the uniqueness of the rational canonical form.

Do A, B have the same minimal polynomial over F? Denote these by pA () and pB ()

You have

pA (A) = 0

and so pA (B) = 0 also because A and B are similar with respect to G. It follows that

pA () must divide pB () . Similar reasoning implies that pB () must divide pA ().

Therefore, these two minimal polynomials having coefficients in F coincide. Otherwise,

you could write

pB () = pA () + r ()

where the degree of r () is smaller than the common degree of pA () and pB ().

Then plugging in = B, you get a contradiction to the definition of pB () as having

the smallest degree possible. Say the common minimal polynomial over F is

q

Y

i ()mi

i=1

where i () is irreducible over F. Thus, each of these i is a polynomial with coefficients in F and hence in G, and so

i (A) , i (B)

128

Exercises

are similar over G. Letting vi and wj denote the A cyclic sets associated with

ker (i (A))

mi

, ker (i (B))

mi

respectively, it follows from Lemma 10.8.3 that the lengths of these vi and wi comprise exactly the same set of positive numbers. These lengths are identified with the

size of the Jordan blocks of the two similar matrices, i (A) , i (B). However, the

proof of this lemma exhibits a basis which yields the Jordan form, which happens to

involve only the field F. The companion matrices corresponding to two of these which

have the same length are therefore exactly the same, both being companion matrices

k

of i () for a suitable k. It follows that the rational canonical form for A, B over F

is exactly the same. Hence, by uniqueness, the two matrices are similar over F.

F.29

Exercises

11.4

1. Suppose the migration matrix for three

.5

.3

.2

locations is

0 .3

.8 0 .

.2 .7

Find a comparison for the populations in the three locations after a long time.

.5 0 .3

1 0 0

0.5

0

0.3

.3 .8 0 0 1 0 = 0.3 0.2

0

.2 .2 .7

0 0 1

0.2

0.2 0.3

1

3

0

0

1 0 53 0

2

10

9

3

15

0

0 , row echelon form: 0 1 10

0

10

1

1

3

0

0

0

0

10 0

5

5

It follows that the populations are

.6

t .9

1

where t is chosen such that the sum of the entries corresponds to the total amount at

the beginning.

P

2. Show that if i aij = 1, then if A = (aij ) , then the sum of the entries of Av equals

the sum of the entries of v. Thus it does not matter whether aij 0 for this to be so.

P P

P P

P

i

j aij vj =

j

i aij vj =

j vj

3. If A satisfies the conditions of the above problem, can it be concluded that limn An

exists?

Not necessarily. Consider the following matrix.

4

9

3 8

4

4

9

311 936

=

etc. The eigenvalues of this matrix are

3 8

312

937

1, 5. Thus there is no way that the limit can exist. Just do the powers of the matrix

to the eigenvector which corresponds to 5.

129

Exercises

4. Give

1.

1

0

0

0

0

1

0

1 , eigenvalues: 1, 1

0

5. Show that when a Markov matrix is non defective, all of the above theory can be proved

very easily. In particular, prove the theorem about the existence of limn An if the

eigenvalues are either 1 or have absolute value less than 1.

It suffices to show that limn An v exists for each v Rn . This is because you could

then assert that limn eTi An ej exists. But this is just the ij th entry of An . Let the

eigen pairs be

(1, v1 ) , (i , vi ) , i = 2, , n

and by assumption, these vi form a basis. Therefore, if v is any vector, there exist

unique scalars ci such that

X

v=

ci vi

i

It follows that

Am v = c1 v1 +

n

X

j=2

ci m vi c1 v1

6. Find a formula for An where

5

2

5

A=

7

2

7

2

12

0

12

12

0

0

1

2

1

4

25

2

Does limn An exist? Note that all the rows sum to 1. Hint: This matrix is similar

to a diagonal matrix. The eigenvalues are 1, 1, 12 , 12 .

5

21 0 1

2

5

0

0 4

7

2 12 52

2

7

21 0 2

2

1 1 1 0

1 0 0 0

1

0 0 1

1

3 2 1 0 0

0 0

1 0 2

2

=

5

2

0 0 1 0 1

1

1

1

0 1

4

2 1 1 0

0 0 0 12

41 41 1 21

Now it follows that

1 1 1 0

3 2 1 0

An =

2 5 1 1

4

2 1 1 0

1

n

1

2n (1) + 1

2n

2n 3 (1)n + 1 2n

2

2

=

1n 2 (1)n + 1 1n

2

2

n

1

1

2n 2 (1) + 1

2n

(1)

0

0

0

1

1

1

1

0

1

0

1

0

1

1

2n

0

0 1

1

1

1

1

0

2n

4

4

n

2

0

(1) 2n + 1

0 3 (1)n 24n + 1

1

2 (1)n 23n + 1

2n

n

0 2 (1) 22n + 1

0

0

0

1

0

0 1

0 2

0 1

1 12

130

Exercises

2

4

A=

5

2

3

21

0

21

21

1

4

2

2

1

2

1

1

1

2

Note that the rows sum to 1 in this matrix also. Hint: This matrix is not similar

to a diagonal matrix but you can find the Jordan form and consider this in order to

obtain a formula for this product. The eigenvalues are 1, 1, 21 , 12 .

2 21 12 1

4

0

1 4

5

2 1 2

2

3 21 12 2

97 1 61 79

1 0 0 0

1 0

0

1

23 1 1 14 0 1 0 0 1 1

0

1

9

3

9

=

13 1 1 4 0 0 1 1 0 6 14 20

9

6

9

2

0 0 0 12

1 0

3

2

16

1 61 79

9

Now consider the Jordan block on the bottom.

It when raising this to the nth power, you get

=

=

1

2

X

n

2nk

k

0

k=0

1

0

2n

+n

1

0

n

1 21n

2

n

2n

1

0

n

2

0

1

2nk

1

2n1

0

1

2n1

0

0

1

0

k

0 1

0 0

n

97 1 61 97

(1)

0 0

0

1 0

0

1

23 1 1 14

1 1

0

1 0

0

0

1

9

3

9

13

1

4

1

1n

0

0

2

n

0

6

14

20

2n

9

6

9

1

7

1

1

0

3

2

16

1

0

0

0

2n

9

6

9

n

n

7

1 1n

7

1

1 1n

7

16

1 1n

2

n

(1)

+

1

1

n

n+1

92n

6

9

2n

22

9 (1) 92n 3 2

n

n

23

32

2 1n

1n

14 n 1 21n n 23 (1) + 1 2n 1

2

n

n+1

3

9

2

9 (1) 92n 3 2

92

4 n 1 21n n 13 (1)n + 1 1n 1 1 21n n + 1n 13 (1)n 22 n 1 21n n + 1

92

6

9

2

2

2

9

92

3

n

n

7

1 1n

1

1 1n

16

16

1 1n

n 16

n

n+1

92n 6 2

9 (1) + 1

2n 1

22

9 (1) 92n 3 2

8. Find limn An if it exists for the matrix

1

12

2

1

1

2

2

A=

1

1

2

3

2

2

3

2

21

21

3

2

3

2

0

0

0

1

The matrix equals

131

Exercises

1

1 1 1 0

1 1

0

0

0

2 1 1 0 0

4 3 1 1

0

0

1

2

0

0

0

0

1

0

0

1

1

0

0 1

0

1

0 1

1 1

1 0

2 1

1 1 1 0

1 0 0 0

1

1 1

0 1n 0 0 1

0

0

2

2 1 1 0 0 0 1 0 1

4 3 1 1

0

0 0 0 1

you get

1 1 1 0

1 0 0 0

1

1 1

0 0 0 0 1

0

0

2 1 1 0 0 0 1 0 1

4 3 1 1

0 0 0 1

0

0 1 1 0

1 0 1 0

=

1

1

2 0

3

3

3 1

0

1

1

2

0

1

1

2

1

1

0

1

1

1

0

1

0

0

0

1

0

0

and so, in the limit,

0

1

0

0

0

1

9. Given an example of a matrix A which has eigenvalues which are either equal to 1,1,

or have absolute value strictly less than 1 but which has the property that limn An

does not exist.

1 0 0

An easy example is 0 0 1 , eigenvalues: 1, 1

0 1 0

1 0 0

1 0 0

0 0 1 alternates between the identity and 0 0 1 .

0 1 0

0 1 0

10. If A is an n n matrix such that all the eigenvalues have absolute value less than 1,

show limn An = 0.

The Jordan form is

J (1 )

..

.

J (r )

m

0.

11. Find an example of a 3 3 matrix A such that limn An does not exist but

limr A5r does exist.

An easy example is the following.

i(2/5)

e

0

0

A=

0

ei(2/5)

0

i(2/5)

0

0

e

i(2/5)

5

e

0

0

= I.

A5 =

0

ei(2/5)

0

i(2/5)

0

0

e

Therefore, limr A5r exists. However, limr Ar cannot exist because it will just

yield diagonal matrices which have various fifth roots of 1 on the diagonal.

132

Exercises

12. If A is a Markov matrix and B is similar to A, does it follow that B is also a Markov

matrix?

No. Start with a Markov matrix

1/2 1/3

1/2 2/3

1

1

1 2

1/2 1/3

1 2

2

=

3 4

1/2 2/3

3 4

1

1

5

3

13. P

In Theorem 11.1.3

P suppose everything is unchanged except that you assume either

a

1

or

ij

j

i aij 1. Would the same conclusion be valid? What if you dont

insist that each aij 0? Would the conclusion hold in this case?

You cant conclude that the limit exists if the sign of aij is not restricted. For example,

4

9

3 8

has an eigenvalue of 5Pand so the limit of powers of this matrix cannot exist. What

about the case where i aij 1 instead of equal to 1? In this case, you do still

get the conclusion of this theorem. The condition was only used to get the entries

of An bounded independent of n and this can be accomplished just as well with the

inequality.

14. Let V be an n dimensional vector space and let x V and x 6= 0. Consider x

x,Ax, ,Am1 x where

Am x span x,Ax, ,Am1 x

and m

is the smallest such

that the above inclusion in the span takes place. Show

that x,Ax, ,Am1 x must be linearly independent. Next suppose {v1 , , vn }

is a basis for V . Consider vi as just discussed, having length mi . Thus Ami vi is a

linearly combination of vi ,Avi , ,Am1 vi for m as small as possible. Let pvi () be

the monic polynomial which expresses this linear combination. Thus pvi (A) vi = 0

and the degree of pvi () is as small as possible for this to take place. Show that the

minimal polynomial for A must be the monic polynomial which is the least common

multiple of these polynomials pvi ().

Let p () be the minimal polynomial. Then there exists ki () , a monic polynomial

such that

p () = ki () pvi () + ri ()

where ri () must be zero since otherwise, you would have ri (A) vi = 0 and this would

contradict the minimality of the degree of pvi (). Therefore, each of these pvi ()

divides p () . If q () is their least common multiple, then it follows that q (A) = 0

because every vector is a linear combination of these vectors vi . Therefore, p ()

divides q (). Thus

q () = p () k ()

Now each pvi () divides p () so it is a common multiple. It follows that k () = 1

since otherwise, q () wouldnt really be the least common multiple.

133

Exercises

15. If A is a complex Hermitian n n matrix which has all eigenvalues nonnegative, show

that there exists a complex Hermitian matrix B such that BB = A.

You have A = U DU for D a diagonal matrix having all nonnegative entries. Then

just note that B = U D1/2 U U D1/2 U, U is unitary. Clearly you should take B =

U D1/2 U.

16. Suppose A, B are n n real Hermitian matrices and they both have all nonnegative

eigenvalues. Show that det (A + B) det (A) + det (B).

Let P 2 = A, Q2 = B where P, Q are Hermitian and nonnegative. Then

P

A+B = P Q

Q

Now apply the Cauchy Binet formula to both sides.

X

det (A + B) =

det (Ci )2 det (P )2 + det (Q)2

i

= det P 2 + det Q2 = det (A) + det (B)

where the Ci are n n submatrices of P Q . Note that this does not depend on

the matrices being real. You could write

P

A + B = P Q

Q

Then

det (A + B)

det (Ci Ci )

det P

+ det Q

c

is an (n + 1) (n + 1) Hermitian nonnegative matrix where

b A

is a scalar and A is n n. Show that must be real, c = b, and A = A , A is

nonnegative, and that if = 0, then b = 0.

b

B =

= B and so you must have is real and b = c while A = A .

c A

Suppose = 0 and b 6= 0. I need to show that B is not nonnegative.

17. Suppose B =

}| {

0 b

x

b A

x

b x

x

= b x+x b + x Ax

b + Ax

z

If for some x, x Ax < 0, then let = 0 and you see that B is not nonnegative.

Therefore, A must be nonnegative. But now you could let x = b and then pick be

large and negative and again see that B is not nonnegative. Hence if = 0, then

b = 0. Now consider b 6= 0 and determine the sign of . Consider

B

z

}| {

b

=

b A

0

=

134

Exercises

18. If A is an n n complex Hermitian and nonnegative matrix, show that there exists

an upper triangular matrix B such that B B = A. Hint: Prove this by induction. It

is obviously true if n = 1. Now if you have an (n + 1) (n

+ 21) Hermitian

nonnegative

b

matrix, then from the above problem, it is of the form

, real.

b A

From the above problem, a generic (n + 1) (n + 1) Hermitian nonnegative matrix is

of the form just claimed where A is nonnegative and Hermitian and is real. In case

= 0, there is nothing to prove. You would just use

0 0

0 0

0 B

0 B

where B B = A. Otherwise, you consider

0

b B

0

b

B

A11 A12

A=

A21 A22

where A11 , A22 are square matrices. Show that det (A) det (A11 ) det (A22 ). Hint:

Use the above problem to factor A getting

B11 0

B11 B12

A=

B12

B22

0

B22

B11 , A22 = B12

B12 + B22

B22 . Use the Cauchy Binet theo

B12 + B22

B22 ) det (B22

B22 ) . Then explain

why

det (A)

= det (B11

) det (B22

) det (B11 ) det (B22 )

= det (B11

B11 ) det (B22

B22 )

B11 , A22 = B12

B12 +B22

B22 follows

from block multiplication. Why is the determinant of a block triangular

matrix

equal

B11 B12

to the product of the determinants of the blocks? Consider

. If the

0

B22

rank of B22 is not maximal, then its determinant is zero and also the determinant of

the matrix is zero. Assume then that it has full rank. Then suitable row operations

can be applied to obtain that the determinant of this matrix equals the determinant

of

B11

0

0

B22

This determinant is clearly equal to the product of the determinants of B11 and B22 .

B12 + B22

B22 ) det (B22

B22 ) follows from the Cauchy Binet

theorem since there are more nonnegative terms in the sum for the determinant on the

left than for the term on the right. Note that these determinants are all real because

the matrices are Hermitian. It follows that

det (A)

= det (B11

B11 ) det (B22

B22 )

B12 + B22

B22 ) = det (A11 ) det (A22 )

135

Exercises

then

Y

det (A)

Aii

i

This follows from induction. It is clearly true if n = 1. Now let A be nn, nonnegative

and Hermitian and assume the theorem is true for all nonnegative matrices which are

of smaller size. Then

B a

A=

a

It follows that B and are both nonnegative. Then from the above,

det (A) det (B) det ()

By induction, this implies

det (A)

F.30

Aii

Exercises

12.7

1. Find the best solution to the system

x + 2y = 6

2x y = 5

3x + 2y = 0

1 2

1 2

14 6

2 1 2 1 =

6 9

3 2

3 2

1 2

6

14 6

x

Solve

= 2 1 5

6 9

y

3 2

0

17

14 6

x

16

15

=

, Solution is:

1

6 9

y

7

45

vector (1, 1, 2).

0

1

1

A basis consists of 1 , 1 , 0

2

0

0

1 0 1

1 1 0

2 0 0

1

1

1

5 6 25 5

6

6

6 6 30

6

6 6

1

1

1

= 61 6

5

6

0 0

6

6 5 6 30 5 6

1

1

1

2

6

5

6

5

5

0

0

3

15

5

5

136

Exercises

Then one example of an orthonormal basis extending the given vector is

2

1

1

5 6

30

6 6

5 5

1 6 , 1 5 6 ,

0

6

6

1

1

1

5

6

5 5

3

15

Define

(x, y) (Ax, y) .

Show this is an inner product on Rn . What does the Cauchy Schwarz inequality say

in this case?

It satisfies all the axioms of an inner product obviously. The only ones which are not

obvious are whether (Ax, x) 0 and so forth. But A = U DU where D is diagonal

and U is orthogonal. Therefore,

2

where is the smallest eigenvalue which is assumed positive. Therefore, also (Ax, x) =

0 if and only if |x| = 0. The Cauchy Schwarz inequality says

|(Ax, y)| (Ax, x)1/2 (Ay, y)1/2

4. Let

||x|| max {|xj | : j = 1, 2, , n} .

T

Show this is a norm on Cn . Here x = x1 xn

. Show

1/2

The axioms of a norm are all obvious except for the triangle inequality. However,

kx + yk = max {|xi + yi | , i n} kxk + kyk

so even this one is all right. Also for some i

kxk

1/2

2

= |xi | = |xi |

5. Let

||x||1

Show this is a norm on Cn . Here x =

n

X

j=1

|xi |

"1/2

= |x| .

|xj | .

x1

xn

1/2

T

. Show

!

"1/2

X

X

2

|xi |

|xi |

?

i

Of course. Just square both sides and the left has mixed terms which are not present

on the right.

137

Exercises

6. Show that if |||| is any norm on any vector space, then

|||x|| ||y||| ||x y|| .

kxk = kx y + yk kx yk + kyk and so

kxk kyk kx yk

Now repeat the argument with the x, y switched. Finally, |kxk kyk| equals either

kxk kyk or kyk kxk either way, you get the inequality.

7. Relax the assumptions in the axioms for the inner product. Change the axiom about

(x, x) 0 and equals 0 if and only if x = 0 to simply read (x, x) 0. Show the Cauchy

Schwarz inequality still holds in the following form.

|(x, y)| (x, x)

1/2

1/2

(y, y)

8. Let H be an inner product space and let {uk }nk=1 be an orthonormal basis for H.

Show

n

X

(x, y) =

(x, uk ) (y, uk ).

k=1

You have x =

(x, uj ) uj and y =

(x, y) =

(y, uj ) uj . Therefore,

(x, uj ) uj ,

X

j

(y, uk ) uk

j,k

n

X

(x, uk ) (y, uk )

k=1

9. Let the vector space V consist of real polynomials of degree no larger than 3. Thus a

typical vector is a polynomial of the form

a + bx + cx2 + dx3 .

For p, q V define the inner product,

(p, q)

Then state

the Cauchy Schwarz inequality in

terms of this inner product. Show 1, x, x2 , x3 is a basis for V . Finally, find an

orthonormal basis for V. This is an example of some orthonormal polynomials.

p1 (x) = 1

p2 (x) =

x(1/2)

1/2 =

3 (2x 1)

1 2

x 2 )

0 (

x2 x+ 16

=

1

2

2

2

2

kx (x , 3(2x1)) 3(2x1)(x ,1)1k

kx 6 3 3(2x1) 13 k

x(x,p1 )p1

kx(x,p1 )p1 k

R

1

p3 (x) =

= 6 5 x2 x + 16

138

Exercises

1

1

3

3

x3 20

56 5(x2 x+ 16 ) 20

3 3(2x1) 41

x3 20

56 5(x2 x+ 16 ) 20

3 3(2x1) 14

=

1

1

1

1

3

3

2

2800

kx 20 56 5(x x+ 6 ) 20 3 3(2x1) 4 k

2800

p4 (x) =

= 20 7 x3 32 x2 + 35 x

1

20

1

3

2

1, 3 (2x 1) , 6 5 x x +

, 20 7 x3 x2 +

6

2

is

3

1

x

5

20

10. Let Pn denote the polynomials of degree no larger than n 1 which are defined on an

interval [a, b] . Let {x1 , , xn } be n distinct points in [a, b] . Now define for p, q Pn ,

(p, q)

n

X

p (xj ) q (xj )

j=1

Show this yields an inner product on Pn . Hint: Most of the axioms are obvious. The

one which says (p, p) = 0 if and only if p = 0 is the only interesting one. To verify this

one, note that a nonzero polynomial of degree no more than n 1 has at most n 1

zeros.

If (p, p) = 0 then p equals zero at n distinct points and so it must be the zero polynomial

because it has degree at most n 1 and so it can have no more than n 1 roots.

11. Let C ([0, 1]) denote the vector space of continuous real valued functions defined on

[0, 1]. Let the inner product be given as

Z 1

(f, g)

f (x) g (x) dx

0

Show this is an inner product. Also let V be the subspace described in Problem 9.

Using the result of this problem, find the vector in V which is closest to x4 .

It equals

x4 , 1 1 + x4 , 3 (2x 1)

3 (2x 1)

1

1

+ x4 , 6 5 x2 x +

6 5 x2 x +

6

6

3

3

3

1

3

1

+ x4 , 20 7 x3 x2 + x

20 7 x3 x2 + x

2

5

20

2

5

20

Now it is just a matter of working these out.

R1

2

x4 , 3 (2x 1) = 0 x4 3 (2x 1) = 15

3

R 1 4

1

2

4

2

2

x , 6 5 x x + 6 = 0 x 6 5 x x + 16 = 35

5

R

1

1

1

1

x4 , 20 7 x3 32 x2 + 53 x 20

= 0 x4 20 7 x3 32 x2 + 35 x 20

= 70

7

Therefore, the closest point is

1

2

2

1

+

3 3 (2x 1) +

56 5 x2 x +

+

5 15

35

6

1

3

3

1

+

720 7 x3 x2 + x

70

2

5

20

2

1

9

= 2x3 x2 + x

7

7

70

139

Exercises

12. A regular Sturm Liouville problem involves the differential equation, for an unknown function of x which is denoted here by y,

and it is assumed that p (t) , q (t) > 0 for any t [a, b] and also there are boundary

conditions,

C1 y (a) + C2 y (a) = 0

C3 y (b) + C4 y (b) = 0

where

C12 + C22 > 0, and C32 + C42 > 0.

There is an immense theory connected to these important problems. The constant,

is called an eigenvalue. Show that if y is a solution to the above problem corresponding

to = 1 and if z is a solution corresponding to = 2 6= 1 , then

Z

(6.37)

= 0,

= 0.

and then integrate. Use the boundary conditions to show that y (a) z (a)z (a) y (a) =

0 and y (b) z (b) z (b) y (b) = 0. The formula, 6.37 is called an orthogonality relation.

It turns out there are typically infinitely many eigenvalues and it is interesting to write

given functions as an infinite series of these eigenfunctions.

Let y go with and z go with .

Subtract.

=

=

0

0

Now integrate from a to b. First note that

z (p (x) y ) y (p (x) z ) =

d

(p (x) y z p (x) z y)

dx

p (b) y (b) z (b) p (b) z (b) y (b) (p (a) y (a) z (a) p (a) z (a) y (a))

+ ( )

140

Exercises

Look at the stuff on the top line. From the assumptions on the boundary conditions,

C1 y (a) + C2 y (a)

C1 z (a) + C2 z (a)

= 0

= 0

and so

y (a) z (a) y (a) z (a) = 0

Similarly,

y (b) z (b) y (b) z (b) = 0

Hence, that stuff on the top line equals zero and so the orthogonality condition holds.

13. Consider the continuous functions defined on [0, ] , C ([0, ]) . Show

Z

(f, g)

f gdx

0

nq

o

2

is an inner product on this vector space. Show the functions

sin

(nx)

are

n=1

an orthonormal set. What does

of the vector space

q the dimension

qthis mean about

2

2

C ([0, ])? Now let VN = span

sin (x) , ,

sin (N x) . For f C ([0, ]) find

a formula for the vector in VN which is closest to f with respect to the norm determined

from the above inner product. This is called the N th partial sum of the Fourier series

of f . An important problem is to determine whether and in what way this Fourier

series converges to the function f . The norm which comes from this inner product is

sometimes called the mean square norm.

y + n2 y = 0, y (0) = y () = 0

It is obvious that sin (nx) solves this boundary value problem. Those boundary conditions in the above problem are implied by these. In fact,

1 sin 0 + (0) cos (0) =

1 sin + (0) cos

0

0

Z

Now also

R

0

sin (nx)

are orthonormal.

1 4 1 1

2 1 2

3

A=

4 9 0

1

5 6 3

4

Find an orthonormal basis for V. Hint: You might first find a basis and then use the

Gram Schmidt procedure.

141

Exercises

1

2

4

5

4

1

9

6

1 1

x

y

2

3

0

1 z

3

4

w

9

7 t3

0

0

4 t3

= , Solution is: 7

0

t3

0

0

This is pretty easy. The space has dimension 1. Therefore, an orthonormal basis is

just

9

146

146

9

4 2 146

1

= 73

7

146

16 + 49 + 81 7 146

0

0

15. The Gram Schmidt process starts with a basis for a subspace {v1 , , vn } and produces an orthonormal basis for the same subspace {u1 , , un } such that

span (v1 , , vk ) = span (u1 , , uk )

for each k. Show that in the case of Rm the QR factorization does the same thing.

More specifically, if

A = v1 vn

and if

A = QR

q1

qn

then the vectors {q1 , , qn } is an orthonormal set of vectors and for each k,

span (q1 , , qk ) = span (v1 , , vk )

This follows from the way we multiply matrices and the definition of an orthogonal

matrix.

16. Verify the parallelogram identify for any inner product space,

|x + y|2 + |x y|2 = 2 |x|2 + 2 |y|2 .

Why is it called the parallelogram identity?

2

|x + y| + |x y| = (x + y, x + y) + (x y, x y)

= |x|2 + |y|2 + 2 (x, y) + |x|2 + |y|2 2 (x, y) .

17. Let H be an inner product space and let K H be a nonempty convex subset. This

means that if k1 , k2 K, then the line segment consisting of points of the form

tk1 + (1 t) k2 for t [0, 1]

is also contained in K. Suppose for each x H, there exists P x defined to be a point

of K closest to x. Show that P x is unique so that P actually is a map. Hint: Suppose

z1 and z2 both work as closest points. Consider the midpoint, (z1 + z2 ) /2 and use the

parallelogram identity of Problem 16 in an auspicious manner.

Suppose there are two closest points to x say k1 , k2 both work. Then

K and

2

2

x k1 + k2 = x k1 + x k2

2

2

2

2

2

k1 +k2

2

is also in

142

Exercises

2

2

k2 k1 2

+ 2 x k1 + 2 x k2

2

2

2

2

2

2

k2 k1

+ d2

2

18. In the situation of Problem 17 suppose K is a closed convex subset and that H

is complete. This means every Cauchy sequence converges. Recall from calculus a

sequence {kn } is a Cauchy sequence if for every > 0 there exists N such that

whenever m, n > N , it follows |km kn | < . Let {kn } be a sequence of points of K

such that

lim |x kn | = inf {|x k| : k K}

n

This is called a minimizing sequence. Show there exists a unique k K such that

limn |kn k| and that k = P x. That is, there exists a well defined projection map

onto the convex subset of H. Hint: Use the parallelogram identity in an auspicious

manner to show {kn } is a Cauchy sequence which must therefore converge. Since K

is closed it follows this will converge to something in K which is the desired vector.

Let x be given and let {kn } be a minimizing squence.

|x kn | inf {|x y| : y K}

Then

2

kn km 2 x kn

x kn 2

x km 2

x km

+

+

= 2

+2

2

2

2

2

2

kn km 2

2

1

2

|x kn | +

2

1

2

|x kn | +

2

2

1

kn + km

2

|km x| x

2

2

1

2

|km x| 2

2

since the space is given to be complete, it follows that this converges to some k. Since

K is closed, it follows that k K. Now

|x k| = lim |x kn | =

n

19. Let H be an inner product space which is also complete and let P denote the projection map onto a convex closed subset, K. Show this projection map is characterized

by the inequality

Re (k P x, x P x) 0

for all k K. That is, a point z K equals P x if and only if the above variational

inequality holds. This is what that inequality is called. This is because k is allowed

to vary and the inequality continues to hold for all k K.

Consider for t [0, 1] the following.

|y (x + t (w x))|

143

Exercises

where w K and x K. It equals

2

f (t) = |y x| + t2 |w x| 2t Re hy x, w xi

Suppose x is the point of K which is closest to y. Then f (0) 0. However, f (0) =

2 Re hy x, w xi . Therefore, if x is closest to y,

Re hy x, w xi 0.

Next suppose this condition holds. Then you have

2

|y (x + t (w x))| |y x| + t2 |w x| |y x|

x is the closest point.

20. Using Problem 19 and Problems 17 - 18 show the projection map, P onto a closed

convex subset is Lipschitz continuous with Lipschitz constant 1. That is

|P x P y| |x y|

hP x P y,y P yi

hP y P x,x P xi

0

0

Thus

hP x P y,x P xi 0

Hence

hP x P y,x P xi hP x P y,y P yi 0

and so

hP x P y,xy (P x P y)i 0

|xy| |P x P y| hP x P y,P x P yi = |P x P y|

21. Give an example of two vectors in R4 x, y and a subspace V such that x y = 0 but

P xP y 6= 0 where P denotes the projection map which sends x to its closest point on

V.

Try this. V is the span of e1 and e2 and x = e3 + e1 , y = e4 + e1 .

P x = (e3 + e1 , e1 ) e1 + (e3 + e1 , e2 ) e2 = e1

P y = (e4 + e1 , e1 ) e1 + (e4 + e1 , e2 ) e2 = e1

P xP y = 1

22. Suppose you are given the data, (1, 2) , (2, 4) , (3, 8) , (0, 0) . Find the linear regression

line using the formulas derived above. Then graph the given data along with your

regression line.

You draw the graphs. You want to solve

a+b=2

2a + b = 4

3a + b = 8

b=0

144

Exercises

Of course there is no solution. The least squares solution is to solve

T

1 1

1 1

1 1

2

2 1 2 1 a

2 1 4

=

3 1 3 1 b

3 1 8

0 1

0 1

0 1

0

13

14 6

a

34

5

=

, Solution is:

Then the best line is

6 4

b

14

52

y=

13

2

x

5

5

23. Generalize the least squares procedure to the situation in which data is given and you

desire to fit it with an expression of the form y = af (x) + bg (x) + c where the problem

would be to find a, b and c in order to minimize the error. Could this be generalized

to higher dimensions? How about more functions?

Say you had ordered pairs given which you wanted the curve to go through. Say these

are (xi , yi ) for i n. Then you would be finding least squares solutions for a, b, c to

yi = af (xi ) + bg (xi ) + c, i = 1, , n

You would simply obtain a least squares solutions to this. The principles are the same

for higher dimensions.

24. Let A L (X, Y ) where X and Y are finite dimensional vector spaces with the dimension of X equal to n. Define rank (A) dim (A (X)) and nullity(A) dim (ker (A)) .

r

Show that nullity(A) + rank (A) = dim (X) . Hint: Let {xi }i=1 be a basis for ker (A)

r

nr

nr

and let {xi }i=1 {yi }i=1 be a basis for X. Then show that {Ayi }i=1 is linearly

independent and spans AX.

m

Let {Ayi }i=1 be a basis for AX and let {xi }i=1 be a basis for ker (A).

Then {y1 , , ym , x1 , , xr } is linearly independent. In fact it spans X because if

z X, Then

m

X

Az =

ci Ayi

i=1

Pm

and so z i=1 ci yi ker (X) and so it is in the span of the {xi }i=1 . Therefore,

z is in the span of the given vectors and so this list of vectors is a basis. Hence the

dimension of X is equal to m + r = rank (A) + dim (ker (A)).

25. Let A be an mn matrix. Show the column rank of A equals the column rank of A A.

Next verify column rank of A A is no larger than column rank of A . Next justify the

following inequality to conclude the column rank of A equals the column rank of A .

rank (A) = rank (A A) rank (A )

= rank (AA ) rank (A) .

r

Hint: Start with an orthonormal basis, {Axj }j=1 of A (Fn ) and verify {A Axj }j=1

is a basis for A A (Fn ) .

Pr

Say you have i=1 ci A Axi = 0. Then for any appropriate y,

!

!

r

r

X

X

0=

ci A Axi , y =

ci Axi , Ay

i=1

i=1

145

Exercises

In particular, you could have y =

i ci xi .

r

X

ci Axi = 0

i=1

Axj and so A Ax is a linear combination of the A Axi which shows that these vectors

are a basis for A AFn . It follows that the rank of A and the rank of A A are the

same. Similarly, the rank of A and AA are the same. Hence the above inequalities

follow right away.

26. Let A be a real m n matrix and let A = QR be the QR factorization with Q

orthogonal and R upper triangular. Show that there exists a solution x to the equation

RT Rx = RT QT b

and that this solution is also a least squares solution defined above such that AT Ax =

AT b.

There exists a soluton to this last equation. Hence you have a solution to

RT QT QRx = RT QT b

But this is just the equations

RT Rx = RT QT b

Of course this kind of thing is very easy for a computer to solve.

F.31

Exercises

12.9

T

1. Here are three vectors in R4 : (1, 2, 0, 3) , (2, 1, 3, 2) , (0, 0, 1, 2) . Find the three

dimensional volume of the parallelepiped determined by these three vectors.

(1, 2, 0, 3) (1, 2, 0, 3) = 14

(1, 2, 0, 3) (2, 1, 3, 2) = 10

(1, 2, 0, 3) (0, 0, 1, 2) = 6

(2, 1, 3, 2) (2, 1, 3, 2) = 18

(2, 1, 3, 2) (0, 0, 1, 2) = 1

(0, 0, 1, 2) (0, 0, 1, 2) = 5

volume is

218

14 10 6

det 10 18 1 = 218

6 1 5

T

2. Here are two vectors in R4 : (1, 2, 0, 3) , (2, 1, 3, 2) . Find the volume of the parallelepiped determined by these two vectors.

(1, 2, 0, 3) (1, 2, 0, 3) = 14

(1, 2, 0, 3) (2, 1, 3, 2) = 10

146

Exercises

(2, 1, 3, 2) (2, 1, 3, 2) = 18

det

volume is

14 10

10 18

= 152

152

T

3. Here are three vectors in R2 : (1, 2) , (2, 1) , (0, 1) . Find the three dimensional

volume of the parallelepiped determined by these three vectors. Recall that from the

above theorem, this should equal 0.

It does equal 0.

4. Find the equation of the plane through the three points (1, 2, 3) , (2, 3, 1) , (1, 1, 7) .

5. Let T map a vector space V to itself. Explain why T is one to one if and only if T is

onto. It is in the text, but do it again in your own words.

This is because if T is one to one, it takes a basis to a basis. If T is onto, then it takes

a basis to a spanning set which must be independent, since otherwise, there would be

a linearly independent set of vectors which would also span and yet have fewer vectors

than a basis.

6. Let all matrices be complex with complex field of scalars and let A be an nn matrix

and B a m m matrix while X will be an n m matrix. The problem is to consider

solutions to Sylvesters equation. Solve the following equation for X

AX XB = C

where C is an arbitrary n m matrix. If q () is a polynomial, show first that if

AX XB = 0, then q (A) X Xq (B) = 0. Next define the linear map T which maps

the n m matrices to the n m matrices as follows.

T X AX XB

Show that the only solution to T X = 0 is X = 0 if and only if (A) (B) = .

Conclude that there exists a unique solution for each C to Sylvesters equation if and

only if (A) (B) = .

q (A) X Xq (B) = q (A) X = 0

Now explain why q (A)1 exists if and only if (A) (B) = .

Consider the first part. Multiply by An1 . Then An1 AX An1 XB = 0. Since

AX = XB, you can make multiple switches and write this as

An X XB n = 0

Now it follows that the desired result holds for a polynomial. Consider T which maps

the vector space of n m matrices to itself. Let q () be the characteristic polynomial

of B as suggested and suppose T X = 0. Then

q (A) X Xq (B) = 0 = q (A) X.

Qm

Let the characteristic polynomial for B be i=1 ( i ) . In case there are no shared

eigenvalues, it follows that

m

Y

q (A) =

(A i I)

i=1

147

Exercises

there is a shared eigenvalue, then the matrix q (A) will no longer be invertible. Hence

there will exist nonzero solutions X to q (A) X = 0. It follows that in this case T is

not one to one and so it is not onto either.

7. Compare Definition 12.8.2 with the Binet Cauchy theorem, Theorem 3.3.14. What is

the geometric meaning of the Binet Cauchy theorem in this context?

Letting U be the matrix having ui as the ith column the above definition states that

1/2

the volume is just det U T U

and by the Cauchy Binet theorem, in the interesting

case that U has more rows than columns, this equals

!

det (UI )

"1/2

where I is a choice of m rows which are selected from U so that UI is the square matrix

which results from these rows. The geometric meaning of the above expression is this.

det (UI ) gives the m dimensional volume of the parallelepiped which is obtained from

a projection onto the m dimensional subspace of Rn determined by setting the other

n m variables equal to 0. Thus the volume is the square root of the sum of the

squares of the areas of these projections.

F.32

Exercises

13.12

Since y is arbitrary, this shows (AB) x = B A x and since x is arbitrary, this shows

the formula.

This is just like the proof of the theorem.

3. Show that if A is an n n matrix which has an inverse then A+ = A1 .

A1 A = I which is also Hermitian.

4. Using the singular value decomposition, show that for any square matrix A, it follows

that A A is unitarily similar to AA .

You have A = U V where is the singular value matrix and U, V are unitary of the

right size. Therefore, A A = V U U V = V 2 V . Similarly, AA = U 2 U. Then

2 = U AA U and so

A A = V U AA U V

Since these matrices are all square, this does it. U V is unitary.

5. Let A, B be a m n matrices. Define an inner product on the set of m n matrices

by

(A, B)F trace (AB ) .

148

Exercises

Show this is an inner product

P satisfying all the inner product axioms. Recall for M an

n n matrix, trace (M ) ni=1 Mii . The resulting norm, ||||F is called the Frobenius

norm and it can be used to measure the distance between two matrices.

This was done earlier.

||A||2F (A, A)F =

2j

2

the singular values.

7. If A is a general n n matrix having possibly repeated eigenvalues, show there is a

sequence {Ak } of n n matrices having distinct eigenvalues which has the property

that the ij th entry of Ak converges to the ij th entry of A for all ij. Hint: Use Schurs

theorem.

A = U T U where T is upper triangular and U is unitary. Change the diagonal entries

of T slightly so that the resulting upper triangular matrix Tk has all distinct diagonal

entries and Tk T in the sense that the ij th entry of Tk converges to the ij th entry

of T . Then let Ak = U Tk U. It follows that Ak A in the sense that corresponding

entries converge.

8. Prove the Cayley Hamilton theorem as follows. First suppose A has a basis of eigenn

vectors {vk }k=1 , Avk = k vk . Let p () be the characteristic polynomial. Show

p (A) vk = p (k ) vk = 0. Then since {vk } is a basis, it follows p (A) x = 0 for all

x and so p (A) = 0. Next in the general case, use Problem 7 to obtain a sequence {Ak }

of matrices whose entries converge to the entries of A such that Ak has n distinct

eigenvalues and therefore by Theorem 7.1.7 Ak has a basis of eigenvectors. Therefore, from the first part and for pk () the characteristic polynomial for Ak , it follows

pk (Ak ) = 0. Now explain why and the sense in which

lim pk (Ak ) = p (A) .

n

that p (A) vk = p (k ) vk . Hence if x is any vector, let x = k=1 xk vk and it follows

that

!

n

n

X

X

p (A) x = p (A)

xk vk =

xk p (A) vk

k=1

n

X

xk p (k ) vk =

k=1

k=1

n

X

xk 0vk = 0

k=1

Hence p (A) = 0. Now drop the assumption that A is nondefective. From the above,

there exists a sequence Ak which is non defective which converges to A and also

pk () p () uniformly on compact sets because these characteristic polynomials

are defined in terms of determinants of the corresponding matrix. See the above

construction of the Ak . It is probably easiest to use the Frobinius norm for the last

part.

kpk (Ak ) p (A)kF kpk (Ak ) p (Ak )kF + kp (Ak ) p (A)kF

149

Exercises

The first term converges to 0 because the convergence of Ak to A implies all entries

of Ak lie in a compact set. The second term converges to 0 also because the entries of

Ak converge to the corresponding entries of A.

9. Prove that Theorem 13.4.6 and Corollary 13.4.7 can be strengthened so that the

condition

on

the Ak is necessary as well as sufficient. Hint: Consider vectors of the

x

form

where x Fk .

0

Say the matrix is positive definite. Then you could just look at

x

T

x

0 M (A)

= xT M (A)k x

0

and this needs to be positive. Hence those principle minors have all positive determinants.

10. Show directly that if A is an n n matrix and A = A (A is Hermitian) then all the

eigenvalues and eigenvectors are real and that eigenvectors associated with distinct

eigenvalues are orthogonal, (their inner product is zero).

(x, x) so = .

Now if x is an eigenvector,

(x, x) = (Ax, x) = (x,Ax) =

A

x =Ax =

x and so x + x

is also an eigenvector. Hence, you can assume that

the eigenvectors are real. If x

is pure imaginary, you could simply multiply by i and

get an eigenvector which is real.

11. Let v1 , , vn be an orthonormal basis for Fn . Let Q be a matrix whose ith column

is vi . Show

Q Q = QQ = I.

This follows from how we multiply matrices.

12. Show that an n n matrix Q is unitary if and only if it preserves distances. This

means |Qv| = |v| . This was done in the text but you should try to do it for yourself.

13. Suppose {v1 , , vn } and {w1 , , wn } are two orthonormal bases for Fn and suppose Q is an n n matrix satisfying Qvi = wi . Then show Q is unitary. If |v| = 1,

show there is a unitary transformation which maps v to e1 .

This is easy because you show it preserves distances.

14. Finish the proof of Theorem 13.6.5.

15. Let A be a Hermitian matrix so A = A and suppose all eigenvalues of A are larger

than 2 . Show

2

(Av, v) 2 |v|

Where here, the inner product is

(v, u)

n

X

vj uj .

j=1

(Ax, x) = (U DU x, x) = (DU x,U x) 2 |U x|2 = 2 |x|2

150

Exercises

16. Suppose A + A has all negative eigenvalues. Then show that the eigenvalues of A

have all negative real parts.

You have

0 >

=

((A + A ) x, x) = (Ax, x) + (A x, x)

(Ax, x) + (Ax, x)

2

|x| = Re () |x|

0 > |x| +

n1

1 X i 2

F (x) = z where zk =

e n jk xj .

n j=0

n1

1 X i 2 jk

F 1 (z) = x where xj =

e n zk

n j=0

2

2

2

ei n 00

ei n 10

ei n 20

2

2

2

ei n 11

ei n 21

ei n 01

2

2

1 ei 2

n 02

ei n 12

ei n 22

U=

n

..

..

..

.

.

.

2

2

2

ei n 0(n1) ei n 1(n1) ei n 2(n1)

where

2

ei n (n1)0

2

ei n (n1)1

2

ei n (n1)2

..

.

ei n (n1)(n1)

Now argue U is unitary and use this to establish the result. To show this verify

each row has length 1 and the inner product of two different rows gives 0. Now

2

2

Ukj = ei n jk and so (U )kj = ei n jk .

Take the inner product of two rows.

n1

X

ei n jk ei n jl

j=0

n1

X

ei n jk ei n jl =

j=0

n1

X

ei n j(lk)

j=0

1e

i 2

n n(lk)

2

1 ei n (lk)

= 0 if l 6= k

Pn1

2

because ei n n(lk) = 1. What if l = k? then the sum reduces to j=0 1n 1n = 1

and so this matrix is unitary. Therefore, its inverse is just the transpose conjugate

which yields the other formula.

18. Let f be a periodic function having period 2. The Fourier series of f is an expression

of the form

n

X

X

ikx

ck e lim

ck eikx

n

k=

k=n

and the idea is to find ck such that the above sequence converges in some way to f . If

f (x) =

ck eikx

k=

151

Exercises

and you formally multiply both sides by eimx and then integrate from 0 to 2,

interchanging the integral with the sum without any concern for whether this makes

sense, show it is reasonable from this to expect

Z 2

1

cm =

f (x) eimx dx.

2 0

Now suppose you only know f (x) at equally spaced points 2j/n for j = 0, 1, , n.

Consider the Riemann sum for this integral obtained

n from using the left endpoint of

the subintervals determined from the partition 2

j

. How does this compare with

n

j=0

the discrete Fourier transform? What happens as n to this approximation?

Multiply by eimx and formally integrate both sides. This leads to the formula for

cm . Now use a left sum. This would be

n1

1 X

2

2

2

im 2

j

n

f

j e

(j + 1)

j

2 j=0

n

n

n

=

n1

2

2

1X

f

j eim n j

n j=0

n

above where f is the vector whose j th component is f

1 f by the

n

2

n j / n.

unitary matrix U

having determinant 1. Show it must have an eigenvalue equal to 1. Note this shows

there exists a vector x 6= 0 such that Ax = x. Hint: Show first or recall that any

orthogonal matrix must preserve lengths. That is, |Ax| = |x| .

If Ax = x, then you can take the norm of both sides and conclude that || = 1. It

follows that the eigenvalues of A are ei , ei and another one which has magnitude

1 and is real. This can only be 1 or 1. Since the determinant is given to be 1, it

follows that it is 1. Therefore, there exists an eigenvector for the eigenvalue 1.

20. Let A be a complex m n matrix. Using the description of the Moore Penrose inverse

in terms of the singular value decomposition, show that

1

lim (A A + I)

0+

A = A+

where the convergence happens in the Frobenius norm. Also verify, using the singular

value decomposition, that the inverse exists in the above formula.

Recall that the Moore Penrose inverse is

1

0

V

U

0

0

0

where A = U

V . The matrices U, V are unitary and of the right size. First

0 0

of all, observe that there is no problem about the existence of the inverse mentioned

in the formula. This is because

((A A + I) x, x) |x|

and so it is a one to one matrix. Now consider the limit assertion. The left side equals

152

Exercises

1

0

0

V

UU

V + I

V

0 0

0 0

0

0+

2

1

0

0

= lim V

V + I

V

U

0 0

0 0

0

lim

Now

2

V

0

=

=

0

0

2

V

0

2

V

0

= V

= V

0

0

0

0

V + I

V + V V

+ I

2 + I 0

0

I

! 2

1

+ I

0

=V

1

1

1

0

0

0

0

0

0

0

0

1

0

U

0 0

"

0

0

U

0 0

1 I

!

1

2 + I

0

0

!

1

Of course and 2 and so forth are diagonal matrices. Thus 2 + I

is also a

! 2

1

diagonal matrix having on the diagonal an expression of the form i +

i and

this clearly converges as 0 to 1

.

Therefore,

the

above

converges

to

i

1

0

V

U = A+

0

0

+

21. Show that A+ = (A A) A . Hint: You might use the description of A+ in terms of

the singular value decomposition.

1

0

0

Recall that A+ = V

U where A = U

V . The was the

0

0

0 0

diagonal matrix which consisting of square roots of eigenvalues of A A, arranged in

decreasing order from top left toward lower right.

2

0

A A = V

V

0 0

2

2

1

0

0

V

V V

U =V

U = A+

0

0

0 0

0

0

153

Exercises

F.33

Exercises

14.7

1. Solve the system

using the

solving it

4 1

1 5

0 2

1 1

x

1

5 2 y = 2

2 6

z

3

4

1

0

1 1

x

1

7 2 y = 2

2 4

z

3

1 1

x

1

7 2 y = 2

2 4

z

3

Gauss Seidel method and the Jacobi method. Check your answer by also

using row operations.

5

1

x

1

94

7

2 y = 2 , Solution is: 94

67

4

z

3

94

using the

solving it

5 1

1 7

0 2

4

1

0

Gauss Seidel method and the Jacobi method. Check your answer by also

using row operations.

9

1

x

1

100

21

2 y = 2 , Solution is: 100

43

6

z

3

100

using the

solving it

4 1

1 7

0 2

5

1

0

Gauss Seidel method and the Jacobi method. Check your answer by also

using row operations.

5

1

x

1

118

9

2 y = 2 , Solution is: 118

42

4

z

3

59

4. If you are considering a system of the form Ax = b and A1 does not exist, will either

the Gauss Seidel or Jacobi methods work? Explain. What does this indicate about

finding eigenvectors for a given eigenvalue?

No. These methods only work when there is a unique solution.

5. For ||x|| max {|xj | : j = 1, 2, , n} , the parallelogram identity does not hold.

Explain.

Let x = (1, 0) , y = (0, 1) . Then

kx + yk2 + kx yk2

2

2 kxk + 2 kyk

= 1+1=2

= 4

x + y

2 < ||x|| = ||y|| .

154

Exercises

Show the norm || which comes from an inner product is strictly convex.

Let x, y be as described.

If x 6= y, then

x + y
2
x y
2

1

1

2

2

2

+

2

2
= 2 kxk + 2 kyk = kxk

x + y
2

2

2
< kxk

7. A norm |||| is said to be uniformly convex if whenever ||xn || , ||yn || are equal to 1 for

all n N and limn ||xn + yn || = 2, it follows limn ||xn yn || = 0. Show the

norm || coming from an inner product is always uniformly convex. Also show that

uniform convexity implies strict convexity which is defined in Problem 6.

The usual norm satisfies the parallelogram law. Thus

|xn yn |

=

=

4 |xn + yn |

8. Suppose A : Cn Cn is a one to one and onto matrix. Define

||x|| |Ax| .

Show this is a norm.

If kxk = 0, then Ax = 0 ans since A is one to one, it follows that x = 0. Clearly

kcxk = |c| kxk. What about the triangle inequality? This is also easy.

kx + yk |A (x + y)| |Ax| + |Ay| = kxk + kyk

9. If X is a finite dimensional normed vector

space

and A, B L (X, X) such that

||B|| < ||A|| , can it be concluded that A1 B < 1?

1 0

1/2 0

1 0

1

Maybe not. Say A =

,B =

. Then A =

and

0 1/2

0 3/4

0 2

1

0

2

A1 B =

so it has norm 3/2 > 1. Also kBk < kAk.

0 32

10. Let X be a vector space with a norm |||| and let V = span (v1 , , vm ) be a finite

dimensional subspace of X such that {v1 , , vm } is a basis for V. Show V is a closed

subspace of X. This means that if wn w and each wn V, then so is w. Next show

that if w

/ V,

dist (w, V ) inf {||w v|| : v V } > 0

is a continuous function of w and

|dist (w, V ) dist (w1 , V )| kw1 wk

Next show that if w

/ V , there exists z such that ||z|| = 1 and dist (z, V ) > 1/2.

For those who know some advanced calculus, show that if X is an infinite dimensional

vector space having norm |||| , then the closed unit ball in X cannot be compact. Thus

closed and bounded is never compact in an infinite dimensional normed vector space.

155

Exercises

P

1/2

Pm

m

2

Say wn = k=1 cnk vk . Define a norm on V by kwk

where w =

k=1 |ck |

Pm

c

v

.

This

gives

a

norm

on

V

which

is

equivalent

to

the

given

norm

on V .

k=1 k k

Pm

Thus if wn w, it follows that cn c in Cm and so w =

c

v

and

so

k=1 k k

V is closed. Let w, w be two points and suppose without loss of generality that

dist (w, V ) dist (w , V ) 0. Let v V be such that kw vk < dist (w , V ) + .

Then

|dist (w, V ) dist (w , V )| =

kw vk kw vk +

kw w k + kw vk kw vk +

kw w k +

Then since is arbitrary, this shows that |dist (w, V ) dist (w , V )| kw w k and

so this is continuous as claimed. Now let w

/ V. Thus dist (w, V ) > 0 because if this

is not so, there wouild exist vn w and by the first part, w V . Let v V be such

wv

that 2 dist (w, V ) > kw vk. Then you can consider z = kwvk

. If v1 V,

kz v1 k =

wv

v1 kw vk

kw vk

kw vk

1

kw v kw vk v1 k

2 dist (w, V )

dist (w, V )

1

=

2 dist (w, V )

2

Since v1 is arbitrary, it follows that dist (z, V ) 12 . If you have an infinite dimensional

normed linear space, the above argument shows that you can obtain an infinite sequence of vectors {xn } , each kxn k = 1 and such that dist (xn+1 , span (x1 , , xn ))

1/2 for every n. Therefore, there is no convergent subsequence because no subsequence

is a Cauchy sequence. Therefore, the unit ball is not sequentially compact.

11. Suppose (A) < 1 for A L (V, V ) where V is a p dimensional vector space having

a norm ||||. You can use Rp or Cp if you like. Show there exists a new norm ||||||

such that with respect to this new norm, |||A||| < 1 where |||A||| denotes the operator

norm of A taken with respect to this new norm on V ,

|||A||| sup {|||Ax||| : |||x||| 1}

Hint: You know from Gelfands theorem that

||An ||

1/n

<r<1

provided n is large enough, this operator norm taken with respect to ||||. Show there

exists 0 < < 1 such that

A

< 1.

You can do this by arguing the eigenvalues of A/ are the scalars / where (A).

Now let Z+ denote the nonnegative integers.

n

A

|||x||| sup n x

nZ+

156

Exercises

First show this is actually a norm. Next explain why

n+1

A

|||Ax||| sup n+1 x |||x||| .

nZ+

A

<1

Just pick larger than the absolute value of all the eigenvalues which are each less

than 1. Now use the norm suggested. First of all, why is this a norm? If |||x||| = 0,

then kxk = 0 because

kxk |||x|||

It is also clear that |||cx||| = |c| |||x|||. What about the triangle inequality?

n

n
n

A

A
A

|||x + y||| sup
n (x + y)
sup

n x
+
n y

nZ+

nZ+

n

n

A

A

n y
|||x||| + |||y|||

+

sup

sup

x

n

nZ

nZ

+

Now why is the operator norm with respect to this new norm less than 1?

n

n+1

A

A

|||Ax||| sup n Ax = sup n+1 x |||x|||

nZ+

nZ+

12. Establish a similar result to Problem 11 without using Gelfands theorem. Use an

argument which depends directly on the Jordan form or a modification of it.

(A) < 1 and A = S 1 J S where J is in modified Jordan form having or 0 on the

super diagonal the diagonal entries of every block being less than 1 in absolute value.

Let < 1 be larger than the absolute values of all eigenvalues. Now what if you define

|||x||| kSxk ? Then

|||Ax|||

kSAxk = kJ Sxk

max i (Sx)i + (Sx)i+1

i

13. Using Problem 11 give an easier proof of Theorem 14.6.6 without having to use Corollary 14.6.5. It would suffice to use a different norm of this problem and the contraction

mapping principle of Lemma 14.6.4.

The suggestion gives it away. You were given B 1 C < 1. Now with the above

problem, there is another norm such that with respect to this other norm, B 1 C is a

contraction mapping. Therefore, it has a fixed point by the easier result on contraction

maps.

157

Exercises

P

14. A matrix A is diagonally dominant if |aii | > j6=i |aij | . Show that the Gauss Seidel

method converges if A is diagonally dominant.

It is the eigenvalues of B 1 C which are important. Say

B 1 Cx = x

Then you must have

det (B C) = 0

Illustrating what happens in the Gauss Seidel method in three dimensions, this requires

det a21 a22 a23 = 0

a31 a32 a33

In other words the above matrix must have a zero determinant. However, since the

original matrix was given to be diagonally dominant, this matrix cannot have a zero

determinant unless || < 1.

P

15. Suppose f () = k=0 an n converges if || < R. Show that if (A) < R where A is

an n n matrix, then

X

f (A)

an An

k0

converges in L (Fn , Fn ) . Hint: Use Gelfands theorem and the root test.

Since the given series converges if || < R, it follows that

lim sup |an n |

1/n

<1

and so

lim sup |an |

1/n

|| < 1 if || < R

lim sup |an |

1/n

n 1/n

1

R

1/n

1/n

kAn k

1/n

<

1

R=1

R

n

n

1/n

1/n

16. Referring to Corollary 14.4.3, for = a + ib show

exp (t) = eat (cos (bt) + i sin (bt)) .

Hint: Let y (t) = exp (t) and let z (t) = eat y (t) . Show

z + b2 z = 0, z (0) = 1, z (0) = ib.

158

Exercises

Now letting z = u + iv where u, v are real valued, show

u + b2 u

= 0, u (0) = 1, u (0) = 0

v + b2 v

= 0, v (0) = 0, v (0) = b.

Next show u (t) = cos (bt) and v (t) = sin (bt) work in the above and that there is at

most one solution to

w + b2 w = 0 w (0) = , w (0) = .

Thus z (t) = cos (bt) + i sin (bt) and so y (t) = eat (cos (bt) + i sin (bt)). To show there

is at most one solution to the above problem, suppose you have two, w1 , w2 . Subtract

them. Let f = w1 w2 . Thus

f + b2 f = 0

and f is real valued. Multiply both sides by f and conclude

!

2

2

d (f )

2f

+b

=0

dt

2

2

Thus the expression in parenthesis is constant. Explain why this constant must equal

0.

The first claim is a routine computation. The next claim is also routine computations.

What about the uniqueness assertion? It suffices to show that if , = 0 then the

only solution is 0. Multiply both sides of the differential equation by w . Then you get

It follows that

1 d

b2 d " 2

2

(w ) +

w =0

2 dt

2 dt

2

(w ) + w2

is a constant. From the initial conditions, this constant can only be 0. The rest follows.

17. Let A L (Rn , Rn ) . Show the following power series converges in L (Rn , Rn ).

k k

X

t A

k=0

k!

You might want to use Lemma 14.4.2. This is how you can define exp (tA). Next show

using arguments like those of Corollary 14.4.3

d

exp (tA) = A exp (tA)

dt

so that this is a matrix valued solution to the differential equation and initial condition

(t) = A (t) , (0) = I.

This (t) is called a fundamental matrix for the differential equation y = Ay. Show

t (t) y0 gives a solution to the initial value problem

y = Ay, y (0) = y0 .

159

Exercises

The series converges absolutely by the ratio test and the observation that
Ak

k

kAk . Now we need to show that you can differentiate it. Writing the difference

quotient gives

k

k

k1 k

k1 k

(t

+

h)

t

Ak

X

X

X

1

ks (k, h)

A

s (k, h)

A

=

=

h

k!

k!

(k 1)!

k=0

k=1

k=1

where s (k, h) is between t and t + h. Thus for each k, limh0 s (k, h) = t. I want to

P k1 Ak

show that the limit of the series is k=1 t(k1)!

.

k1 k

X

s (k, h)

A

k=1

(k 1)!

k1

s (k, h)

tk1 Ak

X

X

tk1 Ak

=

(k 1)!

(k 1)!

k=1

k=1

X

(k 1) p (k, h)k2 (s (k, h) t) Ak

(k 1)!

k=1

X

p (k, h)k2 (s (k, h) t) Ak

(k 2)!

k=2

where p (k, h) is also between t and t + h. Now the norm of this is dominated by

|h|

k2

k

X

|p (k, h)|

kAk

(k 2)!

k=2

k1 k

X

s (k, h)

A

k=1

(k 1)!

X

tk1 Ak

0

(k 1)!

k=1

k1 k1

X

X

tk1 Ak

t

A

=A

= A exp (At)

(k 1)!

(k 1)!

k=1

k=1

When you plug in t = 0, you get I. Therefore, this is the solution to the given

differential equation.

18. In Problem 17 (t) is defined by the given series. Denote by exp (t (A)) the numbers

exp (t) where (A) . Show exp (t (A)) = ( (t)) . This is like Lemma 14.4.7.

Letting J be the Jordan canonical form for A, explain why

(t)

k k

X

t A

k=0

k!

=S

k k

X

t J

k=0

k!

S 1

and you note that in J k , the diagonal entries are of the form k for an eigenvalue

of A. Also J = D + N where N is nilpotent and commutes with D. Argue then that

k k

X

t J

k=0

k!

is an upper triangular matrix which has on the diagonal the expressions et where

(A) . Thus conclude

( (t)) exp (t (A))

160

Exercises

Next take et exp (t (A)) and argue it must be in ( (t)) . You can do this as

follows:

(t) et I

k k

k k

X

X

t A

t

I

k!

k!

k=0

k

X

k=0

k k1

X

t X

k=0

k!

k k1

X

t X

k=0

k=0

t k

A k I

k!

k!

j=1

Akj j (A I)

(6.38)

Akj j

j=1

converges to something in L (R , R ). To do this, use the ratio test and Lemma 14.4.2

after first using the triangle inequality. Since (A) , (t) et I is not one to one

and so this establishes the other inclusion. You fill in the details. This theorem is a

special case of theorems which go by the name spectral mapping theorem.

P (tA)k x

If (A) , then Ax = x and so

= et x and so exp (t (A))

k=0

k!

(exp (tA)) . Now consider exp (tA) . You have A = S 1 JS where J is Jordan form.

Thus it is simple to write that

exp (tA) = S 1

k

X

(tJ)

k=0

k!

Now J is block diagonal and the blocks have a constant down the main diagonal. Say

..

.

. ..

When you raise such a block to the exponent k, you get an upper triangular matrix

which has k down the diagonal. Therefore, exp (tA) is of the form

S 1 BS

where B is an upper triangular matrix which has for its diagonal entries et for

(A). These diagonal entries are the eigenvalues of exp (tA). It follows that

(exp (tA)) exp (t (A)) .

19. Suppose (t) L (V, W ) where V, W are finite dimensional inner product spaces and

t (t) is continuous for t [a, b]: For every > 0 there there exists > 0 such that

if |s t| < then || (t) (s)|| < . Show t ( (t) v, w) is continuous. Here it is

the inner product in W. Also define what it means for t (t) v to be continuous

and show this is continuous. Do it all for differentiable in place of continuous. Next

show t || (t)|| is continuous.

161

Exercises

k (t) v (s) vk k (t) (s)k |v|

so t (t) v is continuous. Differentiable works out similarly. If t (t) is

differentiable, then those other things are too. For example,

k (t + h) v (t) v (t) hvk

o (h) |v| = o (h)

As to continuity of t k (t)k ,

|k (t)k k (s)k| k (t) (s)k

from the triangle inequality.

20. If z (t) W, a finite dimensional inner product space, what does it mean for t z (t)

to be continuous or differentiable? If z is continuous, define

Z

as follows.

w,

z (t) dt W

z (t) dt

a

Show that this definition is well defined and furthermore the triangle inequality,

Z

Z

b

b

z (t) dt

|z (t)| dt,

a

a

and fundamental theorem of calculus,

Z t

d

z (s) ds = z (t)

dt

a

hold along with any other interesting properties of integrals which are true.

Differentiability is the same as usual.

z (t + h) z (t) z (t) h = o (h)

If t z (t) is continuous, then from the above problem, so is t (z (t) , w) and so

Rb

(z (t) , w) dt makes sense. Does there exist an element I of the inner product space

a

Rb

such that (w, I) = a (w, z (t)) dt? This is so in a finite dimensional inner product

space because

Z b

w

(w, z (t)) dt

a

162

Exercises

is a linear map and so it can be represented by a unique I W so that

Z b

(w, I) =

(w, z (t)) dt

a

!

R t+h

Z

z (s) ds

1 t+h

t

w,

h

h t

Since the space is finite dimensional, this is the same as saying that

R t+h

z (s) ds

lim t

= z (t) .

h0

h

You can just apply the above to the finitely many Fourier coefficients. Therefore, the

usual Fundamental theorem of calculus holds. What about the triangle inequality?

Z

!

Z b

b

z (t) dt = sup w,

z (t) dt

a

|w|=1

a

Z

Z b

b

(w, z (t)) dt sup

|(w, z (t))| dt

= sup

|w|=1 a

|w|=1 a

Z b

|z (t)| dt.

a

Z b

(t) dt L (V, W )

a

as follows.

w,

(t) dt (v)

Rb

Show this is well defined and does indeed give a (t) dt L (V, W ) . Also show the

triangle inequality

Z

Z

b

b

(t) dt

|| (t)|| dt

a

a

where |||| is the operator norm and verify the fundamental theorem of calculus holds.

Z t

(s) ds = (t) .

a

Also verify the usual properties of integrals continue to hold such as the fact the

integral is linear and

Z b

Z c

Z c

(t) dt +

(t) dt =

(t) dt

a

and similar things. Hint: On showing the triangle inequality, it will help if you use

the fact that

|w|W = sup |(w, v)| .

|v|1

163

Exercises

You should show this also.

That last assertion follows by the Cauchy Schwarz inequality which shows the right

side is no larger than |w| and then by taking v = w/ |w|. That the integral is well

Rb

defined follows because w a (w, (t) v) dt is a linear mapping and so, by the

Riesz representation theorem, there exists a unique I (v) W such that (w, I (v)) =

Rb

(w, (t) v) dt. Now also, I is a linear function of v because of the fact that each

a

Rb

(t) is linear. Therefore, you can denote by a (t) dt L (V, W ) this I. Thus the

definition of the integral is well defined. The standard properties of the integral are

now fairly obvious. As for the triangle inequality, it goes the same way as in the above

problem.

Z

Z

!

Z b

b

b

(t) dt
sup

(t) dt (v)
= sup sup w,

(t) dt (v)

a

|v|1

|v|1 |w|1

a

a

Z

Z

b

b

= sup sup

(w, (t) (v)) dt

k (t)k dt.

|v|1 |w|1 a

a

Now consider the fundamental theorem of calculus. For simplicity, let h 0 + .

Z

Z

1 t+h

1 t+h

(s) ds (t)
=

(s) (t) ds

h t

h t

Z t+h

1

k (s) (t)k ds

h t

22. Prove Gronwalls inequality. Suppose u (t) 0, continuous, and for all t [0, T ] ,

u (t) u0 +

Ku (s) ds.

u (t) u0 eKt .

Rt

Hint: w (t) = 0 u (s) ds. Then using the fundamental theorem of calculus, w (t)

satisfies the following.

u (t) Kw (t) = w (t) Kw (t) u0 , w (0) = 0.

Now use the usual techniques you saw in an introductory differential equations class.

Multiply both sides of the above inequality by eKt and note the resulting left side is

now a total derivative. Integrate both sides from 0 to t and see what you have got. If

you have problems, look ahead in the book. This inequality is proved later in Theorem

C.4.3.

Following the hint,

" Kt

we

u0 eKt

and therefore,

w (t) e

Kt

u0

Ks

ds = u0

1

eKt

K

K

164

Exercises

Therefore, w (t) u0

eKt

K

1

K

u (t) u0 + Kw (t)

u (t) u0 + K u0

eKt

1

K

K

= u0 eKt

23. With Gronwalls inequality and the integral defined in Problem 21 with its properties

listed there, prove there is at most one solution to the initial value problem

y = Ay, y (0) = y0 .

Hint: If there are two solutions, subtract them and call the result z. Then

z = Az, z (0) = 0.

It follows

z (t) = 0+

Az (s) ds

and so

||z (t)||

From Gronwalls inequality,

kz (t)k 0ekAkt = 0.

24. Suppose A is a matrix which has the property that whenever (A) , Re < 0.

Consider the initial value problem

y = Ay, y (0) = y0 .

The existence and uniqueness of a solution to this equation has been established above

in preceding problems, Problem 17 to 23. Show that in this case where the real parts

of the eigenvalues are all negative, the solution to the initial value problem satisfies

lim y (t) = 0.

Hint: A nice way to approach this problem is to show you can reduce it to the

consideration of the initial value problem

z = J z, z (0) = z0

where J is the modified Jordan canonical form where instead of ones down the main

diagonal, there are down the main diagonal. You have y = S 1 J Sy and so you

just let z = Sy. Then z 0 if and only if y 0.

Then

z = Dz + N z

165

Exercises

where D is the diagonal matrix obtained from the eigenvalues of A and N is a nilpotent

matrix commuting with D which is very small provided is chosen very small. Now

let (t) be the solution of

= D, (0) = I

described earlier as

X

(1)k tk Dk

.

k!

k=0

The reason these commute is that D is block diagonal, each block having constant

entries down the diagonal while the same is true of N . The blocks associated with N

commute with the blocks associated with D and so D and N commute. Thus (t)

commutes with D and N .

Next argue

(t) z (t) + (t) z (t) =

=

(t) N z (t)

Then integrating,

(t) z (t) z0 =

Z t

|| (t) z (t)|| ||z0 || +

||N || || (s) z (s)|| ds.

0

|| (t) z (t)|| ||z0 || e||N ||t

Now look closely at the form of (t) to get an estimate which is interesting. Explain

why

t

e 1

0

..

(t) =

.

n t

0

e

and now observe that if is chosen small enough, ||N || is so small that each component

of z (t) converges to 0.

It follows right away from the definition of (t) that (t) is given by the above

diagonal matrix where here the i are negatives of the eigenvalues of A. Thus their

real parts are all bounded below by some > 0. It follows that

|( (t) x)i | et |xi |

and so

|| (t) z (t)|| et kz (t)k

166

Exercises

Therefore, from the above inequality,

kz (t)k kz0 k et + et

Of course the right side converges to 0 if you choose small enough that kN k < .

25. Using Problem 24 show that if A is a matrix having the real parts of all eigenvalues

less than 0 then if

(t) = A (t) , (0) = I

it follows

lim (t) = 0.

From the above, problem, this happens. If is just defined, then the solution to

xi = Axi (t) , xi (0) = ei is (t) ei . This follows from the fact that (t) ei satisfies the

equation and uniqueness. Thus the ith column of (t) converges to 0. In particular

the entries of (t) converge to 0. Thus k (t)k 0 also.

26. Let (t) be a fundamental matrix satisfying

(t) = A (t) , (0) = I.

n

Show (t) = (nt) . Hint: Subtract and show the difference satisfies = A, (0) =

0. Use uniqueness.

27. If the real parts of the eigenvalues of A are all negative, show that for every positive

t,

lim (nt) = 0.

n

Hint: Pick Re ( (A)) < < 0 and use Problem 18 about the spectrum of (t)

and

for the spectral radius along with Problem 26 to argue that

theorem

Gelfands

(nt) /ent < 1 for all n large enough.

n

P

28. Let H be a Hermitian matrix. (H = H ) . Show that eiH n=0 (iH)

is unitary.

n!

You have H = U DU where U is unitary and D is a real diagonal matrix. Then you

have

i

e 1

n

X

(iD)

..

eiH = U

U = U

U

.

n!

n=0

ein

and this is clearly unitary because each matrix in the product is.

29. Show the converse of the above exercise. If V is unitary, then V = eiH for some H

Hermitian.

To do this, note first that V is normal because V V = V V = I. Therefore, there

exists a unitary U such that V = U DU where D is the diagonal matrix consisting

of the eigenvalues of V down the main diagonal. Since V is unitary, it preserves all

lengths and so each diagonal entry of D is of magnitude 1. Thus you have

i

e 1

k

X

(iD)

.

.

V =U

U

U = U

.

k!

k=0

ein

167

Exercises

where D =

..

.

n

k

X

(iU DU )

k=0

= eiH

k!

1

that

H = i (I U ) (I + U )1

exists, show

U = (I + iH) (I iH)

To see this, note first that (I U ) , (I + U ) commute. Taking the adjoint of the

above operator, it is easy to see that this equals

i (I + U )

(I U )

= i (U U + U )

1

= i (U + I)

= i (I + U )

(U U U )

U U (U I)

(I U ) = i (I U ) (I + U )

1

1

1

I + i i (I U ) (I + U )

I i i (I U ) (I + U )

1

1

1

=

I (I U ) (I + U )

I + (I U ) (I + U )

1

1

1

= (I + U (I U )) (I + U )

(I + U + I U ) (I + U )

1

1

1

1 1

= 2U (I + U )

2I (I + U )

= 2U (I + U )

(I + U ) = U

2

31. Suppose that A L (V, V ) where V is a normed linear space. Also suppose that

kAk < 1 where this refers to the operator norm on A. Verify that

(I A)1 =

Ai

i=0

This is called the Neumann series. Suppose now that you only know the algebraic

1

condition (A) < 1. Is it still the case that the Neumann series converges to (I A) ?

Consider partial sums of the series.

q

q

q

X i
X

i
X

i

A

A

kAk

i=p
i=p

i=p

P

i

1

which converges to 0 as p, q because i=0 kAk = 1kAk

< . Therefore, by

completeness of L (V, V ) , this series converges. Why is is the desired inverse? This is

obvious when you multiply both sides by (I A) . Thus

(I A)

X

i=0

Ai = lim (I A)

n

n

X

i=0

Ai = lim I An+1 = I

n

168

Exercises

n+1

because
An+1
kAk

and this converges to 0 as n because kAk < 1. Yes,

it the series still converges. Let (A) < r < R < 1. Then for all n large enough, you

have

1/n

kAn k

<R

and so kAn k < Rn . It follows that the partial sums of the series form a Cauchy

sequence and so the series converges.

F.34

Exercises

15.3

1. In Example 15.1.10 an eigenvalue was found correct to several decimal places along

with an eigenvector. Find the other eigenvalues along with their eigenvectors.

1 2 3

The matrix was 2 2 1

3 1 4

20

1 2 3

8. 485 7 1015 6. 190 4 1015 1. 188 8 1016

2 2 1 = 6. 190 4 1015 4. 515 9 1015 8. 672 7 1015

3 1 4

1. 188 8 1016 8. 672 7 1015 1. 665 6 1016

= 0.390 23 0.598 82 0.699 38

0.749 39 0.647 89

0.136 60

16

1. 586 4 10

1. 157 3 1016 2. 222 5 1016

0

2. 212 9 1011 8. 403 1 1011

0

0

4. 160 5 1011

1 2 3

0.390 23 0.598 82 0.699 38 2 2 1

0.749 39 0.647 89

0.136 60

3 1 4

0.390 23 0.598 82 0.699 38

0.749 39 0.647 89

0.136 60

6. 662 1

1. 218 8 104 7. 374 9 105

1. 139 5

1. 156 9

= 1. 218 8 104

7. 374 9 105

1. 156 9

0.801 49

0.534 92

The largest eigenvalue is close to 6. 662 1 and an eigenvector is 0.390 23 . The

0.749 39

others can be found by looking at the lower right block. You could do the same as the

above or you could just graph the characteristic function and zoom in on the roots.

They are approximately 1.7 and 1.3. Next use shifted inverse power method to find

them more exactly and to also find the eigenvectors.

1 2 3

1 0 0

0.977 92 5. 047 3

3. 47

2 2 1 1.7 0 1 0 = 5. 047 3 33. 47

21. 136

3 1 4

0 0 1

3. 47

21. 136 13. 281

169

Exercises

Rather than use this method as described, I will just use the variation of it based on

the QR algorithm which involves raising the matrix to a power which was just used

to find the first eigenvalue.

20

0.977 92 5. 047 3

3. 47

5. 047 3 33. 47

21. 136 =

3. 47

21. 136 13. 281

6. 044 8 1031

3. 893 4 1032 2. 458 8 1032

3. 893 4 1032

2. 507 7 1033 1. 583 7 1033 :

32

2. 458 8 10

1. 583 7 1033 1. 000 2 1033

0.990 94

0.838 32

0.529 98

0.127 88

2

0.529 42

0.847 36

4. 112 3 10

32

33

4. 644 3 10

2. 991 4 10

1. 889 2 1033

0

9. 738 8 1027 3. 859 8 1028

0

0

3. 342 9 1027

T

0.130 15 3. 322 8 102

0.990 94

0.838 32

0.529 98

0.127 88

2

0.529 42

0.847 36

4. 112 3 10

0.977 92 5. 047 3

3. 47

5. 047 3 33. 47

21. 136

3. 47

21. 136 13. 281

0.990 94

0.838 32

=

0.529 98

0.127 88

2

0.529 42

0.847 36

4. 112 3 10

47. 602

6. 755 1 105 2. 994 7 105

6. 755 1 105 6. 320 5 102

0.232 90

5

2. 994 7 10

0.232 90

0.190 38

1

1.7

0.130 15

suitable eigenvector is just the first column of the orthogonal matrix 0.838 32 .

0.529 42

How

well

does

it

work?

1 2 3

0.130 15

0.218 53

2 2 1 0.838 32 = 1. 407 5

3 1 4

0.529 42

0.888 91

0.130 15

0.218 52

0.838 32 (1. 679 0) = 1. 407 5

0.529 42

0.888 90

1

1 2 3

1 0 0

2 2 1 + 1.3 0 1 0

3 1 4

0 0 1

8. 119 2

= 7. 810 9 3. 278 5 3. 802 7

8. 119 2 3. 802 7 3. 689 6

170

Exercises

20

16. 948 7. 810 9

8. 119 2

7. 810 9 3. 278 5 3. 802 7 =

8. 119 2 3. 802 7 3. 689 6

1. 745 5 1027 7. 960 3 1026

8. 313 7 1026 :

27

26

1. 823 0 10

8. 313 7 10

8. 682 7 1026

0.380 73 0.461 03

0.801 56

0.397 63 0.700 98

0.592 05

27

4. 584 7 10

2. 090 8 1027 2. 183 7 1027

0

1. 658 9 1022

2. 792 9 1022

0

0

4. 620 1 1021

T

0.834 83 0.544 13 8. 355 9 102

0.380 73 0.461 03

0.801 56

0.397 63 0.700 98

0.592 05

8. 119 2

7. 810 9 3. 278 5 3. 802 7

8. 119 2 3. 802 7 3. 689 6

0.380 73 0.461 03

0.801 56

0.397 63 0.700 98

0.592 05

24. 377

9. 597 9 105

5. 267 105

0.127 02

1. 802 1 102

= 9. 597 9 105

5

2

5. 267 10

1. 802 1 10

0.334 17

1

+1.3

= 24. 377, Solution is: 1. 341. The approximate eigenvector is the first column.

How

well

does it work?

1 2 3

0.834 83

1. 119 5

2 2 1 0.380 73 = 0.510 57

3 1 4

0.397 63

0.533 24

0.834 83

1. 119 5

0.380 73 (1. 341) = 0.510 56

0.397 63

0.533 22

It works well. You could do more iterations if desired.

3 2 1

2. Find the eigenvalues and eigenvectors of the matrix A = 2 1 3 numerically.

1 3 2

In this case the exact eigenvalues are 3, 6. Compare with the exact answers.

13

3 2 1

4. 353 6 109 4. 353 6 109 4. 353 6 109

2 1 3 = 4. 353 6 109 4. 353 6 109 4. 353 6 109 =

1 3 2

4. 353 6 109 4. 353 6 109 4. 353 6 109

0.577 35 0.408 25

0.707 11

0.577 35 0.408 25

0.707 11

7. 540 7 109

7. 540 7 109

7. 540 7 109

0

1. 533 3 1019 1. 533 3 1019

0

0

2. 736 9 1048

Exercises

171

3 2 1

0.577 35 0.408 25

2 1 3

0.707 11

0.577 35 0.408 25

0.707 11

1 3 2

29

0.577 35 0.816 50 1. 437 8 10

0.577 35 0.408 25

0.707 11

0.577 35 0.408 25

0.707 11

6. 000 0

0

0

1. 5

0.866 03

= 1. 262 2 1029

29

2. 524 4 10

0.866 03

1. 5

0.577 35

An eigenvalue is 6 and an eigenvector is 0.577 35 . Now lets find the others.

0.577 35

1

3 2 1

1 0 0

2 1 3 1.5 0 1 0

1 3 2

0 0 1

2. 740 7

0.592 59

1. 925 9

= 0.592 59 7. 407 4 102 0.740 74

1. 925 9

0.740 74

1. 407 4

20

2. 740 7

0.592 59

1. 925 9

0.592 59 7. 407 4 102 0.740 74 =

1. 925 9

0.740 74

1. 407 4

12

3. 034 1 10

8. 129 9 1011 2. 221 1 1012

8. 129 9 1011 2. 178 4 1011

5. 951 5 1011 :

12

11

2. 221 1 10

5. 951 5 10

1. 626 0 1012

0.788 68

0.425 47 0.443 81

0.211 33 0.490 28 0.845 56

0.577 35 0.760 66 0.296 76

0

3. 850 7 106

3. 843 7 107

0

0

1. 927 6 107

0.788 68

0.425 47 0.443 81

2. 740 7

0.592 59

1. 925 9

0.211 33 0.490 28 0.845 56 0.592 59 7. 407 4 102 0.740 74

0.577 35 0.760 66 0.296 76

1. 925 9

0.740 74

1. 407 4

0.788 68

0.425 47 0.443 81

0.211 33 0.490 28 0.845 56

0.577 35 0.760 66 0.296 76

4. 309 4

9. 614 1 106 1. 276 5 105

= 9. 614 1 106

0.223 59

0.195 59

1. 276 5 105

0.195 59

0.136 42

0.788 68

1

0.211 33 . How well

1.5 = 4. 309 4, Solution is: 1. 732 1 and an eigenvector is

0.577 35

does

it

work?

3 2 1

0.788 68

1. 366

2 1 3 0.211 33 = 0.366 02

1 3 2

0.577 35

1.0

172

Exercises

0.788 68

1. 366 1

0.211 33 1. 732 1 = 0.366 04

0.577 35

1.0

Next you can find the other one.

1

3 2 1

1 0 0

2 1 3 + 1.5 0 1 0

1 3 2

0 0 1

2

4. 444 4 10

0.711 11 0.622 22

0.711 11

2. 622 2

2. 044 4

=

0.622 22

2. 044 4

1. 288 9

20

4. 444 4 102 0.711 11 0.622 22

0.711 11

2. 622 2

2. 044 4 =

0.622 22

2. 044 4

1. 288 9

11

2. 178 4 10

8. 129 9 1011 5. 951 5 1011

8. 129 9 1011 3. 034 1 1012 2. 221 1 1012 :

5. 951 5 1011 2. 221 1 1012 1. 626 0 1012

0.788 68 0.425 47 0.443 81

0.577 35

0.760 66 0.296 76

12

1. 030 8 10

3. 847 1 1012 2. 816 3 1012

0

1. 437 1 107

2. 791 7 107

0

0

1. 927 6 107

T

0.211 32 0.490 28 0.845 56

0.788 68 0.425 47 0.443 81

0.577 35

0.760 66 0.296 76

0.211 32 0.490 28 0.845 56

0.711 11

2. 622 2

2. 044 4 0.788 68 0.425 47 0.443 81 :

0.622 22

2. 044 4

1. 288 9

0.577 35

0.760 66 0.296 76

4. 309 4

2. 468 3 106 1. 167 8 105

2. 468 3 106

0.280 95

6. 479 4 102

5

2

1. 167 8 10

6. 479 4 10

0.161 74

0.211 32

1

0.788 68 . How

+1.5 = 4. 309 4, Solution is: 1. 732 1 and an eigenvector is

0.577 35

well

does

it

work?

3 2 1

1 0 0

0.211 32

2 1 3 + 1. 732 1 0 1 0 0.788 68

1 3 2

0 0 1

0.577 35

5

2. 262 8 10

= 6. 262 8 105 which is pretty close to 0 so this worked well.

7. 935 106

3 2 1

3. Find the eigenvalues and eigenvectors of the matrix A = 2 5 3 numerically.

1 3 2

The exact eigenvalues are 2, 4 + 15, 4 15. Compare your numerical results with

the exact values. Is it much fun to compute the exact eigenvectors?

Exercises

173

20

3 2 1

2 5 3 =

1 3 2

2. 713 5 1017 5. 082 3 1017 3. 058 1 1017 =

1. 632 8 1017 3. 058 1 1017 1. 840 1 1017

0.415 99

0.806 58

0.419 97

0.779 18 7. 803 9 102 0.621 92

0.468 86

0.585 95

0.660 94

17

17

3. 482 5 10

6. 522 6 10

3. 924 8 1017

0

1. 539 1013 1. 324 1 1013

0

0

1. 399 9 1012

3 2 1

0.415 99

0.806 58

0.419 97

0.779 18 7. 803 9 102 0.621 92 2 5 3

1 3 2

0.468 86

0.585 95

0.660 94

0.415 99

0.806 58

0.419 97

0.779 18 7. 803 9 102 0.621 92

0.468 86

0.585 95

0.660 94

7. 873 0

8. 768 9 105 3. 295 6 105

1. 746 2

0.641 05

= 8. 768 9 105

5

3. 295 6 10

0.641 05

0.380 82

0.415 99

An eigenvalue is 7. 873 0 and the approximate eigenvector is 0.779 18 . How well

0.468 86

does

it

work?

3 2 1

1 0 0

0.415 99

2 5 3 7. 873 0 0 1 0 0.779 18

1 3 2

0 0 1

0.468 86

4

1. 007 3 10

= 2. 414 105

8. 478 105

It works pretty well. Next we find the other eigenvalues and eigenvectors which go

with them.

1

3 2 1

1 0 0

2 5 3 1.75 0 1 0 =

1 3 2

0 0 1

3. 295 6 1. 006 3

1. 106 9

1. 006 3 0.276 73

0.704 4

2

1. 106 9 0.704 4 2. 515 7 10

20

3. 295 6 1. 006 3

1. 106 9

1. 006 3 0.276 73

0.704 4

1. 106 9 0.704 4 2. 515 7 102

= 2. 998 7 1011 9. 995 6 1010

9. 995 4 1010 =

11

10

2. 998 6 10

9. 995 4 10

9. 995 2 1010

174

Exercises

0.301 52 0.829 69 0.469 80

0.301 51 0.384 47 0.872 51

0

2. 995 0 106

1. 376 106

0

0

5. 368 8 105

3. 295 6 1. 006 3

1. 106 9

0.301 52 0.829 69 0.469 80 1. 006 3 0.276 73

0.704 4

2

0.301 51 0.384 47 0.872 51

1. 106 9 0.704 4 2. 515 7 10

0.301 52 0.829 69 0.469 80 =

0.301 51 0.384 47 0.872 51

4. 000 0

1. 828 8 106 8. 371 4 106

1. 828 8 106

0.155 70

7. 669 2 102

6

2

8. 371 4 10

7. 669 2 10

0.608 53

0.904 53

1

0.301 52 . How well does it

1.75 = 4, Solution is: 2.0. An eigenvector is then

0.301 51

work?

3 2 1

1 0 0

0.904 53

0.000 02

2 5 3 2 0 1 0 0.301 52 = 0.000 03

1 3 2

0 0 1

0.301 51

0.000 03

1

3 2 1

1 0 0

2 5 3 0.38 0 1 0 =

1 3 2

0 0 1

0.493 54

7. 815 4 102 0.449 38

7. 815 4 102

1. 056 5

1. 908 3

0.449 38

1. 908 3

2. 639 1

20

2

0.493 54

7. 815 4 10

0.449 38

7. 815 4 102

1. 056 5

1. 908 3 =

0.449 38

1. 908 3

2. 639 1

7. 593 8 109

4. 459 9 1010

6. 738 1010

4. 459 9 1010 2. 619 4 1011 3. 957 3 1011 :

6. 738 1010

3. 957 3 1011 5. 978 7 1011

0.989 8

0.549 53

0.834 58 3. 863 5 102

0.830 22

0.540 31

0.137 12

10

11

8. 115 9 10

4. 766 6 10

7. 201 3 1011

0

4. 418 3 106

1. 380 8 106

0

0

6. 663 6 105

T

9. 356 7 102 0.107 43

0.989 8

0.549 53

0.834 58 3. 863 5 102

0.830 22

0.540 31

0.137 12

2

0.493 54

7. 815 4 10

0.449 38

7. 815 4 102

1. 056 5

1. 908 3

0.449 38

1. 908 3

2. 639 1

175

Exercises

0.989 8

0.549 53

0.834 58 3. 863 5 102 =

0.830 22

0.540 31

0.137 12

3. 952 9

2. 044 4 105 1. 021 4 105

2. 044 4 105

0.182 25

0.145 62

1. 021 4 105

0.145 62

0.568 55

1

.38 = 3. 952 9,Solution is: 0.127 02 and an approximate eigenvector is then

9. 356 7 102

0.549 53

0.830 22

3 2 1

1 0 0

9. 356 7 102

2 5 3 0.127 02 0 1 0

0.549 53

1 3 2

0 0 1

0.830 22

2. 388 105

= 5. 469 9 105

3. 754 4 105

0 2 1

4. Find the eigenvalues and eigenvectors of the matrix A = 2 5 3 numerically.

1 3 2

I dont know the exact eigenvalues in this case. Check your answers by multiplying

your numerically computed eigenvectors by the matrix.

50

0 2 1

5. 045 4 1042 1. 454 4 1043 8. 826 9 1042

2 5 3 = 1. 454 4 1043 4. 192 3 1043 2. 544 4 1043

1 3 2

8. 826 9 1042 2. 544 4 1043 1. 544 2 1043

0.284 32 0.756 78

0.588 6

= 0.819 59 0.510 39 0.260 32

0.497 42 0.408 40 0.765 36

0

7. 091 8 1038 8. 072 9 1037

0

0

3. 007 5 1038

0.284 32 0.756 78

0.588 6

0.819 59 0.510 39 0.260 32

0.497 42 0.408 40 0.765 36

0.284 32 0.756 78

0.588 6

0.819 59 0.510 39 0.260 32

0.497 42 0.408 40 0.765 36

7. 514 5

9. 130 9 105

9. 130 9 105

0.541 46

1. 248 7 105

0.344 28

0

2

1

2 1

5 3

3 2

1. 248 7 105

0.344 28

2

2. 686 9 10

0.284 32

Thus an eigenvalue is 7. 514 5 and an eigenvector is approximately 0.819 59 . How

0.497 42

well does it work?

176

Exercises

0 2 1

0.284 32

2. 136 6

2 5 3 0.819 59 = 6. 158 9

1 3 2

0.497 42

3. 737 9

0.284 32

2. 136 5

0.819 59 7. 514 5 = 6. 158 8 .

0.497 42

3. 737 9

This has essentially found it. However, we dont know the others yet. begin with the

one which is closest to 0.541 46

0 2 1

1 0 0

2 5 3 + .541 46 0 1 0

1 3 2

0 0 1

5. 323 8

2. 181 4

0.480 23

0.393 89 0.393 38

= 2. 181 4

0.480 23 0.393 38

1. 046 8

20

5. 323 8

2. 181 4

0.480 23

2. 181 4

0.393 89 0.393 38 =

0.480 23 0.393 38

1. 046 8

15

5. 483 4 10

2. 055 8 1015 2. 530 4 1014

2. 055 8 1015 7. 707 7 1014 9. 487 0 1013 :

2. 530 4 1014 9. 487 0 1013 1. 167 7 1013

0.935 48

0.353 06

1. 491 6 102

0.350 73

0.932 81

8. 286 3 102

2

2

4. 316 9 10

7. 228 5 10

0.996 45

15

15

5. 861 6 10

2. 197 6 10

2. 704 9 1014

0

2. 167 1 1010 1. 796 2 109

0

0

8. 270 5 107

T

0.935 48

0.353 06

1. 491 6 102

0.350 73

0.932 81

8. 286 3 102

2

2

4. 316 9 10

7. 228 5 10

0.996 45

5. 323 8

2. 181 4

0.480 23

2. 181 4

0.393 89 0.393 38

0.480 23 0.393 38

1. 046 8

0.935 48

0.353 06

1. 491 6 102

0.350 73

0.932 81

8. 286 3 102 =

2

2

4. 316 9 10

7. 228 5 10

0.996 45

5

6. 163 8

1. 742 0 10

1. 159 3 106

1. 742 0 105

0.513 51

0.577 10

1. 159 3 106

0.577 10

0.979 41

1

+.541 46

0.935 48

0.350 73

2

4. 316 9 10

0 2 1

1 0 0

0.935 48

2 5 3 + 0.703 70 0 1 0

:

0.350 73

2

1 3 2

0 0 1

4. 316 9 10

177

Exercises

6. 276 106

8. 299 106

6. 025 3 106

0 2 1

1.0 1.0 1.0

2 5 3 = 1.0

1.0 2.0

1 3 2

1.0 2.0 4.0

20

1.0 1.0 1.0

1.0

1.0 2.0

1.0 2.0 4.0

1. 287 1 1013

2. 778 8 1013 5. 314 3 1013

5. 999 6 1013 1. 147 4 1014 :

= 2. 778 8 1013

13

5. 314 3 10

1. 147 4 1014 2. 194 2 1014

0.453 05 0.178 55 0.873 42

0.866 43 0.318 81 0.384 26

0

1. 509 3 109

1. 706 9 109

0

0

6. 192 3 109

T

0.209 85 0.930 85 0.299 13

0.453 05 0.178 55 0.873 42

0.866 43 0.318 81 0.384 26

0.209 85 0.930 85 0.299 13

1.0

1.0 2.0 0.453 05 0.178 55 0.873 42 =

1.0 2.0 4.0

0.866 43 0.318 81 0.384 26

5. 288 0

2. 017 9 105 2. 856 3 105

2. 017 9 105

0.916 86

0.727 58

5

2. 856 3 10

0.727 58

0.371 12

0.209 85

1

0.453 05 .

= 5. 288 0, Solution is: 0.189 11. An approximate eigenvector is

0.866 43

How well does it work?

0 2 1

1 0 0

0.209 85

1. 473 4 105

2 5 3 0.189 11 0 1 0 0.453 05 = 1. 628 6 105

1 3 2

0 0 1

0.866 43

9. 422 7 106

.

0 2 1

5. Find the eigenvalues and eigenvectors of the matrix A = 2 0 3 numerically.

1 3 2

I dont know the exact eigenvalues in this case. Check your answers by multiplying

your numerically computed eigenvectors by the matrix.

I will do this one a little differently. You can easily find that the eigenvalues are

approximately 4.9, 2.7, .3. To find the eigen pair which goes with .3 to the following.

1

0 2 1

1 0 0

2 0 3 .3 0 1 0 =

1 3 2

0 0 1

178

Exercises

1. 138 5

4. 788 7 102 0.754 22

4. 788 7 102

0.180 77

0.347 18

0.754 22

0.347 18

0.468 10

20

1. 138 5

4. 788 7 102 0.754 22

4. 788 7 102

0.180 77

0.347 18 =

0.754 22

0.347 18

0.468 10

17805.

3431. 1 12214.

3431. 1

661. 21 2353. 7 =

12214. 2353. 7 8378. 3

0.156 94

0.923 48 0.350 07

0.558 67 0.209 28 0.802 55

21863.

4213.0

14997.

0

2. 261 7 102 6. 348 1 102

0

0

0.281 26

A similar matrix is

T

0.814 41 0.321 53 0.483 08

0.156 94

0.923 48 0.350 07

0.558 67 0.209 28 0.802 55

1. 138 5

4. 788 7 102 0.754 22

4. 788 7 102

0.180 77

0.347 18

0.754 22

0.347 18

0.468 10

0.156 94

0.923 48 0.350 07 =

0.558 67 0.209 28 0.802 55

1. 665 1

7. 437 4 106 1. 163 3 105

7. 437 4 106

0.296 62

0.142 05

5

1. 163 3 10

0.142 05

0.174 36

Thus an eigenvalue is close to 1.6651 and so the eigenvalue for the original matrix

1

is the solution of .3

= 1.6651, Solution is: 0.300 56. Then what about the

eigenvector? We have

1

(A .3I)1 v =

v

.3

is just the first column of that orthogonal matrix above. Check it.

0 2 1

0.814 41

0.244 79

2 0 3 0.156 94 = 0.047 19

1 3 2

0.558 67

0.167 89

0.814 41

0.244 78

0.156 94 (0.300 56) = 4. 717 0 102

0.558 67

0.167 91

This worked very well. Now lets find the pair associated with the eigenvalue being

close to 2.7.

1

1 0 0

0 2 1

2 0 3 + 2.7 0 1 0

1 3 2

0 0 1

Exercises

= 13. 823 25. 248 13. 175

7. 127 4 13. 175 7. 105 8

20

7. 969 8 13. 823 7. 127 4

13. 823 25. 248 13. 175 =

7. 127 4 13. 175 7. 105 8

3. 436 7 1031 6. 226 6 1031 3. 260 7 1031 =

1. 799 7 1031 3. 260 7 1031 1. 707 5 1031

0.439 25

0.891 55

0.110 45

0.795 85

0.443 21

0.412 55

0.416 76 9. 330 8 102 0.904 21

0

7. 478 5 1026 3. 772 7 1026

0

0

3. 494 4 1026

T

0.439 25

0.891 55

0.110 45

0.795 85

0.443 21

0.412 55

0.416 76 9. 330 8 102 0.904 21

0.439 25

0.891 55

13. 823 25. 248 13. 175 0.795 85

0.443 21

7. 127 4 13. 175 7. 105 8

0.416 76 9. 330 8 102

39. 777

1. 791 2 104 6. 881 105

1. 791 2 104

0.336 06

0.128 95

5

6. 881 10

0.128 95

0.210 75

179

0.110 45

0.412 55 =

0.904 21

1

Eigenvalue, +2.7

= 39. 777, Solution is: 2. 674 9. The eigenvector would be the first

column of the above orthogonal matrix on the right.

0 2 1

0.439 25

1. 174 9

2 0 3 0.795 85 = 2. 128 8

1 3 2

0.416 76

1. 114 8

0.439 25

1. 174 9

0.795 85 (2. 674 9) = 2. 128 8 This worked very well.

0.416 76

1. 114 8

1

0 2 1

1 0 0

2 0 3 4.9 0 1 0

1 3 2

0 0 1

= 2. 962 0 4. 446 3 5. 621

3. 668 8 5. 621 6. 735 1

40

1. 753 6 2. 962 0 3. 668 8

2. 962 0 4. 446 3 5. 621 =

3. 668 8 5. 621 6. 735 1

1. 765 1 1044 2. 722 1 1044 3. 337 8 1044

2. 164 3 1044 3. 337 8 1044 4. 092 8 1044

180

Exercises

0.379 20

0.739 6

0.556 06

0.584 81 0.657 26 0.475 40

0.717 08 0.144 92

0.681 76

44

3. 018 2 10

4. 654 7 1044

0

6. 539 3 1039

0

0

0.379 20

0.739 6

0.556 06

0.584 81 0.657 26 0.475 40

0.717 08 0.144 92

0.681 76

2. 962 0 4. 446 3 5. 621

3. 668 8 5. 621 6. 735 1

0.379 20

0.739 6

0.556 06

0.584 81 0.657 26 0.475 40

0.717 08 0.144 92

0.681 76

13. 259

1. 247 5 104

1. 247 5 104

0.142 61

1. 129 2 105 2. 290 2 102

1

4.9

5. 707 5 1044

5. 579 8 1039

4. 797 6 1039

T

1. 129 2 105

2. 290 2 102

0.181 74

0 2 1

0.379 20

1. 886 7

2 0 3 0.584 81 = 2. 909 6

1 3 2

0.717 08

3. 567 8

0.379 20

1. 886 7

0.584 81 4. 975 4 = 2. 909 7 . It worked well.

0.717 08

3. 567 8

3 2 3

T

6. Consider the matrix A = 2 1 4 and the vector (1, 1, 1) . Find the shortest

3 4 0

distance between the Rayleigh quotient determined by this vector and some eigenvalue

of A.

T

1

3 2 3

1

q = 1 2 1 4 1 13 = 7. 333 3

1

3 4 0

1

3 2 3

1

1

2 1 4 1 7. 333 3 1

3 4 0

1

1

|7. 333 3 q |

3

= 0.471 41

1 2 1

T

7. Consider the matrix A = 2 1 4 and the vector (1, 1, 1) . Find the shortest

1 4 5

distance between the Rayleigh quotient determined by this vector and some eigenvalue

of A.

181

Exercises

T

1

1

q= 1 2

1

1

2 1

1

1 4 1 13 = 7

4 5

1

|7 q | = 2. 449 5

3 2 3

T

8. Consider the matrix A = 2 6 4 and the vector (1, 1, 1) . Find the shortest

3 4 3

distance between the Rayleigh quotient determined by this vector and some eigenvalue

of A.

T

1

3 2 3

1

q = 1 2 6 4 1 13 = 8.0

1

3 4 3

1

|q 8|

9.

Using

3

2

3

3

2

3

3. 266 0

2

6

4

3

1

1

4 1 8 1

3

1

1

Gerschgorins

theorem, find upper and lower bounds for the eigenvalues of A =

2 3

6 4 .

4 3

10 12

10. Tell how to find a matrix whose characteristic polynomial is a given monic polynomial.

This is called a companion matrix. Find the roots of the polynomial x3 + 7x2 + 3x + 7.

0 0 7

The companion matrix is 1 0 3

0 1 7

20

0 0 7

8. 062 3 1014 5. 408 5 1015 3. 628 2 1016

1 0 3 = 2. 253 5 1014 1. 511 7 1015 1. 014 1 1016

0 1 7

7. 726 4 1014 5. 183 2 1015 3. 477 1016

0.707 72 0.258 58

0.657 47

= 0.197 82 0.820 86 0.535 78

0.678 23 0.509 24 0.529 79

0

4. 570 4 1010 2. 079 1 1010

0

0

3. 151 4 1011

0.707 72 0.258 58

0.657 47

0 0 7

0.197 82 0.820 86 0.535 78 1 0 3

0.678 23 0.509 24 0.529 79

0 1 7

0.707 72 0.258 58

0.657 47

0.197 82 0.820 86 0.535 78

0.678 23 0.509 24 0.529 79

182

Exercises

6. 708 3

= 4. 147 2 105

1. 391 6 105

5. 850 6

5. 220 9

0.154 76

1. 187 6

0.936 81 0.446 46

Clearly a real root is close to 6. 708 3. Then the other roots can be obtained from

the lower right block.

0.154 76

1. 187 6

, eigenvalues: 0.145 85 + 1. 011i, 0.145 85 1. 011i

0.936 81 0.446 46

How well do these work? Try the last one. Evaluating the polynomial at this value of

x gives

6. 350 5 104 + 2. 065 8 104 i

20

0 0 1

0 0 1

=

1 0 4

0 1 3

36482.

27300.0

49899.

2. 448 9 105

14010.0

9182.0

77199.

1. 949 9 105

27300.0

49899.

2. 448 9 105 5. 442 5 105

0.260 30

6. 965 1 102

0.941 85

0.200 79

9. 996 0 102

0.251 72

0.209 32

0.939 59

0.940 38

0.202 95

0.253 14

0.102 07

0.194 78

0.943 71

7. 090 6 102 0.257 75

0

70622. 2. 084 2 105 3. 625 1 105

0

0

2. 130 5

5. 038

0

0

0

2. 683 4

T

0.260 30

6. 965 1 102

0.941 85

0.200 79

9. 996 0 102

0.251 72

0.209 32

0.939 59

0.940 38

0.202 95

0.253 14

0.102 07

0.194 78

0.943 71

7. 090 6 102 0.257 75

0 0 0 1

1 0 0 1

0 1 0 4

0 0 1 3

0.260 30

6. 965 1 102

0.941 85

0.200 79

9. 996 0 102

0.251 72

0.209 32

0.939 59

0.940 38

0.202 95

0.253 14

0.102 07

0.194 78

0.943 71

7. 090 6 102 0.257 75

0.613 47

4. 251 0

0.485 69

2. 096 8

0.503 89

2. 337 6

0.115 42 0.330 94

4. 611 8 106 7. 585 5 106 0.974 48 0.108 46

11. Find

0

1

0

0

0.613 47 4. 251 0

, eigenvalues: 1. 475 5 + 1. 182 7i, 1. 475 5 1. 182 7i

0.503 89 2. 337 6

183

Exercises

0.157 34 0.304 46

0.974 48 0.108 46

, eigenvalues: 0.024 44 + 0.528 23i, 0.024 44 0.528 23i

4

2

+1 = 2. 887 9 105 3. 009 2 106 i

12. Suppose A is a real symmetric matrix and the technique of reducing to an upper

Hessenberg matrix is followed. Show the resulting upper Hessenberg matrix is actually

equal to 0 on the top as well as the bottom.

Let QT AQ = H where H is upper Hessenberg. Then take the transpose of both sides.

This will show that H = H T and so H is zero on the top as well.

- Matrices and DeterminantsCargado porsudersanaviswanathan
- Cambridge scheCargado poradamadamO
- Wikipedia: MatrixCargado porMaiara e maraiara
- Ansys2Excite_UsersGuide.pdfCargado pordeepali0305
- Mechanical Engg 2014Cargado porvignanaraj
- 19b_FP3_(new)_June_2009Cargado porgkosyvas
- Study Notes on Matrix Algebra PrepLadderCargado pormanikantamnk11
- Intro to Continuum Mechanics - Raymond.pdfCargado porSeth Ferrara
- 100314 Ott 205 Exam Ssoln05o LnCargado porfranciis
- Principal-component-analysis-applied-to-remote-sensing.pdfCargado porCrina Lavinia
- Rotor_Software_Manual_v1Cargado porBernardo Foresti
- GATE Mathematics Questions All Branch By S K Mondal.pdfCargado porHarshit Agarwal
- CE Engineering MathematicsCargado porphanirajinish
- The Determinant FunctionCargado porsebastobon85
- Plus Two Chapter wise important 3,6 and 10 marksCargado porAhamedIqbal
- Lab 5 solnsCargado porPatriceSingson
- invertible matrix theoremCargado porapi-298329103
- 1894270-AQA-MFP4-W-QP-JUN11.pdfCargado porastargroup
- Prac 1 ControlCargado porJack Clarkson
- tut1Cargado porgermanschultze
- hmwk2Cargado porEren Can Kızıldağ
- Low-cost Algorithm for Some Bearing Estimation Methods in Presence of Separable Nuisance ParametersCargado poruranub
- Lecture Note 6Cargado porMathieu Aucejo
- Application of Eigenvectors and EigenvaluesCargado porAnonymous zfmlsb2GjA
- Week13-14Cargado porAbdurrahman Hazer
- Final PrepCargado porŞamil Okan
- UT Dallas Syllabus for math2333.503 06s taught by William Donnell (wxd022000)Cargado porUT Dallas Provost's Technology Group
- UT Dallas Syllabus for math2333.501 05f taught by William Donnell (wxd022000)Cargado porUT Dallas Provost's Technology Group
- Music Baes DfCargado porSindhu Velayudham
- Vibration Control of Unknown Flexible Beam StructuresCargado poroscar201140

- Abstract AlgebraCargado porMarub Asub
- Political Economy (Cambria)Cargado porcantor2000
- Gretl's Guide (June 2017)Cargado porcantor2000
- APAstyle_paperexample1Cargado porGill
- APA Style in-text Citations - BedfordCargado porcantor2000
- Basic Maths and Algebra - US NavyCargado porcantor2000
- APA Sample Paper U PurdueCargado porcantor2000
- APA Style Workshop Purdue_U.pdfCargado porcantor2000
- AVU - Multimedia Design and ApplicationsCargado porcantor2000
- APA Style Manuscript Format - BedfordCargado porcantor2000
- Basic Legal CitationCargado porcantor2000
- Analysis Tools With Applications - Bruce DriverCargado porcantor2000
- Basic Analysis, Introduction to Real Analysis - Lebl JCargado porrivera_markgelo
- AVU - Text-Based Productivity ToolsCargado porcantor2000
- AVU - Probability and StatisticsCargado porcantor2000
- AVU - Multimedia Design and Applications_readingsCargado porcantor2000
- AVU - Linear ProgrammingCargado porcantor2000
- AVU - Introduction to ICTCargado porcantor2000
- AVU - Graphics and Information Management SystemsCargado porcantor2000
- Mathematics-3-Calculus (1).pdfCargado pornoth
- AVU - Analysis 2Cargado porcantor2000
- AVU - Analysis 1Cargado porcantor2000
- [Wissam Raji] an Introductory in Elementary Number TheoryCargado pormp19uy
- An Introduction to Economic ReasoningCargado porapi-3745953
- Human Rights in BriefCargado porcantor2000
- Democracy in BriefCargado porcantor2000
- Algebra & Number Theory - BakerCargado porcantor2000
- Alder - Multivariate CalculusCargado porcantor2000

- A hadamarad TransformCargado porAnup Dakre
- CLRS Solution 4Cargado porNapster
- Advanced MathCargado porRuby Villagracia
- Fraleigh = Linear Algebra.pdfCargado porOmar Butrón Zamora
- Linear Algebra SyllabusCargado porSum Ting Wong
- rgtrwgrCargado porraj
- Procrustes Distance READMECargado porMirQuinto
- Chap.5 Eigenvalues EigenvectorsCargado porndarubagus
- MCQ's of Maths-ICargado porAnsari Sameer
- 1304.4627.pdfCargado porhendra lam
- lecture15.pptCargado porApel King
- Analysis of Two Partial-least-Squares Algorithms for Multivariate Calibration.pdfCargado porAna Rebelo
- QRG_CE.pdfCargado porckvirtualize
- Inverse[1]Cargado porscribdsurya4u
- MIT18 06S10 Final ExamCargado porAbdul Rahim
- Matrix Algorithms Volume II Eigensystems~Tqw~_darksidergCargado pornijo00
- Basic Mathematics NotesCargado porsasyeda
- Hmw1TextCargado porCaseyJ.Li
- Multivariate Statistics - An Introduction 8th EditionCargado porsharingandnotcaring
- Matrix Algebra 2Cargado porManideep Nulu
- II Assignment MmCargado porshantan02
- Matrices TutorialCargado porNelsonMoseM
- Chapter1.pdfCargado porSahaya Grinspan
- Saad.Y.-.Iterative.Methods.for.Sparse.Linear.Systems.(2000).pdfCargado porpalwster
- Applied Linear AlgebraCargado porMatthew Flowers
- graph theoryCargado porTeto Schedule
- MA1101R APC Test SampleCargado portechnikvista
- 64130-mt----linear algebraCargado porSRINIVASA RAO GANTA
- applicatioms of SVDCargado porBilal Barut
- key encoding messages into matricesCargado porapi-327831991