COMPLEX ANALYSIS

Lecture notes for

MA 648

Rudi Weikard

2 2

1 1
0
-2 0

-1

0 -1

1
-2
2

Version of December 2, 2022

 c 2022. This manuscript version is made available under the CC-BY-NC-SA 4.0 license
http://creativecommons.org/licenses/by-nc-sa/4.0/.
 Contents

Preface iii
Chapter 1. The complex numbers: algebra, geometry, and topology 1
1.1. The algebra of complex numbers 1
1.2. The topology and geometry of complex numbers 2
1.3. Sequences and series 3

Chapter 2. Complex-valued functions of a complex variable 5

2.1. Limits and continuity 5
2.2. Holomorphic functions 5
2.3. Integration 6
2.4. Contours 7
2.5. Series of functions 8
2.6. Analytic functions 9
Chapter 3. Cauchy’s theorem and some of its consequences 11
3.1. The index of a point with respect to a closed contour 11
3.2. Cauchy’s theorem 11
3.3. Consequences of Cauchy’s theorem 13
3.4. The global version of Cauchy’s theorem 14
Chapter 4. Isolated singularities 17
4.1. Classifying isolated singularities 17
4.2. The calculus of residues 18

Chapter 5. A zoo of functions 21

5.1. Polynomial functions 21
5.2. Rational functions 22
5.3. Exponential and trigonometric functions 23
5.4. The logarithmic function and powers 24

Chapter 6. Entire functions 27

6.1. Infinite products 27
6.2. Weierstrass’s factorization theorem 27
6.3. Counting zeros using Jensen’s theorem 28
6.4. Hadamard’s factorization theorem 29

Appendix A. Topology in metric spaces 31

A.1. Basics 31
A.2. Compactness 32
A.3. Product spaces 32
i
ii CONTENTS

Glossary 33

List of special symbols 35

Index 37
 Preface

Complex Analysis, also called the Theory of Functions, is one of the most important
and certainly one of the most beautiful branches of mathematics. This is due to the fact
that, in the case of complex variables, differentiability in open sets has consequences which
are much more significant than in the case of real variables. It is the goal of this course to
study these consequences and some of their far reaching applications.
As a prerequisite of the course familiarity with Advanced Calculus1 is necessary and,
generally, sufficient. Some prior exposure to topology may be useful but the necessary
concepts and theorems are provided in the appendix if they are not covered in the course.
In order to help you navigate the notes an index, a list of symbols, and a glossary of
important terms (not explicitly defined in the text) are appended at the end.
There are many textbooks on our subject; several of the following have been consulted
in the preparation of these notes.
• Lars V. Ahlfors, Complex analysis, McGraw-Hill, New York, 1953.
• John B. Conway, Functions of one complex variable, Springer, New York, 1978.
• Walter Rudin, Real and complex analysis, McGraw-Hill, New York, 1987.
• Elias M. Stein and Rami Shakarchi, Complex analysis, Princeton University Press,
Princeton, 2003.
These lecture notes are not intended to be encyclopedic; I tried, rather, to be peda-
gogic. As a consequence there are many occasions where assumptions could be weakened
or conclusions strengthened. Also, very many interesting results living close by have been
skipped and other interesting and fruitful subjects have not even been touched. This is just
the very beginning of the most exciting and beautiful Theory of Functions.
Thanks to my classes of Fall 2013, 2016, and 2019 who caused many improvements to
the notes and found a large number of mistakes (but probably not all).

1
Among many sources my Advanced Calculus notes would serve in this respect. They may be found
at http://people.cas.uab.edu/~weikard/teaching/ac.pdf.

iii
 CHAPTER 1

The complex numbers: algebra, geometry, and topology

1.1. The algebra of complex numbers

1.1.1 The field of complex numbers. The set C of complex numbers is the set of all
ordered pairs of real numbers equipped with two binary operations + and · as follows: if
(a, b) and (c, d) are pairs of real numbers, then
(a, b) + (c, d) = (a + c, b + d)
and
(a, b) · (c, d) = (ac − bd, ad + bc).
The set C with the binary operations + and · is a field, i.e., the following properties
hold:
(A1): (x + y) + z = x + (y + z) for all x, y, z ∈ C (associative law of addition).
(A2): x + y = y + x for all x, y ∈ C (commutative law of addition).
(A3): (0, 0) is the unique additive identity.
(A4): Each x ∈ C has a unique additive inverse −x ∈ C, called the negative of x.
(M1): (x · y) · z = x · (y · z) for all x, y, z ∈ C (associative law of multiplication).
(M2): x · y = y · x for all x, y ∈ C (commutative law of multiplication).
(M3): (1, 0) is the unique multiplicative identity.
(M4): Each x ∈ C \ {(0, 0)} has a unique multiplicative inverse x−1 ∈ C, called the
reciprocal of x.
(D): (x + y) · z = x · z + y · z for all x, y, z ∈ C (distributive law ).
For simplicity we will usually write x − y in place of x + (−y), and xy or x/y for x · y −1 .
It is also common to write xy instead of x · y and to let multiplication take precedence over
addition, i.e., x + yz is short for x + (yz).
1.1.2 Embedding of R into C. The identification of a real number x with the complex
number (x, 0) is an injective map and addition and multiplication of real numbers are faith-
fully reproduced in C, i.e., (x, 0) + (y, 0) = (x + y, 0) and (x, 0)(y, 0) = (xy, 0). Thus we
may treat R as a subfield of C.
Next note that (x, y) = (x, 0) + (0, 1)(y, 0). Abbreviating (0, 1) by i this becomes x + iy,
the standard notation for complex numbers. Numbers of the form iy, y ∈ R, are called
(purely) imaginary numbers. Note that i2 = −1.
Complex numbers are not ordered. Still we will encounter inequalities like z ≥ w and,
if we do, it is tacitly assumed that z, w ∈ R.
1.1.3 Real and imaginary parts of a complex number. If a, b ∈ R and z = a + ib
is a complex number, then a is called the real part of z (denoted by Re z) and b is called
the imaginary part of z (denoted by Im z). The number z = a − ib is called the complex
conjugate of z. If z ∈ C, then z + z = 2 Re z, z − z = 2i Im z, and zz = (Re z)2 + (Im z)2 ≥ 0.
If z, w ∈ C, then z + w = z + w and zw = z w.
1
2 1. THE COMPLEX NUMBERS: ALGEBRA, GEOMETRY, AND TOPOLOGY

1.1.4 Absolute value √ of a complex number. If a, b ∈ R and z = a + ib is a complex

number, then |z| = a2 + b2 ∈ [0, ∞) is called the absolute value or modulus of z. Note
that this definition is compatible with the definition of the absolute value of a real number.
If z ∈ C, then |z|2 = zz, | Re z| ≤ |z|, and | Im z| ≤ |z|. If z 6= 0, then 1/z = z/|z|2 .
The absolute value of a product of two complex numbers equals the product of the absolute
values of the numbers, i.e., |z1 z2 | = |z1 | |z2 |.
1.1.5 The triangle inequality. If x, y ∈ C, then
|x + y| ≤ |x| + |y| and |x + y| ≥ |x| − |y| .

1.2. The topology and geometry of complex numbers

1.2.1 Distance between complex numbers. Given two complex number z1 and z2 we
call |z1 − z2 | their distance. In fact, equipped with this distance function, C becomes a
metric space and thus an object of geometry. When one wants to emphasize the geometry
of C one calls it the complex plane. Appendix A gathers some pertinent information about
metric (and topological) spaces. In particular, it reviews the concepts of convergence and
limit in metric spaces, which are obvious generalizations of those in R.
1.2.2 Polar representation of complex
p numbers. For any non-zero complex number
z = x + iy, x, y ∈ R define r = x2 + y 2 = |z| > 0 and θ ∈ (−π, π] by the relations
cos θ = x/r and sin θ = y/r. These requirements determine r and θ uniquely. Interpreting z
as a point (x, y) in a two-dimensional rectangular coordinate system the number r represents
the distance of z from the origin. If θ ≥ 0, it is the counterclockwise angle from the positive
x-axis to the ray connecting z and the origin, otherwise −θ is the clockwise angle. The
numbers r and θ are called the polar coordinates of z.
1.2.3 Powers and roots. For any z ∈ C we define z 0 = 1 and, inductively, z n = zz n−1
for n ∈ N. When z 6= 0 and n ∈ N we note that (z −1 )n = (z n )−1 and we abbreviate this
number by z −n . The numbers z n , n ∈ Z, are called the (integer) powers of z.
We also define n-th roots of complex numbers when n ∈ N. The complex number a
is called an n-th root of a complex number b, if an = b. Every non-zero complex number
has exactly n n-th roots. To see this note that (cos θ + i sin θ)n = cos(nθ) + i sin(nθ) when
n ∈ N.
1.2.4 Open and closed sets. As a metric space C has open and closed balls which, in
this context, are usually called disks. Thus, as explained in A.1.5, C is a topological space
whose open sets are unions of open disks; in particular open disks are open. The following
characterizations of open and closed sets are often very useful.
• A set S ⊂ C is open if and only if for every x ∈ S there is an r > 0 such that
B(x, r) ⊂ S.
• A set S ⊂ C is closed if and only if for every convergent sequence whose elements
lie in S the limit also lies in S.
From this one may conclude that closed disks are closed.
As a metric space C is the same as R2 . Hence, by the Heine-Borel theorem A.2.5, a
subset of the complex plane is compact precisely when it is closed and bounded.
1.2.5 Line segments and convex sets. Given two points z and w in C we call the set
{(1 − t)z + tw : 0 ≤ t ≤ 1} the line segment joining z and w. The number |z − w| is called
the length of the segment.
 1.3. SEQUENCES AND SERIES 3

A subset S of C is called convex if it contains the line segment joining x and y whenever
x, y ∈ S. Every disk is convex.
1.2.6 Connectedness. A subset S of C is called connected if S ∩ A ∩ B 6= ∅ whenever
A, B ⊂ C are open sets neither of which covers S but whose union does.
The empty set, all singletons, and all convex sets are connected. To prove the last claim
assume that the convex set S is not connected and pick points x ∈ S ∩ A and y ∈ S ∩ B.
The fact that {t ∈ [0, 1] : (1 − t)x + ty ∈ A} has a supremum leads to a contradiction.
1.2.7 Connected components. If S is a subset of the complex plane and x ∈ S, let C(x)
subsets of S which contain x. Any union of sets in C(x)
be the collection of all connected S
is connected. In particular, T = C∈C(x) C is a connected subset of S. Any larger subset
of S is not connected. T is called a component of S. The components of S are pairwise
disjoint and their union equals S.
The components of open sets are open. There are at most countably many of them.

1.3. Sequences and series

1.3.1 Numerical sequences. A sequence n 7→ zn of complex numbers converges1 if and
only if both of the real sequences n 7→ Re(zn ) and n 7→ Im(zn ) converge. In this case
lim zn = lim Re(zn ) + i lim Im(zn ).
n→∞ n→∞ n→∞
This result allows to extend immediately many theorems proved for sequences of real
numbers to sequences of complex numbers. In particular, if the sequences z and w have
limits, then so do z ± w and zw. In fact,
lim (zn ± wn ) = lim zn ± lim wn ,
n→∞ n→∞ n→∞
lim zn wn = lim zn lim wn ,
n→∞ n→∞ n→∞
and, assuming also wn 6= 0 for all n and limn→∞ wn 6= 0,
zn limn→∞ zn
lim = .
n→∞ wn limn→∞ wn
1.3.2 Cauchy sequences. A sequence z of complex numbers is called a Cauchy2 sequence,
if
∀ε > 0 : ∃N > 0 : ∀n, m > N : |zn − zm | < ε.
Theorem. A sequence of complex numbers converges if and only if it is a Cauchy
sequence.
1.3.3 Series. If z : N → C is a sequence of complex numbers, the sequence
n
X
s : N → C : n 7→ sn = zk
k=1
P∞
is a called the sequence of partial sums of z or a series. We will denote s by k=1P
zk .
∞
If the sequence s of partial sums of z converges to L ∈ C, we say the series k=1 zk
converges to L. We then write (abusing notation slightly)
X∞
zk = lim sn = L.
n→∞
k=1

1Refer to A.1.6 for the definition of convergence.

2Augustin-Louis Cauchy (1789 – 1857)
4 1. THE COMPLEX NUMBERS: ALGEBRA, GEOMETRY, AND TOPOLOGY

Of course, a series (i.e., a sequence of partial sums) may also diverge.

P∞
1.3.4
P∞ Absolute convergence of series. The series k=1 zk is called absolutely convergent
if k=1 |zk | is convergent.
P∞ Every absolutely convergent series
P∞is convergent.
P∞ Suppose that
P∞ n=1 n z converges absolutely and let n=1 wn be a rearrangement of
z
n=1 n . Then n=1 nw converges absolutely, and
∞
X ∞
X
wn = zn .
n=1 n=1
This follows from 1.3.1 since the analogous statement is known to hold for series of real
numbers.
P∞ n
1.3.5 The geometric series. If |z| < 1, then the geometric series n=0 z converges
absolutely. Its limit is 1/(1 − z).
 CHAPTER 2

Complex-valued functions of a complex variable

2.1. Limits and continuity

2.1.1 Limits. Let S be a subset of C, f : S → C a function, and a ∈ C a limit point of S.
We say that f converges to the complex number b as z tends to a if the following statement
is true:
∀ε > 0 : ∃δ > 0 : ∀z ∈ S : 0 < |z − a| < δ ⇒ |f (z) − b| < ε.
Note that this definition can easily be extended to any metric space.
As for sequences, f can converge to at most one value as z tends to a. This value is
then called the limit of f as z tends to a. It is denoted by limz→a f (z).
The results on limits of sequences stated in 1.3.1 hold analogously also for functions.
2.1.2 Continuity. Let S be a subset of C and a ∈ S. A function f : S → C is continuous
at a if the following statement is true:
∀ε > 0 : ∃δ > 0 : ∀z ∈ S : |z − a| < δ ⇒ |f (z) − f (a)| < ε.
Any function is continuous at any isolated point of its domain. Suppose S ⊂ C, f :
S → C, and a ∈ S is a limit point of S. Then f is continuous at a if and only if limz→a f (z)
exists and equals f (a). It follows that sums and products and compositions of continuous
functions are continuous.
Important examples of continuous functions (on all of C) are Re, Im and | · |.

2.2. Holomorphic functions

2.2.1 Differentiation. Let S be a subset of C, f : S → C a function, and a ∈ S a limit
point of S. We say that f is differentiable at a if
f (z) − f (a)
lim
z→a z−a
exists. The limit is called the derivative of f at a and is commonly denoted by f 0 (a). If
f : S → C is differentiable at every point a ∈ S, we say that f is differentiable on S.
This definition is completely analogous to that of a derivative of a function of a real
variable. In fact, when S is a real interval (one of the cases we are interested in) things
are hardly different from Real Analysis; the fact that f assumes complex values is not very
important. However, if S is an open set and z may approach a from many directions, the
existence of the limit has far-reaching consequences.
2.2.2 Basic properties of derivatives. Suppose S, S1 and S2 are subsets of C. Then
the following statements hold true:
(1) A function f : S → C is differentiable at a point a ∈ S if and only if there is a
number F and a function h : S → C which is continuous at a, vanishes there, and
satisfies f (z) = f (a) + F (z − a) + h(z)(z − a). In this case we have F = f 0 (a).
5
6 2. COMPLEX-VALUED FUNCTIONS OF A COMPLEX VARIABLE

(2) If f : S → C is differentiable at a ∈ S, then it is continuous at a.

(3) Sums and products of differentiable functions with common domains are again
differentiable.
(4) The chain rule holds, i.e., if f : S1 → S2 is differentiable at a and g : S2 → C is
differentiable at f (a), then g◦f is differentiable at a, and (g◦f )0 (a) = g 0 (f (a))f 0 (a).
2.2.3 Differentiation with respect to real and complex variables. Suppose S is a
real interval. Then the function f : S → C is differentiable at a ∈ S if and only if Re f and
Im f are differentiable there. In this case, f 0 (a) = (Re f )0 (a) + i(Im f )0 (a).
The situation is completely different if S is an open set in C. For instance, the function
z 7→ f (z) = z 2 is differentiable on all of C and f 0 (z) = 2z. However, its real and imaginary
parts are differentiable only at z = 0. In fact, if a real-valued function defined in B(a, r) is
differentiable at a, then its derivative must be zero.
2.2.4 Holomorphic functions. Let Ω be a non-empty open set. A function f : Ω → C
is called holomorphic on Ω if it is differentiable at every point of Ω. A function which is
defined and holomorphic on all of C is called entire.
It is easy to think that the notions of differentiability and holomorphicity are the same.
To avoid this mistake note that differentiability is a pointwise concept while holomorphicity
is defined on open sets. Nevertheless, we might say that a function is holomorphic at a
point, if it is holomorphic on a neighborhood of that point.
2.2.5 Basic properties of holomorphic functions. Holomorphic functions are contin-
uous on their domain.
Sums, differences, and products of holomorphic functions (on common domains) are
holomorphic. The composition of holomorphic functions is also holomorphic. The usual
formulas hold, including the chain rule.
All polynomials (of a single variable) are entire functions. The function z 7→ 1/z is
holomorphic on C \ {0}. The quotient f /g of two holomorphic functions f, g : Ω → C is
holomorphic on Ω \ {z ∈ Ω : g(z) = 0}.

2.3. Integration
2.3.1 Integrals of complex-valued functions over intervals. Let [a, b] be a closed
interval in R and f a complex-valued function on [a, b]. Then Re f and Im f are real-valued
functions on [a, b]. We say that f is Riemann integrable over [a, b] if Re f and Im f are. The
integral is defined to be
Z b Z b Z b
f= Re f + i Im f.
a a a
The integral is linear and satisfies
Z b Z b
f ≤ |f |.
a a
Rb
To prove this inequality let z = a
f and, when this is different from 0, set α = |z|/z. Then
Rb
consider a αf which is real.
2.3.2 Paths. Let Ω ⊂ C be a non-empty open set and [a, b] a non-trivial bounded interval
in R. A continuous function γ : [a, b] → Ω is then called a path in Ω. If the derivative γ 0
exists and is continuous on [a, b], γ is called a smooth path.
 2.4. CONTOURS 7

The points γ(a) and γ(b) are called initial point and end point of γ, respectively. A
path is called closed if its initial and end points coincide. The range of γ, i.e., the set
{γ(t) : t ∈ [a, b]} is denoted by γ ∗ . Note that γ ∗ is compact and connected. The number
Rb 0
a
|γ (t)|dt is called the length of the smooth path γ.
Let γ : [a, b] → Ω be a smooth path. The coordinate transform ϕ(t) = (1 − t)a + tb,
t ∈ [0, 1], gives rise to a new smooth path γ̃ = γ ◦ ϕ defined on [0, 1] with the same initial
and end points and the same range as γ. Similarly, η(t) = γ(a + b − t), t ∈ [a, b], called the
opposite of γ, also has the same range as γ but switches initial and end points. We use the
notation η = γ.
2.3.3 Integrals along smooth paths. Suppose γ : [a, b] → C is a smooth path in C and
f : γ ∗ → C is continuous. Then the number
Z b
(f ◦ γ)γ 0
a
R
is well-defined. It is called the integral of f along γ and denoted by γ
f . We have the
estimate Z
f ≤ sup(|f |(γ ∗ ))L(γ)
γ

where L(γ) is the length of γ.

∗
R Suppose R function defined on γ and γ̃(t) = γ((1 − t)a + tb). Then
R f is a continuous
R
γ
f = − γ f and γ̃ f = γ f . The latter identity shows that, when we are interested in
integrals, we may always choose a parametrization with any fixed interval as domain.
2.3.4R Example. Let γ : [0, 2π] → C : t 7→ cos(t) + i sin(t) and f : C \ {0} → C : z 7→ 1/z.
Then γ f = 2πi.

2.4. Contours
2.4.1 Contours. A contour in Ω is a finite ordered list of smooth paths in the non-empty
open set Ω. Among the contours we introduce an associative binary relation ⊕ which assigns
to a and b, lists of length m and n, respectively, the concatenation of these lists, denoted
by a ⊕ b, i.e., a listLof length m + n. We will write γ1 γ2 for γ1 ⊕ ( γ2 ).
n Sn
Suppose Γ = k=1 γk is a contour in Ω. Then we denote the compact set k=1 γk∗ by
Γ∗ . Moreover, if f is continuous on Γ∗ we define
Z Xn Z
f= f.
Γ k=1 γk

2.4.2 Closed contours. A contour γ1 ⊕...⊕γn is called closed when there is a permutation
π of {1, ..., n} such that the end point of γk coincides with the initial point of γπ(k) for
k = 1, ..., n.
2.4.3 Connected contours. A particularly important instance of a contour Γ = γ1 ⊕...⊕γn
is when for k = 1, ..., n − 1, the end point of γk coincides with the initial point of γk+1
(perhaps after reordering the indices). A contour of this type is called a connected contour .
The initial point of γ1 is called the initial point of Γ while the end point of γn is called the
end point of Γ. We say a contour connects x to y, if it is a connected contour with initial
point x and end point y.
If Γ is a connected contour, then Γ∗ is connected.
8 2. COMPLEX-VALUED FUNCTIONS OF A COMPLEX VARIABLE

Theorem. Let S be a subset of C. If any two points of S are connected by a connected

contour in S, then S is connected. Conversely, if S is open and connected and x, y are two
points in S, then there is a connected contour in S which connects x and y.
Sketch of proof. The first claim follows from 1.2.7. For the second claim let C(x)
denote the set of all y ∈ S for which there is a contour in S connecting some fixed point x
and y. If S is open, then so are C(x) and S \ C(x). 

2.4.4 Polygonal contours. Let z1 , ..., zn+1 ∈ C. We define γk (t) = (1 − t)zk + tzk+1 for
t ∈ [0, 1] and k = 1, ..., n. Then γk is a smooth path with initial point zk and end point
zk+1 . Note that γk∗ is the line segment joining zk and zk+1 . The length of the line segment
equals the length of γk . Moreover, γ1 ⊕ ... ⊕ γn is a connected contour which we denote by
hz1 , ..., zn+1 i. For instance, when n = 3 and z4 = z1 , Γ = hz1 , z2 , z3 , z1 i is a closed contour
tracing the circumference of a (possibly degenerate) triangle.
2.4.5 Primitives. Suppose Ω is a non-empty open set and F 0 = f on Ω. Then F is called
a primitive of f . If F is a primitive of f and c is a constant, then F + c is also a primitive
of f .
IfRγ : [a, b] → Ω is a smooth path and the continuous function f has a primitive F in Ω,
then γ f = F (γ(b)) − F (γ(a)).
The primitives of the zero function in a non-empty connected open set are precisely the
constant functions.

2.5. Series of functions

2.5.1 Pointwise and uniform convergence. Let S ⊂ C and, for each n ∈ N, let fn be
a function from S to C. The map n 7→ fn is called a sequence of functions. We say that
n 7→ fn converges pointwise to a function f : S → C if for each point z ∈ S the numerical
sequence n 7→ fn (z) converges to f (z). We say that n 7→ fn converges uniformly to a
function f : S → C if, loosely speaking, the rate of convergence does not depend on z ∈ S.
More precisely, n 7→ fn converges to f pointwise, if
∀z ∈ S : ∀ε > 0 : ∃N ∈ R : ∀n > N : |fn (z) − f (z)| < ε
is true. On the other hand, n 7→ fn converges to f uniformly, if
∀ε > 0 : ∃N ∈ R : ∀z ∈ S : ∀n > N : |fn (z) − f (z)| < ε
holds. Note the order of the quantifiers in these statements.
Obviously, uniform convergence implies pointwise convergence but not vice versa.
A series of functions is defined as the sequence of partial sums of functions with a
common domain. Thus the definitions of pointwise and uniform convergence extend also to
series of functions.
2.5.2 Cauchy criterion for uniform convergence. Let S ⊂ C and fn : S → C for
n ∈ N. The sequence of functions n 7→ fn converges uniformly on S if and only if for
all ε > 0 there exists N ∈ R such that for all n, m > N and all z ∈ S it holds that
|fn (z) − fm (z)| < ε.
2.5.3 The Weierstrass1 M -test. Let S ⊂ C and suppose n 7→ gn is a sequence of
complex-valued functions defined on S. Assume that there are non-negative numbers Mn

1Karl Weierstraß (1815 – 1897)

 2.6. ANALYTIC FUNCTIONS 9

P∞
such that |gn (z)| ≤ Mn for all z ∈ S and that the series n=1 Mn converges. Then the
following statements hold:
P∞
(1) The series Pn=1 gn (z) converges absolutely for every z ∈ S.
∞
(2) The series n=1 gn converges uniformly in S.
2.5.4 Uniform convergence and continuity. Let S ⊂ C and n 7→ fn a sequence of
continuous complex-valued functions on S which converges uniformly to a function f : S →
C. Then f is continuous on S.
2.5.5 Uniform convergence and integration. Let [a, b] be a bounded interval in R.
Suppose that n 7→ fn : [a, b] → C is a sequence of functions which are Riemann integrable
over [a, b] and that this sequence converges uniformly to a function f : [a, b] → C. Then f
is Riemann integrable and
Z b Z b
f = lim fn .
a n→∞ a

Of course, the analogous result holds for series: Suppose that n 7→ gn : [a, b] →
P∞C is a
sequence of functions which are Riemann integrable over [a, b] and that the series n=0 gn
converges uniformly to a function g : [a, b] → C. Then g is Riemann integrable and
Z b ∞ Z b
X
g= gn .
a n=0 a

2.6. Analytic functions

2.6.1 Power Series. Let an , n ∈ N0 , be a sequence of complex numbers. Then the series
P ∞ n
n=0 an z , z ∈ C, is called a power series. If the series converges for every z ∈ C we say
that it has infinite radius of convergence. Otherwise there is a non-negative number R such
that the series converges absolutely whenever |z| < R and diverges whenever |z| > R. The
number R is called the radius of convergence of the series. The disk B(0, R) is called the
disk of convergence. P∞ k n
If k is a non-negative integer than the series n=0 n an z have the same radius of
convergence regardless of k.
2.6.2 Power series define holomorphic functions. A power series with a positive or
infinite radius P
of convergence defines a holomorphic function inPits disk of convergence. In
∞ ∞
fact, if f (z) = n=0 an (z − z0 )n for |z − z0 | < R, then f 0 (z) = n=1 nan (z − z0 )n−1 in the
same disk, i.e., the power series may be differentiated term by term and f is holomorphic
on B(z0 , R).
P∞
Sketch of proof. Let g(w) = n=1 nan (w−z0 )n−1 which converges for w ∈ B(z0 , R)
by 2.6.1. Set r = max{|z − z0 |, |w − z0 |} and use twice the identity
n−1
X
an − bn = (a − b) ak bn−k−1
k=0

with a = z − z0 and b = w − z0 to show that

∞
f (z) − f (w) X n(n − 1)
− g(w) ≤ |z − w| |an |rn−2 .
z−w n=2
2

10 2. COMPLEX-VALUED FUNCTIONS OF A COMPLEX VARIABLE

P∞ We say that a function f : B(z0 , R) → C is represented by a

2.6.3 Analytic functions.
power series if f (z) = n=0 an (z − z0 )n for all z ∈ B(z0 , R).
Let Ω be a non-empty open set. A function f : Ω → C is called analytic in Ω, if Ω is a
union of open disks in each of which f is represented by a power series.
For example, z 7→ 1/(1 − z) is analytic in the (open) unit disk as well as in all of C \ {1}.
Theorem. Every analytic function is holomorphic.
The central and most astonishing theorem of complex analysis is that the converse is
also true, i.e., every holomorphic function is analytic, see Theorem 3.3.2 below.
P∞
2.6.4 Taylor series. Suppose f (z) = n=0 an (z−z0 )n for |z−z0 | < R, i.e., f : B(z0 , R) →
C is analytic. Then f is infinitely often differentiable and an = f (n) (z0 )/n!. We call
∞
X f (n) (z0 )
(z − z0 )n
n=0
n!
the Taylor2 series of f about z0 .
2.6.5 The exponential function. The series
∞
X zn
exp(z) =
n=0
n!
converges absolutely for every z ∈ C and uniformly in every closed disk about the origin. It
represents therefore an entire function, called the exponential function. It is an extension of
the exponential function defined in Real Analysis.
Theorem. If t ∈ R, then exp(it) = cos(t) + i sin(t); in particular, exp(0) = 1. Moreover
exp0 (z) = exp(z), and exp(a + b) = exp(a) exp(b). The exponential function has no zeros.
It is periodic with period 2πi. Any period of the exponential function is an integer multiple
of 2πi. In particular, exp(z) = 1 if and only if z is an integer multiple of 2πi.
2.6.6 Analytic functions defined by integrals. Suppose ψ and φ are continuous
complex-valued functions on [a, b]. Then the function f defined by
Z b
ψ(t)
f (z) = dt
a φ(t) − z
is analytic in Ω = C \ φ([a, b]). In fact, if z0 ∈ Ω, if r = inf{|φ(t) − z0 | : t ∈ [a, b]}, and if
z ∈ B(z0 , r), then
X∞
f (z) = an (z − z0 )n
n=0
where Z b
ψ(t)dt
an = .
a (φ(t) − z0 )n+1
Sketch of proof. Since φ(t) − z = (φ(t) − z0 )(1 − (z − z0 )/(φ(t) − z0 )) we may use
properties of the geometric series which is uniformly convergent on appropriate disks. 

2Brook Taylor (1685 – 1731)

 CHAPTER 3

Cauchy’s theorem and some of its consequences

3.1. The index of a point with respect to a closed contour

3.1.1. Let Γ be a contour in C and Ω = C \ Γ∗ . Since Γ∗ is compact Ω is open. Ω has
precisely one unbounded connected component. This component contains the set B(0, R)c
when R is chosen so that Γ∗ ⊂ B(0, R).
3.1.2. If γ : [a, b] → C is a smooth path and z ∈ C \ γ ∗ , define F : [a, b] → C by
Z s
γ 0 (t)

F (s) = exp dt .
a γ(t) − z

Then s 7→ F (s)/(γ(s) − z) is constant.

3.1.3 The index of a point with respect to a closed contour. Let Γ be a closed
contour in C and Ω = C \ Γ∗ . The function IndΓ defined by
Z
1 du
IndΓ (z) =
2πi Γ u − z
on Ω is an analytic function assuming only integer values. It is constant on each connected
component of Ω and, in particular, zero near infinity.

Sketch of proof. Let Γ = γ1 ⊕ ... ⊕ γn where the γk are smooth paths defined on
[ak , bk ], respectively. Using 2.6.6 we see that IndΓ is analytic. Now define Fk for each γk as
in 3.1.2 so that Fk /(γk − z) is constant. In particular, Fk (bk ) = (γk (bk ) − z)/(γk (ak ) − z).
Hence
n
Y γk (bk ) − z
exp(2πi IndΓ (z)) = = 1.
γk (ak ) − z
k=1
By Theorem 2.6.5 IndΓ (z) is an integer. Continuity shows that the value of IndΓ is constant
on sufficiently small disks within Ω. Hence the sets {z ∈ Ω : IndΓ (z) = k} are open for every
k ∈ Z. 

3.1.4 Example. Consider the smooth paths γ1 (t) = exp(it), γ2 (t) = exp(it)/2, and γ3 (t) =
exp(−it)/2 all defined on [0, 2π] and the contours Γ1 = γ1 , Γ2 = γ1 ⊕ γ2 , and Γ3 = γ1 ⊕ γ3 .
The contours are closed so that IndΓk (z) is defined for z = 0, z = 3i/4, and z = −2.
The number IndΓ (z) is also called the winding number of the contour Γ around z. The
above examples hint at the origin of the name.

3.2. Cauchy’s theorem

3.2.1 Cauchy’s theorem for functions with primitives. Suppose Ω is a non-empty
open set,R f a continuous function on Ω which has a primitive F . If Γ is a closed contour in
Ω, then Γ f = 0.
11
12 3. CAUCHY’S THEOREM AND SOME OF ITS CONSEQUENCES

SketchR of proof. Suppose Γ = γ1 ⊕ ... ⊕ γn with smooth paths γk defined on [ak , bk ].

Evaluate γk f using 2.4.5 and recall that γk (bk ) = γπ(k) (aπ(k) ) for some permutation π of
the set {1, ..., n}. 

3.2.2 Cauchy’s theorem for integer powers. Suppose n ∈ Z. Unless n = −1 the

power function z 7→ z n has the primitive z 7→ z n+1 /(n + 1) in either Ω = C \ {0} or Ω = C
depending on whetherRn is negative or not. Consequently, if Γ is a closed contour in Ω and
if n ∈ Z \ {−1}, then Γ z n dz = 0.
The exceptional case n = −1 gives Rrise to interesting complications which we will discuss
later. Recall, though, from 2.3.4 that γ dz/z = 2πi when γ(t) = exp(it) for t ∈ [0, 2π].
3.2.3 Triangles. Let a, b, c be three pairwise distinct complex numbers. The set ∆ =
{t1 a + t2 b + t3 c : tj ∈ [0, 1], t1 + t2 + t3 = 1} is called a solid triangle with vertices a, b, and
c. The line segments joining each pair of vertices are called edges of the triangle. They are
obtained by setting t1 , t2 , and t3 in turn equal to zero and are thus subsets of ∆.
The diameter of a triangle is the length of its longest edge. To show this set D =
max{|c − b|, |b − a|, |a − c|}, x = t1 a + t2 b + t3 c, and y = s1 a + s2 b + s3 c. Note that
|x − y| = |(t1 − s1 )(a − c) + (t2 − s2 )(b − c)| = |(t2 − s2 )(b − a) + (t3 − s3 )(c − a)| =
|(t3 − s3 )(c − b) + (t1 − s1 )(a − b)|. Two of the numbers t1 − s1 , t2 − s2 , and t3 − s3 are of
the same sign unless one of them is zero.
3.2.4 Quartering a triangle. Let ∆ be a solid triangle with vertices a, b, and c. If c0 =
(a + b)/2, b0 = (a + c)/2, and a0 = (b + c)/2 (the midpoints of the edges), the sets {a0 , b0 , c0 },
{a, b0 , c0 }, {a0 , b, c0 }, and {a0 , b0 , c} define respectively four new triangles ∆0 , ..., ∆3 . Their
union is ∆ and the intersection of any two of these is either a midpoint a0 , b0 , or c0 or one
of the segments joining two midpoints. To show this note that ∆0 = {t1 a + t2 b + t3 c : tj ∈
[0, 1/2], t1 +t2 +t3 = 1} and ∆1 = {t1 a+t2 b+t3 c : t1 ∈ [1/2, 1], t2 , t3 ∈ [0, 1/2], t1 +t2 +t3 = 1}
with similar expressions for ∆2 and ∆3 .
Also ∆0 , ..., ∆3 are congruent and are similar to ∆ itself. The circumference and the
diameter of each one of the ∆k is half of that of ∆.
3.2.5 Integration along a triangle. Suppose a, b, c are pairwise distinct complex
numbers. Suppose a0 ∈ hb, ci∗ , b0 ∈ hc, ai∗ , and c0 ∈ ha, bi∗ . Let Γ = ha, b, c, ai and
0 0 0 0 0 0 0 0 0 0 ∗
R = ha,R c , b , ai ⊕ hb, a , c , bi ⊕ hc, b , a , ci ⊕ hc , a , b , c i. If f is continuous on Λ , then
Λ
Γ
f = Λ f.
3.2.6 Solid triangles are compact. To prove this claim let us first consider the solid
triangle with vertices 1, i and 0. The points of the sequence zn = sn + itn are in the triangle
precisely when sn , tn , and un = 1 − sn − tn are in [0, 1]. Assuming that zn is convergent,
1.3.1 gives that sn and tn and hence un are convergent. Their limits s, t, and u are in [0, 1]
and satisfy s + t + u = 1 implying that zn converges to a point in the triangle. Thus the
triangle is a closed set. It is, of course, bounded and, by the Heine-Borel theorem, compact.
A general triangle is the image of this triangle under a continuous map and also compact
according to A.2.2.
3.2.7 Cauchy’s theorem for triangles. Suppose Ω is a non-empty open set, z0 a point
in Ω, and f : Ω → C a continuous function which is holomorphicR on Ω \ {z0 }. If ∆ is a solid
triangle in Ω with vertices a, b, and c, let Γ = ha, b, c, ai. Then Γ f = 0.

Sketch of proof. Assume first that z0 6∈ ∆. T∞Let ∆0 = ∆. Given ∆n let ∆n+1 be one
of the four triangles constructed in 3.2.3. Then n=1 ∆n consists of one point only, say w.
 3.3. CONSEQUENCES OF CAUCHY’S THEOREM 13

By 2.2.2 the function h defined by f (z) = f (w) + f 0 (w)(z − w) + h(z)(z − w) is continuous

and
R vanishes at w. Hence, given ε > 0 and using 3.2.2 and 2.3.3, there is an n such that
| Γn f | ≤ 3εD2 4−n when D denotes the diameter of ∆. By making, at each step, the
right choice of triangle in the above recursion one can also show that | Γn f | ≥ 4−n | Γ f |.
R R

This settles the first case. If z0 is a vertex of ∆, we repeat the Rconstruction

R above but
choosing ∆n always to be the one which has z0 as a vertex so that Γ f = Γn f . Finally, if
z0 ∈ ∆ \ {a, b, c} construct three triangles in ∆ which have z0 as a vertex. 
3.2.8 Cauchy’s theorem for convex sets. Suppose Ω is a non-empty open convex set,
z0 a point in Ω, and f : Ω → C a continuous
R function which is holomorphic on ΩR\ {z0 }. Fix
a ∈ Ω and define F : Ω → C by F (z) = ha,zi f . Then F is a primitive of f and Γ f = 0 for
every closed contour Γ in Ω.

3.3. Consequences of Cauchy’s theorem

3.3.1 Cauchy’s integral formula for convex sets. Suppose Ω is a non-empty open
convex set, f : Ω → C a holomorphic function on Ω, and z is a point in Ω. Then
(
f (u)−f (z)
u−z if u 6= z
g(u) =
f 0 (z) if u = z
R
satisfies the hypothesis of Cauchy’s theorem for convex sets 3.2.8 so that Γ g = 0 when Γ
is a closed contour in Ω.
Therefore, if z ∈ Ω \ Γ∗ , then
Z
1 f (u)
f (z) IndΓ (z) = du.
2πi Γ u − z
This identity is known as Cauchy’s integral formula.
3.3.2 Holomorphic functions are analytic. The conclusions in 3.3.1 and 2.6.6 imply
the central theorem of Complex Analysis.
Theorem. A holomorphic function is analytic on its domain of definition.
From now on the words holomorphic and analytic may be considered synonymous (many
authors do not ever make a distinction between them).
Corollary. Any holomorphic function may be expanded into a Taylor series about
any point in its domain.
3.3.3 RMorera’s1 theorem. Let Ω be a non-empty open set. If f : Ω → C is continuous
and if Γ f = 0 for every triangular contour Γ ∈ Ω, then f is holomorphic in Ω.
3.3.4 Sequences of holomorphic functions. Let Ω be a non-empty open set. Suppose
n 7→ fn is a sequence of holomorphic functions defined on Ω which converges uniformly to
f : Ω → C. Then f is holomorphic.
3.3.5 General integral formulas for convex sets. Suppose Ω is a non-empty open
convex set and f : Ω → C a holomorphic function on Ω. If Γ is a closed contour in Ω,
z ∈ Ω \ Γ∗ , and n ∈ N0 , then
Z
(n) n! f (u)
f (z) IndΓ (z) = du.
2πi Γ (u − z)n+1
1Giacinto Morera (1856 – 1909)
14 3. CAUCHY’S THEOREM AND SOME OF ITS CONSEQUENCES

Sketch of proof. Using 2.6.4 we may express f (k) (z0 ) in terms of a Taylor coefficient
and then compute that coefficient employing a combination of 3.3.1 and 2.6.6. 

3.3.6 Cauchy’s estimate. If f : B(z0 , R) → C is a holomorphic function, then the radius

of convergence of its Taylor series about z0 is at least equal to R. If |f (z)| ≤ M whenever
z ∈ B(z0 , R), then
|f (n) (z0 )| M
|an | = ≤ n
n! R
for all n ∈ N0 .
3.3.7 Liouville’s2 theorem. Every bounded entire function is constant.

P3.3.8
∞
Zeros and their order. Let z0 be a zero of the holomorphic function f and
n
n=0 n (z − z0 ) the Taylor series of f about z0 . Then z0 is an isolated point in the
a
set of all zeros of f unless an = 0 for all n ∈ N0 . If z0 is an isolated zero of f , the number
m = min({k ∈ N : ak 6= 0}) is called the order or the multiplicity of z0 as a zero of f .
In this case we have f (z) = (z − z0 )m g(z) where g is holomorphic in the domain of f and
g(z0 ) 6= 0.
3.3.9 The set of zeros of a holomorphic function. Let Ω be a non-empty open
connected subset of C and f a holomorphic function on Ω. Let Z(f ) be the set of zeros of
f , i.e., Z(f ) = {a ∈ Ω : f (a) = 0}, and A the set of limit points of Z(f ) in Ω. Then A and
Ω \ A are open and, in consequence, either Z(f ) = Ω or else Z(f ) is a set of isolated points.
3.3.10 Analytic continuation. Let f and g be holomorphic functions on Ω, a non-empty
open connected subset of C. Then the following statements hold:
(1) If f (z) = g(z) for all z in a set which contains a limit point of itself, then f = g.
(2) If f (n) (z0 ) = g (n) (z0 ) for some z0 and all n ∈ N0 , then f = g.
Suppose f is a holomorphic function on Ω and that Ω0 is another open connected set
which intersects Ω. Then there is at most one holomorphic function on Ω∪Ω0 which coincides
with f on Ω. This function (if it exists) is called the analytic continuation of f to Ω ∪ Ω0 .

3.4. The global version of Cauchy’s theorem

3.4.1 Continuity of parametric integrals. Let Ω be an open set,
R Γ a contour in C, and
ϕ a continuous complex-valued function on Γ∗ × Ω. Define g(z) = Γ ϕ(·, z) on Ω. Then g
is continuous on Ω.

Sketch of proof. By A.3.1 Γ∗ ×B(z0 , r) is compact and A.2.3 gives that ϕ|Γ∗ ×B(z0 ,r)
is uniformly continuous so that, for all t, |ϕ(γ(t), z) − ϕ(γ(t), z0 )| < ε if |z − z0 | is sufficiently
small. 

3.4.2 Differentiation under the integral. Let Ω, Γ, ϕ, and g be defined as in 3.4.1. Fur-
thermore suppose that ϕ(u, ·) is holomorphic in Ω for each u ∈ Γ∗ and denote its derivative
by ψ(u, ·), so that ϕ(u, ·) is a primitive of ψ(u, ·) for every fixed u ∈ Γ∗ . If ψ : Γ∗ × Ω → C
is continuous, then g is holomorphic in Ω.

2Joseph Liouville (1809 – 1882)

 3.4. THE GLOBAL VERSION OF CAUCHY’S THEOREM 15

R
Sketch of proof. Note that we have ϕ(u, z)−ϕ(u, z0 ) = hz0 ,zi ψ(u, ·) and let h(z0 ) =
R
Γ
ψ(·, z0 ). Then
n Z bk Z 1
g(z) − g(z0 ) X
− h(z0 ) ≤ |ψ(γk (t), u(s)) − ψ(γk (t), z0 )|ds|γk0 (t)|dt
z − z0 ak 0
k=1

and this is less than ε times the total length of Γ. 

3.4.3 A special (but important) case. Suppose Ω is an open subset of C and f is a
holomorphic function on Ω. Define ϕ, ψ : Ω × Ω → C by
(
(f (u) − f (z))/(u − z) if u =
6 z,
ϕ(u, z) =
f 0 (z) if u = z
and (
(f (u) − f (z) − f 0 (z)(u − z))/(u − z)2 if u 6= z,
ψ(u, z) = 1 00
2 f (z) if u = z.
Then ϕ and ψ are continuous on Ω × Ω and ψ(u, ·) is the derivative of ϕ(u, ·).
Sketch of proof. Fix (u0 , z0 ) ∈ Ω × Ω. There is no difficulty if u0 6= z0 . If u0 = z0
R1 R1 R1
we use ϕ(u, z) = 0 f 0 ((1 − t)z + tu)dt and ψ(u, z) = 0 t 0 f 00 ((1 − st)z + stu)dsdt. 
3.4.4 Cauchy’s integral formula, global version. Suppose Ω is an open subset of the
complex plane, f is a holomorphic function on Ω, and Γ is a closed contour in Ω such that
IndΓ (w) = 0 whenever w is not an element of Ω. Then
Z
1 f (u)
f (z) IndΓ (z) = du
2πi Γ u − z
for every z ∈ Ω \ Γ∗ .
0
R
Sketch of proof. On Ω define R g(z) = Γ ϕ(u, z)du with ϕ as in 3.4.3. On Ω =
{z ∈ C : IndΓ (z) = 0} define h(z) = Γ f (u)/(u − z)du. Then create an entire function by
showing that h = g on Ω ∩ Ω0 . This function must be zero. 
3.4.5 Cauchy’s theorem, R global version. Suppose Ω, f , and Γ satisfy the same hy-
potheses as in 3.4.4. Then Γ f = 0.
3.4.6 Laurent3 series. Suppose 0 ≤ r1 < r2 and let f be holomorphic in the annulus
Ω = {z ∈ C : r1 < |z − a| < r2 }. Then, f can be expressed by a Laurent series, i.e.,
∞
X
f (z) = an (z − a)n .
n=−∞

This is actually an abbreviation for

∞
X ∞
X
f (z) = an (z − a)n + a−n (z − a)−n ,
n=0 n=1

i.e., convergence of a Laurent series requires convergence of both series of positive and
negative powers.

3Pierre Laurent (1813 – 1854)

16 3. CAUCHY’S THEOREM AND SOME OF ITS CONSEQUENCES

The coefficients an are given by the integrals

Z
1 f (u)
an = du
2πi γ (u − a)n+1
where γ = a + r0 exp(it), t ∈ [0, 2π] and r0 ∈ (r1 , r2 ).
Sketch of proof. Fix z ∈ Ω. For j = 1, 2 let γj (t) = a + rj0 exp(it) for t ∈ [0, 2π]
and suitably chosen rj0 . Then use Cauchy’s integral formula for Γ = γ2 γ1 . To find
an for n ≥ 0 employ 2.6.6; for the others use a variant of its proof after noting that
u − z = −(z − a)(1 − (u − a)/(z − a)). In the formula for an the radii r10 and r20 may be
replaced by any r0 ∈ (r1 , r2 ). 
The Taylor series of an analytic function is, of course, a special case of a Laurent series.
3.4.7 Cauchy’s theorem and integral formula for simply connected open sets.
A connected open set Ω in the complex plane is called simply connected if C \ Ω has no
bounded component. The set C \ {0} is not simply connected but C \ (−∞, 0] is.
If Ω is a non-empty open simply connected subset of C and Γ a closed contour in Ω,
then IndΓ (w) = 0 whenever w is not an element of Ω. Consequently, if f is a holomorphic
function on Ω, then Z
1 f (u)
f (z) IndΓ (z) = du
2πi Γ u − z
for every z ∈ Ω \ Γ∗ and Z
f = 0.
Γ
 CHAPTER 4

Isolated singularities

4.1. Classifying isolated singularities

Throughout this section Ω denotes an open subset of the complex plane.
4.1.1 Isolated singularities. Suppose z0 ∈ Ω and f : Ω \ {z0 } → C is holomorphic. Then
z0 is called an isolated singularity of f .
The following are typical examples: z 7→ (z 2 − z02 )/(z − z0 ), z 7→ 1/(z − z0 ), and
z 7→ exp(1/(z − z0 )).
4.1.2 Laurent series about an isolated singularity. A punctured disk {z ∈ C : 0 <
|z − z0 | < r} is a special case of an annulus and thus a holomorphic function defined on it
has a Laurent series expansion. In particular, if z0 is an isolated singularity of f , then there
is a punctured disk {z ∈ C : 0 < |z − z0 | < r} (possibly the punctured plane) on which
∞
X
f (z) = an (z − z0 )n .
n=−∞

We call the point z0 a removable singularity, if an = 0 for all n < 0. It is called a pole of
order m or of multiplicity m (where m > 0), if an = 0 for all n < −m and a−m 6= 0. The
point z0 is called an essential singularity, if it is neither a pole nor a removable singularity,
i.e., if the set {n ∈ Z : an 6= 0} is not bounded below.
4.1.3 Removable singularities. If z0 is a removable singularity of a function f with a
Laurent expansion on the punctured disk B(z0 , r) \ {z0 }, then there is an analytic contin-
uation of f to B(z0 , r). In particular, in this case f has a limit at z0 . Conversely, if z0
is an isolated singularity of f and limz→z0 (z − z0 )f (z) = 0, then z0 is in fact a removable
singularity of f . To see this define h(z0 ) = 0 and h(z) = (z − z0 )f (z) for z 6= z0 . Then one
may show that h is holomorphic near z0 .
4.1.4 Poles. The following statements about poles hold.
(1) The point z0 is a pole of order m of a function f if and only if 1/f has an analytic
continuation for which z0 is a zero of order m.
(2) If z0 is a pole of f and M is any positive real number, then there is a positive δ
such that |f (z)| ≥ M for all z ∈ B(z0 , δ) \ {z0 }.
(3) If z0 is an isolated singularity of f and there is a natural number m such that
limz→z0 (z − z0 )m+1 f (z) = 0, then z0 is either a removable singularity of f or else
a pole of order at most m.
(4) If f has a pole of order m at z0 , then there are complex numbers c1 , ..., cm so that
m
X ck
z 7→ f (z) −
(z − z0 )k
k=1

17
18 4. ISOLATED SINGULARITIES

Pm k
has a removable singularity at z0 . The sum k=1 ck /(z−z0 ) is called the principal
part or singular part of f at z0 .
4.1.5 Meromorphic functions. Let P be a set of isolated points in Ω without a limit
point in Ω and f a holomorphic function on Ω \ P . If no point of P is an essential singularity
of f , then f is called meromorphic on Ω.
4.1.6 The Casorati1-Weierstrass theorem. Suppose f is a holomorphic function defined
on B 0 = B(z0 , r) \ {z0 }. If z0 is an essential singularity of f , then f (B 0 ) is dense in C.

Sketch of proof. Assume on the contrary that there is a a ∈ C and δ > 0 such that
B(a, δ) ∩ f (B 0 ) = ∅. Then consider the function g(z) = 1/(f (z) − a) on B 0 which has a
removable singularity at z0 . If z0 is a zero of g of order m ∈ N then f has a pole of order
m. If m = 0, then f has a removable singularity. 

4.2. The calculus of residues

4.2.1 Residues. P
If the holomorphic function f has the isolated singularity z0 and the
∞
Laurent expansion n=−∞ an (z − z0 )n , then the number
Z
1
a−1 = f,
2πi γ
where γ traces counterclockwise a sufficiently small circle about z0 , is called the residue of
f at z0 and is denoted by Res(f, z0 ).
If limz→ z0 (z − z0 )f (z) = a 6= 0, then z0 is a simple pole, i.e., a pole of order one, of f
and a = Res(f, z0 ).
4.2.2 The residue theorem. Suppose Ω is an open subset of the complex plane, f is
a holomorphic function on Ω0 = Ω \ {z1 , ..., zn }, and Γ is a closed contour in Ω0 such that
IndΓ (w) = 0 whenever w is not an element of Ω. Then
n Z
X 1
Res(f, zk ) IndΓ (zk ) = f.
2πi Γ
k=1

Sketch of proof. Let mk = IndΓ (zkL R k t) for t ∈ [0, 2π] and

), γk (t) = zk + Rrk exp(im
n
sufficiently small but positive rk , and Γ0 = k=1 γk . Then Γ f = Γ0 f . 

4.2.3 Some examples.

Z ∞
1
dx
−∞ 1 + x4
Z ∞
1
dx
−∞ exp(x) + exp(−x)
Z 2π
1
dx
0 2 + sin x
The following hints might be useful: For the first integral consider a contour made up
from the interval [−R, R] and a semi-circle of radius R. For the second one consider a
rectangle with vertices ±R and ±R + πi. Finally, for the last integral let u = exp(it) and
note that 2i sin(t) = u − 1/u.
1Felice Casorati (1835 – 1890)
 4.2. THE CALCULUS OF RESIDUES 19

4.2.4 Counting zeros and poles. Suppose f is a non-zero meromorphic function on the
open set Ω, Z is the set of its zeros, and P is the set of its poles. For any point z0 ∈ Ω there is
a unique integer M (z0 ) denoting the smallest index for which the coefficient in the Laurent
expansion of f about z0 is non-zero. In fact, if z0 is a zero or a pole of f its multiplicity is
M (z0 ) or −M (z0 ), respectively.
Let Γ be a closed contour in Ω\(Z ∪P ) such that IndΓ (w) = 0 whenever w ∈ C\Ω. There
are at most finitely many zeros and poles in those components of Ω which have a non-zero
index. To see this note that A ∪ Γ∗ is a compact subset of Ω if A = {z ∈ C : IndΓ (z) 6= 0}.
It follows that Z 0
1 f X
= M (z) IndΓ (z).
2πi Γ f
z∈Z∪P

In particular, if f is holomorphic on Ω and IndΓ assumes only the values 0 and 1,

1
R f0
then 2πi Γ f
is the number of zeros (counted according to their multiplicities) in the set
{z ∈ C : IndΓ (z) = 1}.
4.2.5 Rouché’s2 theorem. Let f and g be meromorphic functions on the open set Ω
and assume that B(a, r) ⊂ Ω. Let nz (f ) be the number of zeros of f in B(a, r), counted
according to their multiplicity. Similarly, np (f ) is the count of poles, and nz (g) and np (g)
are the analogous quantities for g. If no zero or pole lies on the circle C = {z : |z − a| = r}
and if |f (z) − g(z)| < |f (z)| + |g(z)| for all z ∈ C, then nz (f ) − np (f ) = nz (g) − np (g).
Sketch of proof. Let γ parameterize the circle and set h = f /g. Then γ h0 /h =
R

2πi Indh◦γ (0) = 0 since (h ◦ γ)∗ ∩ (−∞, 0] = ∅. 

4.2.6 Behavior of a holomorphic functions near a zero. Suppose f : Ω → C is
holomorphic and z0 is a zero of order m of f . Then there are positive numbers r and R such
that the following statement is true. If |w| < R, then f − w has precisely m zeros (counting
multiplicities) in B(z0 , r).
Sketch of proof. Note that f (z) = a(z − z0 )m h(z) where h is holomorphic on Ω,
h(z0 ) = 1, and a 6= 0. Since h − 1 is bounded near z0 and vanishes at z0 , there is an
r > 0 such that |h(z) − 1| ≤ 1/2 for |z − z0 | ≤ r. Now let F (z) = a(z − z0 )m h(z) − w and
G(z) = a(z − z0 )m − w. If |z − z0 | = r and |w| < R = |a|rm /2, we have |F (z) − G(z)| ≤
|a|rm /2 < |a(z − z0 )m − w| ≤ |G(z)| + |F (z)|. Now apply Rouché’s theorem and recall that
w/a has precisely m m-th roots. 
4.2.7 The open mapping theorem. Suppose f is a non-constant holomorphic function
on the open and connected set Ω. Then f (Ω0 ) is open whenever Ω0 is an open subset of Ω.
Sketch of proof. Suppose f (z0 ) = w0 ∈ f (Ω0 ). Let F = f − w0 and G = f − w.
Then |F (z0 + r exp(it))| ≥ δ > 0 for some r > 0 and all t ∈ [0, 2π]. The proof follows once
we determine how close w has to be to w0 so that Rouché’s theorem 4.2.5 applies to F and
G. 
4.2.8 The maximum modulus theorem. Suppose f is a holomorphic function on the
open and connected set Ω and |f (z0 )| is a local maximum of |f |. Then f is constant.

2Eugène Rouché (1832 – 1910)

 CHAPTER 5

A zoo of functions

In this chapter we investigate briefly the most elementary functions of analysis. Polyno-
mials and exponential functions have been introduced earlier since they are too important
to postpone their use.

5.1. Polynomial functions

5.1.1 Polynomial functions. If n is a non-negative integer and a0 , a1 , . . ., an are complex
numbers, then the function p : C → C defined by
n
X
p(z) = ak z k
k=0

is called a polynomial (of a single variable). The integer n is called the degree of p if an is
different from 0. It is then called the leading coefficient of p. The zero function is also a
polynomial but no degree is assigned to it. Any polynomial is an entire function.
If p and q are polynomials of degree n and k, respectively, then p + q and pq are also
polynomials. The degree of p + q is the larger of the numbers n and k unless n = k in which
case the degree is at most n. The degree of pq equals n + k.
5.1.2 The fundamental theorem of algebra. Suppose p is a polynomial of degree n ≥ 1.
Then there exist numbers a and z1 , ..., zn (not necessarily distinct) such that
n
Y
p(z) = a (z − zk ).
k=1

To prove this use either Rouchés theorem or the fact that 1/p would be entire if p had no
zeros.
5.1.3 Zeros of polynomials. The coefficients ak of a monic polynomial are given as
symmetric polynomials in terms of the zeros. As a matter of fact,
X
an−k = (−1)k zj1 ...zjk .
1≤j1 <...<jk ≤n
Pn Qn
In particular, an−1 = − k=1 zk and a0 = (−1)n k=1 zk . They vary continuously with the
zeros.
That the converse is also true can be proved Pwith the aid of Rouché’s
Pn theorem. The
n
precise statement is as follows: Suppose f (z) = k=0 ak z k and g(z) = k=0 bk z k are two
monic polynomials of degree at most n. If z0 is the only zero of f in B(z0 , r) and if the
multiplicity of z0 is m then the following holds: For every ε ∈ (0, r) there is a δ > 0 such
that, if |ak − bk | < δ for k = 0, ..., n, then g has precisely m zeros in B(z0 , ε) (counting
multiplicities).
21
22 5. A ZOO OF FUNCTIONS

5.1.4 Among the entire functions only polynomials grow like powers. Suppose f
is an entire function and k 7→ rk is an increasing unbounded sequence of positive numbers.
Furthermore, suppose there are positive real numbers N and C such that |f (z)| ≤ C|z|N
for all z lying on any of the circles |z| = rk . Then f is a polynomial whose degree is less
then or equal to N .
It is in fact enough to require Re(f (z)) ≤ C|z|N to arrive at the same conclusion. To see
R 2π
this use the Laurent expansion 3.4.6 of f to obtain 2πrkn an = 0 f (γk (t)) exp(−int)dt for
R 2π
n ∈ Z. Therefore, using that an = 0 if n < 0, we get πrkn an = 0 Re f (γk (t)) exp(−int)dt
R 2π R 2π
for n ∈ N and 2π Re a0 = 0 Re f (γk (t))dt. We also have 0 CrkN exp(−int)dt = 0 so that
2π
πrkn |an | ≤ 0 (CrkN − Re f (γk (t)))dt = 2πCrkN − 2π Re a0 . Hence an = 0 when n > N .
R

Thus the real (or the imaginary part) of a polynomial grows roughly at the same pace as
the modulus. This is, in fact, a special case of a more general result, the Borel-Carathéodory1
theorem, which allows to estimate the modulus of an entire function by its real (or imaginary)
part.

5.2. Rational functions

5.2.1 The extended complex plane. We define the extended complex plane C∞ as the
union of C and another point, not in C, which we denote by ∞. We call a subset of C∞
open, if it is either an open subset of C or else the union of {∞} and the complement of a
compact subset of C. The set of all open subsets of C∞ is a topology for C∞ .
Under this topology C∞ is a compact space sometimes called the one-point compactifi-
cation of C.
5.2.2 The Riemann sphere. The set S 2 = {(x, y, z) ∈ R3 : x2 + y 2 + z 2 = 1} is called
the unit sphere. The point (0, 0, 1) is called the north pole of the sphere. Let (x, y, z) be a
point in S 2 other than the north pole. The map t 7→ (tx, ty, tz + 1 − t), t ∈ R, describes
a straight line passing through the north pole and the point (x, y, z). This line intersects
the equator plane (i.e., the plane where the third component is 0) in the unique point
(x/(1 − z), y/(1 − z), 0). We have thus defined the map
x + iy
π : S 2 \ {(0, 0, 1)} → C : (x, y, z) 7→ .
1−z
This map, called the stereographic projection, is bijective. Points in the southern hemi-
sphere correspond to points in the unit disk, points on the equator correspond to points
on the unit circle, and points in the northern hemisphere correspond to points outside the
closed unit disk.
With this interpretation the sphere S 2 is called the Riemann sphere. Note that points
on the sphere close to the north pole are those whose image points in the plane have large
absolute values. The Riemann sphere is homeomorphic to the extended complex plane C∞ .
5.2.3 Continuity and the extended complex plane. We can view a meromorphic
function on Ω as a function from Ω to C∞ by defining
(
f (u) if u is not a pole
f˜(u) =
∞ if u is a pole.

Note that f˜ is continuous at every point of Ω, in the sense of A.1.2, even the poles.
1Constantin Carathéodory (1873 – 1950)
 5.3. EXPONENTIAL AND TRIGONOMETRIC FUNCTIONS 23

5.2.4 Rational functions. Let p and q be polynomials and Q = {z ∈ C : q(z) = 0}.

Assume Q 6= C, i.e., q is not the zero polynomial. Then, the function r : (C \ Q) → C given
by
p(z)
r(z) =
q(z)
is called a rational function.
As a meromorphic function on C a rational function may also be interpreted as a
continuous function from C to C∞ . We can even extend the domain of r to C∞ and
define r(∞) = limz→∞ r(z) (which may again be infinity).
5.2.5 Möbius transforms. A rational function of the type z 7→ (az + b)/(cz + d) where
ad − bc 6= 0 is called a Möbius2 transform. The set of all Möbius transforms forms a group
under composition. Special cases of Möbius transforms are the translations z 7→ z + b, the
dilations and rotations z 7→ az where a > 0 and |a| = 1, respectively, and the inversion
z 7→ 1/z. Any Möbius transform is a composition of at most five of these.
A Möbius transform may be interpreted as a homeomorphism from the Riemann sphere
to itself. Conversely, an injective meromorphic function on C is a Möbius transform.
Möbius transforms have many interesting properties to which an entire chapter might
be devoted.

5.3. Exponential and trigonometric functions

5.3.1 The exponential function. We defined the exponential function already in 2.6.5.
Hence most of the following will be a repetition of results obtained there. The function
exp : C → C defined by
∞
X zk
exp(z) =
k!
k=0

is called the exponential function. It is an entire function and exp0 (z) = exp(z).
If x, y ∈ C, then exp(x + y) = exp(x) exp(y). In particular, exp(z) is never zero and
exp(−z) = 1/ exp(z).
The exponential function has period 2πi, i.e., exp(z + 2πi) = exp(z). If p is any period
of exp, then there is an integer m such that p = 2mπi. Let a be a fixed real number and
Sa = {z ∈ C : a < Im(z) ≤ a + 2π}. Then exp |Sa maps Sa bijectively to C \ {0}.
5.3.2 Trigonometric functions. The trigonometric functions are defined by
exp(iz) + exp(−iz)
cos(z) = ,
2
and
exp(iz) − exp(−iz)
sin(z) = .
2i
They are entire functions, called the cosine and sine function, respectively.
The following properties, familiar from the real case, extend to the complex case:
(1) Derivatives: sin0 (z) = cos(z) and cos0 (z) = − sin(z).
(2) The Pythagorean theorem: (sin z)2 + (cos z)2 = 1 for all z ∈ C.

2August Ferdinand Möbius (1790 – 1868)

24 5. A ZOO OF FUNCTIONS

(3) Addition theorems:

sin(z + w) = sin(z) cos(w) + cos(z) sin(w)
and
cos(z + w) = cos(z) cos(w) − sin(z) sin(w)
for all z, w ∈ C.
(4) Taylor series:
∞
X (−1)n z 2n+1
sin(z) =
n=0
(2n + 1)!
and
∞
X (−1)n z 2n
cos(z) = .
n=0
(2n)!
In particular, sin and cos are extensions of the functions defined in Real Analysis.
(5) cos(z) = sin(z + π/2).
(6) sin(z) = 0 if and only z is an integer multiple of π.
(7) The range of both, the sine and the cosine function, is C.

5.4. The logarithmic function and powers

5.4.1 The real logarithm. Since exp : R → (0, ∞) is a bijective function, it has an inverse
ln : (0, ∞) → R. One shows that ln(1) = 0 and ln0 (x) = 1/x so that, by the fundamental
theorem of calculus Z x
dt
ln(x) =
1 t
for all x > 0.
5.4.2 The logarithm. Suppose Ω is a simply connected non-empty open subset of C \ {0}
and Γ is a contour in C\{0} connecting 1 to z0 ∈ Ω. Then we define the function LΓ : Ω → C
by setting
Z
du
LΓ (z) =
Γ⊕β u

where β is a contour in Ω connecting z0 to z. As the notation suggests LΓ (z) does not

depend on β as long as β remains in Ω. The function LΓ is called a branch of the logarithm
on Ω.
Suppose now that Γ0 is another contour in C \ {0} connecting 1 to a point z00 ∈ Ω. If γ
is a contour in Ω connecting z0 and z00 , then
LΓ (z) − LΓ0 (z) = 2πi IndΓ⊕γ Γ0 (0).

Thus there are (at most) countably many different functions LΓ which are defined this way,
even though there are many more contours Γ connecting 1 to a point in Ω. In fact, for
each m ∈ Z there is a Γ0 such that LΓ (z) − LΓ0 (z) = 2mπi. This shows that defining the
logarithm as an antiderivative comes with certain difficulties (to say the least). At the same
time this behavior gives rise to a lot of interesting mathematics.
LΓ is a holomorphic function on Ω with derivative 1/z. Moreover, exp(LΓ (z)) = z for
all z ∈ Ω but LΓ (exp(z)) may well differ from z by an integer multiple of 2πi.
 5.4. THE LOGARITHMIC FUNCTION AND POWERS 25

5.4.3 The principal branch of the logarithm. Let Ω = C \ (−∞, 0]. Then
Z
du
log(z) = ,
γ u

where γ is contour in Ω connecting 1 and z, is uniquely defined for any z ∈ Ω (i.e., log = LΓ
where Γ : [0, 1] → C : t 7→ 1). It is called the principal branch of the logarithm. If z ∈ Ω has
polar representation z = r exp(it) where t ∈ (−π, π) and r > 0, then
log(z) = ln(r) + it.
The range of log is the strip {z ∈ C : | Im(z)| < π}. In particular, log(exp(z)) = z if and
only if | Im(z)| < π.
The Taylor series of z 7→ log(1 + z) about z = 0 has radius of convergence 1 and is given
by
∞
X (−z)n
log(1 + z) = − .
n=1
n
If Re(a) and Re(b) are positive, then log(ab) = log(a) + log(b). The conclusion may be
wrong when the hypothesis is not satisfied.
5.4.4 Powers. Suppose a is a non-zero complex number and L is a branch of the logarithm
whose domain includes a. If b ∈ Z, then ab = exp(bL(a)). We may therefore define
ab = exp(bL(a))
for any b ∈ C. Be aware, though, that, in general, the value of ab depends on the branch of
the logarithm chosen and is therefore ambiguous when b 6∈ Z. In most cases one chooses, of
course, the principal branch of the logarithm and defines ab = exp(b log(a)) (forsaking the
definition of powers of negative numbers). In this case we get ez = exp(z) for all z ∈ C,
after we define e = exp(1).
5.4.5 Power functions. Suppose Ω is a simply connected non-empty open subset of
C \ {0}. Let p be a complex number and L : Ω → C a branch of the logarithm. The function
Ω → C : z 7→ z p = exp(pL(z)) is called a branch of the power function. Each such branch
is a holomorphic function which never vanishes. Note that 1/z p = z −p .
If p ∈ N0 all branches of the associated power function may be analytically extended to
the entire complex plane and any branch gives then rise to one and the same function. The
same is true if p ∈ Z is negative, except that the function may not be extended to 0 since 0
is then a pole of the function.
If p ∈ Q there are only finitely many different branches of z 7→ z p in any simply
connected open set not containing 0. In fact, if m/n is a representation of p in lowest terms,
there will be n branches of z 7→ z p .
5.4.6 Holomorphic functions without zeros. Suppose f is a holomorphic function
defined on a simply connected open set Ω. If f has no zeros, then f 0 /f has an antiderivative
g̃ and f exp(−g̃) is constant. It follows that there is an holomorphic function g on Ω such
that f (z) = exp(g(z)).
5.4.7 A definite integral. We will later use the following result (which is also an instruc-
tive example):
Z 2π
log |1 − exp(it)|dt = 0.
0
26 5. A ZOO OF FUNCTIONS

To see this let f (z) = z −1 log(1 − z) R and note that 0 is a removable singularity of f .
Hence we get for small positive δ that γ1 γ2 f = 0 when γ1 (t) = exp(it), t ∈ [δ, 2π − δ] and
γ2 (t) = (1−t) exp(iδ)+t exp(−iδ), t ∈ [0, 1]. Since Re(log(1−z)) = ln |1−z|, |1−eit | ≥ |t|/2
for sufficiently small t, |1 − γ2 (t)| ≥ |1 − cos δ| ≥ δ 2 /4, and t ln t − t is an antiderivative of
ln t, we obtain the claim upon taking the limit δ → 0.
 CHAPTER 6

Entire functions

6.1. Infinite products

6.1.1 Infinite products. If n 7→ zn is a sequence of complex numbers, we denote the
Qk Q∞
sequence k 7→ pk = n=1 zn of partial products by n=1 zn . IfQ limk→∞ pk exists, we will
∞
call it the infinite product of the numbers zn and also denote it by n=1 zn (abusing notation
as we do for series).
P∞
6.1.2 Convergence criteria for infinite products. Suppose n=1 (zn − 1) is absolutely
convergent.
P∞ Then there is an n0 ∈ N such Q that |zn − 1| < 1/2 for all n ≥ n0 . Moreover,
∞
n=n0 log(z n ) is absolutely convergent and n=n0 zn converges to a non-zero number. The
latter limit is independent of the order of the factors since
∞
Y ∞
X
zn = exp( log(zn )).
n=n0 n=n0
Q∞
We willQthen say that n=1 zn converges absolutely.
∞
If n=1 zn converges absolutely to 0, then finitely many and only finitely many of the
zn vanish.
6.1.3 Infinite products of analytic functions. Let Ω be a non-empty open set and
n 7→ fnP a sequence of holomorphic functions on Ω none Q∞ of which is identically equal to
∞
zero. If n=1 |fn − 1| converges uniformly on Ω, then n=1 fn converges to a holomorphic
function f on Ω. If z0 is a zero of f , then it is a zero of only finitely many of the functions
fn and the order of z0 as a zero of f is the sum of the orders of z0 as a zero of fn .

6.2. Weierstrass’s factorization theorem

6.2.1 Elementary factors. The functions
E0 (z) = 1 − z
Ep (z) = (1 − z) exp(z + ... + z p /p)
are called elementary factors. Note that Ep (·/a) is entire and has a simple zero at a but no
other zero.
If |z| ≤ 1/2 we have the following estimates:
|Ep (z) − 1| ≤ 2e|z|p+1
and
−2e|z|p+1 ≤ log(|Ep (z)|) ≤ 2e|z|p+1 .
6.2.2 Prescribing zeros of an entire function. Suppose an is a (finite or infinite)
sequence of non-zero complex numbers with no finite limit point. Then there is a sequence
27
28 6. ENTIRE FUNCTIONS

n 7→ pn ∈ N0 , e.g., pn = n, such that

X
∀r > 0 : (r/|an |)pn +1 < ∞.
n

Moreover,
Y
Epn (z/an )
n
converges to an entire function f . The zeros of f are precisely the numbers an and the order
of z0 as a zero of f equals #{n : z0 = an }.
6.2.3 The Weierstrass factorization theorem. Let f be an entire function and suppose
that an , n ∈ N, are the non-zero zeros of f repeated according to their multiplicity. Then
there is an integer m, an entire function g, and a sequence n 7→ pn ∈ N0 such that
Y
f (z) = z m exp(g(z)) Epn (z/an ).
n

6.3. Counting zeros using Jensen’s theorem

6.3.1 Order of an entire function. An entire function is said to be of finite order if the
set
S = {a ∈ [0, ∞) : ∃r0 > 0 : |z| > r0 =⇒ |f (z)| ≤ exp(|z|a )}
is not empty and otherwise of infinite order. In the former case the number inf S is called
the order of f .
6.3.2 Exponent of convergence. Let n 7→ an be a sequence of non-zero complex numbers.
Then
∞
X
τ = inf{s > 0 : |an |−s < ∞},
n=1
if it exists, is called the exponent of convergence of the sequence n 7→ an . Note that, if τ is
finite, the sequence n 7→ an has no finite limit point.
6.3.3 Counting zeros. If f is an entire function we define nf (r) to be the number of
zeros, counting multiplicities, of f in B(0, r).
6.3.4 Jensen’s1 theorem. Suppose f is an entire function and f (0) 6= 0. Denote its
non-zero zeros, repeated according to their multiplicity and ordered by size, by ak , k ∈ N .
Then
n(r) Z 2π
X 1
log(r/|ak |) + log |f (0)| = log |f (r exp(it))|dt.
2π 0
k=1
6.3.5 The growth of an entire function controls the number of its zeros. Let f
be an entire function of finite order ρ for which f (0) = 1 and let ε be a positive number.
Then there is an R > 0 such that
nf (r) ≤ (er)ρ+ε
for all r ≥ R.

1Johan Jensen (1859 – 1925)

 6.4. HADAMARD’S FACTORIZATION THEOREM 29

6.3.6 Relating growth order and exponent of convergence. Suppose f is an entire

function of finite order ρ. Then the exponent of convergence τ of the sequence of non-zero
zeros of f is not more than ρ.
Weierstrass’s factorization theorem takes the form
Y
f (z) = z m exp(g(z)) Ep (z/an )
n

where p ∈ N0 satisfies p ≤ τ < p + 1.

Sketch of proof. We may assume that f (0) = 1 and that the non-zero zeros of f
are labeled by an which are repeated according to their multiplicity and ordered by absolute
value. Thus, if s > ρ and 0 < ε < s − ρ, we obtain from 6.3.5
|an |−s ≤ es nf (|an |)−s/(ρ+ε) ≤ es n−s/(ρ+ε)
when |an | = r > R. 

6.4. Hadamard’s factorization theorem

6.4.1 Canonical products. Let n 7→ an be a sequence of non-zero complex numbers
Q finite exponent of convergence τ . If p ∈ N0 is such that p ≤ τ < p + 1 then k(z) =
with
n Ep (z/an ) is called the canonical product for the sequence n 7→ an .
The function k is an entire function of order τ .
6.4.2 Controlling the decay of canonical products. Suppose an is a (finite or infinite)
sequence of non-zero complex numbers with a finite exponent of convergence τ . Let p ∈ N0
be such that p ≤ τ < p + 1. If τ < s ≤ p + 1 and |z − an | ≥ |an |−p−1 for all n ∈ N , then
Y
|Ep (z/an )| ≥ exp(−c|z|s )
n

for some c > 0.

Note that the hypothesis |z − an | ≥ |an |−p−1 can be achieved for all z on a sequence of
sufficiently large circles of radius rk where k < rk < k + 1 and all n ∈ N.
6.4.3 Hadamard’s2 Factorization Theorem. Let f be an entire function of order ρ
with non-zero zeros (repeated according to their multiplicities) an , n ∈ N . Let τ be the
exponent of convergence of the sequence n 7→ an and define non-negative integers p and q
by requiring p ≤ τ < p + 1 and q ≤ ρ < q + 1. Also let m ∈ N0 be the order of 0 as a zero
of f . Then there exists a polynomial g of degree at most q such that
Y
f (z) = z m exp(g(z)) Ep (z/an ).
n∈N

6.4.4 The very little Picard theorem. An entire function of finite order that misses
two values is constant.
To prove this assume that the entire function f of finite order misses the values α and
β, where α 6= β. Then h = (f − α)/(β − α) is entire, of finite growth order, and misses the
values 0 and 1. Hadamard’s theorem 6.4.3 shows that h = exp(g) for some polynomial g.
Since g has no zero it must be constant by the fundamental theorem of algebra, Theorem
5.1.2.

2Jacques Hadamard (1865 – 1963)

30 6. ENTIRE FUNCTIONS

Picard’s3 little theorem states that this so for every entire function. Picard’s great
theorem states that near an isolated singularity a holomorphic function takes every value,
with at most one exception, infinitely often; this is an extension of the Casorati-Weierstrass
theorem.
6.4.5 An entire function of finite non-integral order assumes every value infin-
itely often. If f − α has no zeros it must be of integral order according to Hadamard’s
theorem 6.4.3. The claim follows now since a canonical product with finitely many zeros
has order 0.

3Émile Picard (1856 – 1941)

 APPENDIX A

Topology in metric spaces

The following material is covered in a course on topology.

A.1. Basics
A.1.1 Topology for a set. Let X be a set. The system τ of subsets of X is called a
topology for X if it has the following properties:
(1) ∅ and X are in τ
(2) Any union of elements of τ is again in τ .
(3) The intersection of two elements of τ is again in τ .
If τ is a topology for X, the pair (X, τ ) is called a topological space. The elements of τ
are called open sets, while their complements are called closed sets.
A set V is called a neighborhood of x ∈ X, if it is open and contains x.
A.1.2 Continuity. Suppose X and Y are topological spaces and f is a function from X
to Y . Then f is called continuous at x ∈ X, if for every neighborhood V of f (x) there is a
neighborhood U of x such that f (U ) ⊂ V . The function f : X → Y is called continuous, if
it is continuous at every point x ∈ X. Equivalently, f is continuous, if the pre-image of any
open subset of Y is an open subset of X.
A.1.3 Metric spaces. Let X be a set. A function d : X × X → [0, ∞) is called a metric
or a distance function in X if it has the following properties:
(1) d(x1 , x2 ) = 0 if and only if x1 = x2 .
(2) d(x1 , x2 ) = d(x2 , x1 ) for all x1 , x2 ∈ X.
(3) d(x1 , x3 ) ≤ d(x1 , x2 ) + d(x2 , x3 ) for all x1 , x2 , x3 ∈ X (triangle inequality).
(X, d) is then called a metric space and the number d(x1 , x2 ) is called the distance
between x1 and x2 .
If X 0 is a subset of X then (X 0 , d0 ) is again a metric space when d0 is the restriction of
d to X 0 × X 0 . Pn
If Rn is equipped with the distance function d defined by d(x, y)2 = k=1 (xk − yk )2 , it
becomes a metric space called the Euclidean space.
A.1.4 Open and closed balls. Let (X, d) be a metric space, x0 a point in X, and r a
non-negative real number. Then the set B(x0 , r) = {x ∈ X : d(x, x0 ) < r} is called the
open ball of radius r centered at x0 while B(x0 , r) = {x ∈ X : d(x, x0 ) ≤ r} is called the
closed ball of radius r centered at x0 . Note that B(x0 , 0) = ∅ so that the empty set is an
open ball. Similarly, the singleton {x0 } = B(x0 , 0) is a closed ball.
A subset of a metric space is called bounded if it is contained in some ball.
A.1.5 Metric spaces are topological spaces. Let (X, d) be a metric space. Then
τ = {U ⊂ X : U is a union of open balls}
31
32 A. TOPOLOGY IN METRIC SPACES

is a topology for X. This follows easily once one proves with the help of the triangle
inequality that the intersection of two open balls is a union of open balls and hence an open
set.
A.1.6 Sequences and their limits. Let (X, d) be a metric space. A function x : N →
X : n 7→ xn is called a sequence in X. We say that the sequence x is convergent if the
following statement holds:
∃L ∈ X : ∀ε > 0 : ∃N > 0 : ∀n > N : d(xn , L) < ε.
If a sequence is not convergent, we say that it diverges or is divergent.
If a sequence x is convergent there is only one element in X to which it converges. This
element is called the limit of the sequence x and is denoted by limn→∞ xn .
A.1.7 Continuity. Let (X, d) and (Y, ρ) be metric spaces and f a function from X to
Y . The definition of continuity given in A.1.2 translates in the context of metric spaces as
follows. The function f is called continuous at a ∈ X if the following statement is true:
∀ε > 0 : ∃δ > 0 : ∀x ∈ X : d(x, a) < δ ⇒ ρ(f (x), f (a)) < ε.

A.2. Compactness
A.2.1 Compact sets. A subset S of a topological space is called compact if every open
cover of SShas a finite subcover . More precisely, whenever V is a family of open
Sn sets such
that S ⊂ V ∈V V then there is a finite subset {V1 , ..., Vn } of V such that S ⊂ k=1 Vk .
A closed subset of a compact set is compact.
A.2.2 Images of compact sets under continuous functions are compact.
A.2.3 Uniform continuity. Let (X, d) and (Y, ρ) be metric spaces and f a function from
X to Y . Then f is called uniformly continuous if the following statement holds:
∀ε > 0 : ∃δ > 0 : ∀x, x0 ∈ X : d(x, x0 ) < δ ⇒ ρ(f (x), f (x0 )) < ε.
If X is compact and f : X → Y is continuous, then it is uniformly continuous.
A.2.4 Sequences in compact metric spaces have convergent subsequences.
A.2.5 The Heine1-Borel2 theorem. The Heine-Borel theorem states that a subset of
Rn is compact if and only if it is closed and bounded. It is a very important result since
it makes it easy to check compactness which in turn is an essential ingredient for several
important results.
A.2.6 The Bolzano3-Weierstrass theorem. Suppose X is a metric space. Every se-
quence in X has a convergent subsequence if and only if X is compact.

A.3. Product spaces

A.3.1 Product spaces. Let (X, d) and (X 0 , d0 ) be metric spaces. The function
ρ : (X × X 0 ) × (X × X 0 ) → [0, ∞) : ((x, x0 ), (y, y 0 )) 7→ max{d(x, y), d0 (x0 , y 0 )}
is a distance function in X × X 0 turning it into a metric space.
Theorem. The cartesian product of compact metric spaces is compact.

1Eduard Heine (1821 – 1881)

2Émile Borel (1871 – 1956)
3Bernard Bolzano (1781 – 1848)
 Glossary

annulus: In a metric space an annulus is a set lying between two concentric circles.
binary operation: A binary operation on a set A is a function from A × A to A. It is
customary to express a binary operation as a ? b (or with other symbols in place of ?).
circumference of a triangle: The sum of the edge lengths of the triangle.
congruence of triangles: Two triangles are congruent if their edge lengths respectively
coincide.
dense: A subset A of a metric space (X, d) is called dense in X if every point of X \ A is
a limit point of A.
diameter: In a metric space (X, d) the diameter of a set A ⊂ X is defined to be sup{d(x, y) :
x, y ∈ A}.
group: A set in which an associative binary operation is defined so that an identity and
inverses to any element exist is called a group.
homeomorphic: Two topological spaces X and Y are called homeomorphic, if there is a
bijection f : X → Y such that both f and f −1 are continuous.
identity: An identity (in a set A with a binary operation ?) is an element e ∈ A such that
e ? a = a ? e = a for all a ∈ A.
interior point: A point z is called an interior point of a set S ⊂ C if there is an open disk
about z which is contained in S.
inverse: The inverse of an element a (in a set A with a binary operation ? and identity e)
is an element b ∈ A such that a ? b = b ? a = e.
isolated point: A point x in a metric space (X, d) is called an isolated point of a set
S ⊂ X if there exists an open disk B about x such that B ∩ S = {x}.
limit point: A point x in a metric space (X, d) is called a limit point of a set S ⊂ X, if
B(x, 1/n) ∩ S \ {x} =
6 ∅ for all n ∈ N.
linear: A map F (defined on a vector space) is called linear if F (αu+βv) = αF (u)+βF (v)
for all α, β ∈ C and all u, v in the domain of F .
local maximum: The value u(z0 ) is called a local maximum of a function u : S → R if
u(z0 ) ≥ u(z) for all z in some neighborhood V = B(z0 , r) ∩ S of z0 .
monic: A polynomial is called monic if its leading coefficient is equal to 1.
monomial: A function of the form m : Ck → C : (z1 , ..., zk ) 7→ az1n1 ...zknk when a is
a complex number and n1 , ..., nk are non-negative integers. The number a is called the
coefficient of the monomial while z1 , ..., zk are called the variables.
33
34 GLOSSARY

periodic: A function f defined on C (or R) is called periodic if there exists a complex (or
real) number a 6= 0 such that f (z + a) = f (z) for all z ∈ C (or z ∈ R). The number a is
called a period of f .
permutation: Let A be a finite set. A permutation of A is a bijective map from A to
itself.
polynomial: A finite sum of monomials.
punctured disk: A set is called a punctured disk if it is a disk with the center removed.
P∞ P∞
rearrangement: The series n=1 wn is a rearrangement of the series n=1 zn if there is
bijective sequence k : N → N such that wn = zkn .
similarity of triangles: Two triangles are similar if the ratios of their respective edge
lengths coincide.
singleton: A singleton is a set containing precisely one element.
symmetric polynomial: A polynomial is called symmetric if no permutation of its vari-
ables changes its values.
unit circle: The set of points in C of modulus 1.
unit disk: The set of points in C of modulus less than 1.
zero: x is called a zero of a function f if f (x) = 0.
 List of special symbols

γ ∗ : the range of the path or contour γ, 7

|z|: the absolute value or modulus of the complex number z, 2

B(x0 , r): the open disk of radius r centered at x0 , 31

B(x0 , r): the closed disk of radius r centered at x0 , 31

C: the set of complex numbers, 1

z: the complex conjugate of the complex number z, 1

Im z: the imaginary part of the complex number z, 1

nf (r): the number of zeros of an entire function f in the disk B(0, r), 28

Re z: the real part of the complex number z, 1

Z(f ): the set of zeros of f , 14

35
 Index

n-th root, 2 connected, 7

convergence
absolute convergence of a function, 5
of a series, 4 of a sequence, 32
of an infinite product, 27 of a series, 3
absolute value, 2 pointwise, 8
analytic, 10 uniform, 8
analytic continuation, 14 convex, 3
annulus, 33 cosine function, 23
associative law of addition, 1
associative law of multiplication, 1 degree, 21
dense, 33
ball, 31 derivative, 5
binary operation, 1, 33 diameter, 33
bounded, 31 differentiable, 5
branch distance, 2, 31
of the logarithm, 24 distance function, 31
branch of the power function, 25 divergence
of a sequence, 32
canonical product, 29 of a series, 4
Cauchy sequence, 3
Cauchy’s integral formula, 13 edge of a triangle, 12
circumference of a triangle, 33 elementary factor, 27
closed, 31 end point
closed contour, 7 of a contour, 7
commutative law of addition, 1 of a path, 7
commutative law of multiplication, 1 entire, 6
compact, 32 essential singularity, 17
complex conjugate, 1 exponent of convergence, 28
complex plane, 2 exponential function, 10
congruence of triangles, 33 extended complex plane, 22
connected, 3
simply, 16 finite subcover, 32
continuity, 31 function
at a point, 31 exponential, 23
in metric spaces, 32 rational, 23
contour, 7
closed, 7 group, 33
37
38 INDEX

Heine-Borel theorem, 32 one-point compactification, 22

holomorphic, 6 open, 31
homeomorphic, 33 open cover, 32
opposite of a path, 7
identity, 33 order
additive, 1 of a pole, 17
multiplicative, 1 of a zero, 14
imaginary number, 1 order of an entire function, 28
imaginary part, 1
infinite product, 27 partial sums, 3
initial point path, 6
of a contour, 7 smooth, 6
of a path, 7 periodic, 34
interior point, 33 permutation, 34
inverse, 33 polar coordinates, 2
additive, 1 pole, 17
multiplicative, 1 polynomial, 21, 34
isolated point, 33 power, 2
isolated singularity, 17 power series, 9
primitive, 8
Laurent series, 15 principal branch
leading coefficient of the logarithm, 25
of a polynomial, 21 principal part, 18
length punctured disk, 34
of a line segment, 2
of a smooth path, 7 radius of convergence, 9
limit real part, 1
of a function, 5 rearrangement, 34
limit of a sequence, 32 reciprocal, 1
limit point, 33 removable singularity, 17
line segment, 2 residue, 18
linear, 33 Riemann integrability, 6
local maximum, 33 Riemann sphere, 22
logarithm, 25
sequence
Möbius transform, 23 in a metric space, 32
meromorphic, 18 of functions, 8
metric, 31 sequence of partial sums, 3
metric space, 31 series, 3
modulus, 2 geometric, 4
monic, 33 similarity of triangles, 34
monomial, 33 simple pole, 18
multiplicity simply connected, 16
of a pole, 17 sine function, 23
of a zero, 14 singleton, 34
singular part, 18
negative of a number, 1 singularity
neighborhood, 31 essential, 17
north pole, 22 isolated, 17
 INDEX 39

removable, 17 for the absolute value, 2

sphere
unit, 22 unit circle, 34
stereographic projection, 22 unit disk, 34
symmetric polynomial, 34 unit sphere, 22

topological space, 31 vertex, 12

topology, 31
triangle, 8, 12 winding number, 11
triangle inequality
for a metric, 31 zero, 34