last edited on

November 21, 2023

Complex Analysis

Oliver Lorscheid

Lecture notes,
Groningen, 2023/24
2

These lecture notes were written in the fall semester of 2023/24, with the help of Anna
Van Gaal, Thomas Zwartveld, Sebastian Holwitt, Lars Wieringa, Margherita Bonan,
Albert Silvans, Prakhar Agarwal, Jamara Admiraal, ...
Contents

Contents 3

0 Preliminaries 5
0.1 The complex numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
0.2 Polar coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
0.3 Sequences and series . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
0.4 Open and closed subsets . . . . . . . . . . . . . . . . . . . . . . . . . 8
0.5 Continuous functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
0.6 Connected subsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
0.7 Compact subsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
0.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1 Holomorphic functions 13
1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2 The Cauchy-Riemann equations . . . . . . . . . . . . . . . . . . . . . 15
1.3 Power series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.4 The logarithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2 Path integrals 27
2.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2 Path integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3
 chapter last edited on
November 15, 2023

Chapter 0

Preliminaries

In this chapter, we revise knowledge from previous courses that we need for this course
in Complex Analysis.

0.1 The complex numbers

Let R be the real numbers. The complex numbers are the set C = R2 together with the
vector addition    0
x + x0
 
x x
+ 0 =
y y y + y0
and the multiplication    0  0
xx − yy0

x x
· 0 =
y y xy0 + x0 y
which turn C into a field. In particular, the additive neutral element of C is 00 and

the multiplicative unit is 10 . The additive inverse of xy is − xy = −x

−y and the
x

multiplicative inverse of a nonzero element y is
 −1  
x 1 x
= 2 2
· .
y x +y −y

We denote by C× = C \ {0} the unit group of
C.

The real part of a complex number xy is Re xy = x and its imaginary part is

Im xy = y. We call { 0x | x ∈ R} the real axis of C and { 0y | y ∈ R} its imaginary axis.

Note that the real axis forms a subfield of C that
is isomorphic to R. In the following,
we identify R with the real axis and write x for 0x .
The imaginary unit is the element i = 01 , which satisfies i2 =
−1



0 = −1, i.e. i is a
square root of −1. This allows us to rewrite a complex number xy as
     
x x y
= +i· = x + iy.
y 0 0

5
6 Preliminaries

In this notation, the previous formulas for the sum, product and inverse of complex
numbers z = x0 + iy and z0 = x0 + iy0 can be rewritten as

z + z0 = (x + x0 ) + i(y + y0 ),
z · z0 = (xx0 − yy0 ) + i(xy0 + x0 y),
x − iy
z−1 = 2 .
x + y2

The complex conjugate of z = x + iy is z̄ = x − iy. The absolute value of z = x + iy is

√ p
|z| = z · z̄ = x2 + y2 .

Thus we have z−1 = z̄/|z|2 .

We typically illustrate C = R2 as a plane with 1 pointing to the right and i pointing
upwards. Complex conjugation corresponds to the reflection in the x-axis.

Insert an illustration

0.2 Polar coordinates

Let R>0 = {x ∈ R | x > 0} be the set of positive real numbers and R/2πZ the quotient
of the additive group of R by the subgroup 2πZ. The map

Φ : R>0 × (R/2πZ) −→ C×
(r, ϕ) 7−→ r(cos ϕ + i sin ϕ)

is a bijection whose inverse sends z ∈ C× to Ψ(z) = (|z|, arg z) where arg z is the argument
of z;1 this is, up to multiples of 360 degrees, 360
2π · arg z is the angle between the positive
real axis and the halfline spanned by z in counterclockwise direction. The tuple (|z|, arg z)
is called the polar coordinates of z.

Insert an illustration

1 Note that we consider the argument as a function to R/2πZ, i.e. the argument is only well-defined
up to integers multiples of 2π. This allows us to add angles without further explanations, in contrast to the
so-called principal argument, which takes values in (−π, π].
0.3. Sequences and series 7

We can express several formulas in polar coordinates:

Ψ(z · z0 ) = (|z| · |z0 |, arg z + arg z0 ),

Ψ(z−1 ) = (|z|−1 , − arg z),
Ψ(z̄) = (|z|, − arg z).

In particular, the multiplication with a complex number z of absolute value |z| = 1 is a

counter-clockwise rotation of C = R2 by the angle 360
2π · arg ϕ around 0.

Insert an illustration

0.3 Sequences and series

A sequence in C is an indexed family {zn }n∈N of complex numbers zn ∈ C. A limit
of a sequence {zn }n∈N is a complex number z ∈ C such that for all  > 0, there is an
N ∈ N such that |zk − z| <  for all k > N. A sequence {zn }n∈N converges if a limit exists.
Otherwise it is said to diverge. We write

z = lim zn or zn −→ z
n→∞ n→∞

to indicate that z is a limit of {zn }n∈N . A Cauchy sequence in C is a sequence {zn }n∈N in
C such that for all  > 0, there is an N ∈ N such that for all k, l > N, we have |zk − zl | < .

Insert an illustration

Fact 0.3.1. Let {zn }n∈N be a sequence in C.

(1) Limits are unique, i.e. {zn }n∈N has at most one limit.
(2) The sequence {zn }n∈N has a limit if and only if it is a Cauchy sequence.

A series in C is an expression of the form ∑∞

i=0 zn for complex numbers zn ∈ C. The
n
partial sums of ∑i=0 zi are the sums Sn = ∑i=0 zi . The series ∑∞
∞
i=0 zi converges (to z)
∞
if the sequence {Sn }m∈N converges (to z). The series ∑i=0 zi converges absolutely if
∑∞i=0 |zi | converges as a series in R.
8 Preliminaries

Fact 0.3.2. A series in C converges if it converges absolutely.

Definition 0.3.3. Let {ai }i∈N be a sequence of real numbers ai . The superior limit of
{ai } is

lim sup ai = lim sup{ai | i > n}
n→∞

which we interpret as ∞ if sup{ai | i > n} = ∞ for all n and as −∞ if sup{ai | i > n} has
no lower bound.

Note that the superior limit always exists since sup{ai | i > n} is a monotonically
decreasing sequence in n. If {ai } converges, then lim sup{ai } = lim{ai }.

Fact 0.3.4. Let {ai } and {bi } be sequences with superior limits in R. Then

lim sup ai ± lim sup bi = lim sup(ai ± bi )

and
lim sup ai · lim sup bi = lim sup(ai · bi ).
If ai 6= 0 for all i ∈ N and lim sup ai 6= 0, then

lim sup(a−1 −1
i ) = (lim sup ai ) .

Insert an illustration

0.4 Open and closed subsets

We are mostly concerned with the topology of C = R2 , but at times, we also apply
topological properties to R = R1 . Therefore we revise the following topological concepts
and facts in the generality of Rn for n > 0.
The Euclidean norm of a vector x = (x1 , . . . , xn ) in Rn is
q
||x|| = x12 + . . . + xn2 .

For x ∈ R = R1 , the Euclidean norm is equal to the usual Euclidean absolute value
||x|| = |x|. For z ∈ C = R2 , the Euclidean norm is equal to the complex absolute value
||z|| = |z|.
Let r > 0 and a ∈ Rn . The open ball of radius r and with center a is the set

Br (a) = {x ∈ Rn | ||x − a|| < r}.

0.5. Continuous functions 9

In R = R1 , the open ball Br (a) is the open interval (a − r, a + r). In C = R2 , the open
ball Br (a) is the open disc Dr (a) = {z ∈ C | |z − a| < r}.
A subset U of Rn is open if for every a ∈ U, there is an  > 0 such that B (a) ⊂ U.
In other words, U is open if and only if it is the union of open discs. A subset A of Rn is
closed if Rn \ A is open.

Fact 0.4.1. A subset A of Rn is closed if and only if every Cauchy sequence {zn }n∈N in
A has a limit z = lim zn in A.

The interior of a subset W of Rn is

W◦ =
[
U = {x ∈ W | B (w) ⊂ W for some  > 0},
U⊂W
open

which is the largest open subset of W . The closure of W is

\
W = A = {w ∈ Rn | w = lim zn for {zn }n∈N ⊂ W },
W ⊂A
closed

which is the smallest closed subset of Rm that contains W . The boundary of W is the
difference ∂W = W \W ◦ between the closure and the interior of W . In particular, we
have
W ◦ = W \ ∂W and W = W ∪ ∂W.

Insert an illustration

A subset V of W is open in W if it is of the form V = U ∩W for an open subset U of

Rn . This topology for W is called the induced topology or the subspace topology.

0.5 Continuous functions

Let W be a subset of Rn . A function f : W → Rm has a limit a ∈ Rm in w ∈ W if for
every  > 0, there is a δ > 0 such that f (Bδ (w) ∩W ) ⊂ B (a). We write

a = lim f (z) or f (z) −→ a

z→w z→w

in this case. Note that the limit limz→w f (z) of f in w is unique if it exists.

Fact 0.5.1. Let f , g : W → Rm be two functions for which the limit in w ∈ W exists. Then

lim ( f ± g)(z) = lim f (z) ± lim g(z)

z→w z→w z→w
10 Preliminaries

and
lim ( f · g)(z) = lim f (z) · lim g(z).
z→w z→w z→w

If limz→w f (z) 6= 0, then

 −1
lim 1/ f (z) = lim f (z) .
z→w z→w

The function f : W → Rm is continuous if f (w) = limz→w f (z) for every w ∈ W .

Fact 0.5.2. Let W ⊂ Rn be a subset and f : W → Rm a function. Then the following are
equivalent:

(1) The function f is continuous.

(2) For every open subset U in Rm , the inverse image f −1 (U) is open in W .
(3) For every sequence {zn }n∈N in W with limit z ∈ W , the image f (z) is a limit of
{ f (zn )}n∈N .

0.6 Connected subsets

A subset W of Rn is connected if it is not contained in the disjoint union U1 tU2 of two
open subsets U1 and U2 of Rn that do neither contain W . A path in W is a continuous
map γ : I → W from the unit interval I = [0, 1] to W . A subset W of Rn is path connected
if for all x, y ∈ W , there is a path γ : I → W such that γ(0) = x and γ(1) = y.

Fact 0.6.1. Every path-connected subset W of Rn is connected. If W is open, then W is

connected if and only if it is path-connected.

0.7 Compact subsets

A subset W of Rn is bounded if there is an r > 0 such that W ⊂ Br (0). A subset A of Rn
is compact if it is closed and bounded. For example, a subset of R = R1 is compact if and
only if it is a finite union of closed intervals. The closed disc Dr (a) = {z ∈ C | |z−a| 6 r}
of radius r and with center a ∈ C is a compact subset of C.

Theorem 0.7.1 (Heine-Borel and Bolzano-Weierstrass). Let A be a subset of Rn . The

following are equivalent.

(1) A is compact.
(2) If A ⊂ i∈I Ui for open subsets Ui of Rn , then there exists a finite subset J of I such
S

that A ⊂ i∈J Ui .
S

(3) For every sequence {zn }n∈N in A, there is an order-preserving injection σ : N → N

such that {zσ(n) }n∈N has a limit in A.
0.8. Exercises 11

Lemma 0.7.2. Let A ⊂ Rn be a compact subset and f : A → Rm a continuous function.

Then f (A) is a compact subset of Rm . In particular, if A ⊂ C is compact, then every
continuous function f : A → R assumes its maximum, i.e. there is an a ∈ A such that
f (z) 6 f (a) for all z ∈ A.

Lemma 0.7.3 (Lebesgue’s Lemma). Let A ⊂ Rn be a compact subset that is contained

in the union i∈I Ui of open subsets Ui of Rn . Then there is a δ > 0 such that for every
S

a ∈ A, there is an i ∈ I such that Bδ (a) ⊂ Ui .

In particular, if [0, 1]n = i∈I Ui for open subsets Ui of [0, 1]n (in the subset topology
S

for [0, 1]n ), then there is an N > 0 such that for all k1 , . . . , kn ∈ {1, . . . , N}, there is an
i ∈ I such that    
k1 − 1 k1 kn − 1 kn
, ×···× , ⊂ Ui .
N N N N

0.8 Exercises
Exercise 0.8.1. Verify all formulas of section 0.1 and section 0.2.

Exercise 0.8.2. Show that limn→∞ (n + 1)1/n = 1, which can be done as follows:
p
n
(1) For an integer n > 1, show that n + 1 6 1 + 2/(n + 1) by comparing n + 1
with the binomial expension of the right hand side.
p
(2) Use this together with 1 6 (n + 1)1/n and lim 1 + 2/(n + 1) = 1 to conclude
that lim(n + 1)1/n = 1.

Exercise 0.8.3. Let z ∈ C have |z| < 1. Show that the partial sum Sn = ∑ni=0 zi equals
(zn+1 − 1)/(z − 1) and conclude that the series ∑∞ i −1
i=0 z converges absolutely to (1 − z) .

Exercise 0.8.4 (Warsow curve). Consider the following subset of R2 :

X = {(x, sin( 1x ) | x > 0} ∪ {(0, y) | −1 6 y 6 1}.

Show that X is closed and connected. Show that X is not path-connected.

Exercise 0.8.5. Let f : X → Y be a continuous map, X compact and Z ⊂ X closed. Prove

that both Z and f (X) are compact.

Exercise 0.8.6. Show that X is connected if and only if there is no continuous surjection
X → {0.1} with respect to the discrete topology for {0, 1} (i.e. every subset is open).

Exercise 0.8.7. Prove the facts in this chapter.

 chapter last edited on
November 15, 2023

Chapter 1

Holomorphic functions

1.1 Definitions
Definition 1.1.1. Let U and V be open subsets of C and a ∈ U. A function f : U → V
is complex differentiable in a if
f (z) − f (a)
f 0 (a) = lim
z→a z−a
exists. A function f : U → V is holomorphic if it is complex differentiable in all a ∈ U.
An entire function is a holomorphic function f : C → C.
Mostly the codomain V plays a subordinate role and it does not matter if we replace
V by C, or vice versa. Therefore we consider f : U → C often as a function in C.
In some situations, it is however important to work with proper subsets V of C as a
codomain, such as in Theorem 1.2.5.
Remark 1.1.2. A function f : U → C is complex differentiable in a ∈ U if and only if
there is a w ∈ C such that
f (z) − f (a) − w · (z − a)
−→ 0.
|z − a| z→a

In this case f 0 (a) = w.

Note further that if f 0 (a) exists, then

lim f (z) − f (a) = f 0 (a) · lim (z − a) = 0,

z→a z→a

which shows that f is continuous in a.

Proposition 1.1.3. Let U ⊂ C be an open subset, a ∈ U, c ∈ C and f , g : U → C two
functions that are complex differentiable in a. Then
(1) ( f ± g)0 (a) = f 0 (a) ± g0 (a);

13
14 Holomorphic functions

(2) (c · f )0 (a) = c · f 0 (a);

(3) ( f · g)0 (a) = f 0 (a) · g(a) + f (a) · g0 (a); (Leibniz rule)
(4) (1/ f )0 (a) = − f 0 (a)/( f (a))2 if f (a) 6= 0.
(5) If V ⊂ C is an open subset with f (U) ⊆ V and h : V → C is complex differentiable
in f (a), then (h ◦ f )0 (a) = h0 ( f (a)) · f 0 (a). (Chain rule)

Proof. We only prove the most difficult case (5) and leave (1)–(4) as an exercise. The
chain rule follows from the direct computation

(h ◦ f )(z) − (h ◦ f )(a) (h ◦ f )(z) − (h ◦ f )(a) f (z) − f (a)

lim = lim · lim
z→a z−a z→a f (z) − f (a) z→a z−a
= h0 ( f (a)) · f 0 (a),

where we use that limits interchange with products (Fact 0.5.1) for the first equality.
For the second equality, we use that f is continuous in a and thus f (z) → f (a) when
z → a.

Example 1.1.4. (1) Let n > 0. Then the function

f : C −→ C
z 7−→ zn

is entire with derivative f 0 (z) = nzn−1 . (We leave the verification of this claim as
an exercise.)
(2) Let n < 0 and C× = C \ {0}. Then the function

f : C× −→ C
z 7−→ zn

is entire with derivative f 0 (z) = nzn−1 . (We leave the verification of this claim as
an exercise.)
(3) A polynomial function is a function f : C → C of the form f (z) = ∑nn=0 cn zn
for some c0 , . . . , cn ∈ C. A polynomial function is entire with derivative f 0 (z) =
∑n−1
n=0 (n + 1)cn+1 z .
n

(4) Let f , g : U → C be holomorphic and assume that g(z) 6= 0 for z ∈ U. Then

f
g : U → C is holomorphic with derivative

 0
f f 0 g − f g0
(z) = (z).
g g2
f
A rational function is a function of the form g : U → C for two polynomial
functions f and g.
1.2. The Cauchy-Riemann equations 15

1.2 The Cauchy-Riemann equations

In this section, we investigate how complex differentiability relates to real differentiabil-
ity. For this, we recall the concept of differentiability in two real variables.
Recall that x + iy = xy under the identification C = R2 . Let U ⊂ C be an open

subset and f : U → C a function. Then there exist unique functions u, v : U → R such

that
f (x + iy) = u(x, y) + iv(x, y)
for all x, y ∈ R; namely

u(x, y) = Re( f (x + iy)) and v(x, y) = Im( f (x + iy)).

We write f = u + iv for short to express these relations.

Definition 1.2.1. Let a = bc ∈ U. The function f : U → R2 is (real) differentiable in a

if there is a real 2 × 2-matrix J f (a) such that

x
b



f (x, y) − f (b, c) − J f (a) · −
 
y c 0
lim = .
|| xy − bc ||



(xy)→(bc) 0

The function f : U → R2 is differentiable if it is differentiable in all a ∈ U.

The matrix J f is unique if it exists, and it is called the Jacobian matrix of f . Its
coefficients are denoted as follows:
∂u ∂u
! !
∂ x (a) ∂ y (a) ux (a) uy (a)
J f (a) = ∂ v ∂v
= .
∂x (a) ∂y (a) v x (a) vy (a)

The key lemma to relate complex and real differentiation is the following.
x

Lemma 1.2.2. Let A be a real 2 × 2-matrix and w = r + is ∈ C. Then wz = A · y for
all z = x + iy ∈ C if and only if A = rs −s


r .

Proof. The computation

rx−sy
r −s

x

wz = (r + is) · (x + iy) = (rx − sy) + i(ry + sx) = ry+sx = s r y

shows that z 7→ wz is an R-linear map R2 → R2 that corresponds to the (uniquely

determined) matrix A = rs −s


r .

Theorem 1.2.3. Let U ⊂ C be an open subset and f = u + iv : U → C a function. Then

f is holomorphic if and only if f is differentiable and if the Cauchy-Riemann equations
hold:
ux = vy and uy = −vx .
If f is holomorphic, then f 0 (a) = ux (a) + ivx (a) for all a ∈ U.
16 Holomorphic functions

b

Proof. Let a = c ∈ U. The function f is complex differentiable in a if and only if

f (z) − f (a) − f 0 (a) · (z − a)

−→ 0
|z − a| z→a

with f 0 (a) = r + is for some r, s ∈ R. By Lemma 1.2.2, this is equivalent to

f (x,y)− f (b,c)−J f (a)·((xy)−(bc))

−→ 0
||(xy)−(bc)||
(xy)→(bc)

with J f (a) = rs −s


r for some r, s ∈ R.
By the definition of the partial differentials ux , uy , vx and vy , we have r = ux (a) =
vy (a) and s = −ux (a) = vy (a). Thus f is holomorphic if and only if f is (real) differen-
tiable and if the Cauchy-Riemann equations ux = vy and uy = −vx hold.

Example 1.2.4. (1) The function

f : C −→ C
z 7−→ z2

satisfies

f (x + iy) = (x + iy)2 = (x2 − y2 ) + i(2xy) = u(x, y) + iv(x, y)

for u(x, y) = x2 − y2 and v(x, y) = 2xy, which satisfy Cauchy-Riemann equations:

ux = 2x = vy and uy = −2y = −vy .

This re-establishes that z 7→ z2 is an entire function.

(2) The complex conjugation
f : C −→ C
z 7−→ z̄
satisfies
f (x + iy) = x − iy = u(x, y) + iv(x, y)
for u(x, y) = x and v(x, y) = −y. Since ux = 1 6= −1 = vy , the complex conjugation
is not holomorphic.

Theorem 1.2.5 (Inverse function theorem). Let U and V be open subsets of C and
f : U → V a holomorphic bijection with inverse bijection g : V → U. If g is continuous
and if f 0 (a) 6= 0 for all a ∈ U, then g is holomorphic with derivative

1
g0 (b) =
f 0 (g(b))

for all b ∈ V .
1.3. Power series 17

Proof. Consider w, b ∈ V with images z = g(w) and a = g(b). Since both f and g are
continuous, w converges to b if and only if z converges to a (where we consider w and z
as variables). Since f 0 (a) 6= 0, we can exchange the limit with (−)−1 (Fact 0.5.1), which
yields

f (z) − f (a) −1
 
0 g(w) − g(b) z−a
g (b) = lim = lim = lim
w→b w−b z→a f (z) − f (a) z→a z−a
1 1
= 0 = 0 .
f (a) f (g(b))
This shows, in particular, that g is complex differentiable in b.
Remark 1.2.6. The assumption that g is continuous is not necessary for Theorem 1.2.5
to hold. This can be seen by applying the implicit function theorem of (real) multivariate
analysis, which shows that g is (real) differentiable and therefore continuous. We will
show this later on with an argument from complex analysis.

1.3 Power series

Definition 1.3.1. A (complex) power series is an expression of the form
∞
∑ an zn
n=0
with a0 , a1 , . . . ∈ C. When the context is clear, we write ∑ an zn for short.
In the following we investigate the question for which z ∈ C such a power series con-
verges and write ∑ an zn = f (z) if ∑ an zn converges to f (z) (where f (z) is an expression
in z). Subsequently, we show that power series define holomorphic functions whenever
they converge on an open disc of the form Dr (a) = {z ∈ C | |z − a| < r}.
Lemma 1.3.2. The geometric series ∑ zn converges absolutely
∞
1
∑ zn =
1−z
n=0
for all z ∈ D1 (0).
Proof. Consider the partial sums
n
1 − |z|n+1
Sn = ∑ |zi| = 1 − |z|
i=1
of absolute values where the identity on the left hand side results from multiplying both
sides by 1 − |z|. For z ∈ Dz (0), we have limn→∞ |z|n+1 = 0. Thus
∞
1 − |z|n+1 1
∑ |z|n = lim Sn = lim
n→∞ n→∞ 1 − |z|
=
1 − |z|
n=0
converges. In other words, ∑ zn converges absolutely for z ∈ D1 (0). The same computa-
1
tion with |z| replaced by z shows that ∑ zn converges to 1−z for z ∈ D1 (0).
18 Holomorphic functions

As a consequence, we see that the power series ∑∞ n

n=0 an z defines a holomorphic
1
function f : D1 (0) → C. In fact, 1−z extends to a holomorphic function f : C \ {1} → C,
which is a first example for analytic continuation, which we discuss in a later chapter.
Definition 1.3.3. The radius of convergence of a power series ∑ an zn is
1
R =
lim sup |an |1/n
1 1
where we define 0 = ∞ and ∞ = 0.
Theorem 1.3.4 (Cauchy-Hadamard). Let ∑ an zn be a power series with radius of con-
vergence R. Then ∑ an zn converges absolutely if |z| < R and it diverges if |z| > R.
Proof. Let L = lim sup |an |1/n = R1 . Assume that |z| < R or, equivalently, that L · |z| <
1. Then there is an  > 0 such that q = (L + ) · |z| < 1. By the definition of L =
lim sup |an |1/n , there is an N > 0 such that |an |1/n 6 L +  for all n > N. Thus
∞ ∞
n ∞
∑ |an | · |z|n 6 C + ∑ (L + ) · |z| = C + ∑ qn
n=0 n=0 n=0

for some sufficiently large C ∈ R. Since q < 1, Lemma 1.3.2 implies that ∑ qn converges,
which shows that ∑ an zn converges absolutely for |z| < R.
Assume that |z| > R or, equivalently, that L · |z| > 1. It suffices to show that the
sequence of partial sums Sn = ∑nn=0 an zn is not a Cauchy sequence, in order to show that
∑ an zn converges. First note that there is an  > 0 such that q = (L − ) · |z| > 1. By the
definition of L = lim sup |an |1/n , there are infinitely many n ∈ N such that |an |1/n > L − .
For such n, we have

n
|Sn − Sn−1 | = |an zn | > (L − ) · |z| = qn ,
which does not tend to 0 for increasing n. This shows that {Sn } is not a Cauchy sequence
and that ∑ an zn diverges.
Remark 1.3.5. For |z| = R, it depends on the series and the specific z whether ∑ an zn
converges or diverges.
Example 1.3.6. (1) The exponential series
∞
1 n
ez = ∑ z
n=0 n!
has radius of convergence
1 1
R = = = ∞
lim sup | n!1 |1/n 1
∞

since q

1/n
|n!|1/n > ( 2n )n/2 = n
−→
2 n→∞ ∞.
This defines the exponential function exp : C → C with exp(z) = ez .
1.3. Power series 19

(2) Both
in n in−1 n
cos z = ∑ z and sin z = ∑ z
n∈N even n! n∈N odd n!

have radius of convergence R = ∞ and thus define functions cos : C → C and

sin : C → C.

Theorem 1.3.7. Let ∑ an zn be a power series with radius of convergence R. Then the
function
f : DR (0) −→ C
z 7−→ ∑ an zn
is holomorphic with derivative
∞
f 0 (z) = ∑ (n + 1) · an+1zn,
n=0

which is a power series with the same radius of convergence R.

Proof. For the sake of this proof, let g(z) = ∑∞ n

n=0 (n + 1) · an+1 z . Using m = n − 1 and
Exercise 0.8.2, we have
1
m/(m−1)
lim sup |m · am |1/(m−1) = lim sup m1/(m−1) · lim sup |am | m
| {z }
=1
1
= lim sup |am |1/m = ,
R
which shows that the radius of convergence of the power series ∑(n + 1) · an+1 zn is equal
to that of ∑ an zn , which defines g as a function from DR (0) to C.
In order to prove that ∑ an zn is holomorphic on DR (0) with derivative f 0 (z) = g(z),
we divide the power series into its partial sum and its “tail”

N ∞
SN (z) = ∑ an zn and EN (z) = ∑ an zn ,
n=0 n=N+1

respectively. Consider a fixed element z ∈ DR (0) and |z| < r < R. For an element
h ∈ Dr−|z| (0) \ {0}, we compute

f (z + h) − f (z)
− g(z)
h     
SN (z + h) − SN (z) 0 0 EN (z + h) − EN (z)
= − SN (z) + SN (z) − g(z) +
h h

We have
SN (z + h) − SN (z)
− SN0 (z) −→ 0,
h h→0
20 Holomorphic functions

by definition of the derivative SN0 (z), and

∞
SN0 (z) − g(z) = − ∑ (n + 1) · an+1 zn −→ 0.
N→∞
n=N+1

Since an − bn = (a − b) · ∑n−1 i n−1−i for a = z + h and b = z, we have for small h (for

i=0 a b
which |z + h| < r):

∞ n−1 ∞
EN (z + h) − EN (z) h
h
= ∑ |an | ·
h
· ∑ (z + h)izn−1−i 6 ∑ |an | · n · rn−1 ,
n=N+1 i=0 n=N+1

which tends to 0 as N → ∞. Thus

∞
f (z + h) − f (z)
lim − g(z) 6 SN0 (z) − g(z) + ∑ |an | · n · rn−1 −→ 0,
h→0 h n=N+1
N→∞

which shows that

f (z + h) − f (z)
g(z) = lim
h→0 h
is the differential f 0 (z) of f in z. This shows, in particular, that f is holomorphic as a
function in DR (z).

As an immediate consequence, we have:

Corollary 1.3.8. The functions exp, cos and sin are entire.

Proposition 1.3.9. Consider two power series f (z) = ∑ an zn and g(z) = ∑ bn zn that are
absolutely convergent on Dr (0) with r > 0 (possibly r = ∞) and let c ∈ C. Then
∞
( f + g)(z) = ∑ (an + bn)zn;
n=0
∞
(c · f )(z) = ∑ (c · an)zn;
n=0
∞  
( f · g)(z) = ∑ ∑ ak bl zn (Cauchy product)
n=0 k,l>0 s.t.
k+l=n

for all z ∈ Dr (0). In particular, all power series converge absolutely for z ∈ Dr (0).

Proof. All three formulas follow by proving them for the partial sums of the correspond-
ing power series and then letting the number of terms in the partial sums go to infinity.
Since the limit of the partial sums converges, the power series in these equations have
radius of convergence R > r and thus converge absolutely for z ∈ Dr (0). We omit the
details.
1.4. The logarithm 21

Corollary 1.3.10. For all complex numbers z = x + iy and w, we have

(1) exp0 (z) = exp(z) (4) eiz = cos z + i sin z (7) cos z = 21 (eiz + e−iz )
(2) sin0 (z) = cos(z) (5) ez = ex (cos y + i sin y) (8) sin z = 1 iz
2i (e − e )
−iz

(3) cos0 (z) = − sin(z) (6) ez+w = ez · ew (9) i arg z

z = |z| · e

The proof of these identifies follow from a direct manipulation of the corresponding
power series, which is left as an exercise. The identity eiz = cos z + i sin z is known as
Euler’s formula.

1.4 The logarithm

The real exponential function exp : R → R>0 is a bijection with inverse log : R>0 → R.
In this section, we investigate the corresponding behaviour of the complex exponential
function exp : C → C.
As a first observation, we compute for z = x + iy ∈ C that

ex · cos y + i sin y) ∈ C×
ez = |{z}
| {z }
∈R>0
∈S1

where S1 = {z ∈ C | |z| = 1} is the unit sphere. Therefore the image of exp : C → C is

C× = R>0 × S1 .

Insert an illustration

If w ∈ C× has polar coordinates (|w|, arg w) with |w| ∈ R>0 and arg w ∈ R/2πZ,
then
exp−1 (w) = x + iy ∈ C x = log |w|, y arg w + 2πZ .


In particular, exp−1 (1) = 2πiZ. This shows that exp : C → C× is not injective, in
contrast to exp : R → R>0 .

Insert an illustration

Idea: restrict exp to a bijection between subsets of the forms

Hα = {z ∈ C | Im(z) ∈ (α, α + 2π)} and Uϑ = C× \ {λ · ϑ | λ > 0}

22 Holomorphic functions

for α ∈ R and ϑ = eiα ∈ S1 .

Insert an illustration

Definition 1.4.1. The α-branch of the logarithm is the inverse logα : Uϑ → Hα to the
bijection exp : Hα → Uϑ for α ∈ R>0 and ϑ = eiα . The principal branch of log is
log = log−π : U−1 → H−π .

Insert an illustration

Proposition 1.4.2. Let α ∈ R and ϑ = eiα . The function logα : Uϑ → Hα is holomorphic

with derivative
1
log0α (z) = .
z
Proof. As a first step, we show that logα is continuous, i.e., for every a ∈ Uϑ with image
b = logα (a) and for every  > 0 there is a δ > 0 such that


logα Dδ (a) ∩Uθ ⊂ D (b).

After making  and δ suitably small, we can assume that D (a) ⊂ Uθ and that D (b) ⊂
Hα . Since exp ◦ logα is the identity on Dδ (a), the above inclusion follows from the
inclusion Dδ (a) ⊂ exp(D (b)) of the respective images under exp. Let Q be a small
square around b contained in D (b). Then the horizontal boundary lines of the square
are mapped to straight lines that lie on rays originating from the origin of C, and the
vertical boundary lines of the square lie on circles around the origin. This makes clear
that exp(Q) is a deformed square that contains a in its interior (as illustrated below) and
thus also an open disc Dδ (a) for suitably small δ. This shows that logα is continuous.

Insert an illustration
1.4. The logarithm 23

Therefore we can apply the inverse function theorem (Theorem 1.2.5) to exp : Hα →
Uϑ , which shows that its inverse logα is holomorphic with derivative

1 1 1
log0α (z) =
=
= ,
exp0 logα (z) exp logα (z) z

where we use Exercise 3 from List 2 for exp0 (w) = exp(w).

Remark 1.4.3. (1) We have

logα (z) − logβ (z) ∈ 2πiZ

for all α, β ∈ R and z ∈ Ueiα ∩ Ueiβ .

(2) For α ∈ R and ϑ = eiα , we have
 
−i i
lim logα (ϑ · e ) − logα (ϑ · e ) = 2πi.
→0
with >0

Insert an illustration

(3) We have logα (z · w) ≡ logα (z) + logα (w) (mod 2πi), but in general these terms
are not equal. For example:
 3 3
  1 
log e 4 πi · e 4 πi = log e− 2 πi = − 12 πi
 3   3 
πi
3 3 3
+ log e 4 πi .



6= 2 πi = 4 πi + 4 πi = log e 4

Insert an illustration
24 Holomorphic functions

Application 1: Complex powers

Heuristically, we could attempt to define the power zw for z, w ∈ C with w 6= 0 as
?
zw = (elogα (z) )w = ew logα (z)

for suitable α ∈ R. However, this expression depends on the choice of α.

A better definition of zw is as the set

zw = ew·logβ (z) β ∈ R = ew·logα (z)+k·2πi k ∈ Z ,

 

which does not depend on the choice of α, but rather combines all possible such choices.
Lemma 1.4.4. Let z, w ∈ C with w 6= 0. Then the set zw is a singleton if and only if
w ∈ Z and it is finite if and only if w ∈ Q.
Proof. The set  w·log (z)+k·2πi
zw = e α k∈Z ,
is finite if and only if kw ∈ Z for some k > 0, which is the case if and only if k ∈ Q. This
proves the second claim. The first claim follows since zw is a singleton if and only if
kw ∈ Z for all k > 0, which means that w ∈ Z.
Example 1.4.5. For positive n ∈ N, we have
√  1 k
z = z1/n = e| n logα (z) n ·2πi
{z } · |e {z } k = 1, . . . , n .
n

=ρ0 =ζnk

2πi
where ρ0 is an n-th root of z (which depends on the choice of α) and ζn = e n is a
primitive n-th root of unity.

Insert an illustration

Proposition 1.4.6. Let w ∈ C× , α ∈ R and ϑ = eiα . Then the map

f : Uϑ −→ C
z 7−→ ew·logα (z)

is holomorphic with derivative

f 0 (z) = w · e(w−1)·logα (z) .

If we interpret f (z) as (a branch of) zw , then Proposition 1.4.6 shows that f 0 (z) is (a
branch of) w · zw−1 .
1.4. The logarithm 25

Proof. As a composition of holomorphic functions, f is holomorphic. We use the rules

from Proposition 1.1.3 to compute:

1
f 0 (z) = exp0 (w · logα (z)) · w · log0α (z) = exp(w · logα (z))w ·
z
= w · e−1·logα (z) · ew·logα (z) = w · e(w−1)·logα (z) .

Application 2: Inverse trigonometric functions

Let z = cos(w). If we try to express w in dependency of z, we find that

1 iw
z = cos(w) = (e + e−iw )
2
⇔ 2z = eiw + e−iw
⇔ (eiw )2 − 2zeiw + 1 = 0
p

⇔ w = −i · logα z ± i 1 − z2

for a suitable α ∈ R. The principal branch of arccos is

arccos : C \ {x ∈ R | |x| > 1} −→ C √

z 7−→ −i · log z + i 1 − z2

where α = −π and 1 − z2 ∈ U−1 .

Insert an illustration

Similarly, we define the principal branch of arcsin as

arccos : C \ {x ∈ R | |x| > 1} −→ C√

z 7−→ −i · log iz + 1 − z2 .
 chapter last edited on
November 15, 2023

Chapter 2

Path integrals

Motivation
Rb
Question. Does the expression f (z) dz make sense for f : U → C and a, b ∈ U ⊂ C?
a

Thought 1. If f = F 0 for a holomorphic function F : U → C, then we expect that

Zb
f (z) dz = F(b) − F(a).
a

Thought 2. Attempt to construct F:

• Choose F(a) arbitrarily.

• For small ∆z = b − a, we expect that

F(b) ≈ F(a) + ∆z · f (a).

• For general b, we would like to define

n
F(b) = F(a) + n→∞
lim ∑ ∆zi · f (ai−1).
∆zi →0 i=1

Insert an illustration

27
28 Path integrals

Problem.

Insert an illustration

Thought 3. Integrate along path γ from a to b.

R
Question. Is f (z) dz independent of γ?
γ

2.1 Preliminaries
Definition 2.1.1. Let a 6 b be real numbers and f : [a, b] → C continuous. We define

Zb Zb Zb
f (t) dt = Re( f (t)) dt + i · Im( f (t)) dt
a a a

and
Za Zb
f (t) dt = − f (t) dt.
b a

Proposition 2.1.2. Let a 6 b be real numbers and f : [a, b] → C continuous. Then the
following hold.
Rb
(1) a f (t) dt is C-linear in f , i.e.,

Zb Zb Zb
( f + g)(t) dt = f (t) dt + g(t) dt
a a a

and
Zb Zb
(c · f )(t) dt = c · f (t) dt
a a

for all continuous g : [a, b] → C and c ∈ C.

(2) For a = a0 6 . . . 6 an = b,

Zb n Zai
f (t) dt = ∑ f (t) dt
a i=1 a
i−1
2.2. Path integrals 29

∂ F/∂ x

(3) If f = F 0 = ∂ F/∂ y
for differentiable F : [a, b] → C, then
Zb
f (t) dt = F(b) − F(a).
a

(4) Let α : [c, d] → [a, b] be a continuous map where c 6 d. Then

α(d)
Z Zb
f (t) dt = f (α(s)) · α0 (s) ds.
α(c) a

(5)
Zb Zb
f (t) dt 6 | f (t)| dt 6 (b − a) · max{| f (t)| | a 6 t 6 b}.
a a

Note that {| f (t)| | a 6 t 6 b} is compact as the image of the compact interval [a, b]
under the continuous map | f | : [a, b] → R, and therefore its maximum exists.

2.2 Path integrals

Definition 2.2.1. Let U be an open subset of C. A path in U is a continuous map
γ : [a, b] → U where a 6 b. A path γ : [a, b] → U is closed if γ(a) = γ(b). It is smooth
if it continuously differentiable. We denote its derivative by
d 
0 d dt Re(γ(t))
γ (t) = γ(t) = d .
dt dt Im(γ(t))
Definition 2.2.2. Let U ⊂ C be open, γ : [a, b] → U smooth and f : U → C continuous.
The path integral of f along γ is
Z Z Zb
f = f (z) dz = f (γ(t)) · γ 0 (t) dt.
γ γ a

The arc length of γ is

Zb
`(γ) = |γ 0 (t)| dt.
a

Example 2.2.3. (1) Let p ∈ U. The constant path based at p is the
path γ : [0, 1] → U
with γ(t) = p for all t ∈ [0, 1]. Its derivative is γ 0 (t) = 00 for all t ∈ [0, 1].
Therefore
Z1 Z Z1
0
`(γ) = |γ (t)| dt = 0 and f = f (γ(t)) · γ 0 (t) dt = 0
0 γ 0
for any continuous function f : U → C.
30 Path integrals

(2) Let p, q ∈ U. The linear path from p to q is the path γ : [0, 1] → U given by
γ(t) = p + t · (q − p) for t ∈ [0, 1]. Its derivative is γ 0 (t) = q − p for all t ∈ [0, 1].
Therefore
Z1 Z Z1
`(γ) = |q − p| dt = |q − p| and f = f (γ(t)) · (q − p) dt
0 γ 0

for any continuous function f : U → C. For example, if γ(t) = t is the linear path
from 0 = γ(0) to 1 = γ(1), then

Z Z1
f = Re( f (t))dt
γ 0

is the same as the usual integral of the real valued function Re ◦ f : [0, 1] → R.
(3) Let w ∈ C and r > 0. The (parametrized) circle of radius r around w is the path

Cr (w) : [0, 1] −→ C
t 7−→ w + e2πi·t

Insert an illustration

Its derivative is Cr (w)0 (t) = 2πi · re2πi·t and its length is

Z1 Z1
2πi·t
`(Cr (w)) = |2πi · re | dt = 2πi · r · |e2πi·t | dt = 2πi · r.
| {z }
0 0 =1

Let w = 0 and r = 1. Consider f : C× → C with f (z) = zn for n ∈ Z. If n 6= 1,

then f = F 0 for F(z) = n+1
1 n+1
z and thus
Z Z
(!)
f = zn dz = F(1) − F(1) = 0,
C1 (0) C1 (0)

as we show in Proposition 2.2.4. For n = −1, this integral is more interesting:

Z2π Z2π Z2π

1 1  d it  1
Z
it
dz = · e dt = · i · e dt = i dt = 2πi.
C1 (0) z eit dt eit
0 0 0
2.2. Path integrals 31

Proposition 2.2.4. Let U ⊂ C be open, f : U → C continuous and γ : [a, b] → U a

smooth path. Then
R
(1) γ f is C-linear in f .
(2) For a = a0 6 . . . 6 an = b and γi = γ|[ai−1 ,ai ] ,
Z n Z
f = ∑ f.
γ i=1 γ
i

(3) If f = F 0 for a holomorphic function F : U → C, then

Z
f = F(γ(b)) − F(γ(a)).
γ

(4) Let α : [c, d] → [a, b] be continuously differentiable with α(c) = a and α(d) = b.
Then Z Z
f = f.
γ◦α γ

(5) Z
f 6 `(γ) · max{| f (z)| | z ∈ im(γ)}.
γ

(6) Let γ − : [a, b] → U be defined by γ − (t) = γ(a + b − t). Then

Z Z
f = − f.
γ− γ

Proof. Properties (1) and (2) follow directly from the corresponding properties of
Proposition 2.1.2.
Property (3) follows from the direct computation
Z Zb Zb
0 0
f = F (γ(t)) · γ (t) dt = (F ◦ γ)0 (t) dt = F(γ(b)) − F(γ(a))
γ a a

where we use the chain rule for multivariate real-valued functions for the second equality
and Proposition 2.1.2.(3) for the third equality.
Property (4) follows from the direct computation
Z Zd
f = f (γ ◦ α(s)) · (γ ◦ α)0 (s) ds
γ◦α c
Zd
= f (γ(α(s))) · γ 0 (α(s)) · α0 (s) ds
c
32 Path integrals

Zb Z
= f (γ(t)) · γ 0 (t) dt = f
a γ

where we use Proposition 2.1.2.(4) for the third equality.

Property (5) follows from the direct computation

Z Zb
f = f (γ(t)) · γ 0 (t) dt
γ a
Zb
6 | f (γ(t))| · |γ 0 (t)| dt
a
6 `(γ) · max{| f (z)| | z ∈ im(γ)}

where we use Proposition 2.1.2.(5) for the first inequality. The second inequality follows
from | f (γ(t))| 6 max{| f (z)| | z ∈ im(γ)}.
Property (6) follows analogously to (4), using that γ − = γ ◦ α for the map α :
[a, b] → [a, b] with α(t) = a + b − t. The difference is that the end points are reversed,
i.e., α(a) = b and α(b) = a, which leads to

Z α(b)
Z Zb
0
f = f (γ(t)) · γ ( f ) dt = − f.
γ− α(a) a

Outlook: Cauchy’s integral theorem

Theorem (Cauchy’s integral theorem). Let U ⊂ CR be open, f : U → C holomorphic
and γ : [0, 1] → C a path. If γ is contractible, then f = 0.
γ

Insert an illustration

Question.

• What does it mean that a path is contractible?

R
• What is f for a path γ that is not smooth?
γ

Strategy.
R
(1) Goursat: γ f = 0 for triangular paths.
2.2. Path integrals 33

(2) Primitives for f on discs.

R
(3) γ f for arbitrary paths.
R
(4) Homotopy invariance of γ f.
(5) Cauchy’s integral theorem and more.