Page 56

5 Calculus of residues

5.1 Singularities

The final bit of theory in this section is devoted to Cauchy’s residue theorem and its appli-
cations. Recall that Cauchy’s theorem stated that
I
f (z)dz = 0

so long as f (z) was analytic everywhere inside C. What if f (z) is not analytic within C ?
The answer is provided by Cauchy’s residue theorem.
We defined a singular point of a function f (z) as a point where f (z) is not analytic. This
sometimes arises because the function is discontinuous, such as at a branch cut or branch
point. More commonly, the reason for the singular behaviour is an infinity.
Suppose z = a is one of the singularities of f (z), but there are no other singular points in
its immediate neighbourhood. Then z = a is called an isolated singular point.
In this case f (z) will possess a Laurent expansion around z = a which will be valid as
far as the next nearest singular point, let’s say this is z = b. Therefore, near z = a, more
precisely for all 0 < |z − a| < |b − a|, then the expansion
+∞
an (z − a)n
X
f (z) =
n=−∞

is a convergent Laurent series representation of f (z).

Figure 14: The singular points z = a and z = b

In fact we can construct a patchwork of circles around each isolated singularity each with
an appropriate Laurent series. In this way, it is possible to cover the entire region by a
collection of series representations of the function f (z), except of course at the singularities
themselves where the function becomes infinite. It is time to classify these infinities.

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Singularities Page 57

Poles and essential singularities

If the Laurent expansion of f (z) in the neighbourhood of an isolated singular point z = a

contains a finite number of negative powers of z − a, then z = a is called a pole of f (z).

Poles: If (z − a)−m is the highest negative power in the Laurent expansion around z = a,
the pole is said to be of order m.

Essential singularity: If the Laurent expansion of f (z) in the neighboured of an isolated

singular point z = a contains infinitely many negative powers of z − a, then z = a is called
an essential singularity of f (z).

So it is possible to cure the infinity of a function with a pole by multiplying f (z) by a term
like (z − a)m . It is not possible to remove an essential singularity in this way, because it
is extremely singular. We should consider essential singularities as the most severe form of
singularity.

For instance, the function:

12
g(z) = + 3 + 5z 2 + 7z 4 + · · ·
z2
From this equation, it is clear that the function g(z) has a an infinity at z = 0. More
precisely, it has a pole at z = 0, of order 2.
1
Another example is the function: q(z) = z(z−1) 2 . Obviously there are singularities (infini-

ties) when z = 0 and z = 1. These are the only singularities and since they are isolated
we can obtain Laurent series. Suppose we require a series expansion near the point z = 1,
centred around z = 1. Such a Laurent series can be defined in a ring 0 < |z − 1| < 1.
Remember the region of convergence is always bounded by the nearest singularity.

1 1
q(z) = =
z(z − 1)2 [1 + (z − 1)](z − 1)2
= [1 + (z − 1)]−1(z − 1)−2
= [1 − (z − 1) + (z − 1)2 − (z − 1)3 + · · ·](z − 1)−2

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Singularities Page 58

1 1
= 2
− + 1 − (z − 1) + (z − 1)2 + · · ·
(z − 1) z−1
this Laurent expansion converges (absolutely) for all values of z such that, 0 < |z − 1| < 1.
The Laurent series clearly indicates a textbook case of a function with a pole of order 2 at
z=1
The function p(z) = e1/z can be written as a series. It is represented for all values of z
except z = 0 by the series, that is absolutely convergent for all |z| > 0

1 1 1
p(z) = e1/z = 1 + + + + ···
z 2!z 2 3!z 3
However it is also clear that this function has an essential singularity at z = 0.
It is also important to understand the behaviour of functions as z → ∞. A function like
f (z) = 3z 2 + z + 1, is perfectly analytic for all finite z. But what about z → ∞ ? If we
change variable (a simple example of a complex transformation)
1
ζ=
z
then every z has a unique value of ζ and vice versa. In particular the ‘point’ z = ∞
corresponds to the point ζ = 0. We call ζ = 0, or z = ∞, the point at infinity.
Suppose we have a function, say f (z) = 2z 2 , then:
f (z) = g(ζ)
where
2
g(ζ) =
ζ2
So, from the ζ perspective, g(ζ) has a pole of order 2 at ζ = 0. So we can say that, f (z) has
a pole of order 2 at z = ∞. Consider another case: f (z) = ez . Clearly f (z) is analytic and
well-behaved everywhere in the finite plane. However, the function does become infinite as
|z| → ∞. So we can think of the point at infinity as a singularity. In this case consider what
happens as z → ∞, that is as ζ → 0.
1 1 1
f (z) = ez = e1/ζ = 1 + + 2
+ + ···
ζ 2!ζ 3!ζ 3
this indicates that ζ = 0 is an essential singularity. So we say that the function f (z) = ez
has an essential singularity at infinity.

Example 1

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Singularities Page 59

Question:
Identify and classify the singularities of
e2z
f (z) = 3
z (1 + z 2 )
Answer:
Infinities arise in both numerator and denominator.

The denominator has a term z 3 so : there is a pole of order 3 at z = 0.

The term (1 + z 2 ) can also be zero, and there are two possible values. Note that
(1 + z 2 ) = (z + 1)(z − i) = 0, when z = i and z = −i. There is a pole of order 1
at z = i, and a pole of order 1 at z = −i.
Finally the numerator can be infinite as z → ∞. We have already seen that ez has
an essential singularity at z = ∞, so there is an essential singularity of f (z) at
z = ∞.

Example 2

Question:
Identify and classify the singularities of
z
f (z) =
(4 + z 2 ) sin(πz)
Answer:
Infinities arise from:
4 + z2 = 0
That is (z + 2i)(z − 2i) = 0. Thus, there is a pole of order 1 at z = 2i and a pole
of order 1 at z = −2i

Infinities might also arise from

sin πz = 0
That is z = 0, ±1, ±2, ±3, . . .. Near each of these values, say near the integer q,
we can write the Taylor series:
2
 
2  −π sin πq 
sin πz ≈ sin πq + (z − q) [π cos πq] + (z − q) + ···
2!
So that, since cos πq = (−1)q :
sin πz ≈ (z − q)π(−1)q

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Residues Page 60

Then z = q is clearly a pole of order 1. So the function also has poles of order 1 at
z = ±1, ±2, ±3, . . ..
The point z = 0, requires closer examination. Consider the behaviour of the func-
tion for small z, then sin πz ≈ πz − (πz)3 /6 + · · ·, so :
z
f (z) ≈ |z|  1
(4 + z 2 )πz(1 − π 2 z 2 /6 + · · ·)
So the zero in the denominator is exactly cancelled by the zero in the numerator,
and the function has a finite limit as z → 0.
Thus:
1
lim f (z) =
z→0 4π
and z = 0 is not a singularity.

5.2 Residues

If a function f (z) has a Laurent series near the singularity, z = a:

am a−2 a−1
f (z) = m
+ · · · + 2
+ + a0 + +a1 (z − a) + a2 (z − a)2 + · · ·
(z − a) (z − a) (z − a)
The coefficient a−1 is of great importance because of its connection with the integral of the
function. Recall Laurent’s theorem which defines the coefficient a−1 in terms of a contour
integral:

1 I
a−1 = f (z) dz
2πi C
where C encircles z = a in the positive (anticlockwise) sense.

The coefficient of (z − a)−1 in the Laurent expansion of f (z) in the neighbourhood is called
the residue of f (z) at z = a.

Now consider a simple closed curve C enclosing several isolated singularities of a function
f (z). Now shrinking the contour, deforming it smoothly without crossing any singularities
we can arrange it so that C is converted to several minicircles surrounding the singularities,

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Residues Page 61

connected by pairs of lines.

1 I 1 I 1 Z
f (z) dz = f (z) dz + f (z) dz
2πi C 2πi C1 2πi D12
1 I 1 Z
+ f (z) dz + f (z) dz + · · ·
2πi C2 2πi D23
1 I
+ f (z) dz
2πi Cn
Then the integrals along the connecting pairs of lines all vanish
Z Z
f (z) dz = 0 f (z) dz = 0 · · ·
D12 D23

1 I 1 I 1 I
f (z) dz = f (z) dz + · · · + f (z) dz
2πi C 2πi C1 2πi Cn
But all of the integrals on RHS are, by definition, just the residues of f (z) at the isolated
singularities within C.

Theorem 5.2.1 Cauchy’s residue theorem

If C is a closed curve and if f (z) is analytic within and on C except at a finite number of
singular inside C, then
I
f (z) dz = 2πi(r1 + r2 + · · · + rn )
C
where r1, r2, . . . , rn are the residues of f (z) at its singular points within C.

A simple way of getting the residues is as follows. Suppose first that f (z) has a first-order
pole at z = a, then:

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Residues Page 62

a−1
f (z) = + a0 + a1 (z − a) + · · ·
z−a
Multiply both sides by z − a, gives

(z − a)f (z) = a−1 + a0 (z − a) + a1 (z − a)2 + · · ·

Now, letting z approach a, all the terms on the RHS, except the first, go to zero. Then

a−1 = z→a
lim[(z − a)f (z)]
If f (z) has a second-order pole at z = a, then
a−2 a−1
f (z) = + + a0 + a1 (z − a) + · · ·
(z − a)2 (z − a)
To find the residue a−1 , multiply both sides by (z − a)2 .
(z − a)2 f (z) = a−2 + a−1 (z − a) + a0 (z − a)2 + a1 (z − a)3 + · · ·
Then differentiate with respect to z

d h
(z − a)2 f (z) = 0 + a−1 + 2a0 (z − a)2 + 3a1 (z − a)2 + · · ·
i

dz
Now, letting z → a:

d
[(z − a)2 f (z)]
a−1 = z→a
lim
dz
The same procedure can be extended to poles of higher order. Thus, for a pole of order m
let

Theorem 5.2.2 If f (z) has a pole of order m at z = a, then the residue of f (z) at z = a is

1 dm−1
a−1 = lim [(z − a)m f (z)]
(m − 1)! z→a dz m−1

Example 3

Find the integral of

−3z + 4
f (z) =
z(z − 1)(z − 2)
Methods of Applied Mathematics – Section B: Complex variables 2003-4
Residues Page 63

around the circle |z| = 3/2.

There are three singular points of f (z), namely poles of order one at z = 0, z = 1,
and z = 2. However, of these only z = 0 and z = 1 lie within the contour. We can
calculate the integral using the Cauchy residue theorem.
We require the residue at each pole. Consider the pole of order one at z = 0. For a
pole of order one, the residue is defined as:
(−3z + 4) 4
r1 = lim zf (z) = lim = =2
z→0 z→0 (z − 1)(z − 2) 2
Similarly at z = +1, a pole of order one, the residue is:
(z − 1)(−3z + 4) 1
r2 = lim (z − 1)f (z) = lim = = −1
z→+1 z→+1 z(z − 1)(z − 2) 1 × −1

Hence, the residue of f (z) at z = 1 is equal to −1.

Therefore, according to the residue theorem,
I −3z + 4
I= dz = 2πi[(2) + (−1)] = 2πi
C z(z − 1)(z − 2)

Example 4

Question: Find the residue of

(1 + z)
f (z) =
(1 − cos z)
at z = 0.
Answer:
Examine the function near z = 0. To simplify matters, replace cos z by its Taylor
series, near z = 0. Then cos z ≈ 1 − z 2 /2 + z 4 /24 − · · ·, so:

1+z 2(1 + z)
f (z) = =
1 − (1 − z 2 /2 + z 4 /24 − . . .) z 2 (1 − z 2 /12 + . . .)
and the factor z 2 in the denominator now identifies the pole as second order.
Now we can use one of two methods:

(a) applying the formula with m = 2, we have

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Residues Page 64

 
d  2 2(1 + z)
a−1 = lim z 2 
z→0 dz z (1 − z 2 /12 + . . .)
(1 − z 2 /12 + . . .) − (1 + z)(−z/6 + . . .)
= lim 2
z→0 (1 − z 2 /12 + . . .)2
= 2

(b) Find the coefficient of 1/z directly by examining the Laurent series. Expand
the denominator as a binomial series:
(1 + z) 2 −1 (1 + z)
f (z) = 2 2
(1 − z /12 + . . .) = 2 2
(1 + z 2 /12 + . . .)
z z
Multiplying out the terms:
2 2 3 2 2 1 1
f (z) = (1 + z + z /12 + z /12 + . . .) = + + + z + ···
z2 z 2 z 12 12
So near z = 0 the coefficient of 1/z is 2, and this is the residue.

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 65

5.3 The evaluation of real definite integrals

Many real definite integrals can be evaluated by using Cauchy’s residue theorem.

Functions of sin θ and cos θ

If we require an integral of the form:
Z 2π
I= R(cos θ, sin θ) dθ
0
where R is some function, then this integral can be changed to a complex contour integral,
that in turn, can be evaluated by Cauchy’s residue theorem.
The first step is the change of variable from θ to z. The equation is:
z = eiθ
So as θ varies from 0 to 2π, z circles around the point z = 0 at a constant distance |z| = 1.
Then we have
1 1
dz = eiθ i dθ that is dθ = iθ dz = dz
ie iz
So we can replace
Z 2π I
1
dθ = dz
0 iz
Next changing the sin θ and cos θ terms. we note:

1 1 1 1 1
! !
1 iθ −iθ
cos θ = 2 (e +e )= z+ ; sin θ = (eiθ − e−iθ ) = z− ;
2 z 2i 2i z

Example 1

Evaluate the integral

Z 2π
1
J= dθ where 0 < p < 1
0 1 − 2p cos θ + p2
Since the denominator can be written

2p 1
!
2
1 − 2p cos θ + p = 1 − z+ + p2
2 z
1h 1
1 − 2p cos θ + p2 = −p + (1 + p2)z − pz 2 = (1 − pz)(z − p)
i

z z

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 66

The integral becomes:

Z 2π 1 I z 1
J= dθ = dz
0 1 − 2p cos θ + p2 C (1 − pz)(z − p) iz

where C is the contour |z| = 1. The integrand

1
f (z) =
i(1 − pz)(z − p)
has two singularities, one at z = p and the other z = 1/p. Since 0 < p < 1, then
only z = p lies inside C. It is a pole of order one, and has the residue:
1 1
Residue = z→p
lim(z − p)f (z) = z→p
lim(z − p) =
i(1 − pz)(z − p) i(1 − p2)
Using Cauchy’s residue theorem:

1 2π
J = 2πi × =
i(1 − p2 ) 1 − p2
Therefore Z 2π 1 2π
dθ =
0 1 − 2p : cos θ + p2 1 − p2

Integrals of the form:

Z +∞
I= Q(x)
−∞

We consider the integrand Q(x) to be a function of a complex variable z: Q(z). We consider

a complex contour integral in the z-plane, anticlockwise around a very large semicircle with
centre at z = 0. The base of the semicircle (denoted by L) is the line along the real axis
from z = −R to z = +R. The arc, Γ, is defined by z = Reiθ where 0 ≤ θ ≤ π.

Example 2

Question
Show that: Z +∞
dx 3
I= 2 3
= π
−∞ (1 + x ) 8
Answer:

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 67

Consider the complex integral:

I
J= f (z)dz
C

where C is the semicircle discussed above, and f (z) is the complex function:
1
f (z) =
(1 + z 2 )3
Then we will shows that (a) I is closely related to J, and (b) J can be calculated
easily using Cauchy’s residue theorem.
Beginning with (a): we note that:
I Z Z
J= f (z)dz = f (z)dz + f (z)dz
C L Γ
Then, on L, we can see that z can be written as z = x and
Z Z +R Z +R
1
J1 ≡ f (z)dz = f (x)dx = dx
L −R −R (1 + x2 )3
Then consider: J2 ≡ Γ f (z)dz. We have that z = Reiθ . Then, using the Darboux
R

inequality:  
Z 
1 
J2 ≡ f (z)dz ≤ max  × πR
 
Γ (1 + z 2 )3 
 


Since R  1 and z = Reiθ , then

   

max
 1  
≤
  1 


(1 + z 2 )3   (1 − R2 )3 
   


Therefore:

πR
|J2 | ≤
(R2 − 1)3
Now let R → +∞, this gives:
|J2 | → 0
and
lim J = lim J1 + lim J2 = I + 0 = I
R→∞ R→∞ R→∞
Next, part (b), if we consider
1
f (z) =
(1 + z 2 )3

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 68

then this function has singularities at z = i and z = −i. The singularity at z = i is

a pole of order 3 inside C, while z = −i is a pole of order 3 outside C. According
to Cauchy’s residue theorem:
I
J= f (z)dz = 2πi × r1
where r1 is the residue of f (z) at z = i. To calculate the residue at z = i, we can
work out the Laurent series:
1
f (z) =
(z + i)3 (z − i)3
1
=
(z − i) (2i) [1 + (z − i)/2i]3
3 3

1 (z − i) (−3)(−4) (z − i)2
= [1 + (−3) + + · · ·]
−8i(z − i)3 2i 2! 4i2
Then the coefficient of (z − i)−1 is
(−3)(−4) 3
Residue = =
(−8i)2!4i2 16i
So I
3 3
J= f (z)dz = 2πi × = π
16i 8
Therefore: Z +∞ dx 3
= π
−∞ (1 + x2 )3 8

Example 3

Question: Evaluate the integral,

Z ∞ x2
dx
−∞ (x2 + a2 )(x2 + b2 )

Answer

We can use exactly the same technique. Consider the function of the complex
variable:
z2
f (z) = 2
(z + a2 )(z 2 + b2 )
So long as | Γ f (z) dz| → 0 then the integral is determined by the poles. Let’s
R

begin with showing that this is true. Consider

Z z2
J2 ≡ dz
Γ (z 2 + a2 )(z 2 + b2 )

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 69

over the semicircle radius R in the upper half plane. Again using Darboux:

z2
 
 
|J2 | ≤ max  × πR
 
(z 2 2 2 2
+ a )(z + b ) 
 


on Γ. Then
R2
 
 
|J2| ≤  × πR
 

 (R2 − a2 )(R2 − b2) 
 

and as R → ∞, it is clear that

π
|J2| ≤ →0
R
Having established that the arc Γ contributes nothing, turning to the residue theo-
rem, the only poles of f (z) are at z = ±ia, ±ib. Of these, only z = ia and z = ib
lie in the upper half plane, and both are poles of order 1. At z = ia the residue is

z2
lim (z − ia)
z→ia (z − ia)(z + ia)(z 2 + b2)
−a2 a
= =
2ai(b2 − a2 ) 2i(a2 − b2 )

From symmetry, the residue at z = ib is obviously b/[2i(b2 −a2 )]. Hence, the value
of the integral is

x2
 
Z ∞ a b π
dx = 2πi  +  =
−∞ (x2 + a2 )(x2 + b2 ) 2i(a2 − b2) 2i(b2 − a2 ) a+b

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 70

Theorem 5.3.1 Jordan’s lemma (1894)

If |Q(z)| → 0 uniformly when |z| → 0, in the range 0 ≤ arg z ≤ π, and if Q(z) is analytic
in the upper half of the z-plane except at a finite number of poles, then:
Z 
imz
lim e Q(z) dz = 0
R→∞ Γ

where m > 0, and Γ is the semicircular arc, centre z = 0, radius R in the upper-half
z-plane.

Proof: Given a very small positive number ε, then it is always possible to find R such
that:|Q(z)| < ε. This is the mathematical meaning of the phrase: ” |Q(z)| → 0 uniformly
when |z| → 0, in the range 0 ≤ arg z ≤ π”.
Z  Z π 
imz
e Q(z) dz  = eim(R cos θ+iR sin θ) Q(Reiθ ) Re iθ
idθ
   
 
Γ 0
 

Then using the Darboux inequality:

Z π  Z π
eim(R cos θ+iR sin θ) Q(Reiθ ) Re iθ
idθ < |eimR cos θ |e−mR sin θ ε|iReiθ dθ
 

0 0


Z π Z π/2
|eimR cos θ |e−mR sin θ ε| |iReiθ | dθ = (εR)2 e−mR sin θ dθ
0 0

Z π Z π/2
imR cos θ −mR sin θ iθ
|e |e ε| |iRe | dθ = (εR)2 e−mR sin θ dθ
0 0
Now since sin θ ≥ θ/( 21 π) in this range, then:
Z  Z π/2
imz
e Q(z) dz  < (εR)2 e−2mRθ/π dθ
 

Γ 0


Then working out the integral:

−π
!
Z  iπ/2
eimz Q(z) e−2mRθ/π
h
dz  < (εR)2
 


Γ 2mR 0

Z 
imz
e Q(z) dz  < επ/m
 

Γ


Hence
Z 
lim eimz Q(z) dz = 0
R→∞ Γ

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 71

Example 4

Question: Evaluate the integral

Z ∞
cos mx
dx
−∞ a2 + x2

Answer: To do this, we consider the related function

eimz
f (z) = 2
= eimz Q(z)
1+z
Consider the integral of f (z) around the semicircle C shown below.

Then according to Cauchy’s residue theorem: f (z) dz = 2πi×Residues inside C.

H

The only pole of this function in the upper half plane is at z = ia, and the residue
there is

eimz e−ma ie−ma

lim (z − ia) = =−
z→ia (z − ia)(z + ia) 2ia 2a
Hence,

I Z
cos mx + i sin mx
Z ∞ Z ∞
cos mx
f (z) dz = f (z)dz + dx = 0 + dx + 0
Γ −∞ a2 + x 2 −∞ a2 + x2

The sin term vanishes because the function is odd, the arc Γ term goes to zero
because of Jordan’s lemma. So finally
Z ∞ cos mx π −ma
dx = e
−∞ a2 + x2 a
Methods of Applied Mathematics – Section B: Complex variables 2003-4
The evaluation of real definite integrals Page 72

In fact, it is easy to show the integral over Γ goes to zero without invoking Jordan’s
lemma. A simple Darboux inequality argument would do just as well:
eimz  imz 
 
 e
Z
dz < max  × πR


Γ a2 + z 2  1 + z2 
 

eimz 1
Z !
dz < × πR
Γ a2 + z 2 1 − R2
and as R → ∞, then this integral goes to zero.

Jordan’s lemma is indispensable for the calculation of certain integrals. Consider the fol-
lowing example.

Example 5

Question:
Calculate the following integral:
Z +∞ sin x
I= dx
−∞ x
Answer:

This is a classic problem, with a well-known but not obvious solution. Consider
the integral of the function:
eiz
f (z) =
z
around the closed contour C shown below.

This contour is the familiar semicircle in the upper half plane, but with an inden-
tation to avoid the (a pole of order one) singularity at z = 0. Clearly with C there
are no infinities (singularities) of f (z) within C, and so Cauchy’s theorem applies:

Methods of Applied Mathematics – Section B: Complex variables 2003-4

The evaluation of real definite integrals Page 73

I
J= f (z) dz = 0
C
Now C is composed of 3 parts. A large arc Γ, centre z = 0, radius R a small arc
γ, centre z = 0, radius r; an integral along the real axis from −R to −r, and from
+r to +R. In the following we let R → ∞ and r → 0.
Consider the large semicircle Γ. Then using Jordan’s lemma, we can identify
Q(z) = 1/z and obviously Q(z) → 0 as |z| → ∞, so:
eiz 
 
Z
lim  dz = 0
R→∞ Γ z

So
eiz 
 
Z Z −r Z +∞  cos x + i sin x
J = 0 = lim  dz + lim + dx
r→0 γ z r→0 −∞ +r x
Focussing on the integral γ; on this path z = reiθ , therefore

eiz  eir cos θ−r sin θ iθ 

   
Z Z 0
lim  dz = lim  re idθ
r→0 γ z r→0 π reiθ
Then letting r → 0 inside the integral:

eiz 
  !
Z Z 0
lim  dz = lim idθ = −πi
r→0 γ z r→0 π

Then this results in:

Z −r Z +∞ 
cos x + i sin x
lim + dx = πi
r→0 −∞ +r x

sin x Z +∞ Z +∞
cos x
dx = π P dx = 0
−∞ x −∞ x
where P denotes the Cauchy principal value:
Z +∞ Z −r Z +∞ 
P = lim +
−∞ r→0 −∞ +r

The result
sin x Z +∞
dx = π
−∞ x
can also be obtained by another ‘device’ using a result derived previously.
Z ∞
cos mx π −ma
I(m, a) = dx = e
−∞ a2 + x2 a
Methods of Applied Mathematics – Section B: Complex variables 2003-4
The evaluation of real definite integrals Page 74

If we consider I(m, a) as a function of the parameters m and a, then we could

consider the derivative:

∂I Z ∞ x sin mx π a −ma
!
− = dx = e
∂m −∞ a2 + x2 a
Now letting a → 0:
∂I sin mx
" !# Z ∞
lim − = dx = π
a→0 ∂m −∞ x

Example 6
Question:
Calculate
3z 2 + 2 I
dz
C z(z + 1)

using Cauchy’s integral formula and/or Cauchy’s residue theorem, where

C is the circle |z| = 3.
Answer

The function has pole at z = 0 and z = −1. These are both of order one
and within the contour C. So using the residue theorem:
I
f (z)dz = 2πi × [Sum of Residues inside]
C
We have the residue at z = 0:
z(3z 2 + 2) 
 

Residue = lim [zf (z)] lim  =2

z→0 z→0 z(z + 1)
We have the residue at z = −1:
(z + 1)(3z 2 + 2) 
 

Residue = lim [(z + 1)f (z)] lim  = −5

z→−1 z→−1 z(z + 1)
Therefore:
I 3z 2 + 2
dz = 2πi × [(+2) + (−5)] = −6πi
C z(z + 1)

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Using residues to evaluate infinite series Page 75

5.4 Using residues to evaluate infinite series

It is possible to write certain series as contour integrals. For example we can show that:

+∞
X 1 π2
=
n=1 n2 6
The method relies on special properties of the function:

π
f (z) =
cot πz
z2
This function is analytic everywhere except at z = 0 and at z = ±1, ±2, ±3, . . ..
The singularities at z = ±1, ±2, ±3, . . . can be identified as poles of order 1, since, near
z = n, n = ±1, ±2, ±3:
cos πz cos πn + (z − n)(−π) sin πn + 12 (z − n)2(π 2 ) cos πn + · · ·
cot πz = ≈
sin πz sin πn + (z − n)(π) cos πn + 12 (z − n)2(π 2 ) sin πn + · · ·
(−1)n + 12 (z − n)2(π 2 )(−1)n + · · ·
=
(z − n)(π)(−1)n + · · ·
The residue at each of these points (z = n) is
1
n2
There is also a pole at z = 0 of order 3, at its residue can be found from the Laurent series,
near z = 0:
π  1 − 12 (πz)2 + · · · 
 
π 1  1 2

f (z) = 2 cot πz ≈ 2 ≈ 1 − (πz) + · · ·
z z πz − 16 (πz)3 + · · · z3 3

The residue,at z = 0 is
π2
−
3
Now if we integrate: I
I= f (z) dz
around a square with vertices: z1 = (N + 12 )(1 + i), z2 = (N + 21 )(−1 + i), and z3 =
(N + 21 )(−1 − i), z4 = (N + 21 )(1 − i),
Use the Darboux inequality (estimation theorem) to show the integral on the lines of the
square go to zero as the size of the square increases.

Methods of Applied Mathematics – Section B: Complex variables 2003-4

Using residues to evaluate infinite series Page 76

Figure 15:

Now as N → ∞ on the vertical sides of the square, we note that, between z1 and z4 ,
z = (N + 12 ) + iy, and therefore:
cos πz = cos π(N + 12 ) + iπy = cos π(N + 21 ) cosh πy + sin π(N + 12 )i sinh πy = i sinh πy
h i

Similarly:
sin πz = sin π(N + 12 ) + iπy = sin π(N + 21 ) cosh πy + cos π(N + 21 )i sinh πy = cosh πy
h i

| cos πz| | sinh πy|

| cot πz| = = = | tanh πy| < 1
| sin πz| | cosh πz|
The same is true for the other vertical side. So that as N → ∞ then the
π π
|f (z)| < 2 < 2
|z| N
And this goes to zero as N → ∞.
Similarly along the horizontal sides of the square: z = x + i(N + 12 ),

| cos πz| = | cos πx + (N + 12 )iπ | = | cosh π(N + 21 ) cos πx − i sinh π(N + 12 ) sin πx|
h i

It follows that,
r
| cos πz| = | cosh2 π(N + 21 ) cos2 πx + sinh2 π(N + 21 ) sin2 πx|

Methods of Applied Mathematics – Section B: Complex variables 2003-4