Contents

1 The Complex Plane 1

1.1 Basic Properties of Complex Numbers . . . . . . . . . . . . . . . . . . 2
1.2 Regions in C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Complex Functions 6
2.1 Limits and Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 The Riemann Sphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Holomorphic Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Power Series and Analytic functions . . . . . . . . . . . . . . . . . . . 18
2.5 A Hamper of Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3 Complex Integration 30
3.1 Contour Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2 The Estimation Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Cauchy’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.4 Cauchy’s Integral Formula . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.5 Applications of Cauchy’s Integral Formula . . . . . . . . . . . . . . . . 44

4 Residues and Integration 49

4.1 Laurent Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 Singularities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.3 Cauchy’s Residue Theorem . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4 Applications of Cauchy’s Residue Theorem . . . . . . . . . . . . . . . 61
Chapter 1

The Complex Plane

The set of complex numbers is denoted by C. An element z ∈ C is of the form

z = x + iy where x, y ∈ R, where i is an “imaginary” number that satisfies i2 = −1.
We call x and y the real part and the imaginary part of z, respectively, and we write

x = Re(z) and y = Im(z).

The real numbers are those complex numbers with zero imaginary parts, making
R ⊂ C. A complex number with zero real part is called purely imaginary. One can
identify the set of complex numbers C with the set of points on the real plane R2 via
x + iy 7→ (x, y). For example, 0 corresponds to the origin, 1 corresponds to (1, 0) and
i corresponds to (0, 1). The x and the y axis of R2 are called the real axis and the
imaginary axis, as they correspond to the real and purely imaginary numbers in C.
This simple observation helps us to visualise C and to capture many of its “geometric”
properties, but the “algebraic” properties of C and R2 , as R × R, are vastly different.
The similarities with R2 come in handy at points, but it is the crucial differences
between C and R2 that makes complex analysis significantly different to real analysis.
Recall that R2 , as the cartesian product of R with itself, is not a field. This is because
with the component-wise multiplication (x1 , y1 )·(x2 , y2 ) = (x1 x2 , y1 y2 ) some elements
have no inverse. Indeed, as the identity element is (1, 1), any pair of the form (a, 0)
or (0, b) with a, b ∈ R∗ has no multiplicative inverse, where R∗ = R\{0}.
We would like to endow C with a different multiplication than the one in R2 , so
that it becomes a field. Such multiplication comes naturally by simply treating all
numbers (in z = x + iy) as if they are real, and keeping in mind that i2 = −1. So, for
z1 = x1 + iy1 and z2 = x2 + iy2 we have

z1 + z2 = (x1 + x2 ) + i(y1 + y2 )
z1 z2 = (x1 + iy1 )(x2 + iy2 )
= x1 x2 + ix1 y2 + ix2 y1 + i2 y1 y2
= (x1 x2 − y1 y2 ) + i(x1 y2 + x2 y1 ).

The reader should check that these two operations make C a field.

1
1.1 Basic Properties of Complex Numbers

The notion of length, or absolute value, of a complex number is identical to the notion
of Euclidean length in R2 . More precisely, we define the modulus, or the absolute
value, of z = x + iy to be q
|z| = x2 + y 2 .
This analogy provides the triangle inequality from real analysis at no cost

|z + w| ≤ |z| + |w| for all z, w ∈ C

as z and w can be thought of as two vectors in R2 . There are other useful inequalities
that can be derived from the definition of the triangle inequality:

|Re(z)| ≤ |z| for all z ∈ C

|Im(z)| ≤ |z| for all z ∈ C
||z| − |w|| ≤ |z − w| for all z, w ∈ C

We will prove these in the exercises.

The complex conjugate of z = x + iy is defined by

z = x − iy

and it is visualised as the reflection across the real axis in the plane. Indeed, a complex
number z is real if and only if z = z, and it is purely imaginary if and only if z = −z.
The reader should have no difficulty checking that
z+z z−z
Re(z) = and Im(z) = .
2 2i
Also, for z, w ∈ C we have
|zw| = |z| · |w|
as well as
1 z
zz = |z|2 which implies = 2 whenever z , 0.
z |z|

Any non-zero complex number z can be written in polar form

z = reiθ ,

where r = |z| > 0 and θ ∈ R is the argument of z, the angle (with positive counter-
clockwise orientation) between the positive real axis and the half-line starting at the
origin and passing through z. Clearly, arg(z) = θ is not uniquely determined by z as
reiθ = rei(θ+2kπ) for any k ∈ Z. The principal value of arg(z) is defined as the unique
value of the argument that falls in (−π, π], and is denoted by Arg(z).
Given the definition of θ and r one can easily deduce the Euler formula

reiθ = r(cos θ + i sin θ).

2
Using the polar form is useful in many ways. For example, writing z = reiθ and
w = seiφ then
zw = rsei(θ+φ)
which suggests the multiplication in C is a rotation composed with scaling. It can
also be used to easily prove any of the following for z, w ∈ C:
z
|zw| = |z| · |w|, arg(zw) = arg(z) + arg(w), arg( ) = arg(z) − arg(w) with w , 0
w
One could also play around with the polar form to recover many of the known trigono-
metric identities. For example, if we set r = 1 and take the n-th power of z we obtain
z n = (eiθ )n = einθ
which immediately implies de Moivre’s theorem:
(cos θ + i sin θ)n = cos nθ + i sin nθ for all θ ∈ R.

The equation x2 +1 = 0 has no real solutions, which is a fact that makes R inadequate
for solving polynomial equations. On the other hand, the fundamental theorem of
algebra asserts that any polynomial of degree n ∈ N with coefficients in C has n roots
(counted with multiplicity). Some polynomials occur repeatedly in complex analysis.
One such polynomial is z n − 1. Solving it, i.e., finding all z ∈ C for which z n = 1, can
be done as follows. First write z = rei θ. Obviously r = 1 as we must have
|z n | = |1| =⇒ |z|n = 1 =⇒ rn = 1 =⇒ r = 1.
We can also write 1 = ei2kπ for k ∈ Z, and deduce that
z n = 1 =⇒ (eiθ )n = ei2kπ =⇒ einθ = ei2kπ
and the latter results in
z = ei2kπ/n for k = 0, . . . , n − 1.
These numbers are known as the n-th roots of unity. They lie at the vertices of
a regular n-gon centred at 0 ∈ C with one vertex at 1. There is a single real root 1
when n is odd and exactly two real roots ±1 when n is even. For example,
z4 = 1 if and only if z = ±1 or z = ±i
and the third roots of unity are 1, ω and ω 2 , where
√ √
−1 + i 3 −1 − i 3
ω=e 2πi/3
= , and consequently ω = e
2 4πi/3
= .
2 2
One can prove the geometric identity
(1 − z)(1 + z + · · · + z m ) = 1 − z m+1
by simply multiplying the left hand side and noticing that all but two terms cancel
out. Taking m = 2 we obtain
(1 − z)(1 + z + z 2 ) = 1 − z 3 .
Combining this with the result above on the 3-rd roots of unity, we see that the
equation 1 + z + z 2 = 0 has exactly two (non-real) roots ω and ω 2 .

3
1.2 Regions in C

For a fixed point z0 ∈ C and a positive real number r, the circle of radius r centred
at z0 is the set of all complex numbers z ∈ C with distance r to z:

Cr (z0 ) = {z ∈ C such that |z − z0 | = r}.

The open disc of radius r centred at z0 is defined as the interior of Cr (z0 ):

Dr (z0 ) = {z ∈ C such that |z − z0 | < r}.

The set Dr (z0 ) is also called an r-neighbourhood of z0 .

Similarly, a closed disc of radius r centred at z0 is the closure of Dr (z0 ) and is
defined as:
Dr (z0 ) = {z ∈ C such that |z − z0 | ≤ r}.

A point z0 is called an interior point of a set A ⊆ C if a neighbourhood of z0 can

be found that is completely contained within A. Similarly, a point z0 ∈ C is called a
boundary point of A if every neighbourhood of z0 contains at least one point in A
and at least one point not in A (note that we did not require z0 ∈ A). The set of all
boundary points of A is called the boundary of A, and is denoted by ∂A. Similarly,
the set of all interior points of A is called the interior of A, denoted by A◦ .
Clearly, the boundary of both Dr (z0 ) and Dr (z0 ) is the circle Cr (z0 ).

Dr (z0 )
|z − z0 |
z0 z

Cr (z0 ) : the dashed circle

A set A ⊂ C is called open if it contains none of its boundary points; in other words
all its points are interior. A set A is called closed if its complement Ac = C − A
is open; or equivalently it contains all its boundary points. The closure of A is the
smallest closed set that contains A; or equivalently it is A = A ∪ ∂A. One can check
that:
∂A = A − A◦
Note that both C and Ø are open and closed!
In a diagram, we use a solid line to indicate a portion of the edge of the set that
includes boundary points, and a dashed line to indicate a portion that does not
include boundary points. The end of a curve is denoted by a closed circle if it is
included in the set, and by an open circle ◦ if it is not.

4
A set A ⊆ C is called connected if any two points z, w ∈ A can be joined by a
path entirely contained in A (this notion is known as path-connected in topology).
Alternatively, an open set A is connected if A = A1 ∪ A2 with both A1 and A2 open
and non-empty (and disconnected) implies A1 = A or A2 = A. One could prove that
these two notions are equivalent for open sets in C. We are not concerned about that
in this module, and refer to more topological text-books if you are interested.
A set A is called a domain (or a region) if it is open and connected.
Examples

(a) |z − 2| < 1
This is the set of complex numbers with distance less than 1 from the complex
number 2:
{z ∈ C | |z − 2| < 1}
This is a domain as it is both connected and and open:

0 2

(b) {z | Rez < 0} ∪ {z | |z − 2| ≤ 1}

The set Rez < 0 is open and connected, so it is a domain. The set {z | |z −2| ≤ 1}
on the other hand is closed (and connected), hence not a domain. The union of
the two sets is neither open nor connected, hence not a domain.

Rez < 0
|z − 2| ≤ 1
open
closed
0 2

5
Chapter 2

Complex Functions

Let D ⊆ C be nonempty. A function f : D → C is a rule that assigns to each point

z ∈ D an image f (z) ∈ C, also known as the value of f at z. Following conventional
calculus, the set D is called a “domain”. However, it should not be confused with
the definition of (topological) domain given in the previous chapter; in particular, it
is not required that the domain of a function is open or connected. If f is defined
without specifying D, we take the domain as the largest subset of C in which f is
“meaningful”. For example, if not specified, the domain of the function f (z) = z1 is
the set C∗ = C \{0}.
Write z = x + iy. Then f being a function is equivalent to having two real-valued
functions u(x, y) and v(x, y), i.e.,
u : R2 → R and v : R2 → R
such that
f (z) = u(x, y) + iv(x, y).
We use the notation Ref (z) = u(x, y) and Imf (z) = v(x, y).
For example, for f (z) = z 2 we can write
f (x + iy) = (x + iy)2 = x2 − y 2 + 2ixy
which shows that u(x, y) = x2 − y 2 and v(x, y) = 2xy.
The reader can check that
f (z) = z =⇒ u(x, y) = x and v(x, y) = −y,
f (z) = |z|
2
=⇒ u(x, y) = x2 + y 2 and v(x, y) = 0.
The function f (z) = |z|2 is called a real-valued function of complex variable z.
Example 2.0.1. Find the real and imaginary parts of the function f (z) = z1 , z , 0.
We do this in two different ways. First, we use the coordinate definition of a complex
number:
1 x − iy x − iy
f (x + iy) = = = 2 ,
x + iy (x + iy)(x − iy) x + y2
which implies
x −y
u(x, y) = 2 and v(x, y) = 2 .
x +y 2 x + y2

6
Alternatively, we use the polar coordinates:

1 e−iθ cos θ − i sin θ

f (reiθ ) = iθ
= = ,
re r r
which implies
cos θ − sin θ
Ref (z) = and Imf (z) = .
r r

2.1 Limits and Continuity

Definition 2.1.1. Let D be a domain and z0 ∈ D. For a function f : D → C, we

say that lim f (z) = w0 (or, equivalently, f (z) tends to w0 as z tends to z0 ) if for all
z→z0
ε > 0 there exists δ > 0 such that

z∈D and 0 < |z − z0 | < δ =⇒ |f (z) − w0 | < ε.

Note that, as usual, we used 0 < |z − z0 | instead of 0 ≤ |z − z0 |, that is z , z0 .

Indeed, the limit (if it exists) is determined by the behaviour of f (z) as z approaches
z0 , and the value of f at z0 is irrelevant. In fact, f may not even be defined at z0 .
For example, if f (z) = 1 if z , 0 and f (0) = 0, then lim f (z) = 1, and in particular
z→0
lim f (z) , f (0).
z→0
Also, this definition means that f (z) tends to w0 as z tends to z0 along any path
(not just the straight lines, and definitely not just the coordinate directions)!
If the real and the imaginary parts of f are known, then the usual real analysis
methods can be handy. This is basically due to the following fact.

Theorem 2.1.2. Given f (z) = u(x, y) + iv(x, y), z0 = x0 + iy0 and w0 = u0 + iv0 ,
then lim f (z) = w0 if and only if
z→z0

lim u(x, y) = u0 and lim v(x, y) = v0 .

(x,y)→(x0 ,y0 ) (x,y)→(x0 ,y0 )

Proof. Left as exercise (Hint: use exercise 7 in the Exercise sheet 1). □

In particular, the algebra of complex limits (sums, products, etc.) and other elemen-
tary properties can be developed exactly as in the real analysis. We will use these
freely without proving them.
Examples
1. lim(3z 2 + 2z − 1) = 3i2 + 2i − 1 = −3 + 2i − 1 = −4 + 2i.
z→i
z
2. lim = 0
= 0.
z→0 z2 +1 1

7
 z
3. lim : we are going to argue that this limit does not exist! Recall that the
z→0 |z|
limit would exists (as a point w0 ) if f (z) tends to w0 as z tends to z0 from
“every direction”. We choose two different directions through which f (z) tends
to two different values! For this function, we choose these two directions to be
both on the real axis, tending to 0 from the right and from the left:
on the positive real axis z = x with x > 0, hence we have
z x
lim = lim =1
x→0+ |z| x→0+ x
and along the negative real axis z = x with x < 0, which gives
z x
lim = lim = −1.
x→0− |z| x→0− −x
The limit taken along the two paths are different, and hence the limit as z → 0
does not exist.
Alternatively, put z = reiθ and observe that for a fixed θ we have
z reiθ
lim = lim = lim eiθ = eiθ .
z→0 |z| r→0 r r→0

The result clearly depends on the angle of approach θ and hence the limit as z
tends to 0 does not exist.

As you can guess, some rules in real limits that depend on the order in R do not (easily)
transfer to complex limits (recall that C is not an ordered field). For example, recall
the squeeze (or the sandwich) theorem over R: let f, g, h be real-valued real functions
defined on an interval I such that g(x) ≤ f (x) ≤ h(x) for all x ∈ I, and suppose a is
a limit point of I, such that lim g(x) = lim h(x) = ℓ. Then lim f (x) = ℓ.
x→a x→a x→a
This theorem, as stated above, depends on the order in R, hence it does not directly
apply to complex functions. However, it has a complex version as follows.
Theorem 2.1.3 (The Squeeze Theorem). Suppose f, g : D → C are two functions
such that |g(z)| ≤ |f (z)| for all z ∈ Dr (z0 )\{z0 } ⊆ D, for some r ∈ R+ . If
lim f (z) = 0, then lim g(z) = 0.
z→z0 z→z0

Proof. Let ε > 0. By assumption there exists δ > 0 such that if z ∈ D with 0 <
|z − z0 | < δ then |f (z)| < ε. Hence, for any z ∈ D with 0 < |z − z0 | < δ we have
|g(z)| ≤ |f (z)| < ε. □

A complex function f is called bounded on a set D if there exists a real number

M > 0 such that |f (z)| ≤ M for all z ∈ D.
Corollary 2.1.4. Suppose that lim f (z) = 0 and g is bounded in Dr (z0 )\{z0 }. Then
z→z0
lim f (z)g(z) = 0.
z→z0

Proof. This follows from the fact that |f (z)g(z)| ≤ M |f (z)|, together with the squeeze
theorem. □

8
Definition 2.1.5. Let D be a domain and f : D → C a function. We say that f is
continuous at z0 ∈ D if lim f (z) = f (z0 ).
z→z0
We say that f is continuous on D if it is continuous at every point of D.

Continuity over C is very similar to continuity in real analysis. In particular, if

f, g : D → C are complex functions continuous at z0 ∈ D, then
f (z) + g(z), f (z)g(z), cf (z)(c ∈ C)
f (z)
are all continuous at z0 . Similarly, is continuous provided that g(z0 ) , 0.
g(z)
These all follow from the fact below, which can be easily proven using the definition
and Theorem 2.1.2.
Theorem 2.1.6. The function f (z) = u(x, y) + iv(x, y) is continuous at z0 = x0 + iy0
if and only if u and v are continuous at (x0 , y0 ).

It follows that all polynomials, as expected, are continuous functions over C. Similarly,
f (z) = |z|2 is continuous for all z ∈ C.

2.2 The Riemann Sphere

In real analysis, we defined +∞ and −∞ to capture limiting behaviour of functions
1
at infinity. This also allows the study of limits of function like as x → 0. Taking
x
the same approach over C is necessary, but at the same time not exactly the same as
in R. Notably, there are infinitely many directions to take for ∞ over C if we used the
same intuition as in R. The idea of infinity in C is provided by an ingenious device,
due to Riemann. Not only it will allow us to do limits with simple arithmetic, but it
also provides a uniform treatment for lines and circles in C.
Let us embed C in the Euclidean space R3 by identifying z = x + iy with (x, y, 0).
Now, define the Riemann sphere as
1 1
S := {(x, y, u) ∈ R3 |x2 + y 2 + (u − )2 = }.
2 4
It touches the plane C at the origin (0, 0, 0):

u p

z′
y
z o
C

9
The stereographic projection allows to set up a one-to-one correspondence between
the points of C and the points of S, excluding p the north pole of S. Geometrically,
the line from any point z ∈ C to p cuts S\{p} in precisely one point z ′ , and, for every
point z ′ ∈ S\{p}, the line through p and z ′ meets the plane C in a unique point z (as
in the picture). The irritation of the north pole being left out can be removed: we
add to C an extra point ∞ < C and define the extended complex plane C e to be
C ∪ {∞}. we now have a perfect one-to-one correspondence between S and C, e which
is given by

x y r2
C ∋ z = x + iy = reiθ ←→ z ′ = ( , , ) ∈ S,
1 + r2 1 + r2 1 + r2
∞ ←→ (0, 0, 1).

Having a single point at infinity is very natural in C (unlike in R). For instance, if we
let z = reiθ with θ fixed and allow r to become arbitrarily large, then z will approach
p regardless of the value of θ: “all roads lead to ∞”!

2.2.1 Arithmetic in the extended plane

Regarding the Riemann sphere as the extension of C, we extend the arithmetic on C

to C,
e with some provision, using the following conventions:

z
z±∞=∞±z =∞ , = 0 ∀z ∈ C,
∞
z
z·∞=∞·z =∞ , = ∞ ∀z ∈ C\{0},
0
∞+∞=∞·∞=∞=∞

We allow division of a nonzero complex number by zero (in the extended complex
∞
plane). However, as expected we have not assigned any meanings to ∞ − ∞ and .
∞
1
In the extended complex plane, an r-neighbourhood of ∞ is the set {z ∈ C| < r}.
|z|
Definition 2.2.1. Let f be a function defined at all points z in some neighbourhood
of ∞. We say that lim f (z) = w (for some w ∈ C) if for any ε > 0 there exists δ > 0
z→∞
1
such that < δ implies |f (z) − w| < ε.
|z|
Similarly, let f be a function defined and nonzero at all points in a deleted neigh-
bourhood of z0 . We say lim f (z) = ∞ if for all ε > 0 there exists δ > 0 such that
z→z0
1
0 < |z − z0 | < δ implies < ε.
|f (z)|
And finally, let f be a function defined and nonzero at all points of some neighbourhood
1
of ∞. We say lim f (z) = ∞ if for all ε > 0 there exists δ > 0 such that <δ
z→∞ |z|
1
implies < ε.
|f (z)|

Given all these, one can deduce the following useful theorem.

10
Theorem 2.2.2. If z0 , w0 ∈ C then
1
lim f (z) = ∞ if and only if lim = 0,
z→z0 z→z0 f (z)

lim f (z) = w0 if and only if lim f (1/z) = w0 ,

z→∞ z→0
1
lim f (z) = ∞ if and only if lim = 0,
z→∞ z→0 f (1/z)

Examples.

iz + 3
(a) lim
z→−1 z + 1
iz + 3
Let f (z) = and consider
z+1
1 z+1 0
lim = lim = = 0.
z→−1 f (z) z→−1 iz + 3 3−i
iz + 3
Hence we conclude that lim = ∞.
z→−1 z + 1

2z + i
(b) lim
z→∞ z + 3
2z + i
Let f (z) = and consider
z+3

1 2
+i 2
+i z 2 + iz 2
lim f ( ) = lim z
= lim z
· = lim = = 2.
z→0 z z→0 1
z + 3 z→0 1
z +3 z z→0 1 + 3z 1

2z + i
Hence we conclude that lim = 2.
z→∞ z + 3

2z 3 − 1
(c) lim
z→∞ z 2 + 1

2z 3 − 1
Let f (z) = 2 and consider
z +1

1 ( 1 )2 + 1 ( 1 )2 + 1 z 3 z + z3 0
lim = lim z1 3 = lim z1 3 · 3 = lim = = 0.
z→0 f (1/z) z→0 2( ) − 1 z→0 2( ) − 1 z z→0 2 − z 3 2
z z

2z 3 − 1
Hence we conclude that lim = ∞.
z→∞ z 2 + 1

11
2.3 Holomorphic Functions

Recall from real analysis that a function f : (a, b) → R is differentiable at a point

x0 ∈ (a, b) if
f (x) − f (x0 )
f ′ (x0 ) = lim
x→x0 x − x0
exists. In that case, f ′ (x0 ) is called the derivative of f at x0 . We then say that f
is differentiable (on (a, b)) if it is differentiable at all points x0 ∈ (a, b). Intuitively,
f ′ (x0 ) is the slope of the tangent line to the graph of the function f at the point
(x0 , f (x0 )). A key part of that definition is to take the limit both from the right and
from the left as we approach x0 . For example, you may recall that the function |x| is
not differentiable at 0 precisely because

|x| − 0 |x| − 0
lim = −1 and lim = 1.
x→0− x−0 x→0+ x−0

This fact perfectly captures the geometric intuition that there is no well-defined tan-
gent to |x| at the origin as it has an edge. This example also makes it clear why we
consider open sets in the definition of derivatives.

2.3.1 Complex Differentiation

The generalisation of differentiability to complex functions is as you expect:

Definition 2.3.1. Let D ⊆ C be a domain and f : D → C a function. We say that f

is complex differentiable at z0 ∈ D if

f (z) − f (z0 )
f ′ (z0 ) = lim
z→z0 z − z0

exists. We call f ′ (z0 ) the derivative of f at z0 , and we say f is holomorphic in

D if it is complex-differentiable at every point of D. A function that is defined and
holomorphic on C is called entire. A function f is said to be holomorphic at a
point z0 ∈ C if it is defined and holomorphic on Dr (z0 ) for some r > 0.

Holomorphy for a complex function is a much stronger condition that differentiability

for a real function. For example, a holomorphic function is infinitely many times com-
plex differentiable! In other words, the existence of the first derivative guarantees the
existence of derivatives of any order. Recall that there are differentiable real functions
with no second derivative. If fact, one could make an even stronger statement: every
holomorphic function is analytic, that is, it has a power series expansion near every
point (we will discuss power series later). For this reason, the synonym analytic is
sometimes used for holomorphic functions.
In comparison with real analysis, we emphasise that complex derivative cannot be
used to calculate local minima or maxima. Indeed, the notion of local minimum or
maximum does not make sense for complex functions, as C is not ordered. Similarly,
the notion of slope of the tangent line is best forgotten for complex functions!

12
Example. Let f (z) = z 2 , defined over C. For any point z0 ∈ C we have

f (z) − f (z0 ) z 2 − z02 (z − z0 )(z + z0 )

lim = lim = lim = lim (z + z0 ) = 2z0 .
z→z0 z − z0 z→z 0 z − z0 z→z 0 z − z0 z→z0

Hence, f ′ (z0 ) = 2z0 for all z0 ∈ C, and f is holomorphic at every point (z 2 is entire).
The standard rules of differentiable functions continue to hold over C:

Proposition 2.3.2. Let f and g be holomorphic on D, and c ∈ C. Then we have:

(a) (f + g)′ = f ′ + g ′ ,

(b) (cf )′ = cf ′ ,

(c) (f g)′ = f ′ g + f g ′ ,
f f ′g − f g′
(d) ( )′ = ,
g g2
(e) (f ◦ g)′ = (f ′ ◦ g) · g ′ .

We will not prove these. The proof usually involves a modification of the one for
real-valued functions (replacing the absolute value of real numbers in the definition
of the limit with complex modulus).

Proposition 2.3.3. Holomorphy implies continuity.

Proof. Let f be a function that is holomorphic at z0 ∈ C. We need to show that

lim f (z) = f (z0 ). This is done as follows:
z→z0

f (z) − f (z0 )
lim f (z) − f (z0 ) = lim (z − z0 )
z→z0 z→z0 z − z0
f (z) − f (z0 )
= lim × lim (z − z0 ) = f ′ (z0 ) × 0 = 0
z→z0 z − z0 z→z0

Example 2.3.4. The function f (z) = |z|2 is continuous for all z ∈ C, which fol-
lows, for example, from continuity of the real function x2 + y 2 . Let us examine its
differentiability on C, by examining the limit

f (z) − f (z0 )
lim .
z→z0 z − z0
Put z = x + iy and z0 = x0 + iy0 , and consider the limit above as z approaches z0
along the line y = y0 . On this line, z → z0 as x → x0 , hence

f (z) − f (z0 ) |z|2 − |z0 |2 x2 + y02 − (x20 + y02 )

lim = lim = lim
x→x0 z − z0 x→x0 z − z0 x→x0 x − x0
2
x − x0 2
= lim = lim x + x0 = 2x0 .
x→x0 x − x0 x→x0

13
Similarly, on the x = x0 line, we have

f (z) − f (z0 ) |z|2 − |z0 |2 x2 + y 2 − (x20 + y02 )

lim = lim = lim 0
y→y0 z − z0 y→y0 z − z0 y→y0 i(y − y0 )
2
y − y0 2
= lim = lim −i(y + y0 ) = −2iy0 .
y→y0 i(y − y0 ) y→y0

The two limits disagree, unless x0 = y0 = 0. Hence, we conclude that f is not

differentiable for any point on C − {0}. However, this is not enough to conclude
differentiability at z = 0. For this, note that

f (z) − f (0) |z|2 zz

lim = lim = lim = lim z = 0.
z→0 z z→0 z z→0 z z→0

Hence, the function |z|2 is everywhere continuous on C, but it is only differentiable

at 0 and it is nowhere holomorphic. Another example that goes even further than
this is f (z) = z, which is continuous everywhere but nowhere differentiable. We will
check this later!

2.3.2 The Cauchy-Riemann equations

So far, we have almost freely related complex notions to their real counterparts.
As you have seen, complex differentiability distinguishes itself from the real case.
However, they can still be related via a mysterious relation! Before we state this,
recall that for a real-valued function h : R2 → R the partial derivatives ∂h/∂x and
∂h/∂y are defined as follows:

∂h h(x + t, y) − h(x, y) ∂h h(x, y + s) − h(x, y)

(x, y) = lim , (x, y) = lim
∂x t→0 t ∂y s→0 s

Theorem 2.3.5 (the Cauchy-Riemann theorem). Let f : D → C, D ⊆ C a domain,

and write f (x + iy) = u(x, y) + iv(x, y). If f is differentiable at z0 = x0 + iy0 then the
partial derivatives ∂u/∂x, ∂u/∂y, ∂v/∂x, ∂v/∂y exist at (x0 , y0 ) and the Cauchy-
Riemann equations hold:
∂u ∂v ∂u ∂v
(x0 , y0 ) = (x0 , y0 ) and (x0 , y0 ) = − (x0 , y0 )
∂x ∂y ∂y ∂x

Proof. As f is differentiable at z0 , we know that

f (z) − f (z0 )
lim = f ′ (z0 )
z→z0 z − z0
from any direction that z approaches z0 . We concentrate on two specific directions:
the horizontal and the vertical approach to z0 . Let t ∈ R and consider z0 + t =
(x0 + t) + iy0 . Then as t → 0 we have z0 + t → z0 .

14
Hence
f (z0 + t) − f (z0 )
f ′ (z0 ) = lim
t→0 t
u(x0 + t, y0 ) + iv(x0 + t, y0 ) − u(x0 , y0 ) − iv(x0 , y0 )
= lim
t→0 t
u(x0 + t, y0 ) − u(x0 , y0 ) v(x0 + t, y0 ) − v(x0 , y0 )
= lim + i lim
t→0 t t→0 t
∂u ∂v
= (x0 , y0 ) + i (x0 , y0 ). (2.1)
∂x ∂x
Now, let s ∈ R and consider z0 + is = x0 + i(y0 + s). Then as s → 0 we have
z0 + is → z0 . Hence
f (z0 + is) − f (z0 )
f ′ (z0 ) = lim
s→0 is
u(x0 , y0 + s) + iv(x0 , y0 + s) − u(x0 , y0 ) − iv(x0 , y0 )
= lim
s→0 is
u(x0 , y0 + s) − u(x0 , y0 ) v(x0 , y0 + s) − v(x0 , y0 )
= lim + i lim
s→0 is s→0 is
∂u ∂v
= −i (x0 , y0 ) + (x0 , y0 ). (2.2)
∂y ∂y

Comparing the real and imaginary parts of (2.1) and (2.2) gives the Cauchy-Riemann
equations. □

Let us examine the differentiability of the function f (z) = z using the Cauchy-
Riemann equations. Note that f (x + iy) = x − iy, which reads u(x, y) = x and
v(x, y) = −y. Calculating the partial derivative gives:

∂u ∂u ∂v ∂v
= 1, = 0, = 0, = −1
∂x ∂y ∂x ∂y
This proves that the function is nowhere differentiable as the Cauchy-Riemann equa-
tions are never satisfied.
The function z is a very easy to write example of a function that is everywhere continu-
ous but nowhere differentiable. Such functions exist in real analysis, but they are much
harder to write down (e.g., the Weierstrass function w(x) = ∞ k=0 2
−kα cos(2πbk x)
P

for α ∈ (0, 1) and b ≥ 2).

As the example above shows, Cauchy-Riemann equations can be used to prove non-
differentiability of a function. In converse, on their own they are not enough to
guarantee differentiability. For example, let
(
1 if neither x nor y is zero,
f (z) = f (x + iy) =
0 otherwise.

At z = 0 we have ∂u/∂x = ∂u/∂y = ∂v/∂x = ∂v/∂y = 0, so the Cauchy-Riemann

f (z) − f (0)
equations hold. However, lim does not exist, as this limit via the direction
z→0 z

15
of the x-axis or the y-axis is 0 but through other straight line directions it is ∞. Note
that f is not even continuous at 0.
The problem with the example above is that in the definition of differentiability we
let z tend to z0 in an arbitrary way but in calculating the partial derivatives we only
know the limiting behaviour in horizontal or vertical approach. This issue is resolved
with adding some extra conditions as in the proposition below.

Proposition 2.3.6. Let f (x + iy) = u(x, y) + iv(x, y) be a continuous function in a

neighbourhood of z0 = x0 +iy0 . Suppose that ∂u/∂x, ∂u/∂y, ∂v/∂x, ∂v/∂y exists and
are continuous at (x0 , y0 ), and further suppose that the Cauchy-Riemann equations
hold at z0 . Then f is differentiable at z0 .

Let us now look at some applications of the Cauchy-Riemann theorem. It is about

certain conditions that force a holomorphic function to be constant. The first two con-
ditions are not surprising but the third case is the first example of strong implications
of holomorphy.

Theorem 2.3.7. Suppose that f : D → C is holomorphic on the domain D. Then

any of the following conditions forces f to be constant in D.

(i) f ′ (z) = 0 for all z ∈ D;

(ii) |f | is constant in D;

(iii) f (z) is real for all z ∈ D.

Proof. For simplicity, assume D = D1 (0). The full proof follows from the fact that
any two points on D can be joined by a path (consisting of horizontal and vertical
line segments).
The proof of the Cauchy-Riemann theorem shows that for z = x + iy ∈ D we have
∂u ∂v ∂v ∂u
f ′ (z) = +i = −i .
∂x ∂x ∂y ∂y
∂u ∂v ∂u ∂v
Suppose f ′ is identically zero. Then = = = = 0 throughout D. Fix
∂x ∂x ∂y ∂y
two points p = a + ib and q = c + id in D. We will prove that f (p) = f (q). At least
one of s = c + ib and t = a + id lies in D (draw the picture of the disc D to see this).
Without loss of generality suppose that s ∈ D. Each of the real-valued functions
x 7→ u(x, b) and y 7→ u(c, y) has zero derivative, and so is constant (for example by
the Mean Value Theorem). Hence

u(a, b) = u(c, b) and u(c, b) = u(c, d)

and similarly,
v(a, b) = v(c, b) and v(c, b) = v(c, d),
which shows that f (p) = f (s) = f (q). This proves (i).

16
Now consider (ii): suppose that |f (z)| = c for all |z| < 1, where c is a constant. We
then have u2 + v 2 = c2 , which implies
∂u ∂v ∂u ∂v
u +v =0 and u +v = 0.
∂x ∂x ∂y ∂y
Hence, by the Cauchy-Riemann equations,
∂u ∂u ∂u ∂u
u −v =0 and u +v = 0.
∂x ∂y ∂y ∂x
∂u
elimination of gives
∂y
∂u ∂u
0 = (u2 + v 2 ) = c2 .
∂x ∂x
∂u
If c = 0 then trivially f is zero. Otherwise, = 0 throughout D. Similarly, we can
∂x
∂v ∂u ∂v
conclude that = = = 0. This proves the claim, as in the previous case.
∂x ∂y ∂y
∂v ∂v
For (iii) note that f is real-valued, so v = 0, which implies = = 0. By the
∂x ∂y
∂u ∂u
Cauchy-Riemann equations, = = 0 as well. Hence, f must be constant. □
∂x ∂y

Note that D being a domain is crucial in the theorem above. We have implicitly used
its openness by considering that it is covered by discs. Connectedness is also crucial.
For instance, consider the function f : D → C, where D = D1 (−2) ∪ D1 (2), defined
by (
1 if z ∈ D1 (−2),
f (z) =
−1 if z ∈ D1 (2).
We see that all conditions of the theorem are satisfied but f is not constant.

17
2.4 Power Series and Analytic functions

Recall from real analysis that the Maclaurin expansion of a function provides

f ′′ (0) 2 f (n) (0) n

f (x) = f (0) + f ′ (0)x + x + ··· + x + ···
2 n!
Some key examples of this expansion for real x are
x xn
ex = 1 + x + + ··· + ···
2 n!
x3 x2n+1
sin x = x − + · · · + (−1)n + ···
3! (2n + 1)!
x2 x2n
cosh x = 1 + + ··· + + ···
2! (2n)!

We would like to have complex analogues of these, especially as geometric intuition

of trigonometric functions are no longer available.
When using expansion of the form

f (n) (a)
f (z) = f (a) + f ′ (a)(z − a) + · · · + (z − a)n + · · ·
n!
for a complex function f with a ∈ C, there are two issues to address: the existence
of the derivatives f (n) and the convergence. It turns out that the situation is much
better in complex analysis: for a function f that is holomorphic on a domain D, all
f (n) (z) exist for n ∈ N and for z ∈ D. So, we first investigate complex series and
their convergence in general, having in mind the example of powers of (z − a).

2.4.1 Convergence of complex series

As usual, we will regard a complex series as limits of sequences: let zk ∈ C. We

say that the series ∞ k=0 zk converges if the sequence of partial sums sn =
P Pn
k=0 zk
converges. The limit of this sequence of partial sums is called the sum of the series.
A series that does not converge is called divergent. One can check, following the
definition, that ∞ only if both ∞
P∞
k=0 zk converges if and k=0 Re(zk ) and k=0 Im(zk )
P P
P∞
converge (as real series). We say that k=0 zk is absolutely convergent if the real
series ∞ k=0 |zk | is convergent. Given this definition and what was said above, one
P

can deduce several useful facts from real analysis:

1. Suppose that the complex series ∞ n=0 zn converges, then n→∞

lim zn = 0 and there
P

exists a real constant M such that |zn | ≤ M for all n ∈ N.

2. Suppose that ∞ zn and ∞ n=0 wn are convergent complex series. Then the
P P
Pn=0
∞
complex series n=0 (zn + cwn ) converges for any c ∈ C and we have
∞ ∞ ∞
(zn + cwn ) = zn + c
X X X
wn .
n=0 n=0 n=0

18
 3. If ∞n=0 zn is absolutely convergent then it is convergent. The converse is not
P

necessarily true.
4. (comparison test) suppose ∞ n=0 an is a convergent real series with an ≥ 0
P

for all n and suppose that |zn | ≤ can for all n and for some real number c > 0.
Then ∞ n=0 zn converges absolutely, and hence converges.
P

5. (ratio test) assume that

|zn+1 |
ℓ = lim .
n→∞ |zn |
If ℓ < 1 the ∞ ∞ P∞
n=0 |zn | converges, and so does n=0 zn . If ℓ > 1 then n=0 |zn |
P P

diverges. If ℓ = 1 then the test gives no information.

6. (Cauchy’s n-th root test) Assume that ℓ = lim |zn |. If ℓ < 1 the
n
p
P∞ P∞ n→∞ P
∞
n=0 |zn | converges, and so does n=0 zn . If ℓ > 1 then n=0 |zn | diverges. If
ℓ = 1 then the test gives no information.

Example. Consider the series

∞ n
X i
.
n=0
3n
in
Write zn = and use the ratio test to show that the series is convergent:
3n
zn+1 in+1 3n i 1
| | = | n+1 n | = | | = .
zn 3 i 3 3
zn+1 1
Hence lim | | = < 1. The claim follows from the ratio test. Similarly, the root
n→∞ zn 3
test would have given
1 in 1
q q r
n n n
lim |zn | = < 1 as |zn | = | |= .
n→∞ 3 3n 3

2.4.2 Geometric series

The series ∞ k=0 z is called the geometric series. We have already seen the geo-
k
P

metric identity
(1 − z)(1 + z + · · · + z n ) = 1 − z n+1 .
this immediately implies
1 − z n+1
1 + z + z2 + · · · + zn = (z , 1).
1−z
We know that the sequence (z n+1 ) converges to 0 if |z| < 1. Hence ∞ k=0 z converges
k P
1
to whenever |z| < 1. Otherwise, if |z| ≥ 1 the series diverges as lim z n , 0.
1−z n→∞

In conclusion, the geometric series converges if and only if |z| < 1, and
∞
1
zn = (|z| < 1). (2.3)
X

n=0
1−z

19
The formula in (2.3) can be viewed in two ways: either as an infinite series or as
expanding the function (1 − z)−1 when |z| < 1. Taking the second viewpoint, we may
derive many related expansions. Some examples:
∞
1
= 1 − z + z2 − · · · = (−1)n z n (|z| < 1)
X
•
1+z n=0
∞
1
= 1 + + + = (|z| < 1)
X
2 4
• z z · · · z 2n
1 − z2 n=0
∞
1 1 zn
=− =− (|z| < 4)
X
•
z−4 4(1 − (z/4)) n=0
4n+1
More generally, for a, b , 0 we can write:
∞
1 1 an b
= = (−1)n n+1 z n (|z| < | |)
X
az + b b(1 + (a/b)z) n=0 b a

We could go further than this and compute more complicated expansion. For example,
let a , b, then using partial fractions we have
1 1 1 1
 
= −
(z − a)(z − b) (a − b) z−a z−b
∞ 
1 1 1
 !
= (|z| < min{|a|, |b|}).
X
n
− z
(a − b) n=0
bn+1 an+1
And, another example:
∞
1 1−z
= = (z 3n − z 3n+1 ) (|z| < 1)
X
1 + z + z2 1 − z 3 n=0

2.4.3 Power series

We have seen expansions of various rational functions (f /g where f and g are poly-
nomials) in powers of z − a. We now turn things around and systematically study
such series, called power series.
an (z − z0 )n , where z0 ∈ C and an ∈ C
P
Definition 2.4.1. A series of the form
(n ∈ N) is called a power series at z0 .

By a simple change of variables, and replacing z − z0 by z, we need only consider

power series at 0, that is power series of the form
∞
X
an z n .
n=0

We can attach a unique number to this series by defining

∞
( )
R = sup |z| |an z | is convergent .
X
n

n=0

If the series is convergent for z ∈ C with arbitrary large modulus then we write
R = ∞.

20
 P∞ n
Theorem 2.4.2. Let n=0 an z be a power series and R be defined as above. Then
P∞ n
(i) n=0 an z converges absolutely for |z| < R;
P∞ n
(ii) n=0 an z diverges for |z| > R.

Proof. Let z ∈ C with |z| < R. Choose w ∈ C such that |z| < |w| < R and such
that an wn converges (check that such w exists). As an wn converges, it follows
P P

that lim an wn = 0. Hence, lim |an wn | = 0, and consequently |an wn | is a bounded

n→∞ n→∞
z
sequence: there exists M > 0 such that |an wn | < M for all n. Let x = | |. By
w
assumption, x < 1. Now,
z n
|an z n | = |an wn | < M xn .
w
M
We know that M xn = , so by the comparison test we conclude that |an z n |
P P
1 −P x
converges. Hence, the series an z n absolutely converges, and therefore converges.
Now suppose, for a contradiction, that there is some z with |z| > R for which an z n
P

is convergent. Then there exists M such that |an z n | ≤ M for all n. Pick w such that
R < |w| < |z|. Then
wn w n
|an wn | = |an z n |
≤ M .
zn z
The geometric series |w/z|n converges, as |w/z| < 1. Hence, by the comparison
P

test, |an wn | converges. This contradicts the definition of R.

P
□

Following this theorem, for a power series an z n we call R, as defined above, the
P

radius of convergence of the power series.

2.4.4 Calculating the radius of convergence

Given a power series an z n , the radius of convergence R can often be computed

P

using the ratio or the root test. We explain this by some examples:
P∞ n.
Example 2.4.3. Consider the power series n=0 nz
P∞
Apply the ratio test to n=0 |nz
n |. for z , 0 we have
(n + 1)z n+1 1
n
= (1 + )|z| → |z| as n → ∞.
nz n
P∞
Hence n=0 |nz | converges if |z| < 1 and fails to converge if |z| > 1, so R = 1.
n

zn
Example 2.4.4. Consider the power series ∞
P
n=0 .
n!
P∞ zn
Apply the ratio test to n=0 | |. For z , 0 we have
n!
n!z n+1 |z|
= → 0 as n → ∞.
(n + 1)!z n n+1
P∞ zn
Hence n=0 | | converges for all z ∈ C, so R = ∞.
n!

21
 P∞ nzn.
Example 2.4.5. Consider the power series n=0 n

The form of the series suggests the root test may be a good choice:
(
0 if z = 0
q
n
|nn z n | = n|z| →
∞ otherwise.
P∞
Therefore n=0 |n
nzn| converges only for z = 0, so R = 0.
P∞ n,
Example 2.4.6. Consider the power series n=0 an z where
(
m if n = 2m (for m = 0, 1, · · · ),
an =
0 otherwise.

We cannot directly apply the ratio test to ∞ n z | because some of the terms are
n=0 |aP
n P
∞ m
zero. However, we can apply the test to the sum m=1 |mz 2 |. For z , 0 we have
m+1
|(m + 1)z 2 m + 1 (2m+1 −2m ) m + 1 2m
(
| 0 if |z| < 1,
= |z |= |z| →
|mz 2m | m m ∞ if |z| > 1.

We conclude that R = 1.

2.4.5 Differentiating power series

We know that for a polynomial p(z) = a0 + a1 z + a2 z 2 + · · · + an z n the derivative is

given by p′ (z) = a1 + 2a2 z + · · · + nan z n−1 . This suggests that a power series
∞
f (z) =
X
an z n
n=0

can be differentiated term by term to give

∞
f ′ (z) =
X
nan z n−1 .
n=0

Note that ∞ ∞
d
nan z n−1 =
X X
an z n
n=0 n=0
dz
whereas
∞
!
′ d
f (z) =
X
n
an z .
dz n=0
The differentiation and summation are performed in different orders here. They both
include taking a limit. In general, limiting processes need not commute with one
another. So it is not immediate that term-by-term differentiation of a power series is
valid. We do not even know that nan z n−1 converges for |z| < R. We will see that
P

these are true for complex power series. They have technical and not-so-informative
proofs. What really matters here is the appreciation of the need for such justifications,
rather than their proofs.

22
Lemma 2.4.7. Let f (z) = ∞ n=0 an z have radius of convergence R. Then g(z) =
n
P
P∞ n−1 converges for |z| < R.
n=0 nan z

P∞
Proof. Let |z| < R and choose r such that |z| < r < R. Then converges n=0 an r
n
|z|
absolutely. Hence there exists M > 0 such that |an rn | < M for all n ≥ 0. Let x =
r
and note that 0 < x < 1. Then
z n−1 M n−1
|nan z n−1 | = n|an | rn−1 < n x .
r r
But ∞ n−1 converges to (1−x)−2 . Hence by the comparison test, ∞
n=0 |nan z
n−1 |
P P
n=1 nx P∞
converges. Hence n=0 nan z n−1 converges absolutely and so converges. □

Theorem 2.4.8. Let f (z) = ∞ n radius of convergence R. Then f (z) is

P
n=0 an z have
holomorphic on the disc DR (0) and f (z) = ∞
′ n−1 .
P
n=1 nan z

Proof. Let g(z) = ∞ n−1 . By Lemma 2.4.7 we know that this converges for
P
n=1 nan z
|z| < R. We must show that if |w| < R then f (z) is differentiable at w and the
derivative is equal to g(w). In other words, we must show that if |w| < R then
f (z) − f (w)
f ′ (w) = lim = g(w)
z→w z−w
or equivalently
f (z) − f (w)
 
lim − g(w) = 0.
z→w z−w
For any m ≥ 1 we have
∞
f (z) − f (w) z n − wn
 
− g(w) =
X
an − nan wn−1
z−w n=1
z − w
∞  
= an (z n−1 + wz n−2 + · · · + wn−2 z + wn−1 ) − nan wn−1
X

n=1
∞
= an (z n−1 + wz n−2 + · · · + wn−2 z + wn−1 − nwn−1 )
X

n=1
= (z) + (z),
X X

1,m 2,m

where m
(z) = an (z n−1 + wz n−2 + · · · + wn−2 z + wn−1 − nwn−1 )
X X

1,m n=1

and ∞
(z) = an (z n−1 + wz n−2 + · · · + wn−2 z + wn−1 − nwn−1 ).
X X

2,m n=m+1

Let ε > 0, and choose r such that |w| < r < R. Then as in the proof of Lemma 2.4.7
P∞ n−1 is absolutely convergent. Hence we can choose m such that
n=1 nan r
∞
X ε
|nan rn−1 | < .
n=m+1
2

23
Since |w| < r, provided that z is close enough to w so that |z| < r we have
∞
ε
(z)| ≤ 2n|an |rn−1 < .
X X
|
2,m n=m+1
2

Now consider 1,m (z). This is a polynomial in z and so is a continuous function.

P

Note that 1,m (w) = 0. Hence lim 1,m (z) = 0. Hence, provided that z is close
P P
z→w
enough to w we have
ε
| (z)| <
X
.
1,m
2

It then follows that if z is close enough to w we get

f (z) − f (w) ε ε
− g(w) = | (z) + (z)| ≤ | (z)| + | (z)| < + = ε.
X X X X
z−w 1,m 2,m 1,m 2,m
2 2

The above two results have a very important consequence: if f (z) = ∞ n P

n=0 an z
converges on DR (0) then we can differentiate it as many times as we like within the
disc of convergence DR (0)

Corollary 2.4.9. Let f (z) = ∞ n

P
n=0 an z have radius of convergence R. Then all the
higher derivatives f , f , f , · · · , f , · · · of f exist for z ∈ DR (0) and
′ ′′ ′′′ (m)

∞ ∞
n!
f (m) (z) = n(n − 1) · · · (n − m + 1)an z n−m =
X X
an z n−m .
n=m n=m (n − m)!

Proof. This is a simple induction on m. □

Example. The geometric series z n has radius of convergence 1, and provides a

P

power series expansion of (1 − z) for |z| < 1. By differentiation we can produce

−1

1 d 1
= ( ) = 1 + 2z + 3z 2 + · · · (|z| < 1).
(1 − z)2 dz 1 − z

24
2.5 A Hamper of Functions

You are familiar with trigonometric, exponential, and hyperbolic functions. They
are often defined using geometric properties, and then one works out their Maclaurin
expansions. This approach is intuitive, but not so rigorous. A better way is perhaps
to define these functions as power series, and then derive their properties (including
the geometric intuitions). Of course, the intuitions fail for complex functions. So we
will take the latter approach to begin with.

2.5.1 The exponential function

P∞ xn
You may recall that the real exponential function is ex = n=0 . Similarly, we can
n!
define the complex exponential function as
∞
zn
ez = (z ∈ C).
X

n=0
n!

We have already seen in the exercises that it has infinite radius of convergence. Note
that we still do not know if this is compatible with our earlier use of the symbol eiθ
in eiθ = cos θ + i sin θ.
It follows immediately from Theorem 2.4.8 that the exponential function is entire. We
also obtain the following with a straightforward computation:
∞ ∞ ∞ ∞
d z d X zn nz n−1 z n−1 zn
e = = = = = ez .
X X X
dz dz n=0 n! n=1
n! n=1
(n − 1)! n=0
n!

We already knew this was true in the real-valued case. We also know that for x, y ∈ R
we have ex+y = ex ey . Let us prove this, and some other basic properties, in the
complex case.
Recall that, from Theorem 2.3.7 of f is holomorphic on a domain D and f ′ (z) = 0
for all z ∈ D then f is constant.
Proposition 2.5.1. (properties of exponential)

(i) e0 = 1;
(ii) ez+w = ez ew for all z, w ∈ C;
(iii) ez , 0 for all z ∈ C.

Proof. (i) follows immediately from the series definition. For (ii), fix c ∈ C and
consider f (z) = ez ec−z . By differentiation, and the chain rule, we have
f ′ (z) = ez ec−z − ez ec−z = 0.
Therefore by Theorem 2.3.7, there exists a constant M ∈ C such that f (z) = M for a
ll z ∈ C (note that M depends on c). To find M , we put z = c and obtain M = ec ec−c ,
so M = ec by (i). Thus ec = ez ec−z for all z, c ∈ C. Choosing c = w +z we obtain (ii).
From (i) and (ii) we have ez e−z = 1, so (iii) follows. □

25
Let z = x + iy with x, y ∈ R. Then |ez | = ex , and in particular |eiy | = 1 for all y ∈ R.
To prove this, observe that

|ez |2 = ez ez = ez ez = ez+z = e2x = (ex )2 .

Hence, |ez | = ex as both sides are real and positive. The assertion |eiy | = 1 can be
seen by taking x = 0.

2.5.2 Trigonometric functions

For z ∈ C, define
∞
z2 z4 z 2n
cos z = 1 − + − ··· = (−1)n
X
2! 4! n=0
(2n)!
∞
z3 z5 z 2n+1
sin z = z − + − ··· = (−1)n
X
3! 5! n=0
(2n + 1)!

One can easily check, using the ratio test, that both these series have radius of
convergence ∞, hence both sin z and cos z are entire functions. Moreover, we have

cos′ z = − sin z and sin′ z = cos z.

It also follows immediately from the definition that

eiz = cos z + i sin z (for z ∈ C).

In particular, the Euler formula eiθ = cos θ + i sin θ for θ ∈ R holds, as a special case.
Using this, we also recover |eiθ | = 1 (and cos2 θ + sin2 θ = 1), which we proved above
without using this fact. We can also write

ex+iy = ex eiy = ex (cos y + i sin y).

On the other hand, we have

1 1 iz
cos z = (eiz + e−iz ) and sin z = (e − e−iz ).
2 2i
Taking the squares and adding them shows that for all z ∈ C we have

cos2 z + sin2 z = 1.

These are identities we knew that they hold for real numbers, but now we proved that
they are also valid in C. One could go further and prove all kinds of trigonometric
identities, e.g., cos(z + w) = cos z cos w − sin z sin w.

26
2.5.3 Hyperbolic functions

Similarly, for z ∈ C, we can define

∞
z2 z4 z 2n
cosh z = 1 + + + ··· =
X
2! 4! n=0
(2n)!
∞
z3 z5 z 2n+1
sinh z = z + + + ··· =
X
3! 5! n=0
(2n + 1)!

Using the ratio test, we see that both these series have radius of convergence ∞, hence
both sinh z and cosh z are entire functions. Moreover, we have

cosh′ z = sinh z and sinh′ z = cosh z.

Similar to the trigonometric case, we can prove the following using the definitions:
1 1
cosh z = (ez + e−z ) and sinh z = (ez − e−z )
2 2
cosh2 z − sinh2 z = 1 (for all z ∈ C)
One also has the Osborn’s rules:

cos iz = cosh z and sin iz = i sinh z

We can use Osborn’s rules for z = x + iy to show that

cos(x + iy) = cos x cos iy − sin x sin iy = cos x cosh y − i sin x sinh y

sin(x + iy) = sin x cos iy + cos x sin iy = sin x cosh y + i cos x sinh y

Remark 2.5.2. Unboundedness

It follows immediately from the definition of the hyperbolic functions that cosh x and
sinh x tend to ∞ as the real variable x → ∞. From Osborn’s rules we see that

| cos iy| = | cosh y| → ∞ as y → ∞,

| sin iy| = | sinh y| → ∞ as y → ∞.

This is in contrast to the behaviour of the real functions, as | cos x| ≤ 1 and | sin x| ≤ 1
for x ∈ R. Hence, the complex functions cos z and sin z are unbounded in C.

Definition 2.5.3. Let f : C → C. We say that a number p ∈ C is a period for f if

f (z + p) = f (z) for all z ∈ C.

Clearly if p ∈ C is a period and n ∈ Z then np is also a period. For the exponential

function,
e2πi = cos 2π + i sin 2π = 1
so that
ez+2πi = ez e2πi = ez .

27
Hence 2πi is a period for the exponential function, as is any 2nπi for n ∈ Z. We will
check in the exercises that these are the only periods for the exponential. We will also
check that the only complex periods for cos and sin are 2nπ. To solve such questions
is the same as solving equations of type ez = 1 (say, in proving ez+w = ew for all
w ∈ C). One then obtains the following:
ez = 1 ⇐⇒ z = 2nπi (n ∈ Z)
e = −1 ⇐⇒ z = (2n + 1)πi (n ∈ Z)
z

1
cos z = 0 ⇐⇒ z = (n + )π (n ∈ Z)
2
sin z = 0 ⇐⇒ z = nπ (n ∈ Z)
1
cosh z = 0 ⇐⇒ z = (n + )πi (n ∈ Z)
2
sinh z = 0 ⇐⇒ z = nπi (n ∈ Z)
In it, the latter two must be surprising in comparison with their real counterparts.
Knowing these, one could proceed with defining other trigonometric functions. For
example, we can deduce that tan z = sin z/ cos z is defined and is holomorphic in
1
C\{(n + )π | (n ∈ Z)}.
2

2.5.4 Multi-valued functions

We are now ready to see some examples of multi-valued complex functions. Indeed,
we have already met one such function: the arg z. Recall that the angle θ in the
polar expression z = |z|eiθ is not uniquely determined. This is the cause of the multi-
valuedness in complex analysis. For any z , 0, we define the argument of z to
be
arg z = {θ ∈ R | z = |z|eiθ }.
This “function” assigns a set of numbers to the point z. It is in fact an infinite set of
numbers of the form θ + 2nπ for n ∈ Z, where θ is any fixed real number for which
eiθ = z/|z|.
Another such function is the logarithmic function. We know that the real logarithmic
function, defined on (0, ∞) ⊂ R, is the inverse of the exponential. We are going to
define the complex logarithms similarly. However, we exclusively work with the base
e, so we write log x instead of ln x). Note that the complex exponential function
differs from its real analogue: it has periods.
Suppose z ∈ C and z , 0. Put z = ew = eu+iv with u, v real. Then
|z| = |eu eiv | = eu and arg z = {v + 2nπ | n ∈ Z}.
This gives the following important relation:
ew = z ⇐⇒ w = log |z| + i arg z
Definition 2.5.4. Let z ∈ C, z , 0. Then a complex logarithm of z is
log z = log |z| + i arg z
where arg z is any argument of z.

28
It is clearly a multi-valued function. The principal value of log z is the value of log z
where arg z has its principal value Arg z, that is, the unique value of the argument in
(−π, π]. We denote the principal logarithm by Log z:
Log z = log |z| + i Arg z

Indeed, in defining Arg z we made a choice: the convention of working with (−π, π].
We could have made any other choice of interval with the same length, for example
[0, 2π). This choice is made to make it easier to work with multi-valued functions.
Let us fix this set for Arg z to be (−π, π] and see what it means in the complex plane.
To do so, we put a cut (also called a branch cut) in the plane along the non-positive
real axis (−∞, 0] and we forbid z to cross the cut! We regard the cut as having two
edges: on the upper edge, θ = π and on the lower edge, θ = −π. The point 0 is called
a branch point.

• z = |z|eiθ

cut θ
•
0

A key point of doing this branch cutting is to deduce that the function Log z is
continuous, and holomorphic, on the cut plane. This allows differentiation and much
more. We will not prove continuity here and take it as a fact (the proof is not
complicated, but slightly tedious).
Proposition 2.5.5. The principal logarithm Log z is holomorphic on the cut plane
and
d 1
Log z = .
dz z

Proof. Let w = Log z. Then z = ew . Let Log(z + h) = w + k. Then by continuity of

Log z on the cut plane we have that k → 0 as h → 0. Then
d Log(z + h) − Log(z)
Log z = lim
dz h→0 h
!−1
(w + k) − w ew+k − ew
= lim w+k = lim
k→0 e − ew k→0 k
!−1 −1
ew+k − ew d w 1

= lim = e = .
k→0 k dw z
□

Having defined the complex logarithm, we can define complex powers: for z, w ∈ C
with z , 0, we define the principal value of z w to be
z w = ew Log z
and the subsidiary values to be ew log z .

29
Chapter 3

Complex Integration

For the real integral

Z b
f (x)dx
a
there is no ambiguity on what is integrated: it is simply the integral of f from a
to b, and there is only one way to get from a to b (a straight forward move). One
may argue that there are other ways. For instance, you could start from a and go
half-way towards b, then go backwards a little and stop and carry on towards b again.
Of course we know that such manoeuvre will not change the integral as we add and
subtract some irrelevant integrals in there. For two complex numbers z0 , z1 ∈ C, how
would one define the following integral?
Z z1
f (z)dz
z0

the idea is to start from z0 and move through the complex plane to z1 and integrate
f as we go. However, there are many ways of moving from z0 to z1 . Let γ be a path
from z0 to z1 . Similar to calculus, or real analysis over R2 , we can define
Z
f (z)dz,
γ

the integral of f along γ. We will see that in complex analysis the integration is
independent of the choice of γ, in many cases.

3.1 Contour Integration

Definition 3.1.1. A path is a continuous function γ : [a, b] → C, where [a, b] is a

real interval.

For any a ≤ t ≤ b, the image γ(t) is a point on the path. We say that the path γ starts
at γ(a) and ends at γ(b). Note that by definition a path is a function. Conveniently,
we identify the function γ with its image in C. However, we bear in mind that this
set comes with a specified orientation. With this intuition, we sometimes call the
function γ(t) a parametrisation of the path γ.

30
Note that the same path can have different parametrisations. For example,
γ1 (t) = t + it and γ2 (t) = t2 + it2 for 0 ≤ t ≤ 1
are both parametrisations of the straight line that starts at 0 and ends at 1 + i.
We will prove later that an integral along a path is independent of the choice of
parametrisation.
A path γ : [a, b] → C is called a closed path, or a closed loop, if γ(a) = γ(b), i.e., the
path starts and ends at the same point.
A path γ : [a, b] → C is called smooth if γ is differentiable and γ ′ is continuous (of
course we mean one-sided differentiable at a and b).
The length of a smooth path γ is defined to be
Z b
length(γ) = |γ ′ (t)|dt.
a

The reverse of the path γ : [a, b] → C travels backwards in comparison to γ. It starts

at γ(b) and ends at γ(a). We denote it by −γ, called it the reverse path of γ, and
define it formally by −γ : [a, b] → C with
−γ(t) = γ(a + b − t).

Straight line: one could parametrise the straight line from z0 to z1 by

γ(t) = (1 − t)z0 + tz1 (0 ≤ t ≤ 1).
We sometimes denote this path by [z0 , z1 ]. It is easy to see that
length([z0 , z1 ]) = |z1 − z0 |.

Unit circle: this is the circle of radius 1 centred at 0. It can be considered as a

closed path (regarded anti-clockwise) and is parametrised by
γ(t) = eit = cos t + i sin t (0 ≤ t ≤ 2π).
More generally, the circle CR (0) can be parametrised by γ(t) = Reit for 0 ≤ t ≤ 2π.
Its length, as expected, is
Z 2π Z 2π
length(CR (0)) = |Re |dt =
it
Rdt = 2πR.
o o

Definition 3.1.2. Let f : D → C be a function defined on a domain D. Let γ : [a, b] →

D be a smooth path in D. The integral of f along γ is defined to be
Z Z b
f (z)dz = f (γ(t))γ ′ (t)dt.
γ a

We often write f for f (z)dz.

R R
γ γ
Note that we should write f (γ(t))γ ′ (t) = u(t)+iv(t), where u, v : [a, b] → R and define
Z Z b Z b
f= u(t)dt + i v(t)dt.
γ a a

31
Example 3.1.3. Let f (z) = z 2 and γ(t) = t + it, 0 ≤ t ≤ 1.

Then f (γ(t)) = (t + it)2 = 2it2 and γ ′ (t) = 1 + i. Hence

Z Z 1 Z 1
f (z)dz = f (γ(t))γ ′ (t)dt = (2it2 )(1 + i)dt
γ 0 0
2
Z 1
−2
= 2i(1 + i) t2 dt = +i .
0 3 3

You may take a different parametrisation of [0, 1 + i] in the example above. For
instance, you could take the path η(t) = t2 + it2 , 0 ≤ t ≤ 1, which is a dif-
ferent parametrisation of the same straight line. A similar calculation shows that
= (do this computation yourself!). The following proposition states
R 2 R 2
η z dz γ z dz
this in general.

Proposition 3.1.4. Let γ : [a, b] → C be a path and let ϕ : [c, d] → [a, b] be an in-
creasing smooth bijection. Then γ ◦ ϕ : [c, d] → C is a path, with the same image as γ.
Moreover, Z Z
f= f
γ◦ϕ γ

for any continuous function f .

Proof. It is clear that both γ and γ ◦ ϕ have the same image, hence they are different
parametrisations of the same path. We have
Z Z d
f= f (γ(ϕ(t)))(γ(ϕ(t)))′ dt
γ◦ϕ c
Z d
= f (γ(ϕ(t)))γ ′ (ϕ(t))ϕ′ (t)dt by the chain rule
c
Z d
= f (γ(ϕ(t)))γ ′ (ϕ(t))dϕ(t)
c
Z b
= f (γ(x))γ ′ (x)dx by a change of variables x = ϕ(t)
Za
= f.
γ

Note that if ϕ in the proposition above is a decreasing smooth bijection then γ ◦ ϕ

has the same image as γ but the path transverses this in the opposite direction, i.e.,
γ ◦ ϕ parametrises −γ.
Often we integrate over a number of paths joined together. A typical example is
below.

Definition 3.1.5. A contour γ is a collection of smooth paths γ1 , · · · , γn where the

end point of γk coincides with the start point of γk+1 for 1 ≤ k ≤ n − 1. We write

γ = γ1 + · · · + γn .

32
If the end point of γn coincides with the start point of γ1 then we call γ a closed
contour. One can think of a contour as a path that is smooth except possibly at
finitely many points.
Example: Let 0 < r < R be real numbers and define

γ1 : [r, R] → C γ1 (t) = t
γ2 : [0, π] → C γ2 (t) = Reit
γ3 : [−R, −r] → C γ3 (t) = t
γ4 : [−π, 0] → C γ4 (t) = re−it .

Then γ = γ1 + γ2 + γ3 + γ4 is the closed contour in the picture shown below.

−R −r r R

The length of the contour γ = γ1 + · · · + γn is defined to be

length(γ) = length(γ1 ) + · · · + length(γn ).

Suppose γ = γ1 + · · · + γn is a contour in a domain D. Then we define

Z Z Z
f= f + ··· + f.
γ γ1 γn

The following basic properties of contour integration follow from this definition.

Proposition 3.1.6. Let f, g : D → C be continuous and let c ∈ C. Suppose γ, γ1 , γ2

are contours in D. Suppose that the end point of γ1 is the start point of γ2 . Then

(a) Z Z Z
f= f+ f;
γ1 +γ2 γ1 γ2

(b) Z Z Z
(f + g) = f+ g;
γ γ γ

(c) Z Z
cf = c f;
γ γ

(d) Z Z
f =− f.
−γ γ

33
Recall from calculus that one can calculate integral of f by finding its anti-derivative
(if it exists),R i.e., a function F with F ′ = f . The fundamental theorem of calculus then
states that ab f (x)dx = F (b) − F (a). The analogue exists for the complex integrals.
Let f : D → C be a continuous function. We say that a function F : D → C is an
anti-derivative of f on D if F ′ = f .

Theorem 3.1.7 (The Fundamental Theorem of Contour Integration). Suppose that

f : D → C is continuous, F : D → C is an anti-derivative of f on D, and γ is a
contour in D from z0 to z1 . Then
Z
f = F (z1 ) − F (z0 ).
γ

Proof. It suffices to prove the theorem for smooth paths. Let γ : [a, b] → D be
a smooth path with γ(a) = z0 and γ(b) = z1 . Let g(t) = f (γ(t))γ ′ (t) and let
G(t) = F (γ(t)). Then by the chain rule

G′ (t) = F ′ (γ(t))γ ′ (t) = f (γ(t))γ ′ (t) = g(t).

Write g(t) = u(t) + iv(t) and G(t) = U (t) + iV (t) so that U ′ = u and V ′ = v. Then
Z Z b
f= f (γ(t))γ ′ (t)dt
γ a
Z b
= g(t)dt
a
Z b Z b
= u(t)dt + i v(t)dt
a a
= U (t)|ba + iV (t)|ba by the Fundamental Theorem of Calculus
= G(t)|ba = F (z1 ) − F (z0 ).

Note that the integration above does not depend on the choice of the path γ. All we
need to know is the anti-derivative for f on a domain that contains z0 and z1 .
Example. Let f (z) = z 2 and γ be any contour from 0 to 1 + i. Then F (z) = z 3 /3
is the anti-derivative of f and
1 1 2 2
Z
z 2 dz = (1 + i)3 − 03 = − + i.
γ 3 3 3 3

Remark 3.1.8. Let f be a function with an anti-derivative on a domain D. Suppose

that γ is a closed path in D. Then the fundamental theorem of contour integration
implies that γ f = 0. However possessing an anti-derivative is a very strong condition
R

on f .

Consider for example f (z) = 1/z defined on D = C\{0}. A natural candidate for
the antiderivative of f would be Log z. However, Log z is only continuous on the

34
cut-plane. Hence Log z is not differentiable on D and so is not an anti-derivative
of 1/z.
Let us integrate f (z) = 1/z to examine this for the closed path C1 (0), parametrised
by γ(t) = eit , 0 ≤ t ≤ 2π. Then

1
Z Z 2π Z 2π
f= f (γ(t))γ ′ (t)dt = ieit dt = 2πi.
γ 0 0 eit

If f had an anti-derivative on a domain that contains C1 (0) (or in fact any circle
Cr (0) for r R∈ (0, ∞)) then by the fundamental theorem of contour integration we
would have Cr (0) f = 0. Hence, 1/z does not have an anti-derivative on any domain
that contains 0.
In general, looking for an anti-derivative is not the best way of calculating complex
integrals. There are some powerful techniques that allow us to calculate many complex
integrals without needing anti-derivatives.

3.2 The Estimation Lemma

In real integration we know the following by considering the integral of f (x) from a
to b as the area under the graph of f :
Z b Z b
f (x)dx ≤ |f (x)|dx.
a a

If the function is bounded on [a, b], that is, there exists M ∈ R such that |f (x)| ≤ M
for all x ∈ [a, b], then
Z b
f (x)dx ≤ M (b − a).
a
The complex analogues of these facts are the topic of this section. As you can imagine,
the absence of the intuition above in the complex setting makes the proofs here more
complicated.

Lemma 3.2.1. Let u, v : [a, b] → R be continuous functions. Then

Z b Z b
(u(t) + iv(t))dt ≤ |u(t) + iv(t)|dt.
a a

Rb
Proof. Write a (u(t) + iv(t))dt = α + iβ. Then
Z b
α + β = (α − iβ)(α + iβ) = (α − iβ)
2 2
(u(t) + iv(t))dt
a
Z b
= (α − iβ)(u(t) + iv(t))dt
a
Z b Z b
= (αu(t) + βv(t))dt + i (αv(t) − βu(t))dt.
a a

35
However, α2 + β 2 is real, hence the imaginary part of the above expression must be
zero: Z b
(αv(t) − βu(t))dt = 0
a
so that Z b
α +β =
2 2
(αu(t) + βv(t))dt.
a
Note that αu(t) + βv(t) is the real part of (α − iβ)(u(t) + iv(t)). Recall that for any
complex number z we have Re(z) ≤ |z|, hence we get

αu(t) + βv(t) ≤ |(α − iβ)(u(t) + iv(t))|

q
= |α − iβ||u(t) + iv(t)| = α2 + β 2 |u(t) + iv(t)|

Therefore
Z b
α2 + β 2 = (αu(t) + βv(t))dt
a
q Z b
≤ α2 + β 2 |u(t) + iv(t)|dt.
a

Now, cancelling the term α2 + β 2 gives

p

Z b q Z b
(u(t) + iv(t))dt = |α + iβ| = α2 + β 2 ≤ |u(t) + iv(t)|dt
a a

as claimed. □

Now we can prove the following important result.

Lemma 3.2.2. (The Estimation Lemma) Let f : D → C be continuous and let γ
be a contour in D. Suppose that there exists M ∈ R such that |f (z)| ≤ M for all z
on γ. Then Z
f ≤ M length(γ).
γ

Proof. Using the previous lemma, we have

Z Z b
f = f (γ(t))γ ′ (t)dt
γ a
Z b Z b
≤ |f (γ(t))||γ ′ (t)|dt ≤ M |γ ′ (t)|dt = M length(γ).
a a

We will use the Estimation lemma in two different ways: if f is a function with small
modulus values along a contour γ thenR γ f is small; if f is a continuous function and
R

γ is a contour with small length then γ f is small.

z
Example. Let us obtain a bound for the integral of 2 along γ, the semicircle
z +1
z = 3eiθ , θ ∈ [0, π].

36
The length of γ is 3π. On γ, |z| = 3 and we use it to bound |f |. By the reverse
triangle inequality we have

|z 2 + 1| ≥ ||z|2 − 1| = 8 along γ.

Hence, along γ,
z 3
≤ .
+1 8 z2
Therefore, by the Estimation Lemma, we have
z 9π
Z
dz ≤ .
γ z2 +1 8

1
Example. Let f (z) = and let γ(t) = 5eit , 0 ≤ t ≤ 2π, be the circle of
z2 + z + 1
radius 5 centred at 0.
For any z ∈ γ we have |z| = 5. Hence

|z 2 + z + 1| ≥ |z|2 − |z + 1| by the reverse triangle inequality

≥ |z|2 − |z| − 1 by the triangle inequality
= 25 − 5 − 1 = 19.

Hence, for any z ∈ γ we have

1 1
|f (z)| = ≤ .
z2 +z+1 19
On the other hand length(γ) = 10π, so that by the Estimation Lemma we get

dz 10π
Z
≤ .
γ z2 +z+1 19
Z
Example. Obtain a bound on ez dz, where C is the semicircle z = eiθ , 0 ≤ θ ≤ π.
C
It is easy to see that length(C) = π. On C,

|ez | = |ex+iy | = ex ≤ e

as x = cos(θ) ≤ 1 for θ real. Hence by the Estimation Lemma we have

Z
ez dz ≤ πe.
C

37
3.3 Cauchy’s Theorem

Suppose that f : D → C is a function

Z with an anti-derivative. Then for a closed
contour γ ⊂ D we know that f = 0 by the Fundamental theorem of Contour
γ
Integration. As we have seen, for a function to have an anti-derivative is a big ask.
Nonetheless, such integrals over closed contours could still be zero even in the absence
of anti-derivatives. There are many theorems that states the vanishing of integrals
along closed contours with a variety of hypothesis. They are all somewhat attributed
to Cauchy.
Here we state a simple version for a warmup and then look into some more general
forms. The issue is that any direct proof of Cauchy’s theorem is lengthy and tedious,
and the techniques are rarely used in other statements. Although, Cauchy’s theorem
itself is crucial in many proofs later on. Let us see the first (very restricted) version!
y
Let C be a closed contour and consider
it in the xy-plane. Suppose that for y = g(x)
a ≤ x ≤ b the lower part of C may C
be written as y = f (x), and the upper
A
part of C as y = g(x).
The strong assumptions on C imme-
diately restrict the applicability of the y = f (x)
theorem below, as not all closed con-
tours can be expressed in this way. a b x

Theorem 3.3.1 (Cauchy’s Theorem - easy version). Let C be a contour as above,

and suppose f is a function that is holomorphic on and within C. Write f (z) =
u(x, y) + iv(x, y) and assume partial derivatives of u and v are continuous throughout
A where A is the domain within C. Then
Z
f (z) dz = 0.
C

∂u
Z Z
Proof. Put f (z) = u(x, y) + iv(x, y) and consider dA . Then, continuity of
A ∂y
∂u/∂y on A results in

g(x)
Z b Z g(x) ! Z b !
∂u ∂u
Z Z
dA = dy dx = u(x, y) dx
A ∂y a f (x) ∂y a f (x)
Z b
= u(x, g(x)) − u(x, f (x)) dx
 
a
Z a Z b Z
=− u(x, g(x)) dx − u(x, f (x)) dx = − u dx.
b a C

Similar arguments may be used to show that

∂v ∂u ∂v
Z Z Z Z Z Z Z Z Z
dA = − v dx, dA = u dy, dA = v dy.
A ∂y C A ∂x C A ∂x C

38
Hence
Z Z Z
f (z) dz = (u + iv)(dx + i dy) = [u dx − v dy + i(v dx + u dy)]
C C C
∂u ∂v ∂v ∂u
Z   
= − − +i − + dA = 0
A ∂y ∂x ∂y ∂x
because of the Cauchy-Riemann equations, as f is holomorphic on and within C. □

This is a relatively short proof, but it implicitly uses the Green’s Theorem from
calculus. So it is not based on complex analysis. Also, the assumption that C has
this very particular shape is too restrictive. Also, the assumption on continuity of
the partial derivatives is completely unnecessary, as this is automatically valid for
holomorphic functions. Although, the proof of this fact uses the Cauchy’s Theorem.
As it was mentioned earlier, there are direct proofs for Cauchy’s Theorem, which
mostly start by proving it when C is a triangle and then build the complete proof
from there. Interested students can consult any of the recommended books for a
complete proof (warning: the proof usually takes about a chapter!).
Theorem 3.3.2 (Cauchy’s Theorem for contours). Suppose that f is holomorphic on
and within a closed contour γ. Then
Z
f (z) dz = 0.
γ

Before we state the most general form of Cauchy’s Theorem we need to get familiarised
with the notion of winding numbers.
The winding number w(γ, z0 ) is defined for a closed path (or contour) γ and a
point z0 that is not on γ. Intuitively, it is the number of times γ turns around z0 . To
visualise it, imagine you tie a piece of string at one end to a pencil, with the tip of the
pencil on z0 , and the other end of the string tracing around γ. When you get back
to where you started (one loop around γ) the string has wrapped around the pencil
w(γ, z0 ) times. By convention, this counts the anti-clockwise turns positively and the
clockwise turns negatively. Mathematically this number can be calculated using the
formula below (this may be proven in the lectures we have spare time):
1 dz
Z
w(γ, z0 ) =
2πi γ z − z0

It is clear from the definition that if z0 is not located within γ then w(γ, z0 ) = 0.
When calculating the winding number we use the visual definition and do the counting
by eye. However, we use the mathematical definition in the proofs.

z0

z0 z0 z0

γ1 γ2 γ3 γ4

w(γ1 , z0 ) = 1, w(γ2 , z0 ) = 0, w(γ3 , z0 ) = −1, w(γ4 , z0 ) = 2.

39
The following are geometrically obvious consequences of this definition.

Proposition 3.3.3. Let γ, γ1 and γ2 be closed paths not passing through z0 . Then

w(γ1 + γ2 , z0 ) = w(γ1 , z0 ) + w(γ2 , z0 ) and w(−γ, z0 ) = −w(γ, z0 ).

Proof.
1 dz
Z
w(γ1 + γ2 , z0 ) =
2πi γ1 +γ2 z − z0
1 dz 1 dz
Z Z
= + = w(γ1 , z0 ) + w(γ2 , z0 )
2πi γ1 z − z0 2πi γ2 z − z0

1 dz −1 dz
Z Z
w(−γ, z0 ) = = = −w(γ, z0 ) □
2πi −γ z − z0 2πi γ z − z0

The following theorem is a version of Cauchy’s theorem that incorporates winding

numbers.

Theorem 3.3.4 (Cauchy’s theorem). Let f be holomorphic on a domain D. If γ is

a closed contour in D with w(γ, z) = 0 for all z < D then γ f = 0.
R

We omit the proof of this theorem. It immediately implies the following statement.

Theorem 3.3.5 (Cauchy’s Theorem). Let f be holomorphic on a domain D. Let

γ1 , . . . , γn be closed contours in D such that w(γ1 , z) + · · · + w(γn , z) = 0 for all
z < D. Then Z Z
f + ··· + f = 0.
γ1 γn

Proof. Suppose that γr starts and ends at zr ∈ D, 1 ≤ r ≤ n. Choose any z0 ∈ D and

contours σ1 , . . . , σn in D which join z0 to z1 , . . . , zn , respectively (as in the picture
below). For each r, σr + γr − σr is a closed contour that starts and ends at z0 and,
moreover, for z < D we have w(σr + γr − σr , z) = w(γr , z). We see that

γ = σ1 + γ1 − σ1 + · · · + σn + γn − σn

is a closed contour that starts and ends at z0 . For z < D we have

n n
w(γ, z) = w(σ1 +γ1 −σ1 +· · ·+σn +γn −σn , z) = w(σr +γr −σr , z) = w(γr , z) = 0.
X X

r=1 r=1

Therefore f = 0 by Cauchy’s Theorem. Hence, as f =− f , we have

R R R
γ −σr σr

n Z n Z Z Z  Z
f= f+ = f = 0.
X X
f− f □
r=1 γr r=1 σr γr σr γ

40
 z2
σ2
z0
σ1
γ2 z1

γ1

3.3.1 Contour deformation

The idea of manipulating the contours in the proof of (the general form of) Cauchy’s
Theorem can sometimes be used to simplify computations. The process is called
deformation of the contour and it goes as follows.
Consider two positively oriented simple closed contours C1 and C2 (simple means it
does not cross itself, or in other words the winding number with respect to any interior
point is 1), such that C2 is contained within C1 , and suppose that a function f is
analytic on and between these contours (C1 and C2 may touch, or even coincide over
part of their length). Denote by C12 a contour linking C1 to C2 , then by Cauchy’s
theorem

Z Z Z
f (z) dz + f (z) dz + f (z) dz
C1 C12 −C2 C1
Z
+ f (z) dz = 0 C2
−C12

but Z Z
f (z) dz = − f (z) dz
−C2 C2

and
Z Z C1
f (z) dz = − f (z) dz
−C12 C12 C12 −C2
so that −C12
Z Z
f (z) dz = f (z) dz.
C1 C2

Note that the value of the integral does not depend on the shape of the integration
contour.

41
 dz
Z
Example. Evaluate where C is the square joining the four points z = ±1 ± i
C z
taken in the positive sense.
The integral could be evaluated by breaking the contour down into four sections, one
along each side of the square. However, the integrand is holomorphic between and on
C and the circle C1 (0), on which z = eiθ , θ ∈ [0, 2π], so that Cauchy’s theorem gives

1 dz 1
Z 2π Z 2π  
dz dz
Z Z
= = dθ = (ieiθ ) dθ = 2πi.
C z C1 (0) z 0 z dθ 0 eiθ

In fact the integrand is analytic between and on C and CR (0) for any R > 0 so that

dz dz
Z Z
= .
C z CR (0) z

The contour may be deformed ‘inwards’ or ‘outwards’ provided the deformation does
not pass over any points where the integrand is not holomorphic.

42
3.4 Cauchy’s Integral Formula

Cauchy’s Theorem was fascinating in its nature: it tells us that if we know about
differentiability of a function then we may be able to integrate it. But the applicability
of that theorem is enormous. The first door it opens is the content of this section:
Cauchy’s Integral Formula. Once we have established this formula we are able to
prove several extraordinary facts about complex functions.
Theorem 3.4.1 (Cauchy’s Integral Formula). Let f be holomorphic inside and
on a positively oriented closed contour γ. Then for any a inside γ we have
1 f (z)
Z
f (a) = dz.
2πi γ z−a

Proof. Fix a inside γ. As a is inside γ, there exists R > 0 such that CR (a) is entirely
within γ (and f is holomorphic on and within it). For any r < R we have
f (z) f (z)
Z Z
dz = dz
γ z−a Cr (a) z−a
by Cauchy’s Theorem (note that both γ and Cr (a) are positively oriented).
As a is fixed, we can write
f (a) 1
Z Z
dz = f (a) dz = 2πif (a).
Cr (a) z−a Cr (a) z−a
Now we have
1 f (z) 1 f (z) − f (a)
Z Z
dz − f (a) = dz
2πi γ z−a 2πi Cr (a) z−a
1 2π f (a + reiθ ) − f (a)
Z
= ireiθ dθ
2πi 0 reiθ
1
≤ × 2π × sup |f (a + reiθ ) − f (a)|.
2π θ∈[0,2π]

The latter inequality is obtained by the Estimation Lemma. Since f is continuous at

a, the supremum above tends to zero as r → 0. That proves that the expression on
the left-hand side of the display, which is independent of r, must be zero. □

Examples
cos z
Z
(1) dz = 2πi cos(0) = 2πi.
C3 (2) z
z2
Z
(2) dz.
C1 (i) z2 + 1
z2
Note that the only points at which is not holomorphic are i and −i, but
z2 + 1
only i is inside the contour C1 (i). Hence, we can write
z2 z2 i2
Z Z
dz = dz = 2πi = −π.
C1 (i) z2 + 1 C1 (i) (z − i)(z + i) i+i

43
 iπz
e
Z
2
(3) dz.
C2 (0) z2 − 1
iπz
e 2
We cannot directly use Cauchy’s integral formula here as the function is
z2 − 1
not holomorphic at two points 1 and −1 and they are both inside the contour
C2 (0). However, we can use partial fractions to write
iπz iπz iπz
e 1 e 2 1 e 2
Z Z Z
2
dz = dz − dz
C2 (0) z −1
2 2 C2 (0) z − 1 2 C2 (0) z + 1
1 iπz 1 iπz 2πi 1 π π
   
= 2πi( e 2 − e 2 )= × sin =
2 z=1 2 z=−1 2 2i 2 2

3.5 Applications of Cauchy’s Integral Formula

The first application of Cauchy’s Integral Formula we study is the so-called Liouville’s
theorem, which has no real analogue. We have seen already that functions such as
cos z and sin z are unbounded over C, unlike their real counterparts. This theorem
shows that this behaviour is typical for complex entire functions.

Theorem 3.5.1 (Liouville’s Theorem). Let f be holomorphic and bounded in the

complex plane C. Then f is constant.

Proof. Suppose that there exists M > 0 such that |f (z)| ≤ M for all z ∈ C. Fix a and
1
b in C. We will show that f (a) = f (b). Take R ≥ 2 max{|a|, |b|}, so that |z − a| ≥ R
2
1
and |z − b| ≥ R whenever |z| = R. Now, apply Cauchy’s integral formula with
2
contour CR (0):

1 1 1
Z  
f (a) − f (b) = f (z) − dz
2πi CR (0) z−a z−b
a−b f (z)
Z
= dz.
2πi CR (0) (z − a)(z − b)

Now, by the Estimation Lemma, we have

1 |a − b| 4M |a − b|
|f (a) − f (b)| ≤ 2πRM R 2 = .
2π (2) R

The right-hand side can be made arbitrarily small by taking R → ∞. Hence, we have
that f (a) = f (b) for any a and b in C. □

Liouville’s Theorem comes with an unexpected bonus:

Fundamental Theorem of Algebra. Let p(z) be a non-constant polynomial with
complex coefficients. Then there exists ζ ∈ C such that P (ζ) = 0.
Consequently, a complex polynomial of degree n > 1 has n roots (counted with
multiplicity).

44
Proof. Suppose, for a contradiction, that p(z) , 0 for all z ∈ C. If p(z) , 0 for all
1 1
z ∈ C, then is holomorphic for all z ∈ C. We will show that is bounded
p(z) p(z)
and the use Liouville’s theorem to conclude that p is constant.
1
As lim p(z) = ∞, there exists R > 0 such that | | < 1 for |z| > R.
z→∞ p(z)
1
(to see this, recall Definition 2.2.1, and in it put ε = 1 and take R = )
δ
1
Hence is bounded on the set {z ∈ C | |z| > R}. We show that it is also bounded
p(z)
on {z ∈ C | |z| ≤ R}. Let w ∈ C with |w| ≤ R, and consider the circle Cr (w)
centred at w with radius r. Choosing r sufficiently large, we can assume that Cr (w)
is contained in {z ∈ C | |z| > R}. Now, we can combine Cauchy’s integral formula
and the Estimation lemma to prove the desired boundedness:
1
1 1 1 1
Z
p(z)
= dz ≤ × 2πr × = 1.
p(w) 2πi Cr (w) z−w 2π r

Note that in the estimate above, we have used the fact that for any z on the circle
1
Cr (w) the function is bounded by 1 (by the fact that Cr (w) ⊂ {z ∈ C | |z| > R})
p(z)
and that |z − w| = r. □

Next, we will prove the much anticipated Taylor’s theorem, which implies that a
holomorphic function is analytic.

Theorem 3.5.2 (Taylor’s Theorem). Suppose that f is holomorphic in the domain

D. Then all of the higher derivatives of f exist in D and, on the disc DR (a) ⊂ D, f
has a Taylor series expansion
∞
f (n) (a)
f (z) = (z − a)n .
X

n=0
n!

Furthermore, if 0 < r < R then

n! f (z)
Z
f (n) (a) = dz. (Cauchy’s Integral Formula for Derivatives)
2πi Cr (a) (z − a)n+1

Proof. Recall that for any w ∈ C we have

1 − wm+1
1 + w + w2 + · · · + wm = .
1−w

h
Put w = to get
z−a
 m+1  m+1
h hm 1− h
z−a 1− h
z−a
1+ + ··· + = = × (z − a).
z−a (z − a)m 1 h
− z−a z−a−h

45
Rearrange this to obtain:

1 1 1 h hm hm+1
= = + +· · ·+ +
z − (a + h) z−a−h z − a (z − a)2 (z − a)m+1 (z − a)m+1 (z − a − h)

Fix h such that 0 < |h| < R and suppose that |h| < r < R. Then Cauchy’s Integral
Formula, together with the identity above, provides:

1 f (z)
Z
f (a + h) = dz
2πi Cr (a) z − (a + h)
1 1
!
hm hm+1
Z
= f (z) + ··· + + dz
2πi Cr (a) z−a (z − a)m+1 (z − a)m+1 (z − a − h)
m
= a n h n + εm ,
X

n=0

where
1 f (z) 1 f (z)hm+1
Z Z
an = dz and εm = .
2πi Cr (a) (z − a)n+1 2πi Cr (a) (z − a)m+1 (z − a − h)

We show that εm → 0 as m → ∞. As f is differentiable on Cr (a), it is bounded

(check this!). Hence, there exists M > 0 such that |f (z)| ≤ M for all z ∈ Cr (a). By
the reverse triangle inequality, using the fact that |h| < r = |z − a| for z on Cr (a), we
have that
|z − a − h| ≥ ||z − a| − |h|| = r − |h|.
Hence, by the Estimation Lemma we get
m
1 M |h|m+1 M |h| |h|

|εm | ≤ 2πr = .
2π r m+1 (r − |h|) r − |h| r

Since |h| < r, this tends to zero as m → ∞. Hence

∞
1 f (z)
Z
f (a + h) = for |h| < R with an =
X
an hn dz
n=0
2πi Cr (a) (z − a)n+1

provided that r satisfies |h| < r < R. However, the integral is unchanged if we vary
r in the whole range 0 < r < R. hence, the formula is valid for all r with 0 < r < R.
Now, putting h = z − a we obtain
∞
f (z) = an (z − a)n
X
|z − a| < R.
n=0

We also know that we can differentiate power series term-by-term, so we get

f (n) (a)
an = .
n!
□

46
Example. We can use Cauchy’s integral formula for derivatives to compute integrals.
For example, note that in the integral

z 2 dz
Z

C3 (0) (z − 2i)2

the denominator is a square. Hence, if we put f (z) = z 2 and a = 2i in Cauchy’s

integral formula for the first derivative, we get:

z 2 dz
Z
= 2πif ′ (a) = 2πi2(2i) = −8π.
C3 (0) (z − 2i)2

Next, we explore another powerful statement in complex analysis about the global
behaviour of holomorphic functions. Before we do this, let us define limit points for
a given set, similar to the real analysis case. A point a ∈ C is a limit point of a
set S if for every ε > 0 we have Dε′ (a) ∩ S , Ø, where Dε′ (a) is the punctured disc
{z ∈ C | 0 < |z − a| < ε}.

Lemma 3.5.3. Suppose the function f is holomorphic on Dr (a) for some r > 0 and
a ∈ C. Then either
(1) f is identically zero in Dr (a), or
(2) the zero set of f at a is isolated, that is, there exists ε > 0 such that the punctured
disc Dε′ (a) contains no zeros of f .

Proof. By Taylor’s Theorem

∞
f (z) = cn (z − a)n (z ∈ Dr (a)).
X

n=0

There are two possibilities. If all cn = 0 then (1) holds. Otherwise, there exists a
smallest integer m > 0 such that cm , 0 and we may write
∞
f (z) = (z − a)m g(z), where g(z) = cm+k (z − a)k .
X

k=0

The series defining g has radius of convergence at least r. Hence, g is continuous on

Dr (a). Because g(a) = cm , 0 and g is continuous at a, there exists some ε > 0 such
that g(z) , 0 for all z ∈ Dr′ (a). Hence, throughout this punctured disc we also have
f (z) , 0. □

Theorem 3.5.4 (Identity Theorem). Let D be a domain and suppose that f is holo-
morphic on D. Assume that the set Z(f ) of zeros of f has a limit point in D. Then
f is identically zero in D. In particular, if f ≡ 0 on some open disc Dr (a) ⊆ D then
f is identically zero on D.

Proof. Let E be the set of limit points of Z(f ). the strategy is to prove two things:
(i) E ⊆ Z(f ),
(ii) E and D\E are both open.

47
As E is nonempty (by assumption) and D is a domain it follows from (ii) that D = E.
Hence, by (i), we conclude that f ≡ 0 on D.
Let a ∈ E. For each n ∈ N there exists an ∈ D′1 (a) such that f (an ) = 0. By
n
continuity of f we have f (a) = 0, so a ∈ Z(f ). This proves E ⊆ Z(f ).
To prove (ii), first let a ∈ E. Then by Lemma 3.5.3 f ≡ 0 in Dr (a) for some r > 0.
But then Dr (a) ⊆ E, so E is open. To show that D\E is open, take a ∈ D\E. Since
a is not a limit point of Z(f ), there exists r > 0 such thatf (z) is never zero on Dr′ (a).
No point of Dr (a) belongs to E, so Dr (a) ⊆ D\E. □

Once again, the condition that D is a domain crucial. For instance, if D is an open set
but not connected, then the theorem may fail. As an example, let D = D1 (−2)∪D1 (2)
and define
(
1 if z ∈ D1 (−2),
f (z) =
0 if z ∈ D1 (2).
Clearly, f is not identically zero in D while every point of D2 (1) is a limit point of
Z(f ).
This theorem has some interesting consequences. For example, if f and g are two
functions that are holomorphic on a domain D, and that f (z) = g(z) for all z in a
set S so that S has a limit point in D, then f ≡ g in D. (can you prove this?)
1 1
For example, suppose that f is an entire function such that f ( ) = sin( ) for all
n n
1 1 1
n = 1, 2, . . . . Then 0 is a limit point of S = {1, , , , · · · } and f (z) = sin z for all
2 3 4
z ∈ S. Therefore, f (z) = sin z for all z ∈ C.
On the other hand, all conditions of the theorem must be satisfied to make such
conclusions. For example, let f be a holomorphic function in C\{0} with f (z) =
1 1 1
sin( ) whenever z = for n = 1, 2, · · · . It does not follow that f (z) = sin( ) in
z nπ z
1 1 1
C\{0}. This is because the only limit point of the set { , , , · · · } is 0, which
π 2π 3π
does not belong to the domain of holomorphy of f (or sin(1/z)).

48
Chapter 4

Residues and Integration

Many basic tools of complex analysis, including differentiation and integration, have
been introduced thus far. The most striking results were derived from Cauchy’s
integration techniques, where differences between complex and real analysis became
apparent. In this chapter we continue building the theory of complex analysis by
further investigating the behaviour of functions that fail to be holomorphic at some
points in the complex plane. At the end, we explore some unexpected applications in
real calculus.

4.1 Laurent Series

We have already seen that a holomorphic function f in a domain D with a ∈ D can

be expressed as a Taylor series:
∞
f (z) = an (z − a)n
X

n=0

for suitable coefficients an , which is valid for z with |z − a| < R, for some R > 0.
Laurent’s theorem provides a neat substitute for Taylor’s theorem for functions that
are holomorphic except at isolated points.
Before we state the theorem, let us explore some examples.
Binomial expansion: We have seen some of this before! Recall that for |z| < 1 we
1
can expand in positive powers of z:
1−z
∞
1
= (|z| < 1).
X
zn
1 − z n=0

However, for |z| > 1 the right hand side no longer converges. Instead, we can do a
1
little trick: note that |z| > 1 if and only if | | < 1 so that
z
∞ −1
1 1 1 1
=− · =− =− (|z| > 1).
X X
zm
1−z z 1− 1
z n=0
z n+1 m=−∞

49
 1
Similarly, we can expand as a series in positive powers of z if |z| < a and as a
z−a
series in negative powers of z if |z| > a.
For example, note that, in the annulus {z ∈ C | 1 < |z| < 3} we have
−1 ∞
4 1 1 zn
= + = zn + (−1)n n+1 .
X X
(1 − z)(3 + z) 1 − z z + 3 n=−∞ n=0
3

4.1.1 Laurent’s expansion

By definition, a series ∞
P∞
n=−∞ an converges (to s = s1 + s2 ) if converges (to
P
n=0 an
s1 ) and ∞ converges (to s2 ).
P
n=1 a−n

If f is holomorphic in a disc Dr (a) except at a, where something nasty happens, then

we cannot hope for a power series expansion
∞
f (z) = cn (z − a)n
X

n=0

since the power series on the right-hand side behaves nicely at a. Motivated by the
examples above we aim for a doubled ended series (Laurent series) so that
∞
f (z) = cn (z − a)n (0 < |z − a| < r).
X

n=−∞

This is done via Lauren’s theorem, which is slightly more general. We will omit
the proof of this theorem, as it requires some results on uniform convergence to
justify interchange of summation and integration. Interested students may consult
any relevant textbooks for the proof.

Theorem 4.1.1 (Laurent’s Theorem). Let f be holomorphic on (the annulus)

A = {z ∈ C | R1 < |z − a| < R2 } with 0 ≤ R1 < R2 ≤ ∞).

Then ∞
f (z) = cn (z − a)n (z ∈ A),
X

n=−∞

where
1 f (w)
Z
cn = dw,
2πi γ (w − a)n+1
with γ = Cr (a) for some (any) R1 < r < R2 .

Remark 4.1.2. Note that we cannot conclude here, as in Taylor’s theorem, that
f (n) (a)
cn = as we do not know if f is differentiable at a, or even defined at a.
n!

50
4.1.2 Computation of Laurent expansion

The following result tells us that the coefficients in the Laurent expansion are uniquely
determined. It may look obvious and rather superficial at first. however, as we will
see it is conceptually more useful that the Laurent theorem itself!
Theorem 4.1.3. Suppose that on the annulus A = {z ∈ C | R1 < |z − a| < R2 },
where 0 ≤ R1 < R2 , we have that
∞ ∞
an (z − a) = bn (z − a)n
X X
n

n=−∞ n=−∞

for all z ∈ A. Then an = bn for all n ∈ Z.

Proof. Let m ∈ Z be an arbitrary integer. Then

∞ ∞
an (z − a)n−m+1 = bn (z − a)n−m+1
X X

n=−∞ n=−∞

for all z ∈ A. Let R1 < r < R2 . Then integrating over Cr (a) we have
∞ Z Z ∞
2πiam = (z − a) dz = an (z − a)n−m−1 dz
X X
n−m−1
an
n=−∞ Cr (a) Cr (a) n=−∞
Z ∞
= bn (z − a)n−m−1 dz
X
Cr (a) n=−∞
∞ Z
= (z − a)n−m−1 dz
X
bn
n=−∞ Cr (a)

= 2πibm .

Therefore am = bm for all m ∈ Z so our proof is complete. In the above we used

two facts. The first is that we can integrate termwise1 . The second is the following
formula which you should check
(
2πi if n − m + 1 = −1;
Z
(z − a) n−m−1
dz =
Cr (a) 0 otherwise.

We are interested in calculating the Laurent expansion for a given function f . If

we were to use Laurent’s theorem directly then we would have had to compute the
coefficients directly from the integral
1 f (w)
Z
cn = dw
2πi γ (w − a)n+1
which is tedious, if not impossible. In practice, we combine simple known expansions
(Taylor or Laurent) to obtain Laurent expansions of more complicated functions.
1
This is something we have not formally justified

51
Then the theorem above guarantees that what we have calculated is indeed the Lau-
rent expansion, as it is unique. As we will see later, the coefficient c−1 in the Laurent
expansion is of special significance.
A key example that we often use is the binomial expansion:
 ∞
|z| < 1,
X
zn,





1 n=0


=
1−z  X 1 ∞
|z| > 1.

− ,


zn


n=1

1
Example 1. Let us consider the Laurent expansion of the function f (z) =about
z
z = −2. The Taylor expansion of f about z = −2, valid for |z + 2| < 2, can be
computed using binomial expansion as follows:
∞ ∞
1 −1 1 −1 X z+2 n (z + 2)n
= = ( ) =− (|z + 2| < 2).
X
·
z 2 1− z+2 2 n=0 2 n=0
2n+1
2

To find a Laurent series valid for |z + 2| > 2 put z + 2 = w and then from the binomial
expansion we get
∞ ∞ ∞
1 1 1 1 2 2n−1 2n−1
"  n #
f (z) = = =− = = =
X X X
− −
z w−2 2(1 − 2 w)
1 2 n=1
w n=1
wn n=1
(z + 2)n
w
which converges for | | > 1, that is |z + 2| > 2 as required.
2
1
Example 2. f (z) = around 1. Note that f is holomorphic in A1 = {z ∈
z(1 − z)2
C | 0 < |z − 1| < 1} and in A2 = {z ∈ C | |z − 1| > 1}.
For z ∈ A1 the binomial expansion gives
1 1 1
   
f (z) = = 1 − (z − 1) + (z − 1) 2
− · · · .
(z − 1)2 1 + (z − 1) (z − 1)2
Hence, we have
∞
f (z) = (−1)n (z − 1)n (0 < |z − 1| < 1).
X

−2

On the other hand, for |z − 1| > 1 we write w = z − 1 to ease the notation (hence
1 1
z = 1 − (−w)), so that f can be written as · 2 , consider it expansion at
1 − (−w) w
1 1
w = 0. The expression 2 is already a Laurent series! And for we use
w 1 − (−w)
binomial expansion to get:
∞ 3
1 1 1 1
!
= 2 = (−1)n+1 wn
X X
· −
w2 1 − (−w) w n=1
(−w) n
n=−∞

52
Hence, we conclude that
3
f (z) = (−1)n+1 (z − 1)n (|z − 1| > 1).
X

n=−∞

53
 3
Example 3. f (z) = about z = 0.
(1 + z)(2 − z)
There are different Laurent expansions for f
in each open annulus around z = 0 for which
f is holomorphic and for which there is a
point of non-holomorphy on the boundary.
The points at which f is not holomorphic are
|z| < 1
z = −1, 2, which implies there are three an-
nular regions; hence, there are separate series −1 0 2
for f in |z| < 1, 1 < |z| < 2, |z| > 2. By
partial fractions 1 < |z| < 2
3 1 1
f (z) = = +
(1 + z)(2 − z) 1+z 2−z |z| > 2
and then using binomial expansion, the individual terms have series
 ∞ ∞
(−z)n = (−1)n z n , |z| < 1,
 X X



1 1

n=0 n=0

= =
1+z 1 − (−z)  ∞ 
1
 n ∞
(−1)n+1
= |z| > 1,
X X
− − ,


zn


n=1
z n=1

and ∞  n ∞
1X zn

z
= |z| < 2,
X
,


 2 n=0 2 2n+1


1 1

n=0
= =
2−z 2(1 − z/2)  ∞  n
1X 2 ∞
2n−1
= |z| > 2.
 X
− − ,


2 n=1 z n


n=1
z
The series for f valid for |z| < 1 is obtained from the series expansions of 1/(1 + z)
and 1/(2 − z) that are valid in that region, that is the first series in each case. It
doesn’t matter that the second series is valid in a larger region; the aim is to select
series that are both valid in |z| < 1. With each region treated in a similar way it is
found that ∞ 
1

(−1) + n+1 z n , |z| < 1,
X

 n
2



n=0



∞ ∞
(−1)n+1 X

zn

X
f (z) = + , 1 < |z| < 2,


 n=1
zn n=0
2n+1


∞
(−1)n+1 − 2n−1



|z| > 2.
X
,




zn
n=1

54
 3
Example 4. Find all series expansions of f (z) = about z = i.
z(z − 3i)
The function f is not holomorphic at z = 0, 3i. 3i
The distance from the expansion point z = i to 1 < |z − i| < 2
the nearest one at z = 0 is 1, while the distance
to the second point at z = 3i is 2. Thus, it is
necessary to consider the three regions |z − i| < 1
A1 : {|z − i| < 1}, A2 : {1 < |z − i| < 2}, i
A3 : {|z − i| > 2} separately.
Put w = z − i so that
3 3 0
f (z) = =
z(z − 3i) (w + i)(w − 2i)
1

1 1
 |z − i| > 2
= − .
i w − 2i w + i
The individual terms have series
∞ 
i X iw n
 
|w| < 2,


− ,
 2 n=0 2



1 1 i/2

= = =
w − 2i −2i(1 − w/(2i)) 1 − (−iw/2)) 
i X∞ 
2 n

|w| > 2,


−
 2 − ,
iw


n=1

and  ∞
(iw)n , |w| < 1,
 X



 −i
1 1 −i n=0


= = =
w+i i(1 + w/i) 1 − iw  ∞ 
1
n
 i , |w| > 1.
  X
iw


n=1

The appropriate combinations of series for each of the three annular regions are:

|z − i| < 1,
∞ ∞
1X i 1 1
  n   n+1 !
f (z) = + i (iw) = 1− − in (z − i)n ;
X
n
−
i n=0 2 2 n=0
2

1 < |z − i| < 2,
∞ ∞  ∞ ∞
1 i X 1 n 1X i (−i)n
" n  # n
−iw
 
f (z) = −i = (z−i)n − ;
X X
−
i 2 n=0 2 n=1
iw 2 n=0 2 n=1
(z − i)n

|z − i| > 2,
∞ ∞ 
1X i 1  (−i)n
  n
f (z) = − (−2)n − i = (−2)n−1 − 1
X
.
i n=1 2 iw n=1
(z − i)n

Note that the final expressions may be written in a number of equally-satisfactory

ways.

55
4.2 Singularities

For a function f , a point a ∈ C is called a regular point of f if the function is

holomorphic at that point. In other words, if there exists r > 0 such that f is
differentiable at every z ∈ Dr (a). A point a ∈ C is called a singularity of f if a is a
limit point of regular points but is not itself regular. If a is a singularity of f while f
is holomorphic in some punctured disc Dr′ (a), for some r > 0, then a is an isolated
singularity; otherwise a is an non-isolated singularity. Non-isolated singularities
(also called essential singularities) are not of interest in this course. This is mainly
because the existence (and uniqueness) of Laurent expansion for a function near a
point relies on the function being holomorphic in some punctured disc around the
point.
Here, we discuss different types of isolated singularities, their behaviour and how to
characterise them.
1
For example, the function f (z) = is not defined at 0, hence cannot be differentiated
z
at 0 and is not holomorphic at that point. So 0 is a singularity of f . However, f is
holomorphic at all nonzero points, hence 0 is an isolated singularity of f .
Classification of isolated singularities. Suppose that f has an isolated singularity
at a. Then f is holomorphic in some annulus {z ∈ C | 0 < |z − a| < r}, and therefore
has a unique Laurent expansion
∞
f (z) = cn (z − a)n .
X

n=−∞

We may write this as

−1 ∞
f (z) = cn (z − a)n + cn (z − a)n .
X X

n=−∞ n=0

The second term on the right-hand side is holomorphic in Dr (a) and it cannot be
responsible for the singularity at a. Hence, the singularity is caused by the first sum,
−1
cn (z − a)n
X

n=−∞

which is known as the principal part of the Laurent expansion. Isolated singularities
are classified into three types according to the behaviour of the coefficients cn for
n < 0. The point a is said to be

• a removable singularity if cn = 0 for all n < 0;

sin z
Example: f (z) = , (z , 0) has a removable singularity as
z
sin z z2 z4
=1− + − ··· .
z 3! 5!
Indeed, we can define f (0) = 1 to be compatible with the expansion above (so that
we superficially “remove” that singularity).

56
• a pole of order m (m ≥ 1) if c−m , 0 and cn = 0 for all n < −m;
sin z
Example: f (z) = 3 , (z , 0) has a pole of order 2 at z = 0, as
z
sin z 1 1 z2
= − + − ··· .
z3 z 2 3! 5!

• an isolated essential singularity if there does not exist m such that cn = 0 for
1
all n < −m. Example: f (z) = sin , (z , 0) has an isolated essential singularity at
z
0 as
1 1 1 1
sin = − + − ··· .
z z 3!z 3 5!z 5
Poles of order 1, 2, 3, . . . are called simple, double, triple, . . . .
Isolated essential singularities are difficult to deal with and their study is beyond
this course.

Definition 4.2.1. Let D ⊆ C be a domain. A function f : D → C is said to be

meromorphic if it is holomorphic in D except at finitely many poles or removable
singularities.

Recall that a function, defined on a domain D, is said to have a zero at a ∈ D if

f (a) = 0. A zero a is called isolated if there exists ε > 0 such that f (z) , 0 for all z
with 0 < |z − a| < ε. Suppose that f is holomorphic at a. Then on some disc around
a we have a Taylor expansion
∞
f (z) = an (z − a)n .
X

n=0

We say that f has a zero of order m at a if a0 = a1 = · · · = am−1 = 0 but am , 0.

We say that f has a simple zero (double, triple,...) at a if m = 1 (m = 2, 3, . . . ).

Theorem 4.2.2. Suppose that f is holomorphic in some open disc Dr (a). Then f
1
has an isolated zero of order m at a if and only if has a pole of order m at a.
f

This theorem is intuitively plausible, but its proof requires some (minor) technical
tools, which are not covered in this text. However, it is useful in computations as
writing down the Laurent expansion to find order of poles can be tedious.
We end this section by a discussion on a remarkable result on the Riemann sphere
(this is optional). The definition of meromorphic functions, and singularities and so
on, can be extended to the Riemann sphere. As we have already seen functions with
poles can be complicated in nature. However, one can prove that any meromorphic
function on the Riemann sphere is a rational function. Similarly, any holomorphic
function on the Riemann sphere is a polynomial function.

57
4.3 Cauchy’s Residue Theorem

Definition 4.3.1. Suppose that f is holomorphic on a domain D except for a pole at

a ∈ D. The residue of f at a is defined to be the (unique) coefficient c−1 of (z − a)−1
in the Laurent expansion of f about a. We denote it by Res(f, a).

Suppose that R > 0 so that we write the Laurent expansion of f in the annulus
0 < |z − a| < R as
∞
f (z) = cn (z − a)n .
X

n=−∞

Then by Laurent’s theorem, for 0 < r < R, we have

1
Z
Res(f, a) = f (z)dz
2πi Cr (a)

where the Cr (a) is the circular anticlockwise path centred at a with radius r. One
observation is that residues are related to integration.
Recall that a contour γ is said to be a simple closed loop if for every point z not
on γ the winding number is either w(γ, z) = 0 (when z is outside γ) and w(γ, z) = 1
(when z is inside γ). In other words, a simple closed loop is a loop that goes round
anticlockwise once and never intersects itself.

Theorem 4.3.2 (Cauchy’s Residue Theorem). Let D be a domain containing a simple

closed loop γ and the points inside γ. Suppose that f is holomorphic on D except at
finitely many poles a1 , a2 , . . . , am inside γ. Then
Z m
f (z)dz = 2πi Res(f, ai ).
X
γ i=1

Proof. Since singularities are only poles (in particular isolated) for each i = 1, 2, · · · , m
we can find a positively oriented circle Ci = Cri (ai ), for some ri > 0, entirely con-
tained inside γ (hence D) such that the inside of Ci contains no singularity other
than ai . With this construction, w(Ci , zj ) is zero if i , j and it is 1 if i = j. We also
know that w(γ, zi ) = 1 for each i. Hence, −γ and C1 , · · · , Cm satisfy the conditions
of Cauchy’s theorem (note that f is holomorphic in the domain γ minus Ci s and their
interiors). In other words, we have
Z Z Z Z Z Z
f+ f + ··· + f =0 which implies f= f + ··· + f.
−γ C1 Cm γ C1 Cm

On the other hand, it follows from Laurent’s theorem (applied to f in each annulus
0 < |z − ai | < ri ) that
1
Z
Res(f, ai ) = f (z)dz.
2πi Ci
Hence, Z
f = 2πi (Res(f, a1 ) + · · · + Res(f, am )) .
γ
□

58
Calculating residues by computing the entire Laurent expansion may be tedious,
although sometimes necessary. Often we are able to calculate the residues with simple
manipulations. For example:

Lemma 4.3.3. If f has a simple pole at a ∈ C then

Res(f, a) = lim (z − a)f (z).

z→a

Proof. If f has a simple pole at a then its Laurent series looks like
∞
c−1
f (z) = + cn (z − a)n (0 < |z − a| < R).
X
z − a n=0

Hence,
∞
(z − a)f (z) = c−1 + cn (z − a)n+1 =⇒ lim (z − a)f (z) = c−1 .
X
z→a
n=0

□
p(z)
Lemma 4.3.4. If f (z) = where p, q are differentiable, p(a) , 0, q(a) = 0 but
q(z)
q ′ (a) , 0, then
p(a)
Res(f, a) = .
q ′ (a)

Proof. The hypothesis implies that f has a simple pole at a. Then by Lemma 4.3.3
and the fact that q(a) = 0, the residue is calculated as

(z − a)p(z) p(z) p(a)

Res(f, a) = lim = lim = .
z→a q(z) z→a q(z)−q(a) q ′ (a)
z−a

□
cos πz
Example. Consider the function . It has a simple pole at z = 1, and satisfies
z2 − 1
the assumptions in the lemma above. Hence,
cos π −1
Res(f, 1) = = .
2(1) 2

We can produce similar results for poles of order m:

Lemma 4.3.5. Suppose that f has a pole of order m at a. Then

1 dm−1
Res(f, a) = lim ((z − a)m f (z)).
z→a (m − 1)! dz m−1

59
Proof. If f has a pole of order m at a then its Laurent expansion looks like
∞
c−m c−1
f (z) = + + + cn (z − a)n (0 < |z − a| < R, for some R > 0).
X
· · ·
(z − a)m z − a n=0

Multiplying tis with (z − a)m gives

∞
(z − a)m f (z) = c−m + (z − a)cm−1 + · · · + (z − a)m−1 c−1 + cn (z − a)n+m .
X

n=0

Differentiating both sides m − 1 times gives

∞
dm−1 (m + n)!
((z (z)) = (m 1)!c + cn (z − a)n+1 .
X
m
− a) f − −1
dz m−1
n=0
(n + 1)!

Now, dividing by (m − 1)! and taking the limit z → a completes the proof. □

z 4 + 2z + 1
Example. The function f (z) = has a triple pole at z = 2. Using
(z − 2)3
(z − 2)3 f (z) = z 4 + 2z + 1 and the lemma above, we can calculate

1 d2 4 1
Res(f, 2) = lim (z + 2z + 1) = lim (12z 2 ) = 24.
z→2 2 dz 2 z→2 2

1
Example. Let us calculate residues of the function f (z) = 2 . This function
z sin z
has singularities whenever the denominator is zero, that is when z = kπ for k ∈ Z.
Amongst those z = 0 is special as it contributes more zeros! Let us calculate the
residue at zero first, using Laurent expansion. Using the power series expansion of
sin z we can write
!−1
1 1 1 z2
= ! = 1− + ···
z 2 sin z z 3 z 3 6
z2 z− + ···
6
1 1 1
!
z2
= 3 1+ + ··· = + + ··· .
z 6 z 3 6z

Note that in computations we omit higher order terms, and especially those that do
1 1
not contribute to the coefficient of . This shows that Res(f, 0) = .
z 6
Calculating the residue at other poles (kπ for k , 0) is much easier, as any of these
points is a simple pole of f (z) so we can use one of the formulae above. To do this,
p(z)
set f (z) = with p(z) = 1 and q(z) = z 2 sin z so that q ′ (z) = z 2 cos z + 2z sin z,
q(z)
and
1 (−1)k
Res(f, kπ) = ′ = .
q (kπ) (kπ)2

60
4.4 Applications of Cauchy’s Residue Theorem

In this section we explore some applications of Cauchy’s residue theorem in real

calculus. All applications are illustrated by examples.

4.4.1 Integrals round the circle

Consider the (real) integral

Z 2π
dt
0 1 + 8 cos2 t
We can use existing techniques of parametrisation to convert this into a complex
integral which we can easily solve using residues. To do this, first parametrise the
unit circle by z = γ(t) = eit with 0 ≤ t ≤ 2π. Then, on this contour, we have

dz dz
= γ ′ (t) = ieit = iz =⇒ dt =
dt iz
and
1 1 1
cos t = (eit + e−it ) = (z + ).
2 2 z
Substituting these into the integral above, we get
Z 2π
dt dz
Z
=
0 1 + 8 cos2 t γ iz(1 + 2(z 2 + 2 + z −2 ))

1 zdz zdz
Z Z
= = −i .
i γ 2z + 5z + 2
4 2
γ (2z + 1)(z + 2)
2 2

±i √
The function has four simple poles: √ inside γ, and ±i 2 outside γ. Those poles
2
located outside the contour do not contribute to the integral. Hence

i
Z 2π
dt z z −i
 
= −i × 2πi Res( , √ ) + Res( , √ )
0 1 + 8 cos t
2 (2z + 1)(z + 2)
2 2
2 (2z + 1)(z + 2)
2 2
2
 
   

= 2π 
z
+
z  = 2π .
 
3
  
 2(z + √i )(z 2
2
+ 2) 2(z − √i
2
)(z 2 + 2)
−i

z= √i z= √
2 2

Various integrals of the form

Z 2π
F (cos t, sin t)dt
0

can be computed using this method.

61
4.4.2 Infinite real integrals

The overall aim in this subsection is to compute real integrals of the form
Z ∞
f (x)dx
−∞

or simple variations of those. We begin by considering a simple example:

Z ∞
dx
I: = .
0 1 + x4
First, we consider this integral as
Z R
dx
I = lim .
R→∞ 0 1 + x4
Then, we use the fact that the integrand is an even function, so that
Z R Z R
dx dx
2 = .
0 1 + x4 −R 1 + x4
Now, let γ be the semicircular contour that consists of the (real) line segment [−R, R]
and the upper semicircle centred at 0 with radius R, going anticlockwise:

3πi πi
e 4 e4

−R R

Because of the choice of γ we made, we have

iReiθ dθ
Z R Z π
dz dx
Z
= + .
γ 1 + z4 −R 1 + x4 0 1 + R4 e4iθ
As R → ∞, the first term on the right-hand side tends to 2I. We will show that
the second term on the right-hand side tends to zero. Hence 2I is the value of the
complex integral on the left-hand side, which we will calculate using Cauchy’s residue
theorem.
Using the estimation lemme, together with the reverse triangle inequality and the
fact that the length of the semicircle is πR, we obtain the following bound

iReiθ dθ πR2
Z π
≤
0 1 + R4 e4iθ |R4 − 1|

62
But the right-hand side tends to 0 as R → ∞.
dz
Z
Let us now calculate the value of the integral using residues. Note that the
γ 1 + z4
1
function is holomorphic except for simple poles at the zeros of 1 + z 4 , which
1 + z4
are (2k+1)πi
e 4 (k = 0, 1, 2, 3).
iπ 3πi
For large enough R, only z1 = e 4 = 1+i √ and z2 = e 4 = −1+i
2
√
2
lie inside γ. Hence,
1
for large enough R, the function is holomorphic inside and on γ except at z1
1 + z4
and z2 . We can also calculate the residues at these points:
1 −(1 + i) 1 1−i
Res( , z1 ) = √ and Res( , z2 ) = √ .
1+z 4
4 2 1+z 4
4 2
Hence, by Cauchy’s residue theorem, we have
dz 2πi π
Z
= √ (−(1 + i) + (1 − i)) = √ .
γ 1 + z4 4 2 2
We conclude that Z ∞
dx π
= √ .
0 1 + x4 2 2
Key elements in the strategy above:

• we relateR the required integral I, or an approximation of it, to some contour

integral γ f ;

• we must be able to apply Cauchy’s residue theorem, so that f must have at

most finitely many poles inside the contour γ and none on it;

• the contour γ is chosen so that the integral of f along each portion of it either
contributes to the integral we want, or can be handled by estimation or in some
other way.

Optional Discussion: It may be that the function has a pole on the real line, in
which case the method above does not apply. However, this can be overcome by
“walking around” that point as in the picture below, having control over the integral
(it being very small for instance) on the smaller semicircle. Interested readers may
find various examples and their treatments in usual textbooks.

63
4.4.3 Summation of series
∞
1
Consider the sum . We can calculate the exact value of this infinite sum, using
X

n=1
n2
1
Cauchy’s residue theorem. For this, let f (z) = and consider the function
z2
cot πz cos πz
f (z) cot πz = = 2 ,
z 2 z sin πz

This function has a simple pole for each z = n, n ∈ Z and n , 0, and has a triple pole
at z = 0. Let us calculate the residues. At n , 0, we can use Lemma 4.3.4 to get
cot πz cos πn 1
Res( , n) = = .
z2 πn2 cos πn + 2n sin πn πn2
Now consider the pole at 0. Using the expansion for cot:

1 z z3 2z 5
cot z = − − − − ···
z 3 45 945
we get
cot πz 1 π π 3 z 2π 5 z 3
= − − − − ···
z2 πz 3 3z 45 945
which concludes that the residue at 0 is 3 .
−π

Now, fix m ∈ N and consider the square shaped contour γm in the complex plane,
going anti-clockwise, with vertices at
1 1 1 1
(m + ) − i(m + ) , (m + ) + i(m + )
2 2 2 2
1 1 1 1
−(m + ) + i(m + ) , −(m + ) − i(m + )
2 2 2 2
Each side of this square has length 2m + 1.

−(m + 1) 0 m+1
−m −1 1 m

64
Let us now bound cot πz on γm : write z = x + iy. On the horizontal sides of γm ,
|y| > 1/2, hence

eiπz + e−iπz |eiπz | + |e−iπz | e−πy + eπy π

| cot πz| = ≤ ≤ ≤ coth |πy| ≤ coth .
iπz
e −e −iπz iπz
|e | − |e −iπz | e −πy −e πy 2

On the vertical side of γm , we have z = ±(m + 12 ) + iy. Hence,

eiπz + e−iπz e2iπz + 1 eiπ−2πy + 1 −e−2iπy + 1 1 − e−2πy

| cot πz| = = = = = ≤ 1.
eiπz − e−iπz e2iπz − 1 eiπ−2πy − 1 −e−2iπy − 1 1 + e2πy
These two inequalities show that | cot πz| ≤ max{1, coth π/2} ≤ 2 for all z ∈ γm .
Now, by the Estimation Lemma, we have

cot πz 2 8(2m + 1)
Z
2
dz ≤ 2 length(γm ) =
γm z m m2

which tends to zero as m → ∞, which shows that

cot πz
Z
lim dz = 0.
m→∞ γm z2

On the other hand, by Cauchy’s residue theorem we have

m
cot πz cot πz
Z
2πi Res( , n) =
X
dz
n=−m
z2 γm z2

Hence,
m
cot πz
lim Res( , n) = 0.
X
m→∞
n=−m
z2
Also, from the computation of the residues above we get
m −1 m
cot πz cot πz cot πz cot πz
Res( = Res( + Res( 0) + Res( 2 , n)
X X X
2
, n) 2
, n) 2
,
n=−m
z n=−m
z z n=1
z
m
1 π
2
X
− .
n=1
πn 2 3

Combining all these concludes that

∞
1 π2
=
X
.
n=1
n2 6

65