The Residue Theorem

“Integration Methods over Closed Curves for Functions with Singular-

ities”

We have shown that if f (z) is analytic inside and on a closed curve C,

then Z
f (z)dz = 0.
C
We have also seen examples where f (z) is analytic on the curve C, but
not inside the curve C and
Z
f (z)dz 6= 0
C

(for example f (z) = 1/z over the unit circle centered at 1). In the
latter instance however, we had to calculate the integral directly by
brute force. In the following few sections, we shall develop methods
to integrate functions with singularities over closed curves which avoid
direct computation and then we shall use them to solve other related
(and seemingly unrelated) problems in math.

1. The Cauchy Residue Theorem

Before we develop integration theory for general functions, we observe
the following useful fact.
Proposition 1.1. Suppose that f (z) has an isolated singularity at z0
and
X∞
f (z) = ak (z − z0 )k
k=−∞

is its Laurent expansion in a deleted neighbourhood of z0 . Then if C is

any circle surrounding z0 and containing no other isolated singularities
and it is oriented counterclockwise, then
Z
f (z)dz = 2πia−1 .
C

Proof. By our earlier results, in the Laurent expansion for f (z) around
z0 , for a given k we have
1 f (z)
Z
ak = dz,
2πi C (z − z0 )k+1
so using k = 1, the result follows.

1
2

The coefficient a−1 will be very important for our uses so we give it its
own name.
Definition 1.2. If
∞
X
f (z) = ak z k
k=−∞

in a deleted neighbourhood of z0 , then we call the coefficient a−1 the

residue of f at z0 and we denote it by Res(f ; z0 ).
Evaluation of residues is fairly straight forward and we do not (always)
have to find the Laurent expansion explicitly to find residues. Specifi-
cally, if we have a pole, we can use the following results.
Proposition 1.3. Suppose z0 is a pole of f (z) = A(z)/B(z).
(i ) If z0 is a simple pole (of order 1) and B(z) has a simple zero
at z0 , then
A(z)
a−1 = lim (z − z0 )f (z) =
z→z0 B ′ (z)
(ii ) If z0 is a pole of order k, then
1 dk−1
a−1 = lim (z − z0 )k f (z)
z→z0 (k − 1)! dk

Proof. (i ) By assumption, we have

a−1
f (z) = + a0 + a1 (z − z0 ) + . . .
z − z0
so
(z − z0 )f (z) = a−1 + a0 (z − z0 ) + a1 (z − z0 )2 + . . .
giving
lim (z − z0 )f (z) = lim (a−1 + a0 (z − z0 ) + a1 (z − z0 )2 + . . . ) = a−1 .
z→z0 z→z0

Alternatively, we have
A(z) (z − z0 )
a−1 = lim (z − z0 )f (z) = lim (z − z0 ) = lim A(z)
z→z0 z→z0 B(z) z→z 0 B(z) − B(z0 )
since B(z0 ) = 0, so we get
(z − z0 ) A(z) A(z0 )
a−1 = lim A(z) = lim B(z)−B(z0 )
=
z→z0 B(z) − B(z0 ) z→z0 B ′ (z0 )
z−z0

noting that B ′ (z0 ) 6= 0 since the pole is of order 1 (using the

homework problem).

 3

This proposition can be used to evaluate the residue for functions with
simple poles very easily and can be used to evaluate the residue for
functions with poles of fairly low order. However, it becomes increas-
ingly difficult the higher the order of the pole, and impossible with
essential singularities. In these cases, we have no choice but to return
to the Laurent expansion.
2
Example 1.4. Find the residues of f (z) = sin (z)/z 2 and g(z) = e−1/z
at z = 0 and use it to evaluate
Z
f (z)dz
C
and Z
g(z)dz
C
where C is the unit circle centered at the origin..
(i ) We could apply the above results, but first we would need
to determine what the order of the pole of f (z) at z = 0 is
(it looks like a pole of order 2, but recall that sin (z)/z has a
removable singularity at z = 0). With this in mind, we instead
use the Laurent expansion. We have
z3 z5
sin (z) = z − + − ...
3! 5!
so
sin (z) 1 z z3
= − + −...
z2 z 3! 5!
so the residue is 1 (and in fact z = 0 is a pole of order 1).
Using the earlier proposition, we have
Z
f (z)dz = 2πi ∗ 1 = 2πi.
C

(ii ) We have
z2 z3
ez = 1 + z + + + ...
2! 3!
so
2 1 1 1
e−1/z = 1 − 2
+ 4
− + ...
z 2!z 3!z 6
so the residue is 0. Using the earlier proposition, we have
Z
f (z)dz = 2πi ∗ 0 = 0.
C

By the first proposition we gave, we can use residues to evaluate inte-

grals of functions over circles containing a single. To evaluate general
integrals, we need to find a way to generalize to general closed curves
which can contain more than one singularity. First we recall some
simple facts and definitions about closed curves.
4

Definition 1.5. A closed curve is called simple if it does not intersect

itself (the book calls such curves “regular curves”).
Theorem 1.6. Any closed curve in C can be written as a union of
simple closed curves.
Proposition 1.7. If C is a closed curve and C = C1 ∪· · ·∪Cn (with the
relevant imposed orientations) where the Ci are simple closed curves,
then
Z Xn Z
f (z)dz = f (z)dz
C i=1 Ci

Example 1.8. Evaluate

sin (z)
Z
dz
C z2
where C is one of the following curves:
(i ) We can break up C into two curves, one oriented counter-
clockwise around the origin - C1 , and the other clockwise - C2 .
Next note that since there are no singularites contained be-
tween C1 and a small circle C3 centered at the origin oriented
counterclockwise, and C2 and C3 , we have
Z Z
f (z)dz = f (z)dz
C1 C3

and Z Z
f (z)dz = − f (z)dz
C2 C3

Then using our earlier proposition, we have

Z
f (z)dz = 2πi
C3

It follows that
Z
f (z)dz = 2πi − 2πi = 0.
C
 5

(ii ) We can break up C into two curves, both oriented clockwise

around the origin. Using the same argument as above, it fol-
lows that
Z
f (z)dz = −2πi − 2πi = −4πi.
C

(iii ) We can break up C into two curves, both oriented counter-

clockwise around the origin. Using the same argument as
above, it follows that
Z
f (z)dz = 2πi + 2πi = 4πi.
C

With these results and observing the the examples above, it suffices to
determine a formula to integrate a function f (z) over simple closed de-
pending upon its orientation. Before we develop the formula however,
we have a couple of necessary definitions.
Definition 1.9. If C is a closed simple curve, we call the compact
region bounded by C the “inside” of C.
Definition 1.10. We say a simple closed curve C is oriented coun-
terclockwise if as a particle moves around C in the direction of the
orientation, the “inside” of C is to the left of the particle.
We are now ready to prove the main result.
Theorem 1.11. (Cauchy’s Residue Theorem) Suppose f (z) is analytic
in a simply connected region D except for isolated singularities. Let γ
be a simple closed curve in D which does not contain any singularities
oriented counterclockwise and suppose the singularities z1 , . . . , zn lie in
6

the inside of C. Then

Z n
X
f (z)dz = 2πi Res(f, zi ).
C i=1

Proof. In order to prove this result, we shall use a generalization of the

previous result we proved stating that the integral over a closed sim-
ple curve is equal to the integral over any closed curve inside provided
f (z) is analytic in between the two curves. Specifically, if z1 , . . . , zn
are the singularities inside C, we can draw small circles around them,
C1 , . . . , Cn which are fully contained in the interior of C and lines con-
nected these circles to the boundary, L1 , . . . , Ln , see illustration.
C

L1 Ln

z_1 z_n
C1
Cn

Starting at the end of L1 on C, we define the curve K by traversing L1

to C1 , traversing C1 counterclockwise, traversing L1 back toward Cand
then traversing C counterclockwise until the end of L2 and continuing
in this fashion until we traverse the whole of C. As with the earlier
proposition, the interior of K is simply connected, and since z1 , . . . , zn
are the only singulairites of f (z), we get
Z
f (z)dz = 0.
K

Next note that since all the Li ’s cancel (since we are traversing each
curve in both directions), we get
Z
f (z)dz = 0
C∪−C1 ∪−C2 ∪···∪−Cn
or
Z n Z
X
f (z)dz = f (z)dz.
C i=1 Ci

Applying our first result, we get

Z n
X
f (z)dz = 2πi Res(f, zi ).
C i=1


 7

Corollary 1.12. Suppose f (z) is analytic in a simply connected region

D except for isolated singularities. Let γ be a simple closed curve in D
which does not contain any singularities oriented clockwise and suppose
the singularities z1 , . . . , zn lie in the inside of C. Then
Z Xn
f (z)dz = −2πi Res(f, zi ).
C i=1

Proof. This is simply due to the fact that C is oriented in the opposite
direction to that given in the previous result. 

2. Application of the Residue Theorem

We shall see that there are some very useful direct applications of the
residue theorem. However, before we do this, in this section we shall
show that the residue theorem can be used to prove some important
further results in complex analysis. We start with a definition.
Definition 2.1. We say f is meromorphic in a domain D if f is analytic
in D except possibly isolated singularities.
Theorem 2.2. Suppose C is a simple closed curve. If f is meromor-
phic inside and on C and contains no zeros or poles on C and if Z =
number of zeros (counted with multiplicity) and P = number of poles
(counted with multiplicity), then
Z ′
1 f (z)
dz = Z − P.
2πi C f (z)
Proof. In order to show this, we shall calculate the residues at each of
the poles of f ′ /f in D, the interior of C and then apply the Residue
Theorem. First note that f ′ /f will be analytic at all points except the
zeroes and poles of f (z), so we consider these two possibilities.
First, if z0 ∈ D is a zero of f (z), of multiplicity k, then f (z) = (z −
z0 )k g(z) for some function g(z) which is analytic and nonzero at z0 .
Then we have f ′ (z) = k(z − z0 )k−1 g(z) + (z − a)k g ′ (z), so
f ′ (z) k(z − z0 )k−1g(z) + (z − a)k g ′(z) k g ′ (z)
= = +
f (z) (z − z0 )k g(z) z − z0 g(z)
near z0 . Since g ′/g is analytic at z0 , it follows that f ′ /f will have a
simple pole at z0 with residue k.
If (z − z0 ) is a pole of f (z) of order k, then f (z) = g(z)(z − z0 )−k for
some function g(z) which is analytic and nonzero at z0 . Then we have
f ′ (z) = −k(z − z0 )−k−1 g(z) + (z − a)−k g ′ (z), so
f ′ (z) −k(z − z0 )−k−1g(z) + (z − a)−k g ′(z) k g ′(z)
= =− +
f (z) (z − z0 )−k g(z) z − z0 g(z)
8

near z0 . Since g ′/g is analytic at z0 , it follows that f ′ /f will have a

simple pole at z0 with residue −k. Applying the residue theorem, the
result follows.

Theorem 2.3. (Argument Principle) Suppose f (z) is analytic and
nonzero on a closed simple curve C, fix some z0 on C and let ∆C Arg(f )
denote the change in argument (measured in total radians of change -
not restricted to 0 6 Arg(z) 6 2π) from z0 when traversing the curve
in a counterclockwise direction. Then
Z ′
1 1 f (z)
∆C Arg(f ) = dz.
2π 2πi C f (z)
Proof. First, if z(t) parameterizes the curve C with a 6 t 6 b, then we
have Z ′ Z b ′
f (z) f (z(t))
dz = ż(t)dt.
C f (z) a f (z(t))

Suppose that f (z) = r(t)eiϑ(t) . Next, recall (as shown in Chapter 3)

that if w(t) = f (z(t)), then w ′ (t) = f ′ (z(t))ż(t) (this is simply the
chain rule for complex functions depending upon a single parameter).
It follows that along smooth arcs making up the curve C, we have
d d
f ′ (z(t))ż(t) = f (z(t)) = (r(t)eiϑ(t) ) = r ′(t)eiϑ(t) + ir(t)eiϑ(t) ϑ′ (t).
dt dt
Therefore
Z b ′ Z b ′
r (t)eiϑ(t) + ir(t)eiϑ(t) ϑ′ (t)
Z ′
f (z) f (z(t))
dz = ż(t)dt = dt
C f (z) a f (z(t)) a r(t)eiϑ
Z b ′ Z b b b
r (t) ′
 
= dt + iϑ (t)dt = ln (r(t)) + ϑ(t) = 0 + ∆C Arg(f )

a r(t) a 1 a
since the first integral is purely real and the second integral is simply
the change in argument from z(a) to z(b) (which is measured as the
actual change, even though z(a) = z(b)).

Corollary 2.4. Suppose C is a simple closed curve. If f is mero-
morphic inside and on C and contains no zeros or poles on C and if
Z = number of zeros (counted with multiplicity) and P = number of
poles (counted with multiplicity), and ∆C Arg(f ) denotes the change in
argument when traversing once around C, then
1
Z −P =
∆C Arg(f ).
2π
The argument principle provides a way to measure the angle change
when traversing a closed curve once. Since there are 2π radians in a
 9

complete revolution, the value of

Z ′
1 f (z)
dz
2πi C f (z)
gives the number of times a curve wraps around the origin (and is
consequently sometimes referred to as the winding number of C). Given
the corollary, it follows that Z − P gives the winding number of a curve
C. A consequence of the argument principle and these observations is
the following.
Theorem 2.5. (Rouchés Theorem) Suppose that f and g are analytic
inside and on a closed regular curve C and that |f (z)| > |g(z)| for
all z ∈ C. Then Z(f + g) = Z(f ) inside C (where Z(f ) denotes the
number of zeros of f inside C).
Proof. First, on C observe that |f (z) + g(z)| > |f (z)| − |g(z)| > 0.
Therefore, since we are assuming |f (z)| > |g(z)| > 0, it follows that
f (z) has no zeros on C and neither does f (z) + g(z). Since both f and
f + g are analytic, it follows that the number of zeros of f and f + g
inside C are (∆C Arg(f ))/2π and (∆C Arg(f + g))/2π respectively.
Next using the basic properties of the argument, we observe that
g f +g
∆C Arg(1 + ) = ∆C Arg( ) = ∆C Arg(f + g) − ∆C Arg(f ).
f f
Therefore it suffices to show that
g
∆C Arg(1 + ) = 0.
f
Observe however that |f | > |g|, so |g/f | < 1 and consequently
  

 1+ g(z) 
− 1 < 1.
 f (z)
However, this means that all the values of 1 + f /g take place inside the
circle of radius 1 centered at z = 1, and consequently there is no way
the function 1 + g/f could wrap around the origin. It follows that
g
∆C Arg(1 + ) = 0
f
and hence Z(f + g) = Z(f ).

We finish with a couple of examples.
Example 2.6. (i ) Show that z 6 + 3z 4 − 2z + 8 has all 6 zeros
satisfying 1 < |z| < 2.
We shall attempt to apply Rouch/’es Theorem. First ob-
serve that on |z| = 2, if f (z) = z 6 and g(z) = 3z 4 − 2z + 8,
then |f (z)| = 64 and |g(z)| 6 3 · 23 + 2 · 2 + 8 = 60, so
10

|f (z)| > |g(z)|. It follows that f (z) + g(z) has the same num-
ber of zeros as f (z) for |z| < 2, so all 6 zeros of z 6 +3z 4 −2z +8
occur in |z| < 2.
Next observe that for |z| = 1, if we take f (z) = z 6 + 8 and
g(z) = 3z 4 −2z, then |f (z)| > 8−1 = 7 and |g(z)| 6 3+2 = 5,
so it follows that f (z) + g(z) has the same number of zeros in
|z| < 1 as f (z) = z 6 + 8. However, all the zeros of z 6 + 8 have
modulus 8( 1/6) > 1, so none occur in |z| < 1. Thus it follows
that no zeros of z 6 + 3z 4 − 2z + 8 occur in |z| < 1. Thus all
zeros occur in the annulus 1 < |z| < 2.
(ii ) Show that the quartic polynomial p(z) = z 4 + z 3 + 1 has one
zero in each quadrant.
First observe that there is no zero on the real axis (since
x4 +x3 > −1 for all real x using elementary calculus). Likewise,
there is no zero on the imaginary axis since (iy 4 ) + (iy)3 + 1 =
y 4 − iy 3 + 1 = (y 4 + 1) − iy 3 always has positive real part i.e.
y 4 + 1 > 0 for all real y. Thus the zeros of p(z) must occur in
the quadrants.
Next note that since p(z) has real coefficients, the zeros of
p(z) must come in conjugate pairs. Therefore, it suffices to
show that exactly one zero occurs in the first quadrant (since
its conjugate will appear in the fourth, and the other pair must
appear in the second and third).
We shall use the argument principle to count the number of
zeros (since f (z) is analytic, the value of
1
∆C Arg(f (z))
2π
will count the number of zeros contained in C). Let C be the
curve which consists of the quarter circle in the first quadrant
centered at z = 0 with radius R and the real and imaginary
line segments making this a closed curve (see illustration).

We now consider the image of C under the map f (z). First,

we have f (0) = 1. Next, the line segment 0 < x < R along the
real axis (A) maps under f (z) to the curve x4 + x3 + 1 which
is a portion of the positive real axis. The line line segment
 11

0 < y < R along the imaginary axis (B) maps under f (z) to
the curve (y 4 + 1) − iy 3 which has strictly positive real part
and strictly negative imaginary part. Finally, the quarter circle
Reiϑ with 0 6 ϑ 6 π/2 (D) maps to
 
4 4iϑ 3 3iϑ 4 4iϑ e−iϑ e−4iϑ
R e +R e +1=R e 1+ + .
R R4
It follows that the image of C under f (z) lookes something
like the following:
R f

B
f(D)
f(A)

A
f(B)

In particular, observe that

 
4 4iϑ e−iϑ e−4iϑ
Re 1+ +
R R4
has argument close to 4ϑ for large values of R, so the image of
D under f (z) will have a change in argument of approximately
2π. Next, since A and B both map to curves with strictly real
part, they could not wrap around the origin, so the change in
ϑ must be 2π. It follows that the number of zeros contained
in C for sufficiently large R will be equal to
1 1
∆C Arg(f (z)) = 2π = 1.
2π 2π

Homework:
Questions from pages 126-127; 1,2,5,6,7,8