Introduction to Complex Analysis

Draft, containing only the first part up to the chapter

on Meromorphic functions

Gestur Ólafsson

Fall 2025
2
Contents

1 Introduction 7

1.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 The topology of the complex plane 9

2.1 Complex numbers and the plane . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.2 The topology of C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.3 Convergence: Sequences and series . . . . . . . . . . . . . . . . . . . . . . . 12

2.3.1 Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.3.2 Compact sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.4 Continuous functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3 Differentiable functions 17

3.1 Differentiable functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.2 Complex differentiable functions . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.3 Real and complex differentiable functions: The Cauchy-Riemann Differential

equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.3.1 Intermezzo: Real linear and complex linear maps on Rn . . . . . . . . 23

3.4 Power series and their properties . . . . . . . . . . . . . . . . . . . . . . . . 24

3.4.1 The geometric series . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.4.2 Power series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3
 3.5 The exponential function and and other well known functions . . . . . . . . 27

3.5.1 The exponential function . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.5.2 The trigonometric functions . . . . . . . . . . . . . . . . . . . . . . . 27

3.5.3 The log function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4 The Cauchy Theorem, holomorphic functions are analytic 31

4.1 Integration along curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.2 Goursat-Pringsheim Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 36

4.3 Cauchy theorem and its applications . . . . . . . . . . . . . . . . . . . . . . 38

4.4 Some application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.4.1 Liouville’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.4.2 The fundamental theorem of algebra . . . . . . . . . . . . . . . . . . 44

4.4.3 The maximum principle . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.5 The winding number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.6 Sequences of holomorphic functions . . . . . . . . . . . . . . . . . . . . . . . 46

4.7 Schwarz reflection principle . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5 Meromorphic functions 51

5.1 Zeros and poles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.2 Singularities and meromorphic functions . . . . . . . . . . . . . . . . . . . . 51

5.3 Application to evaluation of integrals . . . . . . . . . . . . . . . . . . . . . . 51

5.4 Runge’s approximation theorem . . . . . . . . . . . . . . . . . . . . . . . . . 52

6 Open mapping theorem 53

7 Analytic continuation 55

7.1 examplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4
 7.2 Analytic continuation in general . . . . . . . . . . . . . . . . . . . . . . . . . 55

7.3 Further topics: Remarks on Riemannian surfaces . . . . . . . . . . . . . . . 55

8 Weierstrass Theorem 57

9 Applications: The Fourier transform 59

9.0.1 The Paley-Wiener Theorem . . . . . . . . . . . . . . . . . . . . . . . 59

10 Special functions 61

10.1 The Mellin transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

10.2 The Riemann ζ-function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

10.3 Elliptic functions: The ℘-function . . . . . . . . . . . . . . . . . . . . . . . . 61

10.4 The Hardy space and spectral Theory . . . . . . . . . . . . . . . . . . . . . . 61

5
6
 Chapter 1

Introduction
{SIntro}

1.1 Notation

7
8
 Chapter 2

The topology of the complex plane

2.1 Complex numbers and the plane

As a set, and topologically, we have C = R2 = {(x, y) | x, y ∈ R}. We identify the real line R
with the line Re1 , where e1 = (1, 0). We set i = e2 = (0, 1). We can then write any element
in C as z = x + iy ↔ (x, y) with x, y ∈ R. We call x = Rez, the real part of z and y = Imz
the imaginary part of z.

We define the following algebraic operations for z = x + iy and w = r + it

Add: z + w = (x + r) + i(y + t);

Mult: zw = (xr − yt) + i(xt + ry).

In particular the addition is just the usual addition in R2 . We also define the complex
em:FirstFP} conjugate of z by z̄ = x − iy. We note the following simple consequences of those definitions:
Lemma 2.1.1. Let z = x + iy, w = r + it ∈ C. Then the following holds:

1) i2 = −1.

2) If r ∈ R then rz = (rx) + (ry)i, the usual multiplication of the vector (x, y) by the real
number r.

3) z z̄ = x2 + y 2 = |z|2 is the length of the vector (x, y).

4) If z ̸= 0 then
z
z· = 1.
|z|2

9
 5) |z + w| ≤ |z| + |w|.
1 1
6) Rez = (z + z) and Imz = 2i
(z − z)
2
7) |Rez|, |Imz| ≤ |z|.

It is now a simple exercise to show that C with the above defined addition and multipli-
cation is a field with R as a subfield.

If z ̸= 0 then |z/|z|| = 1 and hence there exists θ ∈ R such that

z
= cos(θ) + i sin(θ)
|z|

and
{eq:polar} z = r(cos(θ) + i sin(θ)), r = |z|. (2.1.1)
The coordinates in (2.1.1) are called the polar coordinates and they corresponds to the usual
coordinates (x, y) = r(cos(θ), sin(θ)). The “function” z 7→ θ is denoted by arg(z). Note that
this is only well defined as a function C∗ \ {0} → R/2πZ or one can remove the half-line
(−∞, 0] and then fix the value by setting arg(1) = 0. This is a simple example of what is
called a multi-valued function.

The multiplication of two complex numbers is easy to understand using polar-coordinates:

Lemma 2.1.2. Let z = r(cos(θ) + i sin(θ)), w = s(cos(ψ) + i sin(ψ)). Then

zw = (rs)(cos(θ + ψ) + i sin(θ + ψ)).

Thus
|zw| = |z||w| and arg(zw) = arg(z) + arg(w) mod 2π.

2.2 The topology of C

In this section we give a brief overview of the topology of the complex plane. We assume
that the reader is familiar with basic point set topology and the topology of metric spaces.

For z = x + iy, w = r + it ∈ C we have

p
|z − w| = (x − r)2 + (y − s)2

the usual distance between the vectors z and w in the identification of C as R2 . Furthermore,
if z, v, w ∈ C then
|z − w| ≤ |z − v| + |v − w|.

10
 We define for z0 ∈ C and r > 0 the the ball with center z0 and radius r by

Dr (z0 ) = {z ∈ C | |z − z0 | < r}.

The closed ball is defined by

Dr (z0 ) = {z ∈ C | |z − z0 | ≤ r}.

Definition 2.2.1. 1) A subset U ⊂ C is open if for every z ∈ U there exists r > 0 such that
Dr (z) ⊂ U . A set A ⊂ C is closed if C \ A = Ac = {z ∈ C | z ̸∈ A} is open.

2) If U ⊂ C is a set. The set

U ◦ = {z ∈ U | (∃r > 0)Dr (z) ⊂ U }

is the interior of U and cl(U ) = ((U c )◦ )c is the closure of U (in C).

3) If Ω ⊂ C then a set U ⊂ Ω is open in Ω if there exists an open set V ⊂ C such that

U = V ∩ Ω. A set W ⊂ Ω is closed in Ω if Ω \ W is open in Ω.

Remark 2.2.2. If z = 0 then we simply write Dr for Dr (0).

Example 2.2.3. 1) The empty set ∅ and C are the only sets in C that are both open and
closed.

2) If z0 ∈ C and r > 0, then the disc Dr (z0 ) is open: Let z ∈ Dr (z0 ). Let 0 < s ≤ r − |z − z0 |.
L If w ∈ Ds (z) then

|w − z0 | ≤ |w − z| + |z − z0 | < (r − |z − z0 |) + |z − z0 | = r

{lem:Union} Thus Ds (z) ⊂ Dr (z0 ).

S
Lemma 2.2.4. 1) Let {Uj }j be a collection of open sets. Then j Uj is open.
T
2) Let {Aj } be a collection of closed set, then Aj is closed.

S
Proof. 1) Let W = j Uj . Suppose z ∈ W . Then there exists j such that z ∈ Uj . As Uj is
open it follows that there exist r > 0 such that Dr (z) ⊂ Uj . But then Dr (z) ⊂ W and W is
open.

2) This follows from (1) and the definition because

\ c [
Aj = Acj .
j

11
 Lemma 2.2.5. Let U ⊂ C. Then the following holds:
[
1) U ◦ = W is open and in fact maximal open subset contained in U .
W ⊆U W open

\
2) cl(U ) = A and cl(U ) is the smallest closed set containing U .
U ⊂A,A closed

Proof. We will only prove (1) as the second part follows by taking complements. S For the
moment let U be the collection
S of all open sets contained in U . Thus V = W ⊆U W open W
can be written as V = W ∈U W . It follows by Lemma 2.2.4 that V is open. It is contained in
U by definition. If z ∈ U and Dr (z) ⊂ U , then, as Dr (z) is open, it follows that Dr (z) ⊂ V .
Hence U ◦ ⊂ V . Let z ∈ V . Then, as V is open it exists r > 0 such that Dr (z) ⊂ V ⊂ U .
Hence V ⊂ U ◦ . It follows that V = U o . If W ⊂ U is open, then W ∈ U so by what we just
showed it follows that W ⊂ U ◦ so U ◦ is in fact the maximal open set contained in U .
Definition 2.2.6. Let U ⊂ C. The set ∂U = cl(U ) \ U ◦ is the boundary of U .
Example 2.2.7. 1) We have (Q + iQ)◦ and cl(()Q + iQ) = C.

2) If U is open then U ◦ = U and if A is closed then Let r > 0 and z ∈ C then

{eq: Sr} ∂Dr (z) = ∂Dr (z) := Sr (z) = {z ∈ C | |z| = r}. (2.2.1)

2.3 Convergence: Sequences and series

Definition 2.3.1. A sequence is a map a : N → C. We write aj for a(j) and we write

a = (aj )j∈N = (aj )∞
j=1

Remark 2.3.2. The above definition works for arbitrary index set I and sometimes it is
helpful to allow more general, but still countable infinite, sets. Every countable set in
isomorphic to N, but that is not always useful. For example for In particular if I = N
then the sequence (aj )j∈I has a “beginning” element a1 where as in the case I = Z there is
no such element, so there is no “natural” choice of z0 . One could number the sequence as
a0 , a1 , a−1 , a2 , a−2 etc or a2 , a2 , a3 , a1 , a0 , a−1 , a4 , a5 , a6 , a−2 , a−3 , a−5 etc.
Definition 2.3.3. Let (zj ) be a sequence and z ∈ C.

(1) We say that (zj ) converges to z, or zj → z if for all ϵ > 0 there exits a n = n(ϵ) ∈ N such
|z − zj | < ϵ for all j ≥ n.

(2) We say that (zk j) is a Cauchy sequence if for all ϵ > 0 there exists n = n(ϵ) ∈ N such
that |zj − zk | < ϵ for all j, k ≥ n.

12
Lemma 2.3.4. Let (zj ) be a sequence. Then (zj ) is convergent if and only if the real
sequences (Rezj ) and (Imzj ) are convergent and in that case we have

lim zj = lim Rezj + i lim Imzj .

Theorem 2.3.5. Let (zj ) be a sequence. Then (zj ) is convergent if and only if (zj ) is a
Cauchy sequence.

Lemma 2.3.6. Let U be a non-empty subset of C. Then z ∈ ∂U if and only if there exists
a sequence (zj ), zj ∈ U such that zj → z

2.3.1 Series
P∞
Let {an } be a sequence of complex numbers. Then we say that n+0 an is a series of
complex numbers or simply a series. To associate a numerical value to the series, if possible,
we consider the finite sum n
X
sn = ak .
k=0
P∞
We say that the series n+0 an exist and equals s if lim sn exists and lim sn = s.

2.3.2 Compact sets

Let U ⊂ CSand let {Uj } be a collection of subsets of C. We say that {Uj }j is a covering of
U if U ⊂ j Uj . An open covering Is a covering {Uj } such that each Uj is open. Finally if
{Uj } is a covering of U , then {Vk }k is a sub-covering if

Definition 2.3.7. 1) A set U ⊂ C is bounded if there exists r > 0 such that U ⊂ Dr (0).

2) A set K ⊂ C is compact if for every open covering U = {Uj }j of K there exists a finite
sub-covering.

Theorem 2.3.8. Let K ⊂ C. Then the following are equivalent:

1) K is compact.

2) K is closed and bounded.

3) For every sequence {zj } in K there exits a convergent subsequence.

S
Proof. “(1) ⇒ (2)” Let Un = Dn , n ∈ N. Then K ⊂ Un . As K is compact and the
sequence {Un }n is increasing, there exists n such that K ⊂ Un . Hence K is bounded.

13
 Suppose that z ∈ cl(K) \ K. For n ∈ N
T S let Un = C \ D1/n (z). Then Un is open and, as
n D1/n (z) = {z} it follows that K ⊂ n Un . As {Un } is an increasing sequence and K
is compact there exists n such that K ⊂ Un . But then K ∩ D1/n (z) = ∅ contradicting the
assumption that z ∈ ∂K.

“(2) ⇒ (3)”:

“(3) ⇒ (1)”: Assume that K is not compact. Then there exists an open covering U = (Uj )j∈J
Sn sub-covering. As C is second countable we can assume that J = N. For n ∈ N
with no finite
let Wn = j=1 Wj . We have Wn ⊂ Wn+1 . If Wn = Wn+1 then we can remove Un+1 from the
collection without changing our assumptions. We can there fore assume that we can find
zn+1 ∈ Wn+1 \ Wn . Let (zjk ) be a converging subsequence with limit z. Then there exists m
such that z ∈ Um ⊂ Wm . But then, as Wm is open, there exists n such that for all j ≥ n we
have zj ∈ Wm contradicting the choice of zj . Hence (Uj ) has a finite sub-covering.

Let U ⊂ C. Then the number

diam(U ) = sup |z − w|
z,w∈U

is called the diameter of U . Note that if U ̸= ∅ then diam(U ) ∈ R≥0 , but diam(∅) = −∞.
{thm:Ints} We will need the following later on:
Theorem 2.3.9. Suppose hat (Kj )j is a decreasing sequence
T of compact sets such that
diam(Kj )swarrow0. Then there exists z ∈ C such that Kj = {z}.

Proof. Pick zj ∈ Kj . For ϵ > 0 let n ∈ N be so that diam(Uj ) < ϵ for all j ≥ n. As the
sequence {Kj } is decreasing it follows that for all j ≥ n we have Uj ⊂ Kn . Thus for j, k ≥ n
we have zj , zk ∈ Kn and |zj − zk | ≤ diam(Kn ) < ϵ. Hence the sequence (zj ) is Cauchy and
therefore convergent. Let z = lim zj . Let n ∈ N. Then there exists k such that zj ∈ Kn
T j ≥ k. As Kn is closed
for all T it follows that z ∈ Kn . As n was arbitrary it follows that
z ∈ Kj . Assume that w ∈ Kj . Then w − z ∈ Kj for all j. Thus |z − w| ≤ diam(Kj ) → 0
and w = z.

2.4 Continuous functions

In this section U will denote a non-empty subset of C In this section we discuss basic theory
of continuous functions defined on U .
Definition 2.4.1. (1) Let z ∈ U . A function f : U → C is continuous at z if for for every
ϵ > 0 there exists δ > 0 such that if w ∈ Dδ (z) ∩ U then |f (w) − f (z)| < ϵ.

(2) The function f : U → C is continuous on U if f is continuous at every point in U .

14
 Remark 2.4.2. If U is given and f : U → C then “f is continuous” will mean “f is
continuous on U ”.

Lemma 2.4.3. Let f : U → C. Then f is continuous on U if and only if for every open set
V ⊂ C then set f −1 (V ) is open in U .

Proof. Assume first that f is continuous. Let V ⊂ C be open. Let z ∈ f −1 (V ) ∩ U . Let

ϵ > 0 be so that Dϵ (f (z)) ⊂ V . Then, as f is continuous, there exits δ > 0 such that
f (Dδ (z) ∩ U ) ⊂ Dϵ (f (z)) ⊂ U . It follows that Dδ (z) ∩ U ⊂ f −1 (U ) and hence f −1 (U ) is open
in U .

To prove the other direction let z ∈ U and ϵ > 0. Then f −1 (Dϵ (z)) is open in U . In particular
there exists δ > 0 such that Dδ (z) ∩ U ⊂ Dϵ (f (z)) and hence f is continuous at z. As z was
arbitrary it follows that f is continuous on U .

Theorem 2.4.4. Assume that K ⊂ C is compact and f : K → C is continuous. Then f (K)

is compact.

Proof. Let (Uj ) be a open covering of f (K). As f is continuous the sets Vj = f −1 (Uj ) is
open in K. Let Wj be open in C such that Vj = K ∩ Wj . Then (Wj ) is an open covering of
K.

Corollary 2.4.5. If K is compact and f : K → R is continuous, then there exists a ≤ b

such that a ≤ f (z) ≤ b for all z ∈ K.

em:ContBas} Here are some more standard facts about continuous functions:

Lemma 2.4.6. Let U be open and non-empty. Let g, f : U → C be continuous and λ ∈ C.

Then the following holds:

1. λf is continuous.

2. f + g and f g are continuous.

3. If f (z) ̸= 0 on U then 1/f is continuous on U .

4. If W is open, h : W → C continuous and f (V ) ⊂ W . Then h ◦ f is continous.

Definition 2.4.7 (Connected sets). Let U ⊂ C. Then U is connected if for two disjoint
open sets O1 and O2 with U ⊂ O1 ∪ O2 we have U ⊂ G1 or U ⊂ G2 .

The notation Ω ⊂ C will always stands for anon-empty open connected subset of C. Such
sets are called domains.

15
Remark 2.4.8. Another we to write this is by saying that U can not be written as a disjoint
union of two non-empty relatively open subsets of U .

Theorem 2.4.9. If U is connected and f : U → C is continuous, then f (U ) is connected.

Proof.

Definition 2.4.10. Let U ⊂ C. A path in U is a continuous map γ : [a, b] → U , a, b ∈ R,

a < b. We say that γ(a) is the initial point and γ(b) is the end-point of the path. We denote
by |γ| = γ([a, b]) ⊂ U .

Remark 2.4.11. The map ψ(t) = (b − a)t + a is a bijection [0, 1] → [a, b] with inverse
ψ −1 (s) = (s − a)/(b − a). Hence, if γ : [a, b] → U is a path, then γ
e = γ ◦ ψ is a new path
such that |γ| = |e
γ |, showing that (a) one can in most cases assume that [a, b] = [0, 1] and (b)
it is important to distinguish between the path γ and |γ|, the corresponding subset of U .

Definition 2.4.12. Let U ⊂ C. We say that U is path-wise connected if for every points
z, w ∈ U there exists a path in U with initial point z and endpoint w.

Theorem 2.4.13. U ⊂ C is connected if and only if it is path connected.

Define what a polygon in U is.

Lemma 2.4.14. Ω a domain, then every two points can be connected by a polygon.

Exercises for Chapter 1

1. Show that (C, +, ·) is a field.

2. Show that Dr (z0 ) is the closure of Dr (z0 ).

3. For a set U ⊂ C show that z ∈ ∂U if and only if for all r > 0 we have Dr (z)∩U, Dr (z)∩
U c ̸= ∅

16
Chapter 3

Differentiable functions

In this section we introduce the notion of differentiable functions, analytic or holomorphic

functions and power series. We then show that an analytic function is infinitely many times
differentiable and in particular holomorphic. In the next chapter we then show that every
holomorphic function is analytic. We note here that there are authors that use the word
analytic for continuously differentiable.

3.1 Differentiable functions

In this section U will always denote an open non-empty subset of C and f will denote a
complex valued function on U . A point z ∈ U is isolated if there exists r > 0 such that
U ∩ Dr (z) = {z}. In the following we will always assume without stating it that z is not
isolated, so in particular there always exists a sequence (wj ) such that wj ̸= z and wj → z.
Note that if z ∈ U ◦ then z is not isolated.

3.2 Complex differentiable functions

Definition 3.2.1. 1) Let z ∈ U . The function f : U → C is differentiable at z if the limit

f (w) − f (z)
lim =: f ′ (z)
U \{z}∋w→z w−z

exists.
2) f is differentiable on U if f is differentiable at every z ∈ U .

17
 3) If f is differentiable on U then we define the function f ′ : U → C by
f (w) − f (z)
f ′ (z) = lim .
U \{z}∋w→z w−z
The function f ′ is called the derivative of f (on U ).
m:DiffCont} 3) If U is open then f is holomorphic on U and f is differentiable on U .
Lemma 3.2.2. If f is differentiable at z ∈ U then f is continuous at z.

Proof. Let ϵ, a > 0. Then there exists a δ > 0 such that

1. δ < ϵ/(|f ′ (z)| + a) and

2. for all w ∈ U \ {z} and |w − z| < δ we have
f (w) − f (z)
− f ′ (z) < a.
w−z

Then for |w − z| < δ:

|f (w) − f (z)| < (|f ′ (z)| + a)|w − z| < ϵ
showing that f is continuous at z.
Lemma 3.2.3. Let f : U → C. Then f is differentiable at z ∈ U if and only if there exists
a complex number A and a function ∆z = ∆ : U → C such that

1. ∆ is continuous at z and lim ∆(w)/(w − z) = 0 and

w→z

2. f (w) = f (z) + A(w − z) + ∆(w)

Proof. Assume that f is differentiable at z. Let A = f ′ (z) and define

∆(w) = f (w) − f (z) − A(w − z).
Then by definition f (w) = f (z) + A(w − z) + ∆(s) and by Lemma 3.2.2 the function ∆ is
continuous. We also have for w ̸= z
∆(w) f (w) − f (z)
= − A → 0, as w → z
w−z w−z
as f is differentiable at z.

Now assume that ∆ as above exits. Then

f (w) − f (z) ∆(w)
−A= → 0 as w → z
w−z w−z
by (1). Hence f is differentiable at z with derivative A.

18
 The results for differentiable functions corresponding to Lemma 2.4.6 for continuous func-
tions is:

Lemma 3.2.4. Suppose f, g : U → C be differentiable at z ∈ U . Let λ ∈ C. Then the

following holds:

Linearity: λf + g is differentiable at z and (λf + g)′ (z) = λf ′ (z) + g ′ .

Product rule: The function f g is differentiable at z and

(f g)′ (z) = f ′ (z)g(z) + f (z)g ′ (z).

1/f : Assume that z ∈ U ◦ and that f Iw) ̸= 0 for w ∈ U . Then, if f ′ (z) ̸= 0, there
exists Dr (z) ⊂ U such that f (w) ̸= f (z) for w ∈ Dr (z), f1 is differentiable and
 ′
1 −f ′ (z)
(z) = .
f f (z)2

Proof. Let us proof (3): Let h = 1/f

h(w) − h(z) f (z) − f (w) 1 −f ′ (z)

= −→ as w −→ z.
ex:DiffPol} w−z w−z f (w)f (z) f (z)2

Example 3.2.5 (Polynomials). The constant functions f (z) = a are differentiable with
derivative 0. We also have with f (z) = z

f (w) − f (z) w−z

= =1
w−z w−z
so f is differentiable with f ′ = 1. Assume now that fk (z) = z k is differentiable for all
k ≤ n ∈ N. Then by the product rule fn+1 = fn f1 is differentiable and
′
fn+1 (z) = fn′ (z)f1 (z) + fn (z)f1′ (z) = nz n−1 z + z n = (n + 1)z n .

It follows that every polynomial p(z) = nk=0 ak z k is differentiable and

P

n−1
X
p′ (z) = (k + 1)ak+1 z k .
k=0

ex:nonDiff} In a moment we will generalize this to infinite sums or power series.

Example 3.2.6 (Non differentiable functions). 1) Let f (z) = z̄. Then with w = z + h,
h ∈ R∗ = R \ {0}, we have
f (w) − f (z) h
= =1
w−z h

19
 and for w = ih, h ∈ R∗
f (w) − f (z) −ih
= = −1 .
w−z ih
More generally we see that
f (z + h) − f (z) h̄
=
(z + h) − z h
and this can be any number in S1 (0). In particular the map z 7→ z̄ is not differentiable.

Theorem 3.2.7 (Chain rule). Assume that f : U → V ⊂ U is differentiable at z ∈ U and

that g : V → C is differentiable at f (z). Then the function g ◦ f : U → C is differentiable at
z and
(g ◦ f )′ (z) = g ′ (f (z))f ′ (z).

Proof. Write ∆f respectively ∆g the above ∆-function for f respectively g. Let

w′ = f ′ (z)(w − z) + ∆f (w)

and note that w′ → 0 and

{eq:wprime} w′ /(w − z) → f (z) as w → z. (3.2.1)

Then with h = g ◦ f :

h(w) = g(f (w))

= g(f (z) + w′ )
= h(z) + g ′ (f ′ (z))w′ + ∆g (f (z) + w′ ))
= h(z) + g ′ (f ′ (z))f ′ (z)(w − z) + (g ′ (f ′ (z))∆f (w) + ∆g (f (z) + w′ )).

Define ∆h (w) = g ′ (f ′ (z))∆f (w)+∆g (f (z)+w′ ) then (2) holds by definition. Furthermore,
as ∆g is continuous at f (z) as well as f and w′ are continuous at z, it follows that ∆h is
continuous at z. We also have
∆h (w) ∆f (w) ∆g (f (z) + w′ )
= g ′ (f ′ (z)) +
w−z w−z w−z
∆f (w) ∆g (f (z) + w′ ) w′
= g ′ (f ′ (z)) +
w−z w′ w−z
→0 as w → z

by (3.2.1) and (1) applied to ∆f .

Lemma 3.2.8. If Ω is connected and f : Ω → C holomorphic with f ′ (z) = 0 for all z ∈ Ω,

then f is constant.

20
 3.3 Real and complex differentiable functions: The Cauchy-
Riemann Differential equation

Let us look at the definition a little closer:

f (z + h) − f (z)
f ′ (z) = lim .
h→0 h
Assume that f is defined on an open set ∅ =
̸ Ω ⊂ C. Let us first take the limit h ∈ R and
then ih ∈ iR. Then we get
f (z + h) − f (z) f (z + ih) − f (z)
eq:reDiff1} f ′ (z) = lim = lim . (3.3.1)
R∋h→0 h h→0 ih
The best way to understand this is to use the identification C ≃ R2 and write
 
u(x, y)
f (x, y) = f (x + iy) = u(x, y) + iv(x, y) ↔ .
v(x, y)

Then (3.3.1) reads

∂u ∂v 1 ∂u ∂v
(x, y) + i (x, y) = (x, y) + (x, y).
∂x ∂x i ∂y ∂y
Thus if f is differentiable at z then
∂u ∂v ∂v ∂u
{eq:CR} (x, y) = (x, y) and (x, y) = − (x, y). (3.3.2)
∂x ∂y ∂x ∂y
The system of differential equations (3.3.2) is called the Cauchy-Riemann equation.

We will often use this in the following way. Define, using the above notation:
   
∂ 1 ∂ 1 ∂ ∂ 1 ∂ 1 ∂
= + and = −
∂z 2 ∂x i ∂y ∂ z̄ 2 ∂x i ∂y

then the CR reads: If f is holomorphic then

∂f
= 0.
∂ z̄

∂2 ∂2
One consequence is the following. Let ∆ = + be the Laplace differential operator
∂x2 ∂y 2
on R2 ≃ C and note that
∂2 ∂2
∆=4 =4 .
∂z∂ z̄ ∂ z̄∂z
Definition 3.3.1. A function F : R2 → R is harmonic if ∆F = 0

21
 Theorem 3.3.2. Assume that f (x, y) = u(x, y)+iv(x, y) is twice continuously differentiable.
Then u and v are harmonic.

Let Ω ⊂ R2 and z ∈ Ω. A function F : Ω → R2 is differentiable at z if there exists a

linear map T : R2 → R2 and a map ∆ : Ω → R2 such that
f (z + h) = f (z) + T (h) + ∆(h), z + h ∈ Ω, ∥∆(h)∥/∥h∥ → 0 as h → 0.
If f is differentiable at z then we write Df (z) = T . The linear map Df (z) is the total
derivative of f at z. The function f is differentiable on Ω if it is differentiable at every
z ∈ Ω. If f is differentiable at z then the partial derivatives ∂fi /∂xj (z) exists and the matrix
representation of T is given by
 ∂f ∂f1 
1
(z) (z)
 ∂x ∂y 
Df (z) =   =: Jf (z).
 
 ∂f ∂f 
2 2
(z) (z)
∂x ∂y
The matrix Jf (z) is the Jacobi matrix of f at z. Furthermore, if the partial derivatives
exists in a neighborhood U of z and are continuous, then f is differentiable on U and
{thm:CR2} Df (w) = Jf (w) for w ∈ U .
Theorem 3.3.3. Suppose ∅ ̸= Ω ⊂ C and that f = u + iv : Ω → C. If the function f is
real differentiable at z ∈ Ω and satisfies the Cauchy-Riemann differential equation then f is
holomorphic and
∂ ∂
f ′ (z) = f (z) = 2 u(z), z ∈ Ω.
∂z ∂z

Proof. As f is differentiable at z there exists a function ∆ : Ω → C such that limh→0 |∆(z +

h)|/|h| = 0 z, z + h ∈ Ω. Write with h = h1 + ih2 :
f (z + h) − f (z) = Df (z)h + ∆(z + h)
= (ux (z)h1 + uy (z)h2 ) + i(vx (z)h1 + vy (z)h2 ) + ∆(z + h)
= (ux (z)h1 + uy (z)h2 ) + i(−uy (z)h1 + ux (z)h2 ) + ∆(z + h)
= (ux (z) − iuy (z))(h1 + ih2 ) + ∆(z + h)
∂
= 2 u(z)h + ∆(z + h).
∂z
Now divide by h and take the limit h → 0 to get
f (z + h) − f (z) ∂ ∆(z + h) ∂
lim = 2 u(z) + lim = 2 u(z).
h→0 h ∂z h→0 h ∂z
∂
Hence f is complex differentiable and f ′ (z) = f . In particular a real differentiable function
∂z
f is holomorphic if and only if it satisfies the Cauchy-Riemann differential equation.

22
3.3.1 Intermezzo: Real linear and complex linear maps on Rn

To understand the Cauchy-Riemann differential equation let us discuss complex linear maps
on Cn = R2n where we as usually identify Cn with R2n via

 
x
Φ(x + iy) = , x = (x1 , . . . , xn )⊤ and y = (y1 , . . . , yn )⊤ .
y

Multiplication my i corresponds to the R-linear map

 
0 −I
J= = Φ ◦ Mi ◦ Φ−1
I 0

as follows from
     
−1 x 0 −I x
i(x + iy) = −y + ix or Φ iΦ = .
y I 0 y

Here I = In stands for the n × n identity matrix and Mi stands for the multiplication by i.

A R-linear map is complex linear

 if and only if it commutes with multiplication by i.
a b
Thus, written as a matrix, A = , a, b, c, d ∈ Mn (R), is complex linear if and only if
c d

    
a b 0 −I b −a
=
c d I 0 d −c
  
0 −I a b
=
I 0 c d
 
−c −d
=
a b

which happens if and only if a = d and b = −c. Thus A is complex linear if and only if in
matrix form
 
a b
A= .
−b a

Thus Theorem 3.3.3 says that a real differentiable function f : Ω → C is holomorphic if and
only if Jf (z) is complex linear and hence a multiplication by a complex number which in
this case is f p rime(z).

23
 3.4 Power series and their properties

3.4.1 The geometric series

Let us start with a simple example by considering the series nn=0 z n which is called the
P

geometric series. Consider the finite sum sN = N n

P
n=0 z and note that
N
X +1
zsN = z n = sN + z N +1 − 1.
n=1

Thus
1 z N +1
(1 − z)sN = 1 − z N +1 or sN = − .
1−z 1−z
We have (
0 if |z| < 1
lim |z N +1 | = .
does not exists if |z| > 1 or z = −1
Thus the infinite sum
∞
X 1 z N +1
z n = lim sN = − lim
n=0
N →∞ 1 − z N →∞ 1 − z

exists if and only if |z| < 1 and in that case

∞
X 1
zn = .
n=0
1−z

3.4.2 Power series

∞
X
A power series is a formal expression an (z − a)n with an , a ∈ C. The numbers an are
n=0
the coefficients and a is the center. A special case is when an = 0 for all n greater or equal
to some fixed naturalP 0number n0 . n In that case the sum is a polynomial hence represent
a function p(z) = nn=0 an (z − a) which is well defined for all z ∈ C. Furthermore the
function p is infinitely many times differentiable. We will now discuss similar questions for
{thm:PS1} infinite sums.
Theorem 3.4.1. Let ∞ n
P
n=0 an (z − a) be a power series. Then there exists R ∈ [0, ∞] such
that if R > 0 the power series converges absolutely for all z with |z − a| < R and is divergent
for all z such that |z − a| > R in case R < ∞. If R = ∞ then the power series converges
for all z ∈ C and if R = 0 then the power series is divergent for all z ̸= a. Finally we
can take R = 1/ lim sup |an |1/n and if an ̸= 0 for all n then this equals 1/L = R where
L = lim sup cn+1
cn
.

24
 Definition 3.4.2. The number P R = Rf above is called the convergence radius or radius of
convergence of the power series ∞ n
n=0 an (z − a) .

Note that if R > 0 then the power series ∞ n

P
n=0 an (z −a) defines a function f : DR (a) → C
given by
∞
X
f (z) = an (z − a)n , |z − a| < R.
{thm:Abel} n=0

an (z − a)n converges for z = z0 ̸= 0 then the

P
Lemma 3.4.3 (Abel). If the power series
power series converges absolute and uniformly for all z ∈ Dr (a) for 0 < r < |z − z0 |.

an (z0 − a)n is convergent it follows that there exists n0 such that for all n ≥ n0
P
Proof. As
we have |an (z0 − a)|n < s < 1. Thus for |z − a| ≤ r < |z0 − a| we get
n n
n n z−a z−a
|an (z − a) | = |an ||z0 − a| ≤ .
z0 − a z0 − a
As |z − a|/|z0 − a| < 1 then comparing with the geometric series, shows that
∞
X n1
X ∞
X
|an (z − a)n | = |an (z − a)n | + |an (z − a)n |
n=0 n=0 n=n1 +1

shows that the series converges uniformly on Dr (a).

Pn−1
Lemma 3.4.4. Let a, b ∈ C and n ∈ N. Then an − bn = (a − b) j=0 aj bn−1−j .

Proof. We have
n−1
X n−1
X n−1
X
j n−1−j j n−1−j
(a − b) ab =a ab −b aj bn−1−j
j=0 j=0 j=0
n−1
X n−1
X
= aj+1 bn−1−j − aj bn−j
j=0 j=0
n
X n−1
X
= aj bn−j − aj bn−j
j=1 j=0
n n
{thm:pSer} =a −b
P∞ n
Theorem 3.4.5. Let f (z) = n=0 an (z − a) be a power series with R > 0. Then the
function f is infinitely many times differentiable on DR (a) and the k th derivative is given by
∞
X
(k)
{eq:pSer} f (z) = an+k (n + k) · · · (n + 1)(z − a)n , |z − a| < R. (3.4.1)
n=0

In particular we have Rf (k) = Rf .

25
 Proof. Let L = lim sup |an |1/n and R = 1/L. Assume first that 0 < L < ∞. If |z − a| < R
let ϵ > 0 be so that (L + ϵ)|z − a| = r < 1. Let n0 be so that for all n ≥ n0 we have
|an |1/n < L + ϵ. Then for n ≥ n0 we have 1/n
P∞|an | |z − a| ≤ r < 1 and the comparison with
the geometric series show that the sum n=0 an (z − a)n converges absolutely and uniformly
on Dr (a). If |z − a| > R then |an |1/n |z − a| > 1 for infinitely many n and hence the series
diverges.

It follows that the series defines a function f : DR (a) → C. As the convergence is uniform
on each of the compact subsets Dr (a) and the finite partial sums are continuous it follows
that f is continuous. As lim n1/n = 1 it follows that

lim sup |nan |1/n = lim sup |an |1/n = L.

To be finished

Definition 3.4.6. A function f P : Ω → C is said to be analytic if for every z0 ∈ Ω there exits

n
Dr (z0 ) ⊂ Ω and a power series n=0 an (z − z0 ) with convergence radius R ≥ r such that
f (z) = n=0 an (z − z0 )n on Dr (z0 ).
P

Corollary 3.4.7. Suppose that Ω is connected and that f is analytic on Ω. Then the fol-
lowing holds:

k
X f (n)
1. If f (z) = ak (z − z0 )n on Dr (z0 ) ⊂ Ω then an = , n ∈ N0 .
n=0
n!

2. If f is analytic and there exits a sequence {zn } in Ω with limit point z0 ∈ Ω such that
f (zn ) = 0 for all n then f = 0 on Ω.

3. If f (k) (z0 ) = 0 for all k ∈ N0 then f = 0.

thm:PrAnal}
P∞ P∞
Theorem 3.4.8. Let f (z) = n=0 an (z − a)n and g(z) = n=0 bn (z − a)n be analytic. Let
X∞
R = min Rf , Rg . Then f g is analytic on DR (a) and f g(z) = cn (z − a)n with
n=0

n
X
cn = ak bn−k .
k=0

Proof. To be added.

26
3.5 The exponential function and and other well known
functions

3.5.1 The exponential function

X zn
Consider the power series f (z) = . Then an = 1/n! and
n=0
n!

|an+1 | (n)! 1
= = → 0.
|an | (n + 1)! n+1

Thus R = ∞ and the power series converges for all z ∈ C. Furthermore f (z) is differentiable
and
∞ ∞
′
X zn X zn
f (z) = (n + 1) = = f (z).
n=0
(n + 1)! n=0 n!

Recall that the usual real exponential function is a solution to the following initial value
d
problem g = g and f (0) = 0. To be extended: It then follows that f (z) defined through
dx
the above power series is an extension of the usual exponential function. We therefore write
∞
X zn
ez = , z ∈ C.
n=0
n!

Lemma 3.5.1. Let z, w ∈ C, the

ez+w = ez ew .
In particular ez ̸= 0 for all z ∈ C and 1/ez = e−z .

Proof. To be added

3.5.2 The trigonometric functions

Let t ∈ R. Then, by separating the even and odd

∞ ∞ ∞
it
X (it)n X t2n n
X t2n+1
e = = (−1) +i (−1)n .
n=0
n! n=0
(2n)! n=0
(2n + 1)!

We know from calculus that the first part is the Taylor series for cos(t) and the second one,
without the i, is the power series for sin(t). We therefore get:

27
 1. If t ∈ R then eit = cos(t) + i sin(t) and every z ∈ C can be written as

z = reiθ with r = |z|

and θ is unique module 2πZ.

eit + e−it eit − e−it
2. [Euler’s formula] cos(t) = and sin(t) =
2 2i
3. Both cos(t) and sin(t) extends to analytic functions on all of C given by the power
series expansion
∞ ∞
X z 2n X z 2n+1
cos(z) = (−1)n and sin(z) = (−1)n , z ∈ C.
n=0
(2n)! n=0
(2n + 1)!

Expressing cos and sin in terms of the exponential functions allows us to prove trigonometric
formulas easily. For example for x, y ∈ R:

cos(x + y) + i sin(x + y) = ei(x+y)

= eix eiy
= (cos(x) + i sin(x))(cos(y) + i sin(y))
= (cos(x) cos(y) − sin(x) sin(y)) + i(cos(x) ∈ (y) + sin(x) cos(y)).

Taking the real and imaginary parts on both sides give:

cos(x + y) = cos(x) cos(y) − sin(x) sin(y) and sin(x + y) = cos(x) sin(y) + sin(x) cos(y).

We also note that all of the above functions are analytic in x and y and extends to analytic
functions on the whole of C in each of the variables. We therefore get:

Lemma 3.5.2. Let z, w ∈ C, then

cos(z + w) = cos(z) cos(w) − sin(z) sin(w)

and
sin(z + w) = cos(z) sin(w) + sin(z) cos(w).

et +e−t et −e−t
Recall that cosh(t) = 2
and sinh(t) = 2
. Using the known power series expansion
for ez and e−z we get.

Lemma 3.5.3. The functions cosh and sinh are analytic on C and given by the power series
expansion X
cosh(z) =
n=0

28
3.5.3 The log function

Maybe move that to later as we have introduced the complex integral

Exercises for Chapter 2

1. Show that lim n1/n = 1.
n→∞

∞an z n respec-
P
2. Let R1 respectively R2 be the radius of convergence for f (x) = n=0
tively g(x) = ∞ n
P
n=0 n z . Show that
b

(a) If |an | ≤ |bn | for all n then R1 ≤ R2 .

(b) f g is analytic around zero with convergence radius R = min R1 , R2 . Furthermore,
X∞ n
X
n
if f g(x) = cn z , then cn = aj bn−j .
n=0 j=0

3. Determine the convergence radius for the following series:

∞
X 1 n
(a) z .
n=1
n
∞
X
(b) log n(z − 1)n .
n=1
∞
X n n
(c) z .
n=0
n+1
∞
X 2n
(d) zn.
n=0
(2n)(n!)
X∞
(e) n!z n .
n=0
X∞
(f) en z n .
n=0

4. Find the power series expansion for the following function around the given point a:
z
(a) , a = 0.
2 − z2
1
(b) , a = 2.
1 + z2
2
(c) ze−z , a = 0,

29
 (d) cos(z) sin(z), a = 0, a = π/2.
sin(z) π
5. Find Re , z ̸∈ + πZ.
cos(z) 2

30
Chapter 4

The Cauchy Theorem, holomorphic

functions are analytic

4.1 Integration along curves

We start this section by several definitions. Recall that a curve, or also called a parametrized
curve is a continuous map γ : [a, b] → C. The curve γ is smooth if γ ′ exists and is continuous.
Here at the endpoint the derivative is the one-sided derivative
γ(a + h) − γ(a) γ(a + h) − γ(a)
γ ′ (a) = lim and γ ′ (b) = lim .
0<h→0 h 0>h→0 h
The curve γ is piece wise smooth if there exists a = a0 < a1 < a2 < . . . < ak = b such that
the curves γ|[aj−1 ,aj ] , j = 1, . . . , k, are smooth. If γ : [a, b] → C and σ : [c, d] → C are so that
γ(b) = σ(c) then we define “the product” of γ and σ, denoted by σγ : [a, d − c + b] → C by
(
γ(t) if t ∈ [a, b]
σγ(t) = .
σ(t + c − b) if t ∈ [b, d − c + b]

We also write γ − (t) = γ(b − a − t) the opposite oriented curve. Two curves smooth γ :
[a.b] → C and σ : [c.d] → C are equivalent if there exists a strictly increasing continuously
differentiable map φ : [a, b] → [c, d] with φ(a) = c and φ(b) = c and such that γ(t) = σ(φ(t)).
Thus γ is obtained from σ by reparametrization. In that case γ ′ (t) = σ ′ (φ(t))φ′ (t).

The curve γ is closed if γ(a) = γ(b) and a closed curve is simple if γ|(a,b) is injective. The
trace of γ denoted by Trγ is the set Trγ = γ([a, b]).
Theorem 4.1.1 (Jordan curve theorem). Let γ : [a, b] → C be a simple closed curve. Then
C \ Trγ has two connected components and exactly one component is bounded and the other
is unbounded.

31
 We call the bounded component in Jordan’s Theorem the interior of the curve and the
unbounded component the exterior. The curve γ is positively oriented if the interior is always
on the left hand site and negatively oriented otherwise. If γ is positively oriented then γ − is
negatively oriented.

Example 4.1.2. 1) For a ∈ C and r > 0 let γ(t) = γr,a (t) = a + reit , t ∈ [0, 2π] then γ is
closed and simple. The trace of γ is the circle Sr (a) := {z ∈ C | |z − a| = r}. The interior is
the disk Dr (a) and the exterior is the unbounded domain {z ∈ C | |z − a| > r}.

2) Polygon

3) Spiral: γ(t) = tu eivt , u, v > 0.

If f = u + iv : [a, b] → C is continuous then we define

Z b Z b Z b X
f (t)dt = u(t)dt + i v(t)dt = lim f (tj )(tj+1 − tj ).
a a a sup |tj+1 −tj |→0

Rb
The integral exists and as in the real valued case the map f 7→ a
f (t)dt is linear.
Z b Z b
Lemma 4.1.3. Let f ∈ C(Ω). Then f (t)dt ≤ |f (t)|dt.
a a

Proof. We have
Z b X X Z b
f (t)dt = lim f (tj )(tt+1 − tj ) ≤ lim |f (tj )| (tt+1 − tj ) = |f (t)|dt. .
a ∆→0 ∆→0 a

From now on we will always assume that γ : [a, b] → C is a pice wise smooth curve. If
f : Ω → C is continuous and Trγ ⊂ Ω then we define the integral of f along γ by
Z Z b
q:CurvInt1} f= f (γ(t))γ ′ (t). (4.1.1)
γ a

If γ is piece wise smooth and γj = γ|[aj−1 ,aj ] smooth, then we define

Z k Z
X
q:DurvInt2} f= f. (4.1.2)
γ j=1 γj

R R
Lemma 4.1.4. If γ and σ are equivalent and f ∈ C(Ω), then γ
f= σ
f.

32
Proof. Let φ : [a, b] → [c, d] be strictly increasing, φ(a) = c, φ(b) = d and such that γ = σ ◦φ
as in the definition of equivalence. Then by the change of variable in one dimensional
integration we get, using that γ ′ (t) = σ ′ (φ(t))φ′ (t):
Z Z d
f= f (σ(t))σ ′ (t)dt
σ c
Z d
= (f ◦ σ)(t)σ ′ (t)dt
c
Z b
= (f ◦ σ)(φ(t))σ ′ (φ(t))φ′ (t)dt
a
Z b
= (f ◦ (σ ◦ φ))(t)(σ ◦ φ)′ (t)dt
a
Z b
= f (γ(t))γ ′ (t)dt
Za
= f.
γ

We define the length of the curve γ as

Z b Z b
′
L(γ) = ∥γ∥ := |γ (t)|dt = (x′ (t)2 + y ′ (t)2 )1/2 dt
a a

where γ(t) = x(t) + iy(t).

Lemma 4.1.5. Let γ be a piece wise smooth curve with Trγ ⊂ Ω and let f ∈ C(Ω). Then
the following holds:

R
1. The map : C(Ω) → C is linear.
γ
Z Z Z
2. If σγ is defined then f = f + f.
σγ σ γ
Z Z
3. f =− f.
γ γ−
Z
4. f ≤ ∥f ∥∞ L(γ).
γ

Definition 4.1.6. Let f ∈ C(Ω) then f has a primitive F ∈ C1 (Ω) if

F′ = f

on Ω.

33
 We say that a curve γ : [a, b] → C connects the points z1 and z2 if γ(a) = z1 and γ(b) = z2 .
The point z1 is then the initial point and z2 is the endpoint.
Lemma 4.1.7. If f ∈ C(Ω) has a primitive F and γ : [a, b] → Ω connects z1 , z2 ∈ Ω, then
Z
f = F (z2 ) − F (z1 ).
γ

Proof. We note that

d
F (γ(t)) = F ′ (γ(t))γ ′ (t) = f (γ(t))γ ′ (t).
dt
Hence, by the fundamental theorem of calculus
Z b Z
d
F (γ(b)) − F (γ(a)) = (F ◦ γ)(t)dt = f
a dt γ

and the claim follows.

Z
Corollary 4.1.8. If f has a primitive then f = 0 for every closed curve γ in Ω.
γ

{ex:FwP} The following example will play an important role in the following:
1
Example 4.1.9 (Function without a primitive). Let f (z) = z−a , a ∈ C and z ∈ C \ {a}.
Let γ(t) = a + reit , t ∈ [0, 2π] then γ is closed, γ ′ (t) = rieit = iγ(t) and
Z Z 2π Z 2π
1 it
f= ire dt = i dt = 2πi.
γ 0 reit 0

{ex:PrimPS} I particular f does not have a primitive on C \ {a}.

Example 4.1.10 (Power series). Let f (z) = ∞ n
P
n=0 an (z − a) be a power series with con-
vergence radius R > 0. Then
an−1 1/n
lim sup = lim sup |an |1/n
n
and hence the power series
∞
X an−1
F (z) = b + (z − a)n , b∈C
n=1
n

converges on DR (a). By (3.4.1) in Theorem 3.4.5 we know that F is differentiable on DR (a)

and X
F ′ (z) = an (z − a)n = f (z).
n=0
Hence every power series has a primitive. Note that F (a) = b can be chosen arbitrary, but
cancels out in Z
f = F (γ(b)) − F (γ(a)).
γ

34
 R
We have just seen that if f has a primitive then γ
f = 0 for every closed curve. We now
{thm:exPr} prove the converse:
R
Theorem 4.1.11. Assume that Ω is path connected, and that f ∈ C(Ω) is such that γ
f =0
for every closed curve in Ω. Then f has a primitive in Ω.

Proof. Fix a point z0 ∈ RΩ. For z ∈ Ω let γ = γz : [a, b] → Ω be so that γ(a) = z0 and
γ(b) = z. Define F (z) = γ f . We claim that the definition of F is independent of the choice
of γ. For that let σ : [c, d] → Ω be another curve with initial point z0 and endpoint z. We
can assume that c = b. Then the curve σ − γ : [a, d] → Ω is closed and hence
Z Z Z
0= f =− f+ f
σ− γ σ γ
R R
showing that γ f = σ f . Let z ∈ Ω. Let r > 0 be so that Dr (z) ⊂ Ω. Let h ∈ C be so that
γh (t) := z + th ∈ Dr (z) for all t ∈ [0, 1]. Then γh γ is a curve connecting z0 and z + h. Thus
Z Z Z Z 1
F (z + h) − F (z) = f− f= f= f (z + th)hdt.
γh γ γ γh 0

Let ϵ > 0. Choose r so that |f (w) − f (z)| < ϵ for all w ∈ D(z). Then
Z 1 Z 1
F (z + h) − F (z)
− f (z) = f (z + th) − f (z)dt ≤ |f (z + th) − f (z)|dt ≤ ϵ.
h 0 0

F (z + h) − F (z)
Hence lim = f (z) as claimed.
lem:LimInt} h→0 h
Lemma 4.1.12. Let (fn )n be a sequence of continuous functions on I = [a, b] that converges
uniformly on I to f . Then f is continuous on I and
Z Z
f (t)dt = lim fn (t)dt.
I n→∞ I

Proof. Let ϵ > 0. Let n0 ∈ N be so that all n ≥ n0 we have |fn (t) − f (t)| < ϵ for all t ∈ I.
As I is compact it follows that each of the function fn is uniformly continuous. Let δ > 0 be
so that |fn0 (s) − fn0 (s)| < ϵ if |s − t| < δ. Then

|f (s) − f (t)| ≤ |f (s) − fn0 (s)| + |fn0 (s) − fn0 (t)| + |fn0 (t) − f (t)| < 3ϵ.

Thus f is continuous and hence integrable. Now for n ≥ n0 :

Z Z Z
| f (t)dt − fn (t)dt| ≤ |f (t) − f( t)|dt ≤ ϵ(b − a).

Hence the claim.

35
em:PowPrim}
∞
X
Lemma 4.1.13. Let f (z) = an (z − a)n be a power series with convergence radius R > 0.
n=0
Let γ : [a, b] → DR (a) be a continuous curve. Then with fn (z) = nk=0 ak (z − a)k we have
P

Z Z
f = lim fn .
γ n→∞ γ
R
In particular if γ is a closed curve in Dr (a) then γ
f =0

Proof. There exists 0 < r < R such that Trγ ⊂ Ds (a) ⊂ DR (a). As fn → f uniformly
on Ds (a) it follows that f◦ γn → f ◦ γ uniformly on [a, b]. The first claim follows from
Pn+1 ak−1 k
LemmaR 4.1.12. The polynomials fn haveRall a primitive Fn (z) = k=1 k (z − a) and
hence γ fn = 0 by Theorem 4.1.11. That γ f = 0 follows then by the first part.

4.2 Goursat-Pringsheim Theorem

R
We will now show that for holomorphic functions the statements γ f = 0 for special class
of closed curves. This is enough to prove the existence of primitives for special domains.

Let D ⊂ C be a bounded closed subset. We set diam(D) = sup{|w − z| | w, z ∈ D}.

We now divide T0 into three smaller triangles by adding the midpoints of each site as a new
vertex.

In the following add pictures

{thm:GP} We will use the following construction in the following.

Theorem 4.2.1 (Goursat-Pringsheim). Let Ω ⊂ C be a domain and let T0 be a triangle such

that T0 and the interior are contained in Ω. Let γ be a positively oriented parametrization
of the boundary of T0 . If f ∈ O(Ω) then
Z
f = 0.
γ

Proof. Denote by a0 , b0 and c0 the sides of T0 and A, B, and C the corners. We can then
take γ = γ0 as 
A + t(B − A)
 t ∈ [0, 1]
γ(t) = B + (t − 1)(C − B) t ∈ [1, 2] .

C + (t − 2)(A − C) t ∈ [2, 3]


36
So the first part is a parametrization of a staring at A and ending at B, etc. Let d0 =
diam(T0 ). We now divide T0 into four new triangles T01 , T02 , T03 , T04 (the last one being in
the middle) be adding new endpoints given by the midpoint of each of the sites. Let γ0J be
a positively oriented parametrization of the new triangles and note that the new sites are
traveled now twice but in opposite direction. Hence the total integrals over those sides is
zero and we get Z Z Z Z Z
f= f+ f+ f+ f
γ0 γ01 γ02 γ03 γ04

and hence Z Z Z Z Z Z
f ≤ f + f + f + f ≤ max f
γ0 γ01 γ02 γ03 γ04 j=1,2,3,4 γ0j

Let γ1 = γ0j be such that

Z Z
max f = f .
j=1,2,3,4 γ0j γ1

Iterating this process we get parametrized triangles with diam(Tj ) ≤ 2−j diam(T0 ) and
R R
γ0
f ≤ 4j | γj f |. Let Dj be the closed triangle given by the closure of the interior of
T
Tj . Then there exists z0 = j Dj . Write

∆(z)
f (z) = f (z0 ) + f ′ (z0 )(z − z0 ) + (z − z0 )
z − z0
∆(z) ∆(z)
and recall that limz→z0 z−z0
= 0 and that by definition z 7→ z−z0
is continuous.

By Example 4.1.10 we see that

Z Z
f (z0 ) = f ′ (z0 )(z − z0 ) = 0,
γj γj

Hence Z Z
∆(z)
f = (z − z0 ) .
γj γj z − z0
∆(z)
Let ϵ > 0. Choose δ > 0 such that z−z 0
< ϵ/d−2
0 with d0 the length of γ0 . if |z − z0 | < δ.
Then let j0 be so that diam(Tj ) ≤ δ. we have with dj be the length of the triangle Tj . Then
dj ≤ 2−j d0 . We also have |z − z0 | ≤ d0 2−j for z ∈ Dj . Then for j ≥ j0 It complete follows
that Z
f ≤ ϵ.
γ0

Definition 4.2.2. A domain Ω is a star domain if there exists a point a ∈ Ω such that for
all z ∈ Ω the line segment γz (t) = a + t(z − a), t ∈ [0, 1], is contained in Ω.

37
 Example 4.2.3. Every convex domain is start shaped. In particular the open disk Dr (a) is
start shaped. A start with arbitrary vertexes is star shaped.

If Ω is a start domain and f ∈ C(Ω) then we can define F : Ω → C by

Z Z 1
F (z) = f = (z − a) f (a + t(z − a))dt.
γz 0

4.3 Cauchy theorem and its applications

We start by some simple generalizations of Groursat-Pringshiem theorem that then allows

us to give a simple proof of Cauchy’s integral theorem.

Theorem 4.3.1. Assume that Ω is a star domain and f ∈ O(Ω) then F is a primitive fort
f.

Proof. Let z ∈ Ω. As before let r > 0 be so that Dr (z) ⊂ Ω. Let γ(t) = z + th, t ∈ [0, 1].
Then γz+h γ − γz− parametrizes a triangle and
R it follows from the fact that Ω is a star domain
that the interior is contained in Ω. Thus γz+h γ − γz− f = 0. It follows that
Z Z
F (z + h) = f = F (z) + f.
γγz γz

It now follows as in the proof of Theorem 4.1.11 that F ′ (z) = f (z).

{lem:HebS1} We can now use this result to give generalizations that will be used in the next section.

Lemma 4.3.2. Let Ω be a convex domain and w ∈ Ω. Assume that f ∈ O(Ω \ {w}) and
f continuous on Ω. Let γ be a positively
R oriented parametrized triangle in Ω such that the
interior of γ is contained in Ω. Then γ f = 0.

Proof. Denote by ∆ the closed triangle determined by γ. We consider four cases.

1) w ∈ Ω \ ∆. This reduces to the Theorem 4.2.1 by shrinking the domain.

2) w is one of the corner, let us say w = A if the triangle has vertices A, B, and C. Let
ϵ > 0. Choose a small triangle ∆1 = ADE with parametrization γ1 and length
ϵ
L(γ1 ) <
supz∈Trγ |f (z)| + 1

38
 and D on AB and E on AC. Drawing the segments DB gives two new triangles ∆2 and
∆3 . Let γj be a parametrization of ∆j and note that all new sites are traveled twice with
opposite orientation. It follows from Theorem 4.2.1 that
Z Z
f= f.
∆ ∆1

Hence Z Z
f = f ≤ sup |f (z)|L(γ1 ) < ϵ
∆ ∆1
R
and it follows that ∆
f = 0.

3) The point w is on one of the sites of ∆, say AB. Then divide the original triangle into
two triangles ∆1 and ∆2 with parametrization γj , j = 1, 2, by drawing the segment wC and
use (2) to show that Z Z Z
f= f+ f = 0.
γ γ1 γ2

4) The final case is now that w is an interior point in the original triangle. Then connect w
to each of the vertices to get for new triangles with parametrization γj , j = 1, 2, 3, 4. Then
Z 4 Z
X
f= f =0
γ j=1 γj

{thm:exPr2} by (2).

Theorem 4.3.3. Let Ω be a convex domain and w ∈ Ω. Assume that f ∈ O(Ω \ {w}) and
f continuous on Ω. Then the following hols:

1. f has a primitive on Ω.
R
2. If γ is a closed curve in Ω then γ
f = 0.

{thm:CIT} Proof.

Theorem 4.3.4 (Cauchy’s integral theorem). Assume that Ω is a convex domain and a ∈ Ω.
Let r > 0 be so that Dr (a) ⊂ Ω and let γ(t) = a + reit be a simple positively oriented
parametrization of the boundary ∂Dr (a) = {w ∈ C | |z − a| = r}, t ∈ [0, 2π]. Let f ∈ O(Ω).
Then Z
1 f (w)
f (z) =
2πi γ w − z
for all z ∈ Dr (a).

39
 Proof. Define (
f (w)−f (z)
w−z
if w ̸= z
g(w) = ′
f (z) if w = z
Then g is holomorphic on Ω \ {z} and continuous on Ω as f is complex differentiable.

Let 0 < s < r be so that Ds (z) ⊂ Dr (a). Let γs,z (t) be a negatively oriented parametriza-
tion of ∂Ds (z). Fix a point w1 ∈ Trγ and let w2 ∈ Trγs.z be the point closest to w1 . Denote
by γ3 the line segment from w1 to 2 . We now start with γ until we reach w1 then travel γs,z
starting at w2 until we reach w2 again. Add a picture. Then we travel back from w2 to
w1 and then finish γ. By Theorem 4.3.3 the integral over this path is zero so
Z Z
g= g.
−
γ γs,z

R
We can therefore assume that z = a is the center. It follows by Theorem 4.3.3 that γ
g = 0.
Hence by Example 4.1.9:

f (w) − f (z)
Z
1
0=
2πi γ w−z
Z Z
1 f (w) f (z) 1
= −
2πi γ w − z 2πi γ w − z
Z
1 f (w)
= − f (z).
2πi γ w − z

Here we used that the integrals in the second line exists as all functions are continuous and
we are integrating over a compact set.

Remark 4.3.5. The statement in the theorem remains true for arbitrary simple closed curve
γ containing z in the interior with the same proof if we can show that
Z
1 1
= 1.
2π γ w − z

hm:HolDiff} This will be discussed in a moment. A very simple version of this is the following.

Theorem 4.3.6. Assume that Ω is a convex domain and f ∈ O(Ω). Then f (k) = dk f /dz k
exists for all k ∈ N0 . Furthermore, if r > 0 be so that Dr (z) ⊂ Ω and γ a simple curve
parametrizing the boundary circle {w ∈ Ω | |w − z| = r} then
Z
(k) k! f (w)
f (z) = .
2πi γ (w − z)k+1

Proof. Induction and interchanging integration and differentiation. Add details later.

40
 Theorem 4.3.7. Assume that Ω is convex and f ∈ O(Ω). Then f is analytic. In fact let
a, z ∈ Ω then
∞
X f (n) (a)
{eq:PwS} f (z) = (z − a)n (4.3.1)
n=0
n!

where the convergence radius for the power series in (4.3.1) is at least any 0 < R <
d(a, ∂Ω) = inf w∈∂Ω |w − a|.

Proof. We write
Z
1 f (w)
f (z) =
2πi γ w−z

where γ(t) = a + reit , t ∈ [0, 2π], is a so that the closed triangle ∆r (a) determined by γ and
its interior is contained in Ω. We have for z ∈ Trγ:

1 1 1
= z−a
w−z 1 − w−a w − a

and
∞  n
z−a 1 X z−a
< 1 and hence z−a = .
w−a 1 − w−a n=0
w−a

Therefore, after interchanging the sum and the integral, (add details) we get using Theorem
4.3.6
Z
1 f (w)
f (z) =
2πi γ w − z
∞  Z 
X 1 f (w)
= n+1
(z − a)n
n=0
2πi γ (w − a)
∞
X f (k) (a)
= (z − a)n .
n=0
n!

For the radius of convergence we note that the only restriction was that ∆r (a) ⊂ Ω or
0 < r < d(a, ∂Ω).

Remark 4.3.8. Note that the convergence radius for (4.3.1) can be bigger than d(a, ∂Ω)
because f might be holomorphic in a bigger domain than Ω.

Corollary 4.3.9. Assume that Ω is connected. Let f, g ∈ O(Ω) and assume there exists a
sequence (zj ) with limit limj zj = z0 ∈ Ω and such that f (zj ) = g(zj ) for all j. Then f = g
on Ω.

41
 Proof. By replacing f by f − g we can assume that g = 0. As f is continuous it follows that
f (z0 ) = 0. Let r > 0 be so that
∞
X
f (z) = an (z − z0 )n .
n=0

Then a0 and
∞
X
f (z) = (z − z0 ) an+1 (z − z0 )n .
n=0

It follows that ∞
f (z) X
f1 (z) = = an+1 (z − z0 )n
z − z0 n=0

is analytic and hence continuous. It follows that f1 (z0 ) = a1 = 0. So the same argument
applies again and inductively we see that for all k ∈ N0 we have
∞
f (z) X
fk (z) = = an+k (z − z0 )n
(z − z0 )k n=0

is analytic and fk (0) = ak = 0. Hence f = 0 on an open disk centered around z. The claim
now follows from the following simpler statement.

Lemma 4.3.10. Assume that Ω is connected and f ∈ O(Ω) is zero on an open disk Dr (z) ⊂
0. Then f = 0.

Proof. Let U = {z ∈ Ω | (∃r > 0)Dr (z) ⊂ Ω, f |Dr (z) = 0}. Then U ̸= ∅ and U is open.
Let w ∈ cl(U ). Then there exists a sequence (zj ) in U such that zj → w. By assumption
f (zj ) = 0 and hence there exists r > 0 such that Dr (w) ⊂ Ω and f = 0 on Dr (w). Hence
{cor:div} cl(()U ) is open in Ω. As Ω is connected it follows that U = Ω and f = 0.

Corollary 4.3.11. Assume that Ω is connected and f ∈ O(Ω), f ̸= 0. Let z0 ∈ Ω. Then

there exits N ∈ N0 such that f (j) (z0 ) = 0 for j = 0, . . . , N and f (N ) (z0 ) ̸= 0 and there exists
g ∈ O(Ω), g(z0 ) ̸= 0 such that
f (z) = (z − z0 )N g(z).

Proof. There exists a finite, but maximal N such that

∞ ∞
X f (n) (z0 ) X f (n+N ) (z0 )
f (z) = (z − z0 )n = (z − n0 )N (z − z0 )n .
n=N
n! n=0
(n + N )!

f (n+N ) (z0 )
Let g(z) = ∞ n
P
n=0 (n+N )! (z − z0 ) . The convergence radius is the same as that form f and
g(z0 ) = f (N ) (z0 )/(n!) ̸= 0.

42
{thm:Mor}
Theorem 4.3.12 (Morera). Suppose Ω is open and f ∈ C(Ω). The f isRholomorphic if and
only if for every disk Dr (a) ⊂ Ω and every triangle ∆ ⊂ Dr (a) we have ∂∆ f = 0.

Proof. The assumption implies that f has a primitive. Theorem 4.3.6 then implies the
statement.

As an application we pint out the following.

Lemma 4.3.13. Let f : Ω × I → C be continuous and such that for every t ∈ I the function
z 7→ f (z, t) is holomorphic. Then
Z
F (z) = f (z, t)dt
I

is holomorphic.

Proof. Let Dr (a) ⊂ Ω and ∆ ⊂ Dr (a) a triangle with positively oriented boundary γ. Then
Z Z Z  Z Z 
F = f (z, t)dt dz = f dt = 0
γ γ I I γ

where we have used Fubini’s Theorem to interchange the order of the integrals.

We collect the main results now in the following theorem:

Theorem 4.3.14. Let ∅ =
̸ Ω be open in C and let f : Ω → C. Then the following statements
are equivalent:

1. f is holomorphic on Ω,
2. f has a primitive locally around each point z ∈ Ω.
3. f is real differentiable and satisfies the Cauchy-Riemann differential equation
4. f has a power series expansion around every point z0 ∈ Ω.
5. f is continuous
R on Ω and or every disk Dr (a) ⊂ Ω and every triangle ∆ ⊂ Dr (a) we
have ∂∆ f = 0

Complete the following:

Theorem 4.3.15. f ∈ O(Ω \ {z}) and bounded (or L1 ) “around” z. Then f is holomorphic
on Ω.

Other applications?

43
4.4 Some application

4.4.1 Liouville’s Theorem

Theorem 4.4.1 (Liouville’s Theorem). Suppose f ∈ O(C) is bounded. Then f is constant.

Proof. Let A = supz∈C |f (z)| < ∞ and let r > 0. Then for z ∈ C:

f (w) 2
Z
′ 1
|f (z)| =
π γr,z w−z
|f (w)|
Z
1
≤
π γr,z |w − z|2
Z
A
≤ 2 1
πr γr.z
2A
= → 0, r → ∞.
r
Hence f ′ = 0 and f is constant.

Corollary 4.4.2. Let f ∈ O(C) not constant. Then f (C) is dense in C.

Proof. Assume that f (C) is not dense. Then there exists w ∈ C and r > 0 such that
f (C) ∩ Dr (w) = ∅. But then z 7→ 1/(f (z) − w) is bounded and hence constant contradicting
the assumption of the Theorem.

Example 4.4.3 (The exponential function). Let f (z) = ez . Then f (C) = C \ {0}. To see
that we know that ez e−z = 1 so f (z) ̸= 0 for all z. If w ∈ C \ {0} then we can write for
r = |w| > 0 w = reiθ = elog r+iθ so w ∈ f (C).

4.4.2 The fundamental theorem of algebra

Let p(z) be a polynomial. Then p is holomorphic on C.

Lemma 4.4.4. Let p be a polynomial on C of degree ≥ 1. Then there exists z1 ∈ C such

that p(z1 ) = 0

Proof. Assume that p(z) ̸= 0 for all z ∈ C. Then f (z) = 1/p(z) is holomorphic on C. Write

p(z) = an z n + an−1 z n−1 + · · · + a0

44
with an ̸= 0. Then
z n−1
 
n a0
p(z) = z an + an−1 n + · · · + n
z z
and note that
z n−1 a0
lim an + an−1 n
+ · · · + n = 0.
|z|→∞ z z
As an ̸= 0 there exists r > 0 and L1 > 0 such that |p(z)| ≥ L1 > 0 for all |z| > r. As Dr (0)
is compact and |p(z)| > 0 for all z ∈ Dr (0) there exists L2 > 0 such that |p(z)| > L2 for all
z ∈ Dr (0). Thus |f (z)| ≤ 1/ min{L1 , L2 } showing that f is bounded and hence a constant.
Thus p(z) is a constant contradicting the assumption that deg p ≥ 1. Thus there exists a
point w such that p(w) = 0.
Theorem 4.4.5. Let p(z) be a polynomial of degree n. Then there exists A, w1 , . . . wn ∈ C
such that
p(z) = A(z − w1 ) · · · (z − wn ).

Proof. We prove this by induction. If p(z) has degree one then p(z) = A(z − w1 ) and we are
done. Assume therefore that deg p > 1, Then there exists w1 ∈ C such that p(w1 ) = 0. By
Corollary 4.3.11 there exits N and q(z) such that p(z) = (z − w1 )N q(z) and q(z0 ) ̸= 0. By
the construction in the proof of Corollary 4.3.11 we see that deg q = n − N < n and hence
q is a product of linear factors and the claim follows by setting w2 = . . . = wN = w1 .

4.4.3 The maximum principle

Theorem 4.4.6. Let assume that Ω ⊂ C is open. Let f ∈ O(Ω) and assume that that |f |
takes a global maximum at the point z ∈ Ω. Then f is constant. If Ω is bounded and f
extends to a continuous function on the closure cl(Ω) then |f | takes its maximum on the
boundary ∂Ω of Ω.

Proof. Assume that f is not constant and that A = |f (z)| = supw∈Ω |f (w)|. Then there
exits z ∈ Ω and Dr (z) ⊂ Ω such that |f (z + reit )| < A for t ∈ I, I = [a, b], 0 ≤ a < b ≤ 2π.
Thus Z 2π
1
|f (z)| ≤ |f (z + reit )|dt = A.
2π 0
But Z 2π Z b Z
1 it 1 1
|f (z + re )|dt = |f (w)|dt + |f (w)| > A
2π 0 2π a 2π [0,2π]\I
a contradiction.

As cl(Ω) is compact and z 7→ |f (z)| is continuous, there exists z ∈ cl(Ω) such that
|f (z)| = A. By what we have already shown we must have z ∈ ∂Ω.

45
4.5 The winding number

To be completed

γ a closed curve, z ∈ C \ Trγ. Then the following number is a count how many times the
curve circles around the point z.
Z
1 1
n(γ, z) = .
2πi γ w − z

n(γ, z) = 0 if z is outside of γ and otherwise an integer. Furthermore n(γ − , z) = −n(γ, z).

Example 4.5.1. Let n ∈ Z and γ(t) = eintR , t1 ∈ 2π. If n ̸= 0. We have seen in the proof of
1
Theorem 4.3.4 that D1 (0) → C, z 7→ 2πi γ w−z is constant. Hence we can take z = 0 and
we get:
Z 2π
1 1
n(γ, 0) = int
ineint dt = n.
2πi 0 e

4.6 Sequences of holomorphic functions

In this section we give conditions for a sequence of holomorphic functions on a domain to

converge to a holomorphic function. A simple example is that if f ∈ O(Ω) and z0 ∈ Ω, then
there exits r > 0 such that
n
X f (k) (z0 )
fn (z) = (z − z0 )k → f (z)
k=0
n!

uniformly on Dr (z0 ). So every holomorphic function can be approximated by a polynomial

on some compact set. In the following Ω is always an open set.

Our main result is:

Theorem 4.6.1. Let Let (fn )∞n=0 be a sequence of holomorphic functions on Ω. Assume that
there exists a function f : Ω → C such that for every compact set K the sequence fn (z)
converges uniformly on K to f (z), then f is holomorphic.

Proof. Let K be compact with non-empty interior. For every z ∈ Z we can find r = r(z) > 0
such that Dr (z) ⊂ Ω. Then K ⊂ ∪z Dr(z) (z). As K is compact there exists finitely many zj
such that
n
[ n
[
K⊂ Dr(zj ) (zj ) ⊂ Dr(zj ) (zj ) ⊂ Ω.
j=1 j=1

46
Hence the theorem follows if we can prove it for any compact set of the form Dr (z0 ) ⊂ Ω.
For that we note that
Z Z
1 fn (w) 1 f (w)
fn (z) = → =: F (z)
2πi γr z − z0 2πi γr z − z0

as the convergence is uniform on the compact set Trγ. Now we already know that F is
holomorphic on Dr (z0 ). As fn (z) → f (z) we must have f = F on Dr (z0 ) and the claim
follows.
Theorem 4.6.2. Let the assumptions be as in the last theorem. Then f (k) → f (k) uniformly
on any compact subset.

To be completed

Proof.

4.7 Schwarz reflection principle

Let Ω ⊂ C be open and such that Ω = {z̄ | z ∈ Ω} = Ω. Let Ω+ = {z ∈ Ω | Rez > 0},
Ω− = −Ω+ and I = Ω ∩ R. We will assume that I is a non-empty interval. We have

Ω = Ω+ ∪ Ω− ∪ I.

Not if f + is holomorphic on Ω+ such that f extends to a continuous function f + on Ω+ ∪ I.

Define f − : Ω− → C by f − (z) = f + (z̄). Then f − is holomorphic on Ω− . If f1 is real valued
on I then the function (
f ± (z) for z ∈ Ω±
f (z) =
f + (x) if x ∈ I
is continuous on the open set Ω = Ω+ ∪ I ∪ Ω− . Then following Theorem shows that f is in
fact holomorphic.
Theorem 4.7.1 (Symmetry principle). Let the notation be as above. Let f ± ∈ O(Ω± ) be
such that f ± extends to a continuous function on Ω± ∪ I. Assume that f + |I = f − |I . Then
the function (
f ± (z) for z ∈ Ω±
f (z) =
f + (x) = f − (x) if x ∈ I

Proof. It is enough to show this for an arbitrary disk Dr (z) ⊂ Ω. So assume that Ω = Dr (x)
x ∈ R (note that the assumption on R Ω implies that the center is real). Let+ ∆ be− a triangle
±
contained in Ω. If Ω ⊂ Ω then ∆ f = 0 be as f is holomorphic in Ω ∪ Ω . We can

47
therefore assume that ∆ ∩ I ̸= ∅. We can also assume that ∆ ∩ Ω+ ̸= ∅. We now consider
R is that ∆ ∩ I is a point, say z0 , which then has to be one of the
several cases. The first
vertices of ∆. That ∆ f = 0 follows then as in the proof of (2) in Lemma 4.3.2. Next
assume that ∆ ∩ I = J is an interval and otherwise ∆ \ J ⊂ Ω+ (add a picture). For ϵ > 0
R edge RJ by JϵR= {z | Imz = ϵ} ∩ ∆.
small Rand consider a new triangle ∆ϵ by replacing the
Then ∆ϵ f = 0. As f is continuous it follows that ∆ϵ f → ∆ f so ∆ f = 0 Add details
using pictures.

Now assume that ∆ ∩ Ω± ̸= ∅. write ∆ as a union of two polygons ∆± separated by the

±
new edge ∆ ∩ I. We can then write each of the polygons R ∆ Pas aR finite union of triangles
∆1 , . . . , ∆k , with intersection at most an edge. Then ∆ = kj=1 ∆j f as integration over
R
common edges cancels. Now each of the integrals ∆j f = 0 by the previous discussion. This
R
implies ∆ f = 0. It follows by Morera’s Theorem, Theorem 4.3.12.

Exercises for Chapter 3

In the following γr,z denotes the curve γr,z (t) = z + reit , t ∈ [0, 2π]. If z = 0 then we write
γr = γr,0

1. Use the Cauchy integral theorem to evaluate the following integrals:

Z
1
(a) 2
,
γ2 (z + 1)(z − i)
Z
1
(b) 2
where γ is the path:
γ 1+z

i. γ(t) = 21 eit , t ∈ [0.2π]

ii. A positively oriented parametrization of the circle {z | |z + i| = 1}.
iii. A positively oriented parametrization of the circle {z | |z − i| = 1}.
iv. A positively oriented parametrization of the circle {z | |zi| = 3}.
(c) Let (fj )j be a sequence of holomorphic functions on Ω. Assume that for every
z ∈ Ω there exists Dr (z) ⊂ Ω such that there existsP∞ 0 < A < 1 such that
−j
for w ∈ Dr (z) we have |fj (w)| ≤ A . Show that j=0 fj exists on Ω and is
holomorphic.

2. Let Ω be a convex domain and Dr (z) ⊂ Ω. Let f ∈ C(Ω). Show that

Z
1 f (w)
F (z) =
2πi γr,z w − z

defines a holomorphic function on Ω.

48
3. A set Γ ⊂ Ω is discrete (in Ω) if for every point z ∈ Γ there exists r > 0 such that
Dr (z) ⊂ Ω and Dr (z) ∩ Γ = {z}. Let Ω be a convex domain and Γ ⊂ Ω discrete.
Suppose that f ∈ C(Ω) is holomorphic on Ω \ Γ. Then f is holomorphic on Ω.

49
50
Chapter 5

Meromorphic functions

5.1 Zeros and poles

5.2 Singularities and meromorphic functions

5.3 Application to evaluation of integrals

51
 The following list can change and also the order

5.4 Runge’s approximation theorem

52
Chapter 6

Open mapping theorem

53
54
Chapter 7

Analytic continuation

7.1 examplex

7.2 Analytic continuation in general

7.3 Further topics: Remarks on Riemannian surfaces

55
56
Chapter 8

Weierstrass Theorem

57
58
Chapter 9

Applications: The Fourier transform

9.0.1 The Paley-Wiener Theorem

59
60
Chapter 10

Special functions

10.1 The Mellin transform

10.2 The Riemann ζ-function

10.3 Elliptic functions: The ℘-function

10.4 The Hardy space and spectral Theory

61