Complex Analysis I

MAST31006

Ritva Hurri-Syrjänen
University of Helsinki

May 11, 2019

Complex Analysis I CONTENTS

Contents
Contents 1

2019 Foreword 4

2018 Foreword 5

1 Background 6
1.1 The vector space ℝ2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 The complex plane ℂ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 The elementary properties of complex numbers 10

2.1 Complex numbers can be written in Cartesian form . . . . . . . . . . . . . . . . 10
Theorem 2.1.10: The triangle inequality . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Complex numbers can be written in polar form . . . . . . . . . . . . . . . . . . 12
Definition 2.2.1: Polar coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 12
Theorem 2.2.4: Euler’s formula . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.9 The geometric meaning of the multiplication of complex numbers . . . . 14
Theorem 2.2.15: De Moivre’s theorem . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.16 Applications of De Moivre’s theorem . . . . . . . . . . . . . . . . . . . 16
2.3 The 𝑛th roots of complex numbers . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.3 Special case: the 𝑛th roots of unity . . . . . . . . . . . . . . . . . . . . . 18
2.4 About history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3 Topological notions of sets in the complex plane 20

3.1 Sets in the complex plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1.7 Connectedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 Limits and continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.1 Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.2 Limits of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2.3 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 The Cauchy convergence principle for ℂ . . . . . . . . . . . . . . . . . . . . . . 26

4 Analytic functions 27
4.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 Elementary properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Theorem 4.2.10: Chain rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Theorem 4.2.11: Derivative of inverse function . . . . . . . . . . . . . . . . . . 30
4.3 The relationship between the complex and real derivatives . . . . . . . . . . . . 30
4.3.18 On harmonic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3.32 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

5 Complex Series 41
5.1 Series of complex terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Proposition 5.1.4: Linear combinations of series . . . . . . . . . . . . . . . . . . 41
Proposition 5.1.8: Absolute convergence vs. convergence . . . . . . . . . . . . . 42
5.1.10 The comparison test . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.11 d’Alembert’s ratio test . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.12 The 𝑛th root test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.2 Power series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Theorem 5.2.7: Hadamard’s theorem . . . . . . . . . . . . . . . . . . . . . . . . 46

6 The exponential, sine and cosine functions 51

6.1 The exponential function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.1.2 Properties of the exponential function . . . . . . . . . . . . . . . . . . . 51
6.1.3 The complex conjugate and the modulus of the exponential function . . . 52
6.1.4 Euler’s formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.2 Trigonometric functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.2.1 The sine and cosine functions of one complex variable . . . . . . . . . . 53

RHS, version May 11, 2019 1

Complex Analysis I CONTENTS

6.3 Complex hyperbolic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

6.3.3 The real and imaginary parts . . . . . . . . . . . . . . . . . . . . . . . . 55
6.3.4 Unboundedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6.4 Zeros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

7 Complex mappings 57
7.1 Some properties of the complex mappings . . . . . . . . . . . . . . . . . . . . . 57

8 Complex logarithms 60
8.0.1 Argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
8.0.2 Complex logarithms: the inverse of the exponential function . . . . . . . 60
8.0.4 Branches of the argument . . . . . . . . . . . . . . . . . . . . . . . . . 61
8.0.5 Branches of the complex logarithm . . . . . . . . . . . . . . . . . . . . 61
8.0.12 General power function . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

9 Integrals 66
9.1 Complex-valued functions of real variables . . . . . . . . . . . . . . . . . . . . 66
9.2 Curves and contours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
9.3 Integration along curves/contours . . . . . . . . . . . . . . . . . . . . . . . . . . 70
9.3.3 Basic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Proposition 9.3.4: Invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Proposition 9.3.6: The integral of the negative . . . . . . . . . . . . . . . . . . . 72
Proposition 9.3.7: Joining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
9.3.11 Integrals along curves with respect to the arclength . . . . . . . . . . . . 73
Lemma 9.3.12: Estimation lemma . . . . . . . . . . . . . . . . . . . . . . . . . 73

10 Theorems related to an antiderivative 74

Theorem 10.0.8: Fundamental theorem of Calculus II . . . . . . . . . . . . . . . 75

11 Integral theorems 79
Theorem 11.0.1: Goursat’s theorem . . . . . . . . . . . . . . . . . . . . . . . . 79
Theorem 11.0.5: The Cauchy–Goursat theorem . . . . . . . . . . . . . . . . . . 86

12 Cauchy’s integral formulas 87

Theorem 12.0.1: Cauchy’s local integral formula . . . . . . . . . . . . . . . . . 87
Theorem 12.0.3: Cauchy’s second formula . . . . . . . . . . . . . . . . . . . . . 88

13 Corollaries of the Cauchy integral formulas 91

Theorem 13.0.2: Morera’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . 91
Corollary 13.0.6: Liouville’s theorem . . . . . . . . . . . . . . . . . . . . . . . 92
13.0.11 Analytic continuation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Theorem 13.0.12: Identity theorem . . . . . . . . . . . . . . . . . . . . . . . . . 94
Corollary 13.0.18: Corollary of the Identity theorem . . . . . . . . . . . . . . . 95
Theorem 13.0.21: Local maximum modulus theorem . . . . . . . . . . . . . . . 95
Theorem 13.0.24: Maximum modulus theorem . . . . . . . . . . . . . . . . . . 96
Lemma 13.0.26: The Schwarz lemma . . . . . . . . . . . . . . . . . . . . . . . 97
Theorem 13.0.27: Minimum modulus theorem . . . . . . . . . . . . . . . . . . . 98

14 Global Cauchy theorem 99

14.1 Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Definition 14.1.2: Winding number . . . . . . . . . . . . . . . . . . . . . . . . . 99
Lemma 14.1.3: Winding number lemma . . . . . . . . . . . . . . . . . . . . . . 99
Lemma 14.1.8: The component lemma . . . . . . . . . . . . . . . . . . . . . . . 101
14.2 Cauchy global integral theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Theorem 14.2.1: Cauchy global integral theorem . . . . . . . . . . . . . . . . . 101
14.3 Three corollaries of Cauchy global integral theorem . . . . . . . . . . . . . . . . 102
Corollary 14.3.3: The case Ω = ℂ ⧵ {0} . . . . . . . . . . . . . . . . . . . . . . 102
14.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
14.5 Deformation theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
14.6 Integrating rational functions on the real line . . . . . . . . . . . . . . . . . . . . 104

RHS, version May 11, 2019 2

Complex Analysis I CONTENTS

14.7 The proof for the Cauchy global integral formula . . . . . . . . . . . . . . . . . 105

15 The extended complex plane 109

15.1 The Riemann sphere and the extended complex plane . . . . . . . . . . . . . . . 109
15.2 Circlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
15.3 The extended complex plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
15.3.1 Behavior of sets at ∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
15.4 Behavior of functions at ∞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

16 Möbius transformations 113

16.1 The image of a circline is a circline . . . . . . . . . . . . . . . . . . . . . . . . . 114
16.2 Finding the image of a circline . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
16.3 The image of a disc and the image of a half-plane . . . . . . . . . . . . . . . . . 115
16.4 Cross-ratios and the triplet representation of a Möbius transformation . . . . . . 116
16.5 Möbius transformations preserve angles . . . . . . . . . . . . . . . . . . . . . . 118

17 Conformal mappings 120

17.1 On conformal mappings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
17.1.4 Conformality at ∞ as analyticity at ∞ . . . . . . . . . . . . . . . . . . . 120
17.2 On conformal mapping problems . . . . . . . . . . . . . . . . . . . . . . . . . . 122

References 124

Alphabetical Index 125

RHS, version May 11, 2019 3

Complex Analysis I CONTENTS

2019 Foreword
These lecture notes form the material to the elementary course on Complex Analysis which I have
taught at the University of Helsinki in 2018-2019. The purpose has been to introduce the founda-
tion of the analysis of complex valued functions defined on the complex plane: Differentiability,
power series expansions, and integration of functions along suitable curves. As an application of
Cauchy’s global integral theorem we have studied integration of rational functions on the real line
as well as integration of trigonometric-rational forms on the real line. For mapping properties we
have considered especially Möbius transformations and their invariance properties.

These notes are based on my lecture notes in Finnish which were based on Kari Astala’s Lecture
Notes and Terry Tao’s Lecture Notes. But I have benefitted also from H. A. Priestley’s book
Introduction to Complex Analysis as well as Elias Stein’s and Rami Shakarchi’s book on Complex
Analysis and also some other sources which are mentioned in the reference list.

I wish to thank my students for attending the lectures and homework class. I wish to express my
gratitude especially to those students who were very dedicated to the course the whole semester.
So thank you- Alicia, Aleksis, Eetu, Elli, Fatima, Henri, Janne, Jose, Joonas, Khalid, Laura,
Lauri, Matti, Miika, Mikael, Miko, Outi, Paul, Pekka, Riku, Risto, Robert, Sara, Sauli, Simo,
Toivo, Tuomas, Ville and Ville. I wish to thank Outi Boman for being an excellent scribe and
organizer and supervisor to the homework writing groups. Also, I am extremely grateful to Ilmari
Lehmusoksa for drawing excellent pictures.

University of Helsinki, Finland

May 2019

Ritva Hurri-Syrjänen

RHS, version May 11, 2019 4

Complex Analysis I CONTENTS

2018 Foreword
These lecture notes form the material to the elementary course on Complex Analysis which I
taught at the University of Helsinki in winter and spring 2018. The purpose has been to introduce
the foundation of the analysis of complex valued functions defined on the complex plane: Dif-
ferentiability, power series expansions, and integration of functions along suitable curves. As an
application of Cauchy’s global integral theorem we have studied integration of rational functions
on the real line. For mapping properties we have considered especially Möbius transformations
and their invariance properties.

These notes are based on my lecture notes in Finnish which were based on Kari Astala’s Lecture
Notes and Terry Tao’s Lecture Notes. But I have benefitted also from H. A. Priestley’s book
Introduction to Complex Analysis as well as Elias Stein’s and Rami Shakarchi’s book on Complex
Analysis and also some other sources which are mentioned in the reference list.

I wish to thank my students for attending the lectures and homework class, and especially for their
encouragement and help for preparing these lecture notes using LATEX. So thank you- Antti J.,
Antti M., Atte, Dennis, Eloy, Emil, Francisco, Heli, Ilmari, Jaakko, Joni, Joonas, Luukas, Mats,
Miika, Mirjeta, Olli, Osku, Tapio, Tomi, Tommi, and Tuuli. I wish to express my gratitude to
Ilmari Lehmusoksa for being an excellent organizer and supervisor for those who typed my lecture
notes using LATEX. I am also extremely grateful to Ilmari Lehmusoksa for drawing excellent
pictures to these lecture notes.

University of Helsinki, Finland

May 2018

Ritva Hurri-Syrjänen

RHS, version May 11, 2019 5

Complex Analysis I 1 BACKGROUND

1 Background
1.1 The vector space ℝ2
We have studied the 2-dimensional vector space,
{ }
ℝ2 = (𝑥1 , 𝑥2 ) ∣ 𝑥1 ∈ ℝ, 𝑥2 ∈ ℝ ,

where ℝ is the set of real numbers. The vector space ℝ2 is equipped with two operations: addition
between elements of ℝ2 and scalar multiplication between an element of ℝ2 and an element of
ℝ. That is, for every 𝑎 = (𝑎1 , 𝑎2 ) and 𝑏 = (𝑏1 , 𝑏2 ),

𝑎 + 𝑏 = (𝑎1 , 𝑎2 ) + (𝑏1 , 𝑏2 ) = (𝑎1 + 𝑏1 , 𝑎2 + 𝑏2 )

and for 𝜆 ∈ ℝ,
𝜆𝑎 = 𝜆(𝑎1 , 𝑎2 ) = (𝜆𝑎1 , 𝜆𝑎2 ).

Proposition 1.1.1. Recall that (ℝ2 , +) is an Abelian group.

Proof.
(A1) For all 𝑎, 𝑏 ∈ ℝ2 we have 𝑎 + 𝑏 ∈ ℝ2 .
(A2) For all 𝑎, 𝑏, 𝑐 ∈ ℝ2 , (𝑎 + 𝑏) + 𝑐 = 𝑎 + (𝑏 + 𝑐).
(A3) There exists a unique 0 ∈ ℝ2 , 0 = (0, 0), such that 𝑎 + 0 = 0 + 𝑎 = 𝑎 for every 𝑎 ∈ ℝ2 .
(A4) For every 𝑎 = (𝑎1 , 𝑎2 ) ∈ ℝ2 there exists exactly one −𝑎 ∈ ℝ2 ,
−𝑎 = −(𝑎1 , 𝑎2 ) = −1(𝑎1 , 𝑎2 ) = (−𝑎1 , −𝑎2 ), such that 𝑎 + (−𝑎) = −𝑎 + 𝑎 = 0.
(A5) For all 𝑎, 𝑏 ∈ ℝ2 : 𝑎 + 𝑏 = 𝑏 + 𝑎.

Remarks 1.1.2.
(A2) the operation + is associative.
(A3) the element 0 is called zero element.
(A4) the element −𝑎 is called the additive inverse of 𝑎
(A5) since the operator is commutative, (ℝ2 , +) is an Abelian group.

𝑦
𝑎+𝑏=𝑏+𝑎
𝑏
𝑏−𝑎
1
𝑎

0 1 𝑥

𝑎−𝑏

Figure 1.1: Addition and subtraction in the vector space (ℝ2 , +) can be interpreted geometrically
via the parallelogram law.

Remark 1.1.3. The vector space (ℝ2 , +) is the Euclidean plane.

Remark 1.1.4. Recall that (ℝ2 , +) is a real vector space, since (ℝ2 , +) is an Abelian group and
the following statements hold:
(V1) for all 𝜆 ∈ ℝ, 𝑣 ∈ ℝ2 ∶ 𝜆𝑣 ∈ ℝ2 .

RHS, version May 11, 2019 6

Complex Analysis I 1 BACKGROUND

(V2) for all 𝜆, 𝜇 ∈ ℝ, 𝑣 ∈ ℝ2 ∶ (𝜆𝜇)𝑣 = 𝜆(𝜇𝑣).

(V3) for the unit element 1 ∈ ℝ and every 𝑣 ∈ ℝ2 ∶ 1𝑣 = 𝑣.
(V4) for all 𝜆 ∈ ℝ2 and for all 𝑣, 𝑢 ∈ ℝ2 ∶ 𝜆(𝑣 + 𝑢) = 𝜆𝑣 + 𝜆𝑢.
(V5) for all 𝜆, 𝜇 ∈ ℝ and 𝑣 ∈ ℝ2 ∶ (𝜆 + 𝜇)𝑣 = 𝜆𝑣 + 𝜇𝑣.

Definition 1.1.5. Recall that we have introduced the inner product between 𝑎 = (𝑎1 , 𝑎2 ) ∈ ℝ2
and 𝑏 = (𝑏1 , 𝑏2 ) ∈ ℝ2 such that ( )
𝑎|𝑏 = 𝑎1 𝑏1 + 𝑎2 𝑏2 .
This is a mapping ℝ2 × ℝ2 → ℝ. Another notation is ⟨𝑎|𝑏⟩.

Remark 1.1.6. The notation 𝑎 ⋅ 𝑏 is not used in this Complex Analysis course, since it is reserved
for the multiplication which makes (ℝ2 , +, ⋅) a field.

1.2 The complex plane ℂ

We introduce multiplication between any two elements 𝑎 and 𝑏 from ℝ2 such that the multiplica-
tion takes as input 𝑎 ∈ ℝ2 and 𝑏 ∈ ℝ2 and gives as output 𝑐 ∈ ℝ2 . We define ⋅ ∶ ℝ2 × ℝ2 → ℝ2
such that
(𝑎1 , 𝑎2 ) ⋅ (𝑏1 , 𝑏2 ) = (𝑎1 𝑏1 − 𝑎2 𝑏2 , 𝑎1 𝑏2 + 𝑎2 𝑏1 )
for every pair 𝑎 = (𝑎1 , 𝑎2 ) ∈ ℝ2 and 𝑏 = (𝑏1 , 𝑏2 ) ∈ ℝ2 . In doing so we obtain that (ℝ2 , +, ⋅) is a
field.
We write ℂ = (ℝ2 , +, ⋅) and call ℂ the complex plane.

Remark 1.2.1. Here lies the difference between our vector calculus (Vektorianalyysi II) course
and this first complex analysis course.

Remark 1.2.2. Multiplication is a little bit messy in Cartesian co-ordinates, but it becomes nicer
in polar form, which we introduce later. This polar form gives a geometric meaning for multipli-
cation of two vectors.

In the vector calculus course we studied functions 𝑓 ∶ ℝ2 → ℝ2 , where ℝ2 is the Euclidean plane.
We looked at their differentiability and the properties of their integrals.
Now we study functions 𝑓 ∶ ℂ → ℂ, where ℂ = (ℝ2 , +, ⋅) is the complex plane.
It turns out that for a function 𝑓 ∶ ℝ2 → ℝ2 to have a derivative at every point in an open ball in
ℝ2 , that is, to be real differentiable in an open ball, is a weaker property than to have a complex
derivative in every point of an open ball, that is, to be complex differentiable in an open ball.
In summary: to be real differentiable in an open ball is easier than to be complex differentiable
in the corresponding ball.

Example 1.2.3. Let 𝑓 ∶ ℝ2 → ℝ2 , (𝑥, 𝑦) ↦ (𝑥, −𝑦). The mapping 𝑓 is real differentiable on the
whole plane ℝ2 , since its partial derivatives exist and are continuous.
However, it turns out that
𝑓 ∶ ℂ → ℂ, (𝑥, 𝑦) ↦ (𝑥, −𝑦),
is not complex differentiable.

Before studying the complex differentiability and analyticity of functions, it will be useful to go
through the elementary properties of the elements of ℂ.

Remark 1.2.4. We give an outline of why (ℝ2 , +, ⋅) =∶ ℂ is a field.

(F1) (ℝ2 , +) is an Abelian group.
(F2) (ℝ2 ⧵ {0}, ⋅) is an Abelian group:
(B1) For all 𝑎 ∈ ℝ2 and 𝑏 ∈ ℝ2 : 𝑎 ⋅ 𝑏 ∈ ℝ2 .

RHS, version May 11, 2019 7

Complex Analysis I 1 BACKGROUND

(B2) For all 𝑎, 𝑏, 𝑐 ∈ ℝ2 : (𝑎 ⋅ 𝑏) ⋅ 𝑐 = 𝑎 ⋅ (𝑏 ⋅ 𝑐).

(B3) There exists (1, 0) ∈ ℝ2 such that for every 𝑎 ∈ ℝ2 :

𝑎 ⋅ (1, 0) = 𝑎 = (1, 0) ⋅ 𝑎.

(B4) For each 𝑎 ∈ ℝ2 ⧵ {0}, 𝑎 = (𝑎1 , 𝑎2 ), there exists a unique inverse element of 𝑎,
( )
𝑎1 −𝑎2
, =∶ 𝑎−1 ∈ ℝ2
2 2 2
𝑎1 + 𝑎2 𝑎1 + 𝑎2 2

( )
such that 𝑎 ⋅ 𝑎−1 = 𝑎−1 ⋅ 𝑎 = 1, 0 =∶ 1.
(B5) For all 𝑎, 𝑏 ∈ ℝ2 : 𝑎 ⋅ 𝑏 = 𝑏 ⋅ 𝑎.
(F3) For all 𝑎, 𝑏, 𝑐 ∈ ℝ2 :
(𝑎 + 𝑏) ⋅ 𝑐 = 𝑎 ⋅ 𝑐 + 𝑏 ⋅ 𝑐.
This is called the 1st distributional law.

Remarks 1.2.5.
(B2) The operation ⋅ is associative.
(B3) The element (1, 0) is called the unit element.
Note: (𝑎1 , 𝑎2 ) ⋅ (1, 0) = (𝑎1 ⋅ 1 − 𝑎2 ⋅ 0, 𝑎1 ⋅ 0 + 𝑎2 ⋅ 1) = (𝑎1 , 𝑎2 ) and (B5) yields

(1, 0) ⋅ (𝑎1 , 𝑎2 ) = (𝑎1 , 𝑎2 ).

(B4) The point 𝑎−1 is the inverse of 𝑎 and we also write 𝑎1 .

Let 𝑎 = (𝑎1 , 𝑎2 ) ∈ ℝ2 ⧵ {0}. We have to find 𝑥 = (𝑥1 , 𝑥2 ) ∈ ℝ2 ⧵ {0} such that

𝑥 ⋅ 𝑎 = 𝑎 ⋅ 𝑥 = (1, 0),

that is,
𝑥 ⋅ 𝑎 = (𝑥1 , 𝑥2 ) ⋅ (𝑎1 , 𝑎2 ) = (𝑥1 𝑎1 − 𝑥2 𝑎2 , 𝑥1 𝑎2 + 𝑥2 𝑎1 ) = (1, 0).

This holds if and only if

{
𝑎1 𝑥1 − 𝑎2 𝑥2 = 1 and
𝑎1 𝑥2 + 𝑎2 𝑥1 = 0;

here 𝑎21 + 𝑎22 > 0. There exists exactly one such (𝑥1 , 𝑥2 ); since the determinant

|𝑎 −𝑎2 ||
| 1
| | = 𝑎21 + 𝑎22 ≠ 0;
|𝑎2 𝑎1 |
| |
the corresponding linear system has a unique solution:
( )
𝑎1 −𝑎2
(𝑥1 , 𝑥2 ) = , .
𝑎21 + 𝑎22 𝑎21 + 𝑎22

Remarks 1.2.6.
1. We sometimes write 𝑧1 ⋅ 𝑧2 = 𝑧1 𝑧2 , where 𝑧1 ∈ ℂ and 𝑧2 ∈ ℂ.
2. Let 𝑧 ∈ ℂ. We define 𝑧0 = (1, 0) and for each 𝑛 ∈ ℕ: 𝑧𝑛 = 𝑧 ⋅ 𝑧𝑛−1 .

RHS, version May 11, 2019 8

Complex Analysis I 1 BACKGROUND

Remark 1.2.7. The set

{ }
(𝑥, 0) ∣ 𝑥 ∈ ℝ ⊂ ℂ
{ }
is a subfield of the field ℂ. There is a mapping 𝑓 ∶ ℝ → (𝑥, 0) ∣ 𝑥 ∈ ℝ , which is a field
isomorphism. This means that we can identify 𝑥 ∈ ℝ with (𝑥, 0) ∈ ℂ. Hence ℝ ⊂ ℂ, that is, the
set of real numbers is a proper subset of the set of complex numbers.
( )
Hence especially we can write 1, 0 = 1.

Definition 1.2.8. The elements of ℂ are called complex numbers.

Definition 1.2.9. The complex number (0, 1) ∈ ℂ is called the imaginary unit and we write

(0, 1) =∶ 𝑖.

Remark 1.2.10. Note that

𝑖2 = 𝑖 ⋅ 𝑖 = (0, 1) ⋅ (0, 1) = (0 ⋅ 0 − 1 ⋅ 1, 0 ⋅ 1 + 1 ⋅ 0) = (−1, 0) = −1,

but we cannot take any roots, yet. We have not defined what the square root of 𝑧 ∈ ℂ is.

RHS, version May 11, 2019 9

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

2 The elementary properties of complex numbers

2.1 Complex numbers can be written in Cartesian form
Proposition 2.1.1. Every complex number 𝑧 ∈ ℂ can be expressed in a unique way 𝑧 = 𝑥 + 𝑖𝑦,
where 𝑥, 𝑦 ∈ ℝ.

Proof. Let 𝑧 ∈ ℂ be fixed. Then 𝑧 = (𝑥, 𝑦) for some 𝑥, 𝑦 ∈ ℝ. In particular,

𝑧 = (𝑥, 𝑦) = (𝑥, 0) + (0, 𝑦) = (1, 0)(𝑥, 0) + (0, 1)(𝑦, 0) = (𝑥, 0) + 𝑖(𝑦, 0) = 𝑥 + 𝑖𝑦.
It is unique:
Let 𝑧 = 𝑥 + 𝑖𝑦 = 𝑎 + 𝑖𝑏, where 𝑥, 𝑦, 𝑎, 𝑏 ∈ ℝ. Hence,
(𝑥, 0) + (0, 1)(𝑦, 0) = (𝑎, 0) + (0, 1)(𝑏, 0).
By calculating the products we get
(𝑥, 0) + (0 ⋅ 𝑦 − 1 ⋅ 0, 0 ⋅ 0 + 1 ⋅ 𝑦) = (𝑎, 0) + (0 ⋅ 𝑏 − 1 ⋅ 0, 0 ⋅ 0 + 1 ⋅ 𝑏),
which implies that (𝑥, 𝑦) = (𝑎, 𝑏). Thus 𝑥 = 𝑎 and 𝑦 = 𝑏.

Definition 2.1.2. Let 𝑧 = 𝑥 + 𝑖𝑦, where 𝑥, 𝑦 ∈ ℝ are fixed. The real part of 𝑧 is
𝑥 =∶ Re(𝑧) ∈ ℝ
and the imaginary part of 𝑧 is
𝑦 =∶ Im(𝑧) ∈ ℝ.
The set {𝑧 ∈ ℂ ∣ Im(𝑧) = 0} is the real axis and the set {𝑧 ∈ ℂ ∣ Re(𝑧) = 0} is the imaginary
axis.
The complex conjugate of 𝑧 is defined by
𝑧 = 𝑥 − 𝑖𝑦,
and the magnitude of 𝑧 is √
|𝑧| = 𝑥2 + 𝑦2 .
Imaginary axis
𝑧

𝑖 |𝑧|

0 1 Real axis

𝑧̄

Figure 2.1: Complex number 𝑧, its magnitude|𝑧| and its complex conjugate 𝑧.

Remark 2.1.3. Two complex numbers are equal if and only if they have the same real part and
the same imaginary part.

Example 2.1.4.
√
𝑧= 3 + 𝑖,
√
𝑧 = 3 − 𝑖,
√
(√ ) 2
|𝑧| = 3 + 12 = 2,
√
Re(𝑧) = 3,
Im(𝑧) = 1

RHS, version May 11, 2019 10

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

Imaginary axis
√
𝑖 || 3+𝑖
√ 𝑖|
|| 3 + |
||

0 1 Real axis

√
3−𝑖

√ |√ |
Figure 2.2: Complex number 3 + 𝑖 from example 2.1.4, its magnitude || 3 + 𝑖|| = 2 and its
√ | |
complex conjugate 3 − 𝑖.

Basic complex operations:

Proposition 2.1.5. For all 𝑧1 , 𝑧2 , 𝑧 ∈ ℂ:

(𝑖) 𝑧1 + 𝑧2 = 𝑧1 + 𝑧2 ,
(𝑖𝑖) 𝑧1 ⋅ 𝑧2 = 𝑧1 ⋅ 𝑧2 ,
(𝑖𝑖𝑖) 𝑧 = 𝑧.

Proof. Let 𝑧1 = 𝑥1 + 𝑖𝑦1 , 𝑧2 = 𝑥2 + 𝑖𝑦2 and 𝑧 = 𝑥 + 𝑖𝑦, where 𝑥1 , 𝑥2 , 𝑦1 , 𝑦2 , 𝑥, 𝑦 ∈ ℝ.

(𝑖) 𝑧1 + 𝑧2 = 𝑥1 − 𝑖𝑦1 + 𝑥2 − 𝑖𝑦2 = (𝑥1 + 𝑥2 ) − 𝑖(𝑦1 + 𝑦2 ) = 𝑧1 + 𝑧2 ,

( )
(𝑖𝑖) 𝑧1 ⋅ 𝑧2 = (𝑥1 , −𝑦1 )(𝑥2 , −𝑦2 ) = 𝑥1 𝑥2 − 𝑦1 𝑦2 , −(𝑥1 𝑦2 + 𝑥2 𝑦1 ) = 𝑧1 ⋅ 𝑧2 ,
( )
(𝑖𝑖𝑖) 𝑧 = (𝑥, −𝑦) = 𝑥, −(−𝑦) = (𝑥, 𝑦) = 𝑧.

Note: given 𝑧1 = 𝑥1 + 𝑖𝑦1 and 𝑧2 = 𝑥2 + 𝑖𝑦2 , we have 𝑧1 + 𝑧2 = (𝑥1 + 𝑥2 ) + 𝑖(𝑦1 + 𝑦2 ).

The proofs to the following propositions are similar direct calculations as in the previous one, and
are left as an exercise to the reader.

Proposition 2.1.6. For all 𝑧 ∈ ℂ:

|𝑧| = |−𝑧| ,
|𝑧| = |𝑧| ,
| |
𝑧 ⋅ 𝑧 = |𝑧|2 .

Remark 2.1.7. 𝑧 ⋅ 𝑧 = |𝑧|2 , thus if 𝑧 ∈ ℂ ⧵ {0}, then

1 𝑧 𝑧
𝑧−1 = = = .
𝑧 𝑧 ⋅ 𝑧 |𝑧|2

Proposition 2.1.8. For every 𝑧 ∈ ℂ,

1
Re(𝑧) = (𝑧 + 𝑧) and
2
1
Im(𝑧) = (𝑧 − 𝑧).
2𝑖

RHS, version May 11, 2019 11

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

The magnitude|𝑧| behaves well with respect to the multiplication and division operations:

Proposition 2.1.9.
|𝑧 ⋅ 𝑧 | = |𝑧 ||𝑧 | , for all 𝑧1 , 𝑧2 ∈ ℂ,
| 1 2 | | 1 || 2 |
| 𝑧 | |𝑧 |
| 1 | | 1|
| |= , for all 𝑧1 ∈ ℂ and 𝑧2 ∈ ℂ ⧵ {0},
| 𝑧2 | |𝑧2 |
| | | |
|𝑧|𝑛 = ||𝑧𝑛 || , for all 𝑧 ∈ ℂ.

With respect to addition and subtraction, we have the triangle inequality:

Theorem 2.1.10: The triangle inequality.

|𝑧1 + 𝑧2 | ≤ |𝑧1 | + |𝑧2 | , for all 𝑧1 , 𝑧2 ∈ ℂ.
| | | | | |
More generally,
|| | | || |
||𝑧1 | − |𝑧2 || ≤ |𝑧1 + 𝑧2 || ≤ ||𝑧1 || + ||𝑧2 || , for all 𝑧1 , 𝑧2 ∈ ℂ.
| |
Proof.
( )
|𝑧1 + 𝑧2 |2 = 𝑧1 + 𝑧2 ⋅ (𝑧1 + 𝑧2 ) = (𝑧1 + 𝑧2 ) ⋅ (𝑧1 + 𝑧2 ) = 𝑧1 𝑧1 + 𝑧1 𝑧2 + 𝑧1 𝑧2 + 𝑧2 𝑧2
| |
( )
= ||𝑧1 || + ||𝑧2 || + 𝑧1 𝑧2 + 𝑧1 𝑧2 = ||𝑧1 || + ||𝑧2 || + 2 Re 𝑧1 𝑧2
2 2 2 2

≤ ||𝑧1 || + ||𝑧2 || + 2||𝑧1 ||||𝑧2 || = (||𝑧1 || + ||𝑧2 ||) ,

2 2 2

which implies ||𝑧1 + 𝑧2 || ≤ ||𝑧1 || + ||𝑧2 ||.

For the lower inequality, we have
|𝑧1 | = |𝑧1 − 𝑧2 + 𝑧2 | ≤ |𝑧1 − 𝑧2 | + |𝑧2 | ,
| | | | | | | |
which implies
|𝑧1 | − |𝑧2 | ≤ |𝑧1 − 𝑧2 | , for all 𝑧1 , 𝑧2 ∈ ℂ.
| | | | | |
Similarly,
|𝑧 | = |𝑧 − 𝑧 + 𝑧 | ≤ |𝑧 − 𝑧 | + |𝑧 | .
| 2| | 2 1 1| | 2 1| | 1|

Note that ||𝑧2 − 𝑧1 || = ||𝑧1 − 𝑧2 ||, which gives us

|𝑧2 | − |𝑧1 | ≤ |𝑧1 − 𝑧2 | , for all 𝑧1 , 𝑧2 ∈ ℂ.
| | | | | |
| |
Hence we have shown that |||𝑧1 || − ||𝑧2 ||| ≤ ||𝑧1 + 𝑧2 ||.
| |

2.2 Complex numbers can be written in polar form

Definition 2.2.1: Polar coordinates. Let us recall the connection between Cartesian coordinates
and polar coordinates from the vector calculus course.
√
If the point (𝑥, 𝑦) ∈ ℝ2 ⧵ {0} is given, we can find 𝑟 = 𝑥2 + 𝑦2 > 0 and 𝛼 ∈ [0, 2𝜋], counting
counter-clockwise, such that
√
⎧𝑟 = 𝑥2 + 𝑦2 ,
⎪
⎨𝑥 = 𝑟 cos 𝛼,
⎪𝑦 = 𝑟 sin 𝛼.
⎩
Conversely, if 𝑟 > 0 and 𝛼 ∈ [0, 2𝜋] are given, we can find 𝑥 and 𝑦.

RHS, version May 11, 2019 12

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

1 (𝑥, 𝑦)
2 =𝑟
√ 𝑥2 +𝑦

𝛼
0 1 x

Figure 2.3: Polar coordinates 𝑟 and 𝛼 of vector (𝑥, 𝑦) ∈ ℝ2 ⧵ {0}.

Let us remove the restriction on 𝛼. Let 𝛼 ∈ ℝ and note that

{
cos 𝛼 = 𝑥𝑟 ,
sin 𝛼 = 𝑦𝑟 .

These equations determine 𝛼 only up to an integer multiple of 2𝜋.

A complex number 𝑧 ∈ ℂ ⧵ {0} can be written in polar form as

𝑧 = 𝑟 cos 𝛼 + 𝑖𝑟 sin 𝛼, (2.2.2)

where 𝑟 = |𝑧| > 0 is called the magnitude, modulus or absolute value of 𝑧 and 𝛼 ∈ ℝ is called the
argument or phase of 𝑧. The argument 𝛼 is, however, only determined up to an integer multiple
of 2𝜋.

Example 2.2.3.
√
𝑧= 3+𝑖
(√ )
𝜋
arg 3 + 𝑖 = + 𝑛2𝜋, 𝑛∈ℤ
6

Im
√
𝑖 𝑧= 3+𝑖
2 2
√
3
𝜋
𝛼= 6 𝜋 𝜋
0 1 Re 3 3
1 1
(a) The complex number 𝑧 and its argument. (b) An equilateral triangle.

Figure 2.4: The complex number 𝑧 from example 2.2.3 and an equilateral triangle, which can be
used to find the argument of 𝑧.

Note: The “set” of all possible arguments of 𝑧 is denoted by Arg(𝑧).

Theorem 2.2.4: Euler’s formula. The formula

cos 𝛼 + 𝑖 sin 𝛼 =∶ 𝑒𝑖𝛼 , 𝛼 ∈ ℝ,

is called Euler’s formula. We take it now as a notation, but we will prove it later.

2.2.5. The relationship between the Cartesian and the polar form is given by Euler’s formula:

𝑧 = 𝑥 + 𝑖𝑦 = 𝑟 cos 𝛼 + 𝑟𝑖 sin 𝛼 = 𝑟(cos 𝛼 + 𝑖 sin 𝛼) = 𝑟𝑒𝑖𝛼 .

RHS, version May 11, 2019 13

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

Remark 2.2.6. If 𝑟1 𝑒𝑖𝛼1 = 𝑟2 𝑒𝑖𝛼2 ≠ 0, then 𝑟1 = 𝑟2 and 𝛼1 = 𝛼2 + 𝑛2𝜋 for some 𝑛 ∈ ℤ.

2.2.7. There are infinitely many choices for the argument of any given complex number. To get
a unique argument we restrict the argument to lie in an interval (𝛼, 𝛽], where 𝛽 − 𝛼 = 2𝜋. The
standard argument Arg(−𝜋,𝜋] is defined as the unique argument, of a complex number 𝑧, that lies
in the interval (−𝜋, 𝜋]. For the general interval (𝛼, 𝛽], where 𝛽 − 𝛼 = 2𝜋, the argument Arg(𝛼,𝛽]
is defined as the unique argument of 𝑧 that lies in the interval (𝛼, 𝛽].

2.2.8. Recall the formulas for the functions sine and cosine when 𝛼, 𝛽 ∈ ℝ:

sin (𝛼 + 𝛽) = sin 𝛼 cos 𝛽 + cos 𝛼 sin 𝛽,

cos (𝛼 + 𝛽) = cos 𝛼 cos 𝛽 − sin 𝛼 sin 𝛽.

2.2.9 The geometric meaning of the multiplication of complex numbers

For two complex numbers 𝑧1 = 𝑟1 𝑒𝑖𝛼1 and 𝑧2 = 𝑟2 𝑒𝑖𝛼2 , by Euler’s formula, the definition of
multiplication and 2.2.8 we get

𝑧1 𝑧2 = 𝑟1 𝑟2 𝑒𝑖𝛼1 𝑒𝑖𝛼2 = 𝑟1 𝑟2 (cos 𝛼1 + 𝑖 sin 𝛼1 )(cos 𝛼2 + 𝑖 sin 𝛼2 )

( )
= 𝑟1 𝑟2 (cos 𝛼1 cos 𝛼2 − sin 𝛼1 sin 𝛼2 ) + 𝑖(cos 𝛼1 sin 𝛼2 + sin 𝛼1 cos 𝛼2 )
( )
= 𝑟1 𝑟2 cos (𝛼1 + 𝛼2 ) + 𝑖 sin (𝛼1 + 𝛼2 ) = 𝑟1 𝑟2 𝑒𝑖(𝛼1 +𝛼2 ) .

Notice the geometric meaning: when we multiply two complex numbers, the output is a complex
number whose magnitude we obtain by multiplying the magnitudes of the input complex numbers.
Likewise, the argument of the output complex number is obtained by summing the arguments of
the input complex numbers.
If 𝑧0 ∈ ℂ ⧵ {0} is a fixed complex number, then in the mapping 𝑧 ↦ 𝑧0 𝑧, the point 𝑧 is the input
and the output is
𝑧0 𝑧 = ||𝑧0 𝑧|| 𝑒𝑖(Arg 𝑧0 +Arg 𝑧) .
This means stretching and rotating in the complex plane ℂ.

Example 2.2.10. Let 𝑓 ∶ ℂ → ℂ, 𝑧 ↦ 𝑖𝑧. The mapping 𝑓 corresponds to a 90 degree rotation

counter-clockwise.

2.2.11. We have proved

𝑖(𝜑 +𝜑 )
𝑧1 𝑧2 = 𝑟1 𝑟2 𝑒1 1 2 ,
|𝑧1 𝑧2 | = |𝑧1 ||𝑧2 | , for all 𝑧1 , 𝑧2 ∈ ℂ,
| | | || |
where
𝑧1 = 𝑟1 𝑒𝑖𝜑1 , 𝑧2 = 𝑟2 𝑒𝑖𝜑2 ,
𝑟𝑗 > 0, 𝜑𝑗 ∈ ℝ, 𝑗 = 1, 2.
As well as ( ) ( ) ( )
Arg 𝑧1 𝑧2 = Arg 𝑧1 + arg 𝑧2 , for all 𝑧1 , 𝑧2 ∈ ℂ ⧵ {0}.

Remark 2.2.12. Let 𝛼 ∈ ℝ and 𝑧1 , 𝑧2 ∈ ℂ ⧵ {0}. It might be the case that

Arg(𝛼,𝛼+2𝜋] (𝑧1 𝑧2 ) ≠ Arg(𝛼,𝛼+2𝜋] (𝑧1 ) + Arg(𝛼,𝛼+2𝜋] (𝑧2 ).

Example. Let 𝑧1 = −𝑖 = 𝑧2 . Then

𝜋
Arg(−𝜋,𝜋] (𝑧1 ) = Arg(−𝜋,𝜋] (𝑧2 ) = −
2
and
Arg(−𝜋,𝜋] (𝑧1 𝑧2 ) = Arg(−𝜋,𝜋] (−1) = 𝜋.

RHS, version May 11, 2019 14

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

Hence,
( )
𝜋 𝜋
Arg(−𝜋,𝜋] (𝑧1 𝑧2 ) = 𝜋 ≠ − + − = −𝜋 = Arg(−𝜋,𝜋] (𝑧1 ) + Arg(−𝜋,𝜋] (𝑧2 ).
2 2

Im
𝑖

𝜋
−1
0 1 Re
− 𝜋2

−𝑖

Figure 2.5: About Remark 2.2.12: 𝑧1 = −𝑖 = 𝑧2 and Arg(−𝜋,𝜋] (𝑧1 ) = Arg(−𝜋,𝜋] (𝑧2 ) = − 𝜋2 .
( )
𝑧1 𝑧2 = (−𝑖)(−𝑖) = −1 but Arg(−𝜋,𝜋] (𝑧1 𝑧2 ) = 𝜋 ≠ Arg(−𝜋,𝜋] (𝑧1 )+Arg(−𝜋,𝜋] (𝑧2 ) = − 𝜋2 + − 𝜋2 =
−𝜋.

Proposition 2.2.13. For all 𝑧1 ∈ ℂ and 𝑧2 ∈ ℂ ⧵ {0}

| 𝑧 | |𝑧 |
| 1 | | 1|
| |= .
| 𝑧2 | |𝑧2 |
| | | |
Furthermore, if 𝑧1 ∈ ℂ ⧵ {0}, then
( )
𝑧1
arg = arg(𝑧1 ) − arg(𝑧2 ).
𝑧2

Proof. Since
| | | |
|𝑧 | = || 𝑧1 𝑧 || = || 𝑧1 |||𝑧 | ,
| 1 | | 𝑧 2 | | 𝑧 || 2 |
| 2 | | 2|
we obtain
| 𝑧 | |𝑧 |
| 1 | | 1|
| |= .
| 𝑧2 | |𝑧2 |
| | | |
Also, ( ) ( )
𝑧1 𝑧1
arg(𝑧1 ) = arg 𝑧 = arg + arg(𝑧2 ).
𝑧2 2 𝑧2

Remark 2.2.14. Let 𝛼 ∈ ℝ and 𝑧1 , 𝑧2 ∈ ℂ ⧵ {0}. It might be the case that

( )
𝑧1
Arg(𝛼,𝛼+2𝜋] ≠ Arg(𝛼,𝛼+2𝜋] (𝑧1 ) − Arg(𝛼,𝛼+2𝜋] (𝑧2 ).
𝑧2

Example. Let 𝑧1 = 1 and 𝑧2 = −2. Then,

Arg(−𝜋,𝜋] (𝑧1 ) = 0,
Arg(−𝜋,𝜋] (𝑧2 ) = 𝜋 and
( ) ( )
𝑧1 1
Arg(−𝜋,𝜋] = Arg(−𝜋,𝜋] − = 𝜋.
𝑧2 2
Hence, ( )
1
Arg(−𝜋,𝜋] − = 𝜋 ≠ 0 − 𝜋 = Arg(−𝜋,𝜋] (𝑧1 ) − Arg(−𝜋,𝜋] (𝑧2 ).
2

RHS, version May 11, 2019 15

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

Im

𝜋
−2
0 1 Re

Figure 2.6: About Remark 2.2.14: 𝑧1 = 1 and 𝑧2 = −2, so Arg(−𝜋,𝜋] (𝑧1 ) = 0 and Arg(−𝜋,𝜋] (𝑧2 ) =
( )
𝑧
𝜋. So 𝑧1 = − 12 , but Arg(−𝜋,𝜋] − 12 = 𝜋 ≠ 0 − 𝜋 = Arg(−𝜋,𝜋] (𝑧1 ) − Arg(−𝜋,𝜋] (𝑧2 ).
2

Theorem 2.2.15: De Moivre’s theorem. Let 𝛼 ∈ ℝ. Then for any 𝑛 ∈ ℤ

( )𝑛
cos 𝛼 + 𝑖 sin 𝛼 = cos 𝑛𝛼 + 𝑖 sin 𝑛𝛼.

Proof. Homework 1.4.

2.2.16 Applications of De Moivre’s theorem

1. Homework 1.5.(2)
2. Trigonometric formulas:

Example.
cos 2𝛽 = 2 cos2 𝛽 − 1, 𝛽 ∈ ℝ.
Proof.
( )
cos 2𝛽 = Re cos 2𝛽 + 𝑖 sin 2𝛽
(( )2 )
= Re cos 𝛽 + 𝑖 sin 𝛽
( )
= Re cos2 𝛽 − sin2 𝛽 + 𝑖2 cos 𝛽 sin 𝛽

= cos2 𝛽 − sin2 𝛽
( )
= cos2 𝛽 − 1 − cos2 𝛽
= 2 cos2 𝛽 − 1.

3. The study of the 𝑛th roots of complex numbers

2.3 The 𝑛th roots of complex numbers

Let 𝑛 ≥ 2, 𝑛 ∈ ℕ. A complex number 𝑧 ∈ ℂ ⧵ {0} is called an 𝑛th root of a complex number
𝑎 ∈ ℂ ⧵ {0}, if
𝑧𝑛 = 𝑎.

Proposition 2.3.1. Let ( )

𝑎 = |𝑎| cos 𝜃 + 𝑖 sin 𝜃 ≠ 0.
If
𝑛 ∈ ℕ, 𝑛 ≥ 2,

RHS, version May 11, 2019 16

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

then 𝑎 has the 𝑛 distinct 𝑛th roots

( )
1 𝜃 + 𝑘2𝜋 𝜃 + 𝑘2𝜋
𝑧𝑘 = |𝑎| 𝑛 cos + 𝑖 sin , 𝑘 = 0, 1, … , 𝑛 − 1.
𝑛 𝑛

Remarks.
1
1. Each of the 𝑛th roots of 𝑎 has modulus |𝑎| 𝑛 . Hence, all the 𝑛th roots of 𝑎 lie on a circle of
1
radius|𝑎| 𝑛 in the complex plane.
2. Since the argument of each successive 𝑛th root exceeds the argument of the previous root
by 2𝜋
𝑛
, the 𝑛th roots are equally spaced on this circle.

Proof. Let ( )
𝑎 = |𝑎| cos 𝜃 + 𝑖 sin 𝜃 ≠ 0.
Our goal is to find all complex numbers
( )
𝑧 = |𝑧| cos 𝜑 + 𝑖 sin 𝜑

such that
𝑧𝑛 = 𝑎.
De Moivre’s theorem implies that
( ) ( )𝑛 ( )
|𝑎| cos 𝜃 + 𝑖 sin 𝜃 = |𝑧|𝑛 cos 𝜑 + 𝑖 sin 𝜑 = |𝑧|𝑛 cos 𝑛𝜑 + 𝑖 sin 𝑛𝜑 .

Hence,

⎧|𝑧|𝑛 =𝑎
⎪
⎨cos 𝑛𝜑 = cos 𝜃
⎪sin 𝑛𝜑 = sin 𝜃
⎩
and we obtain
⎧ 1
⎪|𝑧| = |𝑎| 𝑛
⎨
⎪𝑛𝜑 = 𝜃 + 𝑘2𝜋, 𝑘 ∈ ℤ.
⎩
that is
⎧ 1
⎪|𝑧| = |𝑎| 𝑛
⎨ 𝜃+𝑘2𝜋
⎪𝜑 = 𝑛 , 𝑘 ∈ ℤ.
⎩
Since by choosing 𝑘 = 0, 1, … , 𝑛 − 1, we obtain different values for 𝑧, we have
( )
1 𝜃 + 𝑘2𝜋 𝜃 + 𝑘2𝜋
𝑧𝑘 = |𝑎| 𝑛 cos + 𝑖 sin , where 𝑘 = 0, 1, … , 𝑛 − 1.
𝑛 𝑛

On the other hand, the numbers 𝑧𝑘 satisfy the equation 𝑧𝑛𝑘 = 𝑎.

Example 2.3.2. Find the fourth roots of 𝑎 = −1. That is, solve the equation

𝑧4 = −1,
𝑧4 + 1 = 0.

Solution. Since

𝑎 = −1 = cos 𝜋 + 𝑖 sin 𝜋,
𝜋 + 𝑘2𝜋 𝜋 + 𝑘2𝜋
𝑧𝑘 = cos + 𝑖 sin , where 𝑘 = 0, 1, 2, 3.
4 4

RHS, version May 11, 2019 17

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

That is, as shown in Figure 2.7,

𝜋 𝜋 1+𝑖 3𝜋 3𝜋 −1 + 𝑖
𝑧0 = cos + 𝑖 sin = √ , 𝑧1 = cos + 𝑖 sin = √ ,
4 4 2 4 4 2
5𝜋 5𝜋 −1 − 𝑖 7𝜋 7𝜋 1−𝑖
𝑧2 = cos + 𝑖 sin = √ and 𝑧3 = cos + 𝑖 sin = √ .
4 4 2 4 4 2

Im
𝑖
𝑧1 𝑧0

0 1 Re

𝑧2 𝑧3

Figure 2.7: The four fourth roots of 𝑎 = −1.

2.3.3 Special case: the 𝑛th roots of unity

The solutions of the equation
𝑧𝑛 = 1, 𝑛 ∈ ℕ ⧵ {0},
are called the 𝑛th roots of unity and are given by

𝑘2𝜋 𝑘2𝜋 𝑘2𝜋

𝑧𝑘 = cos + 𝑖 sin = 𝑒𝑖 𝑛 , 𝑘 = 0, 1, … , 𝑛 − 1.
𝑛 𝑛
The roots represent the 𝑛 vertices of a regular polygon of 𝑛 sides inscribed in a circle of radius 1
with center at the origin.

Example 2.3.4. Find fourth roots of unity.

Solution.
𝑧4 = 1 ⇔ 𝑧4 − 1 = 0.
We notice that
( )( ) ( )( )( )( )
𝑧4 − 1 = 𝑧2 − 1 𝑧2 + 1 = 𝑧 − 1 𝑧 + 1 𝑧 − 𝑖 𝑧 + 𝑖 .

So the equation becomes

( )( )( )( )
𝑧−1 𝑧 + 1 𝑧 − 𝑖 𝑧 + 𝑖 = 0.

And the four fourth roots of unity, shown in Figure 2.8, are 1, 𝑖, −1 and −𝑖.

RHS, version May 11, 2019 18

Complex Analysis I 2 THE ELEMENTARY PROPERTIES OF COMPLEX NUMBERS

Im
𝑖 𝑧1

𝑧2 𝑧0
0 1 Re

𝑧3

Figure 2.8: The four fourth roots of unity.

2.4 About history

We mention only some names from the history of complex numbers and complex analysis. As a
further reference, we refer to MacTutor, a web page by the University of St Andrews:
http://www-history.mcs.st-and.ac.uk/
• Girolamo Gardano (1501-1576) Mentioned the square roots of negative numbers in his
book ’Ars Magna’, 1545, but it seems that he thought they were useless.
• Raphael Bombelli (1526-1572) Introduced the algebraic operations with complex numbers
and used them to solve the 3rd degree equations, ’L’Algebra’, 1572.
• Gottfried Wilhelm Leibnitz (1646-1716) Seemed to wonder if complex numbers are useful.
• Leonhard Euler (1707-1783) Started to use complex numbers actively. For example, Euler
studied analytic functions of one complex variable.
• Caspar Wessel (1745-1818) A Norwegian-Danish engineer. The geometrical interpretation
of complex numbers as vectors in the plane appeared in his writing from 1799.
• Jean Robert Argand (1768-1822) A French accountant, bookkeeper and an amateur mathe-
matician. The geometric presentation of complex numbers as vectors in the plane is called
Argand’s diagram, although his publication is from 1806.
• Carl Friedrich Gauss (1777-1855) Started to study complex numbers systematically and to
use their geometric properties.
• William Rowan Hamilton (1805-1865) Introduced the concept of (ℝ2 , +, ⋅).
The complex function theory was developed in about 60 years by the following mathematicians.
The properties of complex functions based on
• differentiation were developed by Bernhard Riemann (1826-1866),
• integration were developed by Augustin-Louis Cauchy (1789-1859),
• the power series were developed by Karl Theodor Wilhelm Weierstrass (1815-1897).
To summarise, complex numbers were discovered in the 1500’s. However, the complex analysis
was developed in about 60 years during the 1800’s.

RHS, version May 11, 2019 19

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

3 Topological notions of sets in the complex plane

We give some simple topological properties which are necessary in our study of functions. The
norm
|⋅| ∶ ℂ × ℂ → [0, ∞),
defines a metric (or a distance function)
( )
𝑑 ∶ ℂ × ℂ → [0, ∞), 𝑑 𝑧1 , 𝑧2 = ||𝑧1 − 𝑧2 || .

Namely, for all 𝑧1 , 𝑧2 , 𝑧3 ∈ ℂ the following statements are valid:

1. ||𝑧1 − 𝑧2 || ≤ ||𝑧1 − 𝑧3 || + ||𝑧3 − 𝑧2 || .
2. ||𝑧1 − 𝑧2 || = ||𝑧2 − 𝑧1 || .
3. ||𝑧1 − 𝑧2 || = 0, if and only if 𝑧1 = 𝑧2 .
The non-negative real number ||𝑧1 − 𝑧2 || is the distance between 𝑧1 , 𝑧2 ∈ ℂ.
( )
The space (ℂ,|⋅|) is a normed space and (ℂ, 𝑑), where 𝑑 𝑧1 , 𝑧2 = ||𝑧1 − 𝑧2 || is defined by the
norm|⋅|, is a complete metric space.

3.1 Sets in the complex plane

Let 𝑎 ∈ ℂ and 𝑟 > 0. The open disc, 𝔻(𝑎, 𝑟), of radius 𝑟 that is centered at 𝑎 is defined by

𝔻(𝑎, 𝑟) = {𝑧 ∈ ℂ ∶ |𝑧 − 𝑎| < 𝑟}.

The closed disc, 𝔻(𝑎, 𝑟), of radius 𝑟 that is centered at 𝑎 is defined by

𝔻(𝑎, 𝑟) = {𝑧 ∈ ℂ ∶ |𝑧 − 𝑎| ≤ 𝑟}.

The boundary of both 𝔻(𝑎, 𝑟) and 𝔻(𝑎, 𝑟) is the circle

𝐶𝑟 (𝑎) = {𝑧 ∈ ℂ ∶ |𝑧 − 𝑎| = 𝑟}.
( )
The unit disc is the disc 𝔻 0, 1 .

𝑟 𝑟
𝑎 𝑎

(a) The open disc 𝔻(𝑎, 𝑟). (b) The closed disc 𝔻(𝑎, 𝑟).

Figure 3.1: Open and closed disc.

Definition 3.1.1. Let 𝑆 be( a set)in ℂ. A point 𝑧0 is an interior point of 𝑆 if there exists a real
number 𝑟 > 0, such that 𝔻 𝑧0 , 𝑟 ⊂ 𝑆. The interior of 𝑆, denoted by int 𝑆, consists of all of its
interior points. That is,

int 𝑆 = {𝑧 ∈ ℂ ∶ 𝑧 is an interior point of 𝑆}.

Definition 3.1.2. A set 𝑆 is open, if every point contained in 𝑆 is an interior point of 𝑆. A set 𝑆
is closed, if its complement, ℂ ⧵ 𝑆, is open.

RHS, version May 11, 2019 20

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

𝑟
𝑧0

Figure
( 3.2:
) Let 𝑆 be a set in ℂ. A point 𝑧0 is an interior point of 𝑆 if there exists 𝑟 > 0 such that
𝔻 𝑧0 , 𝑟 ⊂ 𝑆.

Example 3.1.3 (Examples of open sets).

( )
1. 𝔻 𝑧0 , 𝑟 .
( )
2. The punctured open disc, 𝔻 𝑧0 , 𝑟 ⧵ {𝑧0 }.
3. The punctured complex plane, ℂ ⧵ {0}.
4. The upper half-plane, ℍ+ = {𝑧 ∈ ℂ ∶ Im(𝑧) > 0}.

Example 3.1.4 (Examples of closed sets).

( )
1. 𝔻 𝑧0 , 𝑟 .
2. The circle, 𝐶𝑟 (𝑎).

Definitions 3.1.5.
( )
• A point 𝑧0 ∈ ℂ is a boundary point of 𝑆 if every disc 𝔻 𝑧0 , 𝑟 contains points belonging
to both 𝑆 and its complement, ℂ ⧵ 𝑆.
• The boundary of 𝑆 is defined by

𝜕𝑆 = {𝑧 ∈ ℂ ∶ 𝑧 is a boundary point of 𝑆}.

( )
• A point 𝑧0 ∈ ℂ is an accumulation point of 𝑆, if each punctured disc 𝔻 𝑧0 , 𝑟 ⧵ {𝑧0 }
contains at least one point of 𝑆.
• A point which is not an accumulation point is called an isolated point of 𝑆.
• The closure of 𝑆, denoted by 𝑆, is the union of 𝑆 and its accumulation points.
( )
• The open set 𝑈 ⊂ ℂ is a neighbourhood of 𝑧0 , if 𝑧0 ∈ 𝑈 . For example, 𝔻 𝑧0 , 𝑟 is a
neighbourhood of 𝑧0 .

Remark. Open sets are useful for differentiation!

Definition 3.1.6. A set 𝑆 is called bounded if there exist a real constant 𝑀 > 0, such that|𝑧| < 𝑀
whenever 𝑧 ∈ 𝑆. In other words, the set 𝑆 is contained in some large disc. If 𝑆 is bounded, we
define its diameter by
diam(𝑆) = sup||𝑧1 − 𝑧2 || , 𝑧1 , 𝑧2 ∈ 𝑆.
Conversely, a set is called unbounded, if it’s not bounded.

A set 𝑆 is compact, if 𝑆 is both closed and bounded.

RHS, version May 11, 2019 21

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

( )
𝔻 0, 𝑀
Figure 3.3: A compact set 𝑆 in ℂ.

3.1.7 Connectedness
Recall that a continuous map,
𝛾 ∶ [𝑎, 𝑏] → 𝑆, 𝑎 < 𝑏,
is a path in 𝑆. We write

𝛾(𝑡) = (𝛾1 (𝑡), 𝛾2 (𝑡))

= 𝛾1 (𝑡) + 𝑖𝛾2 (𝑡)
( ) ( )
= Re 𝛾(𝑡) + 𝑖 Im 𝛾(𝑡) , 𝑡 ∈ [𝑎, 𝑏].

The path 𝛾 joins the points 𝛾(𝑎) and 𝛾(𝑏).

Definition. Let 𝑧1 , 𝑧2 be given. The image of the path 𝛾(𝑡) = 𝑧1 + 𝑡(𝑧2 − 𝑧1 ), where 𝑡 ∈ [0, 1],
is a straight line segment:
𝛾([0, 1]) = [𝛾(0), 𝛾(1)] = [𝑧1 , 𝑧2 ].

Definition. A finite number of line segments joined end-to-end forms a polygonal line or a
polygonal route. So if we have points 𝑧0 , … , 𝑧𝑛 ∈ ℂ, we call the union

[𝑧0 , 𝑧1 ] ∪ [𝑧1 , 𝑧2 ] ∪ ⋯ ∪ [𝑧𝑛−1 , 𝑧𝑛 ]

a polygonal route from 𝑧0 to 𝑧𝑛 .

RHS, version May 11, 2019 22

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

𝑎 𝑏

(a) A continuous mapping 𝛾 ∶ [𝑎, 𝑏] → 𝑆, where 𝑎 < 𝑏, is a path in 𝑆.

𝑧2 𝑧1
𝑧0
𝑧4

𝑧2

𝑧3
𝑧1
(b) A line segment in ℂ. (c) A polygonal route in ℂ.

Figure 3.4: A continuous mapping, a line segment and a polygonal route.

Definition. A non-empty subset 𝑆 of ℂ is polygonally connected if, given two points 𝑧1 , 𝑧2 ∈ 𝑆,

there is a polygonal route from 𝑧1 to 𝑧2 that lies entirely in 𝑆.

Definition. An open set 𝑆 is connected if each pair of points 𝑧1 , 𝑧2 ∈ 𝑆 can be joined by a

polygonal line or route that lies entirely in 𝑆.

𝑆 𝑧2

𝑧1

Figure 3.5: A polygonally connected open set 𝑆.

Remark. Connected sets are good for integration!

Definition (Domains). Open sets which are connected are called domains.

RHS, version May 11, 2019 23

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

3.2 Limits and continuity

3.2.1 Sequences
A complex sequence (𝑧𝑛 ) is an assignment of a complex number 𝑧𝑛 to each number 𝑛 ∈ ℕ. We
write (𝑧𝑛 )∞
𝑛=1
or just (𝑧𝑛 ).

Definitions.
• A sequence (𝑧𝑛 ) is bounded if there exists a constant 𝑀 ∈ ℝ, such that ||𝑧𝑛 || ≤ 𝑀 for all 𝑛.
• The sequence (𝑧𝑛 ) converges to a limit 𝑎 ∈ ℂ, written as 𝑧𝑛 → 𝑎 or lim𝑛→∞ 𝑧𝑛 = 𝑎, if,
given 𝜖 > 0, there exists a number 𝑁𝜀 ∈ ℕ, such that ||𝑧𝑛 − 𝑎|| < 𝜀 whenever 𝑛 ≥ 𝑁𝜀 .
• The sequence (𝜔𝑘 ) is a subsequence of the sequence (𝑧𝑛 ), if there exist natural numbers
𝑛1 < 𝑛2 < … , such that
𝜔𝑘 = 𝑧𝑛𝑘 , for 𝑘 = 1, 2, … .

Example. Let 𝑎 ∈ ℂ be given and|𝑎| ≠ 1. Let (𝑧𝑛 ) be a sequence where 𝑧𝑛 = 𝑎𝑛 .

Now, if|𝑎| < 1, then ||𝑧𝑛 || = |𝑎|𝑛 , which converges to 0.
If|𝑎| > 1, then|𝑎|𝑛 → ∞ and hence the sequence (𝑧𝑛 ) has no limit in ℂ.

3.2.2 Limits of functions

Let 𝑓 ∶ 𝑆 → ℂ be a function on a set 𝑆 ⊂ ℂ. Let 𝑎 be an accumulation point of 𝑆. Then

lim 𝑓 (𝑧) = 𝜔
𝑧→𝑎

or
lim
𝑧→𝑎
𝑓 (𝑧) = 𝜔,
𝑧∈𝑆

that is, 𝑓 (𝑧) → 𝜔 as 𝑧 → 𝑎, if, given 𝜀 > 0, there exists 𝛿 = 𝛿𝑎,𝜀 > 0 such that
|𝑓 (𝑧) − 𝜔| < 𝜀,
| |
whenever 𝑧 ∈ 𝑆 and 0 < |𝑧 − 𝑎| < 𝛿.

Remark. The limit, if it exists, is determined by the behaviour of 𝑓 (𝑧) as 𝑧 approaches 𝑎. The
value of 𝑓 (𝑎) is irrelevant and it may not even be defined if 𝑎 ∉ 𝑆.

Im(𝑧)
Example. Let 𝑓 (𝑧) = , 𝑧 ≠ 0.
Re(𝑧)
Now we have

𝑓 (𝑧) = 0, when 𝑧 ∈ ℝ ⧵ {0}

𝑓 (𝑧) = 1, when 𝑧 is on the line 𝑦 = 𝑥.

Hence, lim 𝑓 (𝑧) fails to exist.

𝑧→0

3.2.3 Continuity
Let 𝑓 ∶ 𝑆 → ℂ be a function. Then 𝑓 is continuous at 𝑎 ∈ 𝑆 if, given 𝜀 > 0, there exists
𝛿 = 𝛿𝑎,𝜀 > 0, such that
|𝑓 (𝑧) − 𝑓 (𝑎)| < 𝜀
| |
whenever 𝑎 ∈ 𝑆 and |𝑧 − 𝑎| < 𝛿. The function 𝑓 is continuous on 𝑆 if it is continuous at each
𝑎 ∈ 𝑆.

RHS, version May 11, 2019 24

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

𝛿
𝑓 (𝑧) 𝑎
𝜀
𝑓 (𝑎)
𝑧

Figure 3.6: A function 𝑓 ∶ 𝑆 → ℂ is continuous at 𝑎 ∈ 𝑆 if, given 𝜀 > 0, there exists 𝛿 > 0,
such that ||𝑓 (𝑧) − 𝑓 (𝑎)|| < 𝜀 whenever 𝑎 ∈ 𝑆 and|𝑧 − 𝑎| < 𝛿.

Remarks.
• Sums and products of continuous functions are also continuous.
• Since the notions of convergence for complex numbers and points in ℝ2 are the same, the
function 𝑓 of a complex variable 𝑧 = 𝑥 + 𝑖𝑦 is continuous if and only if it is also continuous
when viewed as a function of two real variables, 𝑥 and 𝑦.
• If 𝑓 is continuous, then the real-valued function

|𝑓 | ∶ ℂ → ℝ, |𝑓 | (𝑧) = ||𝑓 (𝑧)||

is continuous.

The following lemma provides a link between convergence in ℂ and ℝ.

Lemma 3.2.4. Let (𝑧𝑛 ) be a complex sequence. Then (𝑧𝑛 ) converges, if and only if the real-valued
sequences (Re(𝑧𝑛 )) and (Im(𝑧𝑛 )) both converge.
If 𝑧𝑛 → 𝑎, then ||𝑧𝑛 || → |𝑎| and 𝑧𝑛 → 𝑎.
( ) ( )
Example. Let 𝑧𝑛 = 1 + 𝑚1 + 𝑖 1 − 𝑚1 , 𝑚 ∈ ℕ.

Then 𝑧𝑛 → 1 + 𝑖.

Definition 3.2.5. Let 𝑓 ∶ 𝑆 → ℂ, 𝑓 = 𝑢 + 𝑖𝑣, where 𝑢, 𝑣 ∶ ℂ → ℝ. So 𝑢 = Re 𝑓 and 𝑣 = Im 𝑓 .

Now we define the functions 𝑢 = Re 𝑓 ∶ ℂ → ℝ and 𝑣 = Im 𝑓 ∶ ℂ → ℝ by setting

𝑢(𝑧) = (Re 𝑓 )(𝑧) = Re(𝑓 (𝑧)),

𝑣(𝑧) = (Im 𝑓 )(𝑧) = Im(𝑓 (𝑧)).

Then, 𝑓 (𝑧) → 𝜔 implies

Re 𝑓 (𝑧) → Re 𝜔
Im 𝑓 (𝑧) → Im 𝜔.

We also have ||𝑓 (𝑧)|| → |𝜔| and 𝑓 (𝑧) → 𝜔.

Proposition 3.2.6. Let 𝑓 ∶ 𝑆 → ℂ. Then 𝑓 is continuous at 𝑎 ∈ 𝑆 (or on 𝑆), if and only if both
Re 𝑓 and Im 𝑓 are continuous at 𝑎 ∈ 𝑆 (or on 𝑆).

RHS, version May 11, 2019 25

Complex Analysis I 3 TOPOLOGICAL NOTIONS OF SETS IN THE COMPLEX PLANE

3.3 The Cauchy convergence principle for ℂ

Definition. A sequence (𝑧𝑛 ) is called a Cauchy sequence, or simply Cauchy, if
|𝑧𝑛 − 𝑧𝑚 | → 0, as 𝑛, 𝑚 → ∞.
| |

Recall that ℝ is complete: every Cauchy sequence of real numbers converges to a real number.
Since the complex sequence (𝑧𝑛 ) is Cauchy if and only if (Re(𝑧𝑛 )) and (Im(𝑧𝑛 )) are, then we can
conclude that every Cauchy sequence in ℂ has a limit in ℂ.

Theorem. ℂ is complete.

RHS, version May 11, 2019 26

Complex Analysis I 4 ANALYTIC FUNCTIONS

4 Analytic functions
4.1 Definitions
Let Ω be an open set in ℂ and 𝑓 a complex-valued function on Ω (we assume that 𝑓 is a complex-
valued function of one complex variable and Ω ≠ ∅).

Definition. The function 𝑓 is complex differentiable at the point 𝑧0 ∈ Ω if the quotient

𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 )
(4.1.1)
ℎ
converges to a limit when ℎ → 0, where ℎ ∈ ℂ ⧵ {0} and 𝑧0 + ℎ ∈ Ω. Recall ℂ = (ℝ2 , +, ⋅) is a
field, and hence the quotient is well defined.

The limit of 4.1.1, when it exists, is denoted by 𝑓 ′ (𝑧0 ) and is called the complex derivative of 𝑓
at 𝑧0 :
𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 )
𝑓 ′ (𝑧0 ) = lim . (4.1.2)
ℎ→0 ℎ
Note that ℎ ∈ ℂ ⧵ {0} must approach 0 from any direction.

Remark. Complex differentiability which is defined only at one point or two points is not enough
to build an interesting,
( meaningful
) theory, so we need complex differentiability to be defined also
on a small disc 𝔻 𝑧0 , 𝜀 , 𝜀 > 0.
( )
Definition 4.1.3. The function 𝑓 is said to be analytic at 𝑧0 if there exists a disc 𝔻 𝑧0 , 𝑟 such
that 𝑓 is complex differentiable at every point in the disc.

𝑟
𝑧0

( )
Figure 4.1: Open disc 𝔻 𝑧0 , 𝑟 .

Remark. In order for 𝑓 to be analytic at 𝑧0 , 𝑓 should have a complex derivative at every point in
a small neighbourhood of 𝑧0 .

Definition 4.1.4. Let Ω be an open set in ℂ and 𝑓 a complex-valued function on Ω. The function
𝑓 is analytic on Ω if it is analytic at every point of Ω.

Remark. Thus, analyticity is a stronger condition than differentiability. Now, we give some ex-
amples.

Examples 4.1.5.
1. Any constant function is analytic on the whole complex plane. Let 𝑓 (𝑧) = 𝑐 ∈ ℂ for all
𝑧 ∈ ℂ. Then 𝑓 ′ (𝑧) = 0:
𝑓 (𝑧 + ℎ) − 𝑓 (𝑧) 𝑐 − 𝑐
= = 0.
ℎ ℎ

2. The function 𝑓 (𝑧) = 𝑧 is analytic on any open set Ω in ℂ and 𝑓 ′ (𝑧) = 1:

𝑓 (𝑧 + ℎ) − 𝑓 (𝑧) 𝑧 + ℎ − 𝑧
= = 1.
ℎ ℎ

RHS, version May 11, 2019 27

Complex Analysis I 4 ANALYTIC FUNCTIONS

3. The function 𝑓 (𝑧) = 𝑧2 is analytic on any open set Ω in ℂ and 𝑓 ′ (𝑧) = 2𝑧:

𝑓 (𝑧 + ℎ) − 𝑓 (𝑧) (𝑧 + ℎ)2 − 𝑧2 𝑧2 + 2𝑧ℎ + ℎ2 − 𝑧2 ℎ→0

= = = 2𝑧 + ℎ → 2𝑧.
ℎ ℎ ℎ

Example 4.1.6. The function 𝑓 (𝑧) = 𝑧 is not analytic. We have:

{
𝑓 (𝑧 + ℎ) − 𝑓 (𝑧) 𝑧 + ℎ − 𝑧 ℎ 1 if ℎ ∈ ℝ ⧵ {0}
= = =
ℎ ℎ ℎ −1 if Re(ℎ) = 0, ℎ ≠ 0.

Hence there is no limit as ℎ → 0.

Note that, in terms of real variables, the function 𝑓 (𝑧) = 𝑧 corresponds to the map 𝑔 ∶ (𝑥, 𝑦) ↦
(𝑥, −𝑦), which is differentiable in the real sense. So, the existence of the real derivative need not
guarantee that 𝑓 is analytic.

Remark. The adjectives “regular” and “holomorphic” are sometimes used instead of “analytic”.

Definition 4.1.7. If a function is analytic on the entire complex plane, it is said to be entire.
Entire functions are the very best class of complex functions. (T. Tao)

Remark. Note that, if 𝐷 is a domain, then 𝑓 is analytic if and only if 𝑓 is complex differentiable
at every point in 𝐷.

4.2 Elementary properties

Lemma 4.2.1. A function 𝑓 is complex differentiable at 𝑧0 ∈ Ω if and only if there exists a
complex number 𝑎 such that

𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) − 𝑎ℎ = ℎ𝜑(ℎ), (4.2.2)

where 𝜑 is a function defined for all ℎ, when |ℎ| is very small and limℎ→0 𝜑(ℎ) = 0.
Here, 𝑎 = 𝑓 ′ (𝑧0 ).

Remark. The function 𝜑 depends on 𝑓 and 𝑧0 , that is, 𝜑 = 𝜑𝑓 ,𝑧0 .

Proof.
𝑓 (𝑧0 +ℎ)−𝑓 (𝑧0 )
“⇒”: Let us define 𝜑(0) = 0, and 𝜑(ℎ) = ℎ
− 𝑓 ′ (𝑧0 ), ℎ ≠ 0.
𝑓 (𝑧0 +ℎ)−𝑓 (𝑧0 )
“⇐”: From (4.2.2), ℎ
= 𝑎 + 𝜑(ℎ) → 𝑎 as ℎ → 0.

Corollary 4.2.3. Let Ω be an open set in ℂ. Let 𝑓 be a complex-valued function on Ω. If 𝑓 is

differentiable at 𝑧0 ∈ Ω, then 𝑓 is continuous at the point 𝑧0 ∈ Ω.

Proof. From (4.2.2), limℎ→0 𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) = 0, and this is enough since 𝑧0 ∈ Ω is an
accumulation point of the open set Ω.

Corollary 4.2.4. An analytic function is continuous.

Remark. There are continuous functions which are not analytic.

Example. 𝑓 (𝑧) = 𝑧 is continuous on the whole plane but it is not analytic.

4.2.5 (Differentiation formulas). Let Ω be an open set in ℂ. Let 𝑓 and 𝑔 be complex-valued

functions on Ω. If the complex derivatives of 𝑓 and 𝑔 exist at a point 𝑧0 ∈ Ω, then

RHS, version May 11, 2019 28

Complex Analysis I 4 ANALYTIC FUNCTIONS

1. the function 𝑓 + 𝑔 ∶ Ω → ℂ is complex differentiable at 𝑧0 and

(𝑓 + 𝑔)′ (𝑧0 ) = 𝑓 ′ (𝑧0 ) + 𝑔 ′ (𝑧0 ),

2. the function 𝑓 ⋅ 𝑔 ∶ Ω → ℂ is complex differentiable at 𝑧0 and

(𝑓 𝑔)′ (𝑧0 ) = 𝑓 ′ (𝑧0 )𝑔(𝑧0 ) + 𝑓 (𝑧0 )𝑔 ′ (𝑧0 ).

𝑓 ( )
3. when 𝑔(𝑧0 ) ≠ 0, then the function 𝑔
is well defined in some disc 𝔻 𝑧0 , 𝑟 , 𝑟 > 0 and
( )′
𝑓 𝑓 ′ (𝑧0 )𝑔(𝑧0 ) − 𝑓 (𝑧0 )𝑔 ′ (𝑧0 )
(𝑧0 ) = .
𝑔 𝑔(𝑧0 )2

Proof. The proof is left as homework to the reader. Use lemma 4.2.1 or the definition.

Corollary 4.2.6. Any polynomial

𝑝(𝑧) = 𝑎0 + 𝑎1 𝑧 + … + 𝑎𝑛 𝑧𝑛 ,

where 𝑛 ∈ ℕ = {1, 2, …}, 𝑎𝑘 ∈ ℂ, 𝑘 = 0, 1, … , 𝑛 and 𝑎𝑛 ≠ 0, is analytic on the entire plane, and

𝑝′ (𝑧) = 𝑎1 + … + 𝑛𝑎𝑛 𝑧𝑛−1 .

𝑝(𝑧)
Corollary 4.2.7. A rational function 𝑞(𝑧) , where 𝑝(𝑧) and 𝑞(𝑧) are polynomials, is analytic on
any open set in which 𝑞(𝑧) is never zero.

Examples 4.2.8.
1
1. The function 𝑓 (𝑧) = 𝑧
is analytic on any open set in ℂ that does not contain the origin and
𝑓 ′ (𝑧) = − 𝑧12 .
1
2. The rational function 𝑓 (𝑧) = 𝑧2 +1
is analytic on any open set in ℂ that does not contain ±𝑖.

Example 4.2.9. When is the function 𝑔(𝑧) = |𝑧|2 analytic? Since 𝑓 (𝑧) = 𝑧 is not analytic in any
2
open set and 𝑧 = |𝑧𝑧 | when 𝑧 ≠ 0, the function 𝑔 is not analytic anywhere.

Remark. The function 𝑔 is complex differentiable at the origin.

𝑔(0 + ℎ) − 𝑔(0) |ℎ|2 ℎ→0

= = ℎ → 0.
ℎ ℎ

Remark. Note that the corresponding map 𝑘 ∶ (𝑥, 𝑦) ↦ 𝑥2 + 𝑦2 is differentiable in the real sense
on the whole ℝ2 .

Theorem 4.2.10: Chain rule. Let Ω and 𝑈 be open sets in ℂ. Let 𝑓 be analytic on Ω and let 𝑔
be analytic on 𝑈 and 𝑓 (Ω) ⊆ 𝑈 . The composite function 𝑔◦𝑓 , given by (𝑔◦𝑓 )(𝑧) = 𝑔(𝑓 (𝑧)), is
analytic on Ω and for all 𝑧 ∈ Ω,

(𝑔◦𝑓 )′ (𝑧) = 𝑔 ′ (𝑓 (𝑧)) ⋅ 𝑓 ′ (𝑧).

RHS, version May 11, 2019 29

Complex Analysis I 4 ANALYTIC FUNCTIONS

( )
𝑓 Ω 𝑈

𝑓 𝑔

ℂ
Ω
𝑔◦𝑓

Figure 4.2: Chain rule (Theorem 4.2.10): the composite function (𝑔◦𝑓 )(𝑧) = 𝑔(𝑓 (𝑧)) is analytic
on Ω if functions 𝑔 and 𝑓 are analytic on their respective domains.

Theorem 4.2.11: Derivative of inverse function. Let Ω be an open set in ℂ and let 𝑓 ∶ Ω → ℂ
be differentiable at 𝑧 ∈ Ω, with 𝑓 ′ (𝑧) ≠ 0. If there is a neighbourhood 𝑈 of the point 𝑤 ∶= 𝑓 (𝑧)
such that 𝑓 has a continuous inverse function 𝑓 −1 on 𝑈 , then 𝑓 −1 is differentiable at 𝑤 and
( )′
1 1
𝑓 −1 (𝑤) = = ( ).
𝑓 ′ (𝑧) 𝑓 ′ 𝑓 −1 (𝑤)

Proof. Since 𝑓 has a derivative at 𝑧, we have

𝑓 (𝑧 + ℎ) − 𝑓 (𝑧) = 𝑓 ′ (𝑧)ℎ + ℎ𝜀(ℎ),

where 𝜀(ℎ) → 0 when ℎ → 0.

Let |𝑘| be so small that 𝑤 + 𝑘 ∈ 𝑈 . Then for every such 𝑘 ∈ ℂ there exists ℎ ∈ ℂ with
𝑓 −1 (𝑤 + 𝑘) = 𝑧 + ℎ. Since 𝑓 −1 is continuous in 𝑈 ,

lim (𝑧 + ℎ) = lim 𝑓 −1 (𝑤 + 𝑘) = 𝑓 −1 (𝑤) = 𝑧,

𝑘→0 𝑘→0

so ℎ → 0 when 𝑘 → 0. Hence
𝑓 −1 (𝑤 + 𝑘) − 𝑓 −1 (𝑤) 𝑧+ℎ−𝑧 ℎ
= =
𝑘 𝑤 + 𝑘 − 𝑤 𝑓 (𝑧 + ℎ) − 𝑓 (𝑧)
ℎ 1
= ′ = ′
𝑓 (𝑧)ℎ + ℎ𝜀(ℎ) 𝑓 (𝑧) + 𝜀(ℎ)
1
→ ′ ,
𝑓 (𝑧)
when 𝑘 → 0.

4.3 The relationship between the complex and real derivatives

The notion of complex differentiability differs from the notion of real differentiability of a function
of two real variables. Let us associate to a complex valued function 𝑓 = 𝑢 + 𝑖𝑣, of one complex
variable, the mapping
( )
𝑔 ∶ ℝ2 → ℝ2 , 𝑔(𝑥, 𝑦) = 𝑢(𝑥, 𝑦), 𝑣(𝑥, 𝑦) .

Recall from the vector calculus course that the function 𝑔 is differentiable at a point (𝑥0 , 𝑦0 ), if
there exists an ℝ-linear transformation 𝐿 ∶ ℝ2 → ℝ2 such that
( )
𝑔 (𝑥0 , 𝑦0 ) + ℎ − 𝑔(𝑥0 , 𝑦0 ) = 𝐿ℎ +‖ℎ‖ 𝜀(ℎ),

with ‖ ‖
‖𝜀(ℎ)‖ → 0 as‖ℎ‖ → 0.

RHS, version May 11, 2019 30

Complex Analysis I 4 ANALYTIC FUNCTIONS

Equivalently, we can write

‖ ( ) ‖
‖𝑔 (𝑥0 , 𝑦0 ) + ℎ − 𝑔(𝑥0 , 𝑦0 ) − 𝐿ℎ‖
‖ ‖ → 0,
‖ℎ‖
as‖ℎ‖ → 0, where ℎ = (ℎ1 , ℎ2 ) ∈ ℝ2 ⧵ {0}.

Remark. For a complex number 𝑧 = 𝑥+𝑖𝑦, we write 𝑓 (𝑧) = 𝑢(𝑥, 𝑦)+𝑖𝑣(𝑥, 𝑦), where 𝑢 ∶ ℝ2 → ℝ
and 𝑣 ∶ ℝ2 → ℝ.

The ℝ-linear transformation 𝐿 is unique and is called the derivative of 𝑔 at (𝑥0 , 𝑦0 ). If 𝑔 is differ-
entiable in the real sense, the partial derivatives of 𝑢 and 𝑣 exist, and the ℝ-linear transformation
is described in the standard basis {𝑒1 , 𝑒2 } = {(1, 0), (0, 1)} of ℝ2 by the Jacobian matrix of 𝑔:
⎡ 𝜕𝑢 𝜕𝑢 ⎤
⎢ 𝜕𝑥 𝜕𝑦 ⎥
𝑔 (𝑥, 𝑦) = ⎢ ⎥ ∽ 𝐿.
⎢ 𝜕𝑣 𝜕𝑣 ⎥
⎣ 𝜕𝑥 𝜕𝑦 ⎦

The derivative in the real case is a matrix, while the complex derivative is a complex number
𝑓 ′ (𝑧0 ).
Our goal is to find the relationship between these two notions.

4.3.1. Let us assume that 𝑓 has a complex derivative at the point 𝑧0 . We consider the limit in
(4.1.2), where ℎ ∈ ℝ, i.e. ℎ = ℎ1 + 𝑖ℎ2 with ℎ2 = 0. Let us write 𝑧 = 𝑥 + 𝑖𝑦, 𝑧0 = 𝑥0 + 𝑖𝑦0 , and
𝑓 (𝑧) = 𝑓 (𝑥, 𝑦). Then,
(( ) ) ( )
( )4.1.2
𝑓 𝑥 0 , 𝑦0 + ℎ − 𝑓 𝑥0 , 𝑦0
𝑓 ′ 𝑧0 = lim
ℎ→0 ℎ
( ) ( )
𝑓 𝑥0 + ℎ1 , 𝑦0 − 𝑓 𝑥0 , 𝑦0
= lim
ℎ1 →0 ℎ1
𝜕𝑓 ( )
= 𝜕𝑥 𝑥0 , 𝑦0 ,
𝜕
where 𝜕𝑥 denotes the usual partial derivative in the first variable by the definition of the partial
derivatives.
Next we consider the limit (4.1.2) when ℎ = 𝑖ℎ2 , so ℎ1 = 0. We obtain
(( ) ) ( )
( ) 𝑓 𝑥0 , 𝑦0 + ℎ − 𝑓 𝑥0 , 𝑦0
𝑓 ′ 𝑧0 = lim
ℎ→0 ℎ
( ) ( )
𝑓 𝑥0 , 𝑦0 + ℎ2 − 𝑓 𝑥0 , 𝑦0
= lim
ℎ2 →0 𝑖ℎ2
1 ( )
= 𝜕𝜕𝑦𝑓 𝑥0 , 𝑦0 ,
𝑖
where 𝜕
𝜕𝑦
is the partial derivative in the 2nd variable. Recall that 𝑖 = (0, 1) and so 𝑖ℎ2 = (0, ℎ2 ).

4.3.2. Hence, if 𝑓 is analytic, we have shown that

𝜕𝑓 1 𝜕𝑓
= .
𝜕𝑥 𝑖 𝜕𝑦

1
Writing 𝑓 = 𝑢 + 𝑖𝑣, we find, after separating the real and imaginary parts and writing 𝑖
= −𝑖,
that the partial derivatives of 𝑢 and 𝑣 exist. Moreover, they satisfy the equations
⎧ 𝜕 𝑓 = 𝜕 (𝑢 + 𝑖𝑣) = 𝜕 𝑢 + 𝑖 𝜕 𝑣
⎪ 𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑥
⎨
⎪ 1 𝜕 𝑓 = −𝑖 𝜕 (𝑢 + 𝑖𝑣) = −𝑖 𝜕 𝑢 + 𝜕𝑣
= 𝜕𝑣 𝜕𝑢
− 𝑖 𝜕𝑦 ,
⎩ 𝑖 𝜕𝑦 𝜕𝑦 𝜕𝑦 𝜕𝑦 𝜕𝑦

RHS, version May 11, 2019 31

Complex Analysis I 4 ANALYTIC FUNCTIONS

which gives us the following identities:

𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
= and =− . (4.3.3)
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦

These are called the Cauchy–Riemann (C–R) equations. The Cauchy–Riemann equations link
together real and complex analysis.

4.3.4. Let us define two differential operators:

( )
𝜕 1 𝜕 𝜕
= 𝜕𝑥
− 𝑖 𝜕𝑦
𝜕𝑧 2
and ( ) ( )
𝜕 1 𝜕 1 𝜕 1 𝜕 𝜕
= 𝜕𝑥
− 𝜕𝑦 = 𝜕𝑥
+ 𝑖 𝜕𝑦 .
𝜕𝑧 2 𝑖 2

Theorem 4.3.5. If 𝑓 is analytic at 𝑧0 , then

𝜕𝑓
(𝑧0 ) = 0, (4.3.6)
𝜕𝑧
and

𝜕𝑓 𝜕𝑢
𝑓 ′ (𝑧0 ) = (𝑧 ) = 2 (𝑧0 ). (4.3.7)
𝜕𝑧 0 𝜕𝑧

If we write 𝑔(𝑥, 𝑦) = 𝑓 (𝑧), then 𝑔 is differentiable in the sense of real variables and

| |2
det 𝑔 (𝑥0 , 𝑦0 ) = |𝑓 ′ (𝑧0 )| . (4.3.8)
| |
Proof. For (4.3.6): Taking the real and imaginary parts, we see that the Cauchy–Riemann equa-
tions are equivalent to
𝜕𝑓
(𝑧0 ) = 0.
𝜕𝑧
The C–R equations imply:

( ) ( )
𝜕𝑓 1 𝜕𝑓 𝜕𝑓 1 𝜕𝑢 𝜕𝑣 𝜕𝑢 2 𝜕𝑣
𝜕𝑧
= + 𝑖 = + 𝑖 + 𝑖 + 𝑖
2 (𝜕𝑥 𝜕𝑦 2 𝜕𝑥 )𝜕𝑥 𝜕𝑦 𝜕𝑦
( ) (C–R)
1 𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
= − + 𝑖 𝜕𝑥 + 𝜕𝑦 = 0.
2 𝜕𝑥 𝜕𝑦

Conversely, if (
𝜕𝑓 ( ))
1 𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
0= = 𝜕𝑥
− 𝜕𝑦
+𝑖 𝜕𝑥
+ 𝜕𝑦
,
𝜕𝑧 2
then
𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
= and =− .
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦

For (4.3.7):
1 ′ 1
𝑓 ′ (𝑧0 ) = 𝑓 (𝑧0 ) + 𝑓 ′ (𝑧0 )
2 2 ( )
(4.3.1) 1 𝜕 𝑓 1 1 𝜕𝑓
= (𝑧 ) + (𝑧 )
2 𝜕𝑥 0 2 𝑖 𝜕𝑦 0
(4.3.4) 𝜕 𝑓
= 𝜕𝑧 (𝑧0 ).

RHS, version May 11, 2019 32

Complex Analysis I 4 ANALYTIC FUNCTIONS

Furthermore, we have
( )
𝜕 𝑓 (4.3.4) 1 𝜕𝑓
𝜕𝑧
= 𝜕𝑥
− 𝑖 𝜕𝜕𝑦𝑓
2
( ( ))
1 𝜕𝑢 𝜕𝑣 𝜕𝑢
= 𝜕𝑥
+ 𝑖 𝜕𝑥 −𝑖 𝜕𝑦
+ 𝑖 𝜕𝜕𝑦𝑣
2
(C-R)
( )
1 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢
= − 𝑖 − 𝑖 +
2 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥
( )
1 𝜕𝑢 𝜕𝑢 𝜕𝑢
=2 − 𝑖 𝜕𝑦 = 2 𝜕𝑧 .
2 𝜕𝑥

We defined two differential operators

( ) ( )
𝜕 1 𝜕 𝜕 𝜕 1 𝜕 𝜕
= − 𝑖 and = + 𝑖 𝜕𝑦 .
𝜕𝑧 2 𝜕𝑥 𝜕𝑦 𝜕𝑧 2 𝜕𝑥

Let’s recall theorem 4.3.5:

If 𝑓 is analytic at 𝑧0 , then
𝜕𝑓 𝜕𝑓 𝜕𝑢
(𝑧0 ) = 0 and 𝑓 ′ (𝑧0 ) = (𝑧 ) = 2 (𝑧0 ).
𝜕𝑧 𝜕𝑧 0 𝜕𝑧
Also, if we write
𝑔(𝑥, 𝑦) = 𝑓 (𝑧),
then 𝑔 is differentiable in the sense of two real variables, and
| |2
det 𝑔 ( 𝑥0 , 𝑦0 ) = |𝑓 ′ (𝑧0 )| .
⏟⏟⏟ | |
=𝑧0

Let us write 𝑧0 = (𝑥0 , 𝑦0 ). Now, our goal is to show that 𝑔 is differentiable in the real sense. We
have to show that there exists an ℝ-linear transformation

𝐿𝑔 ∶ ℝ → ℝ

such that
𝑔((𝑥0 , 𝑦0 ) + 𝐻) − 𝑔(𝑥0 , 𝑦0 ) = 𝐿𝑔 𝐻 +‖𝐻‖ 𝜀(𝐻), (4.3.9)
where ‖ ‖
‖𝜀(𝐻)‖ → 0 as‖𝐻‖ → 0, 𝐻 = (ℎ1 , ℎ2 ) ∈ ℝ and ℎ1 + 𝑖ℎ2 ∈ ℂ.
2

Recall from the vector calculus course that the ℝ-linear transformation 𝐿𝑔 is unique and is called
the derivative of 𝑔 at (𝑥0 , 𝑦0 ).
Since 𝑓 is complex differentiable at 𝑧0 and the C–R equations are satisfied, we know that
𝜕𝑓 𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
𝑓 ′ (𝑧0 ) = (𝑧 ) = +𝑖 = −𝑖 .
𝜕𝑥 0 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦
By the definition of complex differentiability,

𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) = 𝑓 ′ (𝑧0 ) ⋅ ℎ + ℎ𝜀(ℎ)

( )
= 𝜕𝜕𝑦𝑣 − 𝑖 𝜕𝑦 𝜕𝑢
ℎ + ℎ𝜀(ℎ)
( )
= 𝜕𝜕𝑦𝑣 − 𝑖 𝜕𝑦 𝜕𝑢
(ℎ1 + 𝑖ℎ2 ) + ℎ𝜀(ℎ),

where limℎ→0 𝜀(ℎ) = 0.

Now we identify a complex number with the pair of its real and imaginary parts. Hence, by
equation 4.3.9 and calculations from the start of this section,

𝑔((𝑥0 , 𝑦0 ) + 𝐻) − 𝑔(𝑥0 , 𝑦0 ) = 𝑔((𝑥0 , 𝑦0 ) + (ℎ1 , ℎ2 )) − 𝑔(𝑥0 , 𝑦0 )

RHS, version May 11, 2019 33

Complex Analysis I 4 ANALYTIC FUNCTIONS

= 𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 )
( )
𝜕𝑢 𝜕𝑢
= 𝜕𝑥 − 𝑖 𝜕𝑦 (ℎ1 + 𝑖ℎ2 ) + ℎ𝜀(ℎ)
⎡ 𝜕𝑢 𝜕𝑢 ⎤
⎢ 𝜕𝑥 𝜕𝑦 ⎥ ⎡ℎ1 ⎤
=⎢ ⎥⎢ ⎥ + ℎ𝜀(ℎ)
⎢ 𝜕𝑣 𝜕 𝑣 ⎥ ⎢ℎ ⎥
⎣ 𝜕𝑥 𝜕𝑦 ⎦
⎣ 2⎦
= 𝐿𝑔 ℎ + ℎ𝜀(ℎ) = 𝐿𝑔 𝐻 +‖𝐻‖ 𝜀(𝐻),
where 𝜀(ℎ) = (𝜀1 (ℎ), 𝜀(ℎ2 )) and 𝐻 ≠ (0, 0).
Note that
(ℎ1 𝜀1 (ℎ) − ℎ2 𝜀2 (ℎ), ℎ2 𝜀2 (ℎ) + ℎ2 𝜀1 (ℎ))
ℎ ⋅ 𝜀(ℎ) = ‖𝐻‖ ,
‖𝐻‖
⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟
=∶𝜀(𝐻)

where ‖ ‖
‖𝜀(𝐻)‖ → 0 as‖𝐻‖ → 0.
So 𝑔 is differentiable in the sense of two real variables.
We obtain
𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
det 𝑔 (𝑥0 , 𝑦0 ) = 𝜕𝑥 𝜕𝑦
− 𝜕𝑥 𝜕𝑦
( )2 ( )2
𝜕𝑢 𝜕𝑢
= 𝜕𝑥
+ 𝜕𝑦
| 𝜕𝑢 |2
| 𝜕 𝑢 |2 | ′ |2
= || 𝜕𝑥 𝜕𝑢 |
− 𝑖 𝜕𝑦 | = ||2 𝜕𝑧 || = ||𝑓 (𝑧0 )|| .
| |

The next theorem gives an important converse:

Theorem 4.3.10. Suppose that 𝑓 = 𝑢 + 𝑖𝑣 is a complex-valued function defined on an open set

Ω ⊂ ℂ. If 𝑢 and 𝑣 are differentiable in the real sense at the point 𝑧0 ∈ Ω, and their partial
derivatives satisfy the C–R equations at point 𝑧0 , then 𝑓 has a complex derivative at 𝑧0 and
𝜕𝑢 𝜕𝑣
𝑓 ′ (𝑧0 ) = (𝑧 ) + 𝑖 (𝑧0 ).
𝜕𝑥 0 𝜕𝑥

Proof. Let 𝑧0 = (𝑥0 , 𝑦0 ) ∈ Ω and ℎ = ℎ1 + 𝑖ℎ2 = (ℎ1 , ℎ2 ). Since 𝑢 and 𝑣 are differentiable at 𝑧0 ,
we have
( ) ( ) ( )
𝑢 𝑧0 + ℎ − 𝑢 𝑧0 = D𝑢 𝑧0 ℎ +‖ℎ‖ 𝜀1 (ℎ)
( )
= (∇𝑢 𝑧0 |ℎ) +‖ℎ‖ 𝜀1 (ℎ)
𝜕𝑢 𝜕𝑢
= (𝑧 )ℎ
𝜕𝑥 0 1
+ (𝑧 )ℎ
𝜕𝑦 0 2
+‖ℎ‖ 𝜀1 (ℎ)

and
( )
𝑣(𝑧0 + ℎ) − 𝑣(𝑧0 ) = D𝑣 𝑧0 ℎ +‖ℎ‖ 𝜀2 (ℎ)
( )
= (∇𝑣 𝑧0 |ℎ) +‖ℎ‖ 𝜀2 (ℎ)
𝜕𝑣 𝜕𝑣
= (𝑧 )ℎ
𝜕𝑥 0 1
+ (𝑧 )ℎ
𝜕𝑦 0 2
+‖ℎ‖ 𝜀2 (ℎ),

where 𝜀𝑘 (ℎ) → 0 as ℎ → 0, 𝑘 = 1, 2.
Let us write 𝜀(ℎ) = 𝜀1 (ℎ) + 𝑖𝜀2 (ℎ). We obtain using the complex functions
𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) = (𝑢(𝑧0 + ℎ) − 𝑢(𝑧0 )) + 𝑖(𝑣(𝑧0 + ℎ) − 𝑣(𝑧0 ))
( ) ( )
𝜕𝑢 𝜕𝑢 𝜕𝑣
= 𝜕𝑥 (𝑧0 )ℎ1 + 𝜕𝑦 (𝑧0 )ℎ2 + 𝑖 𝜕𝑥 (𝑧0 )ℎ1 + 𝜕𝜕𝑦𝑣 (𝑧0 )ℎ2 +‖ℎ‖ 𝜀(ℎ).

By the C–R equations

( ) ( )
𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢
𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) = (𝑧 )ℎ
𝜕𝑥 0 1
− (𝑧 )ℎ
𝜕𝑥 0 2
+𝑖 (𝑧 )ℎ
𝜕𝑥 0 1
+ (𝑧 )ℎ
𝜕𝑥 0 2
+‖ℎ‖ 𝜀(ℎ)

RHS, version May 11, 2019 34

Complex Analysis I 4 ANALYTIC FUNCTIONS

( )
𝜕𝑢 𝜕𝑣
= 𝜕𝑥
+ 𝑖 𝜕𝑥 (ℎ1 + 𝑖ℎ2 ) +‖ℎ‖ 𝜀(ℎ)
( ) ‖ℎ‖ 𝜀(ℎ)
𝜕𝑢 𝜕𝑣
= 𝜕𝑥
+ 𝑖 𝜕𝑥 ⋅ℎ+ℎ ,
ℎ

‖ℎ‖ 𝜀(ℎ)
where 𝜀0 = , if ℎ ≠ 0, 𝜀0 (0) = 0 and 𝜀0 (ℎ) → 0 as ℎ → 0.
ℎ
Hence 𝑓 is complex differentiable at 𝑧0 and
𝜕𝑢 𝜕𝑣
𝑓 ′ (𝑧0 ) = (𝑧 ) + 𝑖 (𝑧0 ).
𝜕𝑥 0 𝜕𝑥

Corollary 4.3.11. The function 𝑓 = 𝑢 + 𝑖𝑣 is complex differentiable at 𝑧0 if and only if 𝑢 and 𝑣

are differentiable in the real sense at 𝑧0 and their partial derivatives satisfy the Cauchy–Riemann
equations at 𝑧0 .

Proof. One implication is Theorem 4.3.10. The other implication follows from the proof of The-
orem 4.3.5.

Remark 4.3.12. Recall the following sufficient condition for 𝑢 ∶ ℝ2 → ℝ to be differentiable in

the real sense:
Let Ω ⊂ ℝ2 be an open set and 𝑢 ∶ Ω → ℝ. If the partial derivatives of 𝑢 exist at every point in
Ω and are continuous at every point in Ω, then 𝑢 is differentiable on Ω.
There is a weaker sufficient condition: If the partial derivatives of 𝑢 exist in a small ball 𝔹(𝑧, 𝑟)
with some 𝑟 > 0 and are continuous at 𝑧, then 𝑢 is differentiable at 𝑧.

In real life the following corollary is very important:

Corollary 4.3.13. Let Ω be an open set in ℂ and let 𝑓 = 𝑢 + 𝑖𝑣 be a complex-valued function on

Ω. If 𝑢 and 𝑣 have continuous partial derivatives on Ω and satisfy the C–R equations on Ω, then
𝑓 is analytic on Ω.

Examples 4.3.14.
1. Let 𝑓 (𝑧) = 2𝑥𝑦 + 𝑖(𝑥2 + 𝑦2 ). Now, let 𝑢 ∶ ℝ2 → ℝ and 𝑣 ∶ ℝ2 → ℝ be such that 𝑢(𝑥, 𝑦) =
2𝑥𝑦 and 𝑣(𝑥, 𝑦) = 𝑥2 + 𝑦2 . The partial derivatives of 𝑢 and 𝑣 are
𝜕𝑢 𝜕𝑢
𝑢𝑥 = 𝜕𝑥
= 2𝑦, 𝑢𝑦 = 𝜕𝑦
= 2𝑥 and
𝜕𝑣 𝜕𝑣
𝑣𝑥 = 𝜕𝑥
= 2𝑥, 𝑣𝑦 = 𝜕𝑦
= 2𝑦.

Now 𝑢𝑥 , 𝑢𝑦 , 𝑣𝑥 , 𝑣𝑦 exists and are continuous, thus we know that 𝑓 is differentiable in the
real sense of two variables. C–R equations are
{
𝑢𝑥 = 𝑣𝑦 and
𝑢𝑦 = −𝑣𝑥 .

The only points where the C–R equations are satisfied are the points where 𝑥 = 0, i.e., the
points on the imaginary axis. Hence, 𝑓 is complex differentiable at these points and

𝑓 ′ (𝑖𝑦) = 𝑢𝑥 (𝑖𝑦) + 𝑖𝑣𝑥 (𝑖𝑦) = 2𝑦.

Therefore 𝑓 fails to be analytic (there does not exist an open disk where the function has
complex derivatives, see Figure 4.3).
2. Let 𝑓 (𝑧) = (𝑧)2 . Is 𝑓 analytic?
Now, let 𝑢 ∶ ℝ2 → ℝ, 𝑢(𝑥, 𝑦) = Re 𝑓 (𝑥, 𝑦) = 𝑥3 − 3𝑥𝑦2 and 𝑣 ∶ ℝ2 → ℝ, 𝑣(𝑥, 𝑦) =
Im 𝑓 (𝑥, 𝑦) = 𝑦3 − 3𝑥2 𝑦. Homework.

RHS, version May 11, 2019 35

Complex Analysis I 4 ANALYTIC FUNCTIONS

Im

Re

Figure 4.3: Function 𝑓 in example 4.3.14.1 is complex differentiable at the imaginary axis, but
it is not analytic anywhere as there does not exist an open disk where the function has complex
derivatives.

√
Remark 4.3.15. Let 𝑓 (𝑥 + 𝑖𝑦) = |𝑥| |𝑦|, where 𝑥, 𝑦 ∈ ℝ. Show that 𝑓 satisfies the C–R
equations at the origin, but 𝑓 is not complex differentiable at the origin.

Remark 4.3.16. Let 𝑓 (𝑥 + 𝑖𝑦) = (𝑥2 + 𝑦2 ) + 𝑖2𝑦𝑥. Now the partial derivatives
⎧ 𝜕𝑓 = 2𝑥 + 𝑖2𝑦 and
⎪ 𝜕𝑥
⎨
⎪ 𝜕𝑓 = −2𝑦 + 𝑖2𝑥
⎩ 𝜕𝑦
satisfy the C–R equations. Do as in example 4.3.14.

Remark 4.3.17. We defined the operators

( ) ( )
𝜕 1 𝜕 𝜕 𝜕 1 𝜕 𝜕
= − 𝑖 and = + 𝑖 .
𝜕𝑧 2 𝜕𝑥 𝜕𝑦 𝜕𝑧 2 𝜕𝑥 𝜕𝑦

The notation for this is clear/reasonable if we write

1 1
𝑥 = (𝑧 + 𝑧) and 𝑦 = (𝑧 − 𝑧),
2 2𝑖
working formally using the chain rule, we have the following equations:
( )
𝜕𝑓 𝜕𝑓 𝜕𝑥 𝜕𝑓 𝜕𝑦 1 𝜕𝑓 1 𝜕𝑓 1 𝜕𝑓 1 𝜕𝑓
𝜕𝑧
= 𝜕𝑥 𝜕𝑧 + 𝜕𝑦 𝜕𝑧 = 𝜕𝑥 + 𝜕𝑦 = + 𝜕𝑦 ,
2 2𝑖 2 𝜕𝑥 𝑖
( )
𝜕𝑓 𝜕𝑓 𝜕𝑥 𝜕𝑓 𝜕𝑦 1 𝜕𝑓 1 𝜕𝑓 1 𝜕𝑓 1 𝜕𝑓
𝜕𝑧
= 𝜕𝑥 𝜕𝑧 + 𝜕𝑦 𝜕𝑧 = 𝜕𝑥 − 𝜕𝑦 = − 𝜕𝑦 .
2 2𝑖 2 𝜕𝑥 𝑖

4.3.18 On harmonic functions

Let 𝑓 = 𝑢 + 𝑖𝑣. The (C–R) equations

⎧ 𝜕𝑢 𝜕𝑣
⎪ 𝜕𝑥 = 𝜕𝑦
⎨ 𝜕𝑣 𝜕𝑢
⎪ 𝜕𝑥 = − 𝜕𝑦
⎩
have an interesting consequence:
If we only know the real part of an analytic function, we can deduce the imaginary part 𝑣 (up to
a constant), by integrating the C–R equations.

Theorem 4.3.19. If 𝑓1 = 𝑢 + 𝑖𝑣1 and 𝑓2 = 𝑢 + 𝑖𝑣2 are two analytic functions which are defined
on the same domain and which have the same real part 𝑢, then we know that 𝑣1 = 𝑣2 + 𝐶 for
some constant 𝐶.
We need following Lemma 4.3.20 from the vector calculus course:

RHS, version May 11, 2019 36

Complex Analysis I 4 ANALYTIC FUNCTIONS

Lemma 4.3.20. If 𝑘 is a function on a domain 𝐷 such that

𝜕𝑘 𝜕𝑘
= =0 on 𝐷,
𝜕𝑥 𝜕𝑦
then 𝑘 is a constant.
𝜕𝑘
Proof. Since 𝜕𝑥
= 0, 𝑘 is constant on every horizontal line segment in 𝐷 by the fundamental
theorem of calculus. As also 𝜕𝜕𝑦𝑘 = 0 similarly 𝑘 is constant. Since every two points in 𝐷 can be
connected be vertical and horizontal line segments, 𝑘 is constant.

Proof. (of the theorem) From the C–R equations we have that

𝜕𝑢 𝜕𝑣 𝜕𝑣 𝜕𝑢 𝜕𝑣 𝜕𝑣
= 1 = 2 and = − 1 = − 2,
𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑥
thus
𝜕(𝑣1 − 𝑣2 ) 𝜕(𝑣1 − 𝑣2 )
=0= .
𝜕𝑦 𝜕𝑥
Then by the above Lemma 4.3.20 𝑣1 − 𝑣2 = 𝐶.

Definition 4.3.21. If 𝑢 + 𝑖𝑣 is an analytic function, then 𝑣 is called a harmonic conjugate of 𝑢,

and vice versa.

The previous Theorem 4.3.19 tells us that up to a constant, every function 𝑢 has only one harmonic
conjugate. Note that not every function has a harmonic conjugate.
Let us take the C–R equations:
𝜕𝑢 𝜕𝑣 𝜕𝑢 𝜕𝑣
(1) = and (2) =− ,
𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥
and differentiate (1) with respect to 𝑥 and (2) with respect to 𝑦.

𝜕2𝑢 𝜕2𝑣 𝜕2𝑢 𝜕2𝑣

(3) = and (4) =− .
𝜕𝑥2 𝜕𝑥𝜕𝑦 𝜕𝑦2 𝜕𝑦𝜕𝑥

Now we assume that

𝜕2𝑣 𝜕2𝑣
= .
𝜕𝑥𝜕𝑦 𝜕𝑦𝜕𝑥
It is enough to assume that 𝑣 ∶ ℝ2 → ℝ is 𝐶 2 in the real sense. Adding (3) and (4) together we
obtain
𝜕2𝑢 𝜕2𝑢
(5) + = 0,
𝜕𝑥2 𝜕𝑦2
which is known the Laplace equation Δ2 𝑢 = 0. Functions which satisfy this rule are called
harmonic.

Remark 4.3.22. We have just shown that in order for 𝑢 to be the real part of an analytic function,
then 𝑢 must be harmonic.

Recall the Laplace operator:

𝜕2 𝜕2
Δ ∶= + . (4.3.23)
𝜕𝑥 2 𝜕𝑦2

Let Ω be an open set and 𝑓 ∶ Ω → ℝ be a 𝐶 2 -function (a function which is twice continuously

differentiable in the real sense, we write 𝑓 ∈ 𝐶 2 (Ω)).
If Δ𝑓 (𝑥, 𝑦) = 0 for all (𝑥, 𝑦) ∈ Ω, then 𝑓 is called harmonic. Using the C–R equations we obtain
the following property for analytic functions:

Proposition 4.3.24. The real part and the imaginary part of an analytic function are harmonic.

RHS, version May 11, 2019 37

Complex Analysis I 4 ANALYTIC FUNCTIONS

Warning: We will use the fact that an analytic function is twice continuously differentiable (in
fact, it is 𝑓 ∈ 𝐶 ∞ (Ω)), but we will prove this property later.
Let Ω be an open set. Let 𝑓 ∶ Ω → ℂ be analytic, and the real part of 𝑓 , Re 𝑓 ∶ Ω → ℂ,
(Re 𝑓 )(𝑧) = Re 𝑓 (𝑧). By the C–R equations, we have:
𝜕 2 Re 𝑓 𝜕 2 Re 𝑓
Δ Re 𝑓 = 𝜕𝑥 2 + 𝜕𝑦2
𝜕 𝜕 𝜕 𝜕
= 𝜕𝑥 𝜕𝑥
Re 𝑓 + 𝜕𝑦 𝜕𝑦
Re 𝑓
𝜕 𝜕 𝜕 𝜕
= 𝜕𝑥 𝜕𝑦
Im 𝑓 − 𝜕𝑦 𝜕𝑥
Im 𝑓 = 0.

Hence, Δ Re 𝑓 = 0 and Re 𝑓 is harmonic. In the same way, Δ Im 𝑓 = 0, and Im 𝑓 is harmonic.

There exists the following theorem, which we do not prove:
( )
Theorem. If 𝑢 is harmonic on 𝔻 𝑧0 , 𝑟 with 𝑟 > 0, then there exists an analytic map 𝑓 on
( )
𝔻 𝑧0 , 𝑟 such that 𝑢 = Re 𝑓 .

Example. 𝑢(𝑥, 𝑦) = 𝑎𝑥 + 𝑏𝑦 + 𝑐, where 𝑎, 𝑏 ∈ ℝ, is harmonic since Δ𝑢(𝑥, 𝑦) = 0.

Theorem 4.3.25. Let 𝑓 be analytic on a disc 𝔻. If Re 𝑓 is constant in 𝔻, then 𝑓 is constant in

𝔻.

Proof. Let 𝑓 = 𝑢 + 𝑖𝑣 = Re 𝑓 + 𝑖 Im 𝑓 . If 𝑢 = Re 𝑓 is constant, then 𝑢𝑥 = 𝑢𝑦 = 0. Since 𝑓 is

analytic, the C–R equations imply that 𝑣𝑥 = 𝑣𝑦 = 0.
From the vector calculus course we know that this means that 𝑓 is constant.

Theorem 4.3.26. If 𝑓 is analytic on a disc 𝔻 and Im 𝑓 or |𝑓 | or arg 𝑓 is constant, then 𝑓 is

constant.

Remark 4.3.27. Different ways to show that 𝑓 is not complex differentiable at 𝑧0 :

1. If 𝑓 is discontinuous at 𝑧0 , then 𝑓 is not complex differentiable at 𝑧0 .
𝑓 (𝑧)−𝑓 (𝑧 )
2. If the limit for 𝑧−𝑧 0 fails to exist when 𝑧 → 𝑧0 , then 𝑓 is not complex differentiable at
0
𝑧0 . For an example, see Homework 3.1.
3. if the C–R equations are not satisfied at (𝑥0 , 𝑦0 ) = 𝑧0 , then 𝑓 is not complex differentiable
at 𝑧0 . Here 𝑓 = 𝑢 + 𝑖𝑣.
4. If 𝑓 is not differentiable in the sense of two real variables at (𝑥0 , 𝑦0 ), then 𝑓 is not complex
differentiable at 𝑧0 = (𝑥0 , 𝑦0 ).
5. Use the differentiation formulas for sum, multiplication and quotient. For an example, see
Homework 3.3.

Important! Let 𝑓 = 𝑢 + 𝑖𝑣 ∶ Ω → Ω, Ω ⊂ ℂ open. In this chapter we proved:

Theorem 4.3.28. If 𝑓 is complex differentiable at 𝑧0 ∈ Ω, then the partial derivatives of 𝑢 and 𝑣

exist and they obey the R–C equations:
𝜕𝑢 𝜕𝑣 𝜕𝑢 𝜕𝑣 𝜕𝑢 ( ) 𝜕𝑣 ( )
= and =− and 𝑓 ′ (𝑥) = 𝑧0 + 𝑖 𝑧 .
𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑥 0

Example 4.3.29. Function 𝑧 ↦ 𝑧2 is analytic in the whole complex plane, as

(𝑥 + 𝑖𝑦) ↦ (𝑥 + 𝑖𝑦)2 = 𝑥2 − 𝑦2 + 𝑖2𝑥𝑦 .

⏟⏟⏟ ⏟⏟⏟
𝑢 𝑣

RHS, version May 11, 2019 38

Complex Analysis I 4 ANALYTIC FUNCTIONS

Now
⎧ 𝜕𝑢 𝜕𝑣
⎪ 𝜕𝑥 = 2𝑥 = 𝜕𝑦
⎨ 𝜕𝑣 𝜕𝑢
⎪ 𝜕𝑦 = −2𝑌 = − 𝜕𝑥
⎩
so 𝑧2 satisfies C–R equations. By the product formula 𝑓 ′ (𝑧0 ) = 2𝑧0 and
( ) ( ) 𝜕𝑢 ( ) 𝜕𝑣 ( )
𝑓 ′ 𝑥0 + 𝑖𝑦0 = 2 𝑥0 + 𝑖𝑦0 = 𝑧0 + 𝑖 𝑧 .
𝜕𝑥 𝜕𝑥 0

Theorem 4.3.30. Suppose that 𝑓 = 𝑢 + 𝑖𝑣 ∶ Ω ↦ ℂ, Ω ⊂ ℂ open. If 𝑢 and 𝑣 are differentiable

in the real sense at 𝑧0 ∈ Ω and their partial derivates obey the C–R equations at 𝑧0 , then 𝑓 has
a complex derivative at 𝑧0 and
( ) 𝜕𝑢 ( ) 𝜕𝑣 ( )
𝑓 ′ 𝑥0 + 𝑖𝑦0 = 𝑧0 + 𝑖 𝑧 .
𝜕𝑥 𝜕𝑥 0

Remark 4.3.31. (Important!) If the partial derivatives 𝜕𝜕𝑥𝑓 and 𝜕𝜕𝑦𝑓 exist and are continuous at 𝑧0 ,
then the following is true:
𝑓 (𝑧) is complex differentiable at 𝑧0 if and only if the C–R equations

𝜕𝑓 ( ) 1 𝜕𝑓 ( )
𝑧 = 𝑧
𝜕𝑥 0 𝑖 𝜕𝑦 0
hold.

4.3.32
Let Ω ⊂ ℝ2 be an open set. Consider 𝑓 ∶ Ω → ℝ2 , a function of two variables. Recall that 𝑓 is
differentiable at the point 𝑧0 if there exists an ℝ-linear mapping from ℝ2 to ℝ2 – the derivative,
or differential 𝑓 (𝑧0 ) – such that

𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) = 𝑓 (𝑧0 )ℎ +‖ℎ‖ 𝜀(ℎ),

where‖𝜀‖ → 0 as‖ℎ‖ → 0 and 𝑧0 , ℎ ∈ ℝ2 .

On the other hand, the fact that the corresponding function 𝑓 ∶ Ω → ℂ, Ω ⊂ ℂ, is complex
differentiable at 𝑧0 means that there exists a complex number 𝛼 such that

𝑓 (𝑧0 + ℎ) − 𝑓 (𝑧0 ) = 𝛼 ⋅ ℎ + ℎ𝜀(ℎ),

where limℎ→0 𝜀(ℎ) = 0.

Hence, 𝑓 is complex differentiable at 𝑧0 if and only if 𝑓 is differentiable in the real sense, and the
derivative 𝑓 (𝑧0 ) is of the form 𝑓 (𝑧0 )ℎ = 𝛼 ⋅ ℎ, with 𝛼 ∈ ℂ.
But the ℝ-linear mappings from ℝ2 to ℝ2 of this type are exactly the ℂ-linear mappings. So we
have:
( )
Remark. The function 𝑓 , defined in 𝔻 𝑧0 , 𝑟 ⊂ ℂ, is complex differentiable at 𝑧0 if and only if
𝑓 is differentiable in the real sense and the derivative 𝑓 (𝑧0 ) is ℂ-linear.

The plane ℝ2 (ℂ) is an ℝ-vector space. Hence, we can consider the mappings 𝐿 ∶ ℂ → ℂ, which
are ℝ-linear:

𝐿(𝑧 + 𝑤) = 𝐿(𝑧) + 𝐿(𝑤), 𝑧, 𝑤 ∈ ℂ

𝐿(𝜆𝑧) = 𝜆𝐿(𝑧), 𝜆 ∈ ℝ.

There is a corresponding matrix of 𝐿 with respect to the basis {(1, 0) = 1, (0, 1) = 𝑖}:
( )
𝑎11 𝑎12
𝐿= , 𝑎𝑘𝑗 ∈ ℝ.
𝑎21 𝑎22

RHS, version May 11, 2019 39

Complex Analysis I 4 ANALYTIC FUNCTIONS

So that

𝐿 ∶ (𝑥, 𝑦) ↦ (𝑎11 𝑥 + 𝑎12 𝑦, 𝑎21 𝑥 + 𝑎22 𝑦)

𝐿(1, 0) = (𝑎11 , 𝑎21 ) = 𝑎11 + 𝑖𝑎21
𝐿(0, 1) = (𝑎12 , 𝑎22 ) = 𝑎12 + 𝑖𝑎22 .

If 𝑧 = 𝑥 + 𝑖𝑦, then by the ℝ-linearity of 𝐿:

𝐿(𝑧) = 𝐿(𝑥 + 𝑖𝑦)

= 𝑥𝐿(1, 0) + 𝑦𝐿(0, 1)
( )
= 𝑎11 𝑥 + 𝑎12 𝑦 + 𝑖 𝑎21 𝑥 + 𝑎22 𝑦
= 𝛼(𝑥 + 𝑖𝑦) + 𝛽(𝑥 − 𝑖𝑦),
1 ( ) 1 ( )
with 𝛼 = 2
𝑎11 + 𝑎22 − 𝑖𝑎12 + 𝑖𝑎21 and 𝛽 = 2
𝑎11 − 𝑎22 + 𝑖𝑎12 + 𝑖𝑎21 .
Hence, 𝐿(𝑧) = 𝛼𝑧 + 𝛽𝑧, where 𝛼, 𝛽 ∈ ℂ, is the general expression of an ℝ-linear mapping.
On the other hand, ℂ is also a ℂ-vector space. So 𝐿 ∶ ℂ → ℂ is ℂ-linear, if

𝐿(𝑧 + 𝜔) = 𝐿(𝑧) + 𝐿(𝜔), 𝑧, 𝜔 ∈ ℂ

𝐿(𝜆𝑧) = 𝜆𝐿(𝑧), 𝜆 ∈ ℂ.

Now 𝐿(𝑧) = 𝛼𝑧, because

𝐿(𝑥 + 𝑖𝑦) = 𝐿(𝑥) + 𝐿(𝑖𝑦) = 𝐿(𝑥) + 𝑖𝐿(𝑦)

= 𝑥𝐿(1) + 𝑖𝑦𝐿(1)
= (𝑥 + 𝑖𝑦) 𝐿(1).
⏟⏟⏟ ⏟⏟⏟
𝑧 𝛼

We note from the previous calculations that 𝛽 = 0 if and only if 𝑎11 = 𝑎22 and 𝑎12 = −𝑎21 .
Hence, in the Jabocian matrix [ ]
𝑢𝑥 𝑢𝑦
,
𝑣𝑥 𝑣𝑦
𝑢𝑥 = 𝑣𝑦 and 𝑢𝑦 = −𝑣𝑥 , which means that the C–R equations are satisfied.

Example. As a first approximation the surface of soap films (or any elastic surface) is the graph
of harmonic function.

RHS, version May 11, 2019 40

Complex Analysis I 5 COMPLEX SERIES

5 Complex Series
5.1 Series of complex terms
Let (𝑧𝑛 ) be a complex sequence. An expression

∑
∞
𝑧𝑛 = 𝑧1 + 𝑧2 + … (5.1.1)
𝑛=1
is an infinite series. The numbers

∑
𝑁
𝑧𝑛 =∶ 𝑆𝑁 , 𝑁 = 1, 2, … (5.1.2)
𝑛=1

are called the partial sums of the series.

Definition 5.1.3. If the sequence of partial sums converges to some number 𝑆 ∈ ℂ, that is,
lim𝑁→∞ 𝑆𝑁 = 𝑆, then we define
∑∞
𝑧𝑛 = 𝑆,
𝑛=1
∑∞
and we say that the series 𝑛=1 𝑧𝑛 converges to (the sum) 𝑆. We call 𝑆 the sum of the series
∑∞
𝑛=1 𝑧𝑛 . A series which converges to some number 𝑆 ∈ ℂ is called a convergent series. If a
series does not converge, we say that it diverges or that it is divergent.

Remark. It follows from the definition that questions of the convergence of series are questions
of the convergence of sequences.

Proposition 5.1.4: Linear combinations of series. If

∑
∞ ∑
∞
𝑧𝑛 = 𝑠 and 𝑤𝑛 = 𝑡,
𝑛=1 𝑛=1

then
∑
∞
( ) ∑
∞
𝑧𝑛 + 𝑤𝑛 = 𝑠 + 𝑡 and 𝜆𝑧𝑛 = 𝜆𝑠, 𝜆 ∈ ℂ.
𝑛=1 𝑛=1
∑∞
Proposition 5.1.5. The complex series 𝑛=1 𝑧𝑛 converges if and only if the real series

∑
∞
( ) ∑
∞
( )
Re 𝑧𝑛 and Im 𝑧𝑛
𝑛=1 𝑛=1

both converge.

The propositions above tell us that we can write

∑
∞ ∑
∞
( ) ∑
∞
( )
𝑧𝑛 = Re 𝑧𝑛 + 𝑖 Im 𝑧𝑛 ,
𝑛=1 𝑛=1 𝑛=1

whenever the two series on the right-hand side converge, or the series on the left-hand side con-
verges.

Remarks 5.1.6.
1. If a series of complex numbers converges, then the 𝑛th term converges to zero as 𝑛 tends to
∑
infinity. Note that the converse is generally not true: ∞ 1 1
𝑛=1 𝑛 diverges, but lim𝑛→∞ 𝑛 = 0.
∑
2. The terms of a convergent series are bounded. That is, when the series ∞ 𝑛=1 𝑧𝑛 converges,
then there exists a positive constant 𝑀 such that ||𝑧𝑛 || ≤ 𝑀 for all positive integers 𝑛.

RHS, version May 11, 2019 41

Complex Analysis I 5 COMPLEX SERIES

∑∞
Definition 5.1.7. The series 𝑛=1 𝑧𝑛 is said to be absolutely convergent if the series
∑
∞
|𝑧 |
| 𝑛|
𝑛=1
√
( )2 ( )2
of real numbers ||𝑧𝑛 || = Re 𝑧𝑛 + Im 𝑧𝑛 converges.

Proposition 5.1.8: Absolute convergence vs. convergence. The absolute convergence of a

series of complex numbers implies the convergence of that series.

In establishing the fact that the sum of a given series is a given number 𝑆, it is often convenient
to define the remainder 𝜚𝑁 after 𝑁 terms:
∑
Definition 5.1.9. Let 𝑆 ∈ ℂ be given. Let 𝑆𝑁 = 𝑁 𝑗=1 𝑧𝑗 be the partial sum of the series
∑∞
𝑧
𝑗=1 𝑗 . The remainder 𝜚 𝑁 after 𝑁 terms is 𝜚 𝑁 = 𝑆 − 𝑆𝑁 . Thus 𝑆 = 𝑆𝑁 + 𝜚𝑁 , and since
|𝑆 − 𝑆𝑁 | = |𝜚𝑁 |, we see that a series ∑∞ 𝑧𝑛 converges to a number 𝑆 ∈ ℂ if and only if the
| | | | 𝑛=1
sequence of remainders tends to zero.

Testing for convergence

∑ ∑∞ | |
Let ∞ 𝑛=1 𝑧𝑛 be a complex series. The associated series 𝑛=1 |𝑧𝑛 | has real, non-negative terms.
Hence well-known tests (from real analysis) for the convergence of a series with non-negative
terms can be applied. Combining this with Proposition 5.1.8 we obtain a sufficient condition for
∑
the convergence of ∞ 𝑛=1 𝑧𝑛 .
∑
Remark. The geometric series 𝑧𝑛
∑
• converges and ∞ 1
𝑛=0 𝑧 = 1−𝑧 , if|𝑧| < 1
𝑛

• fails to converge, if|𝑧| ≥ 1.

5.1.10 The comparison test

∑
Suppose that ∞ 𝑛=1 𝑤𝑛 is a convergent series with 𝑤𝑛 ≥ 0 for all 𝑛 and suppose that for some
∑
constant 𝑘 > 0, ||𝑧𝑛 || ≤ 𝑘𝑤𝑛 for all 𝑛. Then ∞𝑛=1 𝑧𝑛 converges absolutely and hence it converges.

5.1.11 d’Alembert’s ratio test

∑
Suppose that ∞
𝑛=1 𝑧𝑛 is a series of complex terms such that the limit
|𝑧 |
| 𝑛+1 |
lim | | =∶ 𝑞
𝑛→∞| 𝑧𝑛 |
| |
exists.
∑
If 0 ≤ 𝑞 < 1, then ∞𝑛=1 𝑧𝑛 converges absolutely.
∑∞
If 𝑞 > 1, then 𝑛=1 𝑧𝑛 diverges.
If 𝑞 = 1, then the test gives no information.

5.1.12 The 𝑛th root test

( )
Suppose that 𝑧𝑛 is a complex sequence such that the limit
√
lim 𝑛 ||𝑧𝑛 || =∶ 𝓁
𝑛→∞
exists.
∑∞
If 𝓁 < 1, then 𝑛=1 𝑧𝑛 converges absolutely.
∑∞
If 𝓁 > 1, then 𝑛=1 𝑧𝑛 diverges.
If 𝓁 = 1, then the test gives no information.

RHS, version May 11, 2019 42

Complex Analysis I 5 COMPLEX SERIES

Remark (for further reading). There exists a useful test called Raabe’s test.

Examples 5.1.13.
1. The series
∑
∞
(−𝑖)𝑛
𝑛=1 𝑛2
converges, since it is absolutely convergent:

∑∞| 𝑛|
| (−𝑖) | ∑ 1
∞
| 2 |= .
| | 2
𝑛=1| 𝑛 | 𝑛=1 𝑛
∑∞ 1
Recall from real analysis that a series 𝑛=1 𝑛𝑝 converges if 𝑝 > 1 and diverges otherwise.
2. The series
∞ (
∑
)𝑛
−1
𝑖
𝑛=0
𝑛!
(−1)𝑛 ∑∞
is convergent. Let 𝑥𝑛 = 0 and 𝑦𝑛 = . Evidently, the series 𝑛=1 𝑥𝑛 is convergent.
∑ 𝑛!
Likewise, ∞ 𝑛=1 𝑦𝑛 is convergent:

∞ (
∑
)𝑛
−1
= 𝑒−1 .
𝑛=1
𝑛!

And so
∞ (
∑
)𝑛
−1 𝑖
𝑖= .
𝑛=1
𝑛! 𝑒

3. The series ( )
∑
∞
1 1
+𝑖
𝑛2 𝑛
𝑛=1
∑∞ 1
diverges, since the harmonic series 𝑛=1 𝑛 is divergent.
4. The series
∑∞ ( )𝑛−1 ( )2𝑛−1
−1 2𝑖
( )
𝑛=1 2𝑛 − 1 !
converges.
Let ( )𝑛−1 ( )2𝑛−1 ( )𝑛−1 ( )2𝑛
−1 2𝑖 1 −1 2𝑖
𝑧𝑛 = ( ) = ( ) ,
2𝑛 − 1 ! 2𝑖 2𝑛 − 1 !
then
| ( ) ( )𝑛 ( )2𝑛+2 |
| 𝑧 | || 2𝑛 − 1 ! −1 2𝑖
|
|
| 𝑛+1 | | |= 4
| | =|( ) ( ) ( ( ) ) | ( ) ⟶ 0.
| 𝑧𝑛 | | 𝑛−1 2𝑛 | 2𝑛 2𝑛 + 1 𝑛→∞
| | | −1 2𝑖 2 𝑛 + 1 − 1 !|
| |
Hence, the series converges by d’Alembert’s ratio test.
5. Let us prove that
∑
∞
1
𝑧𝑛 = , whenever |𝑧| < 1.
𝑛=0
1−𝑧
( )( )
We achieve this with the help of remainders. Recall that 1 − 𝑧 1 + 𝑧 + ⋯ + 𝑧𝑛 = 1 −
𝑧𝑛+1 , so
1 − 𝑧𝑛+1
1 + 𝑧 + ⋯ + 𝑧𝑛 = , 𝑧 ≠ 1.
1−𝑧

RHS, version May 11, 2019 43

Complex Analysis I 5 COMPLEX SERIES

Now, we may write the partial sum

∑
𝑁
𝑆𝑁 (𝑧) = 𝑧𝑛 = 1 + 𝑧 + ⋯ + 𝑧𝑁 , 𝑧 ≠ 1,
𝑛=0
as
1 − 𝑧𝑁+1
𝑆𝑁 (𝑧) = , 𝑧 ≠ 1.
1−𝑧
1
If 𝑆(𝑧) = 1−𝑧
, then

𝑧𝑁+1
𝜚𝑁 (𝑧) = 𝑆(𝑧) − 𝑆𝑁 (𝑧) = , 𝑧 ≠ 1.
1−𝑧
Thus
𝑁+1
|𝜚𝑁 (𝑧)| = |𝑧|
| | |1 − 𝑧|
| |
and the remainders 𝜚𝑁 (𝑧) tend to zero when|𝑧| < 1, but not when|𝑧| ≥ 1. Hence
∑
∞
1
𝑧𝑛 = , whenever |𝑧| ≤ 1.
𝑛=0
1−𝑧

5.2 Power series

A power series is an expansion of the form
∑
∞
( )𝑛 ( ) ( )𝑛
𝑎𝑛 𝑧 − 𝑧0 = 𝑎0 + 𝑎1 𝑧 − 𝑧0 + ⋯ + 𝑎𝑛 𝑧 − 𝑧0 + … ,
𝑛=0

where 𝑧0 and the coefficients 𝑎𝑛 are complex constants and 𝑧 may be any point in a stated domain
containing 𝑧0 . In such a series involving a variable 𝑧, we shall denote sums, partial sums and
remainders by 𝑆(𝑧), 𝑆𝑁 (𝑧) and 𝜚𝑁 (𝑧), respectively. Our last example 5.1.13(5) gives an example
of a power series.
In the power series expansion we shall often assume, without loss of generality, that 𝑧0 = 0.
∑
Theorem 5.2.1. (Theorem of Abel) If ∞ 𝑎 𝑧𝑛 converges at some 𝑧1 ∈ ℂ, then the series
( )𝑛=0 𝑛
| |
converges absolutely at each 𝑧 ∈ 𝔻 0,|𝑧1 | .

𝑧1

0
𝑧

Figure 5.1: Theorem( 5.2.1:)If a power series converges at some 𝑧1 ∈ ℂ, then it converges abso-
lutely at each 𝑧 ∈ 𝔻 0,||𝑧1 || .
∑ | |
Proof. Since 𝑎𝑛 𝑧𝑛 converges at 𝑧1 ≠ 0, it follows that lim𝑘→∞ |𝑎𝑘 𝑧𝑘1 | = 0. So there exists a
| |
| |
real constant 𝑀 such that |𝑎𝑘 𝑧𝑘1 | ≤ 𝑀 for all 𝑘 ∈ ℕ. Hence for each 𝑘
| |
| |𝑘
| 𝑘 | | 𝑘 || 𝑧 |
|𝑎𝑘 𝑧 | = |𝑎𝑘 𝑧1 || | ≤ 𝑀𝑞 𝑘 ,
| | | || 𝑧1 |
| |

RHS, version May 11, 2019 44

Complex Analysis I 5 COMPLEX SERIES

|𝑧| ( ) ∑
where 𝑞 = < 1, when |𝑧| < ||𝑧1 ||, that is, 𝑧 ∈ 𝔻 0,||𝑧1 || . The series 𝑞 𝑘 converges as a
|𝑧1 |
( ) ∑ ∑| |
geometric series in 𝔻 0,||𝑧1 || . Hence 𝑀𝑞 𝑘 converges and by the comparison test |𝑎𝑘 𝑧𝑘 |
∑ | |
converges, when |𝑧| < ||𝑧1 || and the original series 𝑎𝑘 𝑧𝑘 converges absolutely when |𝑧| < ||𝑧1 ||.

∑ ∑
Corollary 5.2.2. If 𝑎𝑘 𝑧𝑘 fails to converge at some 𝑧2 ∈ ℂ, then 𝑎𝑘 𝑧𝑘 diverges for all|𝑧| >
|𝑧2 |.
| |

𝑧2

Figure 5.2: Corollary 5.2.2: If a power series fails to converge at some 𝑧2 ∈ ℂ, then it diverges
for all|𝑧| > ||𝑧2 ||.

∑
Definition 5.2.3. The radius of convergence of the power series 𝑎𝑛 (𝑧 − 𝑧0 )𝑛 is defined to be
{ ∑ }
𝑅 ∶= sup ||𝑧 − 𝑧0 || ∶ |𝑎𝑛 (𝑧 − 𝑧0 )𝑛 | converges at 𝑧 .
| |
∑| |𝑛
Here we write 𝑅 = ∞, if |𝑎𝑛 (𝑧 − 𝑧0 )| converges for arbitrarily large 𝑧 − 𝑧0 .
∑
Lemma 5.2.4. (Radius of convergence) Let 𝑎𝑘 (𝑧 − 𝑧0 )𝑘 be a power series with radius of con-
vergence 𝑅. Then there are three possibilities:
∑
1. 𝑅 = 0 ∶ the series 𝑎𝑘 (𝑧 − 𝑧0 )𝑘 converges only at the point 𝑧0 .
∑
2. 0 < 𝑅 < ∞ ∶ the series 𝑎𝑘 (𝑧 − 𝑧0 )𝑘 converges absolutely for all 𝑧 with ||𝑧 − 𝑧0 || < 𝑅
∑
and the series 𝑎𝑘 (𝑧 − 𝑧0 )𝑘 fails to converge for any 𝑧 with ||𝑧 − 𝑧0 || > 𝑅
∑
3. 𝑅 = ∞ ∶ the series 𝑎𝑘 (𝑧 − 𝑧0 )𝑘 converges absolutely for all 𝑧 ∈ ℂ.
( )
Definition 5.2.5. If 0 < 𝑅 < ∞ then the disc 𝔻 𝑧0 , 𝑅 is called the disc of convergence. If
𝑅 = 0 the disc of convergence is ∅. If 𝑅 = ∞ then the disc of convergence is ℂ.

𝑅
𝑧0

( )
Figure 5.3: Definition 5.2.5: The disc of convergence 𝔻 𝑧0 , 𝑅 .

RHS, version May 11, 2019 45

Complex Analysis I 5 COMPLEX SERIES

Remark. On the boundary of the disc of convergence, ||𝑧 − 𝑧0 || = 𝑅, the situation is delicate, as
one can have either convergence or divergence.

Examples 5.2.6. For each of the following power series, calculate the radius of convergence:

∑
∞ 𝑛
𝑧 ∑
∞ 𝑛
𝑧 ∑
∞
(1) , (2) , (3) 𝑛!𝑧𝑛 .
𝑛=1
𝑛 𝑛=1
𝑛! 𝑛=1

∑| 1 𝑛 |
(1) We apply the Ratio test to | 𝑛 𝑧 |. For 𝑧 ≠ 0,
| |
| 𝑧𝑛+1 𝑛 |
| | 𝑛
| |= |𝑧| → |𝑧| ,
| 𝑛 + 1 𝑧𝑛 | 𝑛 + 1
| |
∑| 1 𝑛 |
as 𝑛 → ∞. Hence | 𝑛 𝑧 | converges if|𝑧| < 1 and fails to converge if|𝑧| > 1. We conclude
| |
that 𝑅 = 1.
∑| 𝑧𝑛 |
(2) We apply the Ratio test to | 𝑛! |. For 𝑧 ≠ 0,
| |
| 𝑧𝑛+1 𝑛! | |𝑧|
| |
| |= → 0,
| (𝑛 + 1)! 𝑧𝑛 | 𝑛 + 1
| |
∑| 𝑧 𝑛 |
as 𝑛 → ∞. Hence | 𝑛! | converges for all 𝑧 ∈ ℂ. We deduce that 𝑅 = ∞.
| |
∑
(3) We apply the Ratio test to ||𝑛!𝑧𝑛 ||. For 𝑧 ≠ 0,
| (𝑛 + 1)!𝑧𝑛+1 | ( )
| |
| | = 𝑛 + 1 |𝑧| → ∞,
| 𝑛!𝑧𝑛 |
| |
∑| 𝑛 |
as 𝑛 → ∞. Hence |𝑛!𝑧 | converges only if 𝑧 = 0. So 𝑅 = 0.
For calculating the radius of convergence we may apply the Ratio test and the n𝑡ℎ root test. But we
need the limits and sometimes they do not exist. The formula of Hadamard solves the problem!

Note. We use the following convention that

1 1
= ∞ and = 0.
0 ∞

Remark. Recall the definition of limes superior. Let (𝑥𝑘 ) be a sequence of real numbers. Then
{ }
lim 𝑥𝑘 = lim sup 𝑥𝑘 = inf sup 𝑥𝑘 , 𝑥𝑘+1 , … .
𝑘→∞ 𝑘→∞ 𝑘≥1

Example. Let (𝑥𝑘 ) be a sequence where 𝑥𝑘 = (−1)𝑘 , 𝑘 ≥ 1. Note that the sequence

(𝑥𝑘 ) = (−1, 1, −1, … )

fails to converge, but

lim sup 𝑥𝑘 = 1.
𝑘→∞
∑
Theorem 5.2.7: Hadamard’s theorem. The radius of convergence of the power series 𝑎𝑘 (𝑧 −
𝑧0 )𝑘 is given by the formula
1 1
= lim sup||𝑎𝑘 || 𝑘 . (5.2.8)
𝑅 𝑘→∞

Examples 5.2.9. Find the radius of convergence for the following series:

RHS, version May 11, 2019 46

Complex Analysis I 5 COMPLEX SERIES

1.
∑
∞ 𝑛2
𝑧 𝑧4 𝑧9 𝑧16 ∑ ∞
=𝑧+ + + +⋯= 𝑎𝑗 𝑧𝑗 ,
𝑛=1 𝑛2 4 9 16 𝑗=1

where {1
𝑗
, when 𝑗 = 𝑛2 for some 𝑛 ∈ ℕ
𝑎𝑗 =
0, otherwise.
The Hadamard formula 5.2.8 gives
1
1 | |
1
| 1 | 𝑛2 1
= lim sup|𝑎𝑗 | 𝑗 = lim sup|| || = sup 1 = 1,
𝑅 𝑗→∞ | | 𝑛→∞ | 𝑛2 | 𝑘→∞ 𝑘 𝑘

since
1 1 1
𝑘 𝑘 = 𝑒log 𝑘 𝑘 = 𝑒 𝑘 log 𝑘 → 1.

2.
∑
∞
2
𝑎𝑛 𝑧𝑛 , 𝑎 ∈ ℂ.
𝑛=1
The Hadamard formula 5.2.8 gives

1 1 𝑛2
= lim ||𝑎𝑛 || 𝑛 = lim |𝑎| 𝑛 = lim |𝑎|𝑛 .
𝑅 𝑛→∞ 𝑛→∞ 𝑛→∞

Hence,
⎧0, if |𝑎| > 1
⎪
𝑅 = ⎨1, if |𝑎| = 1
⎪∞, if |𝑎| < 1.
⎩

Proof of Hadamard’s theorem. Let 𝐿 = 𝑅1 where 𝑅 is defined by the Hadamard’s formula 5.2.8.
Suppose that 𝐿 ≠ 0, ∞. If|𝑧| < 𝑅, choose 𝜀 > 0 such that

(𝐿 + 𝜀)|𝑧| = 𝑟 < 1.
1
By the definition of L, we have ||𝑎𝑛 || 𝑛 ≤ 𝐿 + 𝜀 for all large 𝑛. Hence,
|𝑎𝑛 ||𝑧|𝑛 ≤ (𝐿 + 𝜀)𝑛 |𝑧|𝑛 = 𝑟𝑛 .
| |
∑ ∑
By the Comparison test with the geometric series 𝑟𝑛 we can show that 𝑎𝑛 𝑧𝑛 converges.
If|𝑧| > 𝑅, then a similar argument proves that there exists a sequence of terms in the series whose
absolute values go to infinity, and hence the sequence, and also the series diverges.

Power series provide a very important class of analytic functions that are easy to manipulate.
The following theorem is very important.

Theorem 5.2.10. The power series

∑
∞
𝑓 (𝑧) = 𝑎𝑛 𝑧𝑛
𝑛=0

defines an analytic function in its disc of convergence. The derivative of 𝑓 is also a power series
obtained by differentiating the series for 𝑓 term by term, that is,

∑
∞
′
𝑓 (𝑧) = 𝑛𝑎𝑛 𝑧𝑛−1 .
𝑛=0

Moreover, 𝑓′ has the same radius of convergence as 𝑓 .

RHS, version May 11, 2019 47

Complex Analysis I 5 COMPLEX SERIES

Remark. There is the obvious candidate for the derivative:

∑
𝑛𝑎𝑛 𝑧𝑛−1 .

But this assumes that we differentiate the series ’term-by-term’. Note that
∑ ∑ d
𝑛𝑎𝑛 𝑧𝑛−1 = 𝑎 𝑧𝑛
d𝑧 𝑛
whereas ( )
d ∑
𝑓 ′ (𝑧) =
𝑎𝑛 𝑧𝑛 ,
d𝑧
the differentiation and summation are performed in different orders here. Both operations are
performed by taking a limit. But recall that limiting processes need not commute with one another.
So, the validity of term-by-term differentiation of a power series needs a proof.
∑
Example 5.2.11. The geometric series 𝑧𝑛 has radius of convergence 1, and provides a power
1
series expansion of 1−𝑧 for|𝑧| < 1. By Theorem 5.2.10,
( )
1 d 1
= = 1 + 2𝑧 + 3𝑧2 + … , when |𝑧| < 1.
(1 − 𝑧)2 d𝑧 1−𝑧

∑ ∑
Lemma 5.2.12. The power series 𝑎𝑛 𝑧𝑛 and 𝑛𝑎𝑛 𝑧𝑛−1 have the same radius of convergence.
∑ ∑
Proof. Let 𝑅 be the radius of convergence of 𝑎𝑛 𝑧𝑛 . We will show that |𝑎𝑛 𝑧𝑛 | converges for
|𝑧| < 𝑅. We choose 𝜌 so that |𝑧| < 𝜌 < 𝑅 and assume 𝑧 ≠ 0.

𝑅
𝜌
𝑧

Figure 5.4: In the proof of Lemma 5.2.12 the disc of convergence is 𝔻(𝑧, 𝑅).

Then ( )𝑛
𝑛 |𝑧|
|𝑛𝑎𝑛 𝑧𝑛−1 | = |𝑎𝑛 𝜌|𝑛 .
|𝑧| 𝜌
( )𝑛 ∑ ( |𝑧| )𝑛
|𝑧|
As 𝜌
< 1, the series 𝑛 𝜌 converges by the ratio test.

⎛ ( )𝑛+1 ⎞
⎜ (𝑛 + 1) |𝑧| ⎟
⎜ 𝜌 𝑛 + 1 |𝑧| |𝑧|
( )𝑛 = → < 1, as 𝑛 → ∞⎟
⎜ |𝑧| 𝑛 𝜌 𝜌 ⎟
⎜ 𝑛 𝜌 ⎟
⎝ ⎠

Hence, there is a constant 𝑀 such that

( )𝑛
|𝑧|
𝑛 ≤ 𝑀, for all 𝑛.
𝜌

RHS, version May 11, 2019 48

Complex Analysis I 5 COMPLEX SERIES

Thus,
| | 𝑀 | 𝑛|
|𝑛𝑎𝑛 𝑧𝑛−1 | ≤ 𝑎 𝜌 .
| | |𝑧| | 𝑛 |
The result follows from the Comparison test.
Conversely, suppose that ∑| |
|𝑛𝑎𝑛 𝑧𝑛−1 |
| |
converges. Then,
|𝑎𝑛 𝑧𝑛 | ≤ |𝑧|||𝑛𝑎𝑛 𝑧𝑛−1 || , 𝑛 ≥ 1,
| | | |
so ∑
|𝑎 𝑧𝑛 |
| 𝑛 |
converges by the Comparison test.

Proof for Theorem 5.2.10. To prove that the power series

∑
∞
𝑎𝑛 𝑧𝑛
𝑛=0

defines an analytic function

∑
∞
𝑓 (𝑧) = 𝑎𝑛 𝑧𝑛
𝑛=0
in its disc of convergence, we must show that the series
∑
∞
𝑔(𝑧) = 𝑛𝑎𝑛 𝑧𝑛−1
𝑛=1

gives the derivative of 𝑓 .

𝑟 𝑅
0

ℎ
𝑧0

( ) ( ) ( )
Figure 5.5: The disc of convergence is 𝔻 0, 𝑅 , the disc 𝔻 0, 𝑟 ⊂ 𝔻 0, 𝑅 and the chosen point
( ) ( ) ( )
𝑧0 ∈ 𝔻 0, 𝑟 with disc 𝔻 𝑧0 , ℎ ⊂ 𝔻 0, 𝑟 in the proof of Theorem 5.2.10.

For that, let 𝑅 be the radius of convergence of 𝑓 . Suppose that |𝑧0 | < 𝑟 < 𝑅. We write
𝑓 (𝑧) = 𝑆𝑁 (𝑧) + 𝜌𝑁 (𝑧)
where
∑
𝑁 ∑
∞
𝑆𝑁 (𝑧) = 𝑎𝑛 𝑧𝑛 and 𝜌𝑁 (𝑧) = 𝑎𝑛 𝑧𝑛 .
𝑛0 𝑁+1

Then, if ℎ is chosen so that ||𝑧0 + ℎ|| < 𝑟, we have

( ) ( ) ( ( ) ( ) )
𝑓 𝑧0 + ℎ − 𝑓 𝑧0 ( ) 𝑆𝑁 𝑧0 + ℎ − 𝑆𝑁 𝑧0 ′
( )
− 𝑔 𝑧0 = − 𝑆𝑁 𝑧0
ℎ ℎ
( ( ) ( ))
( ( ) ( )) 𝜌 𝑧0 + ℎ − 𝜌𝑁 𝑧0
′
+ 𝑆𝑁 𝑧0 − 𝑔 𝑧0 + .
ℎ

RHS, version May 11, 2019 49

Complex Analysis I 5 COMPLEX SERIES

Since 𝑎𝑛 − 𝑏𝑛 = (𝑎 − 𝑏)(𝑎𝑛−1 + 𝑎𝑛−2 𝑏 + … + 𝑏𝑛−1 ), we obtain that

| 𝜌 (𝑧 + ℎ) − 𝜌 (𝑧 ) | ∑∞ | (𝑧0 + ℎ)𝑛 − 𝑧𝑛 |
| 𝑁 0 𝑁 0 | | 0|
| |≤ |𝑎𝑛 | | |
| ℎ | | ℎ |
| | 𝑛=𝑁+1 | |
∑
∞ ∑
𝑛−1
≤ |𝑎𝑛 | |𝑧0 + ℎ|𝑗 |𝑧0 |𝑛−𝑗−1
𝑛=𝑁+1 𝑗=0 ⏟⏟⏟ ⏟⏞⏟⏞⏟
<𝑟 <𝑟
∑
∞
≤ |𝑎𝑛 |𝑛𝑟𝑛−1 .
𝑛=𝑁+1

The expression
∑
∞
|𝑎𝑛 |𝑛𝑟𝑛
𝑛=𝑁+1
is the tail of a convergent series, since 𝑔 converges absolutely on |𝑧| < 𝑅. Therefore, given 𝜀 > 0,
we can find 𝑁1 such that 𝑁 > 𝑁1 implies
| 𝜌 (𝑧 + ℎ) − 𝜌 (𝑧 ) |
| 𝑁 0 𝑁 0 |
| | < 𝜀.
| ℎ |
| |
Also, as
′
lim 𝑆𝑁 (𝑧0 ) = 𝑔(𝑧0 ),
𝑁→∞
we can find 𝑁2 such that 𝑁 > 𝑁2 implies
′
|𝑆𝑁 (𝑧0 ) − 𝑔(𝑧0 )| < 𝜀.

If 𝑁 > 𝑚𝑎𝑥{𝑁1 , 𝑁2 }, then we can find 𝛿 > 0 such that |ℎ| < 𝛿 implies
| 𝑆 (𝑧 + ℎ) − 𝑆 (𝑧 ) |
| 𝑁 0 𝑁 0 |
| − 𝑆 ′ 𝑁(𝑧0 )| < 𝜀,
| ℎ |
| |
because the derivative of a polynomial is obtained by differentiating it term by term. Therefore,
| 𝑓 (𝑧 + ℎ) − 𝑓 (𝑧 ) |
| 0 0 |
| − 𝑔(𝑧0 )| < 3𝜀,
| ℎ |
| |
whenever |ℎ| < 𝛿, thereby concluding the proof of the theorem.

Corollary 5.2.13. (Important!) A power series is infinitely complex differentiable in its disc of
convergence, and the higher derivatives are also power series obtained by termwise differentia-
tion.

∑
∞
Remark 5.2.14. In fact, if 𝑘(𝑧) = 𝑎𝑛 𝑧𝑛 , then 𝑓 is obtained by translating 𝑘 namely
𝑛=0

𝑓 (𝑧) = 𝑘(𝑤), where 𝑤 = 𝑧 − 𝑧0 .

As a consequence everything about 𝑘 also holds for 𝑓 after we made the appropriate translation.
By the chain rule
∑
∞
( )𝑛−1
′ ′
𝑓 (𝑧) = 𝑘 (𝑤) = 𝑛𝑎𝑛 𝑧 − 𝑧0 .
𝑛=0
A function 𝑓 defined on a open set Ω is said to have a power series expansion at a point 𝑧0 ∈ Ω,
∑ ( )𝑛
if there exists series 𝑎𝑛 𝑧 − 𝑧0 centered at 𝑧0 with a positive radius of convergence such that

∑
∞
( )𝑛
𝑓 (𝑧) = 𝑎𝑛 𝑧 − 𝑧0
𝑛=0

for all 𝑧 in a neighbourhood of 𝑧0 .

RHS, version May 11, 2019 50

Complex Analysis I 6 THE EXPONENTIAL, SINE AND COSINE FUNCTIONS

6 The exponential, sine and cosine functions

6.1 The exponential function
(1) Recall the real exponential function 𝑒𝑥 ∶ ℝ → ℝ+ = (0, ∞) has the Taylor expression

∑
∞
𝑥𝑛
𝑒𝑥 = , for all 𝑥 ∈ ℝ.
𝑛=0
𝑛!

∑ 𝑧𝑛
(2) Recall the radius of convergence of the series ∞ 𝑛=0 𝑛! is 𝑅 = ∞. Thus, this series con-
verges absolutely on the whole complex plane ℂ. Hence, by Theorem 5.2.10, the series
∑∞ 𝑧𝑛
𝑛=0 𝑛! defines an analytic function ℂ → ℂ (in fact, ℂ → ℂ ⧵ {0}.)
𝑦

0 1 𝑥

Figure 6.1: The graph of the real exponential function 𝑒𝑥 ∶ ℝ → ℝ+ , here 𝑦 = 𝑒𝑥 .

Definition 6.1.1. We define the exponential function exp ∶ ℂ → ℂ by

∑
∞ 𝑛
𝑧
exp(𝑧) ∶= .
𝑛=0
𝑛!

Remark. Since the Taylor expansion of the real exponential function is

∑
∞
𝑥𝑛
𝑒𝑥 = , for all 𝑥 ∈ ℝ,
𝑛=0
𝑛!

we have exp(𝑥) = 𝑒𝑥 if 𝑥 ∈ ℝ.

Remark. We may write 𝑒𝑧 for exp(𝑧).

6.1.2 Properties of the exponential function

(1) The function exp(𝑧) is analytic in ℂ and

d
exp(𝑧) = exp′ (𝑧) = exp(𝑧), for all 𝑧 ∈ ℂ.
d𝑧
Proof. By example 5.2.6 and Theorem 5.2.10,
( )
∑∞
d 𝑧𝑛 ∑∞
𝑛𝑧𝑛−1 ∑ 𝑧𝑛−1
∞ ∑∞ 𝑛
𝑧
exp′ (𝑧) = = = = = exp(𝑧),
𝑛=0
d𝑧 𝑛! 𝑛=1
𝑛! 𝑛=1
(𝑛 − 1)! 𝑛=0
𝑛!

for all 𝑧 ∈ ℂ.
(2) exp(𝑧 + 𝜔) = exp(𝑧) exp(𝜔), for all 𝑧, 𝜔 ∈ ℂ.

RHS, version May 11, 2019 51

Complex Analysis I 6 THE EXPONENTIAL, SINE AND COSINE FUNCTIONS

Proof. By Mertens’s theorem, the Cauchy product rule and the binomial formula give:
(∞ )( ∞ )
∑ 𝑧𝑛 ∑ 𝜔𝑘 ∑ ∞ ∑ 𝑛
1 𝑝 1
exp(𝑧) exp(𝜔) = = 𝑧 𝜔𝑛−𝑝
𝑛=0
𝑛! 𝑘=0
𝑘! 𝑛=0 𝑝=0
𝑝! (𝑛 − 𝑝)!
∑ 1 ∑ ∑ 𝑛 ( )
1 ∑ 𝑛 𝑝 𝑛−𝑝
∞ 𝑛 ∞
𝑛!
= 𝑧𝑝 𝜔𝑛−𝑝 = 𝑧 𝜔
𝑛=0
𝑛! 𝑝=0 𝑝!(𝑛 − 𝑝)! 𝑛=0
𝑛! 𝑝=0 𝑝
∑∞
1
= (𝑧 + 𝜔)𝑛 = exp(𝑧 + 𝜔).
𝑛=0
𝑛!

( )
(3) exp 0 = 1.
Proof. It follows immediately from the definition.
(4) exp(𝑧) ≠ 0, for all 𝑧 ∈ ℂ.
( )
Proof. By (1) and (2), 1 = exp 0 = exp(𝑧 − 𝑧) = exp(𝑧) exp(−𝑧).
1
(5) exp(−𝑧) = exp(𝑧)
, for all 𝑧 ∈ ℂ.
1
Proof. By (1) and (2), exp(𝑧) exp(−𝑧) = 1, so exp(−𝑧) = exp(𝑧)
.

6.1.3 The complex conjugate and the modulus of the exponential function
( )
(1) exp 𝑧 = exp(𝑧), for all 𝑧 ∈ ℂ.
Proof. By our previous results,

( ) ∑𝑛
𝑧𝑘 ∑
𝑛
𝑧𝑘 ∑
𝑛
𝑧𝑘
exp 𝑧 = lim = lim = lim = exp(𝑧).
𝑛→∞
𝑘=0
𝑘! 𝑛→∞ 𝑘=0 𝑘! 𝑛→∞ 𝑘=0 𝑘!

( ) ∑
(2) ||exp(𝑧)|| ≤ exp |𝑧| , for all 𝑧 ∈ ℂ, since the series ∞𝑛=0
𝑧𝑛
𝑛!
converges absolutely on the
whole complex plane.
(3) ||exp(𝑧)|| = exp(Re 𝑧) = 𝑒Re 𝑧 , for all 𝑧 ∈ ℂ.
Proof. By our previous results, we obtain:
( ) 6.1.2 ( )
|exp(𝑧)|2 = exp(𝑧)exp(𝑧) (1)
= exp(𝑧) exp 𝑧 = exp 𝑧 + 𝑧
| |
( ) ( )2
= exp 2 Re 𝑧 = exp(Re 𝑧 + Re 𝑧) = exp(Re 𝑧) exp(Re 𝑧) = exp(Re 𝑧) .

Since the numbers ||exp(𝑧)|| and exp(Re 𝑧) are both positive real numbers, so the claim
follows.
(4) ||exp(𝑖𝑦)|| = 1, for all 𝑦 ∈ ℝ.
2

Proof. By (3), we have ||exp(𝑧)|| = exp(Re 𝑧). But Re(𝑖𝑦) = 0, hence ||exp(𝑖𝑦)|| = 1, from
which the result follows.

RHS, version May 11, 2019 52

Complex Analysis I 6 THE EXPONENTIAL, SINE AND COSINE FUNCTIONS

Im Im
( )
𝜕𝔻 0, 1

exp(𝑖𝑦)
Re Re

𝑧-plane 𝜔-plane

Figure 6.2: The complex exponential function maps a point 𝑖𝑦, where 𝑦 ∈ ℝ, to a point on the
unit circle.

6.1.4 Euler’s formula

Previously we defined the notation

𝑒𝑖𝛼 ∶= cos 𝛼 + 𝑖 sin 𝛼, where 𝛼 ∈ ℝ.

We verify now that

exp(𝑖𝛼) = cos 𝛼 + 𝑖 sin 𝛼, where 𝛼 ∈ ℝ.
By the Taylor series of the real sine and cosine functions, we obtain

∑
∞
(𝑖𝛼)𝑘 ∑∞ 𝑘
𝑖 𝑘 ∑ (−1)𝑘 2𝑘
∞ ∑∞
(−1)𝑘 2𝑘+1
exp(𝑖𝛼) = = 𝛼 = 𝛼 +𝑖 𝛼 = cos 𝛼 + 𝑖 sin 𝛼.
𝑘=0
𝑘! 𝑘=0
𝑘! 𝑘=0
(2𝑘)! 𝑘=0
(2𝑘 + 1)!

6.2 Trigonometric functions

6.2.1 The sine and cosine functions of one complex variable
Recall that, by Euler’s formula for every real number 𝑥,

exp(𝑖𝑥) = cos 𝑥 + 𝑖 sin 𝑥 and exp(−𝑖𝑥) = cos 𝑥 − 𝑖 sin 𝑥.

Hence,

2𝑖 sin 𝑥 = exp(𝑖𝑥) − exp(−𝑖𝑥) and 2 cos 𝑥 = exp(𝑖𝑥) + exp(−𝑖𝑥),

that is,
1 ( ) 1( )
sin 𝑥 = exp(𝑖𝑥) − exp(−𝑖𝑥) and cos 𝑥 = exp(𝑖𝑥) + exp(−𝑖𝑥) .
2𝑖 2
It is, therefore, natural to define the sine and cosine functions of a complex variable as follows:

1 ( )
sin ∶ ℂ → ℂ sin 𝑧 = exp(𝑖𝑧) − exp(−𝑖𝑧) and
2𝑖 (6.2.2)
1( )
cos ∶ ℂ → ℂ cos 𝑧 = exp(𝑖𝑧) + exp(−𝑖𝑧) .
2

6.2.3. From this, we obtain Euler’s formula for all 𝑧 ∈ ℂ:

cos 𝑧 + 𝑖 sin 𝑧 = exp(𝑖𝑧).

Remark. By Euler’s formula for 𝑧 = 𝑥 + 𝑖𝑦 ∈ ℂ, where 𝑥, 𝑦 ∈ ℝ, we have

exp(𝑧) = exp(𝑥 + 𝑖𝑦) = 𝑒𝑥 (cos 𝑦 + 𝑖 sin 𝑦). (6.2.4)

RHS, version May 11, 2019 53

Complex Analysis I 6 THE EXPONENTIAL, SINE AND COSINE FUNCTIONS

Proposition 6.2.5. The functions sin ∶ ℂ → ℂ and cos ∶ ℂ → ℂ are entire and
𝜕 𝜕
sin 𝑧 = cos 𝑧 and cos 𝑧 = − sin 𝑧.
𝜕𝑧 𝜕𝑧

6.2.6 (Addition formulas).

( )
sin 𝑧1 + 𝑧2 = sin 𝑧1 cos 𝑧2 + sin 𝑧2 cos 𝑧1
( )
cos 𝑧1 + 𝑧2 = cos 𝑧1 cos 𝑧2 − sin 𝑧1 sin 𝑧2 .

Proof. Homework 6.4.

Remark 6.2.7.
∑∞
( )𝑛 𝑧2𝑛 ∑
∞
( )𝑛 𝑧2𝑛+1
cos 𝑧 = −1 ( ) and sin 𝑧 = −1 ( ) .
𝑛=0 2𝑛 ! 𝑛=0 2𝑛 + 1 !

The radius of convergence of the series

∑∞
( )𝑛 𝑧2𝑛 ∑
∞
( )𝑛 𝑧2𝑛+1
−1 ( ) and −1 ( )
𝑛=0 2𝑛 ! 𝑛=0 2𝑛 + 1 !
is 𝑅 = ∞.
We could have defined sin ∶ ℂ → ℂ and cos ∶ ℂ → ℂ as power series. They are entire, as
∑∞
( )𝑛 𝑧2𝑛 ∑
∞
( )𝑛 𝑧2𝑛+1
cos 𝑧 = −1 ( ) and sin 𝑧 = −1 ( ) .
𝑛=0 2𝑛 ! 𝑛=0 2𝑛 + 1 !

Remark 6.2.8. The connections between

𝑧 → exp (𝑧) ,
𝑧 → sin (𝑧) ,
𝑧 → cos (𝑧)
are
exp (𝑧) = cos (𝑧) + 𝑖 sin (𝑧) ,
1 ( )
sin (𝑧) = exp (𝑖𝑧) − exp (−𝑖𝑧) ,
2𝑖
1( )
cos (𝑧) = exp (𝑖𝑧) + exp (−𝑖𝑧) .
2

6.3 Complex hyperbolic functions

We define the complex hyperbolic sine and cosine of one complex variable 𝑧,
sinh ∶ ℂ → ℂ and cosh ∶ ℂ → ℂ,
as follows:
1( ) 1( )
sinh 𝑧 = exp(𝑧) − exp(−𝑧) and cosh 𝑧 = exp(𝑧) + exp(−𝑧) . (6.3.1)
2 2

They are entire functions and

𝜕 𝜕
sinh 𝑧 = cosh 𝑧, cosh 𝑧 = sinh 𝑧.
𝜕𝑧 𝜕𝑧

We obtain the Osborn’s rules:

sin 𝑖𝑧 = 𝑖 sinh 𝑧 and cos 𝑖𝑧 = cosh 𝑧. (6.3.2)

RHS, version May 11, 2019 54

Complex Analysis I 6 THE EXPONENTIAL, SINE AND COSINE FUNCTIONS

6.3.3 The real and imaginary parts

By the addition formulas and Osborn’s rules we obtain for 𝑧 = 𝑥 + 𝑖𝑦

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

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

Remark. Recall the graphs of the real hyperbolic functions.

𝑦 = cosh(𝑥)
1

0 1 𝑥

𝑦 = sinh(𝑥)

Figure 6.3: The graphs of the real hyperbolic functions cosh and sinh.

6.3.4 Unboundedness
Osborn’s rules and the known behaviour of the real cosh ∶ ℝ → [1, ∞) and sinh ∶ ℝ → ℝ imply

|cos 𝑖𝑦| = ||cosh 𝑦|| → ∞, as 𝑦 → ∞

and
|sin 𝑖𝑦| = |𝑖 sinh 𝑦| → ∞, as 𝑦 → ∞.
| | | |
Recall that the real sine and cosine functions are bounded:

|cos 𝑥| ≤ 1 and ||sin 𝑥|| ≤ 1, for all 𝑥 ∈ ℝ.

6.4 Zeros
Recall that if 𝑓 is analytic in some domain 𝐷, then 𝑓1 is analytic in 𝐷 provided 𝑓 (𝑧) ≠ 0 for all
𝑧 ∈ 𝐷. Thus we are often interested in finding out the zeroes of a particular analytic function.
In the case of complex trigonometrical functions, which are defined by the exponential function,
the fundamental equation is ( )
exp 2𝜋𝑖 = 1. (6.4.1)
This equation comes from Euler’s formula

exp(𝑖𝛼) = cos 𝛼 + 𝑖 sin 𝛼,

assuming that sin 2𝜋 = 0 and cos 2𝜋 = 1.

Proposition 6.4.2.

exp(𝑧) = 1 if and only if 𝑧 = 2𝑘𝜋𝑖, 𝑘 ∈ ℤ, and

exp(𝑧) = −1 if and only if 𝑧 = (2𝑘 + 1)𝜋𝑖, 𝑘 ∈ ℤ.

RHS, version May 11, 2019 55

Complex Analysis I 6 THE EXPONENTIAL, SINE AND COSINE FUNCTIONS

Remark 6.4.3. We have already proved that exp(𝑧) ≠ 0 for any 𝑧 ∈ ℂ. On the other hand,
Proposition 6.4.2 tells us that

exp(𝑧) + 1 and exp(𝑧) − 1

both take value zero at infinitely many points. Contrast this with the real case:

𝑒𝑥 − 1 = 0 if and only if 𝑥 = 0

and
𝑒𝑥 + 1 ≠ 0, for all 𝑥 ∈ ℝ.

The fundamental equation is ( )

exp 2𝜋𝑖 = 1.

Proposition 6.4.4.

exp(𝑧) = 1 if and only if 𝑧 = 𝑘2𝜋𝑖, 𝑘 ∈ ℝ.

The complex exponential function is periodic:

( )
exp 𝑧 + 2𝜋𝑖 = exp(𝑧).

We say that the exponential function is periodic with a pure imaginary number 2𝜋𝑖.

Remark 6.4.5. If exp(𝑧) = exp(𝑤), then 𝑧 = 𝑤 + 𝑘2𝜋𝑖 for some 𝑘 ∈ ℤ.

6.4.6. Let 𝑧 ∈ ℂ. Then

𝜋
cos 𝑧 = 0 if and only if 𝑧 = 𝑘𝜋 + , 𝑘 ∈ ℤ,
2
sin 𝑧 = 0 if and only if 𝑧 = 𝑘𝜋, 𝑘 ∈ ℤ,
( )
1
cosh 𝑧 = 0 if and only if 𝑧 = 𝑖 𝑘 + 𝜋, 𝑘 ∈ ℤ, and
2
sinh 𝑧 = 0 if and only if 𝑧 = 𝑘𝜋𝑖, 𝑘 ∈ ℤ.

Now we can define the complete tangent function

sin 𝑧 𝜋
tan 𝑧 ∶= in the set Ω = {𝑧 ∈ ℂ ∶ 𝑧 ≠ 𝑛𝜋 + , 𝑛 ∈ ℤ}.
cos 𝑧 2
The tangent function is analytic in Ω. Similarly,
cos 𝑧
cot 𝑧 ∶= is analytic in ℂ ⧵ {𝑛𝜋 ∶ 𝑛 ∈ ℤ}.
sin 𝑧

Remark 6.4.7. The exponential, sine and cosine and the hyperbolic functions are periodic:

exp(𝑧 + 𝛼) = exp(𝑧), for all 𝑧 ∈ ℂ if and only if 𝛼 = 2𝜋𝑖𝑘, 𝑘 ∈ ℤ,

sin(𝑧 + 𝛼) = sin(𝑧), for all 𝑧 ∈ ℂ if and only if 𝛼 = 2𝜋𝑘, 𝑘 ∈ ℤ,
cos(𝑧 + 𝛼) = cos(𝑧), for all 𝑧 ∈ ℂ if and only if 𝛼 = 2𝜋𝑘, 𝑘 ∈ ℤ,
sinh(𝑧 + 𝛼) = sinh(𝑧), for all 𝑧 ∈ ℂ if and only if 𝛼 = 2𝜋𝑖𝑘, 𝑘 ∈ ℤ,
cosh(𝑧 + 𝛼) = cosh(𝑧), for all 𝑧 ∈ ℂ if and only if 𝛼 = 2𝜋𝑖𝑘, 𝑘 ∈ ℤ.

RHS, version May 11, 2019 56

Complex Analysis I 7 COMPLEX MAPPINGS

7 Complex mappings
Recall that the properties of a real-valued function of a real variable are often expressed by the
graph of the function. But complex functions 𝜔 = 𝑓 (𝑧) are difficult to graph directly, since 𝜔 ∈ ℂ
and 𝑧 ∈ ℂ. Because of this difficulty, it is common to display the 𝑧-plane and the 𝜔-plane side by
side and describe how the points on the 𝑧-plane map to the points in the 𝜔-plane. When a function
𝑓 is considered in this way, it is often referred to as a mapping, a map or a transformation.
More information is usually obtained by sketching the images of curves and domains than by
simply indicating images of individual points.
𝑦

0 1 𝑥

Figure 7.1: Example: the graph of the real exponential function 𝑒𝑥 ∶ ℝ → (0, ∞), 𝑥 ↦ 𝑒𝑥 .

7.1 Some properties of the complex mappings

Recall from 6.1.3(4) that ||exp(𝑖𝑦)|| = 1. This means that the point 𝑖𝑦, where 𝑦 ∈ ℝ, maps to a
2

point on the unit circle.

Proposition 7.1.1. Let us consider the mapping 𝜔 = exp(𝑧).

(1) The lines
{𝑧 ∈ ℂ ∶ Re 𝑧 = 𝑥0 },
which are parallel to the imaginary axis in the 𝑧-plane are mapped into circles

{𝜔 ∈ ℂ ∶ |𝜔| = 𝑒𝑥0 },

with center at the origin of the 𝜔-plane.

(2) The lines
{𝑧 ∈ ℂ ∶ Im 𝑧 = 𝑦0 },
which are parallel to the real axis in the 𝑧-plane are mapped into rays

{𝜔 ∈ ℂ ∶ arg 𝜔 = 𝑦0 }

emanating from the origin in the 𝜔-plane.

Proof.
(1) Let 𝑧 = 𝑥0 + 𝑖𝑦, where 𝑥0 ∈ ℝ is fixed and 𝑦 ∈ ℝ is a variable. Then
( ) ( )
exp(𝑧) = exp 𝑥0 + 𝑖𝑦 = 𝑒𝑥0 cos 𝑦 + 𝑖 sin 𝑦 .

That is, the radius stays constant, while the argument cos 𝑦 + 𝑖 sin 𝑦 varies with 𝑦.
(2) Let 𝑧 = 𝑥 + 𝑖𝑦0 , where 𝑦0 ∈ ℝ is fixed and 𝑥 ∈ ℝ varies. Then
( ) ( )
exp 𝑥 + 𝑖𝑦0 = 𝑒𝑥 cos 𝑦0 + 𝑖 sin 𝑦0 .

That is, the argument stays constant, while the radius 𝑒𝑥 varies with 𝑥.

RHS, version May 11, 2019 57

Complex Analysis I 7 COMPLEX MAPPINGS

Im Im

𝑒𝑥0
exp(𝑧)
𝑥0 Re Re

𝑧-plane 𝜔-plane

Figure 7.2: The mapping 𝜔 = exp(𝑧) in the proposition 7.1.1(1) maps lines
{𝑧 ∈ ℂ ∶ Re 𝑧 = 𝑥0 }, which are paraller to the imaginary axis in the 𝑧-plane, into origin centered
cirles with radius 𝑒𝑥0 in the 𝜔-plane.

Im Im

𝑦0

exp(𝑧) 𝑦0
Re Re

𝑧-plane 𝜔-plane

Figure 7.3: The mapping 𝜔 = exp(𝑧) in the proposition 7.1.1(2) maps lines
{𝑧 ∈ ℂ ∶ Im 𝑧 = 𝑦0 }, which are paraller to the real axis in the 𝑧-plane, into rays
{𝜔 ∈ ℂ ∶ arg 𝜔 = 𝑦0 } emanating from the origin in the 𝜔-plane.

Proposition 7.1.2. The exponential mapping

Ω = {𝑧 ∈ ℂ ∶ 0 ≤ Im(𝑧) < 2𝜋} → ℂ ⧵ {0}, 𝑧 ↦ exp(𝑧),
is one-to-one and onto.

Proof. Note that exp(𝑧) ≠ 0, for all 𝑧 ∈ ℂ.

( ) ( )
One-to-one: Let 𝑧1 = 𝑥1 + 𝑖𝑦1 ∈ Ω and 𝑧2 = 𝑥2 + 𝑖𝑦2 ∈ Ω be such that exp 𝑧1 = exp 𝑧2 .
( ) ( )
We have to show that 𝑧1 = 𝑧2 . Since exp 𝑧1 = exp 𝑧2 , we have
| ( )| | ( )|
𝑒𝑥1 = |exp 𝑧1 | = |exp 𝑧2 | = 𝑒𝑥2 .
| | | |
Hence, 𝑥1 = 𝑥2 . Since
( ( )) ( ( ))
𝑦1 + 2𝜋𝑛 = arg exp 𝑧1 = arg exp 𝑧2 = 𝑦2 + 2𝜋𝑘, 𝑛, 𝑘 ∈ ℤ,

we have 𝑦1 − 𝑦2 = 2𝜋ℎ, where ℎ ∈ ℤ. By the definition of Ω, 0 ≤ 𝑦1 , 𝑦2 < 2𝜋, we obtain

𝑦1 = 𝑦2 . Thus 𝑧1 = 𝑧2 .
Onto: Let 𝜔 ∈ ℂ ⧵ {0} be given. We have to show that there exists 𝑧 ∈ Ω for which exp(𝑧) = 𝜔.
Choose 𝛼 ∈ [0, 2𝜋) such that 𝛼 = arg 𝜔 and write 𝑧 = ln |𝜔| + 𝑖𝛼. Then 𝑧 ∈ Ω and
( ) ( ) ( )
exp(𝑧) = exp ln |𝜔| + 𝑖𝛼 = 𝑒ln |𝜔| cos 𝛼 + 𝑖 sin 𝛼 = |𝜔| cos 𝛼 + 𝑖 sin 𝛼 = 𝜔.

RHS, version May 11, 2019 58

Complex Analysis I 7 COMPLEX MAPPINGS

Im Im
2𝜋

exp(𝑧)
Re Re

𝑧-plane 𝜔-plane

Figure 7.4: The mapping 𝜔 = exp(𝑧) in the proposition 7.1.2 maps the strip
{𝑧 ∈ ℂ ∶ 0 ≤ Im(𝑧) < 2𝜋}, into set ℂ ⧵ {0} in the 𝜔-plane.

RHS, version May 11, 2019 59

Complex Analysis I 8 COMPLEX LOGARITHMS

8 Complex logarithms
8.0.1 Argument
The identity ( )
𝑧 = |𝑧| exp(𝑖𝛼) = |𝑧| cos 𝛼 + 𝑖 sin 𝛼
is not uniquely determined. This is the fundamental cause of many-valuedness in complex func-
tion theory.
Recall that for any 𝑧 ≠ 0 we defined the argument of 𝑧 to be arg(𝑧) = 𝛼 + 2𝜋𝑘, where 𝑘 ∈ ℤ and
𝛼 is any fixed real number such that
𝑧
= cos 𝛼 + 𝑖 sin 𝛼.
|𝑧|
So, arg(𝑧) is not a single number, but all numbers of the form 𝛼 + 2𝜋𝑘, 𝑘 ∈ ℤ.

8.0.2 Complex logarithms: the inverse of the exponential function

( ) ( )
Recall that for 𝑒𝑥 ∶ ℝ → 0, ∞ there exists an inverse function, which is ln ∶ 0, ∞ → ℝ. That
is, for each positive real number 𝑥, there exists a unique real solution 𝑦 = ln 𝑥 to the equation
𝑒𝑦 = 𝑥. We work with logarithms to the base 𝑒 and we write ln 𝑥 for the real logarithm.
𝑦

𝑦 = 𝑒𝑥
1

0 1 𝑥

𝑦 = ln(𝑥)

Figure 8.1: The exponential function 𝑒𝑥 ∶ ℝ → (0, ∞) and its inverse, the natural logarithm
ln ∶ (0, ∞) → ℝ.

In the complex plane we seek solutions to the equation exp(𝑧) = 𝜔. We write log 𝑧 for the complex
logarithm. Recall, however, that the exponential of a complex variable exp(𝑧) ∶ ℂ → ℂ ⧵ {0} is
not bijective. We seek all possible values of log(𝑧).
Suppose that 𝜔 ∈ ℂ ⧵ {0}. Let us write

𝜔 = exp(𝑧) = exp(𝑢 + 𝑖𝑣), and arg 𝜔 = 𝑣 + 2𝜋𝑘, 𝑢, 𝑣 ∈ ℝ and 𝑘 ∈ ℤ.

Note that arg(𝜔) is the collection of all the numbers of the form 𝛼 + 2𝜋𝑘, 𝑘 ∈ ℤ.
For 𝑧 ≠ 0, we define the complex logarithm

log 𝑧 = ln|𝑧| + 𝑖 arg(𝑧) + 2𝜋𝑖𝑘, 𝑘 ∈ ℤ,

where arg(𝑧) is one of the values of the argument. Alternatively, we may say the multi-valued
complex logarithm is given by
log(𝑧) = ln|𝑧| + 𝑖 arg(𝑧),
where arg(𝑧) is the multi-valued argument function.

Examples.
(1) log 1 = ln||1|| + 𝑖2𝜋𝑘 = 𝑖2𝜋𝑘, 𝑘 ∈ ℤ.

RHS, version May 11, 2019 60

Complex Analysis I 8 COMPLEX LOGARITHMS

( ) (√ ) ( )
(2) log 1 + 𝑖 = ln 2 + 𝑖 𝜋4 + 2𝜋𝑘 , 𝑘 ∈ ℤ.
(√ )
Im ln 2 + 𝑖 9𝜋
Im 2𝜋𝑖 4

1+𝑖
(√ )
log(𝑧) ln 2 + 𝑖 𝜋4
1 0
Re Re

(√ )
−2𝜋𝑖 ln 2 − 𝑖 7𝜋
4

𝑧-plane 𝜔-plane
( )
Figure 8.2: Some values of the multi-valued complex logarithms log 1 and log 1 + 𝑖 .

8.0.3. In order to get a single value for 𝜔 = log(𝑧) we need to make restrictions on 𝜔; for example,
we can restrict the imaginary part of 𝜔 to the interval (−𝜋, 𝜋] or [0, 2𝜋). Such a restriction is called
a branch of the complex logarithm function.

8.0.4 Branches of the argument

The multi-valued complex logarithm is given by
log(𝑧) = ln |𝑧| + 𝑖 arg(𝑧).
So we need to find a branch for the multi-valued arg function.
The standard branch of the argument, Arg(−𝜋,𝜋] , which takes values in (−𝜋, 𝜋] is one of a branch of
arg. In fact, for every half-open interval (𝛼, 𝛼 + 2𝜋] we can create a branch of arg, Arg(𝛼,𝛼+2𝜋 ] (𝑧),
which takes values in that interval. That is, 𝜔 = arg(𝑧) such that 𝛼 < arg(𝑧) ≤ 𝛼 + 2𝜋. The
branch Arg(𝛼,𝛼+2𝜋] has a branch cut on the ray
{𝑧 ∶ 𝛼 = arg(𝑧)}.

Im Im

Re 0 Re
0

(a) Arg(−𝜋,𝜋] (𝑧) (b) Arg( 𝜋 9𝜋

] (𝑧)
,
4 4

Figure 8.3: The standard branch of the argument of complex numbers (left) and another branch
of the argument.

8.0.5 Branches of the complex logarithm

Every branch Arg(𝛼,𝛼+2𝜋] gives rise to a branch Log(𝛼,𝛼+2𝜋 ] of the logarithm
Log(𝛼,𝛼+2𝜋 ] (𝑧) = ln |𝑧| + 𝑖 Arg(𝛼,𝛼+2𝜋 ] (𝑧).

RHS, version May 11, 2019 61

Complex Analysis I 8 COMPLEX LOGARITHMS

That is, 𝜔 = log(𝑧), 𝛼 < arg(𝜔) ≤ 𝛼 + 2𝜋.

Example. (√ )
( ) 9𝜋
Log( 𝜋 ] 1 + 𝑖 = ln 2 +𝑖 .
, 5𝜋
2 2 4

8.0.6. The analyticity of a function has been defined in an open set. So we are interested in the
branch of the logarithm Log(𝛼,𝛼+2𝜋 ] when it is defined in an open set

ℂ ⧵ {𝑧 = 𝑟 exp(𝑖𝛼) ∶ 𝑟 ≥ 0},

and Log(𝛼,𝛼+2𝜋 ] maps the above domain onto the set

{𝜔 ∈ ℂ ∶ 𝛼 < Im(𝜔) < 𝛼 + 2𝜋}.

Im (𝛼 + 2𝜋)𝑖 Im

𝛼
Log(𝛼,𝛼+2𝜋 ] (𝑧) 𝛼𝑖
Re Re

𝑧-plane 𝜔-plane

Figure 8.4: The branch Log(𝛼,𝛼+2𝜋 ] of the complex logarithm maps the complex plane, minus a
ray {𝑧 = 𝑟 exp(𝑖𝛼) ∶ 𝑟 ≥ 0}, to a strip {𝜔 ∈ ℂ ∶ 𝛼 < Im(𝜔) < 𝛼 + 2𝜋}.

So we could have defined the branch of the logarithm as follows:

Let 𝛼 ∈ ℝ be given. Write

Ω𝛼 = {𝑧 = 𝑟 exp(𝑖𝛽) ∶ 𝑟 > 0, 𝛼 < 𝛽 < 𝛼 + 2𝜋}.

We define
Log(𝛼,𝛼+2𝜋 ] (𝑧) = ln 𝑟 + 𝑖𝛽,
where (𝑟, 𝛽) is a unique solution of the equation 𝑧 = 𝑟 exp(𝑖𝛽) such that 𝑟 > 0 and 𝛼 < 𝛽 < 𝛼 +2𝜋.

Remark. In the literature one might bump into the following definitions:
(1) Given 𝛼 ∈ ℝ, the half-open strip/belt {𝜔 ∈ ℂ ∶ 𝛼 < Im(𝜔) ≤ 𝛼 + 2𝜋} is called a
fundamental domain of the exponential function with respect to 𝛼, because the exponential
function is invertible on this set.
(2) The principal branch of the complex logarithm is used when 𝛼 is taken from the interval
[−2𝜋, 0] and then fixed. Depending on the book or the lectures, 𝛼 has been chosen, for
example, as 𝛼 = −𝜋 or 𝛼 = 0.

Some remarks
Remark. Note that
( )
exp log (𝑧) = 𝑧, but
( )
log exp (𝑧) = 𝑧 + 2𝑘𝜋𝑖, 𝑘 ∈ ℝ.

RHS, version May 11, 2019 62

Complex Analysis I 8 COMPLEX LOGARITHMS

Remark. All though

log (𝑧𝑤) = log (𝑧) + log (𝑤)

( )
log 𝑤𝑧 = log (𝑧) − log (𝑤) .

This is not true for the branches of the logarithm.

Example 8.0.7. ( )
0 = log(𝜋,𝜋] −1 ⋅ −1 = 2𝜋.

Remark. The log(𝛼,𝛼+2𝜋 ] branch is complex differentiable everywhere except at the

{ }
branch 𝑧 ∶ arg (𝑧) = 𝛼 and at the origin where it is undefined.

Example 8.0.8. From the definition

( )
𝑖𝜋 = exp 𝜋 log (𝑖)
( ( ))
2 1
= exp 𝜋 𝑖 + 2𝑛 , 𝑛∈ℤ
2

and
( )
𝑖𝑖 = exp 𝑖 log (𝑖)
( ( ))
1
= exp −𝜋 + 2𝑛 , 𝑛 ∈ ℤ.
2

Proposition 8.0.9. Let 𝛼 ∈ ℝ be given. Let

Ω𝛼 = {𝑧 = 𝑟 exp(𝑖𝛽) ∶ 𝑟 > 0, 𝛼 < 𝛽 < 𝛼 + 2𝜋}.

Now
(1) The function Log(𝛼,𝛼+2𝜋 ] ∶ Ω𝛼 → ℂ is analytic.

(2) Log(𝛼,𝛼+2𝜋 ] (𝑧) = 1𝑧 ,

𝜕
𝜕𝑧
for all 𝑧 ∈ Ω𝛼 .
( )
(3) Log(𝛼,𝛼+2𝜋 ] Ω𝛼 = {𝜔 ∈ ℂ ∶ 𝛼 < Im(𝜔) < 𝛼 + 2𝜋}.
( )
(4) exp Log(𝛼,𝛼+2𝜋 ] (𝑧) = 𝑧, for all 𝑧 ∈ Ω𝛼 .

(5) If 𝑧1 , 𝑧2 ∈ Ω𝛼 , then
( ) ( ) ( )
Log(𝛼,𝛼+2𝜋 ] 𝑧1 𝑧2 = Log(𝛼,𝛼+2𝜋 ] 𝑧1 + Log(𝛼,𝛼+2𝜋 ] 𝑧2 + 𝑖2𝜋𝑘, for some 𝑘 ∈ ℤ.

Proof.
(1) Homework.
( )
(2) Let us write Log(𝑧) ∶= Log(0,2𝜋 ] (𝑧). Start with exp Log(𝑧) = 𝑧, 𝑧 ≠ 0. Using the chain
rule, we get
( )𝜕
exp Log(𝑧) Log(𝑧) = 1,
𝜕𝑧
and so
𝜕 𝜕 1
𝑧 Log(𝑧) = 1, hence Log(𝑧) = .
𝜕𝑧 𝜕𝑧 𝑧
(3) Ok.
(4) Ok.
(5) Ok.

RHS, version May 11, 2019 63

Complex Analysis I 8 COMPLEX LOGARITHMS

Im ( ) ( )
Log(0,2𝜋 ] 𝑧1 + Log(0,2𝜋 ] 𝑧2

( )
Log(0,2𝜋 ] 𝑧2

( )
Log(0,2𝜋 ] 𝑧1
( )
Log(0,2𝜋 ] 𝑧1 𝑧2

𝑧1 𝑧2 0 1 Re

𝑧1 𝑧2

Figure 8.5: As shown in the example 8.0.10, the complex logarithm has the property 8.0.9(5).
Here 𝑧1 = −1 − 𝑖 and 𝑧2 = 1 − 𝑖 so the product 𝑧1 𝑧2 = (−1 − 𝑖)(1 − 𝑖) = −2. Values of the
corresponding logarithms are also shown.

Example 8.0.10 (To emphasize (5)). Let 𝑧1 = −1 − 𝑖, 𝑧2 = 1 − 𝑖, thus 𝑧1 𝑧2 = −2. Now we have
( ) ( ) ( )
Log(0,2𝜋 ] 𝑧1 𝑧2 = Log(0,2𝜋 ] −2 = ln 2 + 𝑖𝜋,
( ) √ 5𝜋
Log(0,2𝜋 ] 𝑧1 = ln 2 + 𝑖 ,
4
( ) √ 7𝜋
Log(0,2𝜋 ] 𝑧2 = ln 2 + 𝑖 .
4
Hence,
( ) ( ) √ ( ) ( )
Log(0,2𝜋 ] 𝑧1 + Log(0,2𝜋 ] 𝑧2 = 2 ln 2 + 𝑖3𝜋 = ln 2 + 𝑖𝜋 + 𝑖2𝜋 = Log(0,2𝜋 ] 𝑧1 𝑧2 + 𝑖2𝜋.

Examples 8.0.11.
(1) Solve the equation exp(𝑧) = 1 + 𝑖.
( ) √ ( )
Solution. 𝑧 = log 1 + 𝑖 = ln 2 + 𝑖 𝜋4 + 2𝜋𝑘 , 𝑘 ∈ ℤ.

(2) Find all solutions to cos(𝑧) = 2.

we rewrite this as
1( )
exp(𝑖𝑧) + exp(−𝑖𝑧) = 2,
2
which is equivalent to ( )
exp(𝑖𝑧) + exp(−𝑖𝑧) = 4.
Writing exp(𝑖𝑧) = 𝜔, where 𝜔 ≠ 0, we obtain

1
𝜔−4+ = 0,
𝜔
⇔ 𝜔2 − 4𝜔 + 1 = 0,
( )2
⇔ 𝜔 − 2 = 3,
√
⇔ 𝜔 = 2 ± 3.

RHS, version May 11, 2019 64

Complex Analysis I 8 COMPLEX LOGARITHMS

Hence,
√
exp(𝑖𝑧) = 2 ± 3,
( √ ) ( √ )
⇔ 𝑖𝑧 = log 2 ± 3 = ln 2 ± 3 + 𝑖2𝜋𝑘,
( √ ) ( √ )
⇔ 𝑧 = −𝑖 ln 2 ± 3 + 2𝜋𝑘 = 2𝜋𝑘 − 𝑖 ln 2 ± 3 , 𝑘 ∈ ℤ.

8.0.12 General power function

If 𝑧 ∈ ℂ ⧵ {0} and 𝑎 ∈ ℂ, then we define
( )
𝑧𝑎 = exp 𝑎 log 𝑧 .

Since log 𝑧 = ln |𝑧| + 𝑖 arg 𝑧 + 𝑖2𝜋𝑘, where 𝑘 ∈ ℤ, we have

( )
𝑧𝑎 = exp 𝑎 log 𝑧
( ( ))
= exp 𝑎 ln |𝑧| + 𝑖 arg 𝑧 + 𝑖2𝜋𝑘
( ( ))
= 𝑒ln |𝑧| exp 𝑖𝑎 arg 𝑧 + 2𝜋𝑘
𝑎

( ( ))
= |𝑧|𝑎 exp 𝑖𝑎 arg 𝑧 + 2𝜋𝑘 , 𝑘 ∈ ℤ.

RHS, version May 11, 2019 65

Complex Analysis I 9 INTEGRALS

9 Integrals
9.1 Complex-valued functions of real variables
If 𝑓 (𝑡) = 𝑢(𝑡) + 𝑖𝑣(𝑡) is a complex-valued function of a real variable 𝑡, where 𝑢 and 𝑣 are real-
valued, the definite integral of 𝑓 (𝑡) over an interval 𝑎 ≤ 𝑡 ≤ 𝑏 is defined as
𝑏 𝑏 𝑏
𝑓 (𝑡) d𝑡 = 𝑢(𝑡) d𝑡 + 𝑖 𝑣(𝑡) d𝑡, (9.1.1)
∫𝑎 ∫𝑎 ∫𝑎
provided that the individual integrals on the right-hand side exist. Thus
𝑏 𝑏 ( )
Re 𝑓 (𝑡) d𝑡 = Re 𝑓 (𝑡) d𝑡
∫𝑎 ∫𝑎
(9.1.2)
𝑏 𝑏 ( )
Im 𝑓 (𝑡) d𝑡 = Im 𝑓 (𝑡) d𝑡.
∫𝑎 ∫𝑎
[ ]
Recall that if 𝑢 and 𝑣 are piecewise continuous on the interval 𝑎, 𝑏 , then the corresponding
integrals exist.
The fundamental theorem of calculus can be extended to apply to the integrals of 9.1.2. Suppose
that the functions
𝑓 (𝑡) = 𝑢(𝑡) + 𝑖𝑣(𝑡) and 𝐹 (𝑡) = 𝑈 (𝑡) + 𝑖𝑉 (𝑡)
] [ [ ]
are continuous on 𝑎, 𝑏 . If 𝑓 is differentiable, and 𝐹 ′ (𝑡) = 𝑓 (𝑡) when 𝑡 ∈ 𝑎, 𝑏 , then
𝑈 ′ (𝑡) = 𝑢(𝑡) and 𝑉 ′ (𝑡) = 𝑣(𝑡).
Hence
𝑏
𝑓 (𝑡) d𝑡 = 𝐹 (𝑏) − 𝐹 (𝑎).
∫𝑎

Example 9.1.3.
𝜋
𝜋 /
4
( ) ( )
4 1 1 𝜋 𝜋 1 1
exp(𝑖𝑡) d𝑡 = exp(𝑖𝑡) = cos + 𝑖 sin − 1 = √ +𝑖 1− √ .
∫0 𝑖 𝑖 4 4 2 2
0

ℂ
𝑓
𝑎 𝑏
Figure 9.1: Function 𝑓 ∶ [𝑎, 𝑏] ⊂ ℝ → ℂ.

9.2 Curves and contours

[ ]
Let 𝛼, 𝛽 , −∞ < 𝛼 ≤ 𝛽 < ∞, be a closed and bounded interval in ℝ. A parametrized curve 𝛾
[ ] [ ]
with parameter interval 𝛼, 𝛽 is a continuous function 𝛾 ∶ 𝛼, 𝛽 → ℂ. The point 𝛾(𝛼) is the so
called initial point of 𝛾 and 𝛾(𝛽) is the final point of 𝛾. If 𝛾(𝛼) = 𝛾(𝛽), then 𝛾 is said[ to be
] closed.
The parametrized curve 𝛾 is smooth, if it is differentiable and 𝛾 ′ (𝑡) ≠ 0 for all 𝑡 ∈ 𝛼, 𝛽 .

𝛾
𝑎 𝑏
0 1

Figure 9.2: A parametrized curve 𝛾 ∶ [0, 1] ⊂ ℝ → ℂ, 𝛾(𝑡) = 𝑎 + 𝑡(𝑏 − 𝑎).

RHS, version May 11, 2019 66

Complex Analysis I 9 INTEGRALS

𝛾(𝑡) = exp(𝑖𝑡) ( ) ( )
𝛾 0 = 𝛾 2𝜋
0 2𝜋

Figure 9.3: A closed parametrized curve 𝛾 ∶ [0, 2𝜋] ⊂ ℝ → ℂ, 𝛾(𝑡) = exp(𝑖𝑡).

Remark. At the points 𝑡 = 𝛼 and 𝑡 = 𝛽, the quantities 𝛾 ′ (𝛼) and 𝛾 ′ (𝛽) are the one-sided limits
𝛾(𝛼 + ℎ) − 𝛾(𝛼) 𝛾(𝛽 − ℎ) − 𝛾(𝛽)
𝛾 ′ (𝛼) = lim+ and 𝛾 ′ (𝛽) = lim− , ℎ > 0.
ℎ→0 ℎ ℎ→0 ℎ
These quantities are called the right-hand derivative of 𝛾(𝑡) at 𝛼 and the left-hand derivative of
𝛾(𝑡) at 𝛽, respectively.
([ ]) [ ]
The image 𝛾 𝛼, 𝛽 ∶= {𝛾(𝑡) ∶ 𝑡 ∈ 𝛼, 𝛽 } is denoted by |𝛾|. The curve is said to lie in a set
[ ]
𝐴 ∈ ℂ, if |𝛾| ⊂ 𝐴. A parametrized curve is piecewise smooth, if 𝛾 is continuous on 𝛼, 𝛽 and if
there exist points
𝛼 = 𝛼0 < 𝛼1 < ⋯ < 𝛼𝑛 = 𝛽,
[ ]
where 𝛾(𝑡) is smooth on the intervals 𝛼𝑘 , 𝛼𝑘+1 . Here, the right-hand derivative at 𝛼𝑘 may differ
from the left-hand derivative at 𝛼𝑘 , 𝑘 = 1, 2, … , 𝑛 − 1.

Examples 9.2.1.
(1) The piecewise smooth curve in figure 9.4.
(2) Let

( ) [ ]
𝛾1 (𝑡) = exp 2𝜋𝑖𝑡 , 𝑡 ∈ 0, 1
and [ ]
𝛾2 (𝑡) = exp(𝑖𝑡), 𝑡 ∈ 0, 2𝜋 .
These two curves are two different parametrizations of the same curve.
Im

1+𝑖
𝑖 2+𝑖
|𝛾|

0 1 Re

Figure 9.4: A piecewise smooth curve 𝛾.

Definition 9.2.2. Two parametrizations

[ ] [ ]
𝛾 ∶ 𝑎, 𝑏 → ℂ and 𝜇 ∶ 𝑐, 𝑑 → ℂ
[ ] [ ]
are equivalent, if there exists a continuously differentiable bijection 𝑠 ↦ 𝑡(𝑠) from 𝑐, 𝑑 → 𝑎, 𝑏 ,
so that 𝑡′ (𝑠) > 0 and ( )
𝜇(𝑠) = 𝛾 𝑡(𝑠) = (𝛾 ◦ 𝑡) (𝑠).
We say that 𝜇 is a reparametrization of 𝛾.

RHS, version May 11, 2019 67

Complex Analysis I 9 INTEGRALS

𝑡 𝛾

𝑐 𝑑 𝑎 𝑏

𝛾◦𝑡
[ ] [ ]
Figure 9.5: Two parametrizations 𝛾 ∶ 𝑎, 𝑏 → ℂ and 𝜇 ∶ 𝑐, 𝑑 → ℂ are equivalent if there
[ ] [ ]
exists a continuously differentiable bijection 𝑡 ∶ 𝑐, 𝑑 → 𝑎, 𝑏 so that 𝑡′ (𝑠) > 0 and 𝜇(𝑠) =
( )
𝛾 𝑡(𝑠) = (𝛾 ◦ 𝑡) (𝑠).

9.2.3. The curve 𝛾 carries a built-in orientation determined by the direction in which 𝛾(𝑡) traces
out the image |𝛾| as 𝑡 increases from 𝛼 to 𝛽. The positive orientation (counterclockwise) is the
one that is given by the standard parametrization
[ ]
𝛾(𝑡) = 𝑧0 + 𝑟 exp(𝑖𝑡), 𝑡 ∈ 0, 2𝜋 .

The negative orientation (clockwise) is given by

[ ]
𝛾(𝑡) = 𝑧0 + 𝑟 exp(−𝑖𝑡), 𝑡 ∈ 0, 2𝜋 .

(a) counterclockwise (b) clockwise

Figure 9.6: The positive or counterclockwise orientation and the negative or clockwise orientation
of a curve.

[ ]
Example 9.2.4. Let 𝑓 ∶ 0, 2 → ℂ, 𝑓 (𝑡) = 2𝑡 + 𝑖𝑡3 . Now

/2 /2
2 2 2
1 2 1 4 𝑖
𝑓 (𝑡) d𝑡 = 2𝑡 d𝑡 + 𝑖 𝑡3 d𝑡 = 2 𝑡 +𝑖 𝑡 = 4 + ⋅ 36 = 4 + 𝑖4.
∫0 ∫0 ∫0 2 4 4
0 0

[ ]
Example 9.2.5. 1. Let 𝑓 (𝑧) = 𝑧, 𝛾(𝑡) = 1 + 𝑖 − 𝑡, when 𝑡 ∈ 0, 1 . Now
1( )( )
𝑧 d𝑧 = 1+𝑖−𝑡 −1 d𝑡
∫𝛾 ∫0
1( ) 1( )
= 1 + 𝑖 − 𝑡 d𝑡 + 𝑖 −1 d𝑡
∫0 ∫0
/1 /1
( )
1 2 1
= 2
𝑡 − 𝑡 +𝑖 (−𝑡) = 2
− 1 − 𝑖 = − 21 − 𝑖.
0 0

(√ )2
2 [ ] 𝑡4
2. Let 𝑓 (𝑧) = |𝑧|2 , 𝛾(𝑡) = 𝑡 + 𝑖 𝑡2 , when 𝑡 ∈ 0, 1 . Notice that |𝛾(𝑡)|2 = 𝑡2 + 4
and

RHS, version May 11, 2019 68

Complex Analysis I 9 INTEGRALS

𝛾 ′ (𝑡) = 1 + 𝑖𝑡. Now

(√ )2
2
1
𝑡4 ( )
|𝑧| d𝑧 = 𝑡2 + 1 + 𝑖𝑡 d𝑡
∫𝛾 ∫0 4

1( ) 1( )
𝑡4 𝑡5
= 𝑡2 + d𝑡 + 𝑖 𝑡3 + d𝑡
∫0 4 ∫0 4

/1 /1
( ) ( )
1 3 1 5 1 4 1 6
= 3
𝑡 + 20
𝑡 +𝑖 4
𝑡 + 24
𝑡
0 0
( )
1 1 1 1 23 7
= + +𝑖 + = +𝑖 .
3 20 4 24 60 24

1 ( )
Example 9.2.6. 1. Let 𝑓 ∶ ℂ → ℂ, 𝑓 (𝑧) = cosh 𝑧 = 2
exp(𝑧) + exp(−𝑧) . Then

1( )
𝐹 ∶ ℂ → ℂ, 𝐹 (𝑧) = exp(𝑧) − exp(−𝑧) = sinh 𝑧.
2

𝑞
2. Let 𝑓 ∶ ℂ → ℂ, 𝑓 (𝑧) = . Then
sin2 𝑧

cos 𝑧
𝐹 ∶ ℂ ⧵ {𝑛𝜋 ∶ 𝑛 ∈ ℤ} → ℂ), 𝐹 (𝑧) = + 𝐶, 𝐶 ∈ ℤ.
sin 𝑧

9.2.7. Given 𝛾, there exists a curve −𝛾 (or 𝛾 − ) with the same image set |𝛾|, but the opposite
orientation: [ ]
−𝛾(𝑡) = 𝛾(𝛼 + 𝛽 − 𝑡), 𝑡 ∈ 𝛼, 𝛽 .
The curve −𝛾 is called the negative of 𝛾.

Example. The negative of [ ]

𝛾(𝑡) = exp(𝑖𝑡), 𝑡 ∈ 0, 2𝜋
is ( ( )) [ ]
−𝛾(𝑡) = exp(−𝑖𝑡) = exp 𝑖 2𝜋 − 𝑡 , 𝑡 ∈ 0, 2𝜋 .

9.2.8. In the literature, one might bump into the term “an arc”. A set of points 𝑧 = (𝑥, 𝑦) in the
complex plane is said to be an arc if 𝑥 = 𝑥(𝑡) and 𝑦 = 𝑦(𝑡), where 𝛼 ≤ 𝑡(≤ 𝛽 and 𝑥(𝑡)
) and 𝑦(𝑡) are
continuous functions of the real variable 𝑡. In this case we have 𝛾(𝑡) = 𝑥(𝑡), 𝑦(𝑡) .
[ ]
9.2.9. Let 0 ≤ 𝛼 ≤ 𝛼1 ≤ 𝛽1 ≤ 𝛽 and let 𝛾 ∶ 𝛼, 𝛽 → ℂ be a curve. By restricting the curve 𝛾
[ ]
to 𝛼1 , 𝛽1 , we obtain a new parametrized curve, namely 𝛾|[𝛼1 ,𝛽1 ] . Suppose that 𝛼 < 𝑠 < 𝛽 and
𝛾1 = 𝛾|[𝛼,𝑠] and 𝛾2 = 𝛾|[𝑠,𝛽 ] . The final point of 𝛾1 coincides with the initial point of 𝛾2 and |𝛾| is
traced by first tracing ||𝛾1 || and then tracing ||𝛾2 ||.
[ ] [ ] ( ) ( )
Let 𝛾1 ∶ 𝛼1 , 𝛽1 → ℂ and 𝛾2 ∶ 𝛼2 , 𝛽2 → ℂ be curves. As long as 𝛾1 𝛽1 = 𝛾2 𝛼2 we can
form a join 𝛾1 + 𝛾2 (or 𝛾1 ∪ 𝛾2 ):
{ [ ]
( ) 𝛾1 (𝑡), 𝑡 ∈ 𝛼1 , 𝛽1
𝛾1 + 𝛾2 (𝑡) = ( ) [ ]
𝛾2 𝑡 + 𝛼2 − 𝛽1 , 𝑡 ∈ 𝛽1 , 𝛽1 + 𝛽2 − 𝛼2 .

A path is the join of finitely many smooth parametrized curves (H. A. Priestley). We, however,
use the term contour.

RHS, version May 11, 2019 69

Complex Analysis I 9 INTEGRALS

( ) ( )
𝛾1 𝛽 1 = 𝛾2 𝛼 2

𝛾2
𝛾1

𝛼1 𝛽1 𝛼2 𝛽2

Figure 9.7: A contour (or a path) is the join of finitely many smooth parametrized curves.

Example 9.2.10. Let

𝛾1 (𝑡) = 𝑡, 0 ≤ 𝑡 ≤ 1,
𝛾2 (𝑡) = 1 + 𝑖𝑡, 0 ≤ 𝑡 ≤ 1,
𝛾3 (𝑡) = 1 + 𝑖 − 𝑡, 0 ≤ 𝑡 ≤ 1,
𝛾4 (𝑡) = 𝑖 − 𝑖𝑡, 0 ≤ 𝑡 ≤ 1,
Then 𝛾 = 𝛾1 + 𝛾2 + 𝛾3 + 𝛾4 is a contour.
Im

1+𝑖
𝑖

0 1 Re

Figure 9.8: The contour 𝛾 = 𝛾1 + 𝛾2 + 𝛾3 + 𝛾4 in the example 9.2.10.

9.2.11. A smooth or a piecewise smooth curve is said to be simple, if it is not self-intersecting,

that is 𝛾(𝑡) ≠ 𝛾(𝑠) unless 𝑠 = 𝑡. If the curve is closed to begin with, then it is simple if 𝛾(𝑡) ≠ 𝛾(𝑠),
for all 𝑠, 𝑡 ∈ (𝛼, 𝛽) whenever 𝑡 ≠ 𝑠.

9.3 Integration along curves/contours

[ ]
Let 𝑓 ∶ 𝐴 → ℂ be continuous and 𝐴 ⊂ ℂ. The integral along a smooth curve 𝛾 ∶ 𝛼, 𝛽 → 𝐴 is
defined as
𝛽 ( )
𝑓 (𝑧) d𝑧 = 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡) d𝑡. (9.3.1)
∫𝛾 ∫𝛼
[ ]
Example 9.3.2 (Important example). Let 𝛾(𝑡) = 𝜔 + 𝑟 exp(𝑖𝑡), 𝑡 ∈ 0, 2𝜋 and 𝜔 ∈ ℂ. What are
the solutions to the integrals
d𝑧
and (𝑧 − 𝜔)𝑛 d𝑧, 𝑛 ∈ ℤ ⧵ {−1} ?
∫𝛾 𝑧 − 𝜔 ∫𝛾
We begin by calculating the derivative of 𝛾:
𝛾 ′ (𝑡) = 𝑖𝑟 exp(𝑖𝑡).

RHS, version May 11, 2019 70

Complex Analysis I 9 INTEGRALS

Now, for the first integral we have

2𝜋 2𝜋
d𝑧 𝑖𝑟 exp(𝑖𝑡)
= d𝑡 = 𝑖 d𝑡 = 2𝜋𝑖,
∫𝛾 𝑧 − 𝜔 ∫0 𝜔 + 𝑟 exp(𝑖𝑡) − 𝜔 ∫0

and for the second integral we get

2𝜋 ( )𝑛
(𝑧 − 𝜔)𝑛 d𝑧 = 𝜔 + 𝑟 exp(𝑖𝑡) − 𝜔 𝑖𝑟 exp(𝑖𝑡) d𝑡
∫𝛾 ∫0
2𝜋 ( ( ) )
= 𝑖𝑟𝑛+1 exp 𝑖 𝑛 + 1 𝑡 d𝑡
∫0
/
2𝜋
( ( ) )
𝑖𝑟𝑛+1
= ( ) exp 𝑖 𝑛 + 1 𝑡 = 0.
𝑖 𝑛+1
0

9.3.3 Basic properties

Proposition 9.3.4: Invariance. The right-hand side integral of (9.3.1) is independent of the
parametrization chosen for 𝛾.

𝜑 𝛾

𝑐 𝑑 𝑎 𝑏

𝛾◦𝜑
[ ]
Figure 9.9: Reminder for the proof of theorem 9.3.4. Two parametrizations 𝛾 ∶ 𝑎, 𝑏 → ℂ and
[ ] [ ]
𝜈 ∶ 𝑐, 𝑑 → ℂ are equivalent if there exists a continuously differentiable bijection 𝜑 ∶ 𝑐, 𝑑 →
[ ] ( )
𝑎, 𝑏 so that 𝜑′ (𝑠) > 0 and 𝜈(𝑠) = 𝛾 𝜑(𝑠) = (𝛾 ◦ 𝜑) (𝑠).
[ ] [ ]
Proof. Let 𝜈 ∶ 𝑐, 𝑑 → ℂ be a reparametrization of 𝛾. Then, there exists a function 𝜑 ∶ 𝑐, 𝑑 →
[ ] [ ]
𝑎, 𝑏 such that 𝜑 is continuously differentiable and 𝜑′ (𝑡) > 0 for all 𝑡 ∈ 𝑐, 𝑑 . The change of
variables theorem and the chain rule imply that
𝑑 ( )
𝑓 (𝑧) d𝑧 = 𝑓 𝜈(𝑠) 𝜈 ′ (𝑠) d𝑠
∫𝜈 ∫𝑐
𝜈=𝛾◦𝜑
𝑑 ( )
= 𝑓 (𝛾◦𝜑) (𝑠) (𝛾◦𝜑)′ (𝑠) d𝑠
∫𝑐
𝑑 ( ( )) ( )
= 𝑓 𝛾 𝜑(𝑠) 𝛾 ′ 𝜑(𝑠) 𝜑′ (𝑠) d𝑠
∫𝑐
𝜑(𝑠)=𝑡
𝑏 ( )
= 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡) d𝑡 = 𝑓 (𝑧) d𝑧.
∫𝑎 ∫𝛾

Proposition 9.3.5. Integration of continuous functions along smooth curves is linear, that is, if
𝜆, 𝜇 ∈ ℂ and 𝑓 , 𝑔 are continuous functions on a smooth curve 𝛾, then
[ ]
(𝜆𝑓 )(𝑧) + (𝜇𝑔)(𝑧) d𝑧 = 𝜆 𝑓 (𝑧) d𝑧 + 𝜇 𝑔(𝑧) d𝑧.
∫𝛾 ∫𝛾 ∫𝛾

Proof. The claim follows from the definition and the linearity of the Riemann integral. We refer
to (9.1).

RHS, version May 11, 2019 71

Complex Analysis I 9 INTEGRALS

Proposition 9.3.6: The integral of the negative. If 𝛾 − is 𝛾 with the reverse orientation, then

𝑓 (𝑧) d𝑧 = − 𝑓 (𝑧) d𝑧.

∫𝛾 − ∫𝛾

Proof. By the invariance proposition, we may choose the parameter interval for 𝛾 as [0, 1]. Then,
by the definition of the negative,
[ ] ( )
𝛾 − ∶ 0, 1 → ℂ, 𝛾 − (𝑡) = 𝛾 1 − 𝑡 .

The change of variable formula with 𝑠 = 1 − 𝑡 yields

1 ( )
𝑓 (𝑧) d𝑧 = 𝑓 𝛾 − (𝑡) (𝛾 − )′ (𝑡) d𝑡
∫𝛾 − ∫0
1 ( ( )) ( )( )
= 𝑓 𝛾 1 − 𝑡 𝛾 ′ 1 − 𝑡 −1 d𝑡
∫0
0 ( ) ( )( )
= 𝑓 𝛾(𝑠) 𝛾 ′ (𝑠) −1 −1 d𝑠
∫1
1 ( )
=− 𝑓 𝛾(𝑠) 𝛾 ′ (𝑠) d𝑠 = − 𝑓 (𝑧) d𝑧.
∫0 ∫𝛾

[ ]
Proposition 9.3.7: Joining. Let 𝛾 ∶ 𝛼, 𝛽 → ℂ be a smooth curve. Let 𝛼 < 𝑠 < 𝛽 and 𝛾1 =
𝛾|[𝛼,𝑠] and 𝛾2 = 𝛾|[𝑠,𝛽 ] . If 𝑓 is a continuous function on 𝛾, then

𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧 + 𝑓 (𝑧) d𝑧.

∫𝛾 ∫𝛾1 ∫𝛾2

Definition 9.3.8. If 𝛾 is a piecewise smooth curve, then the integral of 𝑓 over 𝛾 is the sum of the
integrals of 𝑓 over smooth parts of 𝛾; so if 𝛾(𝑡) is a piecewise smooth parametrization, then

∑
𝑛−1 𝛼𝑘+1 ( )
𝑓 (𝑧) d𝑧 = 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡) d𝑡.
∫𝛾 ∫𝛼𝑘
𝑘=0

[ ]
Definition 9.3.9. The length of a smooth curve 𝛾 ∶ 𝛼, 𝛽 → ℂ is
𝛽
| ′ |
length(𝛾) = |𝛾 (𝑡)| d𝑡.
∫𝛼 | |

This definition is independent of the parametrization chosen for 𝛾 (arguing as in the invariance
proposition). If 𝛾 is only piecewise smooth, then its length is the sum of the lengths of its smooth
parts.
[ ]
Example 9.3.10. Let 𝛾0 ∶ 0, 2𝜋 → ℂ, 𝛾0 (𝑡) = exp(𝑖𝑡). The length of 𝛾0 is

( ) 2𝜋 2𝜋
length 𝛾0 = |𝑖 exp(𝑖𝑡)| d𝑡 = d𝑡 = 2𝜋.
∫0 | | ∫ 0
[ ] ( )
Let 𝛾1 ∶ 0, 𝜋 → ℂ, 𝛾1 (𝑡) = exp 4𝑖𝑡 . The length of 𝛾1 is

( ) 𝜋
| ( )| 𝜋
length 𝛾1 = |4𝑖 exp 4𝑖𝑡 | d𝑡 = 4 d𝑡 = 4𝜋.
∫0 | | ∫0

RHS, version May 11, 2019 72

Complex Analysis I 9 INTEGRALS

9.3.11 Integrals along curves with respect to the arclength

[ ]
Given a smooth curve 𝛾 in ℂ parametrized by 𝛾 ∶ 𝛼, 𝛽 → ℂ and a continuous function 𝑓 on 𝛾,
we define the integral of 𝑓 along 𝛾 with respect to the arclength by
𝛽 ( )| |
𝑓 ||d𝑧|| = 𝑓 (𝑧)||d𝑧|| = 𝑓 𝛾(𝑡) |𝛾 ′ (𝑡)| d𝑡.
∫𝛾 ∫𝛾 ∫𝛼 | |

Remark. Invariance, linearity and joining property of this integral are as the properties of the
integral (9.3.1). The difference is in determining the integral along the negative:

𝑓 (𝑧)||d𝑧|| = 𝑓 (𝑧)||d𝑧|| .
∫𝛾 − ∫𝛾

Remark. If a curve 𝛾 is parametrized with respect to its arclength, it means that

[ ]
𝛾 ∶ 0, length(𝛾) → ℂ.

Lemma 9.3.12: Estimation lemma.

| |
| |
| 𝑓 (𝑧) d𝑧| ≤ sup|𝑓 (𝑧)| length(𝛾).
|∫𝛾 | 𝑧∈𝛾
| |
Proof.
| | | 𝛽 ( ) |
| | | |
| 𝑓 (𝑧) d𝑧| = | 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡) d𝑡|
|∫𝛾 | |∫𝛼 |
| | | |
| ( ) |
𝛽
| ′ |
≤ sup |𝑓 𝛾(𝑡) | |𝛾 (𝑡)| d𝑡
𝑡∈[𝛼,𝛽]| | ∫𝛼 | |
= sup|𝑓 (𝑧)| length(𝛾).
𝑧∈𝛾

RHS, version May 11, 2019 73

Complex Analysis I 10 THEOREMS RELATED TO AN ANTIDERIVATIVE

10 Theorems related to an antiderivative

Definition 10.0.1. Let 𝑓 be a function defined on an open set Ω ⊂ ℂ. A primitive function (or
an antiderivative) for 𝑓 on Ω is a function 𝐹 that is analytic on Ω and verifies that 𝐹 ′ (𝑧) = 𝑓 (𝑧)
for all 𝑧 ∈ Ω.
𝑧4 𝑧4
Example 10.0.2. Both functions 𝐹 (𝑧) = 4
and 𝐹̃ (𝑧) = 4
+ 1010 are antiderivatives of the
function 𝑓 (𝑧) = 𝑧3 .

Theorem 10.0.3. If a continuous function 𝑓 has an antiderivative 𝐹 in Ω and 𝛾 is a smooth curve

in Ω with the initial point 𝜔1 and the end point 𝜔2 , then
( ) ( )
𝑓 (𝑧) d𝑧 = 𝐹 𝜔2 − 𝐹 𝜔1 .
∫𝛾
[ ]
Proof. Let 𝛾 ∶ 𝛼, 𝛽 → ℂ be a parametrization for 𝛾. Then 𝛾(𝛼) = 𝜔1 and 𝛾(𝛽) = 𝜔2 . We have
that ( ) ( )
(𝐹 ◦𝛾)′ (𝑡) = 𝐹 ′ 𝛾(𝑡) 𝛾 ′ (𝑡) = 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡),
and hence
𝛽 ( ) d
𝛽
𝑓 (𝑧) d𝑧 = 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡) d𝑡 =
(𝐹 ◦𝛾)(𝑡) d𝑡
∫𝛾 ∫𝛼 ∫𝛼 d𝑡
( ) ( ) ( ) ( )
= 𝐹 𝛾(𝛽) − 𝐹 𝛾(𝛼) = 𝐹 𝜔2 − 𝐹 𝜔1 ,

where we have applied the Fundamental Theorem of Calculus.

𝛾(𝛽) = 𝜔2

𝛾
𝛾(𝛼) = 𝜔1

𝛼 𝛽

Figure 10.1: A parametrized smooth curve 𝛾 in the proof of Theorem 10.0.3.

Remark 10.0.4. If 𝛾 is piecewise smooth, we obtain, using Theorem (10.0.3) and

definition (9.3.8), a telescoping sum and we have

∑
𝑛−1 𝛼𝑘+1 ( )
𝑓 (𝑧) d𝑧 = 𝑓 𝛾(𝑡) 𝛾 ′ (𝑡) d𝑡
∫𝛾 ∫𝛼𝑘
𝑘=0
( ( ( ( )))
∑
𝑛−1
( ))
= 𝐹 𝛾 𝛼𝑘+1 − 𝐹 𝛾 𝛼𝑘
𝑘=0
( ( )) ( ( )) ( ) ( )
= 𝐹 𝛾 𝛼𝑛 − 𝐹 𝛾 𝛼0 = 𝐹 𝛾(𝛽) − 𝐹 𝛾(𝛼) .

[ ] 2
Example 10.0.5. Let 𝛾 ∶ 0, 𝜋 → ℂ, 𝛾(𝑡) = 𝑡 + 𝑖 𝑡𝜋 , and 𝑓 ∶ ℂ → ℂ ⧵ {0}, 𝑓 (𝑧) = exp(𝑧). An
antiderivative of 𝑓 is 𝐹 (𝑧) = exp(𝑧) and hence
( )
𝜋2 ( )
exp(𝑧) d𝑧 = exp 𝜋 + 𝑖 − exp 0 = 𝑒𝜋 exp(𝑖𝜋) − 1
∫𝛾 𝜋
= 𝑒𝜋 (cos(𝜋) + 𝑖 sin(𝜋)) − 1 = −𝑒𝜋 − 1.

RHS, version May 11, 2019 74

Complex Analysis I 10 THEOREMS RELATED TO AN ANTIDERIVATIVE

Corollary 10.0.6. If 𝛾 is a closed smooth curve in an open set Ω and the function 𝑓 is continuous
and has an antiderivative, then
𝑓 (𝑧) d𝑧 = 0.
∫𝛾

Remark. Since 𝛾 is closed, the initial point and the end point coincide.

1
Example 10.0.7 (Important example). The function 𝑓 (𝑧) = 𝑧
does not have an antiderivative in
the open set ℂ ⧵ {0}.
Namely, if 𝛾(𝑡) = exp(𝑖𝑡), 𝑡 ∈ [0, 2𝜋], we calculated that

𝑓 (𝑧) d𝑧 = 2𝜋𝑖 ≠ 0.
∫𝛾

Im

Re

Figure 10.2: Curve 𝛾(𝑡) = exp(𝑖𝑡), 𝑡 ∈ [0, 2𝜋] in the open set ℂ ⧵ {0} as in Example 10.0.7. This
means that the function 𝑓 (𝑧) = 1𝑧 does not have an antiderivative in this set.

Theorem 10.0.8: Fundamental theorem of Calculus II. Suppose that 𝑓 is a continuous func-
tion on a domain 𝐷 in ℂ. If
𝑓 (𝑧) d𝑧 = 0,
∫Γ
for all closed contours Γ lying entirely in 𝐷, then 𝑓 (𝑧) has a primitive 𝐹 (𝑧) throughout 𝐷.

𝛾2
𝛾𝑧0 𝑧 ≕ 𝛾1 𝑧

𝑧0

Figure 10.3: Two piecewise smooth curves 𝛾1 and 𝛾2 joining points 𝑧0 and 𝑧 in a domain 𝐷.
As shown in the first part of the proof of the theorem 10.0.8, the integration is independent of a
chosen piecewise smooth curve.

Proof. We start by showing that integration is independent of a piecewise smooth curve.

Let 𝑧0 ∈ 𝐷 be arbitrarily chosen and then fixed. Let 𝑧 ∈ 𝐷 be any other point in 𝐷.

RHS, version May 11, 2019 75

Complex Analysis I 10 THEOREMS RELATED TO AN ANTIDERIVATIVE

Since 𝐷 is a domain, 𝐷 is open and connected, and in fact 𝐷 is polygonally connected. Hence,
there exists a polygonal curve in 𝐷 from 𝑧0 to 𝑧, that is
[ ] [ ]
𝑧0 , 𝑧1 ∪ ⋯ ∪ 𝑧𝑛−1 , 𝑧𝑛 = 𝑧 ⊂ 𝐷.
[ ]
So we have 𝛾𝑧0 𝑧 ∶ 𝛼, 𝛽 → 𝐷, which is a piecewise smooth curve whose initial point is 𝛾𝑧0 𝑧 (𝛼) =
𝑧0 and end point is 𝛾𝑧0 𝑧 (𝛽) = 𝑧. Write 𝛾𝑧0 𝑧 = 𝛾1 .
If there exists another piecewise smooth curve 𝛾2 joining 𝑧0 to 𝑧, we have

0= 𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧 + 𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧 − 𝑓 (𝑧) d𝑧.

∫𝛾1 ∗𝛾2 ∫𝛾1 ∫𝛾 − ∫𝛾1 ∫𝛾2
2

Thus ∫𝛾 𝑓 (𝑧) d𝑧 = ∫𝛾 𝑓 (𝑧) d𝑧 and therefore the integration is independent of the piecewise
1 2
smooth curve in 𝐷 under the assumptions of this theorem.

𝑟
𝛾2
𝑧1
𝛾1
𝛾[𝑧1 ,𝑧1 +ℎ]
𝑧1 + ℎ
𝑧0 ( )
𝔻 𝑧1 , 𝑟

Figure 10.4: The second part of the proof of the theorem 10.0.8. Two piecewise smooth curves
𝛾1 and 𝛾2 joining fixed points 𝑧0 and 𝑧1 in a domain 𝐷 and a smooth curve 𝛾[𝑧1 ,𝑧1 +ℎ] joining point
( )
𝑧1 to any other point ℎ ∈ 𝔻 𝑧1 , 𝑟 .

Now, we can define the function

𝐹 (𝑧) = 𝑓 (𝜔) d𝜔, for all 𝑧 ∈ 𝐷.

∫𝛾𝑧
0𝑧

We have to show that 𝐹 ′ (𝑧) = 𝑓 (𝑧) where 𝑧 ∈ 𝐷.

We fixed the starting point 𝑧0 ∈ 𝐷 in the beginning of the proof. Now we fix the end point as
well,
( and ) label it 𝑧1 ∈ 𝐷. We let 𝑧1 + ℎ be any ( point
) distinct from 𝑧1 , which lies in some disc
𝔻 𝑧1 , 𝑟 , 𝑟 > 0 that is small enough so that 𝔻 𝑧1 , 𝑟 ⊂ 𝐷.
[ ]
We have 𝛾𝑧0 𝑧1 joining 𝑧0 and 𝑧1 . Let 𝑧1 , 𝑧1 + ℎ be the line segment joining 𝑧1 and 𝑧1 + ℎ. The
parametrized curve in this case is
[ ]
𝛾[𝑧1 ,𝑧1 +ℎ] (𝑠) = 𝑧1 + ℎ𝑠, 𝑠 ∈ 0, 1 .

Because the integration is independent of the piecewise smooth curve, we have

( )
𝐹 𝑧1 + ℎ = 𝑓 (𝜔) d𝜔 = 𝑓 (𝜔) d𝜔.
∫𝛾𝑧 ∫𝛾𝑧 ∗𝛾 𝑧 ,𝑧 +ℎ
0 𝑧1 +ℎ 0 𝑧1 [1 1 ]
Then,

1
( ( ) ( )) ( ) 1⎛ ⎞ ( )
𝐹 𝑧1 + ℎ − 𝐹 𝑧1 − 𝑓 𝑧1 = ⎜ 𝑓 (𝜔) d𝜔 − 𝑓 (𝜔) d𝜔⎟ − 𝑓 𝑧1 .
ℎ ℎ ⎜∫𝛾𝑧 𝑧 ∗𝛾 ∫𝛾𝑧 𝑧1 ⎟
⎝ 0 1 [𝑧1 ,𝑧1 +ℎ] 0 ⎠

RHS, version May 11, 2019 76

Complex Analysis I 10 THEOREMS RELATED TO AN ANTIDERIVATIVE

Note that since

𝑓 (𝜔) d𝜔 = 𝑓 (𝜔) d𝜔 + 𝑓 (𝜔) d𝜔,

∫𝛾𝑧 ∗𝛾 𝑧 ,𝑧 +ℎ ∫𝛾𝑧 ∫𝛾
0 𝑧1 [1 1 ] 0 𝑧1 [𝑧1 ,𝑧1 +ℎ]
we are left with

1
( ( ) ( )) ( ) 1⎛ ⎞ ( )
𝐹 𝑧1 + ℎ − 𝐹 𝑧1 − 𝑓 𝑧1 = ⎜ 𝑓 (𝜔) d𝜔⎟ − 𝑓 𝑧1 .
ℎ ℎ ⎜∫𝛾 ⎟
⎝ [𝑧1 ,𝑧1 +ℎ] ⎠

Recall that
1
d𝜔 =
ℎ d𝑠 = ℎ,
∫𝛾 ∫0
[𝑧1 ,𝑧1 +ℎ]
( ) ( )
and hence ∫𝛾 𝑓 𝑧1 d𝜔 = 𝑓 𝑧1 ℎ. With this, we have
[𝑧1 ,𝑧1 +ℎ]

1
( ( ) ( )) ( ) 1⎛ ( ) ⎞
𝐹 𝑧1 + ℎ − 𝐹 𝑧1 − 𝑓 𝑧1 = ⎜ 𝑓 (𝜔) − 𝑓 𝑧1 d𝜔⎟ .
ℎ ℎ ⎜∫𝛾 ⎟
⎝ [𝑧1 ,𝑧1 +ℎ] ⎠

By the estimation lemma 9.3.12, we have

| ( ) ( ) | | |
| 𝐹 𝑧1 + ℎ − 𝐹 𝑧1 ( )| | 1 ||| ( ) ||
| | | |
− 𝑓 𝑧1 | = | || (𝑓 (𝜔) − 𝑓 𝑧1 d𝜔|
|
| ℎ | | ℎ |||∫𝛾[𝑧 ,𝑧 +ℎ] |
|
| | | 1 1 |
( ) ( )
1 | |
≤ sup |𝑓 (𝜔) − 𝑓 𝑧1 | length 𝛾[𝑧1 ,𝑧1 +ℎ] .
|ℎ| 𝜔∈[𝑧1 ,𝑧1 +ℎ]| |
⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏟
=|ℎ|

| ( )|
Since 𝑓 is continuous, we have |𝑓 (𝜔) − 𝑓 𝑧1 | → 0 as 𝜔 → 𝑧1 , that is ℎ → 0.
| |
( ) ( )
This means that 𝐹 ′ 𝑧1 = 𝑓 𝑧1 .

So far, we have proved the following

Theorem 10.0.9. Suppose that 𝑓 is a continuous function on a domain 𝐷 in ℂ.

Then, 𝑓 has a primitive on 𝐷 if and only if

𝑓 (𝑧) d𝑧 = 0
∫Γ

for all closed contours Γ in 𝐷.

[ ]
Examples 10.0.10. 1. Let 𝛾 ∶ 0, 2𝜋 → ℂ, 𝛾(𝑡) = exp(𝑖𝑡). Then
2𝜋 2𝜋
𝑧 d𝑧 = exp(𝑖𝑡)𝑖 exp(𝑖𝑡) d𝑡 = 𝑖 |exp(𝑖𝑡)|2 d𝑡 = 2𝜋𝑖 ≠ 0.
∫𝛾 ∫0 ∫0 | |
⏟⏞⏟⏞⏟
=1

Hence, the function 𝑧 ↦ 𝑧 does not have a primitive on any domain which contains the
unit circle.
2. Since the function 𝑓 (𝑧) = 𝑧2 has a primitive, namely 𝐹 (𝑧) = 13 𝑧3 , we know that

𝑧2 d𝑧 = 0
∫Γ

for every closed contour in ℂ.

RHS, version May 11, 2019 77

Complex Analysis I 10 THEOREMS RELATED TO AN ANTIDERIVATIVE

Remarks.
• The notation ∮ is often used instead of ∫ when integrating over a closed contour.
• By convention, if an orientation of a closed contour is not specified, it is assumed to be
counterclockwise.
• It is not always easy to check whether a function has a primitive or not.
( )
Example. Let 𝑓 (𝑧) = exp 𝑧2 .

RHS, version May 11, 2019 78

Complex Analysis I 11 INTEGRAL THEOREMS

11 Integral theorems
Theorem 11.0.1: Goursat’s theorem. Let( 𝑅 be) a closed rectangle whose sides are parallel to
the coordinate axes. Suppose that 𝑅 ⊂ 𝔻 0, 𝑟 = {𝑧 ∈ ℂ ∶ |𝑧| < 𝑟} and that 𝑓 is analytic in
( )
𝔻 0, 𝑟 . Then
𝑓 (𝑧) d𝑧 = 0.
∫𝜕𝑅

𝜕𝑅

( )
𝔻 0, 𝑟
( )
Figure 11.1: The Goursat’s theorem 11.0.1 states that given a closed rectangle 𝑅 ⊂ 𝔻 0, 𝑟 ⊂ ℂ,
( )
with sides parallel to coordinate axes, and an analytic function 𝑓 in 𝔻 0, 𝑟 then ∫𝜕𝑅 𝑓 (𝑧) d𝑧 = 0.

Remark. This is one of the key results in this Complex Analysis I course.

Proof. The proof is based on the method of bisection.

Figure 11.2: Bisections of the rectangle 𝑅 and bisections of the bisected parts of the 𝑅.

Let us write  ∶= ∫𝜕𝑅 𝑓 (𝑧) d𝑧. Let us divide 𝑅 into four congruent rectangles 𝑅𝑘 , 𝑘 = 1, 2, 3, 4.
Suppose that the boundary of 𝑅 and the boundaries of 𝑅𝑘 are all oriented counterclockwise.
Hence
∑
4
𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧.
∫𝜕𝑅 ∫𝜕𝑅𝑘
𝑘=1

RHS, version May 11, 2019 79

Complex Analysis I 11 INTEGRAL THEOREMS

By the triangle inequality for some 𝑅𝑘 , we have

| |
| |
|| ≤ 4| 𝑓 (𝑧) d𝑧| .
|∫𝜕𝑅 |
| 𝑘 |
This 𝑅𝑘 is denoted by 𝑅1 and we set

1 = 𝑓 (𝑧) d𝑧.
∫𝜕𝑅1

𝑅2 𝑅1

𝑅3 𝑅4

Figure 11.3: The rectangle 𝑅 divided into four congruent rectangles 𝑅𝑘 , 𝑘 = 1, 2, 3, 4.

Now we divide 𝑅1 into four congruent rectangles 𝑅1𝑘 . Again, for some 𝑘 = 1, 2, 3, 4, we denote
𝑅1𝑘 = 𝑅2 and get
| |
|1 | ≤ 4|| |
𝑓 (𝑧) d𝑧| .
| | |∫𝜕𝑅2 |
| |
This process can be repeated indefinitely and we obtain a sequence of nested rectangles

𝑅 ⊃ 𝑅1 ⊃ 𝑅2 ⊃ ⋯ ⊃ 𝑅𝑛 ⊃ ⋯ .

If we write 𝑗 = ∫𝜕𝑅𝑗 𝑓 (𝑧) d𝑧, we have the following property for these rectangles

| | | |
|𝑗 | ≤ 4|𝑗+1 | .
| | | |
| |
Hence|| ≤ 4𝑗 |𝑗 |.
| |
If the sidelengths
⋂∞ of𝑗 𝑅 are 𝐿1 and 𝐿2 , then the sidelengths of 𝑅 are 2 𝐿1 and 2 𝐿2 . We show
𝑗 −𝑗 −𝑗

that the set 𝑗=1 𝑅 contains exactly one point.

√ ⋂
Since diam(𝑅𝑗 ) = 2−𝑗 𝐿21 + 𝐿22 → 0 as 𝑗 → ∞, the set ∞ 𝑗=1 𝑅 contains at most one point.
𝑗

Let 𝑧𝑗 ∈ 𝑅𝑗 , 𝑗 = 1, 2, … be arbitrarily chosen points. Since 𝑧𝑘 ∈ 𝑅𝑗 whenever 𝑘 ≥ 𝑗, the

sequence (𝑧𝑘 ) is a Cauchy sequence and there exists

𝑧∗ = lim 𝑧𝑘 .
𝑘→∞
⋂∞
Since each 𝑅𝑘 is closed and 𝑧∗ is in each 𝑅𝑘 , we have 𝑧∗ ∈ 𝑘=1 𝑅
𝑘.

RHS, version May 11, 2019 80

Complex Analysis I 11 INTEGRAL THEOREMS

𝑅𝑗

Figure 11.4: The integral 𝑗 = ∫𝜕𝑅𝑗 𝑓 (𝑧) d𝑧 is the integration of the function 𝑓 over the boundary
of the rectangle 𝑅𝑗 counterclockwise.

Let 𝜀 > 0 be arbitrarily

( )chosen and then fixed. We can choose 𝛿 > 0 to be small enough so that
𝑓 is analytic in 𝔻 𝑧∗ , 𝛿 . Then

| ( ) ( )( )|
|𝑓 (𝑧) − 𝑓 𝑧∗ − 𝑓 ′ 𝑧∗ 𝑧 − 𝑧∗ | < 𝜀||𝑧 − 𝑧∗ || (11.0.2)
| |

whenever ||𝑧 − 𝑧∗ || < 𝛿.

( )
We may choose 𝑗 to be large enough so that 𝑅𝑗 ⊂ 𝔻 𝑧∗ , 𝛿 . So 𝑗 is fixed now.
Since the constant function 1 has a primitive, i.e. 𝑧, and the function 𝑔(𝑧) = 𝑧 has a primitive,
i.e. 21 𝑧2 , we know that
d𝑧 = 0
∫𝜕𝑅𝑗
and
𝑧 d𝑧 = 0.
∫𝜕𝑅𝑗
Hence,
| | | ( ( ) ( )( )) ||
| | | | |
|𝑗 | = | 𝑓 (𝑧) d𝑧| = | 𝑓 (𝑧) − 𝑓 𝑧∗ − 𝑓 ′ 𝑧∗ 𝑧 − 𝑧∗ d𝑧| .
| | |∫𝜕𝑅𝑗 | |∫𝜕𝑅𝑗 |
| | | |
By the estimation lemma 9.3.12 and 11.0.2 we have
| | ( )
|𝑗 | ≤ 𝜀||𝑧 − 𝑧∗ || length 𝜕𝑅𝑗 .
| |
Here,
( ) ( ) ( ) √
length 𝜕𝑅𝑗 = 2 2−𝑗 𝐿1 + 2−𝑗 𝐿2 = 2−𝑗+1 𝐿1 + 𝐿2 and ||𝑧 − 𝑧∗ || ≤ 2−𝑗 𝐿21 + 𝐿22 ,

since 𝑧, 𝑧∗ ∈ 𝑅𝑗 . Hence,
( √ )(
| | ( )) | | ( )3
|𝑗 | ≤ 𝜀 2−𝑗 𝐿21 + 𝐿22 2−𝑗+1 𝐿1 + 𝐿2 and || ≤ 4𝑗 |𝑗 | ≤ 2𝜀 𝐿1 + 𝐿2 2 .
| | | |
| |
Since 𝜀 > 0 was arbitrarily chosen, we have shown that|| = |∫𝜕𝑅 𝑓 (𝑧) d𝑧| = 0.
| |
As the following, very important theorem shows, the assumption in Goursat’s theorem can be
weakened.
( )
Theorem 11.0.3. Let 𝜔0 ∈ ℂ. Let 𝑟 > 0 be given. If 𝑓 is analytic in 𝔻 0, 𝑟 ⧵ {𝜔0 } and 𝑅 is a
( ) ( )
closed rectangle in 𝔻 0, 𝑟 such that 𝜔0 ∈ 𝑅 and 𝑓 is continuous in 𝔻 0, 𝑟 , then

𝑓 = 0.
∫𝜕𝑅

RHS, version May 11, 2019 81

Complex Analysis I 11 INTEGRAL THEOREMS

𝜕𝑅

𝑅0
𝜔0

( )
𝔻 0, 𝑟

Figure 11.5: For the proof of the theorem 11.0.3 we divide the rectangle 𝑅 into nine parts.

Proof. Suppose that 𝜔0 ∈ int𝑅. We divide 𝑅 into nine rectangles as shown in figure 11.5 and
apply Goursat’s theorem to all but the rectangle 𝑅0 . If the corresponding equations

𝑓 = 0, 𝑗 = 1, 2, … , 8
∫𝜕𝑅

are added, we obtain, after cancellations,

𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧.

∫𝜕𝑅 ∫𝜕𝑅0
( )
Suppose that length 𝜕𝑅0 = 𝜀 > 0. By the estimation lemma 9.3.12 we have

| | | |
| | | |
| 𝑓| =| 𝑓 | ≤ 𝑀𝜀, for some 𝑀 ∈ ℝ.
|∫𝜕𝑅 | |∫𝜕𝑅0 |
| | | |
Since 𝜀 > 0 was arbitrarily chosen, the claim of the theorem follows.
In the case that 𝜔0 ∈ 𝜕𝑅, a similar procedure gives the claim.
( )
Remark. Later on we will find that, in fact, the assumptions that 𝑓 is analytic in 𝔻 0, 𝑟 ⧵ {𝜔0 }
( ) ( )
and continuous on 𝔻 0, 𝑟 imply that 𝑓 is analytic in 𝔻 0, 𝑟 .

First we will prove the existence of antiderivatives in a disc as a consequence of Goursat’s theo-
rem.

Theorem 11.0.4. An analytic function in an open disc has an antiderivative in that disc.

Proof. Fix 𝑧0 ∈ ℂ. Our goal is to show that there exists an analytic function 𝐹 such that
( )
𝐹 ′ (𝑧) = 𝑓 (𝑧), for all 𝑧 ∈ 𝔻 𝑧0 , 𝑅 ,

i.e. an antiderivative of 𝑓 .
( )
Denote 𝑧0 = 𝑥0 + 𝑖𝑦0 and, for each 𝑧 = 𝑥 + 𝑦𝑖 ∈ 𝔻 𝑧0 , 𝑅 , let

𝛾𝑧 = 𝛾[𝑧0 ,𝑥+𝑖𝑦0 ] ∗ 𝛾[𝑥+𝑖𝑦0 ,𝑧] ,

i.e. the join of the line segments connecting the points 𝑧0 , 𝑥 + 𝑖𝑦0 and 𝑧 (as demonstrated
( ) in
Figure 11.6). Note that the point 𝑥 + 𝑖𝑦0 and the contour 𝛾𝑧 are included in 𝔻 𝑧0 , 𝑅 for any

RHS, version May 11, 2019 82

Complex Analysis I 11 INTEGRAL THEOREMS

( )
𝑧 ∈ 𝔻 𝑧0 , 𝑅 ; this can be seen by drawing the line segment between 𝑧0 and 𝑧 and noticing that it
is the hypotenuse of a right-angled triangle whose catheti form the contour 𝛾𝑧 . The hypotenuse
( )is
always longer than either cathetus, and here its length is also less than 𝑅 because 𝑧 ∈ 𝔻 𝑧0 , 𝑅 .

𝑧 = 𝑥 + 𝑖𝑦

𝛾𝑧
𝑧0 = 𝑥0 + 𝑖𝑦0 𝑥 + 𝑖𝑦0

( )
Figure 11.6: The contour 𝛾𝑧 is a path between two points 𝑧0 and 𝑧 in the open disc 𝔻 𝑧0 , 𝑅 .

( )
Let us then define a function 𝐹 ∶ 𝔻 𝑧0 , 𝑅 → ℂ such that

𝐹 (𝑧) = 𝑓 (𝑢) d𝑢.

∫𝛾𝑧

To show that this is the primitive of 𝑓 , it is enough to show that

𝜕 𝜕
𝐹 (𝑧) = 𝑖𝑓 (𝑧) and 𝐹 (𝑧) = 𝑓 (𝑧). (∗)
𝜕𝑦 𝜕𝑥

The reason for this is the following: if 𝐹 = 𝑈 + 𝑖𝑉 and 𝑓 = 𝑢 + 𝑖𝑣, where 𝑈 , 𝑉 , 𝑢, 𝑣 ∶ ℝ2 → ℝ,

we then have
𝜕 𝜕 𝜕
𝐹 (𝑧) = 𝑈 + 𝑖 𝑉 = 𝑖𝑓 (𝑧) = 𝑖(𝑢 + 𝑖𝑣) = −𝑣 + 𝑖𝑢, and
𝜕𝑦 𝜕𝑦 𝜕𝑦
𝜕 𝜕 𝜕
𝐹 (𝑧) = 𝑈 + 𝑖 𝑉 = 𝑓 (𝑧) = 𝑢 + 𝑖𝑣.
𝜕𝑥 𝜕𝑥 𝜕𝑥
Hence,
𝜕𝑈 𝜕𝑈 𝜕𝑉 𝜕𝑉
= 𝑢, − 𝑣, =𝑣 and = 𝑢,
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦
and thus the Cauchy–Riemann equations hold for 𝐹 . Since 𝑓 is continuous, also 𝑢 and 𝑣, which
are the partial derivatives of 𝐹 , are continuous. Hence 𝑈 and 𝑉 are differentiable in the real
sense and satisfy the Cauchy–Riemann equations, so by Theorem 4.3.10, 𝐹 is differentiable and
𝐹′ = 𝑓.
We proceed to show that (∗) holds.
(I) Let 𝜀 > 0, and let ℎ ∈ ℝ ⧵ {0}. It is straightforward to check that

𝑖ℎ = d𝑢.
∫𝛾[𝑧,𝑧+𝑖ℎ]

Also notice that 𝛾𝑧+𝑖ℎ = 𝛾𝑧 ∗ 𝛾[𝑧,𝑧+𝑖ℎ] (see Figure 11.7). Then

( )
| 𝐹 (𝑧 + 𝑖ℎ) − 𝐹 (𝑧) | || 1 |
|
| | | 𝑖ℎ
| − 𝑖𝑓 (𝑧)| = | 𝑓 (𝑢) d𝑢 − 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧)||
| | ∫ ∫
| ℎ | || ℎ 𝛾𝑧+𝑖ℎ 𝛾𝑧 ℎ |
|
| ( ) |
|1 1 |
= || 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 − 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧) d𝑢||
| ℎ ∫𝛾𝑧 ∫𝛾[𝑧,𝑧+𝑖ℎ] ∫𝛾𝑧 ℎ ∫𝛾[𝑧,𝑧+𝑖ℎ] |
| |

RHS, version May 11, 2019 83

Complex Analysis I 11 INTEGRAL THEOREMS

𝑧 + 𝑖ℎ = 𝑥 + 𝑖(ℎ + 𝑦)
𝛾𝑧+𝑖ℎ
𝑧 = 𝑥 + 𝑖𝑦

𝛾𝑧
𝑧0 = 𝑥0 + 𝑖𝑦0 𝑥 + 𝑖𝑦0

( )
Figure 11.7: Extension of the path 𝛾𝑧 to the path 𝛾𝑧+𝑖ℎ in the open disc 𝔻 𝑧0 , 𝑅 . The path is
now between points 𝑧0 and 𝑧 + 𝑖ℎ.

| |
|1 1 |
= || 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧) d𝑢||
| ℎ ∫𝛾[𝑧,𝑧+𝑖ℎ] ℎ ∫𝛾[𝑧,𝑧+𝑖ℎ] |
| |
| |
| 1 || |
= || |||| (𝑓 (𝑢) − 𝑓 (𝑧)) d𝑢|| .
| ℎ ||∫𝛾[𝑧,𝑧+𝑖ℎ] |
| |
Noting that length(𝛾[𝑧,𝑧+𝑖ℎ] ) = |ℎ| and using the estimation lemma 9.3.12, we get that
| |
1 || |
|≤ 1 | |
|
|ℎ| |∫𝛾[𝑧,𝑧+𝑖ℎ]
(𝑓 (𝑢) − 𝑓 (𝑧)) d𝑢| |ℎ| sup |𝑓 (𝑢) − 𝑓 (𝑧)| ⋅ length(𝛾[𝑧,𝑧+𝑖ℎ] )
| 𝑢∈𝛾[𝑧,𝑧+𝑖ℎ]
| |
1
= sup |𝑓 (𝑢) − 𝑓 (𝑧)|||ℎ|
|ℎ| 𝑢∈𝛾[𝑧,𝑧+𝑖ℎ] |
= sup ||𝑓 (𝑢) − 𝑓 (𝑧)|| .
𝑢∈𝛾[𝑧,𝑧+𝑖ℎ]

Since 𝑓 is continuous, there exists 𝛿 > 0 such that whenever|𝑢 − 𝑧| < 𝛿, we have |𝑓 (𝑢) −
𝑓 (𝑧)| < 𝜀. Thus if we let |ℎ| < 𝛿, we get that |𝑢 − 𝑧| ≤ |ℎ| < 𝛿 for all 𝑢 ∈ ℂ such that
𝑢 − 𝑧 ∈ 𝛾[𝑧,𝑧+𝑖ℎ] and hence sup𝑢∈𝛾[𝑧,𝑧+𝑖ℎ] ||𝑓 (𝑢) − 𝑓 (𝑧)|| < 𝜀.
To conclude the proof so far, we have found a 𝛿 such that for our fixed 𝜀, we have

| 𝐹 (𝑧 + 𝑖ℎ) − 𝐹 (𝑧) | | 1 ||| |

|
| |
| − 𝑖𝑓 (𝑧)| = || |||| (𝑓 (𝑢) − 𝑓 (𝑧)) d𝑢||
| ℎ | | ℎ ||∫𝛾 |
| | | [𝑧,𝑧+𝑖ℎ] |
≤ sup ||𝑓 (𝑢) − 𝑓 (𝑧)||
𝑢∈𝛾[𝑧,𝑧+𝑖ℎ]

< 𝜀,
𝜕
whenever|ℎ| < 𝛿. Hence 𝜕𝑦
𝐹 (𝑧) = 𝑖𝑓 (𝑧).

(II) Let ℎ ∈ ℝ ⧵ {0}. We first make some observations about contours:

(1) 𝛾𝑧 = 𝛾[𝑧0 ,𝑥+𝑖𝑦0 ] ∗ 𝛾[𝑥+𝑖𝑦0 ,𝑧] by definition, and
(2) 𝛾𝑧+ℎ = 𝛾[𝑧0 ,𝑥+𝑖𝑦0 ] ∗ 𝛾[𝑥+𝑖𝑦0 ,𝑧+ℎ+𝑖𝑦0 ] ∗ 𝛾[𝑥+ℎ+𝑖𝑦0 ,𝑧+ℎ] .
Then
( )
| 𝐹 (𝑧 + ℎ) − 𝐹 (𝑧) | || 1 |
|
| | |
| − 𝑓 (𝑧)| = | 𝑓 (𝑢) d𝑢 − 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧)||
| ℎ | | ℎ ∫𝛾 ∫𝛾𝑧 |
| | | 𝑧+ℎ |
| ⎛
|1
|
=| ⎜ 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢
| ℎ ⎜∫𝛾 ∫𝛾[𝑥+𝑖𝑦 ,𝑧+ℎ+𝑖𝑦 ] ∫𝛾[𝑥+ℎ+𝑖𝑦 ,𝑧+ℎ]
| ⎝ [𝑧0 ,𝑥+𝑖𝑦0 ]
| 0 0 0

RHS, version May 11, 2019 84

Complex Analysis I 11 INTEGRAL THEOREMS

⎞ |
|
⎟ |
− 𝑓 (𝑢) d𝑢 − 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧)|
∫𝛾[𝑧 ,𝑥+𝑖𝑦 ] ∫𝛾[𝑥+𝑖𝑦 ,𝑧] ⎟ |
⎠ |
0 0 0
|
| ⎛ ⎞ |
|1 |
| ⎜ |
=| 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 − 𝑓 (𝑢) d𝑢⎟ − 𝑓 (𝑧)|
| ℎ ⎜∫𝛾 ∫ ∫ ⎟ |
| ⎝ [𝑥+𝑖𝑦0 ,𝑧+ℎ+𝑖𝑦0 ] 𝛾[𝑥+ℎ+𝑖𝑦 ,𝑧+ℎ] 𝛾[𝑥+𝑖𝑦 ,𝑧]
⎠ |
| 0 0
|
| ⎛ ⎞ |
|1 |
| ⎜ ⎟ |
=| 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧)| .
| ℎ ⎜∫𝛾 ∫ ∫ ⎟ |
| ⎝ [𝑥+𝑖𝑦0 ,𝑧+ℎ+𝑖𝑦0 ] 𝛾[𝑥+ℎ+𝑖𝑦 ,𝑧+ℎ] 𝛾[𝑧,𝑥+𝑖𝑦 ]
⎠ |
| 0 0
|
Goursat’s theorem gives us (see Figure 11.8) the identity

𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 = 0,

∫𝛾[𝑥+𝑖𝑦 ∫𝛾[𝑥+ℎ+𝑖𝑦 ,𝑧+ℎ] ∫𝛾[𝑧+ℎ,𝑧] ∫𝛾[𝑧,𝑥+𝑖𝑦 ]
0 ,𝑧+ℎ+𝑖𝑦0 ] 0 0

whence we get that

− 𝑓 (𝑢) d𝑢 = 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢.

∫𝛾[𝑧+ℎ,𝑧] ∫𝛾[𝑥+𝑖𝑦 ,𝑧+ℎ+𝑖𝑦 ] ∫𝛾[𝑥+ℎ+𝑖𝑦 ,𝑧+ℎ] ∫𝛾[𝑧,𝑥+𝑖𝑦 ]
0 0 0 0

Using the above and the fact that

d𝑢 = ℎ,
∫𝛾[𝑧,𝑧+ℎ]

we can continue:
| ⎛ ⎞ |
|1 |
| ⎜ ⎟ |
| 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 + 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧)|
| ℎ ⎜∫𝛾 ∫ ∫ ⎟ |
| ⎝ [𝑥+𝑖𝑦0 ,𝑧+ℎ+𝑖𝑦0 ] 𝛾[𝑥+ℎ+𝑖𝑦0 ,𝑧+ℎ] 𝛾[𝑧,𝑥+𝑖𝑦0 ] ⎠ |
| |
| |
| 1 ℎ|
= ||− 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧) ||
∫
| ℎ 𝛾[𝑧+ℎ,𝑧] ℎ|
| |
| |
|1 1 |
= || 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧) d𝑢||
| ℎ ∫𝛾[𝑧,𝑧+ℎ] ℎ ∫𝛾[𝑧,𝑧+ℎ] |
| |
| ( )|
|1 |
= || 𝑓 (𝑢) d𝑢 − 𝑓 (𝑧) d𝑢 ||
| ℎ ∫𝛾[𝑧,𝑧+ℎ] ∫𝛾[𝑧,𝑧+ℎ] |
| |
| |
| 1 || |
= || |||| (𝑓 (𝑢) − 𝑓 (𝑧)) d𝑢|| .
ℎ ∫
| || 𝛾[𝑧,𝑧+ℎ] |
| |

𝑧 = 𝑥 + 𝑖𝑦 𝑧+ℎ

𝛾𝑧+ℎ

𝛾𝑧
𝑧0 = 𝑥0 + 𝑖𝑦0 𝑥 + 𝑖𝑦0 𝑥 + 𝑖𝑦0 + ℎ = (𝑥 + ℎ)𝑖𝑦0

( )
Figure 11.8: Extension of the path 𝛾𝑧 to the path 𝛾𝑧+ℎ in the open disc 𝔻 𝑧0 , 𝑅 . The path is now
between points 𝑧0 and 𝑧 + ℎ.

RHS, version May 11, 2019 85

Complex Analysis I 11 INTEGRAL THEOREMS

Now exactly the same way as in (I), we can estimate the above to be at most

sup ||𝑓 (𝑢) − 𝑓 (𝑧)|| ,

𝑢∈𝛾[𝑧,𝑧+ℎ]

which can be made sufficiently small whenever|ℎ| is small enough. This shows that
𝜕
𝐹 (𝑧) = 𝑓 (𝑧).
𝜕𝑥

Now we have shown that (∗) holds, finishing the proof of the theorem.

Theorem 11.0.5: The Cauchy–Goursat theorem. If 𝑓 is analytic in a disc, then

𝑓 (𝑧) d𝑧 = 0,
∫𝛾

for any closed smooth curve in that disc.

|𝛾|

Figure 11.9: A closed smooth curve 𝛾 in a disc 𝔻.

Proof. By 11.0.4, since 𝑓 has an antiderivative, we can apply corollary 10.0.6 of the Fundamental
Theorem of Calculus.

RHS, version May 11, 2019 86

Complex Analysis I 12 CAUCHY’S INTEGRAL FORMULAS

12 Cauchy’s integral formulas

Integral representation formulas are important in mathematics.

Theorem 12.0.1: Cauchy’s local integral formula. Suppose that 𝑓 is analytic in a domain Ω
( )
and that 𝔻 = 𝔻 𝑧0 , 𝑟 is a disc such that 𝔻 ⊂ Ω. If 𝜕𝔻 denotes the boundary of 𝔻 with positive
orientation, then
1 𝑓 (𝑢)
𝑓 (𝑧) = d𝑢, for any 𝑧 ∈ 𝔻.
2𝜋𝑖 ∫𝜕𝔻 𝑢 − 𝑧

𝜕𝔻

𝑟
𝑧0

( )
Figure 12.1: The domain Ω, a disc 𝔻 = 𝔻 𝑧0 , 𝑟 such that the closure 𝔻 ⊂ Ω and the boundary
𝜕𝔻 with positive orientation. The value of an analytical function 𝑓 in this disc can be found by
the Theorem 12.0.1.

Remark.
(1) Now we have an integral formula expressing the value of 𝑓 inside a disc in terms of the
values of 𝑓 on the boundary of the disc.
(2) Any analytic function in a small disc can be expressed as an integral.
( )
Examples 12.0.2. Let 𝜕𝔻 denote the boundary of the unit disc 𝔻 0, 1 with positive orientation.
Evaluate the following integrals:
(1)
exp(𝑧) exp(𝑧) ( )
d𝑧 = d𝑧 = 2𝜋𝑖 exp 0 = 2𝜋𝑖.
∮𝜕𝔻 𝑧 ∮𝜕𝔻 𝑧 − 0

(2)
exp(𝑧) ( )
exp(𝑧) 𝑧−2 exp 0
( ) d𝑧 = d𝑧 = 2𝜋𝑖 = −𝜋𝑖,
∮𝜕𝔻 𝑧 𝑧 − 2 ∮𝜕𝔻 𝑧 − 0 0−2
( )
exp(𝑧)
since 𝑧 ↦ 𝑧−2
is analytic in 𝔻 0, 32 .

RHS, version May 11, 2019 87

Complex Analysis I 12 CAUCHY’S INTEGRAL FORMULAS

Im
𝑖

0 1 Re

( )
Figure 12.2: The boundary (with a positive orientation) of the unit disc 𝔻 0, 1 .

Before we prove the Cauchy local integral formula, we give its corollary. We have a second
remarkable fact about analytic functions, namely their regularity:

Theorem 12.0.3: Cauchy’s second formula. If 𝑓 is analytic in a domain Ω, then 𝑓 has infinitely
many complex derivatives in Ω. Moreover, if 𝔻 ⊂ Ω is a disc so that the boundary 𝜕𝔻 is positively
oriented, then
𝑛! 𝑓 (𝑢)
𝑓 (𝑛) (𝑧) = d𝑢, for all 𝑧 ∈ int 𝔻.
2𝜋𝑖 ∫𝜕𝔻 (𝑢 − 𝑧)𝑛+1

𝜕𝔻

Figure 12.3: The positively oriented boundary 𝜕𝔻 of the disc 𝔻 and 𝔻 ⊂ Ω, where Ω is a domain.
( )
Proof of the Cauchy local integral formula 12.0.1. Let 𝑧 ∈ 𝔻 𝑧0 , 𝑟 be fixed. Let us define the
function {
𝑓 (𝑢)−𝑓 (𝑧)
𝑢−𝑧
, when 𝑢 ∈ Ω ⧵ {𝑧},
𝐹 (𝑢) = ′
𝑓 (𝑧), when 𝑢 = 𝑧.
Then 𝐹 is analytic in Ω ⧵ {𝑧} and continuous in Ω.

RHS, version May 11, 2019 88

Complex Analysis I 12 CAUCHY’S INTEGRAL FORMULAS

𝑧 𝑟
𝑧0

Figure 12.4: (The function

) 𝐹 in the proof of the Cauchy local integral formula 12.0.1 is defined
in the disc 𝔻 𝑧0 , 𝑟 with a fixed point 𝑧.

( ) ( )
Since 𝔻 𝑧0 , 𝑟 ⊂ Ω and Ω is a domain, there exists an open disc 𝔻1 such that 𝔻 𝑧0 , 𝑟 ⊂ 𝔻1 ⊂ Ω.
By the weakened form of Goursat’s theorem we know that
𝑓 (𝑢) 𝑓 (𝑧)
0= 𝐹 (𝑢) d𝑢 = d𝑢 − d𝑢
∫𝜕𝔻(𝑧0 ,𝑟) ∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧 ∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧
𝑓 (𝑢) d𝑢
= d𝑢 − 𝑓 (𝑧) ,
∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧 ∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧
( )
since 𝐹 is analytic in 𝔻1 ⧵ {𝑧} and continuous in 𝔻1 and 𝜕𝔻 𝑧0 , 𝑟 is a closed contour in 𝔻1 .
It remains to show that
𝑓 (𝑧) d𝑢
d𝑢 = = 2𝜋𝑖,
∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧 ∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧

from which the claim follows.

( )
Let 𝛾0 ∶ [0, 1] → Ω, 𝛾0 (𝑡) = 𝑧0 + 𝑟 exp 2𝜋𝑖𝑡 . Then
( )
1 𝑟2𝜋𝑖 exp 2𝜋𝑖𝑡
d𝑢
= ( ) d𝑡
∫𝛾0 𝑢 − 𝑧 ∫0 𝑧0 + 𝑟 exp 2𝜋𝑖𝑡 − 𝑧
( )
1 𝑟 exp 2𝜋𝑖𝑡
= 2𝜋𝑖 ( ) d𝑡
∫0 𝑧0 − 𝑧 + 𝑟 exp 2𝜋𝑖𝑡
1
d𝑡
= 2𝜋𝑖 𝑧0 −𝑧
∫0 +1
𝑟 exp(2𝜋𝑖𝑡)
𝑧0 −𝑧 ( ) 𝑧0 −𝑧 ( )
1 1+ exp −2𝜋𝑖𝑡 1 exp −2𝜋𝑖𝑡
𝑟 𝑟
= 2𝜋𝑖 ( ) d𝑡 − 2𝜋𝑖 ( ) d𝑡
∫0 1 + 𝑧0 −𝑧 exp −2𝜋𝑖𝑡 ∫0 𝑧 −𝑧
1 + 0𝑟 exp −2𝜋𝑖𝑡
𝑟
𝑧0 −𝑧 ( )
1
𝑟
exp −2𝜋𝑖𝑡
= 2𝜋𝑖 − 2𝜋𝑖 𝑧 −𝑧 ( ) d𝑡.
∫0 1 + 0
exp −2𝜋𝑖𝑡
𝑟

Now, let us write

( ) 𝑧0 − 𝑧 ( )
𝛾1 ∶ [0, 1] → 𝔻 0, 1 , 𝛾1 (𝑡) = exp −2𝜋𝑖𝑡 ,
𝑟
and hence 𝑧0 − 𝑧 ( )
𝛾1′ (𝑡) = −2𝜋𝑖 exp −2𝜋𝑖𝑡 .
𝑟
Thus, we get
1 𝛾1′ (𝑡) d𝑢
d𝑡 = .
∫0 1 + 𝛾1 (𝑡) ∫𝛾1 1 + 𝑢

RHS, version May 11, 2019 89

Complex Analysis I 12 CAUCHY’S INTEGRAL FORMULAS

( ) ( )
Since ||𝑧0 − 𝑧|| < 𝑟, we have ||𝛾1 || ⊂ 𝔻 0, 1 . Since 𝑔(𝑢) = 1
1+𝑢
is analytic in 𝔻 0, 1 , we obtain by
Theorem 11.0.5 that
d𝑢
= 0.
∫𝛾1 1 + 𝑢
Thus
𝑓 (𝑢) ( )
0= d𝑢 − 𝑓 (𝑧) 2𝜋𝑖 + 0 ,
∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧
and the claim follows.

Proof of the Cauchy’s second formula 12.0.3. We assume that the boundary 𝜕𝔻 has positive ori-
entation. The proof is by induction on 𝑛. The case 𝑛 = 1 is the Cauchy integral formula.
Suppose that 𝑓 has up to 𝑛 − 1 complex derivatives and
( )
𝑛−1 ! 𝑓 (𝑢)
𝑓 (𝑛−1)
(𝑧) = d𝑢.
2𝜋𝑖 ∫𝜕𝔻 (𝑢 − 𝑧)𝑛

Now, for ℎ small, the difference quotient for 𝑓 (𝑛−1) takes the form:
( ) ( )
𝑓 (𝑛−1) (𝑧 + ℎ) − 𝑓 (𝑛−1) (𝑧) 𝑛−1 ! 1 1 1
= 𝑓 (𝑢) − d𝑢.
ℎ 2𝜋𝑖 ∫𝜕𝔻 ℎ (𝑢 − 𝑧 − ℎ)𝑛 (𝑢 − 𝑧)𝑛

( )
Recall that 𝑎𝑛 − 𝑏𝑛 = (𝑎 − 𝑏) 𝑎𝑛−1 + 𝑎𝑛−2 𝑏 + … + 𝑎𝑏𝑛−2 + 𝑏𝑛−1 .
1 1
With 𝑎 = 𝑢−𝑧−ℎ
and 𝑏 = 𝑢−𝑧
, we have
( )
1 1 1
𝑛 − 𝑛 = 𝑎𝑛−1 + 𝑎𝑛−2 𝑏 + … + 𝑎𝑏𝑛−2 + 𝑏𝑛−1 .
(𝑢 − 𝑧 − ℎ) (𝑢 − 𝑧) (𝑢 − 𝑧 − ℎ) (𝑢 − 𝑧)
Since ℎ is small enough, the points 𝑧 + ℎ and 𝑧 stay at a finite distance from the boundary. Hence,
as ℎ → 0, the quotient converges to
( )
𝑛−1 ! 1 𝑛 𝑛! 𝑓 (𝑢)
𝑓 (𝑢) d𝑢 = d𝑢,
2𝜋𝑖 ∫𝜕𝔻 2
(𝑢 − 𝑧) (𝑢 − 𝑧) 𝑛−1 2𝜋𝑖 ∫ 𝜕𝔻 (𝑢 − 𝑧)
𝑛+1

which completes the induction argument and proves the theorem.

( )
Example 12.0.4. Evaluate in a disc 𝔻 0, 𝑟

sin 𝑧
( ) d𝑧.
∫ 𝑧 − 2 exp (𝑧)

1. The case 𝑟 = 1:
sin 𝑧
𝑓 (𝑧) = ( )
𝑧 − 2 exp (𝑧)
( )
is analytic in 𝔻 0, 32 . So the C-G theorem implies that ∫ 𝑓 (𝑧) d𝑧 = 0.

2. The case 𝑟 = 3:
sin 𝜔
𝑓 (𝜔) =
exp (𝜔)
( ) ( ) ( ) ( )
is analytic in 𝔻 0, 4 21 and 2 ∈ 𝔻 0, 3 and 𝔻 0, 3 ⊂ 𝔻 0, 4 21 . The Cauchy local
integral formula gives
sin(𝜔) ( )
exp(𝜔) sin 2
d𝜔 = 2𝜋𝑖 .
∫𝜕𝔻(0,3) 𝜔 − 2 𝑒2

RHS, version May 11, 2019 90

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

13 Corollaries of the Cauchy integral formulas

Corollary 13.0.1. Let Ω be an open set in ℂ. If 𝑓 ∶ Ω → ℂ has an antiderivative, then 𝑓 is
analytic.

Proof. Let 𝐹 be an antiderivative of 𝑓 . Then, by Definition 10.0.1, 𝐹 is analytic and 𝐹 ′ = 𝑓 .

By Theorem 12.0.3, 𝑓 is analytic.

Theorem 13.0.2: Morera’s theorem. Let Ω be an open set in ℂ. Suppose that 𝑓 ∶ Ω → ℂ is

continuous and
𝑓 (𝑧) d𝑧 = 0
∫Γ
along every closed contour in Ω. Then 𝑓 is analytic.

Proof. Theorem 10.0.9 and Corollary 13.0.1.

13.0.3. Cauchy’s inequalities

( )
Suppose that 𝑓 is analytic in an open set Ω and 𝔻 𝑧0 , 𝑅 ⊂ Ω. Then:

| (𝑛) ( )| 𝑛!
|𝑓 𝑧0 | ≤ 𝑛 ‖𝑓 ‖𝜕𝔻(𝑧0 ,𝑅) ,
| | 𝑅

where ‖𝑓 ‖𝜕𝔻(𝑧0 ,𝑅) = sup𝑧∈𝜕𝔻(𝑧0 ,𝑅) ||𝑓 (𝑧)||.

( )
Proof. We apply the Cauchy integral formula for 𝑓 (𝑛) 𝑧0 :

| | | ( ) |
| | 𝑛! || |
| (𝑛) ( )| | 𝑛!
2𝜋 𝑓 𝑧 + 𝑅 exp(𝑖𝜙) 𝑖𝑅 exp(𝑖𝜙)
𝑓 (𝑢) | 0 |
|𝑓 𝑧0 | = | d𝑢 | = | d𝜙 |
| | | 2𝜋𝑖 ∫𝜕𝔻(𝑧 ,𝑅) ( )𝑛+1 | 2𝜋𝑖 |∫ ( )𝑛+1 |
| 0 𝑢 − 𝑧0 | | 0 𝑅 exp(𝑖𝜙) |
| | | |
| 2𝜋 ( ) |
𝑛! || 𝑓 𝑧0 + 𝑅 exp(𝑖𝜙) |
= | ( )𝑛 d𝜙||
2𝜋 | 0∫ 𝑅 exp(𝑖𝜙) |
| |
𝑛! ‖𝑓 ‖𝜕𝔻(𝑧0 ,𝑅) 𝑛!
≤ 2𝜋 = 𝑛 ‖𝑓 ‖𝜕𝔻(𝑧0 ,𝑅) .
2𝜋 𝑅𝑛 𝑅

( )
Theorem 13.0.4. Suppose that 𝑓 is analytic in an open set Ω and 𝔻 𝑧0 , 𝑅 ⊂ Ω. Then 𝑓 has a
power series expansion at 𝑧0
∑
∞
( )𝑛
𝑓 (𝑧) = 𝑎𝑛 𝑧 − 𝑧0 ,
𝑛=0
( )
for all 𝑧 ∈ 𝔻 𝑧0 , 𝑅 and the coefficients are given by
( )
𝑓 (𝑛) 𝑧0
𝑎𝑛 = , for all 𝑛 = 0, 1, 2 …
𝑛!

Note 13.0.5. This is a local result.

( )
Proof. Let 0 < 𝑟 < 𝑅 and 𝑧 ∈ 𝔻 𝑧0 , 𝑅 . By the Cauchy’s integral formula:

1 𝑓 (𝑢) ( )
𝑓 (𝑧) = d𝑢, 𝑧 ∈ 𝔻 𝑧0 , 𝑅 .
∫
2𝜋𝑖 𝜕𝔻(𝑧0 ,𝑅) 𝑢 − 𝑧

RHS, version May 11, 2019 91

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

( )
For each 𝑢 ∈ 𝜕𝔻 𝑧0 , 𝑟 we can write

1 1 1 1
= =
𝑢 − 𝑧 𝑢 − 𝑧0 − (𝑧 − 𝑧0 ) 𝑢 − 𝑧0 1 − 𝑧−𝑧0
𝑢−𝑧0
( )𝑘
1 ∑
∞
𝑧 − 𝑧0
= ,
𝑢 − 𝑧0 𝑘=0
𝑢 − 𝑧0

since
| 𝑧 − 𝑧 | |𝑧 − 𝑧 |
| 0| | 0|
| |= < 1.
| 𝑢 − 𝑧0 | 𝑟
| |
Hence,
( )𝑘
1 ∑∞
𝑓 (𝑢) 𝑧 − 𝑧0
𝑓 (𝑧) = d𝑢.
2𝜋𝑖 ∫ 𝑢 − 𝑧0 𝑢 − 𝑧0
𝑘=0
𝜕𝔻(𝑧0 ,𝑟)
( )
Since 𝑓 is continuous on the compact set 𝜕𝔻 𝑧0 , 𝑟 , 𝑓 is bounded there and |𝑓 (𝑢)| ≤ 𝑀 ≤ ∞
( )
for all 𝑢 ∈ 𝜕𝔻 𝑧0 , 𝑟 . Thus
| ( ) |
𝑧 − 𝑧0 𝑘 || 𝑀 ||𝑧 − 𝑧0 ||
𝑘
| 𝑓 (𝑢)
| ⋅ ≤ .
|𝑢 − 𝑧 𝑢 − 𝑧0 ||
| 0 𝑟 𝑟𝑘
| |
( )𝑘
∑
∞
𝑀 |𝑧−𝑧0 |
Here 𝑟 𝑟
converges as a geometric series (with 𝑞 < 1). By the Weiestrass criteria
𝑘=0
( )𝑘
∑∞
𝑓 (𝑢) 𝑧 − 𝑧0
𝑘=0
𝑢 − 𝑧0 𝑢 − 𝑧0
( )
converges uniformly on 𝜕𝔻 𝑧0 , 𝑟 . Hence, the order of the summation and integration can me
changed. By changing the order of summation and integration, we get
( )
1 ∑
∞
𝑓 (𝑢) ( )𝑘
𝑓 (𝑧) = d𝑢 𝑧 − 𝑧0 .
2𝜋𝑖 𝑘=0 ∫𝜕𝔻(𝑧0 ,𝑟) (𝑢 − 𝑧) 𝑘+1

By the Cauchy’s 2nd local formula

𝑓 (𝑢) 2𝜋𝑖 (𝑘) ( )
d𝑢 = 𝑓 𝑧0 .
∫𝜕𝔻(𝑧0 ,𝑟) (𝑢 − 𝑧)𝑘+1 𝑘!

Thus ( ) ( )
∑
∞
𝑓 (𝑘) 𝑧0 ( )𝑘 𝑓 (𝑘) 𝑧0
𝑓 (𝑧) = 𝑧 − 𝑧0 , where 𝑎𝑘 = .
𝑘=0
𝑘! 𝑘!
Note that 𝑟 can be chosen as close as we like near 𝑅 and since the coefficients are unique, we get
always the same series. Hence

∑
∞
( )𝑘 1 𝑓 (𝑢)
𝑓 (𝑧) = 𝑎𝑘 𝑧 − 𝑧0 , with 𝑎𝑘 = ( )𝑘+1 d𝑢, 0 < 𝑟 < 𝑅,
2𝜋𝑖 ∫ 𝑢 − 𝑧0
𝑘=0
𝜕𝔻(𝑧0 ,𝑟)
( )
is valid in the whole 𝔻 𝑧0 , 𝑅 .

Remark. This theorem gives another proof that an analytic function is automatically indefinitely
differentiable.

Remark. In particular, if 𝑓 is entire (that is, 𝑓 is analytic on whole ℂ), the theorem implies that
∑
𝑓 has a power series expansion around 0, 𝑓 (𝑧) = ∞ 𝑛=0 𝑎𝑛 𝑧 , that converges in all of ℂ.
𝑛

Corollary 13.0.6: Liouville’s theorem. If 𝑓 is entire and bounded, then 𝑓 is constant.

RHS, version May 11, 2019 92

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

Remark 13.0.7. In the proof, we use the following lemma: If 𝑓 is analytic in a domain Ω and
𝑓 ′ = 0, then 𝑓 is constant.

Proof of remark 13.0.7. Fix 𝜔0 ∈ Ω. It suffices to show that 𝑓 (𝜔) = 𝑓 (𝜔0 ) for all 𝜔 ∈ Ω. Since
Ω is connected, for any 𝜔 ∈ Ω, there exists a suitable curve 𝛾𝜔 which joins 𝜔0 to 𝜔. Since 𝑓 is
an antiderivative of 𝑓 ′ , we have, by Fundamental Theorem of Calculus,

𝑓 ′ (𝑧) d𝑧 = 𝑓 (𝜔) − 𝑓 (𝜔0 ).

∫𝛾𝜔

By assumption, 𝑓 ′ = 0, so the integral on the left hand side is 0, and we obtain 𝑓 (𝜔) = 𝑓 (𝜔0 ).

Proof of Liouville’s theorem 13.0.6. It suffices, because of Remark 13.0.7, to prove that 𝑓 ′ = 0,
since ℂ connected. For each 𝑧0 ∈ ℂ and 𝑅 > 0, the Cauchy inequality yields

| ′ | 𝑀
|𝑓 (𝑧0 )| ≤
| | 𝑅
where 𝑀 is a bound for 𝑓 .
Letting 𝑅 → ∞, gives that ||𝑓 ′ || = 0. This implies (a small lemma is needed) that 𝑓 ′ = 0.

We can give an elegant proof for the Fundamental Theorem of Algebra.

Corollary 13.0.8. Every non-constant polynomial 𝑝(𝑧) = 𝑎𝑛 𝑧𝑛 +𝑎𝑛−1 𝑧𝑛−1 +…+𝑎0 with complex
coefficients has a root in ℂ.

1
Proof. If 𝑝 has no root, then 𝑝
is a bounded analytic function. To see this, let us assume that
𝑎𝑛 ≠ 0 and write ( )
𝑝(𝑧) 𝑎𝑛−1 𝑎
= 𝑎𝑛 + + … + 𝑛0 ,
𝑧𝑛 𝑧 𝑧
whenever 𝑧 ≠ 0. Since every term in the brackets goes to 0 as |𝑧| → ∞, there exists 𝑅 > 0 so
|𝑎 |
that ||𝑝(𝑧)|| ≥ 2𝑛 |𝑧|𝑛 whenever|𝑧| > 𝑅. Hence 𝑝 is bounded from below when|𝑧| > 𝑅.
Since 𝑝( is continuous
) and has no roots in the disc |𝑧| ≤ 𝑅, it is bounded from below in the
disc 𝔻 0, 𝑅 as well. So 1𝑝 is indeed a bounded analytic function. By Liouville’s theorem, 1𝑝 is
constant. But this contradicts our assumption that 𝑝 is non-constant and proves this corollary.

Corollary 13.0.9. Every polynomial 𝑝(𝑧) = 𝑎𝑛 𝑧𝑛 + 𝑎𝑛−1 𝑧𝑛−1 + … + 𝑎0 of degree 𝑛 ≥ 1 has

precisely 𝑛 roots in ℂ. If these roots are denoted by 𝜔1 , 𝜔2 , … , 𝜔𝑛 , then

𝑝(𝑧) = 𝑎𝑛 (𝑧 − 𝜔1 )(𝑧 − 𝜔2 ) … (𝑧 − 𝜔𝑛 ).

Remark 13.0.10. The following claims are essentially equivalent in an open disk 𝔻:
1. 𝑓 is analytic,
2. 𝑓 has a primitive and
3. for all closed 𝔻
𝑓 (𝑢) d𝑢 = 0.
∫
𝛾

The link from the 1. to 2. comes from Cauchy-Goursat theorem. The link from the 2. to
3. comes from The fundamental theorem of calculus. The link from the 2. to 1. comes
from Cauchy’s 2nd theorem. The link from the 3. to 2. comes from The 2nd fundamental
theorem of calculus.

RHS, version May 11, 2019 93

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

13.0.11 Analytic continuation

In the following theorem, Ω is assumed connected.

Theorem 13.0.12: Identity theorem. Suppose that 𝑓 is an analytic function in a domain Ω that
vanishes on a sequence of distinct points with a limit point in Ω. Then 𝑓 is identically 0.

Proof. Suppose that 𝑧0 ∈ Ω is a limit point for the sequence (𝜔𝑘 )∞ 𝑘=1
and 𝑓 (𝜔𝑘 ) = 0.
( )
First, we show that 𝑓 ≡ 0 in a small disc containing 𝑧0 . For that, we choose a disc 𝔻 𝑧0 , 𝑟 ⊂ Ω
∑
with 𝑟 > 0 fixed. We consider the power series expansion of 𝑓 in this disc, 𝑓 (𝑧) = ∞ 𝑛=0 𝑎𝑛 (𝑧 −
𝑧0 ) .
𝑛
( )
If we assume that 𝑓 is not identically 0 in 𝔻 𝑧0 , 𝑟 , then there exists a smallest integer 𝑚 such
that 𝑎𝑚 ≠ 0. But then
𝑓 (𝑧) = 𝑎𝑚 (𝑧 − 𝑧0 )𝑚 (1 + 𝑔(𝑧 − 𝑧0 )),
where 𝑔(𝑧−𝑧0 ) converges to 0 as 𝑧 → 𝑧0 . Taking 𝑧 = 𝜔𝑘 ≠ 𝑧0 for a sequence of points converging
to 𝑧0 , we get
𝑎𝑚 (𝜔𝑘 − 𝑧0 ) ≠ 0 and (1 + 𝑔(𝑧 − 𝑧0 )) ≠ 0,
which is a contradiction, since 𝑓 (𝜔𝑘 ) = 0.
Now, we apply the fact that Ω is connected. Let

Ω1 = int{𝑧 ∈ Ω ∶ 𝑓 (𝑧) = 0}.

Then Ω1 is open by definition and Ω is non-empty by the first part. However, Ω1 is also closed:
if 𝑧𝑛 ∈ Ω1 and 𝑧𝑛 → 𝑧, then 𝑓 (𝑧) = 0, since 𝑓 is continuous and 𝑓 ≡ 0 in a small open disc
containing 𝑧 by the first part. Hence 𝑧 ∈ Ω1 .
Since Ω is connected and 𝑧0 ∈ Ω1 , and hence Ω1 ≠ ∅, we have Ω1 = Ω.

Remark 13.0.13. In other words, if the zeros of an analytical function 𝑓 in a domain Ω accumu-
lates in Ω, then 𝑓 = 0.

Remark 13.0.14. Note that in this theorem the points should accumulate in Ω.

Example 13.0.15. The function sin 𝑧 is entire and

sin 𝑧 = 0 ⇔ 𝑧 = 𝑛𝜋, 𝑛 = 0, ±1, ±2, … , .

Let 𝑓 (𝑧) = sin 1𝑧 . Then 𝑓 (𝑧𝑘 ) = 0 at the points 𝑧𝑘 = 1

𝑘𝜋
, where 𝑘 = ±1, ±2, … .
But 0 is the accumulation point for
{ }
1
∶ 𝑛 = ±1, ±2, … .
𝑛𝜋

However, 0 is not in the domain of analyticity of 𝑓 .

Note that sin 1𝑧 is analytic in ℂ ⧵ {0}.
( ) { } { }
Example 13.0.16. Let 𝑟(𝑧) = exp 1𝑧 − 1, 𝑧 ∈ ℂ ⧵ 0 . Then 𝑔 is analytic in ℂ ⧵ 0 . Note that

1 𝑖 { }
𝑔(𝑧) = 0 ⇔ 𝑧𝑛 = =− , when 𝑛 ∈ ℤ ⧵ 0 .
2𝜋𝑖𝑛 2𝜋𝑛
{ }
Here the points 𝑧𝑛 accumulate to the origin, but 0 ∉ ℂ ⧵ 0 . And 𝑔 is not identically zero.

Corollary 13.0.17. Suppose that 𝑓 and 𝑔 are analytic in a domain Ω and 𝑓 (𝑧) = 𝑔(𝑧) for all 𝑧
in some sequence of distinct points with a limit point in Ω. Then 𝑓 (𝑧) = 𝑔(𝑧) throughout Ω.

RHS, version May 11, 2019 94

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

Corollary 13.0.18: Corollary of the Identity theorem. The Uniqueness theorem. Suppose that
𝑓 and 𝑔 are analytic in a domain Ω and 𝑓 (𝑧) = 𝑔(𝑧) for all 𝑓 in some non-empty open subset of
Ω. Then
𝑓 (𝑧) = 𝑔(𝑧), throughout Ω.

Remark 13.0.19. Suppose that we are given a pair of functions 𝑓 and 𝑔 which are analytic in
domains Ω and Ω, ̃ respectively with Ω ⊂ Ω ̃ (Figure 13.1.) If 𝑓 and 𝑔 agree on the smaller set Ω,
we say that 𝑔 is an analytic continuation of 𝑓 into domain Ω.̃ The corollary 13.0.18 guarantees
that there can be only one such continuation, since 𝑔 is uniquely determined by 𝑓 .

𝑓 analytic

𝑔 analytic
̃
Ω

Figure 13.1: If analytic functions 𝑓 and 𝑔 agree on domain Ω ⊂ Ω,̃ then 𝑔 is an analytic continua-
tion of 𝑓 and as 𝑔 is uniquely determined by 𝑓 there can be only one such continuation. (Remark
13.0.19.)

13.0.20 (The Maximum modulus principle). If 𝑓 is a non-constant analytic function in a domain

Ω, then|𝑓 | can not attain a maximum in Ω.
( )
Theorem 13.0.21: Local maximum modulus theorem. Suppose that 𝑓 is analytic in 𝔻 𝑧0 , 𝑅 .
( )
If |𝑓 | has a local maximum point in 𝔻 𝑧0 , 𝑅 , ie.
( )
|𝑓 (𝑧)| ≤ |𝑓 (𝑧0 )| for all 𝑧 ∈ 𝔻 𝑧0 , 𝑅 . (13.0.22)
| | | |
Then 𝑓 is constant.

RHS, version May 11, 2019 95

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

( )
𝜕𝔻 𝑧0 , 𝑟

𝑅
𝑟
𝑧0

Figure
( 13.2:
) In the proof of Theorem
[ ]13.0.21, the integration path around the point 𝑧0 is 𝛾(𝑡) =
𝜕𝔻 𝑧0 , 𝑟 = 𝑧0 + 𝑟 exp(𝑖𝑡), 𝑡 ∈ 0, 2𝜋 .

Proof. Fix 𝑟 such that 0 < 𝑟 < 𝑅. (Figure 13.2.) By the Cauchy integral formula, we can write

1 𝑓 (𝑢)
𝑓 (𝑧0 ) = d𝑢
2𝜋𝑖 ∫𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧0
2𝜋 𝑓 (𝑧0 + 𝑟 exp(𝑖𝑡))𝑖𝑟 exp(𝑖𝑡)
1
= d𝑡
2𝜋𝑖 ∫0 𝑟 exp(𝑖𝑡)
2𝜋
1
= 𝑓 (𝑧0 + 𝑟 exp(𝑖𝑡)) d𝑡.
2𝜋 ∫0

Hence
2𝜋
|𝑓 (𝑧0 )| ≤ 1 |𝑓 (𝑧 + 𝑟 exp(𝑖𝑡))| d𝑡 ≤ |𝑓 (𝑧 )| ,
| | 2𝜋 ∫ | 0 | | 0 |
0
and therefore
2𝜋
|𝑓 (𝑧0 )| = 1 |𝑓 (𝑧0 + 𝑟 exp(𝑖𝑡))| d𝑡.
| | 2𝜋 ∫ | |
0
We now have
2𝜋
|𝑓 (𝑧0 )| − |𝑓 (𝑧0 + 𝑟 exp(𝑖𝑡))| d𝑡 = 0,
∫0 | | | |

where ||𝑓 (𝑧0 )|| − ||𝑓 (𝑧0 + 𝑟 exp(𝑖𝑡))|| ≥ 0 by our assumption and the continuity of 𝑓 .
Hence
|𝑓 (𝑧 )| − |𝑓 (𝑧 + 𝑟 exp(𝑖𝑡))| ≡ 0
| 0 | | 0 |
( )
for all 𝑡. This is true for all 𝑟 < 𝑅. Thus,|𝑓 | is constant in 𝔻 𝑧0 , 𝑅 , meaning that 𝑓 is constant
there as well.

Corollary 13.0.23. Suppose that 𝑓 is analytic in a domain Ω. If |𝑓 | has a local maximum point
in Ω, then 𝑓 is constant.

Proof. Apply Theorem 13.0.21 and the identity theorem 13.0.12.

Theorem 13.0.24: Maximum modulus theorem. Suppose that Ω is a domain with a compact
closure Ω. If 𝑓 is analytic in Ω and continuous on Ω, then

sup||𝑓 (𝑧)|| ≤ sup ||𝑓 (𝑧)|| .

𝑧∈Ω 𝑧∈Ω⧵Ω

RHS, version May 11, 2019 96

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

Proof. Since Ω is bounded and closed, and 𝑓 is continuous on Ω, |𝑓 | attains its supremum, 𝑀,
at some point of Ω.
Now, assume that|𝑓 | does not attain the value 𝑀( on Ω)⧵ Ω. Then ||𝑓 (𝑧𝑚 )|| = 𝑀 for some 𝑧𝑚 ∈ Ω.
Since Ω is open, there exists 𝑅 > 0 such that 𝔻 𝑧𝑚 , 𝑅 ⊂ Ω and
( )
|𝑓 (𝑧)| ≤ |𝑓 (𝑧𝑚 )| for all 𝑧 ∈ 𝔻 𝑧𝑚 , 𝑅 .
| | | |
( )
Thus, by the local maximum modulus theorem 13.0.21, 𝑓 is constant on 𝔻 𝑧𝑚 , 𝑅 . Using the
identity theorem 13.0.12 gives us that 𝑓 is constant on Ω. Finally, by the continuity of 𝑓 , 𝑓 is
also constant on Ω.
So 𝑓 attains its supremum, 𝑀 at every point of Ω = Ω ∪ 𝜕Ω, contrary to hypothesis.

Example 13.0.25. Let Ω be the open 1st quadrant bounded by the positive half-line 𝑥 ≥ 0 and
the corresponding imaginary line 𝑦 ≥ 0. Let us consider the function
( )
𝐹 (𝑧) = exp −𝑖𝑧2 .

Then 𝐹 is entire and 𝐹 is continuous on Ω, and|𝐹 (𝑥)| = 1 on the two boundary lines 𝑧 = 𝑥 and
𝑧 = 𝑖𝑦. However, 𝐹 (𝑧) is unbounded in Ω, since for example when we pick up
( ) √
𝜋 ( ) 2
𝑧 = 𝑟 exp 𝑖 =𝑟 1+𝑖 , 𝑟 > 0,
4 2
( )
then 𝐹 (𝑧) = exp 𝑟2 .

( )
Lemma 13.0.26: The Schwarz lemma. Suppose that 𝑓 is analytic in 𝔻 0, 𝑅 , 𝑓 (0) = 0 and
( )
|𝑓 (𝑧)| ≤ 𝑀 for all 𝑧 ∈ 𝔻 0, 𝑅 .
| |
Then
|𝑓 (𝑧)| ≤ 𝑀 |𝑧| for all |𝑧| ≤ 𝑅.
| | 𝑅
( )
Proof. Since 𝑓 (0) = 0, there exists an analytic function 𝑔 on 𝔻 0, 𝑅 such that
( )
𝑓 (𝑧) = 𝑧𝑔(𝑧) for all 𝑧 ∈ 𝔻 0, 𝑅 .

On 0 < |𝑧| = 𝑟 < 𝑅 (see Figure 13.3), we have

|𝑔(𝑧)| ≤ 𝑓 (𝑧) ≤ 𝑀 .
| | |𝑧| 𝑟
By the maximum modulus theorem 13.0.24 for 𝑔, we get

|𝑔(𝑧)| ≤ 𝑀 for all |𝑧| ≤ 𝑟.

| | 𝑟
Now, we let 𝑟 → 𝑅 and we obtain

|𝑔(𝑧)| ≤ 𝑀 for all |𝑧| < 𝑅 and 𝑧 ≠ 0.

| | 𝑅
At 𝑧 = 0, we had 𝑓 (𝑧) = 0 and therefore the required inequality holds for all|𝑧| ≤ 𝑅.

RHS, version May 11, 2019 97

Complex Analysis I 13 COROLLARIES OF THE CAUCHY INTEGRAL FORMULAS

𝑅
𝑟
𝑧0

Figure 13.3: In the proof of Schwarz Lemma (13.0.26), choose 𝑟 such that 0 < 𝑟 < 𝑅.

Remark. If equality occurs in

|𝑓 (𝑧)| ≤ 𝑀 |𝑧|
| | 𝑅
for some 𝑧 with|𝑧| < 𝑅, then there exists 𝜆 ∈ ℝ such that
𝑀 ( )
𝑓 (𝑧) = 𝑧 exp(𝑖𝜆) for 𝑧 ∈ 𝔻 0, 𝑅 .
𝑅
This means that 𝑓 is a rotation around the origin. The proof follows the proof of the maximum
modulus theorem 13.0.24.

Theorem 13.0.27: Minimum modulus theorem. Let Ω ⊂ ℂ be a bounded domain and 𝑓 ∶ Ω →

ℂ an analytic function with 𝑓 (𝑧) ≠ 0 for all 𝑧 ∈ Ω. Assume that 𝑓 is a non-constant function
and that 𝑓 ∶ Ω → ℂ is continuous.
Then
|𝑓 (𝑧)| ≥ min |𝑓 (𝜉)| , for all 𝑧 ∈ Ω.
| | 𝜉∈𝜕Ω| |

Proof. If 𝑓 (𝑧0 ) = 0 for some 𝑧0 ∈ 𝜕Ω, the claim is true. Therefore, we can assume that 𝑓 (𝑧0 ) ≠ 0
for all 𝑧0 ∈ 𝜕Ω.
1 1
Now 𝑓 (𝑧)
is analytic, because 𝑓 (𝑧) ≠ 0 for all 𝑧 ∈ Ω and 𝑓 is analytic. In addition, 𝑓 (𝑧)
is non-
1
constant, because 𝑓 was non-constant. Also, 𝑓 (𝑧)
is continuous on Ω, because 𝑓 was continuous
and non-zero on Ω.
Because Ω is a bounded domain, by the maximum modulus theorem 13.0.24, we have

1 1
>
𝑓 (𝑧0 ) 𝑓 (𝑧)

for some 𝑧0 ∈ 𝜕Ω and all 𝑧 ∈ Ω. Therefore ||𝑓 (𝑧)|| > 𝑓 (𝑧0 ) for all 𝑧 ∈ Ω.

RHS, version May 11, 2019 98

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

14 Global Cauchy theorem

14.1 Cycles
Definitions 14.1.1. Let 𝛾1 , … , 𝛾𝑘 be closed, piecewise smooth (𝐶 1 ) curves in the complex plane.
Let 𝑚1 , … , 𝑚𝑘 be integers.
• A cycle is a formal sum
𝜎 = 𝑚1 𝛾1 + 𝑚2 𝛾2 + ⋯ 𝑚𝑘 𝛾𝑘 .

• The trace of a cycle 𝜎 is

|𝜎| = ||𝛾1 || ∪ ||𝛾2 || ∪ ⋯ ∪ ||𝛾𝑘 || ,
[ ]
where the trace of a parametrized curve 𝛾𝑗 ∶ 𝑎, 𝑏 → ℂ is its image
{ }
| |
|𝛾𝑗 | = 𝛾𝑗 (𝑡) ∣ 𝑎 ≤ 𝑡 ≤ 𝑏 .
| |

A cycle in 14.1.1 is a collection of curves where the parametrized curve 𝛾𝑗 appears 𝑚𝑗 times if
𝑚𝑗 > 0, and the negative of 𝛾𝑗 , denoted by

𝛾𝑗− = −𝛾𝑗 ,

appears −𝑚𝑗 times if 𝑚𝑗 < 0.

If 𝑓 is continuous in an open set Ω and|𝜎| ⊂ Ω, then

∑
𝑘
𝑓 (𝑧) d𝑧 = 𝑚𝑗 𝑓 (𝑧) d𝑧.
∫𝜎 ∫𝛾𝑗
𝑗=1

Definition. The length of the cycle 𝜎 is

∑
𝑘
| |
length(𝜎) = |𝑚𝑗 | length(𝛾𝑗 ).
| |
𝑗=1

Remarks.
1. For a curve 𝛾,
𝑓 (𝑧) d𝑧 = − 𝑓 (𝑧) d𝑧,
∫𝛾 − ∫𝛾
where 𝛾 − = −𝛾 is the negative of 𝛾.
2. The negative of a cycle 𝜎 is

−𝜎 = −𝑚1 𝛾1 − 𝑚2 𝛾2 − ⋯ − 𝑚𝑘 𝛾𝑘

and ∫−𝜎 𝑓 (𝑧) d𝑧 = − ∫𝜎 𝑓 (𝑧) d𝑧.

Definition 14.1.2: Winding number. Suppose Ω is an open set in ℂ and 𝛾 is a closed piecewise
smooth curve in Ω and 𝑎 ∈ Ω ⧵ |𝛾|. Then

1 d𝑢
𝑛(𝛾; 𝑎) ∶=
2𝜋𝑖 ∫𝛾 𝑢 − 𝑎

is called the winding number around/about 𝑎.

Remark. 𝑎 ↦ 𝑛(𝛾, 𝑎) is an integer-valued function.

Lemma 14.1.3: Winding number lemma. For any closed piecewise smooth curve 𝛾 and 𝑎 ∉ |𝛾|,
the number 𝑛(𝛾, 𝑎) is an integer.

RHS, version May 11, 2019 99

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

Proof. Define a function ℎ ∶ [0, 1] → ℂ by letting

𝑥
𝛾 ′ (𝑡) [ ]
ℎ(𝑥) = d𝑡, for all 𝑥 ∈ 0, 1 .
∫0 𝛾(𝑡) − 𝑎

Then
1
𝛾 ′ (𝑡) d𝑧
ℎ(1) = d𝑡 = = 2𝜋𝑖 𝑛(𝛾; 𝑎). (∗)
∫0 𝛾(𝑡) − 𝑎 ∫𝛾 𝑧 − 𝑎

Since 𝛾 is piecewise smooth, there are only a finite number of points where 𝛾 ′ is not continuous.
Thus, for any 𝑥 such that 𝛾 ′ is continuous at 𝑥,

𝛾 ′ (𝑥)
ℎ′ (𝑥) = ,
𝛾(𝑥) − 𝑎

so 𝛾 ′ (𝑥) − ℎ′ (𝑥)(𝛾(𝑥) − 𝑎) = 0. Define 𝜑 ∶ [0, 1] → ℂ by 𝜑(𝑥) = (𝛾(𝑥) − 𝑎) exp(−ℎ(𝑥)). Then

d ( )
𝜑′ (𝑥) = (𝛾(𝑥) − 𝑎) exp(−ℎ(𝑥))
d𝑥
= 𝛾 ′ (𝑥) exp(−ℎ(𝑥)) + (𝛾(𝑥) − 𝑎)(−ℎ′ (𝑥)) exp(−ℎ(𝑥))
= exp(−ℎ(𝑥))(𝛾 ′ (𝑥) − ℎ′ (𝑥)(𝛾(𝑥) − 𝑎))
= exp(−ℎ(𝑥)) ⋅ 0
=0

for all but a finite number of 𝑥 ∈ [0, 1].

Since 𝜑′ (𝑥) = 0 almost everywhere, it is constant almost everywhere. But because 𝛾 and ℎ are
continuous, 𝜑 is continuous and, as such, must be a constant function, since [0, 1] is connected.
Notice that

𝜑(0) = (𝛾(0) − 𝑎) exp(−ℎ(0))

= (𝛾(0) − 𝑎) exp(0) = (𝛾(0) − 𝑎) ⋅ 1
= 𝛾(0) − 𝑎,

so because 𝜑 is constant, 𝛾(0) − 𝑎 = 𝜑(0) = 𝜑(1) = (𝛾(1) − 𝑎) exp(−ℎ(1)), i.e.

𝛾(0) − 𝑎
exp(−ℎ(1)) = 1.
𝛾(1) − 𝑎
Because 𝛾 is a closed curve, 𝛾(0) = 𝛾(1) and so exp(−ℎ(1)) = 1. This means that −ℎ(1) = 2𝜋𝑖𝑚
for some 𝑚 ∈ ℤ. But then by (∗),
( )
ℎ 1
𝑛(𝛾; 𝑎) = = −𝑚 ∈ ℤ,
2𝜋𝑖
which completes the proof.

Definition (Winding number of a cycle). The winding number of a cycle 𝜎 is

1 d𝑢
𝑛(𝜎; 𝑎) =
2𝜋𝑖 ∫𝜎 𝑢 − 𝑎

where 𝑎 ∉ |𝜎|.

Examples 14.1.4.
( )
1. If 𝛾 represents the boundary of the disc 𝔻 𝑧0 , 𝑅 (traversed counter clockwise), then
{ ( )
1 if 𝑎 ∈ 𝔻 𝑧0 , 𝑅
𝑛(𝛾; 𝑎) = ( )
0 if 𝑎 ∈ ℂ ⧵ 𝔻 𝑧0 , 𝑅

RHS, version May 11, 2019 100

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

( )
2. Let 𝛾𝑚 (𝑡) = exp 𝑖2𝜋𝑚𝑡 , 𝑡 ∈ [0, 1], 𝑚 ∈ ℤ ⧵ {0}. Then,
( )
1 2𝜋𝑚 exp 𝑖2𝜋𝑚𝑡
1 d𝑧 1
𝑛(𝛾𝑚 ; 0) = = ( ) d𝑡 = 𝑚.
2𝜋𝑖 ∫𝛾𝑚 𝑧 2𝜋 ∫0 exp 𝑖2𝜋𝑚𝑡

This explains the name “winding number”.

Remarks 14.1.5.
( )
1. 𝜎𝑚 = 𝑚𝛾1 is a cycle where 𝛾1 (𝑡) = exp 𝑖2𝜋𝑡 , 𝑡 ∈ [0, 1]; 𝑛(𝜎𝑚 ; 0) = 𝑚 ∈ ℤ ⧵ {0}.
2. 𝜎 = 𝛾𝑚 is a cycle.

14.1.6. Notice that an open set Ω ⧵ |𝜎| in the complex plane is the disjoint union of domains.
These distinct components are the maximal connected open sets.

Remark 14.1.7. Since |𝜎| is bounded, there exists 𝑅 > 0 such that

{𝑧 ∈ ℂ ∶ |𝑧| > 𝑅} ∩ |𝜎| = ∅.

This means that we have one unbounded component and the other components are in the disc
( )
𝔻 0, 𝑅 .

Lemma 14.1.8: The component lemma. Suppose that 𝜎 is a cycle in the complex plane. The
mapping 𝑎 ↦ 𝑛(𝜎; 𝑎) is a constant on each connected component of ℂ⧵|𝜎|. Moreover the number
𝑛(𝜎; 𝑎) = 0 in the unbounded component of ℂ ⧵ |𝜎|.

Proof. Homework.

14.2 Cauchy global integral theorem

Theorem 14.2.1: Cauchy global integral theorem. Suppose that 𝜎 is a cycle in the complex
plane, Ω is a domain and 𝑓 ∶ Ω → ℂ is an analytic function. If 𝜎 is a cycle in Ω such that
𝑛(𝜎; 𝑏) = 0 for all 𝑏 ∈ ℂ ⧵ Ω, then:

1 𝑓 (𝑢)
𝑛(𝜎; 𝑧)𝑓 (𝑧) = d𝑢 for all 𝑧 ∈ Ω ⧵ |𝜎| (14.2.2)
2𝜋𝑖 ∫𝜎 𝑢 − 𝑧

and
𝑓 (𝑧) d𝑧 = 0. (14.2.3)
∫𝜎

Remark. It is not difficult to see that the first equation 14.2.2 implies the second. Let 𝑧0 ∈ Ω ⧵|𝜎|
be arbitrary. Since 𝑓 is analytic in Ω, also the function 𝐺 ∶ Ω → ℂ, 𝐺(𝑧) = (𝑧 − 𝑧0 )𝑓 (𝑧) is
analytic in Ω. Thus we may apply equation 14.2.2 to it to get that

1 1 (𝑧 − 𝑧0 )𝑓 (𝑧)
𝑓 (𝑧) d𝑧 = d𝑧 = 𝑛(𝜎; 𝑧0 )𝐺(𝑧0 ) = 𝑛(𝜎; 𝑧0 )(𝑧0 − 𝑧0 )𝑓 (𝑧0 ) = 0.
2𝜋𝑖 ∫𝜎 2𝜋𝑖 ∫𝜎 𝑧 − 𝑧0

Note that because the integration variable 𝑧 ∈ |𝜎| is never equal to the point 𝑧0 ∈ Ω ⧵ |𝜎|,
multiplying and dividing by 𝑧 − 𝑧0 in the integral is justified.

Remark. If Ω = ℂ ⧵ {0}, then it is enough to require 𝑛(𝜎; 0) = 0.

Definition 14.2.4. A domain Ω in ℂ is simply connected if 𝑛(𝛾; 𝜔) = 0 for each closed curve 𝛾
in Ω and each 𝜔 ∈ ℂ ⧵ Ω. This means there are no holes in Ω.

RHS, version May 11, 2019 101

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

14.3 Three corollaries of Cauchy global integral theorem

Corollary 14.3.1. Suppose that Ω is a simply connected domain in ℂ and 𝑓 ∶ Ω → ℂ is an
analytic function. Then, every cycle 𝜎 in Ω satisfies the following
1 𝑓 (𝑢)
𝑛(𝜎; 𝑧)𝑓 (𝑧) = d𝑢 for all 𝑧 ∈ Ω ⧵ |𝜎| and 𝑓 (𝑧) d𝑧 = 0.
2𝜋𝑖 ∫𝜎 𝑢 − 𝑧 ∫𝜎

Proof. Since Ω is simply connected, 𝑛(𝜎; 𝑏) = 0 for all ℂ ⧵ Ω.

( )
Corollary 14.3.2. If Ω = 𝔻 𝑧0 , 𝑅 , then we recover the Cauchy local integral formula in the
( )
disc 𝔻 𝑧0 , 𝑟 , 𝑟 < 𝑅, then
1 𝑓 (𝑢) ( )
𝑓 (𝑧) = d𝑢 ∀ 𝑧 ∈ 𝔻 𝑧0 , 𝑟 .
∫
2𝜋𝑖 𝜕𝔻(𝑧0 ,𝑟) 𝑢 − 𝑧

Corollary 14.3.3: The case Ω = ℂ ⧵ {0}.

If 𝑓 ∶ ℂ ⧵ {0} → ℂ is analytic and 𝜎 is cycle in ℂ ⧵ {0} such that 𝑛(𝜎; 0) = 0, then
1 𝑓 (𝑢)
𝑛(𝜎; 𝑧)𝑓 (𝑧) = d𝑢
2𝜋𝑖 ∫𝜎 𝑢 − 𝑧
for all 𝑧 ∉ |𝜎| ∪ {0} and ∫𝜎 𝑓 (𝑢) d𝑢 = 0.

14.4 Examples
sin 𝑧
1. An important example. Let 𝛾(𝑡) = 3 exp(−𝑖𝑡), 𝑡 ∈ [0, 4𝜋]. Evaluate the integral ∫𝛾 4𝑧−𝜋
d𝑧.
(a) By Cauchy global integral formula, since 𝑓 (𝑧) = sin 𝑧 is entire, that is 𝑧 ↦ sin 𝑧 is
analytic in the whole ℂ and 𝛾 winds the point 𝜋4 twice clockwise.
(b) Also ℂ is simply connected, we have with 𝜎 = −2𝛾0 , where 𝛾0 (𝑡) = 3 exp(𝑖𝑡), 𝑡 ∈
[0, 2𝜋],
Now
√ √
sin 𝑧 1 sin 𝑧 1 𝜋 𝜋 1 2 𝜋 2
d𝑧 = d𝑧 = ⋅ 2𝜋𝑖 ⋅ 𝑛(𝜎; ) sin = ⋅ 2𝜋𝑖(−2) =− 𝑖.
∫𝛾 4𝑧 − 𝜋 4 ∫𝜎 𝑧 − 𝜋 4 4 4 4 2 2
4

Another way to solve the problem. By the definition of the integral along a cycle 𝜎 = −2𝛾0 ,
which equals to 𝛾, and where 𝛾0 (𝑡) = 3 exp(𝑖𝑡), 𝑡 ∈ [0, 2𝜋], and Cauchy local integral
formula,
√
sin 𝑧 1 sin 𝑧 ∗ 1 sin 𝑧 ∗∗ 1 2
d𝑧 = 𝜋 d𝑧 = (−2) 𝜋 d𝑧 = − ⋅ 2 sin 𝜋4 = − 𝜋𝑖.
∫𝜎 4𝑧 − 𝜋 4 ∫𝜎 𝑧 − 4 ∫𝛾0 𝑧 − 2 2
4 4

*The definition of the integral along the cycle 𝛾. **Cauchy local formula, 𝛾0 (𝑡) = exp(𝑖𝑡),
𝑡 ∈ [0, 2𝜋].
( ) ( )
2. Let Ω be a domain and let 𝔻 𝑧0 , 𝑅 ⊆ Ω. If 𝜎 = 8𝜕𝔻 𝑧0 , 𝑅 , where 𝛾0 (𝑡) = 𝑧0 +𝑅 exp(𝑖𝑡),
𝑡 ∈ [0, 2𝜋], then by the definition of the integral along 𝜎 and by the Cauchy local integration
formula,
1 𝑓 (𝑢) ∗ 1 𝑓 (𝑢) 1 𝑓 (𝑢)
d𝑢 = 8 d𝑢 = 8 d𝑢
∫
2𝜋𝑖 𝜎 𝑢 − 𝑧 ∫
2𝜋𝑖 𝜕𝔻(𝑧0 ,𝑅) 𝑢 − 𝑧 ∫
2𝜋𝑖 𝜕𝔻(𝑧0 ,𝑅) 𝑢 − 𝑧
∗∗ ( )
= 8𝑓 (𝑧), 𝑧 ∈ 𝔻 𝑧0 , 𝑅 .
*The definition of the integral along the cycle 𝛾. **Cauchy local formula.
Hence, in Cauchy global integral formula,
1 𝑓 (𝑢)
d𝑢 = 𝑛(𝜎; 𝑧)𝑓 (𝑧),
2𝜋𝑖 ∫𝜎 𝑢 − 𝑧
the winding number is needed!

RHS, version May 11, 2019 102

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

Remark. Let Ω = ℂ ⧵ {0}. The condition 𝑛(𝜎; 0) = 0 is necessary. Let 𝛾(𝑡) = exp(𝑖𝑡), 𝑡 ∈ [0, 2𝜋].
Then 𝑛(𝛾; 0) = 1 and ∫𝛾 d𝑧
𝑧
= 2𝜋 ≠ 0, although 𝑓 is analytic in ℂ ⧵ {0}.

14.5 Deformation theorem

Still one easy corollary of Cauchy global integration theorem. This corollary is very useful when
evaluating given integrals.

Theorem 14.5.1. Suppose that Ω is a domain in ℂ and 𝑓 ∶ Ω → ℂ is an analytic function. If 𝜎1

and 𝜎2 are cycles in Ω such that

𝑛(𝜎1 , 𝑏) = 𝑛(𝜎2 , 𝑏), for all 𝑏 ∈ ℂ ⧵ Ω, (14.5.2)

then
𝑓 (𝑢) d𝑢 = 𝑓 (𝑢) d𝑢. (14.5.3)
∫𝜎1 ∫𝜎2

Proof. Let
∑
𝑘 ∑
𝑙
𝜎1 = 𝜂 𝑗 𝛾𝑗 and 𝜎2 = 𝜇𝑗 𝜂𝑗 .
𝑗=1 𝑗=1

We write 𝜎 ∶= 𝜎1 − 𝜎2 . Then 𝑛(𝜎; 𝑏) = 𝑛(𝜎1 ; 𝑏) − 𝑛(𝜎2 , 𝑏) = 0 by (14.5.2) for all 𝑏 ∈ ℂ ⧵ Ω. By

Cauchy global integral theorem,

0= 𝑓 (𝑧) d𝑧 = 𝑓 (𝑢) d𝑢 − 𝑓 (𝑢) d𝑢.

∫𝜎 ∫𝜎1 ∫𝜎2

cos 𝑧
Example 14.5.4. Evaluate the integral ∫Γ 𝑧2 −2𝑧
d𝑧, where

Γ = 𝛾[−1−𝑖,1−𝑖] ∗ 𝛾[1−𝑖,1+𝑖] ∗ 𝛾[1+𝑖,−1+𝑖] ∗ 𝛾[−1+𝑖,−1−𝑖]

is the edge of the rectangle in ℂ.

• First method. The Cauchy global integral formula:
cos 𝑧
𝑧−2 ( )
d𝑧 = 2𝜋𝑖𝑓 0 = −𝜋𝑖,
∫Γ 𝑧
( )
as 𝑛 Γ; 2 = 0.
• Second method. By the deformation theorem and using two different cycles Γ and 𝛾:
1
𝛾(𝑡) = 2
exp(𝑖𝑡), 𝑡 ∈ [0, 2𝜋]
cos 𝑧
𝑓∶ 𝑧↦ 𝑧(𝑧−2)
is analytic in ℂ ⧵ {0, 2}

𝑛(Γ; 0) = 𝑛(𝛾; 0) = 1 and 𝑛(Γ; 2) = 𝑛(𝛾; 2) = 0

( )
𝑔 ∶ 𝑧 ↦ cos
𝑧−2
𝑧
is analytic in 𝔻 0, 3
2
and 𝑔(0) = − 21 .

Then: cos 𝑧
cos 𝑧 ∗ cos 𝑧 𝑧−2 ∗∗
d𝑧 = d𝑧 = d𝑧 = 2𝜋𝑖𝑔(0) = −𝜋𝑖.
∫Γ 𝑧(𝑧 − 2) ∫𝛾 𝑧(𝑧 − 2) ∫𝛾 𝑧
cos 𝑧 { }
*Deformation theorem. **Cauchy local formula, as 𝑔(𝑧) = 𝑧−2
is analytic in ℂ ⧵ 2 .

Example 14.5.5. Let 𝑧1 , 𝑧2 ∈ ℂ and 𝑟 > 0, such that ||𝑧1 || < 𝑟 < ||𝑧2 ||. Let 𝜕 = 7𝛾 be a cycle
where 𝛾 ∶ [0, 2𝜋] ↦ ℂ and 𝛾(𝑡) = 𝑟 exp(𝑖𝑡). Show that

d𝑧 14𝜋𝑖
= .
∫𝜕 (𝑧 − 𝑧1 )(𝑧 − 𝑧2 ) 𝑧1 − 𝑧2

RHS, version May 11, 2019 103

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

1
Let us define 𝑓 ∶ ℂ ⧵ 𝑧2 ↦ ℂ, 𝑓 (𝑧) = 𝑧−𝑧2
. Then 𝑓 is analytic in ℂ ⧵ 𝑧2 . Now 𝑛(𝜕; 𝑧2 ) = 0. Then

1
d𝑧 𝑧−𝑧2
= d = 2𝜋𝑖𝑛(𝜕; 𝑧1 )𝑓 (𝑧1 )
∫𝜕 (𝑧 − 𝑧1 )(𝑧 − 𝑧2 ) ∫𝜕 𝑧 − 𝑧1
1 14𝜋𝑖
= 2𝜋𝑖 ⋅ 7 ⋅ = .
𝑧1 − 𝑧2 𝑧1 − 𝑧2

14.6 Integrating rational functions on the real line

Recall that if the limits
𝑎 0
lim 𝑓 (𝑥) d𝑥 and lim 𝑓 (𝑥) d𝑥
𝑎→∞ ∫0 𝑏→−∞ ∫𝑏

exist, then
∞ 0 𝑎
𝑓 (𝑥) d𝑥 = lim 𝑓 (𝑥) d𝑥 + lim 𝑓 (𝑥) d𝑥, (14.6.1)
∫−∞ 𝑏→−∞ ∫𝑏 𝑎→∞ ∫0

and the integral

𝑅
lim 𝑓 (𝑥) d𝑥 (14.6.2)
𝑅→∞ ∫−𝑅

exists and is equal to 14.6.1. On the other hand, the existence of 14.6.2 does not imply the exis-
tence of 14.6.1.

Example. Let 𝑅 > 0 and 𝑓 (𝑥) = 𝑥. Then

/
𝑅
𝑅
1
𝑥 d𝑥 = 𝑥2 = 0.
∫−𝑅 2
−𝑅

∞
But the integral ∫−∞ 𝑥 d𝑥 does not exist:
𝑎 0
lim 𝑥 d𝑥 + lim 𝑥 d𝑥 = ∞ − ∞.
𝑎→∞ ∫0 𝑏→∞ ∫𝑏

𝑃 (𝑧)
Remark. Let 𝑓 (𝑧) = 𝑄(𝑧)
, where 𝑃 and 𝑄 are polynomials such that deg 𝑄 ≥ 2 + deg 𝑃 and 𝑄
∞
does not have real roots. Then ∫−∞ 𝑓 (𝑥) d𝑥 exists and we can integrate
∞
𝑃 (𝑧)
d𝑥
∫−∞ 𝑄(𝑧)

with the help of complex analysis.

Example. Evaluate the integral

∞
d𝑥
.
∫−∞ 𝑥2 − 2𝑥 + 5
Note that 𝑥2 − 2𝑥 + 5 does not have real roots. Because this integral exists, and we can write
∞ 𝑅
d𝑥 d𝑥
= lim .
∫−∞ 𝑥2 − 2𝑥 + 5 𝑅→∞ ∫−𝑅 𝑥2 − 2𝑥 + 5
Assume that 𝑅 ≫ 10. Let us write
1 1
𝑓 (𝑧) = =( ( ) )( ( )) ;
𝑧2 − 2𝑧 + 5 𝑧 − 1 + 2𝑖 𝑧 − 1 − 2𝑖

RHS, version May 11, 2019 104

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

𝑓 is analytic in ℂ ⧵ {1 + 2𝑖, 1 − 2𝑖}. We may write

∗
𝑓 (𝑧) d𝑧 + 𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧.
∫𝛾[−𝑅,𝑅] ∫𝐶𝑅 ∫𝛾[−𝑅,𝑅] ∗𝐶𝑅

*Definition of integration and using the Deformation theorem and the Estimation lemma for ∫𝐶 .
𝑅
( )
Let 𝛾(𝑡) = 1 + 2𝑖 + exp 𝑖2𝜋𝑡 , 𝑡 ∈ [0, 1]. By the Deformation theorem

𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧,

∫𝛾[−𝑅,𝑅] ∗𝐶𝑅 ∫𝛾

because 𝑓 is analytic in ℂ ⧵ {1 + 2𝑖, 1 − 2𝑖} = Ω and 𝛾[−𝑅,𝑅] and 𝛾 are cycles in Ω and
( ) ( )
𝑛 𝛾[−𝑅,𝑅] ∗ 𝐶𝑅 ; 1 + 2𝑖 = 1 = 𝑛 𝛾; 1 + 2𝑖 and
( ) ( )
𝑛 𝛾[−𝑅,𝑅] ∗ 𝐶𝑅 ; 1 − 2𝑖 = 0 = 𝑛 𝛾; 1 − 2𝑖 .

By the Cauchy global integral formula we get

1
𝑧−(1−2𝑖)
( )
( ) d𝑧 = 2𝜋𝑖𝑛 𝛾[−𝑅,𝑅] ∗ 𝐶𝑅 ; 1 + 2𝑖 𝑔(1 + 2𝑖)
∫𝛾[−𝑅,𝑅] ∗𝐶𝑅 𝑧 − 1 + 2𝑖
1 𝜋
= 2𝜋𝑖 ( )= ,
1 + 2𝑖 − 1 − 2𝑖 2
1
where 𝑔 ∶ 𝑧 ↦ 𝑧−(1−2𝑖)
is an analytic function in the simply connected domain
] [ ] ( )[
Ω = −𝑅 − 1, 𝑅 + 1 × −𝑖, 𝑖 𝑅 + 1 .

By the estimation lemma 9.3.12

| |
| | 1 𝜋
| 𝑓 (𝑧) d𝑧| ≤ 𝜋𝑅 = → 0, as 𝑅 → ∞.
|∫𝐶 | 2
𝑅 − 2𝑅 + 5 𝑟 − 2 5
| 𝑅 | 𝑅
Hence,
𝜋
= 𝑓 (𝑧) d𝑧 = 𝑓 (𝑧) d𝑧 + 𝑓 (𝑧) d𝑧,
2 ∫𝛾[−𝑅,𝑅] ∗𝐶𝑅 ∫𝛾[−𝑅,𝑅] ∫𝐶𝑅
where the last term approaches to zero when 𝑅 approaches to infinity thus
𝑅 ∞
𝜋 d𝑥 𝜋
lim 𝑓 (𝑧) d𝑧 = and = .
𝑅→∞ ∫−𝑅 2 ∫−∞ 𝑥2 − 2𝑥 + 5 2

14.7 The proof for the Cauchy global integral formula

The proof given here is based on a proof by Dixon in [10]. The proof of Theorem 14.2.1, more
specifically equation (14.2.2), proceeds as follows:
1. Define 𝐹 ∶ Ω × Ω → ℂ by
𝑓 (𝑧) − 𝑓 (𝜉)
𝐹 (𝑧, 𝜉) = when 𝜉 ≠ 𝑧 (14.7.1)
𝑧−𝜉
and 𝐹 (𝑧, 𝜉) = 𝑓 ′ (𝜉) = 𝑓 ′ (𝑧) when 𝜉 = 𝑧. We will prove that 𝐹 is continuous.
2. Define ℎ ∶ Ω → ℂ by
1
ℎ(𝑧) = 𝐹 (𝑧, 𝑢) d𝑢. (14.7.2)
2𝜋𝑖 ∫𝜎
We will prove that ℎ is continuous and analytic in Ω. Notice that for all 𝑧 ∈ Ω ⧵ |𝜎|, we
have
1 1 𝑓 (𝑢)
ℎ(𝑧) = 𝐹 (𝑧, 𝑢) d𝑢 = −𝑓 (𝑧)𝑛(𝜎; 𝑧) + d𝑢, (14.7.3)
2𝜋𝑖 ∫𝜎 2𝜋𝑖 ∫𝜎 𝑢 − 𝑧
so our aim is to prove that ℎ(𝑧) = 0. This is done via Liouville’s theorem, for which we
need to extend ℎ to all of ℂ.

RHS, version May 11, 2019 105

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

3. Define 𝑔 ∶ ℂ ⧵ |𝜎| → ℂ by
1 𝑓 (𝑢)
𝑔(𝑧) = d𝑢. (14.7.4)
∫
2𝜋𝑖 𝜎 𝑢 − 𝑧
We will show that 𝑔 is analytic in ℂ ⧵|𝜎|.
4. Putting ℎ and 𝑔 together, define 𝐻 ∶ ℂ → ℂ by
{
ℎ(𝑧), 𝑧 ∈ Ω,
𝐻(𝑧) = (14.7.5)
𝑔(𝑧), 𝑧 ∈ ℂ ⧵ |𝜎|, 𝑛(𝜎; 𝑧) = 0.

We show that that 𝐻 is well-defined and analytic in all of ℂ, so it is our desired extension
of ℎ.
5. Lastly, we see that 𝐻 is bounded, so it is constant by Liouville’s theorem. The constant
value is seen to be 0, and so the claim follows.
To properly begin the proof, define 𝐹 as in equation (14.7.1). Outside of the diagonal
{ }
(𝜔, 𝜔) ∶ 𝜔 ∈ Ω ⊆ Ω × Ω,

the function 𝐹 is clearly continuous. Thus it remains to verify continuity on the diagonal: let
𝜔 ∈ Ω and 𝜀 > 0 be fixed.
Since the function 𝑓 is analytic in Ω, it is infinitely differentiable in Ω by Cauchy’s second formula
12.0.3. In particular, 𝑓 ′ is continuous in Ω, so there exists a 𝛿 > 0 so that |𝑓 ′ (𝑢) − 𝑓 ′ (𝑣)| < 𝜀
whenever 𝑢, 𝑣 ∈ 𝔻(𝜔, 𝛿).
Take any two points 𝑧, 𝜉 ∈ 𝔻(𝜔, 𝛿). If 𝜉 = 𝑧 we immediately have

|𝐹 (𝑧, 𝜉) − 𝐹 (𝜔, 𝜔)| = ||𝑓 ′ (𝑧) − 𝑓 ′ (𝜔)|| < 𝜀

| | | |
which verifies continuity. Suppose then that 𝜉 ≠ 𝑧 and let 𝛾[𝑧,𝜉] ∶ [0, 1] → ℂ be the straight line
𝑡 ↦ 𝜉 + 𝑡(𝑧 − 𝜉) from 𝑧 to 𝜉, which has a constant derivative 𝛾[𝑧,𝜉]′ ≡ 𝑧 − 𝜉. Notice that since the
disc 𝔻(𝜔, 𝛿) is convex, |𝛾[𝑧,𝜉] | is entirely contained in the disc. By the Fundamental Theorem of
Calculus, we get that
1
𝑓 (𝑧) − 𝑓 (𝜉) = 𝑓 ′ (𝑢) d𝑢 = (𝑧 − 𝜉) 𝑓 ′ (𝜉 + 𝑡(𝑧 − 𝜉)) d𝑡.
∫𝛾[𝑧,𝜉] ∫0

This allows us to estimate

| |
|𝐹 (𝑧, 𝜉) − 𝐹 (𝜔, 𝜔)| = || 𝑓 (𝑧) − 𝑓 (𝜉) − 𝑓 ′ (𝜔)||
| | | 𝑧−𝜉 |
| |
| 1 1 |
| |
=| 𝑓 ′ (𝜉 + 𝑡(𝑧 − 𝜉)) d𝑡 − 𝑓 ′ (𝜔) d𝑡|
|∫0 ∫0 |
| |
1
| ′ |
≤ |𝑓 (𝜉 + 𝑡(𝑧 − 𝜉)) − 𝑓 ′ (𝜔)| d𝑡
∫0 | |
1
≤ 𝜀 d𝑡
∫0
= 𝜀,

where we use the fact that 𝜔, 𝜉 + 𝑡(𝑧 − 𝜉) ∈ 𝔻(𝜔, 𝛿) for all 0 ≤ 𝑡 ≤ 1. Thus 𝐹 is continuous.

For the second step, define ℎ as in equation (14.7.2). Since 𝐹 is continuous, the integral exists
and ℎ is well defined. To see that ℎ is continuous, let 𝜔 ∈ Ω and let 𝜔𝑛 → 𝜔 be a sequence such
that 𝜔𝑛 ∈ Ω for all 𝑛.
Since the set 𝔻(𝜔, 𝛿) ×|𝜎| is compact (as it is closed and bounded), the restriction of 𝐹 to that
set is uniformly continuous. Given any 𝜉 ∈ |𝜎|, we have that (𝜔𝑛 , 𝜉) → (𝜔, 𝜉), so by uniform
continuity of 𝐹 we have that 𝐹 (𝜔𝑛 , 𝜉) → 𝐹 (𝜔, 𝜉) uniformly with respect to 𝜉 ∈ |𝜎|. To be more

RHS, version May 11, 2019 106

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

specific, this means that the sequence of functions 𝜉 ↦ 𝐹 (𝜔𝑛 , 𝜉) converges uniformly to the
function 𝜉 ↦ 𝐹 (𝜔, 𝜉).
Since integration commutes with uniform convergence, we get that

1 1
ℎ(𝜔𝑛 ) = 𝐹 (𝜔𝑛 , 𝑢) d𝑢 → 𝐹 (𝜔, 𝑢) d𝑢 = ℎ(𝜔),
∫
2𝜋𝑖 𝜎 2𝜋𝑖 ∫𝜎

so ℎ is indeed continuous. Verifying equation (14.7.3) takes just a calculation: given a 𝑧 ∈ ℂ⧵|𝜎|,
𝑧 is never equal to the integration variable 𝑢 ∈ |𝜎|, so

1
ℎ(𝑧) = 𝐹 (𝑧, 𝑢) d𝑢
2𝜋𝑖 ∫𝜎
1 𝑓 (𝑧) − 𝑓 (𝑢)
= d𝑢
2𝜋𝑖 ∫𝜎 𝑧−𝑢
1 𝑓 (𝑧) 1 𝑓 (𝑢)
= d𝑢 − d𝑢
2𝜋𝑖 ∫𝜎 𝑧 − 𝑢 2𝜋𝑖 ∫𝜎 𝑧 − 𝑢
1 d𝑢 1 𝑓 (𝑢)
= −𝑓 (𝑧) + d𝑢
2𝜋𝑖 ∫𝜎 𝑢 − 𝑧 2𝜋𝑖 ∫𝜎 𝑢 − 𝑧
𝑓 (𝑢)
= −𝑓 (𝑧) 𝑛(𝜎; 𝑧) + d𝑢,
∫𝜎 𝑢 − 𝑧

as claimed. Thus, if we show that ℎ ≡ 0, it follows that

𝑓 (𝑢)
𝑛(𝜎; 𝑧)𝑓 (𝑧) = d𝑢,
∫𝜎 𝑢 − 𝑧

which is the claim of the theorem.

Next, we wish to show that ℎ is analytic, not just continuous. For this, fix 𝜔 ∈ Ω and fix a 𝛿 > 0
so that 𝔻(𝜔, 𝛿) ⊆ Ω. We will show that 𝑓 is analytic in 𝔻(𝜔, 𝛿) by Morera’s theorem 13.0.2.
Let 𝛾 be a closed contour in 𝔻(𝜔, 𝛿). Now, since 𝑧 ↦ 𝐹 (𝑧, 𝜉) is analytic in 𝔻(𝛾, 𝜔) ⧵ {𝜉} and
continuous in the whole disc for any 𝜉 ∈ Ω, we have that

𝐹 (𝑧, 𝜉) d𝑥 = 0
∫𝛾

by (a slightly strengthened version of) the Cauchy–Goursat Theorem 11.0.5. For the function ℎ,
we then get that
( ) ( )
ℎ(𝑧) d𝑧 = 𝐹 (𝑧, 𝜉) d𝜉 d𝑧 = 𝐹 (𝑧, 𝜉) d𝑧 d𝜉 = 0 d𝜉 = 0.
∫𝛾 ∫𝛾 ∫𝜎 ∫𝜎 ∫𝛾 ∫𝜎

Since the integral of ℎ over any closed contour in 𝔻(𝜔, 𝛿) is 0, ℎ is analytic in the disc (and
especially at 𝜔) by Morera’s theorem 13.0.2. Since 𝜔 ∈ Ω was arbitrary, ℎ is analytic in Ω.
(Note: it is true in general that if 𝜎 and 𝛾 are continuously differentiable parametrized curves
from closed, bounded intervals to ℝ2 and 𝐹 is a continuous function of two real variables, then
the order of integration over the paths can be exchanged, i.e.

𝐹 (𝑥, 𝑦) d𝑥 d𝑦 = 𝐹 (𝑥, 𝑦) d𝑦 d𝑥.

∫𝜎 ∫𝛾 ∫𝛾 ∫𝜎

This follows from the theory of iterated integrals, by expanding both path integrals into usual
integrals over closed intervals.)

The next step is to define 𝑔 as in equation (14.7.4) and to show it is analytic. Notice that since
𝑧 ∈ ℂ ⧵|𝜎| and 𝜉 ∈ |𝜎|, the denominator in the integrand never vanishes and so 𝑔 is well defined.
We are going to show that 𝑔 is analytic by finding a power series representation for it.

RHS, version May 11, 2019 107

Complex Analysis I 14 GLOBAL CAUCHY THEOREM

( )
Fix a 𝑧0 ∈ ℂ ⧵|𝜎| to serve as the center, and let 𝛿 > 0 be such that 𝔻 𝑧0 , 2𝛿 ⊆ ℂ ⧵|𝜎|. Also, for
( )
the time being, fix a 𝑧 ∈ 𝔻 𝑧0 , 𝛿 .

Consider then a 𝜉 ∈ |𝜎|. By the above, we have ||𝜉 − 𝑧0 || ≥ 2𝛿 > 2||𝑧 − 𝑧0 ||. We can expand 1
𝜉−𝑧
into a geometric sum as follows:
1 1 1 1 ∑ (𝑧 − 𝑧0 )𝑛
∞
= = = ,
𝜉 − 𝑧 𝜉 − 𝑧0 − (𝑧 − 𝑧0 ) 𝜉 − 𝑧0 1 − 𝑧−𝑧0 𝑛=0 (𝜉 − 𝑧0 )
𝑛+1
𝜉−𝑧0
| 𝑧−𝑧 | |𝑧−𝑧 |
and the sum converges uniformly regardless of 𝜉 ∈ |𝜎|, since || 𝜉−𝑧0 || ≤ 2 𝑧−𝑧0 = 1
which holds
| 0 | | 0| 2
because of how we chose 𝑧. Now we can write the function 𝑔 as
1 𝑓 (𝜉)
𝑔(𝑧) = d𝜉
2𝜋𝑖 ∫𝜎 𝜉 − 𝑧
1 ∑∞
(𝑧 − 𝑧0 )𝑛
= 𝑓 (𝜉) d𝜉
2𝜋𝑖 ∫𝜎 𝑛=0 (𝜉 − 𝑧0 )
𝑛+1
( )
∑∞
1 𝑓 (𝜉)
= d𝜉 (𝑧 − 𝑧0 )𝑛 .
𝑛=0
2𝜋𝑖 ∫𝜎 (𝜉 − 𝑧0 )𝑛+1
Exchanging the order of integration and summation is justified, since the sum converges uniformly
with respect to 𝜉. Since the power series for 𝑔 above converges, ( as is) seen by reading the above
chain of equations backwards, we have that 𝑔 is analytic in 𝔻 𝑧0 , 𝛿 , and as 𝑧0 ∈ ℂ ⧵ |𝜎| was
arbitrary, 𝑔 is analytic in all of ℂ ⧵|𝜎|.
{ }
The fourth step is to define 𝐻 as in equation (14.7.5). Let Υ = 𝑧 ∈ ℂ ⧵|𝜎| ∶ 𝑛(𝜎; 𝑧) = 0 , that
is, Υ is the set in the lower condition in the definition of 𝐻.
Suppose thus that 𝑧 ∈ Ω ∩ Υ, so 𝑧 ∈ Ω, 𝑧 ∈ ℂ ⧵ |𝜎| and 𝑛(𝜎; 𝑧) = 0. It is convenient to use
equation (14.7.3) for the value of ℎ: we get that
1 𝑓 (𝑢) 𝑓 (𝑢)
ℎ(𝑧) = −𝑓 (𝑧)𝑛(𝜎; 𝑧) + d𝑢 = 0 + d𝑢 = 𝑔(𝑧),
2𝜋𝑖 ∫𝜎 𝑢 − 𝑧 ∫𝜎 𝑢 − 𝑧
so 𝐻 is indeed well defined. We see that the domain of 𝐻 is all of ℂ by virtue of the assumption
that 𝑛(𝜎; 𝑏) = 0 for all 𝑏 ∈ ℂ ⧵ Ω. Since both 𝑔 and ℎ are analytic on their domains of definition,
which are both open, also the function 𝐻 is analytic.

The last step is to see that

( the)entire function 𝐻 is bounded. Towards that goal, let 𝑅 > 0 be large
enough so that|𝜎| ⊆ 𝔻 0, 𝑅 . By the component lemma 14.1.8, it holds that 𝑛(𝜎; 𝑧) = 0 when 𝑧
( )
is in the unbounded component 𝑈 of ℂ ⧵|𝜎|, and the complement of the disc 𝔻 0, 𝑅 is a subset
of 𝑈 .
{ }
Also, let 𝑀 = sup ||𝑓 (𝜉)|| ∶ 𝜉 ∈ |𝜎| : this supremum exists, since|𝜎| is compact, and a continu-
( )
ous function obtains a maximum in a compact set. Lastly, for any 𝑧 ∉ 𝔻 0, 𝑅 and 𝜉 ∈ |𝜎|, we
have the following estimate from the reverse triangle inequality:
|𝑧 − 𝜉| ≥ |𝑧| −|𝜉| ≥ |𝑧| − 𝑅 ≥ 0.
( )
Now, suppose that 𝑧 ∈ ℂ ⧵ 𝔻 0, 𝑅 , that is|𝑧| > 𝑅. Then 𝑧 is in the unbounded component 𝑈 , so
𝑛(𝜎; 𝑧) = 0, and 𝑧 ∉ |𝜎|, so the second piecewise definition of 𝐻 applies. Using the estimation
lemma 9.3.12, we get the following upper bound:
| 𝑓 (𝜉) || | 𝑓 (𝜉) |
|𝐻(𝑧)| = || 1 1 | | length(𝜎) 𝑀
| | | 2𝜋𝑖 ∫ 𝜉 − 𝑧 d𝜉 || ≤ 2𝜋 ⋅ sup || 𝜉 − 𝑧 || ⋅ length(𝜎) ≤ 2𝜋 |𝑧| − 𝑅
.
| 𝜎 | 𝜉∈|𝜎|| |
( )
Since the last bound goes to 0 as|𝑧| → ∞, we have that 𝐻 is bounded outside of the disc 𝔻 0, 𝑅 .
( )
But since 𝐻 is analytic and thus continuous, and the disc 𝔻 0, 𝑅 is compact, 𝐻 attains a maxi-
mum there, so 𝐻 is bounded within the disc.
Thus 𝐻 is bounded, so by Liouville’s theorem 13.0.6, 𝐻 is constant. But since ||𝐻(𝑧)|| → 0 as
|𝑧| → ∞, it must be that 𝐻 ≡ 0. In particular, ℎ ≡ 0 in Ω, which concludes the proof.

RHS, version May 11, 2019 108

Complex Analysis I 15 THE EXTENDED COMPLEX PLANE

15 The extended complex plane

15.1 The Riemann sphere and the extended complex plane
Let us consider ℂ as embedded in the Euclidean 3-space ℝ3 by identifying 𝑥 + 𝑖𝑦 with (𝑥, 𝑦, 0).
Let { ( )2 ( )2 }
3 2 2 1 1
𝑆 = (𝑥, 𝑦, 𝑧) ∈ ℝ ∶ 𝑥 + 𝑦 + 𝑧 − = .
2 2
This is a sphere which touches the plane ℂ at the point (0, 0, 0) ∈ 𝑆 as shown in Figure 15.1.

𝑧 ℝ3

𝑆 𝑁 = (0, 0, 1)

𝑃′
ℂ
(0, 0, 0)

𝑃 = 𝑥 + 𝑦𝑖
𝑦

Figure 15.1: The Riemann sphere 𝑆 and the complex plane ℂ embedded in the Euclidean 3-space
ℝ3 . A line segment between points 𝑁 ∈ 𝑆 and 𝑃 ∈ ℂ intersects the set 𝑆 ⧵ {𝑁} at exactly one
point 𝑃 ′ . This way a one-to-one correspondence between the complex plane and the Riemann
sphere (the extended complex plane ℂ) can be constructed.

Stereographic projection allows us to set up a one-to-one correspondence between points in ℂ

and the points of 𝑆, excluding the north pole of 𝑆, namely the point 𝑁 = (0, 0, 1).
The line from any point 𝑃 ∈ ℂ to 𝑁 intersects the set 𝑆 ⧵ {𝑁} at exactly one point 𝑃 ′ , and for
every point 𝑃 ′ in 𝑆 ⧵ {𝑁}, the line through 𝑁 and 𝑃 ′ meets the plane ℂ in a unique point 𝑃 . We
add to ℂ an extra point ∞ ∉ ℂ and we define the extended complex plane ℂ∪{∞} = ℂ = ℂ ̃ = ℂ.
̇

We have a correspondence between ℂ and 𝑆 given by

( )
′ 𝑥 𝑦 𝑟2
𝑃 = 𝑥 + 𝑖𝑦 = 𝑟 exp(𝑖𝜙) ↔ 𝑃 = , , ,
1 + 𝑟2 1 + 𝑟2 1 + 𝑟2
( )
∞ ↔ 0, 0, 1 .

Remark. If we let 𝑃 = 𝑟 exp(𝑖𝜙) with 𝜙 fixed, and allow 𝑟 to become arbitrarily large, then 𝑃 ′
will approach 𝑁 regardless of the value of 𝜙.

Remark. We adopt the following conventions when working with ℂ ∶

𝑎 ± ∞ = ±∞ + 𝑎 = ∞, for all 𝑎 ∈ ℂ,

RHS, version May 11, 2019 109

Complex Analysis I 15 THE EXTENDED COMPLEX PLANE

𝑎
= 0, for all 𝑎 ∈ ℂ,
∞
𝑎 ⋅ ∞ = ∞ ⋅ 𝑎 = ∞, for all 𝑎 ∈ ℂ ⧵ {0},
𝑎
= ∞, for all 𝑎 ∈ ℂ ⧵ {0},
0
∞ + ∞ = ∞ ⋅ ∞ = ∞ = ∞.

∞
Remark. The operations ∞ − ∞, 0 ⋅ ∞ and ∞
are not defined.

15.2 Circlines
The benefits of moving to ℂ become clear when we consider lines and circles. Now we can treat
lines and circles in a unified way.

15.3 The extended complex plane

15.3.1 Behavior of sets at ∞
Let us define
𝔻(∞, 𝑟) = {𝑧 ∈ ℂ ∶ |𝑧| > 𝑟} ∪ {∞}, 𝑟 > 0.
The set 𝔻(∞, 𝑟) is a “disc” centered on ∞ with radius 𝑟 > 0. Using 𝔻(∞, 𝑟), we can define a
topology in ℂ. Now Ω ⊂ ℂ is an open set, if for every 𝑎 ∈ Ω there exists 𝑟(𝑎) > 0, such that
𝔻(𝑎, 𝑟) ⊂ Ω.

Remark. Let (𝑧𝑛 ) be a sequence in ℂ. Then 𝑧𝑛 → ∞, if and only if, for each 𝑟 > 0, there exists
𝑁𝑟 such that
𝑧𝑛 ∈ 𝔻(∞, 𝑟) as soon as 𝑛 > 𝑁𝑟 .
This means ||𝑧𝑛 || > 𝑟 or 𝑧𝑛 = ∞.

15.4 Behavior of functions at ∞

Let 𝑎, 𝑏 ∈ ℂ. Now
lim 𝑓 (𝑧) = 𝑏
𝑧→𝑎

if and only if for each 𝑟 > 0, there exists 𝑠 > 0 such that 𝑓 (𝑧) ∈ 𝔻(𝑏, 𝑟) when 𝑧 ∈ 𝔻(𝑎, 𝑠) ⧵ {𝑎}.
1
We use the inverse map 𝑧 ↦ 𝑧
to analyze what happens at ∞ or near ∞. We extend this mapping
1 1
to ℂ → ℂ by recalling that ∞
= 0 and 0
= ∞.

Remark 15.4.1. Let 𝑎, 𝑏 ∈ ℂ. Since

( )
1 1
𝑧 ∈ 𝔻(∞, 𝑟) if and only if ∈ 𝔻 0, ,
𝑧 𝑟

we have
( )
1
lim 𝑓 (𝑧) = 𝑏 if and only if lim 𝑓 =𝑏
𝑧→∞ 𝑧→0 𝑧
1
lim 𝑓 (𝑧) = ∞ if and only if lim =0
𝑧→𝑎 𝑧→𝑎 𝑓 (𝑧)
1
lim 𝑓 (𝑧) = ∞ if and only if lim ( ) = 0.
𝑧→∞ 𝑧→∞
𝑓 1𝑧

Proof. “⇒”: Let 𝜀 > 0 be given. There exists 𝑅 > 0 such that

|𝑓 (𝑧)| ≥ 1 as soon as |𝑧| > 𝑅.

| | 𝜀

RHS, version May 11, 2019 110

Complex Analysis I 15 THE EXTENDED COMPLEX PLANE

| |
If|𝑧| = ||𝑧 − 0|| < 1
, then | 1𝑧 | > 𝑅 and therefore
𝑅 | |
| |
| |
| 1 |
| ( ) | = (1 ) ≤ 1 = 𝜀.
| |
| 𝑓 1 | || | 1
| | |𝑓 1 || 𝜀
| 𝑧 | | 𝑧 ||
|
⏟⏞⏟⏞⏟
≥ 1𝜀

“⇐”: Let 𝑅 > 0 be given. There exists 𝛿𝑟 such that

| |
| |
| 1 | 1
| ( )| < as |𝑧| < 𝛿𝑟 .
| |
|𝑓 1 | 𝑅
| |
| 𝑧 |

1 1
If|𝑧| > 𝛿𝑟
, then 𝛿𝑟 > |𝑧| . Then we have

|𝑓 (𝑧)| = 1 > 1 = 𝑅.
| | | | 1
| 1 |
| ( )| 𝑅
|𝑓 1 |
| 𝑧 |
| |

( ) Suppose that 𝑓 is defined in 𝔻(∞, 𝑟) and its values are in ℂ. Then 𝑓 is analytic
Definition 15.4.2.
at ∞ if 𝑧 ↦ 𝑓 1𝑧 is analytic at the origin.

( )
Definition 15.4.3. The case 𝑓 (𝑧) = ∞ for some 𝑧 ∈ ℂ. Suppose that 𝑓 is defined in 𝔻 𝑧0 , 𝑟
and 𝑓 (𝑧0 ) = ∞.
1
1. If 𝑧0 ∈ ℂ, then 𝑓 is meromorphic at 𝑧0 , if the function 𝑧 ↦ 𝑓 (𝑧)
is analytic at 𝑧0 .
1
2. If 𝑧0 = ∞, then 𝑓 is meromorphic at 𝑧0 = ∞, if the function 𝑧 ↦ ( ) is analytic at 0.
𝑓 1𝑧

The point 𝑧0 in both cases is called the pole.

Definition 15.4.4. Let Ω ⊂ ℂ be open and 𝑓 ∶ Ω → ℂ. Then 𝑓 is meromorphic in Ω, if 𝑓 is

meromorphic or analytic at each point of Ω.

Remarks.
1. When we consider analytic functions, we assume that their values are in ℂ. It might happen
that ∞ is in the domain/set where the function is defined.
2. When we consider meromorphic functions, we allow the functions to have poles. The
values of the functions are in ℂ.

Examples.
𝑧
1. Let 𝑓 (𝑧) = 𝑧2 −4
.

The function 𝑓 is analytic in ℂ ⧵ {±2}. We extend 𝑓 as a continuous function to ℂ → ℂ

by defining 𝑓 (2) = ∞, 𝑓 (−2) = ∞ and 𝑓 (∞) = 0. Now
( ) 1
1 𝑧2 𝑧
𝑓 = ( )𝑧 =( ) = =∶ 𝑔(𝑧).
𝑧 2 1 − 4𝑧 𝑧 1 − 4𝑧2
2
1
𝑧
−4

RHS, version May 11, 2019 111

Complex Analysis I 15 THE EXTENDED COMPLEX PLANE

The function 𝑔(𝑧) is analytic at 0. Thus, 𝑓 is also analytic at ∞. What about +2 and −2?
Now
1 𝑧2 − 4
= =∶ ℎ(𝑧),
𝑓 (𝑧) 𝑧
which is analytic at +2 and −2. These points are poles and thus 𝑓 is meromorphic at them.
2. Let 𝑓 (𝑧) = 𝑧2 + 𝑧 and 𝑓 (∞) = ∞. Now

1 1 𝑧2
( ) = ( )2 = =∶ 𝑔(𝑧).
𝑓 1 1 1 1+𝑧
𝑧 𝑧
+ 𝑧

The function 𝑔(𝑧) is analytic at 0 and thus 𝑓 is meromorphic at ∞.

1
More generally, set 𝑃 (𝑧) = 𝑎0 + 𝑎1 𝑧 + ⋯ + 𝑎𝑛 𝑧𝑛 , 𝑛 ≥ 1 and 𝑃 (∞) = ∞. Since ( ) is analytic
𝑃 1𝑧
at 0, 𝑃 is meromorphic at ∞.

RHS, version May 11, 2019 112

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

16 Möbius transformations
The extended complex plane is the right setting for studying Möbius transformations.

𝑎𝑧+𝑏
Definition. Let 𝑓 ∶ ℂ → ℂ, 𝑓 (𝑧) = 𝑐𝑧+𝑑 , 𝑎, 𝑏, 𝑐, 𝑑 ∈ ℂ, 𝑎𝑑 − 𝑏𝑐 ≠ 0.
( )
If 𝑐 ≠ 0, then 𝑓 −𝑑
𝑐
= ∞ and 𝑓 (∞) = 𝑎𝑐 .

If 𝑐 = 0, then 𝑓 (∞) = ∞.
Then, 𝑓 is continuous, ℂ → ℂ, and meromorphic. 𝑓 is called a Möbius transformation of the
complex plane.

Examples 16.0.1. Let 𝛼, 𝛽 be given complex numbers and 𝑘 and 𝑡0 real-valued constants. The
following are Möbius transformations:
1. Translation by 𝛽: 𝜔 = 𝑧 + 𝛽.
2. Rotation through 𝑡0 : 𝜔 = 𝑧(exp(𝑖𝑡0 )). If 𝑡0 > 0, the rotation is counterclockwise. If 𝑡0 < 0,
the rotation is clockwise.
( )
(a) 𝑧 ↦ exp −𝑖 𝜋2 𝑧 = −𝑖𝑧 is a rotation clockwise by 𝜋2 .

3. Dilation by a factor 𝑘 > 0: 𝜔 = 𝑘𝑧. If 𝑘 > 1, it is said to be stretching, and if 0 < 𝑘 < 1, it
is said to be shrinking.
( ( ))
(a) 𝑓 (𝑧) = 2𝑖𝑧 = 2 exp 𝑖 𝜋2 𝑧. Now 𝑓 is a stretching by 2 and a counterclockwise
rotation by 𝜋2 .
√ ( ( 𝜋 )) √
(b) 𝑓 (𝑧) = (1 + 𝑖)𝑧 − 𝑖 = 2 exp 𝑖 4 𝑧 − 𝑖. Now 𝑓 is a stretching by 2, a
𝜋
counterclockwise rotation by 4
and a translation by −𝑖.

4. Inversion: 𝜔 = 1𝑧 .

Remarks.
1 𝑧
1. = , 𝑧 ≠ 0.
𝑧 |𝑧|2
1 1
2. In polar coordinates: 𝑧 = 𝑟 exp(𝑖𝜃). Now 𝑧
= 𝑟 exp(𝑖𝜃)
= 𝑟−1 exp(−𝑖𝜃). Also, remember
| |
that | 1𝑧 | = |𝑧|
1
.
| |

Proposition 16.0.2. Let 𝑓 ∶ ℂ → ℂ be a Möbius transformation. Then 𝑓 is one-to-one and onto,

and the inverse of 𝑓 is also a Möbius transformation, namely
𝑑𝜔 − 𝑏
𝑔(𝜔) = .
−𝑐𝜔 + 𝑎

Proposition 16.0.3. Möbius transformations form a group, (𝑀, ◦), under the operation of com-
position of maps.

Remark. The group (𝑀, ◦) is not commutative.

Remarks.
1. August Möbius (1790-1868). Möbius transformations are also known as the fractional
linear transformations, the linear fractional transformations, and the bilinear transforma-
tions.
2. These mappings are useful for converting bounded domains into unbounded domains and
vice versa.

RHS, version May 11, 2019 113

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

3. Translations, rotations, dilations and inversions are all special cases of Möbius transforma-
tions.

Proposition 16.0.4. Every Möbius transformation can be obtained by composing a translation,

an inversion, a stretching/dilation, a rotation, and another translation, missing some out if nec-
essary.

Proof. If 𝑐 = 0, then 𝑓 (𝑧) = 𝑑𝑎 𝑧 + 𝑑𝑏 . That is, 𝑓 = 𝑓2 ◦𝑓1 , where

𝑎
𝑓1 (𝑧) = 𝑧,
𝑑
𝑏
𝑓2 (𝑧) = 𝑧 + .
𝑑
If 𝑐 ≠ 0, we choose
𝑑
𝑓1 (𝑧) = 𝑧 + ,
𝑐
1
𝑓2 (𝑧) = ,
𝑧
𝑏𝑐 − 𝑎𝑑
𝑓3 (𝑧) = 𝑧,
𝑐2
𝑎
𝑓4 (𝑧) = 𝑧 + .
𝑐
Then 𝑓 = 𝑓4 ◦𝑓3 ◦𝑓2 ◦𝑓1 and
𝑏𝑐 − 𝑎𝑑 1 𝑎
𝑓 (𝑧) = + .
𝑐2 𝑧 + 𝑑 𝑐
𝑐

Remark. This composition is not unique.

16.1 The image of a circline is a circline

Möbius transformations map circles and lines to circles and lines. The image of a circle under a
Möbius transformation is always either a circle or a line!
The general equation of a circle in the 𝑧-plane is

𝐴𝑧𝑧 + 𝐵𝑧 + 𝐵𝑧 + 𝐶 = 0,

where 𝐴, 𝐶 ∈ ℝ and 𝐵 ∈ ℂ. If 𝐴 = 0 and 𝐵 ∈ ℂ ⧵ {0}, the circline reduces to a straight line.

Under the transformation of inversion, 𝜔 = 1𝑧 , the above equation becomes

𝐶 𝜔𝜔 +𝐵𝜔 + 𝐵𝜔 + 𝐴 = 0,
⏟⏟⏟
=|𝜔|2

a circline in the 𝜔-plane.

Under the transformations of rotation, and stretching, 𝜔 = 𝑎𝑧, the equation becomes
( ) ( )
𝐴𝜔𝜔 + 𝐵𝑎 𝜔 + 𝐵𝑎 𝜔 + 𝐶𝑎𝑎 = 0.

Also, under the transformation of translation, circlines are transformed into circlines.
Since a Möbius transformation can be obtained as a combination of translations, rotations, stretch-
ings, and inversions, the claim follows.

RHS, version May 11, 2019 114

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

16.2 Finding the image of a circline

If we are interested in finding the image of some circline under a Möbius transformation, we could
use the fact that Möbius transformations map circlines to circlines and the fact that there is one
and only one circline through any triplet of three distinct points in ℂ.

Example 16.2.1. The real axis is the unique circline through 0, 1 and ∞, so its image under
1
𝑧 ↦ 𝑧−1 is a unique circline through -1, ∞ and 0, that is, the real axis.

𝑧+𝑖 ( )
Example 16.2.2. Let 𝑓 (𝑧) = 𝑧−𝑖
. Find the image of the boundary of the unit circle 𝜕𝔻 0, 1 .

Solution. We choose three points from the boundary. Let us pick −𝑖, 1 and 𝑖. The images of these
three points are
𝑓 (−𝑖) = 0, 𝑓 (𝑖) = ∞, 𝑓 (1) = 𝑖.

( ( ))
Möbius transformations map circlines onto circlines. Since ∞ ∈ 𝑓 𝜕𝔻 0, 1 , the image of
( ) ( ( )) ( ( ))
𝜕𝔻 0, 1 is a line. Since 𝑖 ∈ 𝑓 𝜕𝔻 0, 1 and 0 ∈ 𝑓 𝜕𝔻 0, 1 , the image of the boundary is
the imaginary axis.

16.3 The image of a disc and the image of a half-plane

Every straight line 𝐿 cuts ℂ into two disjoint half-planes; we denote them by ℍ1 and ℍ2 .
Since Möbius transformations are bijective, continuous mappings ℂ → ℂ, and the inverse of a
Möbius transformation is also a Möbius transformation, and Möbius transformations preserve
circlines, we obtain the following remark:
( )
Remark 16.3.1. Let 𝑧0 , 𝜔0 ∈ ℂ, 𝑟, 𝑅 > 0. The image of 𝔻 𝑧0 , 𝑟 or the complement of a closed
( ) ( ) ( )
disc, that is, ℂ⧵𝔻 𝑧0 , 𝑟 , or a half-plane, is either a disc 𝔻 𝜔0 , 𝑅 or ℂ⧵𝔻 𝜔0 , 𝑅 or a half-plane.

𝑧+𝑖
Example 16.3.2. Let 𝑓 (𝑧) = 𝑖𝑧+1
. Find the image of the unit disc.
( ( )) ( )
Solution. First we find 𝑓 𝜕𝔻 0, 1 : We pick three points from 𝜕𝔻 0, 1 and then their image
points.
𝑓 (1) = 1, 𝑓 (−1) = −1, 𝑓 (𝑖) = ∞.
Since one of the image points is ∞, we know that the image is a line. Since −1 and 1 are in the
line, the image is the real line with ∞ point.
( ( )) ( ( ))
Because of remark 16.3.1, either 𝑓 𝔻 0, 1 = ℍ+ or 𝑓 𝔻 0, 1 = ℍ− . Since 𝑓 (0) = 𝑖 ∈ ℍ+ ,
( ( ))
we know by 16.3.1 that 𝑓 𝔻 0, 1 = ℍ+ .

Remark 16.3.3.
1. When 𝑓 (𝑧) = 𝑧 + 𝑏, 𝑏 ∈ ℂ, 𝑓 is a translation, then for 𝑤0 = 𝑧0 + 𝑏 and 𝑤1 = 𝑧1 + 𝑏 is the
equation 𝑤0 − 𝑤1 = 𝑧0 − 𝑧1 . The distance between two input points is invariant under a
translation.
2. When 𝑓 (𝑧) = 𝑎𝑧 + 𝑏, 𝑎, 𝑏 ∈ ℂ, 𝑓 is a rigid motion, then the distance between two input
points is not invariant under 𝑓 . But if 𝑤𝑗 = 𝑎𝑧𝑗 + 𝑏, when 𝑗 = 0, 1, 2, then
𝑤0 − 𝑤1 𝑧 − 𝑧1
= 0 .
𝑤0 − 𝑤2 𝑧0 − 𝑧1
The ratio of the distances of these input points is invariant under on rigid motion.
3. What about Möbius maps?

RHS, version May 11, 2019 115

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

16.4 Cross-ratios and the triplet representation of a Möbius transforma-

tion
Theorem 16.4.1. Suppose that each {𝑧1 , 𝑧2 , 𝑧3 } and {𝜔1 , 𝜔2 , 𝜔3 } is an ordered triplet of distinct
points in ℂ. Then, there exists a unique Möbius transformation 𝑓 such that 𝑓 (𝑧𝑘 ) = 𝜔𝑘 , 𝑘 =
1, 2, 3 and this 𝑓 (𝑧) = 𝜔 is given by the following:
(𝜔 − 𝜔1 )(𝜔2 − 𝜔3 ) (𝑧 − 𝑧1 )(𝑧2 − 𝑧3 )
=
(𝜔 − 𝜔2 )(𝜔1 − 𝜔3 ) (𝑧 − 𝑧2 )(𝑧1 − 𝑧3 )

Proof. The map

(𝑧 − 𝑧1 )(𝑧2 − 𝑧3 )
𝑔∶ 𝑧 ↦
(𝑧 − 𝑧2 )(𝑧1 − 𝑧3 )
takes 𝑧1 , 𝑧2 , 𝑧3 to 0, ∞, 1, respectively, and the map

(𝜔 − 𝜔1 )(𝜔2 − 𝜔3 )
ℎ∶ 𝜔 ↦
(𝜔 − 𝜔2 )(𝜔1 − 𝜔3 )

takes 𝜔1 , 𝜔2 , 𝜔3 to 0, ∞, 1, respectively. The composition map 𝑓 ∶= ℎ−1 ◦𝑔 is a Möbius trans-

formation which takes 𝑧𝑘 to 𝜔𝑘 for 𝑘 = 1, 2, 3 and

(𝜔 − 𝜔1 )(𝜔2 − 𝜔3 ) (𝑧 − 𝑧1 )(𝑧2 − 𝑧3 )
= .
(𝜔 − 𝜔2 )(𝜔1 − 𝜔3 ) (𝑧 − 𝑧2 )(𝑧1 − 𝑧3 )

The last step to prove uniqueness is left to the reader.

Definition 16.4.2. The cross ratio of four complex numbers 𝑧0 , 𝑧1 , 𝑧2 , 𝑧3 , denoted

by [𝑧0 , 𝑧1 , 𝑧2 , 𝑧3 ] is the number

(𝑧0 − 𝑧1 )(𝑧2 − 𝑧3 )
[𝑧0 , 𝑧1 , 𝑧2 , 𝑧3 ] = .
(𝑧0 − 𝑧2 )(𝑧1 − 𝑧3 )

If one of the points is ∞, then [𝑧0 , 𝑧1 , 𝑧2 , 𝑧3 ] is defined as a limit.

Remark. Suppose 𝑧2 = ∞. Take 𝑎𝑗 so that lim𝑗→∞ 𝑎𝑗 = ∞, and then

→0
⏞⏞⏞
𝑧3
1−
(𝑧0 − 𝑧1 )(𝑎𝑗 − 𝑧3 ) 𝑧 − 𝑧1 𝑎𝑗 𝑧 − 𝑧1
lim = lim 0 𝑧0 = 0 .
𝑗→∞ (𝑧0 − 𝑎𝑗 )(𝑧1 − 𝑧3 ) 𝑗→∞ 𝑧1 − 𝑧3 𝑧3 − 𝑧1
−1
𝑎𝑗
⏟⏟⏟
→0

Remark. The importance of this cross ratio is, that it is preserved by Möbius transformations,
that is [( ) ( ) ( ) ( )]
[𝑧0 , 𝑧1 , 𝑧2 , 𝑧3 ] = 𝑧0 , 𝑓 𝑧1 , 𝑓 𝑧2 , 𝑓 𝑧3

for all Möbius transformation functions 𝑓 .

The claim is trivial, if 𝑓 is a translation 𝑓 (𝑧) = 𝑧+𝑐, 𝑐 ∈ ℂ. Similarly, if 𝑓 is a dilation 𝑓 (𝑧) = 𝑘𝑧,
since all 𝑘-factors will cancel. Similarly, if 𝑓 is a rotation 𝑓 (𝑧) = exp (𝑖𝛼) 𝑧.
( ) ( ) 𝑧 −𝑧
With inversions note that 𝑓 𝑧1 − 𝑓 𝑧2 = − 𝑧1 𝑧 2 . Then
1 2
( ) ( ) ( ) ( )
𝑓 𝑎1 − 𝑓 𝑎2 𝑓 𝑎3 − 𝑓 𝑎4 𝑎1 − 𝑎2 𝑎3 − 𝑎4
( ) ( )⋅ ( ) ( )= ⋅ .
𝑓 𝑎1 − 𝑓 𝑎3 𝑓 𝑎2 − 𝑓 𝑎4 𝑎1 − 𝑎3 𝑎2 − 𝑎4

Thus inversion does not affect the cross ratio either.

RHS, version May 11, 2019 116

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

Since every Möbius transformation is a combination of transformations, dilations, rotations and

inversions, we see that none of the Möbius transformations affect the cross ration and as a corol-
lary of Theorem 16.4.1, we have the following theorem:

Theorem 16.4.3. The cross ratio is invariant under Möbius transformations.

Remark 16.4.4. One consequence of this formula is that Mobius transformations cannot map any
four given points to any other four given points, unless their cross ratios are equal.

Example 16.4.5. There exists no Mobius transformation that maps 0, 1, 2, 3 to 0, 1, 2, 10.

Remark 16.4.6. However, one can use this cross ratio formula to find a Mobius transformation
that maps three given points 𝑧1 , 𝑧2 , 𝑧3 to any other three given points 𝑤1 , 𝑤2 , 𝑤3 , 𝑤, that is
( )
𝑓 𝑧1 = 𝑤1 ,
( )
𝑓 𝑧2 = 𝑤2 ,
( )
𝑓 𝑧2 = 𝑤3 ,
𝑓 (𝑧) = 𝑤,
[ ] [ ]
The identity should be valid 𝑧1 , 𝑧2 , 𝑧3 , 𝑧 = 𝑤1 , 𝑤2 , 𝑤3 , 𝑤 and one can use this identity to
solve for 𝑓 (𝑧) in terms of 𝑧.

Example 16.4.7. Let the points −1, 0, 1 be given and we like to map them to −1, 𝑖, 1, respectively.
By solving the equation
[ ] [ ]
−1, 0, 1, 𝑧 = −1, 𝑖, 1, 𝑤
( )( ) ( )( )
−1 − 0 1 − 𝑧 1 − 𝑖 1 − 𝑓 (𝑧)
( )( )=( )( )
−1 − 1 0 − 𝑧 −1 − 1 𝑖 − 𝑓 (𝑧)
( )( )
𝑧−1 1 + 𝑖 𝑓 (𝑧)
= ( )
2𝑧 2 𝑓 (𝑧) − 𝑖
𝑧+𝑖
𝑓 (𝑧) =
𝑖𝑧 + 1
we found the mapping 𝑓 .

Example 16.4.8. Find the mapping 𝑓 such that

( )
𝑓 −1 = −1,
( )
𝑓 0 = 𝑖,
( )
𝑓 1 = 1.

Solution. By letting 𝑧1 = −1, 𝑧2 = 0, 𝑧3 = 1, 𝑤1 = −1, 𝑤2 = 𝑖, 𝑤3 = 1 and 𝑤 = 𝑓 (𝑧) we can

use Möbius transformation. Thus the mapping is
( )( ) ( )( )
𝑧 − 𝑧1 𝑧2 − 𝑧3 𝑤 − 𝑤1 𝑤2 − 𝑤3
( )( )=( )( )
𝑧 − 𝑧2 𝑧1 − 𝑧3 𝑤 − 𝑤2 𝑤1 − 𝑤3
( )( ) ( )( )
𝑧+1 0−1 𝑤+1 𝑖−1
( )( )= ( )
𝑧 − 0 −1 − 1 (𝑤 − 𝑖) −1 − 1
𝑧+1( ) 𝑤+1( )
−1 = 𝑖−1
𝑧 𝑤−1
𝑤 + 𝑧𝑤𝑖 = 𝑖 + 𝑧
( )
𝑤 1 + 𝑖𝑧 = 𝑖 + 𝑧
𝑧+𝑖
𝑤= .
𝑧𝑖 + 1

RHS, version May 11, 2019 117

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

Example 16.4.9.
1. Find the Möbius transformation 𝑓 such that:

𝑓 (0) = 0, 𝑓 (1) = 1, 𝑓 (−1) = ∞.

( )
2. Find the image of 𝔻 0, 1 under this mapping.

Solution.
1. By using definition 16.4.2 and Theorem 16.4.1, we get

[𝑧, 0, 1, −1] ⇔ [𝑤, 0, 1, ∞]

𝑧 − 0 1 − (−1) 𝜔 − 0
⋅ =
𝑧 − 1 0 − (−1) 𝜔 − 1
𝑧 𝜔
2 =
𝑧−1 𝜔−1
2𝑧𝜔 − 2𝑧 = 𝜔𝑧 − 𝜔
𝑧𝜔 − 2𝑧 = −𝜔
𝑧𝜔 + 𝜔 = 2𝑧
2𝑧
𝜔= .
𝑧+1
( ( ))
2. Note that 𝑓 (𝑖) = 1 + 𝑖. Since −1 ↦ ∞, 1 ↦ 1, 𝑖 ↦ 1 + 𝑖, then 𝑓 𝜕𝔻 0, 1 = {𝑧 ∶ Re 𝑧 =
( ( ))
1}. Since 𝑓 (0) = 0, 𝑓 𝔻 0, 1 = {𝑧 ∶ Re 𝑧 < 1}.

Remark 16.4.10. If a half-plane ℍ is given and we need to find a Möbius transformation which
maps ℍ onto some disc 𝔻, then we pick three points 𝑧1 , 𝑧2 and 𝑧3 from 𝜕ℍ and three points
𝜔1 , 𝜔2 and 𝜔3 from 𝜕𝔻 to find 𝑓 such that 𝑓 (𝑧𝑘 ) = 𝜔𝑘 , 𝑘 = 1, 2, 3. Then we get a Möbius
transformation such that 𝑓 (ℍ) = 𝔻 or 𝑓 (ℍ) = ℂ ⧵ 𝔻. If 𝑓 (ℍ) = 𝔻 we are done. If 𝑓 (ℍ) = ℂ ⧵ 𝔻,
we use a rotation
( 𝑔 which
) gives 𝑔(ℍ) = ℂ ⧵ ℍ. Then, we combine them to get 𝑓 ◦𝑔 and obtain
𝑓 (𝑔(ℍ)) = 𝑓 ℂ ⧵ ℍ = 𝔻.

16.5 Möbius transformations preserve angles

Möbius transformation preserves the angles of objects, although they do not, in general, preserve
the shape of objects.

Theorem 16.5.1. Let 𝛾1 and 𝛾2 be two smooth 𝐶 1 -curves that intersect at 𝑧0 = 𝛾1 (𝑡0 ) = 𝛾2 (𝑡0 ),
suppose that 𝛾𝑘′ (𝑡0 ) ≠ 0 so that the curves aren’t stationary at the intersection, and suppose that
𝑓 is a Möbius transformation. Then the angle 𝛼 at which the smooth curves intersect each other
at 𝑡0 is simply the difference in the arguments of their tangents at time 𝑡0 :
( ) ( )
𝛼 = Arg 𝛾2′ (𝑡0 ) − Arg 𝛾1′ (𝑡0 ) .

Proof. The transformed curves 𝑓 ◦𝛾𝑘 , 𝑘 = 1, 2 are still smooth (at least when ignoring the pos-
sible pole of the Möbius transformation), and they intersect each other at 𝑓 (𝑧0 ) = 𝑓 (𝛾1 (𝑡0 )) =
𝑓 (𝛾2 (𝑡0 )). Their tangents at time 𝑡0 are given by the chain rule as

(𝑓 ◦𝛾𝑘 )′ (𝑡0 ) = 𝑓 ′ (𝛾𝑘 (𝑡0 )) ⋅ 𝛾𝑘′ (𝑡0 ) = 𝑓 ′ (𝑧0 ) ⋅ 𝛾𝑘′ (𝑡0 ).

Since 𝑓 is a Möbius transformation, 𝑓 ′ (𝑧0 ) ≠ 0, so we can calculate

( ) ( )
Arg 𝑓 ′ (𝑧0 ) ⋅ 𝛾2′ (𝑡0 ) − Arg 𝑓 ′ (𝑧0 ) ⋅ 𝛾1′ (𝑡0 )
( ) ( ) ( ) ( )
= Arg 𝑓 ′ (𝑧0 ) + Arg 𝛾2′ (𝑡0 ) − Arg 𝑓 ′ (𝑧0 ) − Arg 𝛾1′ (𝑡0 )

RHS, version May 11, 2019 118

Complex Analysis I 16 MÖBIUS TRANSFORMATIONS

( ) ( )
= Arg 𝛾2′ (𝑡0 ) − Arg 𝛾1′ (𝑡0 )
= 𝛼.

Thus if two (smooth) curves intersect at an angle 𝛼, when we transform them using a Möbius
transformation, their images also intersect at an angle 𝛼. So angles are not distorted, only distances
are.
( )
The assumption that 𝑓 ′ 𝑧0 ≠ 0 is important!

Example 16.5.2. The real and imaginary axis meet at right angles, but they map under 𝑓 (𝑧) = 𝑧2
to the positive real axis and negative real axis, respectively, which are 180◦ apart!

𝑧−1
Example 16.5.3. Let 𝑓 (𝑧) = 𝑧−3
.
( )
1. What is the image of 𝜕𝔻 0, 1 under 𝑓 ? Now

1 ↦ 0,
2 𝑖
𝑖↦ − ,
5 5
1
−1 ↦ .
2
2
The image is a circle which goes through the points 0, 5
− 𝑖
5
and 21 .
2. What is the image of the real axis? Now

−1 ↦ 1∕2,
0 ↦ 1∕3,
1 ↦ 0.

So the image of the real axis is the real axis.

The unit circle has right angles with the real axis at −1 and 1. Since Möbius transformations
preserves angles, the image of the unit circle meets the image of the real axis at right angles at
( ) ( )
𝑓 −1 = 1∕2 and 𝑓 1 = 0. Hence the image of the unit circle is a circle with radius 41 and
( ) (( ) ) (( ) )
center 41 , 0 . Since 0 ↦ 13 ∈ 𝐷 1
4
, 0 , 1
4
the unit circle maps onto 𝐷 1
4
, 0 , 14 .

RHS, version May 11, 2019 119

Complex Analysis I 17 CONFORMAL MAPPINGS

17 Conformal mappings
17.1 On conformal mappings
Definition 17.1.1. Let Ω be open set in ℂ. Suppose that 𝑓 ∶ Ω → ℂ analytic in Ω. The function
𝑓 is conformal at 𝑧 ∈ Ω if 𝑓 ′ (𝑧) ≠ 0. The function 𝑓 is conformal in Ω if 𝑓 is conformal at every
point in Ω, i.e. 𝑓 ′ (𝑧) ≠ 0 for all 𝑧 ∈ Ω. If 𝑓 is conformal at every 𝑧 ∈ Ω and 𝑓 is bijective, 𝑓 is
called a conformal mapping.

Remark 17.1.2.
1. The geometric property where angles are preserved under the analytic mappings is confor-
mality.
2. Mappings which preserve angles of objects are conformal.

An analytic mapping 𝑓 ∶ Ω → Ω is a conformal mapping if 𝑓 ′ (𝑧) ≠ 0 for all 𝑧 ∈ Ω and 𝑓 is

bijective (that is, one-to-one and onto).

Example 17.1.3. Let

{ }
𝐴 = 𝑧 ∈ ℂ ∶ Re (𝑧) > 0 and Im (𝑧) > 0 and
{ }
𝐵 = 𝑧 ∈ ℂ ∶ Im (𝑧) > 0 = 𝐼𝐻 + .

The mapping 𝑓 ∶ 𝐴 → 𝐼𝐻 + where 𝑧 ↦ 𝑧2 is analytic in 𝐴 and 𝑓 ′ (𝑧) = 2𝑧 ≠ 0 for all 𝑧 ∈ 𝐴.

Hence 𝑓 is conformal.

17.1.4 Conformality at ∞ as analyticity at ∞

Suppose that 𝑓 is defined in 𝔻(∞, 𝑟) and gets values in ℂ there.
( )
• Then 𝑓 is conformal at ∞ if 𝑧 ↦ 𝑓 1𝑧 is conformal at 0.
( ) ( )
Suppose that 𝑓 is defined in an open disc 𝔻 𝑧0 , 𝑟 and 𝑓 𝑧0 = ∞.
1
• If 𝑧0 ∈ ℂ, then 𝑓 is conformal at 𝑧0 , if 𝑧 ↦ 𝑓 (𝑧)
is conformal at 𝑧0 .
1
• If 𝑧0 = ∞, then 𝑓 is conformal at ∞, if 𝑧 ↦ ( ) is conformal at 0.
𝑓 1𝑧

{ }
Example 17.1.5. Let 𝑓 (𝑧) = 𝑧2𝑧−4 , 𝑓 is analytic in ℂ ⧵ −2, 2 . We have to set 𝑓 (∞) = 0 and
( )
𝑓 ±2 = ∞ in order to get a continuous 𝑓 ∶ ℂ ↦ ℂ. Now

( ) 1 𝑧2 − 4 4
𝑓 ±2 = ∞ ∶ = = 𝑧 − is analytic at ± 2 and conformal at ± 2.
𝑓 (𝑧) 𝑧 𝑧
⎧ ( )
1∕𝑧
⎪ 𝑓 1𝑧 = 1∕𝑧2 −4 = 1−4𝑧
𝑧
4, 0 < |𝑧| < 𝑟,
𝑓 (∞) = 0 ∶ 𝑔(𝑧) = ⎨
⎪ 0, when 𝑧 = 0.
⎩
( ) 𝑔 (0+ℎ)−𝑔 (0)
Then in the latter case 𝑔 ′ 0 = limℎ→0 𝑓
= 1 ≠ 0, and 𝑔 is conformal at 0 and 𝑓 is
conformal at 0.

Example 17.1.6.
𝑧+𝑖
1. Find the image of the real axis under the mapping 𝑓 ∶ 𝑧 ↦ 𝑧+1 .
( ) 𝑧+𝑖
2. Find the image of the boundary of 𝔻 0, 1 under the mapping 𝑓 ∶ 𝑧 ↦ 𝑧+1
.

Solution.

RHS, version May 11, 2019 120

Complex Analysis I 17 CONFORMAL MAPPINGS

1
1. We pick 1, 0, −1 ∈ ℝ and calculate that 𝑓 (−1) = ∞, 𝑓 (0) = 𝑖, and 𝑓 (1) = 2
+ 2𝑖 .

{(𝑥, 𝑦) ∶ 𝑦 = −𝑥 + 1} = {𝑧 ∶ Im 𝑧 = − Re 𝑧 + 1}.

2. Now
𝑖+1
𝑓 (−1) = ∞ (i.e. the image is a line), 𝑓 (−𝑖) = 0, 𝑓 (1) =
2
( )
𝜕𝔻 0, 1 ⊥ {𝑧 ∶ Im 𝑧 = 0}
⏟⏞⏞⏞⏞⏞⏞⏞⏟⏞⏞⏞⏞⏞⏞⏞⏟
The real axis
𝜋
The angle at (0, 1) is 2
. By conformality,
( ( )) ( ) ( ) 1 𝑖
𝑓 𝜕𝔻 0, 1 ⊥𝑓 {𝑧 ∶ Im 𝑧 = 0} at 𝑓 1 = + .
2 2
( ( ))
Hence 𝑓 𝜕𝔻 0, 1 = {(𝑥, 𝑦) ∶ 𝑥 = 𝑦}, or we could calculate

( ) ( ) 1+𝑖
𝑓 −1 = ∞, 𝑓 (−𝑖) = 0 and 𝑓 1 = .
2

Examples 17.1.7.
1. Möbius transformations are conformal. We refer to 16.5 (𝑓 ′ (𝑧0 ) ≠ 0).
2. 𝑓 (𝑧) = 𝑧2 is not conformal at 0 but is at other points.

Example 17.1.8. Find the Möbius transformation 𝑓 such that

( )
𝑓 1 = 0,
( )
𝑓 −1 = ∞,
𝑓 (𝑖) = 𝑖.

Now
( )( ) ( ) ( )
𝑧 − 1 −1 − 𝑖 𝑤 − 0 (∞ − 𝑖) 𝑤−0 ( ) 𝑤
( ( )) ( ) = (𝑤 − ∞) (0 − 𝑖) = (0 − 𝑖) −1 = −1 (−𝑖) .
𝑧 − −1 1−𝑖

Thus the transformation is

( )( )
𝑧 − 1 −𝑖 + 1 ( ) 𝑧 − 1
𝑤= ( )( ) −1 = ,
𝑧+1 𝑖−1 𝑧+1

and the given conditions apply

( )
𝑤 1 = 0,
( )2
𝑖−1 1−𝑖 1−𝑖 1 − 2𝑖 − 1
𝑤(𝑖) = =− =− =− = 𝑖,
𝑖+1 1+𝑖 1+1 2
( )
𝑤 −1 = ∞.

17.1.9. We refer to 16.5. Any analytic mapping 𝑓 when 𝑓 ′ (𝑧0 ) ≠ 0 preserves angles between
smooth curves.

RHS, version May 11, 2019 121

Complex Analysis I 17 CONFORMAL MAPPINGS

17.2 On conformal mapping problems

Let Ω1 and Ω2 be two domains in ℂ. The problem or task is to find out, if there exists a bijective
conformal map 𝑓 ∶ Ω1 → Ω2 .
If the mapping exists, then the task is to find it.

Example 17.2.1. Find a bijective conformal map 𝑓 which maps the sector
𝜋
{𝑧 ∶ 0 < Arg 𝑧 < , 0 < |𝑧| < 1}
2
( )
onto 𝔻 0, 1 . See Figure 17.1.

Im Im

𝛿?
Re Re

𝑧-plane 𝜔-plane

Figure 17.1: How to find a bijective

( ) conformal mapping 𝛿 which maps the sector {𝑧 ∶ 0 <
Arg 𝑧 < 𝜋2 , 0 < |𝑧| < 1} onto 𝔻 0, 1 ? (Example 17.2.1.)

Solution. The asked mapping 𝛿 can constructed by composing suitable bijective conformal map-
pings 𝜔1 , 𝜔2 , 𝜔3 and 𝜔4 . Now we need to
1. map the real axis to the real axis using the cross ratio formula,
2. the image if the unit circle is the imaginary axis.
Let (as shown in Figure 17.2)

1+𝑧 𝑧−𝑖
𝜔1 = 𝑧2 , 𝜔2 = , 𝜔3 = 𝜔1 = 𝑧2 , and 𝜔4 = .
1−𝑧 𝑧+𝑖
Now
( )2
( )2 1+𝑧2
−𝑖
1 + 𝑧2 1 + 𝑧2 1−𝑧2
𝜔2 ◦𝜔1 = , 𝜔3 ◦𝜔2 ◦𝜔1 = , 𝜔4 ◦𝜔3 ◦𝜔2 ◦𝜔1 = ( )2 ;
1 − 𝑧2 1 − 𝑧2 1+𝑧2
1−𝑧2
+𝑖

(1 + 𝑧2 )2 − 𝑖(1 − 𝑧2 )2
𝜔 = 𝛿(𝑧) = 𝜔4 ◦𝜔3 ◦𝜔2 ◦𝜔1 (𝑧) =
(1 + 𝑧2 )2 + 𝑖(1 − 𝑧2 )2
And the asked bijective conformal mapping 𝛿 has been constructed.

RHS, version May 11, 2019 122

Complex Analysis I 17 CONFORMAL MAPPINGS

Im Im

𝜔1 = 𝑧2
Re Re

𝑧-plane 𝜔1 -plane
(a) The mapping 𝜔1 , 𝑧 ↦ 𝑧2 maps the quadrant to a half-circle.

Im Im

1+𝑧
𝜔2 = 1−𝑧
Re Re

𝑧-plane 𝜔2 -plane
(b) The mapping 𝜔2 , 𝑧 ↦ 1+𝑧
1−𝑧
maps the half-circle to a quarter-plane.

Im Im

𝜔3 = 𝑧2
Re Re

𝑧-plane 𝜔3 -plane
(c) The mapping 𝜔3 = 𝜔1 , 𝑧 ↦ 𝑧2 maps the quarter-plane to a half-plane.

Im Im

𝑧−𝑖
𝜔4 = 𝑧+𝑖
Re Re

𝑧-plane 𝜔4 -plane
(d) The mapping 𝜔4 , 𝑧 ↦ 𝑧−𝑖
𝑧+𝑖
maps the half-plane to a disc.

Figure 17.2: Construction of the bijective conformal mapping 𝛿 = 𝜔4 ◦𝜔3 ◦𝜔2 ◦𝜔1 in the solution
to the example 17.2.1.

RHS, version May 11, 2019 123

Complex Analysis I REFERENCES

References
[1] Kari Astala, Kompleksianalyysi I, University of Helsinki, 2016.
[2] Joaquim Bruna and Julia Cufi, Complex Analysis, EMS Textbooks in Mathematics, Euro-
pean Mathematical Society, 2013.
[3] James Ward Brown and Ruel V. Churchill, Complex Variables and Applications. Eighth
Edition, McGraw-Hill Higher Education, 2009.
[4] Ritva Hurri-Syrjänen, Kompleksianalyysi I kurssi, (Mainly based on Kari Astala’s notes
from 2005 and Terry Tao’s notes from the web), University of Helsinki, 2012.
[5] Serge Lang, Complex Analysis, Addison Wesley.
[6] H. A. Priestley, Introduction to Complex Analysis, second edition, Oxford University Press,
2003.
[7] Murray R. Spiegel, Complex Variables with an Introduction to Conformal Mapping and
Its Applications, Schaum’s Outline Series, McGraw Hill, 1999.
[8] Elias M. Stein and Rami Shakarchi, Complex Analysis, Princeton Lectures in Analysis II,
Princeton University Press, Princeton, New Jersey, 2003.
[9] Terence Tao, Math 132, Lecture notes in the web
[10] J. D. Dixon, A brief proof of Cauchy’s integral theorem, Proc. Amer. Math. Soc., 29 (1971),
pp. 625–626.

RHS, version May 11, 2019 124

Complex Analysis I ALPHABETICAL INDEX

Alphabetical Index
A Complex plane 7
Accumulation point of a set 21, 24 Complex sequence 24, 41
Addition formulas for sine and cosine Cauchy sequence 26
functions 54 convergence 24, 25
Analytic function 27, 32 Complex series 41
as an integral 87 absolute convergence 42
harmonic real and imaginary part 37 convergence 41
polynomials 29 remainder 42
through antiderivative 91 Component lemma 101
using C-R 35 Conformal mapping 120
using power series 47, 50, 91 Connected set 23
Antiderivative 74, 75 Connection between complex and real
analytic function in an open disc 82 derivatives 32, 39
Continuity of a function 24, 25, 28, 35
B Contour 69
Boundary of a disc 20 Cosine function 53
Boundary point of a set 21 derivative 54
Bounded set 21 Cross ratio 116
Cycle 99
C length 99
Cauchy global integral theorem 101
proof 105 D
Cauchy's inequalities 91 d'Alembert's ratio test 42
Cauchy's local integral formula 87 De Moivre's theorem 16, 17
proof 88 Deformation theorem 103
Cauchy's second formula 88 Domain 23, 28, 55, 75
proof 90 simply connected 101
Cauchy-Goursat theorem 86
Cauchy-Riemann equations 32, 34, 83 E
Chain rule 29 Entire function 28, 92
Closed curve 66 Estimation lemma 73
Closed disc 20 Euclidean plane 6
Closed set 20 Euler's formula 13, 53, 55
Closure of a set 21 Extended complex plane 109
Compact set 21 limits 110
Comparison test 42 open set 110
Complex dierentiable function 28
complex derivative
derivative of the inverse
27, 34
30
F
Fundamental equation 55
dierence quotient 27 Fundamental Theorem of Algebra 93
dierentiation formulas 28
Complex exponential function
one-to-one and onto
51, 55
58 G
properties 51 General power function 65
Complex hyperbolic functions 54 Geometric series 42, 48
Complex logarithm proof of convergence 43
branches 61 Goursat's theorem 79
denition 60
properties 63 H
Complex number Hadamard's
argument 13, 14, 60 formula 46
Cartesian form 10 proof of theorem 47
complex conjugate 10, 52 theorem 46
geometric meaning of the multiplication Harmonic function 37
14
imaginary part 10, 11, 25 I
inverse element 8 Identity theorem 94
modulus 10, 13, 52 Imaginary unit 9
polar form 13 Integrals along closed contours 77
real part 10, 11, 25, 52 Integrals along cycles 99
roots 16 Integrals along piecewise smooth curves 72

RHS, version May 11, 2019 125

Complex Analysis I ALPHABETICAL INDEX

Integrals along smooth curves 70 P

arclength 73 Parametrized curve 66, 99
integral of the negative 72 equivalent parametrizations 67
linearity 71 length of a curve 72
with antiderivative 74 negative 69
Integrals of complex-valued functions of a orientation 68
real variable 66 piecewise smooth 67
Integrals of rational functions on the real line Path 22
104 polygonal line 22
Interior point 20 straight line segment 22
Isolated point of a set 21 Power series 44
disc of convergence 45
L innitely complex dierentiable 50
Laplace operator 37 radius of convergence 45
Limes superior 46
Limit of a function
Liouville's theorem
24
92
R
Real dierentiable function 30, 35
Local maximum modulus theorem 95 Real logarithm 60
Riemann sphere 109
M Root test 42
Möbius transformations 113
angles
circlines
118
114
S
Schwarz lemma 97
images of discs and half-planes 115 Sine function 53
triplet representation 116 derivative 54
Mappings 57 Smooth curve 66
Maximum modulus principle 95
Maximum modulus theorem 96
Meromophic function 111 T
meromorphic 111 Tangent function 56
Metric space 20 Taylor expression 51
Minimum modulus theorem 98 Taylor series 53
Morera's theorem 91 Theorem of Abel 44
triangle inequality 12
N
Neighbourhood 21 U
Norm 20 Unit disc 20

O W
Open disc 20 Winding number 99
Open set 20 for a cycle 100
Osborn's rules 54 Winding number lemma 99

RHS, version May 11, 2019 126