Documentos de Académico
Documentos de Profesional
Documentos de Cultura
David Sirajuddin
Itcanbeshown.com
Contents
1 Introduction 3
6 Residue Theorem 8
6.1 Definition of a Residue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
6.2 Finding Residues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
6.2.1 Functional replacement . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.2.2 Quotient rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.2.3 Multiplication of Series . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.2.4 Division of Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6.2.5 Computing Coefficients Directly . . . . . . . . . . . . . . . . . . . . . 12
9 References 21
1
CONTENTS CONTENTS
10 Acknowledgements 22
11 Appendix 22
11.1 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
11.2 Useful Series Representations . . . . . . . . . . . . . . . . . . . . . . . . . . 22
11.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
11.3.1 Real Integration: Functional Replacement and ML Bound . . . . . . 24
11.3.2 Real Integration: Functional Replacement and Jordan’s Lemma . . . 26
11.3.3 Real Integration: Integrals Containing Trigonometric Functions . . . 29
11.3.4 Real Integration: Strange Contours (Fresnel Integral) . . . . . . . . . 31
11.3.5 Complex Integration: Quotient Rule . . . . . . . . . . . . . . . . . . 35
11.3.6 Finding Residues: Multiplication of Series . . . . . . . . . . . . . . . 36
11.3.7 Finding Residues: Division of Series . . . . . . . . . . . . . . . . . . . 37
2
2 COMPLEX VARIABLES AND . . .
1 Introduction
Complex integration is an intuitive extension of real integration. Since a complex number
represents a point on a plane while a real number is a number on the real line, the analog of a
single real integral in the complex domain is always a path integral. For some special functions
and domains, the integration is path independent, but this should not be taken to be the
case in general. Given the sensitivity of the path taken for a given integral and its result,
parametrization is often the most convenient way to evaluate such integrals. Of particular
importance in many physics and engineering problems is that of the closed path integral,
where both the beginning and ending points of the path are identical. As with real calculus,
such
H integrals are called contour integrals, and are denoted by a circle in the integration sign
( ). This guide will discuss solution strategies for complex closed contour integrals, with
particular emphasis on their use in solving difficult real-valued integrals. The material will be
presented to the reader assuming minimal knowledge of complex mathematics. While formal
definitions and theorems will be provided so as to make this document self-contained, proofs
will not be included. The reader is recommended to see Brown and Churchill’s book (see
References) for discussion and proofs of theorems used throughout this document. General
strategies of complex integration are discussed, while fully worked examples are provided in
the Appendix.
3
2 COMPLEX VARIABLES AND . . . 2.2 Functions
adoption of this number set. This extra dimension can sometimes change things considerably
in regards to complex mathematics and its real counterpart. In general, most results from
real mathematics carry over into complex mathematics; however, the few aspects that are
different – are grossly so. Thus, the reader is cautioned when assuming any property from
real numbers is translatable into complex numbers.
2.2 Functions
A complex function f of an independent complex variable z = x + iy = reiθ , where
{r ≥ 0, (r, θ, x, y)R}, is of the general form:
f (z) − f (z0 )
f 0 (z0 ) = lim (1)
z→z0 z − z0
Since the complex number z0 = x0 + iy0 can be approached from any number z = x + iy by
way of an infinite number of paths, the limit is said to exist if and only if the same, unique
limit is obtained when approached from any path that originates from any complex number
z. The criteria that a function f (x, y) = u(x, y) + iv(x, y) = u(r, θ) + iv(r, θ) must meet in
order for this to be true are known as the Cauchy-Riemann Equations. These equations can
be expressed in either cartesian or polar form.
∂u ∂v ∂u ∂v
= and =− Cartesian Form (2)
∂x ∂y ∂y ∂x
∂u 1 ∂v 1 ∂u ∂v
= and =− Polar Form (3)
∂r r ∂θ r ∂θ ∂r
Thus, while real differentiability only demands the continuity of a function at a point x0
in order for it to be said that its derivative exists at that point, a complex-valued function
is complex differentiable at a point z0 if and only if it is both analytic and satisfies the
Cauchy-Riemann equations at that point.
4
2 COMPLEX VARIABLES AND . . . 2.2 Functions
2.2.2 Analyticity
Complex differentiability is a much more robust property than real differentiability. If
a function is complex differentiable at a point z0 , it is also said to be analytic at z0 . A
function satisfying these conditions inherits another consequence in that the function is
infinitely differentiable. The terms regular and holomorphic are often used interchangeably
with analytic.
∞
X
f (z) = cn (z − z0 )n (R1 < |z − z0 | < R2 ) (7)
n=−∞
f (n) (z0 )
I
1 f (z)
an = = (n = 0, 1, 2, ...)
2πi C (z − z0 )n+1 n!
It is noted that only negative powers of z − z0 are found in functions that hold singularities
interior to an otherwise analytic disk. In this case, a punctured disk, or annular domain,
must be defined that does not contain the singularities, and where the function is complex
differentiable. This will be important in the discussion of residues.
For completeness, different series representations will now be discussed and summarized.
A function that is analytic inside a disk with the exception of a finite number of singularities
can be made meromorphic if these points are omitted in the construction of an appropriate
annular domain. A function that is analytic in such a domain is given by a power series
representation involving both positive and negative powers of z − z0 , called a Laurent series.
A function that is analytic inside a disk |z − z0 | < R2 , centered at z0 is given by a Taylor
series, which includes only positive powers of z − z0 . Finally, a function that is analytic
inside a disk |z| < R2 , centered at z0 = 0, is given by its Maclaurin series.
* By the way, power series can be integrated or differentiated term by term, or in generalized
summation notation.
the domain interior to the closed path of integration. As previously stated, the integral of
a complex function f (z) is always a path integral, since f (z) → f (x, y) is a function of two
variables. While integrals of a real function can have physical interpretations, such as the
area under a curve, or the volume prescribed an area, no such direct explanations exists for
complex integrals. The only integrals discussed in this guide will be closed-contour integrals.
The integral of a function f (z) over a simple closed contour C is denoted as
I
f (z)dz
C
where C is often taken to be circular, centered about a point z0 , with the convenient
parametrization z(t) = z0 + Reit (a ≤ t ≤ b) for some nonnegative radius R. The con-
tour C has a direction, or sense, of integration with the convention of counterclockwise
being positive and clockwise paths as negative. If the contour is depicted on axes, its sense
is indicated with a directed
H line segment, i.e. an arrow. Sometimes the sense is indicated
in the integral symbol ( ) with the arrow being drawn in the circle. These integrals will
be solved via residue theorem (Section 6 ) or parametrization (Section 8 ); however, before
a discussion of residue theorem is presented, two important theorems must first be cited in
order to build its theoretical foundation. The two theorems are presented in the subsequent
sections, followed by their generalization into the so-called residue theorem.
This same result is present in real calculus. Whenever an integrable function f (x) is inte-
grated over a closed path, the value of the integral is zero. This is also true in complex
mathematics provided that, in addition to a complex-valued function being analytic at all
points on its contour, it must also be analytic for all points interior to it. That is to say,
this single result should be expected in some sense; however, the theorem presented in the
following section does not necessarily follow from intuition, and exhibits the first sign of a
strong separation between real and complex calculus.
section of Itcanbeshown.com for further explanation if desired. The statement in the theorem
implies that the value of a complex integral is dependent not only on the path of integration,
but also all points interior to its closed contour.
6 Residue Theorem
Residue theorem is a generalization of both Cauchy’s integral theorem and formula to the
scenario when there exists a single, or multiple, number of singularities within a prescribed
domain of interest. From Cauchy’s integral theorem formula, and a function’s Laurent series,
you can probably postulate this theorem on your own, to verify your own intuition kindly
check with the theorem, stated in full, below:
Theorem. Let C be a simple closed contour, describe in the positive sense. If a function
f is analytic inside and on C except for a finite number of singular points zk (k = 1, 2, . . . , n)
inside C, then
I n
X
f (z)dz = 2πi Res f (z) (11)
C z=zk
k=1
In words, the integral of a function f over any contour containing the singular points zk is
equal to the product of 2πi and the sum of the residues of all the singular points interior to
C. Note the simplicity of the result, by invoking this theorem there is no need for any kind
of tricky, or resourceful, integration. All one need do is compute all the residues of a function
f (z) at all of the contained singularities within the prescribed contour of integration, and
multiple this sum by 2πi! Thus, the real work is only in finding these residues, whatever
those are. A precise definition of a residue follows.
an = 0 (n = 1, 2, . . .)
8
6 RESIDUE THEOREM 6.2 Finding Residues
bn = 0 (n 6= 1)
Letting n = 1 in the definition of bn (Eqn. (6)), it is found to be identical to the form of
Cauchy’s integral formula which enables the following to be written down
Z
f (z)dz = 2πib1
C
Thus, the only contribution to the integral of the function f (z) along the simple closed
contour C that encloses a finite number of singular points comes from the b1 coefficient of the
power series expansion of the function for each contour centered at each of the singularities.
The coefficient b1 is called a residue. A residue is the b1 coefficient of a function centered about
that singular point. In general, in order to extract the residue of a function directly from a
power series representation, one must first express the expansion centered about that point.
For example, in the form of Cauchy’s integral formula in Eqn. (10), the residue corresponds
to that a point z0 . There exist residues for each contained singular point zk (k = 1, 2, . . .)
inside of a prescribed contour C of a function f . The notation for a residue at z = zk for a
function f (z) is given by,
Res f (z)
z=zk
It follows from intuition that the expansion must be written down in an expansion of the
form z − zk , as the integral containing a singular point zk is independent of the radius size of
the contour. The derivation of Cauchy’s integral formula, and the subsequent generalization
for residue theorem, follows from centering a contour at each singularity, and letting these
radii tend to zero so as to approach the constant b1 as expressed in Cauchy’s integral formula.
is worth presenting to eliminate any potential curiosities about their workings. Techniques
for finding residues are now presented.
φ(z)
f (z) = ,
(z − z0 )m
where φ(z) is analytic and nonzero at z0 . Moreover,
and,
φ(m−1) (z0 )
Res f (z) = if m ≥ 2 (13)
z=z0 (m − 1)!
where φ(z)(m−1) denotes the (m − 1)th derivative of φ(z).
p(z) p(z0 )
Res = 0 (14)
z=z0 q(z) q (z0 )
A simple pole is defined in the above section, where m = 1 (i.e. a singular point of order
one).
10
6 RESIDUE THEOREM 6.2 Finding Residues
and assume that it can be separated such that f (z) = G(z)H(z). Both functions G and H
have the series representations centered about zk :
G(z) = g1 + g2 + g3 + · · · + gn + . . .
H(z) = h1 + h2 + h3 + . . . + hn + . . .
where the terms gn and hn contain both their corresponding coefficients and powers of
z − zk . Note that the coefficient of the product gn hn is equal to a2n . Since the series forms
are centered about z = zk , if there exists a residue in the combined series it will correspond
to that at z = zk (if the residue about a different point is desired, try to obtain an expansion
centered about this point). The function f can be found by multiplying the known series,
f (z) = G(z)H(z)
= (g1 + g2 + g3 + . . . )(h1 + h2 + h3 + . . .)
f (z) = g1 h1 + g1 h2 + g1 h3 + g2 h1 + . . .
Furthermore, suppose that some term gi hj in the combined series contains the dependence
(z − zk )−1 , for the appropriate i and j. Then, the residue at z = zk is the coefficient of
this term. These series may be long, or even infinite, and in order to find any residues it
is important to notice that it is not necessary to multiply even most of the terms out, but
only those that matter. The reader is strongly cautioned to make sure they know which
terms matter, so as to not inadvertently, and detrimentally, neglect relevant terms. Lastly,
a specific coefficient an can be found by Leibniz’s rule
n
X n
an = G(r) (z)H (n−r) (z), (15)
r
r=0
(r) (n−r)
where G and H denote the rth and (n − r)th derivatives of G and H respectively.
This result follows from the combined series G(z)H(z) = f (z) actually being a Taylor series
expansion.
A(z) = a1 + a2 + a3 + · · · + an + . . .
B(z) = b1 + b2 + b3 + . . . + bn + . . .
where the terms an and bn contain both their corresponding coefficients and powers of z − zk .
The power series for the function f can be found by dividing the series representations of A
and B by way of conventional polynomial division. For the sake of example, let the functions
have the following series expansions, centered about the point zk = 0:
11
7 GENERAL APPROACH TO . . .
α−1
A(z) = + α1 z + α3 z 3 + . . .
z
B(z) = β0 + β1 z + β2 z 2 + . . .
where the terms αn and βn represent the numerical constants of the series. Then, the series
of f (z) = A(z)/B(z) can be found by division
α−1 −1
β0
z − α−1 ββ12 + . . .
0
2 α−1
β0 + β1 z + β2 z + . . . z
+ α1 z + α3 z 3 + . . .
α−1
z
+ α−1 ββ−1
0
+ α−1 ββ20 z + . . .
− α−1 ββ−1
0
+ α1 − α β2
−1 β0 z + . . .
β2
− α−1 ββ−1
0
− α−1 β12 z + . . .
0
h 2 i
α−1 β1
α1 + β0 β0 − β2 z + . . .
Notice how if the goal is to find a residue that the computation of only a few terms are
required, and how the actual work could have stopped after finding only the first term of
the quotient. This is because the first term contains the dependence z −1 , and its coefficient
(α−1 /β0 ) is exactly that of the function’s residue at z = 0.
12
7 GENERAL APPROACH TO . . .
Where the contour C and the singular points of f are shown in Figure 1.
Figure 1 - A semicircular contour C is chosen in order to integrate the function f (z). The contour
encloses the singularity at the point z1 only, while the singularity at z2 is exterior to the contour. The
residue at z2 , therefore, does not contribute to the value of the integral in the system drawn above.
1. Locate any singular points: Begin by examining the function f and locate the relevant
singular points. The only singularities that contribute to the result will be those that
lie interior to the closed contour. Furthermore, in order to determine whether the
function is integrable, these singularities must not lie on the contour of integration.
If any singular points are present on the contour, then the function is not integrable
along this contour and no solution can be obtained. In the case of the example, only
the singularity z1 need be taken into account since z2 is outside of the closed contour.
Note that if both points z1 and z2 were outside of the contour, the contour would
enclose no singularities, and the integral would then be identical to zero by Cauchy’s
integral theorem (Section 4 ).
2. Integrate: If the function contains singularities interior to its closed contour of integra-
tion, then the integration can be accomplished in one of two ways, by the application of
residue theorem or by parameterizing the contour. It is recommended that whichever
method demands the least amount of work, and whichever is more practical, be used.
If the contour is too strange, then parametrization of the contour can become too
complicated to make this method practical. Both strategies are outlined below.
(a) Residue Theorem – Begin by writing down the statement of residue theorem. In
the case of the example, this becomes
I
f (z)dz = 2πi Res f (z)
C z=z0
Next, find the residue(s) by the methods described in Section 6.2, insert the results
into the above equation, and the integral is solved.
13
8 GENERAL APPROACH TO . . .
(b) Parametrization – If the simple closed contour is circular, then apply the parametriza-
tion z(t) = z0 + Reit (0 ≤ t ≤ 2π), where R is the radius of the contour, and z0 is
its center. Differentiation of z(t) with respect to t then admits the substitution
dz = iReit dt. Insert these terms into the integral and proceed with conventional
integration to obtain a solution. In the example, this process yields the following
I Z 2π
f (z)dz = iR f (z0 + Reit )eit dt
C 0
.
After integration of the above equation, the solution has been arrived at.
These strategies can be applied to any complex integral, and usually necessitate little inte-
gration, in the conventional sense. The most difficulty will arise in finding the residues of
the function contained in the integrand (as parametrization is usually not the easiest way to
solve most integrals of this kind).
1. Go complex: transpose all single variables of the same kind to a complex variable (e.g.
f (x) → f (z))
2. Locate any singular points: For the complex-valued function f (z), find its poles and/or
singularities. These points are located where the function becomes boundless, that is
to say, where the denominator equals zero or if the function contains a logarithm this
point will be for z = 0. This is an important preliminary step, as the location of the
singular points will diagnose which situation is present and indicate the strategy that
14
8 GENERAL APPROACH TO . . .
should be used to go about evaluating the integral. The singular points play a role in
the choice of contour for the complex integral. The contour chosen (step 3) will always
involve the real line since it is desired to obtain the result of a real integral, rather than
a complex integral. There are two distinct categories of paths. An indented path is one
in which at least one singular point is located on the real line, and it is evident that in
this scenario a contour involving the real line will intersect this point, but is not able
to be integrated directly through this point due to the function failing to be analytic.
In this case, a different form of residue theorem must be incorporated in addition to
the standard theorem (Section 8.1 ). The steps in this section should still be followed,
but it changes step x, which needs to be accounted for in order to obtain the correct
result. Alternatively, if the singular points are all either above or below the real line,
the path is said to be unindented. Proceed with the following steps in this section, and
use the conventional form of residue theorem. The example function used throughout
this section falls under this category. Finally, be mindful of branch cuts that must be
made for reasons of preserving analyticity (see Glossary). It is not possible to integrate
across a branch cut, but it is possible to integrate around the branch cut.
Figure 2 - A semicircular contour C is chosen in order to integrate the function f (z). Note how
the contour encloses two singularities atH z0 and z1 . These points will contribute to the value of the
contour integral C f (z)dz by way of residue theorem.
The contour was chosen to contain the singularities located at z0 and z1 , and a simple
semicircle (y ≥ 0) does the trick since: (1) it encloses the singular points, and (2)
15
8 GENERAL APPROACH TO . . .
involves the real axis. Often it is in your best interest to choose a contour that encloses
the least amount of singularities (as this will necessitate the least amount of residue
computations). Bear in mind, however, that limits will be taken (step 5), and a contour
should be chosen such that it encloses all singularities before such limits are taken. As
these singular points will contribute to the value of the real integral before, or after,
limits are taken. That is to say, whether you decide count them or not. Contours
containing no singularities should generally be avoided, as this will admit the trivial
result of the complex integral being equal to zero by consequence of Cauchy’s integral
theorem (e.g. a semicircle in the region y ≤ 0). It is important to realize that in this
instance by breaking up the complex integral into a sum of path integrals (where the
sum of the paths are equal to the original contour), the problem can still be solved;
however, doing this often makes the real integral result more difficult to retrieve.
4. Apply residue theorem: Look at the contour you have drawn, and write down the
residue theorem statement for the closed contour integral being considered. For the
example,
I
f (z)dz = 2πi Res f (z) + Res f (z)
C z=z0 z=z1
where C is the entire semicircular contour. Next, break up the integral into a summa-
tion of a parameterized real integral and complex integrals that makeup the remainder
of the contour. The parametrization of the complex function f (z) into the real function
f (x) follows from the idea that z = x + iy → x on the real line. This is the integral
to which a solution is desired. As in the example, the integral is broken up in the
following way
I Z R Z
f (z)dz = f (x)dx + f (z)dz = 2πi Res f (z) + Res f (z)
C −R CR z=z0 z=z1
where the contour CR is the upper half of the semicircle contour (as shown in Figure
2). For convenient bookkeeping, move any complex integrals to the side of the equation
that contains the residues. The result from above is then a direct statement of the
value of the real integral,
Z R Z
f (x)dx = 2πi Res f (z) + Res f (z) − f (z)dz
−R z=z0 z=z1 CR
It is then only necessary to adjust the limits (step 5), and to solve the RHS of the
above equation to find the value of the integral (steps 6 and 7).
5. Take appropriate limits: Let the bounds tend to those which the original real integral
contained. In the example, this corresponds to letting R → ∞, such that f (x) is
integrated from −∞ < x < ∞.
16
8 GENERAL APPROACH TO . . . 8.1 Indented Paths
Z ∞ Z
f (x)dx = 2πi Res f (z) + Res f (z) − lim f (z)dz
−∞ z=z0 z=z1 R→∞ CR
6. Find the value of the complex integrals or prove they tend to zero in this limit: Deter-
mine the value of these complex integrals by either parametrization, or other means
(Section 8.1 ). If any integral cannot be evaluated, prove they tend to zero by using
either an ML bound or Jordan’s lemma given the limits of the previous step (Section
8.2 ). If you cannot prove they all tend to zero, repeat this procedure beginning with
step 3 in order to choose a different contour that may work.
7. Find the residues: Use the methods described in Section 6.2 to find the residues of the
function to be integrated. Reserve this step as the last in order to save time. If it is
not possible to find values for all the complex integrals, a different contour must be
chosen, and any residues that were computed prior to checking this would have been
for naught.
8. Express the final answer: Depending on whether the path of integration was indented
or unindented, the result obtained will be the sum of the residues multiplied by either
2πi and/or −πi. The solution to the example is
Z ∞
f (x)dx = 2πi Res f (z) + Res f (z)
−∞ z=z0 z=z1
Furthermore, suppose that the function f (x) is even, such that f (−x) = f (x), it is
then possible to determine the value of the semi-infinite integral by dividing the result
by two.
Z ∞
f (x)dx = πi Res f (z) + Res f (z)
0 z=z0 z=z1
This is allowed only for even functions, and is permitted due to symmetry. This is
not the only method to solve for semi-infinite real integrals, but it is often the most
convenient. For an example of a semi-infinite integral solution that is achieved directly
from complex integration, see Section 11.3.5, the Fresnel integral solution.
The solution to the original real integral has been found! This practice is fairly easy, with the
most difficult parts being a proper choice of a contour (subsequent examples in the Appendix
will demonstrate this), and showing that the complex integrals tend to zero. Make sure
to mind the scenario carefully, only a few exceptions exist (which have been outlined), but
neglecting to account for any exception will cause the result to be incorrect.
(ii) Cρ denotes the upper half of a circle |z − x0 | = ρ, CR is the upper half of the circle
|z − x0 | = R, where ρ < R and the clockwise direction is taken,
Then
Z
lim f (z)dz = −πi Res f (z) (16)
ρ→0 Cρ z=x0
Notice how the clockwise direction is taken, as opposed to the counterclockwise direction as
per usual. The theorem is written in this form because contours are usually conveniently
chosen containing this segment in this direction, so as to enable the path CR , and the entire
path as a whole, to be counterclockwise.
Although a figure is not included in this section, the scenario should be clear. A positively
oriented semicircular contour of radius R is desired to be used; however, its path along the
real axis contains a point z = x0 at which the function is not analytic, and therefore not
integrable. Thus, the desired contour is obtained by way of choosing paths that initially
avoid the singular point, but which – upon taking the proper limits – collapse to the desired
contour. A common way of accomplishing this is by choosing a counterclockwise semicircular
path CR . Continuing along its counterclockwise direction, the path traverses the real line,
at which point the path evades the singularity at z = x0 by indenting the path along the
real line. That is to say, at some some distance ρ from the singularity z = x0 , the path along
the real axis is rerouted through a semicircular clockwise path Cρ . At the intersection of Cρ
and the real axis, the path continues along the real line in the +x direction so as to close the
contour. The new contour can be described as tracing out a counterclockwise half-annulus.
During the subsequent analysis, the radius ρ is allowed to tend to zero, such that the value
of the path integral at that point is known by way of this theorem. In this manner, it is
possible to integrate the function through its singular point, and hence along the entirety of
the originally desired contour. Alternatively, if an improper real integral result is desired,
then the steps in Section 8 can be followed henceforth.
result for any of these integrals is unable to be determined, the integral must be shown to
tend to zero when taking proper limits so as to validly neglect the term. This is to say, if
it cannot directly be shown that a complex integral is identical to zero, the actual answer
to the real integral could involve other contributions that have not been accounted for. The
failure to prove this may hint that a different contour should be chosen. For most purposes,
a conventional ML limit will show that the integrands will tend to zero. If this fails, and the
integral is able to factored in a form equivalent to Jordan’s lemma, it then can be employed
to demonstrate the result is zero. For most purposes, these two lemmas should be all that
is needed to adequately prove this.
8.3.1 ML Limit
Let C denote a contour represented by the parametrization z = z(t)(a ≤ t ≤ b), it then
follows that the integration of a function f can be expressed as
Z Z b
f (z)dz = f [z(t)]z 0 (t)dt
C a
In order to obtain a maximum value, a modulus is taken
Z Z b Z b
0
|f [z(t)]||z 0 (t)|dt
f (z)dz = f [z(t)]z (t)dt= (17)
C a a
the function f is taken to be bounded on the contour C, such that |f (z)| ≤ M , for any
nonnegative constant M . By taking the maximum value of f on the contour, and factoring
it out of the integrand, Eqn. (17) becomes
Z Z b
|z 0 (t)|dt
f (z)dz = M
C a
The integral on the RHS is identified as the arc length L of the contour, thus it can be
written as
Z
f (z)dz ≤ M L (18)
C
And, an upper bound to a function along any contour has been reached! Since a complex
value function is dependent on z = x + iy, it is evident that such functions are mapped
into R3 , making plotting difficult. It is easier to understand the concept of an ML bound
by downsizing by one dimension, and considering a real function f (x). The ML bound is
depicted in Figure 3.
19
8 GENERAL APPROACH TO . . . 8.3 Proving Complex Integrals Tend to Zero
Figure 3 - A real-valued function of one variable f (x) is plotted. In the figure, the maximum value M
which occurs at a point x = x0 is labelled along with length L of the “contour.” The product M L (i.e. the
area), has been shaded. As evident from the figure, the integration of f over the interval must be less than
the maximum value M of the function multiplied by the length of the contour C.
In this scenario, the maximum value M , occurring at x = x0 , is labelled alongside the length
of the “contour” C. The contour
R x1 in this single real variable case is the interval 0 ≤ x ≤ x1 .
The value of the integral 0 f (x)dx is the area under the curve, and it is evident that the
value of this integral must always be less than or equal to the product of maximum value M
of the function and the length L of the contour (shown as the shaded region above). While
some of the details in the derivation are not the same in the complex case, and it is not
justifiable to transpose this result from real to complex calculus, the result and idea remains
the same in both real and complex mathematics.
This limit will prove handy when certain integrals must be shown to tend to zero when
applying residue theorem. This can be shown if the upper bound M L = 0 for a complex
integral. Since a modulus is always a positive quantity, the inequality must then actually be
an equality, such that the integral is identical to zero.
lim MR = 0
R→∞
This to say, if an appropriate function g(z) can be factored into a form of Eqn. (19), where
the constituent function f (z) meets the above criterion, then the integral is equal to zero.
20
9 REFERENCES 8.4 Integrals Containing Trigonometric . . .
where the above operators denote taking the real (Re[ ]) and imaginary (Im[ ]) parts of the
function. However, the extraction of either a cosine or a sine is only convenient so long as the
complex exponential is not complicatedly embedded within the function. One of the more
common scenarios wherein this arises is the case when trigonometric function(s) are involved
in sums and/or differences in the denominator. In these instances, it is not possible to easily
take the real or imaginary part of the complex version of the function so as to isolate either
the cosine or sine function. Thus, these integrals may prove difficult in moving to complex
space in order to integrate, and ultimately obtain the desired real result. However, these
integrals can be solved under special circumstances provided that the integral is taken to the
complex domain by slightly different means.
If the limits of integration encompass all 2π radians of a circle, then the real-valued
limits can be taken into the complex domain by way of introducing a circular contour. This
is accomplished by introducing the following parametrization:
z = eiθ (0 ≤ θ ≤ 2π)
and,
dz
dz = ieiθ dθ = izdθ → dθ =
iz
Then, by using the identities relating the sine and cosine functions to z:
z + z −1 z − z −1
cos θ = and sin θ =
2 2i
the integral can be taken into the complex domain and solved with residue theorem. This
result will be equivalent to the original value of the real integral.
9 References
1. Barrett, David E. Math 555: Complex Variables and Applications Class Notes. Septem-
ber 2007 - December 2007.
2. Brown, James Ward. Churchill, Ruel V. Complex Variables and Applications, Seventh
Edition. McGraw-Hill Companies, Inc. New York, NY. 2004.
21
11 APPENDIX
10 Acknowledgements
I would like to thank Benjamin Tong Yee, who at the time of this writing is attending
the University of Michigan for his graduate degree in Nuclear Engineering & Radiological
Sciences, for his suggestion on how to typeset the Residue notation in LATEX. Furthermore, I
found it silly to scour the internet and/or paraphrase theorems. Thus, many of the theorems,
and definitions cited in this guide are taken verbatim from Chuchill’s Complex Variables and
Applications, and are the sole property of the authors. This was the book from which I stud-
ied complex mathematics of a single variable, and I wish to thank the authors for a splendid
text. Finally, a special thanks to my Math 555 professor, Dr. David E. Barrett, at the
University of Michigan for making the course so accessible to an engineering, and not math-
ematics, student like myself as well as for his excellent instruction. I have attempted to do
for all of you what he did for me. This guide, under no circumstances, is to be used for profit,
especially since many of the theorems are attributed entirely to the authors. Please do not
hesitate to send any enquiries, corrections, and/or concerns to me at dsirajud@umich.edu.
11 Appendix
11.1 Glossary
1. Analytic: a function is analytic at a point z0 if it is both continuous, and satisfies the
Cauchy-Riemann equations at z0 .
2. Branch cut: Typically this is a restriction that is applied on the angular value θ of
a multiple-valued function so as to enforce that it is single-valued (e.g. the complex
logarithm function), and inherits analytic properties. The cut can be either an infinite
ray (constant angle), or a curve. The only requirement is that the function along the
cut is discontinuous such that it preserves its single-valuedness.
4. Holomorphic: See analytic, strictly speaking these two terms are not entirely synony-
mous, although the small difference between them is of no consequence for applications
of the type described in this guide.
2. ez = ∞ zn
P
n=0 n! (|z| < ∞)
22
11 APPENDIX 11.2 Useful Series Representations
P∞ n z 2n+1
3. sin z = n=0 (−1) (2n+1)! (|z| < ∞)
P∞ n z 2n
4. cos z = n=0 (−1) (2n)! (|z| < ∞)
P∞ z 2n+1
5. sinh z = n=0 (2n+1)! (|z| < ∞)
P∞ z 2n
6. cosh z = n=0 (2n)! (|z| < ∞)
These series can be obtained from each other. For instance, the cos z series is yielded from
differentiation of the sin z series, and that the cosh z expansion can be found via the relation
cosh z = cos(iz). One can find series for tan z by polynomial division of the series expansions
for sin z/ cos z, etc. Of the above, one of the more useful expressions is series (1), which can
be used to find expressions for any form of the reciprocal of any difference or sum involving
a constant and any power of z. This is accomplished by substitution of z for an appropriate
form to find the desired expansion.
For example, in order to obtain the series representation for 1/(4 + z 2 ), begin with series
(1)
∞
1 X
= wn (|w| < 1)
1 − w n=0
Enforcing the substitution w = −z 2 /4, the above equation becomes
∞
1 X z 2n
= (|z| < 2)
1 + (z 2 /4) n=0
4n
Multiplying and dividing both sides of the equation by a form of unity (i.e. 4/4)
∞
1 X 4 z 2n
4 =
4 + z2 n=0
4 4n
∞
1 X z 2n
= (|z| < 1)
4 + z2 n=0
4n+1
*N.B. Something that was not mentioned in this guide was the idea of the radius of conver-
gence for power series expansions. This is to say that a function f (z) has a valid convergent
power series representation only within a certain radius |z|. The radius of convergence for
each of the above series has been noted parenthetically, and must be payed with close at-
tention. It is important when finding series expansions to mind the radius of convergence,
elsewise the series expression arrived at will not be correct. Luckily, this is of little setback,
as all that is needed to do is to express the function in a fashion such that for all points
23
11 APPENDIX 11.3 Examples
z of a considered domain, the expanded term holds values that are within the radius of
convergence. This is easily accomplished by factorization.
11.3 Examples
11.3.1 Real Integration: Functional Replacement and ML Bound
The real integral
Z ∞ Z ∞
dx 1 dx
2
=
0 2x + 8 2 0 x2+4
can be evaluated by first transposing the function f (x) contained in the integrand into a
complex function f (z),
1 1
f (z) = =
z2 +4 (z + 2i)(z − 2i)
The singularities at z = ±2i are evident from the factorization on the RHS. A contour C
is chosen to be a semicircle in the upper-half plane centered at the origin with a radius
R chosen to be large enough to contain the singularity z = 2i. The semicircular arc that
excepts the real line is denoted by CR . The factor of (1/2) in the problem statement will be
neglected for now, and incorporated at the end. Residue theorem implies
I Z R Z
dz dx dz
2
= 2
+ 2
= 2πi Res f (z)
C z +4 −R x + 4 CR z + 4 z=2i
24
11 APPENDIX 11.3 Examples
1 1
z 2 + 4 = |z 2 + 4|
1
= 2
|z| + 4
1
= 2
R +4
= M
πR
ML = lim
R2 + 4
R→∞
π/R
= lim
R→∞ 1 + 4/R2
= 0
Since the upper bound of the integral is equal to zero, then the integral itself must be zero,
Z
dz
2
=0
CR z + 4
The only term left to compute is residue in Eqn. (20). This is done by functional replacement
of the complex function f (z)
1 φ(z)
f (z) = =
(z + 2i)(z − 2i) z − 2i
where
1
φ(z) =
z + 2i
The residue at z = 2i is then equivalent to φ(2i)
25
11 APPENDIX 11.3 Examples
Z ∞
dx 2πi
=
−∞ x2+4 4i
π
=
2
Notice that the function in the integrand of the real integral is even (f (−x) = f (x)), this
allows one to obtain the result for the semi-infinite integral by dividing the result by a factor
of 2, that is to say
Z ∞
dx π
2
=
0 x +4 4
Finally, incorporating the factor of (1/2) that was contained in the original problem state-
ment on the RHS, the above equation is multiplied by (1/2)
Z ∞
1 dx 1π
=
2 0 x2 +4 24
Multiplying through on the left and right hand sides, yields the final result in its original
form
Z ∞
dx π
2
=
0 2x + 8 8
*N.B. The integral could have been solved via real integration by introducing the trigono-
metric substitutions
26
11 APPENDIX 11.3 Examples
zeiz
f (z) =
z2 + 9
It is more convenient to transpose the sine function into a complex exponential rather than
to a complex-valued sine function, as the sine term can be easily extracted by taking the
imaginary part of this integral, while a complex sine function would provide no such conve-
nience. The singular points are located at the zeros of the polynomial in the denominator,
that is to say, when z 2 + 9 = 0. Solutions to this equation are found easily by factorization
zeiz zeiz
=
z2 + 9 (z + 3i)(z − 3i)
Singularities are located at z = ±3i. A semicircular contour C is chosen, centered at z = 0,
with a sufficiently large radius R chosen such that the point z = 3i is contained within the
semicircular domain. The semicircular arc is denoted as CR . Note, that when choosing this
contour, the point z = −3i will not contribute to the value of the integral and is therefore
neglected in the subsequent analysis.
Applying residue theorem, and breaking up the integral into complex and real-valued
parts admits the following
zeiz dz
I
= 2πi Res f (z)
C z2 + 9 z=3i
Z R ix Z iz
xe dx ze dz
+ = 2πi Res f (z)
−R x2 + 9 CR z2 + 9 z=3i
Z R
xeix dx zeiz dz
Z
= 2πi Res f (z) −
−R x2 + 9 z=3i CR z2 + 9
Adjustments are then made to the limits of integration so that they fit those given in the
original problem, i.e. let R → ∞.
R
xeix dx zeiz dz
Z Z
lim = 2πi Res f (z) − lim
R→∞ −R x2 + 9 z=3i R→∞ C z2 + 9
∞ ix Z R
zeiz dz
Z
xe dx
= 2πi Res f (z) − lim (21)
−∞ x2 + 9 z=3i R→∞ C
R
z2 + 9
Where the bounds have been informally represented as explicit limits of integration for the
real integral, and left in front of the complex integral on the RHS. The limit is not written
as being applied to the residue term, as the residue contains no further dependence on the
radius of the contour, provided that it was all ready contained in the domain interior to
it before the limit was taken. In order to solve the RHS of the above equation, begin by
proving the complex integral is equal to zero.
Since the complex integral contains a complex exponential, and given the limit of the
radius R of the contour tending to infinity, it is evident that the integral cannot be easily
27
11 APPENDIX 11.3 Examples
shown to be zero by way of an ML bound. Jordan’s lemma is invoked to prove this, but
before it can be used the function contained in the integrand must first be factored in the
same form as the statement of the theorem. That is to say, the function f (z) must be shown
capable of being expressed as f (z) = g(z)eiz , for some complex function g. This is easily
accomplished by examining the function in the integrand.
zeiz
= g(z)eiz
z2 + 9
where
z
g(z) =
+9z2
The function g immediately meets all the requirements of the theorem, there exists a radius
R0 < R for a positive semicircle in the upper half plane where g is analytic to all points
exterior. Then, by way of Jordan’s lemma, it can be said that
zeiz dz
Z Z
lim = lim g(z)eiz dz = 0
R→∞ C z 2 + 9 R→∞ C
R R
zeiz
f (z) =
z2 + 9
zeiz
=
(z + 3i)(z − 3i)
φ(z)
=
z − 3i
where
zeiz
φ(z) =
z + 3i
The resulting expression shows a simple pole, the residue of the function f is equivalent to
the value of φ(z) at z = 3i.
∞
xeix dx e−3
Z
= 2 πi
x2 + 9
−∞ 2
−3
= iπe
∞
xeix dx
Z
= Im iπe−3
Im 2
x +9
Z −∞
∞
x sin xdx π
2
= 3
−∞ x + 9 e
z − z −1
sin θ =
2i
and
dz
dθ =
iz
inserting these expressions into the original integral allows the integral to be taken into the
complex domain.
I
dz 1
−1
|z|=1 5 + (4/2i)(z − z ) iz
The function contained in the integrand is manipulated until the residues can be recognized.
29
11 APPENDIX 11.3 Examples
1 1 1 1
−1
= −1
5 + (4/2i)(z − z ) iz 5 + (2/i)(z − z ) iz
1 1
=
5 − 2i(z − z −1 ) iz
1 1
= −1
5 − 2iz + 2iz iz
i
= −
5z − 2iz 2 + 2i
i
= 2
2iz − 5z − 2i
1/2
= 2
z + (5i/2)z − 1
the quadratic formula is used to find the roots, z1 and z2 of the polynomial in the denomi-
nator.
1 11
z1 = i and z2 = − i
4 4
Since the contour is designated as |z| = 1, it is evident that the only root, and hence
singularity, that is interior to it is the root z1 . Given these roots, the integral can be
expressed as
Z
(1/2)dz
|z|=1 (z − z1 )(z − z2 )
φ(z) (1/2)
where φ(z) =
z − z1 z − z2
Thus, the integral in the problem statement is equal to the product of 2πi and the residue
at z = z1 . The residue is equivalent to φ(z1 ):
(1/2)
Res = φ(z1 )
z=z1 (z − z1 )(z − z2 )
(1/2)
=
z1 − z2
1
= − i
6
Thus,
Z
(1/2)dz 1 π
= 2πi Res f (z) = 2πi(− i) =
|z|=1 (z − z1 )(z − z2 ) z=z1 6 3
30
11 APPENDIX 11.3 Examples
Since this integral is equivalent to the original integral, it can then be said that
Z π
dθ π
=
−π 5 + 4 sin θ 3
Figure 4 - A plot is shown of the Fresnel cosine function. The frequency increases with x.
As evident from above, the frequency of the function increases with x, until the wavelength
of the function tends to zero as x → ∞. The function oscillates above and below the x-axis,
suggesting that it is possible in the limit for large x, that sufficient contributions from the
area sweeped out by the function will be cancelled out by its negative and positive portions,
leading to a finite result. This is precisely the case, and this finite value of the integral is
found via complex integration.
The real-valued function f (x) = cos x2 is transposed into the complex domain as a
complex exponential f (z) = exp(iz 2 ). The complex function f is identified to hold no
singularities; however, this only suggests that the complex closed contour integral of this
function about any domain is zero, not that all path integrals making up the closed contour
31
11 APPENDIX 11.3 Examples
are themselves zero. Thus, a solution could still be obtained in this manner. In order
to choose a proper contour, begin by examining the behavior of the complex exponential
function in the complex plane, this is qualitatively shown in the figure below:
2
Figure 5 - The behavior of the function eiz is examined. In (a) the function is described along a positively
oriented closed circular contour. The regions where the function decays and increases are labelled. In (b), a
proposed quarter circle contour is shown that evades the increasing regions of the function. And, in (c) an
eighth circle is given, which is a contour that allows for the integration to yield a proper, finite value.
It is seen by inspection that, when examined along a positively oriented closed circular con-
tour, the complex exponential function exp(iz 2 ) increases and decays in different quadrants.
The integral of the function in a region of growth is not capable of admitting a finite result
since the limits of integration (and hence the radius of any circular path) must eventually be
extended to infinity in accordance with the original problem. However, it would seem that
it is still possible to integrate, so long as the increasing regions of the function are avoided.
Thus, a first choice of contour could be the quarter-circle illustrated in Figure 5(b). However,
when choosing this contour, applying residue theorem, and taking limits, the solution to the
problem does not yield a value, but rather an identity:
Z ∞ Z ∞
2
cos x dx = sin x2 dx (24)
0 0
This is certainly nice to know, but a value of both these functions integrated over the
stipulated limits is still yet to be found. In order to integrate the function over the desired
bounds so as to obtain a finite result, the 1/8th circle, shown in Figure 5(c), is used.
Applying residue theorem to the system provides the following statement
I
2
eiz dz = 0
C
32
11 APPENDIX 11.3 Examples
where C is used to denote the closed contour shown in Figure 5(c). Notice how since there
are no residues contained within the contour, the contour integral of the function is equal
to zero. That is to say, residue theorem is reduced to Cauchy’s integral theorem in this
instance. The integral can further be broken up into a summation of path integrals, where
the sum of the paths is equivalent to the contour C.
Z Z Z
iz 2 iz 2 2
e dz + e dz − eiz dz = 0 (25)
C1 C2 C3
| {z } | {z } | {z }
I II III
The integrals have been labelled as I, II, and III for convenient referencing, and each path
C1 , C2 , and C3 are as shown in Figure 5(c). Each integral is evaluated below.
Integral I
The only work that needs to be done on this integral is to parameterize it along the real
line, and take limits. Thus,
Z Z R
iz 2 2
e dz = eix dx
C1 0
since f (z) = f (x, y), and y = 0 for all points on the real line. Letting R → ∞, Integral I
becomes
Z ∞
2
eix dx. (26)
0
where it is noted that the real part of this integral can be taken to make this equation of the
same form as the original integral in the problem statement (Eqn. (22)).
Integral II
The above equation is now in the form of the statement in Jordan’s lemma, where g(z) =
eiz(z−1) . Since the function g is entire, that is, it is analytic for all points in the complex
plane, it meets the requirements for Jordan’s lemma. Thus, it can be said that
Z
2
eiz dz = 0 (27)
C2
Integral III
33
11 APPENDIX 11.3 Examples
1 + i t0
Z Z h
π 2
iz 2
e dz = √ exp i ei 4 t dt
C3 2 0
1 + i t0
Z 2
i π4 t2
= √ exp i e dt
2 0
1 + i t0
Z
= √ exp(i2 t2 )dt
2 0
1 + i t0 −t2
Z
= √ e dt
2 0
Furthermore, if t0 is allowed to extend to infinity then
1 + i ∞ −t2
Z Z
iz 2
e dz = √ e dt
C3 2 0
The parametrization has rendered the original integral into a Gaussian integral. The above
integral can be evaluated in a number of ways. Such methods include Feynmann’s so-called
parametric integration, or by integrating the square of the integral in polar coordinates, or
by using the
√ gamma function Γ(t). It is identified that the integral above is equivalent to
Γ(3/2) = π/2. Thus, the solution to the above equation can be written as
√
1 + i ∞ −t2
Z r r
1+i π 1 π i π
√ e dt = √ = + (28)
2 0 2 2 2 2 2 2
Inserting Eqns. (25), (26), and (27) into (24), and solving for the real integral reveals
Z ∞ r r
2 1 π i π
eix dx = +
0 2 2 2 2
Taking the real part of the above equation, and combining this result with the identity shown
in Eqn. (24), the solution is found to be
Z ∞ Z ∞ r
2 2 1 π
cos x dx = sin x dx = (29)
0 0 2 2
34
11 APPENDIX 11.3 Examples
sinh(inπ) = 0
where n is an integer. From this statement, all zeroes of the function lie on the y-axis
whereas only one singularity (corresponding to n = 0) is contained within the contour
|z| = 1. Application of residue theorem implies
I
coth zdz = 2πi Res coth z
C z=0
where the residue at z = 0 can be found by invoking the quotient rule of Section 6.2.2 with
p(z) = cosh z and q(z) = sinh z.
cosh z
coth z =
sinh z
p(z)
=
q(z)
Since p(0) 6= 0, q(0) = 0, and q 0 (0) = p(0) 6= 0, using the theorem is valid and the residue is
given by
cosh(0)
Res coth z = =1
z=0 cosh(0)
Thus,
I
coth zdz = 2πi.
|z|=1
35
11 APPENDIX 11.3 Examples
where the latter series follows from the replacement of z with −z 2 in series (1). The term z −1
is all ready in the form of a series expansion centered about z = 0. Beginning by multiplying
the first few terms of following two known Maclaurin series
z 1 1 2 1 3
1 + z + z + z + . . . 1 − z2 + z4 + . . .
e =
1 + z2 2 6
1 1
= 1 + z + z2 + z3 + . . .
2 6
− z2 − z3 + . . .
1 1 5
ez 2
= 1 + z − z2 − z3 + . . .
1+z 2 6
36
11 APPENDIX 11.3 Examples
Finally, the z −1 term is multiplied through yielding the full expansion of the function f (z)
ez 1 1 5 2
= + 1 − z − z + ...
z(1 + z 2 ) z 2 6
Identifying the coefficient in front of the z −1 term as the residue provides the desired result.
ez
Res =1
z=0 z(1 + z 2 )
Thus, while in this particular case it was not the quickest way to find the function’s residue (as
functional replacement would have required much less work), direct extraction of the residue
from the power series expansion was shown to still be a possible method. Furthermore, for
some functions (e.g. exp(cos z)), this method can actually prove to be less tasking, so it is
important to not forget where residues come from.
As discussed earlier, usually only a small number of terms of the series need be computed in
order to find a residue.
37