PMATH 352 Spring 2022

Complex Analysis

Instructor: Alan Talmage

Lecture Notes
Latex by Justin Li
 Contents

1 Complex Number and Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.1 Complex Number 3
1.2 Basic Topology, Limit, Continuity and Differentiability 5
1.3 Modulus and Argument 8

2 Holomorphic functions and CR equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.1 Holomorphic Function 10
2.2 Smooth Curve 18

3 Integration and Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1 Integration 21
3.2 Cauchy’s Theorem and its Integration Formula 29
3.3 Liouvlle Theorem and Maximum Modulus Principle 32
3.4 Morera’s Theorem 34
3.5 Series 35
3.6 Integration II 43
 1. Complex Number and Functions

1.1 Complex Number

Definition 1.1.1 — Complex Number.
The complex number is defined as

C = {a + bi : a, b ∈ R} where i2 = −1

Note: There is no prior distinction between i and −i, then all behavior in C should be invariant under the
map i ←→ −i

Question: Is this a good definition?

For any a, b, c, d ∈ R, we have

(a + bi) + (c + di) = (a + c) + (b + d)i closed under addition

(a + bi)(c + di) = (ac − bd) + (bc + ad)i closed under multiplication

For a, b, c, d ∈ R with c, d 6= 0, we have
a + bi a + bi c − di ac + bd bc − ad
= · = 2 + i closed under division
c + di c + di c − di c + d 2 c2 + d 2
The complex number are closed under its operations, then it’s a good defintion.

Definition 1.1.2 — Conjugation.

The conjugation in C is defined as
a + bi = a − bi
where a, b ∈ R.
4 Chapter 1. Complex Number and Functions

Moreover, we have
(a + bi)(a + bi) = (a + bi)(a − bi) = a2 + b2

 Remark 1.1 There is a conical bijection between C and R2 , that is

a + bi ←→ (a, b)
we usually can write it as C ∼
= R2

Definition 1.1.3 — Norm. The norm in C is defined as

q p
|a + bi| = (a + bi)(a + bi) = a2 + b2

 Remark 1.2 The in for n ∈ N has a cycle 4, that is

i=i i2 = −1 i3 = −i i4 = 1 i5 = i ....
| {z }
cycle4

 Remark 1.3 The polar coordinate is defined differently in C and R2 .

For (x, y) ∈ R2 we have
p y
x = r cos θ y = r sin θ r= x2 + y2 = k(x, y)k θ = arctan
x
For a + bi ∈ C we have
p b
a = r cos θ b = r sin θ r= a2 + b2 = |a + bi| θ = arctan
a

Behaviors in R and C function operation:

Let x ∈ R and z = a + bi, then
x3 + 2x + 1 =⇒ z3 + 2z + 1
For z, w ∈ C, we insist that dd ez = ez and ew+z = ew · ez , then we have
ex =⇒ ez = ea+bi = ea · ebi
Note that i is a constant, so for y ∈ R we have
d iy 1 d iy d 2 iy d d
e = · e = eiy and e = (ieiy ) = i · eiy = i2 eiy = −eiy
d(iy) i dy dy2 dy dy
| {z }
=ieiy
d2
Therefore, the f (y) = eiy satisfies dy2
= − f , then by ODE the f must be:

f (y) = A sin y + B cos y and f 0 (y) = A cos y − B sin y

but we can see that f (0) = B = ei0 = 1, f 0 (0) = iei0 = i. This gives us that
f (y) = eiy = cos y + i sin y
Therefore, we have
ea+bi = ea ebi = ea (cos b + i sin b) = ea cos b + iea sin b
 1.2 Basic Topology, Limit, Continuity and Differentiability 5

1.2 Basic Topology, Limit, Continuity and Differentiability

Definition 1.2.1 — Distance.
The distance between two points w, z ∈ C is

|w − z| (or |z − w|)

Thus C and R2 are isomorphic as metric spaces.

Definition 1.2.2 — Open Set and Closed Set.

An open set S ⊆ C is a set such that for every z ∈ S, there exists ε > 0 s.t.

|z − w| < ε =⇒ w∈S

A set S ⊆ C is closed if C \ S is open.

Definition 1.2.3 — Limit.

Let f : C → C, we say
lim f (z) = L
z→w

if ∀ε > 0, ∃δ > 0 s.t. |z − w| < δ , then

| f (z) − f (w)| < ε

 Example 1.1 Calculate the following:

z
lim
z→0 z
Let’s try to approach in different directions.
Let z = x ∈ R, then
z x x
lim = lim = lim =1
z→0 z x→0 x x→0 x
Let z = iy for y ∈ R, then we have
z iy −iy
lim = lim = lim = −1
z→0 z y→0 iy y→0 iy
so the limit does not exist. 

Definition 1.2.4 — Continuity.

A function is continuous at point z0 if

lim f (z) = f (z0 )

z→z0

A function is continuous on a set S if it’s continuous at each point of S

 Example 1.2 Let f : C → C where f (z) = z2 . Consider ∆z = z − z0 , then

lim f (z) = lim z2 = lim (z0 + ∆z)2 = lim z20 + 2z0 ∆z + ∆z2 = z20
z→z0 z→z0 ∆z→0 ∆z→0

so f is continuous everywhere. 
6 Chapter 1. Complex Number and Functions

Definition 1.2.5 — Correctness, Path-Correctness and Domain.

A set S is connected if S there is no open sets S1 , S2 with S1 ∩ S2 = 0/ such that

S ⊆ S1 ∪ S2

A set S is path-connected if ∀z1 , z2 ∈ S, there exists a path from z1 to z2 lying in S where a path is the
image of [0, 1] under a continuous

A domain is a path-connected open set.

 Example 1.3 Consider a set S ⊆ R2 where

   
1
S= x, sin : x > 0 ∪ {(0, y) : y ∈ R}
x

is a connected set. 

Proposition 1.2.1
Path connected =⇒ connected

Proof: Let X be a path-connected set and x0 ∈ X. For each y ∈ X, find a continuous map f : [0, 1] → X
such that
f (0) = x0 and f (1) = y
Since an interval is connected and the image of continuous map preserve correctness, then f ([0, 1]) is
connected. Therefore, y belongs to the largest connected set that contains x0 , so X is connected.

Definition 1.2.6 — Differentiability.

Let f : C → C if
f (z) − f (z0 )
lim exists
z→z0 z − z0
we say that f is differentiable at z0 and that the limit is its derivative f 0 (z0 )

 Example 1.4 Let f (z) = z2 , then we have

f (z0 + ∆z) − f (z0 ) z2 + 2z0 ∆z + ∆z2

lim = lim 0 = lim 2z0 + ∆z = 2z0
∆z→0 ∆z ∆z→0 ∆z ∆z→0


1.2 Basic Topology, Limit, Continuity and Differentiability 7

 Example 1.5 Let f (z) = |z|, then we have

f (z0 + ∆z) − f (z0 ) |z0 + ∆z| − |z0 |

lim = lim
∆z→0 ∆z ∆z→0 ∆z
√
Let’s consider g(a, b) = a2 + b2 , then we have

∂g a ∂g b
=√ (∗) and =√ (∗∗)
∂a a + b2
2 ∂b a + b2
2

Now we try ∆z = x ∈ R and z0 = a + bi where a, b ∈ R, then

p √
|x + a + bi| − |a + bi| (a + x)2 + b2 − a2 + b2 a
lim = lim =√ by (*)
x→0 x x→0 x a + b2
2

Similarly, if we let ∆z = yi for y ∈ R, then by (**) we have

p √
a2 + (b + y)2 − a2 + b2 1 ∂ g 1 b bi
lim = = √ = −√
y→0 yi i ∂y i a +b
2 2 a + b2
2

Therefore, f (z) = |z| is differentiable nowhere 

Proposition 1.2.2
Let f , g : C → C and c ∈ C, then we have

( f + g)(z)0 = f 0 (z) + g0 (z) (c f )0 (z) = c f 0 (z)

( f g)0 (z) = f (z)g0 (z) + f 0 (z)g(z) ( f ◦ g)0 (z) = f 0 (g(z))g0 (z)

Definition 1.2.7 — Real Part and Imaginary Part.

Let z ∈ C with z = a + bi where a, b ∈ R, then a and b are call real and imaginary parts of z respectively
denoted Re(z) and Im(z).

 Example 1.6 Let f (z) = Re(z), what’s the differentiability of f ?

We look at
f (z + h) − f (z)
lim
h→0 h
Let h = hx ∈ R, then
Re(a + hx + bi) − Re(a + bi) a + hx − a
lim = lim =1
hx →0 hx hx →0 hx
Let h = ihy for hy ∈ R, then

Re(a + ihy + bi) − Re(a + bi) a−a

lim = lim =0
hy →0 ihy hy →0 ihy

Therefore, f (z) = Re(z) is differentiable nowhere. Similarly, Im(z) is differentiable nowhere. 

 8 Chapter 1. Complex Number and Functions

 Remark 1.4

Im(z) = Re(−iz) and z = Re(z) − i Im(z)

differentiable nowhere.

Intuition: Differentiable functions are those that acts on z and are blind to Re(z) and Im(z)

 Example 1.7 The function f : C → C with f (z) = z is a reflection by real-axis 

 Example 1.8 The function f : C → C with f (z) = z2 , if we write z = a + bi, then

f (z) = z2 = (a + bi)2 = (a2 − b2 ) + 2abi

Recall z = a + bi = reiθ , so let z = reiθ , then

f (z) = z2 = r2 ei(2θ )

1
Question: What is i 2 ?

π π π 1 π 1 π π π 1 i
i = ei 2 = cos + i sin = 0 + i = i =⇒ i 2 = (ei 2 ) 2 = ei 4 = cos + i sin = √ + √
2 2 4 4 2 2

1.3 Modulus and Argument

Definition 1.3.1 — Modulus.

Let z = reiθ with 0 ≤ r ∈ R and θ ∈ R. The modulus (or magnitude or absolute value) of z is r = |z|

Definition 1.3.2 — Argument.

Let z = reiθ with 0 ≤ r ∈ R and θ ∈ R. The argument of z is θ = arg(z)

 Remark 1.5 The argument of z is not unique. That’s because

π π π π
i = ei 2 = ei( 2 +2π) = ei( 2 +4π) = .... = ei( 2 +nπ) for n ∈ Z

We also note that

π 1
n π π
o n iπ i5π o
i = ei( 2 +2π) =⇒ i 2 ∈ ei( 2 +2π) , ei( 2 +4π) = e 4 , e 4
1.3 Modulus and Argument 9

Proposition 1.3.1
More generally, if n > 0 and n ∈ Z, z = reiθ , then
θ +2(n−1)π
1 1
n 1 θ 1 θ +2π 1 θ +4π 1
o
z n = (reiθ ) n = r n ei n , r n ei n , r n ei n , ...., r n ei n

so any non-zero z ∈ C, has exactly n distinct nth roots:

1 θ +2kπ
r n ei n where 0 ≤ k < n
 2. Holomorphic functions and CR equations

2.1 Holomorphic Function

Definition 2.1.1 — Holomorphic.
If f : C → C is differentiable on a domain D, we say f is holomorphic on D. Also called (complex)
analytic regular.

Sloppy terminology warning: A function is said holomorphic at point z0 if it is holomorphic on some

open set containing z0

Proposition 2.1.1
Let f : C → C be holomorphic on a domain D and let z ∈ D, then

f (z + h) − f (z)
f 0 (z) = lim extsts
h→0 h
Consider h = hx ∈ R, then
f (z + hx ) − f (z)
f 0 (z) = lim
h→0 hx
and let z = x + iy, then
f (x + hx + iy) − f (x + iy)
f 0 (z) = lim
h→0 hx
2.1 Holomorphic Function 11

Let f (z) = f (x + iy) = u(x, y) + iv(x, y), where u, v : R2 → R, then

u(x + hx , y) + iv(x + hx , y) − u(x, y) − iv(x, y)

f 0 (x + iy) = lim
hx →0 hx
u(x + hx , y) − u(x, y) v(x + hx , y) − v(x, y)
= lim + i · lim
hx →0 hx hx →0 hx
∂u ∂v
= +i
∂x ∂x
= f 0 (x + iy)

Now let h = ihy where hy ∈ R, then

u(x, y + hy ) + iv(x, y + hy ) − u(x, y) − iv(x, y)

f 0 (x + iy) = lim
hy →0 ihy
u(x, y + hy ) − u(x, y) iv(x, y + hy ) + iv(x, y)
= lim + · lim
hy →0 ihy hy →0 ihy
1 ∂u ∂v
= +i
i ∂y ∂y
∂v ∂u
= −i
∂y ∂y
Therefore, we have
∂u ∂v ∂v ∂u ∂u ∂v ∂v ∂u
f 0 (x + iy) = +i = −i =⇒ = and =−
∂x ∂x ∂y ∂y ∂x ∂y ∂x ∂y
These are the Cauchy-Riemann equations. If f = u + iv is holomorphic on D, then u and v satisfy the
CR equations on D.

 Example 2.1 Let f be holomorphic on a domain D and let v(x, y) = Im( f ) = xy on D, find u(x, y). Let
f = u + iv, then
∂u ∂v ∂u ∂v
= =x and − = =y
∂x ∂y ∂y ∂x
so that
∂u ∂u 1 1
=x = −y =⇒ u = x2 +C1 (y) u = − y2 +C2 (x)
∂x ∂y 2 2
This gives us that
1 1
u(x, y) = x2 − y2 +C
2 2
so we have  
1 2 1 2
f (x + iy) = x − y +C + xyi
2 2
Note that
(x + iy)2 x2 − y2 + 2xyi
 
1 2 1 2
= = x − y +C + xyi
2 2 2 2
then
z2
f (z) = +C and f 0 (z) = z
2
12 Chapter 2. Holomorphic functions and CR equations

where z = x + iy 

 Example 2.2 Let f be holomorphic on a domain D and let Re( f ) = x2 y. Let z = x + iy and f (x + iy) =
u(x, y) + iv(x, y), u(x, y) = x2 y. Then we have
∂v ∂u
Z
= = 2xy =⇒ v= 2xydy = xy2 + C1 (x)
∂y ∂x | {z }
cause contradiction

∂v ∂u 1
Z
=− = x2 =⇒ v= x2 dx = x3 + C2 (y)
∂x ∂y 3 | {z }
cause contradiction
this implies that there is no such v exists, so this is a contradiction, there is no such function f . 

Proposition 2.1.2
∂u ∂v ∂v ∂u ∂v ∂v ∂u ∂u
f 0 (z) = +i = −i = +i = −i
∂x ∂x ∂y ∂y ∂y ∂x ∂x ∂y

 Example 2.3 Let f (z) = ez , so consider

f (x + iy) = ex eiy = ex · (cos(y) + i sin(y)) = ex cos(y) + iex sin(y)

Then
∂u ∂v ∂u ∂v
= ex cos(y) = = −ex sin(y) =
∂x ∂y ∂y ∂x


Theorem 2.1.3 Let u, v : R2 → R has continuous particle derivatives at (x0 , y0 ) which satisfy

∂u ∂v ∂u ∂v
= =−
∂x ∂y ∂y ∂x

at (x0 , y0 ). Then f (x + iy) = u(x, y) + iv(x, y) is holomorphic at z0 = x0 + iy0

Proof: Let D ⊆ C be domain with z0 ∈ D. Let z = x + iy ∈ D. Then

   
∂u ∂u
u(x, y) = u(x0 , y0 ) + (x − x0 ) (x0 , y0 ) − ε1 (x, y) + (y − y0 ) (x0 , y0 ) − ε2 (x, y)
∂x ∂y

where ε1 , ε2 are continuous at (x0 , y0 ) and ε1 (x0 , y0 ) = ε2 (x0 , y0 ) = 0. Similarly, we have

   
∂v ∂v
v(x, y) = v(x0 , y0 ) + (x − x0 ) (x0 , y0 ) − ε3 (x, y) + (y − y0 ) (x0 , y0 ) − ε4 (x, y)
∂x ∂y
so  
∂u ∂v
f (x + iy) = u(x, y) + iv(x, y) = f (z0 ) + (z − z0 ) (x0 , y0 ) + i (x0 , y0 ) + ε(z)
∂x ∂x
where
x − x0 y − y0
ε(z) = (ε1 + iε3 ) + (ε2 + iε4 )
z − z0 z − z0
2.1 Holomorphic Function 13

satisfying ε continuous at z0 with ε(z0 ) → 0. Then

f (z) − f (z0 ) ∂u ∂v ∂u ∂v
lim = lim (x0 , y0 ) + i (x0 , y0 ) + ε(z) = (x0 , y0 ) + i (x0 , y0 ) = f 0 (z0 )
z→z0 z − z0 z→z0 ∂ x ∂x ∂x ∂x

 Example 2.4 Let f (z) = z so f (x + iy) = x − iy so u = x and v = −y. Note that

∂u ∂v ∂u ∂v
= 1 6= = −1 =0=−
∂x ∂y ∂y ∂y
This holds nowhere, so f is holomorphic nowhere 

 Example 2.5 Let f (z) = 1z so

 
1 x − iy x − iy x y
f (x + iy) = · = = −i 2
x + iy x − iy x2 + y2 x2 + y2 x + y2
Then we have
∂u y2 − x2 ∂v y2 − x 2 ∂u −2xy ∂v −2xy
= 2 = 2 = 2 − = 2
∂ x (x + y2 )2 ∂ y (x + y2 )2 ∂ y (x + y2 )2 ∂ x (x + y2 )2
so f is holomorphic everywhere except {0}

1 z z
Note: f (z) = = = (*) 
z zz |z|2

Proposition 2.1.4 Let f , g : C → C holomorphic at z0 with g(z0 ) 6= 0, then

f
is holomorphic at z0
g
The converse is false because of the Example 1.13, so let f be holomorphic, h is not holomorphic, then
fh
h is holomorphic at z while h(z) 6= 0.

 Example 2.6 Let f (z) = 1z and z = reiθ , then

1 1
f (reiθ ) = = e−iθ
reθ r
Let z = reiθ , z0 = r0 eiθ0 and f (z) be holomorphic and let f (z) = u(r, θ ) + iv(r, θ ) we have
f (z) − f (z0 )
f 0 (z0 ) = lim
z→z0 z − z0
First fix θ = θ0 and r → r0 , then
u(r, θ0 ) + iv(r, θ0 ) − u(r0 , θ0 ) − iv(r0 , θ0 )
f 0 (reiθ0 ) = lim
r→r0 reiθ0 − r0 eiθ0
u(r, θ0 ) − u(r0 , θ0 ) v(r, θ0 ) − v(r0 , θ0 )
= e−iθ0 lim +i
r→r0 r − r0 r − r0
 
∂u ∂v
= e−iθ0 · (r0 , θ0 ) + i (r0 , θ0 )
∂r ∂r
14 Chapter 2. Holomorphic functions and CR equations

Next let r = r0 and θ → θ0 , so similarly we have

  
1 u(r0 , θ ) − u(r0 , θ0 ) v(r0 , θ ) − v(r0 , θ0 ) θ − θ0
f 0 (r0 eiθ ) = lim +i
r0 θ →θ0 θ − θ0 θ − θ0 eiθ − eiθ0
 
1 1 ∂u ∂v
= iθ
+i
r ie ∂θ ∂θ
 
1 ∂v ∂u
= e−iθ −i
r ∂θ ∂θ
= f 0 (reiθ )
 
−iθ ∂ u ∂v
=e +i
∂r ∂r

Therefore, we have

∂u 1 ∂v ∂v 1 ∂u
= =−
∂r r ∂θ ∂r r ∂θ

Polar Form of CR-Equation 

 Example 2.7 Consider

   
1 1θ θ
z = r cos
n n + i sin
n n

so we have
   
1 θ θ
u(r, θ ) = r cos
n v(r, θ ) = sin
n n

so that
 
∂ u 1 1 −1 θ 1 ∂v
= r n cos =
∂r n n r ∂θ

and
 
∂ v 1 1 −1 θ 1 ∂u
= r n sin =−
∂r n n r ∂θ


2.1 Holomorphic Function 15
Definition 2.1.2 — Branch and Principal Branch.
Let’s define
Arg(reiθ ) = θ −π ≤ θ ≤ π
This is a branch of the multvalued function arg. In particular, Arg is called the principal branch

Other breanches:
f (reiθ ) = θ π < θ ≤ 3π = Arg(reiθ ) + 2π
so we have the value
Arg(z) + 2πm for m ∈ Z
is a branch of arg(z)
1 1 θ
Consider the branches z n = r n ei n for −π < θ π
Question: Where is this function holomorphic?
Answer: This function is not continuous along Arg(z) = π, but it’s holomorphic on C \ R≤0 , this is called
branch cut

1 θ
 Example 2.8 The function f (reiθ ) = r n ei n for −π < θ ≤ π has a branch cut along R≤0 but holomorphic
on C \ R≤0
√ √ 1
We use branches in R as well: 4 = 2 because · means principal branch of the function x n

We can put a branch cut somewhere else:

1 θ
f (reiθ ) = r n ei n for 0 < θ < 2π

 Remark 2.1 Some times we can define a branch cut that is not a straight line.

 Example 2.9 Let’s consider the log function, that is

   
log(z) = log reiθ = log(r) + log eiθ = log(r) + iθ = ln |z| + iArg(θ ) + 2πim = ln |z| + iArg(z) + 2πim

for m ∈ Z. Use log for the multivalued function and log for the principal branch. That is

log(z) = log |z| + iArg(z) = log |z| + 2πim

for m ∈ Z 

 1
 Example 2.10 Let’s consider log z 2 , so we have

 1  1 θ 1 θ 1 1
log z = log r 2 ei 2 = ln r 2 + i = (ln |z| + iθ ) = log(z)
2
2 2 2

16 Chapter 2. Holomorphic functions and CR equations

 Remark 2.2 Note that for w, z ∈ C:

log(wz) 6= log(w) + log(z)

in general. We can see that
 2πi 2πi  2πi 4π 2πi 2πi
log e 3 e 3 = − 6= = +
3 3 3 3
That is because of the branches we use.

 Remark 2.3
eiy + e−iy eiy − e−iy
cos(y) = sin(y) =
2 2i
are holomorphic.
Moreover
d eiz − e−iz ieiz + ie−iz eiz + e−iz
 
d
sin(z) = = = = cos(z)
dz dz 2i 2i 2
and
d eiz + e−iz ieiz − ie−iz −eiz + e−iz
 
d
cos(z) = = = = − sin(z)
dz dz 2 2 2

 Remark 2.4 If two holomorphic funcions are equal on "enough" of a set, they must agree on their domains.

Definition 2.1.3 — Tangent Function.

1 eiz − e−iz
·
tan(z) =
2 eiz + e−iz
and all trig identities carry over in the obvious way.

 Remark 2.5 Trig function are not bounded. Consider cos(iy) for y ∈ R, then

ei(iy) + e−i(iy)
|cos(iy)| =
2

so we have |cos(iy)| → ∞ as y → ±∞. The sin function is similar. Therefore, cos, sin are unbounded in C.

 Remark 2.6
ez − e−z ez + e−z
sinh(z) = cosh(z) = and sinh(iz) = i sin(z) cosh(iz) = cos(z)
2 2

Question: What is ii ?
π iπ
Since log(i) = log ei 2 = 2 + 2πk for k ∈ Z, then
 i iπ
ii = elog(i) = ei log(i) = ei·( 2 +2πk) = e− 2 −2πk
π

Define
d w d d 1 w
z2 = ew log(z) = (z ) = ew log(z) = ew log(z) (w log(z) = w · · ew log(z) = · zw = w · zw−1
dz dz dz z z
as expected.
2.1 Holomorphic Function 17

Question: How many values does zw have?

For k ∈ Z we have
zw = ew log(z) = ew(log(z)+2πik) = ew log(z) · e2πikw
When is e2πikw = e2πinw ?
Then are equal when for some n ∈ Z
e2πikw = e2πinw+2πin
Consider ez = 1, then ez = 1ei·0 = ex+iy = ex+iy = ex eiy implies ex = 1 and y = 2πn. Now we can see that
kw = nw + m for some m ∈ Z, then
m
w=
k−n
p 1
for n, m, k ∈ Z, so the powers zw repeats if and only if w ∈ Q. Now if w = qp , so z2 = z q = (z p ) q has q distinct
values. If z 6= 0, we have 
1 if w ∈ Z

p
zw = q if w = q ∈Q


∞ otherwise

Proposition 2.1.5 — Rotalion Approximation.

Let f be holomorphic at z0 , so that

f (z) − f (z0 )
f 0 (z0 ) = lim
z→z0 z − z0
The modolus and argument must converge individually.

f (z) − f (z0 )
f 0 (z0 ) = lim =⇒ | f (z) − f (z0 )| ≈ f 0 (z0 ) |z − z0 |
z→z0 z − z0

and  
0 f (z) − f (z0 )
arg( f (z0 )) = lim arg
z→z0 z − z0
for some branch of arg holomorphic near z0 , f (z0 ), f 0 (z0 ) so

arg( f 0 (z0 )) = arg( f (z) − f (z0 )) − arg(z − z0 ) =⇒ arg( f (z) − f (z0 )) ≈ arg( f 0 (z0 )) + arg(z − z0 )

near z0 we have
0 (z ))
f (x) ≈ f (z0 ) + ei arg( f 0
f 0 (z0 ) (z − z0 )
this is a roralion of z − z0 by arg( f 0 (z0 )) and a scaling by | f 0 (z0 )|

√ iπ
 Example 2.11 Consider f (z) = z2 so f (reiθ ) = r2 ei2θ . Let z0 = 1 + i =
√ 2e 4 , so f (z0 ) = 2i and f 0 (z0 ) =
π
2 2ei 4 so for small h = z − z0 we have
0 (z ))
f (z0 + h) ≈ f (z0 ) + ei arg( f 0

 Remark 2.7
If f is differentiable on an open interval (a, b) and f 0 (x) = 0 on (a, b), then f is constant on (a, b)
 18 Chapter 2. Holomorphic functions and CR equations

Theorem 2.1.6
If f is holomorphic on a domain D and f 0 (z) = 0 for all z ∈ D, then f is constant on D

Proof:
∂u ∂v ∂v ∂u
f 0 (z) = 0 = +i = −i
∂x ∂x ∂y ∂y
so that
∂u ∂u ∂v ∂v
= = = =0
∂x ∂y ∂x ∂y
so u, v are constant on any horizontal or vertical line segment in D, but D is a domain so it’s open and path
connected. Then any two points in D can be connected by a path of horizontal and vertical line segment, so
u and v are constant on D, that means f = u + iv is constant on D.

1
 Example 2.12 Find a branch of (z2 − 1) 2 holomorphic on |z| > 1
1
Note that the principal branch of z 2 does not work:
1 2 −1
e 2 log(z )

Its branch cut is where z2 − 1 ∈ R with z2 − 1 ≤ 0. But let z = 2i, so z2 − 1 = −5 ≤ 0.

1
Consider the principal branch of f (z) = z(1 − z12 ) 2 , its branch cut lies wherever 1 − z12 ≤ 0 in R, which
is z12 ≥ 1 in R ⇐⇒ z2 ≤ 1 in R =⇒ |z| < 1 

2.2 Smooth Curve

Definition 2.2.1 — Smooth Curve.
A smooth curve in C is the image of the function r : [a, b] → C satisfying:

1. r is continuous and differentiable on [a, b]

2. r0 6= 0 on [a, b]
3. r is one to one

 Remark 2.8 The definition of smooth curve results gaps, sharp corners, pausing, retracing and self-

intersection are not smooth curve.

Definition 2.2.2 — Directed Smooth Curve.

A directed smooth curve is a smooth curve with a fixed direction. i.e. The points on the curve are ordered
and and r must trace them in order.

Definition 2.2.3 — Contour.

Sn
A contour is a directed piecewise smooth curve. i.e. Γ = i=1 Ci where each Ci is directed smooth curve
and the terminal point of Ci is the initial point of Ci+1 .

Definition 2.2.4
A contour is simple if it has no self-intersection
2.2 Smooth Curve 19
Definition 2.2.5
A contour is closed if its initial point coincides with its terminal point.

Definition 2.2.6 A simple closed contour is a contour both simple and closed.

 Example 2.13 Let Γ : r1 (t) = z0t + z1 (1 − t) for t ∈ [0, 1] where z0 , z1 are constants.
Note: Parametrizations are not unique. That means we have
1
r2 (t) = z0 (2t) + z1 (1 − 2t) for t ∈ [0, ] r3 (t) = z0t 2 + z1 (1 − t 2 ) for t ∈ [0, 1]
2


 Example 2.14 Consider Γ : r(t) = R · eit + z0 for t ∈ [0, 2π] 

Definition 2.2.7 If Cr (z0 ) = r · eit + z0 for t ∈ [0, 2π] is the circular contour with radius r and center z0 ,
traversed counterclockwise

 Example 2.15 Consider Γ = C1 ∪C2 ∪C3 where

C1 : r1 (t) = t C2 : r2 (t) = ti + (1 = t) C3 : r3 (t) = (1 − t)i
where t ∈ [0, 1] so we can parametrize Γ by

r1
 t ∈ [0, 1]
r(t) = r2 t ∈ [0, 1]

r3 t ∈ [0, 1]

20 Chapter 2. Holomorphic functions and CR equations

Theorem 2.2.1 — Jordan Curve Theorem.

A simple closed contour divides C into two disjoint regions, a bounded interior and an unbounded exterior.

Definition 2.2.8
A simple closed contour is positively oriented if its interior is to the left when traversed, negatively
oriented clockwise otherwise.
 3. Integration and Series

3.1 Integration

Definition 3.1.1 — Partition. Let Γ be a directed with initial point w0 and terminal point w1 . A partition
of Γ is a set of points where
w0 = z0 , z1 , z2 , ...., zn = w1
such that for all 0 ≤ i < n, zi+1 is further along Γ than zi .

Definition 3.1.2 — Mesh.

The mesh of a partition is the largest distance between two consecutive points zi , zi+1 along Γ

Definition 3.1.3 — Riemann Sum.

Let Γ lie on a domain D and let f : D → C. The Riemann sum of f with respect to Pn is
n
S f (Pn ) = ∑ f (zi ) · (zi − zi−1 )
i=1

Definition 3.1.4
f is integrable along Γ if
lim S(Pn ) exists
mesh(Pn )→0

Definition 3.1.5 — integral.

If f is integrable along Γ the integral of f along Γ is
Z
f= lim S(Pn )
Γ mesh(Pn )→0

 Remark 3.1 This definition does not reference a parametrization of Γ, thus the integral is independent of
22 Chapter 3. Integration and Series

the choice of parametrization of the curve.

Let Γ parametrized by r : [a, b] → Γ, then

n−1
lim
mesh(Pn )→0 i=0
∑ f (zi )(zi+1 − zi )

Let t0 ,t1 , ...,tn be the partition of [a, b] s.t. r(ti ) = zi and 0 ≤ i ≤ n. This gives us

n−1
lim
mesh(Pn )→0 i=0
∑ f (r(ti ))∆zi

where ∆ti = ti+1 − ti and ∆zi = zi+1 − zi , so that

n−1 n−1 Z b
0
lim ∑ f (r(ti ))∆zi = ∆tlim ∑ f (r(ti ))r (ti )∆ti = f (r(t))r0 (t)dt
mesh(Pn )→0 i=0 →0 i i=0 a

That is
Z Z b
f (z)dz = f (r(t))r0 (t)dt very important
Γ a

 Remark 3.2 Let’s define the integral over a contour. First we consider

Γ = Γ1 + Γ2 + .... + Γn

where the Γi are smooth directed curves, to be

Z Z Z Z
f= f+ f + ... + f
Γ Γ1 Γ2 Γn

The contour integral immediately satisfies the followings

Z Z Z Z Z Z Z Z
f +g = f+ g f= f+ f c· f = c· f
Γ Γ Γ Γ1 +Γ2 Γ1 Γ2 Γ Γ

for some constant c ∈ C

3.1 Integration 23

 Example 3.1 Let Γ : r(t) = eit where t ∈ [0, π], then by Remark 1.14 we have
 π
1 2it 1
Z Z π Z π
2it
zdz = it
e (ie )dt = i it
e dt = i e = (e2πi − e0 ) = 0
Γ 0 0 2i 0 2
also we have
1 2
Z Z π Z π
z2 dz = (eit )2 (ieit )dt = i e3it dt = (e3iπ − e3i·0 ) = −
Γ 0 0 3 3


 Example 3.2 Let C1 (0) = r(t) = eit for t ∈ [0, 2π], then by Remark 1.14 we have
Z 2π
1
Z
zdz = eit (ieit )dt = (e4πi − e0 ) = 0
C1 (0) 0 2
and Z 2π Z 2π
1
Z
= ieit dt = idt = 2πi
C1 (0) 0 eit 0


 Example 3.3 Let’s define Γ1 : r1 (t) = t with t ∈ [0, 1] and Γ2 : r2 (t) = 1 + it with t ∈ [0, 1], then by Remark
1.14 we have
Z Z Z
z2 dz = z2 dz + z2 dz
Γ Γ1 Γ2
Z 1 Z 1
2
= t (1)dt + (1 + it)2 (i)dt
0 0
2 2
=− + i
3 3


 Example 3.4 Let C1 (z0 ) : r(t) = z0 + eit for t ∈ [0, 2π], then by Remark 1.14 we have
Z Z 2π Z 2π Z 2π
(z − z0 )n dz = (z0 + eit − z0 )n (ieit )dt = enit · ieit dt = i ei(n+i)t dt
C1 (z0 ) 0 0 0

by solving the integral we get

( (
1 2πi(n+1) − e0 )
Z
n
Z 2π
i(n+i)t n+1 (e if n 6= −1 0 if n 6= −1
(z − z0 ) dz = i e dt = =
C1 (z0 ) 0 2πi otherwise 2πi otherwise


Definition 3.1.6 — Length.

The length of a contour Γ parametrized by r : [a, b] → Γ with
Z b
r0 (t) dt
a
24 Chapter 3. Integration and Series

Theorem 3.1.1
Let f be integrable on Γ and | f (z)| ≤ M on Γ, then
Z Z b Z b Z b
0 0
f (z)dz = f (r(t))r (t)dt ≤ f (r(t))r (t) dt = | f (r(t))| r0 (t) dt
Γ a a a

Definition 3.1.7 — Primitive.

A function F is primitive (or antiderivative) for a function f on a domain D if F is holomorphic on D and
for all z ∈ D with F 0 (z) = f (z)

Let f have a primitive F on D and Γ lie on D, consider

Z Z
f (z)dz = F 0 (z)dz
Γ Γ

Let r : [a, b] → Γ parametrize Γ, then

Z b Z b Z r(b)
dF dr dF
Z
0 0 0
F (z)dz = F (r(t))r (t)dt = (r(t)) dt = dr = F(r(b)) − F(r(a))
Γ a a dr dt r(a) dr

by the fundamental theorem of calculus in R

Theorem 3.1.2
Fundamental Theorem of Calculus in C: If f has a primitive F on a domain D and Γ lies in D with
initial point z0 and terminal point z, then
Z
f (z)dz = F(z1 ) − F(z0 )
Γ

 Example 3.5 Let f (z) = z so it has primitive F(z) = 21 z2 on all of C. Then for Γ containing from z0 to z1 ,

1
Z
zdz = z2 |zz10
Γ 2
If z1 = 1 + i and z0 = 0, so we get
1
Z
zdz = z2 |1+i =i
Γ 2 0


 Example 3.6 Let f (z) = 1z has a primitive log(z), that is any branch of log(z) is a primitive of 1z on its
1
domain. so log(z) is primitive of z on the domain C \ R≤0 , then

1
Z  iπ   −iπ 
dz = log(i) − log(−i) = log e 2 − log e 2 = iπ
Γ z
1
but for f (z) = z has no primitive valid on all of C1 (0) 
3.1 Integration 25

Corollary 3.1.3
If f has a primitive on a domain D and Γ is a closed contour lying in D, then
Z
f (z)dz = 0
Γ

Proof: Note that Z

f = F(z1 ) − F(z0 ) = F(z0 ) − F(z0 ) = 0
Γ
H
that is primitive implies f =0

Lemma 3.1.4
Let f be continuous on a domain D and let I
f =0
Γ
for any closed Γ lying in D. Then given Γ1 , Γ2 in D with the same initial and terminal points, then
Z Z
f= f
Γ1 Γ2

Proof: Note that Γ1 + (−Γ2 ) is closed, so

Z Z Z
f =0= f− f =0
Γ1 +(−Γ2 ) Γ1 Γ2

Lemma 3.1.5
Let f be continuous on a domain D such that for Γ1 , Γ2 in D sharing initial and terminal point
Z Z
f= f
Γ1 Γ2

then f has a primitive on D

Proof: Fix z0 ∈ D and define Z

F(z) = f (z)dz
Γ
where Γ is a contour lying in D with initial point z0 and terminal point z. This is well-defined by path-
independent (and path-connectedness of D). Now consider
R
0 F(z + ∆z) − F(z) E f (z)dz
F (z) = lim = lim
|∆z|→0 ∆z |∆z|→0 ∆z

where E is the line segment running from z to z + ∆z, that is E : r(t) = z + t∆z for t ∈ [0, 1]. Then we have
R Z 1 Z 1
0 E f (z)dz 1
F (z) = lim = lim f (z + t∆z)∆zdt = lim f (z + t∆z)dt
∆z→0 ∆z |∆z|→0 ∆z 0 |∆z|→0 0
26 Chapter 3. Integration and Series

since f is continuous, so
lim f (z + t∆z) = f (z)
∆z→0

so for all ε > 0 ∃δ > 0 s.t. as ∆z < δ , then f (z + t∆z) − f (z) < ε. Now we can see that
Z 1 Z 1
0 ≤ lim f (z + t∆z)dt ≤ lim f (z) + εdt = lim f (z) + ε = f (z)
|∆z|→0 0 |∆z|→0 0 ε→0

Theorem 3.1.6
Let f be continuous on a domain D, TFAE:

1. f has a primitive on D Z
2. For all closed contours Γ lying in D, f =0
Γ
3. For any two contours Γ1 , Γ2 in D that sharing intial and terminal points, then
Z Z
f= f
Γ1 Γ2

Definition 3.1.8 — Cauchy Sequence.

A Cauchy sequence is a sequence {zn }∞
n=1 such that ∀ε > 0 ∃N > 0 such that n1 , n2 > N

|zn1 − zn2 | < ε

Lemma 3.1.7
A Cauchy seuqence in a compact set S ⊆ Rn converges to a point in S

Lemma 3.1.8 Any closed and bounded subset of Rn is compact

Lemma 3.1.9 Let f be holomorphic at z0 , then

f (z) = f (z0 ) + f 0 (z0 )(z − z0 ) + ε(z)(z − z0 )

for some ε(z) satisfying lim ε(z) = 0

z→z0

Proof: Let
f (z) − f (z0 )
ε(z) = − f 0 (z0 )
z − z0
then take lim .
z→z0
3.1 Integration 27

Theorem 3.1.10 — Goursat’s Theorem.

Let f be the holomorphic on a domain D and let T be a triangle lying D with interior in D then
Z
f (z)dz = 0
T

Proof: Divide T into four triangles by connecting the midpoints of its sides. Now
Z Z Z Z Z
f= f+ f+ f+ f
T T1 T2 T3 T4

there exists a Ti such that Z Z

f ≤4 f
T Ti

Note that length(T1 ) ≤ 12 length(T ) and diam(T1 ) ≤ 12 diam(T ). Repeat this process, yielding T = T (0) , T (1) , ...
such that Z Z
f ≤ 4n f
T T (n)

with length(T (n) ) ≤ 21n length(T ) and diam(T (n) ) ≤ 21n diam(T ).
Let zn be a point in the interior of T (n) for each n, then {zn } is a Cauchy sequence. Then lim zn = w where
n→∞
w lies in the interior of each T (n) . Since f is hlomorphic at w, then

f (z) = f (w) − f 0 (w)(z − w) + ε(z)(z − w)

lim ε(z) = 0. Now consider

z→w
Z Z
f (z)dz = f (w) + f 0 (w)(z − w) + ε(z)(z − w)dz
T (n) T (n)

Note that f (w) has primitive z f (w) and f 0 (w)(z − w) has primitive 21 f 0 (w)(z − w)2 so
Z
f (w) + f 0 (w)(z − w)dz = 0
T (n)

so that Z Z
f (z)dz = ε(z)(z − w)dz
T (n) T (n)
Let’s define εn = sup |ε(z)| and then
z∈T (n)

1
|z − w| ≤ diam(T (n) ) ≤ diam(T )
2n
and
1
length(T (n) ) ≤ length(T (n) )
2n
so that
1
Z Z
f (z)dz = ε(z)(z − w)dz ≤ εn diam(T (n) )length(T (n) ) ≤ εn diam(T )length(T )
T (n) T (n) 4n
28 Chapter 3. Integration and Series

thus
1
Z Z
n
f (z)dz ≤ 4 f (z)dz ≤ 4n εn diam(T )length(T ) = εn diam(T )length(T )
T T (n) 4n
let n → ∞ and εn → 0, we get
Z Z
f (z)dz → 0 =⇒ f (z)dz = 0
T (n) T (n)

which completes the proof.

Corollary 3.1.11
The Goursat’s Theorem also works for retangles and polygons.

Corollary 3.1.12
If f is holomorphic on an open disk, then f has a primitive on that disk

Proof: Choose z0 ∈ D and define Z

F(z) = f (z)dz
Γ
so that Z Z Z
F(z + h) − F(z) = f (z)dz + f+ f
Γn 
| ∆ {z }
=0

so that
d
Z
f (z)dz = f (z)
dz Γn

as in last lecture.

 Example 3.7 f holomorphic on domain D does not imply f has primitive on D.

1
Let f (z) = z is holomorphic on
{z ∈ C : 1 < |z| < 2}
and
1
Z
dz = 2πi
C1 (0) z


Definition 3.1.9 — Homotopic.

Let Γ1 , Γ2 be two contours in a domain D with the same initial and terminal point. Γ1 is homotopic (or
continuously deformable) if there exists r : [0, 1]2 → C satisfying:

1. r is continuous on [0, 1]2

2. For a fixed s, r(s,t) is a parametrization of a contour in D with initial and terminal point shared with
Γ1 , Γ2
3. r(0,t) parmetrizes Γ1 , r(1,t) parmetrizes Γ2
 3.2 Cauchy’s Theorem and its Integration Formula 29
Definition 3.1.10 — Simply Connected.
A domain D is simply connected if any two contours in D sharing initial and terminal point are homotopic
to each other.

3.2 Cauchy’s Theorem and its Integration Formula

Theorem 3.2.1 — Cauchy’s Theorem.

Let f be holomorphic on a simply connected domain D and let Γ be a closed contour in D, then
Z
f =0
Γ

Proof: Γ is homotopic and triangle.

 Example 3.8 Let f (z) = z2 , since z2 is entire so by Cauchy’s Theorem,

Z −1
2
Z Z
f (z)dz = f (z)dz = x2 dx = −
Γ Γ1 1 3


 Example 3.9 Let f (z) = z2 1−1 , so f is holomorphic on C \ {1, −1}, then

Z Z Z
f (z)dz = f (z)dz + f (z)dz
C2 (0) Cε (−1) Cε (1)

for ε ∈ (0, 2). Note that  

1 1 1 1
= · −
z2 − 1 2 z−1 z+1
then  
1  1 1  1
Z Z Z
f (z)dz = · dz −  = 2 · 2πi = πi
dz
Cε (1) 2 Cε (1) z − 1 Cε (1) z + 1

| {z }
=0
30 Chapter 3. Integration and Series

Similarly, we have
 
1  1 1  1
Z Z Z
f (z)dz = · dz −  = 2 · −2πi = −πi
dz
Cε (−1) 2 Cε (−1) z − 1 Cε (−1) z + 1

| {z }
=0

so Z
f (z)dz = 0
C2 (0)


 Example 3.10 Say f has a taylor series at z0 :

f (z) = f (z0 ) + a1 (z − z0 ) + a2 (z − z0 )2 + .....

Then we get
 
1 f (z) 1  f (z0 )
Z Z
dz = + a1 + a2 (z − z0 ) + ...dz

2πi z − z 2πi z − z

Γ 0 Γ 0 | {z }
holomorphic
f (z0 ) 1
Z
= dz
2πi Γ z − z0
f (z0 )
= · 2πi
2πi
= f (z0 )

Theorem 3.2.2 — Cauchy Integral Formula.

Let f be function holomorphic on a domain Ω ⊆ C, Γ is a jordan curve (closed contour) contained in Ω
and whose interior in contained in Ω. Let z0 ∈ the interior of Γ, then

1 f (z)
Z
f (z0 ) = dz
2πi Γ − z0
z

Proof: WE can replace Γ with C(r) = {z ∈ C : |z − z0 | = r} for small enough r, then

1 f (z) 1 f (z)
Z Z
dz = dz
2πi Γ z − z 0 2πi C(r) − z0
z
1 f (z0 ) f (z) − f (z0 )
Z Z
= dz + dz
2πi C(r) z − z0 C(r) z − z0
f (z) − f (z0 )
Z
= f (z0 ) +
C(r) z − z0

We will show that

f (z) − f (z0 )
Z
=0
C(r) z − z0
3.2 Cauchy’s Theorem and its Integration Formula 31

Since
f (z) − f (z0 )
f 0 (z0 ) = lim
z→z0 z − z0
f (z)− f (z0 )
exists and z−z0 is bounded on C(r) and its interior S0 , so

f (z) − f (z0 )
Z
dz → 0
C(r) z − z0

as r → 0. so since this integral is independent of r and it must equal 0, which complesthes the proof.

 Example 3.11 Say g is holomorphic on 0 < |z| < R, which of the following implies that
Z
g(z)dz = 0
C(r)

(a) g is holomorphic at 0
(b) g is identically 0 on 0 < |z| < R
(c) |g| is bounded on 0 < |z| < R
(d) g(z) = 2πi on 0 < |z| < R
(e) g is defined and continuouos at 0
(f) limz→0 g(z) = ∞

Answer: (a)(b)(c)(d)(e) 

 Example 3.12 Let Γ = {z ∈ C : |z| = 1}, compute

R cos(z)
(a) Γ z dz
R ez
(b) Γ z−2 dz
R cos(2πz)
(c) 2z−1 dz

By Cauchy Integral Thm/Formula, (a)(b) are 0. 

Proposition 3.2.3

— Cauchy Integral Formula for Derivaives.
1 0 1
Note that 1−x = (1−x)2
, then the CIF:

1 f (z) d 1 d f (z) 1 d f (z)

Z Z Z
f (w) = dz =⇒ f (w) = dz = dz
2πi Γ z−w dw 2πi dw Γ z−w 2πi Γ dw z − w

Taking devaritive again:

1 f (z) 2 f (z)
Z Z
f 0 (w) = dz =⇒ dz
2πi Γ (z − w)2 2πi Γ −w
z
 32 Chapter 3. Integration and Series

Taking devaritive n times:

n! f (z)
Z
f (n) (w) = n+1
dz
2πi Γ (z − w)

so f is infinitely differentiable!

Proposition 3.2.4
Let f be holomorphic on Ω with D = {|z − z0 | < R} ⊆ Ω, then

n!M
f (n) (z0 ) ≤
rn
where M = max | f (z)|.
|z|=r

Proof: Take Γ = {|z − z0 | = r} then apply prop 3.1.15, then we have

n! f (z) n! M n! M n!M
Z Z
f (n) (z0 ) ≤ n+1
dz ≤ n
dz = n
· 2πi = n
2πi Γ (z − z0 ) 2πi Γr 2πi r r

3.3 Liouvlle Theorem and Maximum Modulus Principle

Theorem 3.3.1 — Liouivlle.

A bounded entire (holomorphic on C) function is constant.

Proof: By the CIF for derivatives we have | f 0 (z)| ≤ n!M r for any r > 0, so since M canbe taken inde-
pendent of r, we get | f 0 (z0 )| = 0 for all z0 so f is constant.

Theorem 3.3.2 — Maximum Modulus Principle.

A non-constant holomorphic function on a domain Ω cannot chieve its supremum on Ω. More precisely,
for all z0 ∈ Ω, there is some z1 ∈ Ω with | f (z1 )| > | f (z0 )|

Theorem 3.3.3
Every non-constant complex polynomials has a root in C

Proof: Let p be a complex polynomial with no root in C, we will show that p is constant. Then we
have 1p is entire, let
m(r) = max |p(z)|
|z|=r

then m(r) increases as r → ∞, so g(r) = min |p(z)| decreases as r → ∞, but lim |p(z)| = ∞ is p is not
|z|=r z→∞
constant, which they couldn’t true at the same time, so it’s a contradiction. That means p is constant.

 Example 3.13
If f is entire and non-constant, which of the followings are true?
3.3 Liouvlle Theorem and Maximum Modulus Principle 33

(a) Image( f ) = f (C) must intersect the upper half plane.

(b) Image( f ) must intersect every straight line.
(c) Image( f ) must intersect every non-empty open set.
(d) Image( f ) must contain every point.

Answer: (a)(b)(c) 

 Example 3.14
If p is a polynomial satisfying |p(z)| ≤ |ez | for all z, what is p(z)?

Answer: Only p(z) = 0 by taking negative z with |z| large. 

Theorem 3.3.4 — Maximum Modulus Priciple.

Let f be holomorphic on an open set Ω. If f achieves its maximum on Ω, then f is a constant. That is, if
there is some z0 ∈ Ω such that | f (z0 )| ≥ | f (z)| for all z ∈ Ω, then f is constant.

Proof: Let z0 be a local max of | f | on Ω, let

D = {|z − z0 | ≤ r} ⊆ Ω

be a disc around z0 . Then the Cauchy integral formula says for C(r) = ∂ D = {|z − z0 | = r}:

f (z0 + reiθ )
Z 2π
1 f (z) 1
Z
f (z) = dz = d(z0 + reiθ )
2πi C(r) z − z 0 2πi 0 reiθ
f (z0 + reiθ )
Z 2π
1
= reiθ dθ
2πi 0 reiθ
Z 2π
1
= f (z)dθ
2πi 0
This gives us that
Z 2π
1
| f (z0 )| ≤ | f (z)|dθ ≤ max | f (z)|
2π 0 z∈C(r)

with equality iff | f | is constant on C(r) with | f (z0 )| = max | f (z)| because r is arbitrary (as long as D ⊆ Ω),
z∈C(r)
we will show f is constant on D. Write f = u + iv then u2 + v2 is constant on D.

2uux + 2vvx = 0 2uuy + 2vvy = 0 since ux = vy and uy = −vx

then we get −2uvx + 2vux = 0. That is to solve

    
u v ux 0
=
v −u vx 0
| {z }
det=−u2 −v2

so either u2 + v2 = 0 or ux = vx = 0. The u2 + v2 = 0 implies f = 0 is a constant or ux = vx = uy = vy = 0

implies f is constant. Therefore, f is constant on D, since D is arbitrary so f is constant on Ω.
 34 Chapter 3. Integration and Series

3.4 Morera’s Theorem

Theorem 3.4.1 — Morera’s Theorem.

If f is continuous on a domain Ω with Z
f (z)dz = 0
Γ
for all simple closed curves Γ ⊆ Ω whose interiors are contained in Ω, then f is holomorphic on Ω.

Proof: We will find a holomorphic F with F 0 = f . This will prove that f is holomorphic. Since holomor-
phicity is local, we can assume that Ω = D is a disc. Now choose z0 ∈ D we define
Z
F(z) = f (z)dz
Γ

where Γ is any path from z0 to z because D is simply connected, this F is well defined by hypothesis.
Compute
Z z Z x 
0 F(z) − F(x) 1
F (z) = lim = lim f (y)dy − f (y)dy
x→z z−x x→z z − x z0 z0
Z z
1
= lim f (y)dy
x→z z − x x
Z z Z z 
f (y) − f (z) f (z)
= lim dy + dy
x→z x z−x x z−x
Z z
f (y) − f (x)
= lim f (x) + dy
x→z x z−x
= f (z) + 0

because the Z z
f (y) − f (x) f (m) − f (x)
dy ≤ |z − x| = | f (m) − f (x)|
x z−x z−x
where m = the max value of f on (x, z), completes the proof.

Lemma 3.4.2 — Symmetry Principle.

Let D be a domain symmetric across R, let D+ , D− , I be as indicated. Let f + be holomorphic on D+ , f −
be holomorphic on D, both extend continuously to I and f + (z) = f − (z) for z ∈ I, then
 +
 f (z)
 z ∈ D+
f (z) = f + (z) = f − (z) z ∈ I
 −
z ∈ D−

f (z)

is holomorphic on D.

Proof: Note that if f is continuous on D, then

Z Z  
0
f (z) − f (z) ≤ ε · max f (z) · length(T) → 0
T Tε z∈T
 3.5 Series 35

as ε → 0

Proposition 3.4.3 — Schwarz Reflection Principle.

Let D+ , D− , I as be the ones defined in Lemma 3.1.22. Let f + be holomorphic on D+ and extend continu-
ously to I, then there exists f such that f (z) = f + (z) on D+ and f is holomorphic on D

Proof: Let f + (z) = f (z), by A2 we have f − is holomorphic on D− , then apply the Symmetry Prin-
ciple

Lemma 3.4.4 — Schwarz’s Lemma.

Let D = {z ∈ C : |z| < 1} and f be holomorphic on D, f (0) = 0 and | f (z)| ≤ 1 for all z ∈ D. Then
| f (z)| ≤ |z| for all z ∈ D and | f 0 (0)| ≤ 1. Furthermore, if | f (z)| = |z| for some 0 6= z ∈ D, then f is rotation
f (z) = λ z for some constant |λ | ≤ 1.

Proof: Let ( f (z)

z z 6= 0
g(z) =
0
f (0) z=0
Note that g is holomorphic on D, since

f (z) f (z) − f (0)

lim = lim = f 0 (0)
z→0 z z→0 z−0
Now consider g on |z| < r < 1, then

| f (w)| 1
|g(z)| ≤ max |g(w)| ≤ max ≤ →1
|w|=r |w|=r |w| r

as r → 1. so we have |g(z)| ≤ 1 on D, then for z 6= 0

| f (z)|
≤1 =⇒ | f (z)| ≤ |z|
|z|

for z = 0 we have |g(0)| = | f 0 (0)| ≤ 1. If | f (z)| = |z| at some z ∈ D, then |g(z)| = 1, so by maximum
modulus theorem g is constant on D. Let g(z) = λ and |z| = 1, then we have f (z) = λ z as desired.

 Remark 3.3 If f is holomorphic on domain D, then f is infinitely differentialbe on D

 Remark 3.4 In R, an infinitely differentiable function has a Taylor series representation.

3.5 Series
Definition 3.5.1 — Convergent Series.
n
A series ∑∞
n=1 zn is convergent if lim ∑ zi converges.
n→∞
i=1
36 Chapter 3. Integration and Series
Definition 3.5.2 — Cauchy Series.
∞
A series ∑∞
n=1 zn is Cauchy if lim ∑ zn = 0.
k→∞
n=k

Definition 3.5.3 — Uniformly Convergent.

A sequence { fn } is uniformly convergent on a set if ∀ε > 0, ∃N > 0, ∀z ∈ S, ∃L, ∀n > 0

| fn (z) − L| < ε

Lemma 3.5.1 If fn → f uniformly on S, then

Z Z
fn → f
S S

Definition 3.5.4 — Uniformly Convergent Series.

A series is uniformly convergent if its sequence of partial sum is uniformly convergent.

Definition 3.5.5 — Absolutely Convergent Series.

∞
A series is absolutely convergent if the series ∑ zn converges.
n=0

Definition 3.5.6
Let’s define
Dr (z0 ) = {z ∈ C : |z − z0 | < r}
be the open disk of radius r centered at z0 . Let Dr (z0 ) = Dr (z0 ) ∪Cr (z0 ) be its closure.

Definition 3.5.7
Let {xn } ⊆ R, then
lim sup xn = lim sup xk
n→∞ k>n

Proposition 3.5.2 — Ratio Test.

∞ ∞
zn+1 zn+1
If lim sup zn < 1, then ∑ zn converges absolutely. If lim sup zn > 1, then ∑ zn diverges.
n=0 n=0

Proposition 3.5.3 — Root Test.

1
∞ 1
∞
If lim sup |zn | n < 1, then then ∑ zn converges absolutely. If lim sup |zn | n > 1, then ∑ zn diverges
n=0 n=0
3.5 Series 37

Proposition 3.5.4 — Comparison Test.

∞ ∞
If ∑ xn converges with xn ∈ R and |zn | ≤ xn for all n, then ∑ zn converges absolutely.
n=0 n=0

Proposition 3.5.5 — Weierstrass M Test.

∞ ∞
Let { fn }∞
n=1 satisfy | f n (z)| ≤ Mn for all z ∈ S. If ∑ Mn converges, then ∑ fn (z) converges uniformly on S
n=1 n=1
n
Proof: Let gn (z) = ∑ fk (z), then gn is uniformly Cauchy on S
k=0

Definition 3.5.8 — Power Series.

∞
A Power Series about z0 is a series of the form ∑ an (z − z0 )n where zn ∈ C
n=0

Theorem 3.5.6
∞
If a power series ∑ an (z − z0 )n converges at point z with |z − z0 | = R, then it converges absolutely on
n=0
DR (z0 ) and converges uniformly on any closed subdisk of Dr (z0 )

Proof: Let w ∈ DR (z0 ) and |w − z0 | < r < R, then

an (w − z0 )n  r n
|an (w − z0 )n | = |an (z − z0 )n | · ≤ M ·
| {z } an (z − z0 )n R
→0 so is bounded ≤ M | {z } | r{z }
≤ Rr R <1

∞
r n
∑ an (w − z0 )n converges absolutely by


Now we can see that M R is a convergent geometric series so
n=0
comparison. Apply the Weierstrass M-Test to the above to get uniformly convergence on Dr (z0 )

Theorem 3.5.7 — Taylor’s Theorem.

Let f be holomorphic on Dr (z0 ), then for all z ∈ DR (z0 ),
∞
f (n) (z0 )
f (z) = ∑ (z − z0 )n
n=0 n!

Proof: Choose z ∈ DR (z0 ) and let |z − z0 | < r < R, by Cauchy Integration formula

1 f (w)
Z
f (z) = dw
2πi Cr (z0 ) w − z
38 Chapter 3. Integration and Series

for all w ∈ Cr (z). Then

f (w) ∞ z − z0 n (z − z0 )n
  ∞  
f (w) f (w) f (w) 1
= = · z−z0 = ∑ w − z0 = ∑ f (w) · (w − z0 )n+1
w − z (w − z0 ) − (z − z0 ) w − z0 1 − w−z w − z0 n=0 n=0 0

Now we have
(z − z0 )n |z − z0 |n 1 z − z0 n
   
f (w) · ≤ max | f (w)| · n+1 = · max | f (w)|
(w − z0 )n+1 w∈Cr (z0 ) r r w∈Cr (z0 ) r
z−z
Since r ) < 1, then by weierstrass M-test, this series converges uniformly on Cr (z0 ), thus we may
integrate term by term. Then

1 ∞
(z − z0 )n 1 ∞ f (w) ∞
f (n)(z0 )
Z Z
n
f (z) =
2πi ∑
Cr (z0 ) n=0
f (w)
(w − z0 )n+1
dw = ∑
2πi n=0 Cr (z0 ) (w − z0 )
n+1
dw(z − z0 ) = ∑ n! (z − z0 )n
n=0

 Remark 3.5 The term "analytic" means expressible as n power series many text will use "analytic" in place
of "holomorphic".

 Example 3.15
∞
zn z2
ez = ∑ n! = 1 + z + 2! + ......
n=0

with R = ∞

∞
z2n+1 z3 z5
sin(z) = ∑ (−1)n (2n + 1)! = z − 3! + 5! + ....
n=0

with R = ∞

∞
z2n z2 z4
cos(z) = ∑ (−1)n (2n)! = 1 − 2! + 4! + ....
n=0

with R = ∞

∞
1
= ∑ zn = 1 + z + z2 + ....
1 − z n=0

with R = 1 

 Example 3.16 Taylor series for e2z about 0, we have

∞
(2z)n 4z2 8z3
e(2z) = ∑ = 1 + 2z + + + ...
n=0 n! 2! 3!

Converges for |2z| < ∞, implies that R = ∞. 

3.5 Series 39
1
 Example 3.17 Let’s look at 1 2 , let w = − 21 z2 , then
2 z +1

∞ 
1 2 n (−1)n 2n
∞  ∞
1 1 n
1 2
= = ∑ w = ∑ − z = ∑ n
z
2z +1
1 − w n=0 n=0 2 n=0 2

1 2
√ √
and 2z < 1 implies |z| < 2, so R = 2 

 Example 3.18 Taylor series for cos(z) + i sin(z) about 0:

∞
z2n ∞
z2n+1 z2 z3 z4 ∞ n n
i z
cos(z) + i sin(z) = ∑ (−1)n (2n)!
+ i · ∑ (−1)n
(2n + 1)!
= 1 + iz − − i + + .... = ∑
2! 3! 4! n!
= eiz
n=0 n=0 n=0

 Example 3.19 Recall for r > 0

(
Z
0 n 6= 1
(z − z0 )n =
Cr (z0 ) 2πi n = −1

Let f be analytic on DR (z0 ), then

∞
f (n) (z0 )
f (z) = ∑ (z − z0 )n
n=0 n!

uniform convergence on Dr (z0 ) and the convergence is uniform on closed subdisks.

so for all r < R we have
Z Z ∞ ∞ Z

Cr (z0 )
f (z)dz = ∑ an (z − z0 )n dz =
Cr (z0 ) n=0
∑ an · Cr (z0 )
(z − z0 )n dz = 0
n=0

z
 Example 3.20 Find a series representation for ze2 about z = 0.

ez 1 ∞ zn 1 ∞
zn
= =
∑ n! zn (1 + z + ...) = ∑
z2 z2 n=0 n=0 (n + 2)!

Now consider for r > 0

ez ∞
zn ∞
zn 1
Z Z Z
2
= ∑ =∑ = 2πi · = 2πi
Cr (0) z Cr (0) n=0 (n + 2)! n=0 Cr (0) (n + 2)! (−1 + 2)!

Lemma 3.5.8
If fn → f is uniformly on S, then Z Z
fn → f
S S
40 Chapter 3. Integration and Series

Theorem 3.5.9
Let f be holomorphic on a domain {z ∈ C : r1 < |z − z0 | < r2 }, then on that annulus, f has a Laurent
Series (generalized Cauchy series)
∞
f (z) = ∑ an (z − z0 )n
−∞

which converges on the annulus and converges uniformly on closed subannuli.

Definition 3.5.9 — Isolated Singnlarity.

An isolated singnlarity of a function f is a point z0 where f is not holomorphic, but where f is holomor-
phic on some punnctured disk 0 < |z − z0 | < r

1
 Remark 3.6 If f (z) has zero at z0 , then f (z) has singularity at z0 .

Definition 3.5.10 — Zero of Order m.

f (z) f (z)
An analytic function f has a zero of order m at z0 if (z−z0 )m is analytical at z0 but (z−z0 )m+1
is not.
∞
Equivalently if f (z) = ∑ the order is the smallest n such that an 6= 0
n=0

Definition 3.5.11 — Singularity.

A singularity of f is a point where f is not analytic but is a limit point of the points where f is analytic.

Definition 3.5.12
∞
Let z0 be an isolated singularity of f , let ∑ an (z − z0 )n be the Laurrent series of f at z0
n=0
If a−m 6= 0 but an = 0 for all n > m, we call z0 a pole order of m ⇐⇒ (z − z0 )m f (z) is analytic at
z0 but (z − z0 )m+1 f (z) is not

If a−n = 0 for all n > 0, we call this a removable singularity. In this case, we have
(
f (z) z 6= z0
g(z) =
lim f (z) z = z0
z→z0

is analytic at z0 .

If a−n 6= 0 for infinitely many n > 0, we call this essential singularity.

 Example 3.21 Removable singularity examples:

z3 z5 z2 z4
 
z sin(z) 1
f (z) = f (z) = = · z − + − ... = 1 − + − ...
z z z 3! 5! 3! 5!

3.5 Series 41

 Example 3.22 Let a function

∞
f (z) = ∑ an (z − z0 )n with am 6= 0
n=−m

on a punctured disk 0 < |z − z0 | < r. Let Γ be a simple, closed positively oriented contour in the annulus with
z0 inside the loop, then
Z Z ∞ ∞ Z
f (z)dz = ∑ an (z − z0 )n dz = ∑ an (z − z0 )n = 2πi · a−1
Γ Γ n=−m n=−m Γ

Definition 3.5.13 — Residue.

Given f , z0 , Γ as before, we define the residue of f at z0 to be
1
Z
f (z)dz = a−1 = resz0 ( f )
2πi Γ

Definition 3.5.14 — Meromorphic.

A function f is called meromorphic on a domain D if it’s holomorphic on all of D except for a set of
isolated poles.

Theorem 3.5.10 — Residue Theorem.

Let f be meromorphic on a simply connected domain D and let Γ be a simple, closed, positively oriented
contour lying in D. Let z1 , ..., zk be the poles of f inside Γ, then
Z k
f (z)dz = 2πi ∑ reszi ( f )
Γ i=1

 Example 3.23 Consider f (z) = z21+z = z(z+1)

1
, so for 0 < |z| < 1, then
 
1 1 1 1 ∞ 1
= · = ∑ (−z)n = − 1 + z − z2 ....
z(z + 1) z 1 − (−z) z n=0 z

so it has order 1 and residue 1

for 0 < |z + 1| < 1, then

1 n
 
1 1 1 1 1 −1
= · = ·
z(z + 1) z + 1 z + 1 − 1 z + 1 1 − (z + 1)
=− ∑ (z +1)n = z + 1 −1 −(z +1)−(z +1)2 −..
z + 1 n=0

so is has simple pole and residue −1.

for |z| > 1, then
1
1 n 1 n
 
1 1 1 z 1 1 1 1 1
= · · 1
= 2+ 1
= 2·∑ = .... + 4 + + 3 + 2
z(z + 1) z z + 1 z
z 1+ z z n=0 z z z z

42 Chapter 3. Integration and Series

 Remark 3.7 Recall f has pole of order m at z0 ⇐⇒ (z − z0 )m f (z) has removeable singularity at z0 ,
(z − z0 )m−1 f (z) has a pole at z0 . That is
 
m a−m
(z − z0 ) + ... = a−m + am−1 (z − z0 ) + ....
(z − z0 )m
f (z) f (z)
f has a zero of order m at z0 ⇐⇒ (z−z0 )m has a removeable singgularity at z0 , (z−z0 )m+1
has a pole at z0 .

1
(am (z − z0 )m + ...)) = am + am+1 (z − z0 ) + ..
(z − z0 )m
Let f and g be analytic at z0 , let f have a zero of order m at z0 and let g have a zerp pf order n at z0 , then
f (z) m
f (z) (z−z0 )m (z − z0 )
= g(z)
= (z − z0 )m−n h(z) where h(z) is analytic at z0
g(z) n
(z−z0 )n (z − z0 )

and we see that 

f (z) a zero of order m − n at z0
 if m > n
has a pole of order n − m at z0 if m < n
g(z) 
a removable singularity if m = n


1
 Example 3.24 z(z+1) has simple poles at z = 0 and z = −1 

z+3
 Example 3.25 z3 (z+1)2 (z−2) has order 3 pole at 0, order 2 pole at −1 and simple pole at 2. 

 Example 3.26
z2 z4 1 z2
 
cos(z) − 1 1
2
= 2
1 − + − ... = − + − ...
z z 2! 4! 2! 4!
so it has removeable singularity at z = 0. 

 Example 3.27 Let f has a simple pole at z0 and a Laurent series:

a−1
f (z) = + a0 + a1 (z − z0 ) + ...
z − z0
is same punctured disk about z0 , then

(z − z0 ) f (z) = a−1 + a0 (z − z0 ) + a1 (z − z0 )2 + ...

so that
lim ((z − z0 ) f (z)) = a−1 = resz0 ( f )
z→z0

so if f has a simple pole at z0 , then resz0 ( f ) = lim (z − z0 ) f (z) 

z→z0

1 1 1
 Example 3.28 Let f (z) = z(z+1) , then res0 ( f ) = lim z f (z) = lim z · = = 1. Similarly, we
z→0 z→0 z(z + 1) 0 + 1
1 1
have res−1 ( f ) = lim (z + 1) = =1 
z→−1 z(z + 1) −1
 3.6 Integration II 43

 Example 3.29 Now let f have a pole of order m at z0 , then

a−m a−m+1
f (z) = + + ....
(z − z0 )m (z − z0 )−m+1
so that
(z − z0 )m f (z) = a−m + a−m+1 (z − z0 ) + ...
so
d
((z − z0 )m f (z)) = a−m+1 + 2a−m+2 (z − z0 ) + ... + (m − 1)a−1 (z − z0 )m−2 + .
dz
and
d m−1
((z − z0 )m f (z)) = a−1 (m − 1)! + ...
dzm−1
then
d m−1
lim ((z − z0 )m f (z)) = a−1 (m − 1)!
z→z0 dzm−1

and
1 d m−1
a−1 = resz0 ( f ) = lim m−1 ((z − z0 )m f (z))
(m − 1)! z→z0 dz


z
 Example 3.30 Consider f (z) = e z+1 0
3 , then e + 1 = z 6= 0 so f has an order 3 pole at 0, then

d 2 3 ez + 1
 
1 1
res0 ( f ) = lim 2 z · 3 =
(3 − 1)! z→z0 dz z 2


∞
 Example 3.31 Let f (z) = ∑ an (z − z0 )n be the Laurent series for f in some annulus, so
n=−∞

1
I
a−1 = f (z)dz
2πi Γ

where Γ is simple, closed, positively oriented contour looping around the inner circle of annulus.
Now we see
am 1 f (z)
I
−m−1
(z − z0 ) f (z) = .... + + .... =⇒ am = dz
z − z0 2πi (z − z0 )m+1
f (n) (z0 )
Note: for a Taylor series, this is equivalent to n! by Cauchy’s Integration Theorem. 

3.6 Integration II

g(z)
Proposition 3.6.1 Let f (z) = h(z) where g, h are analytic at z0 . Let g(z0 ) 6= 0 and h(z0 ) = 0, h0 (z0 ) 6= 0.
That is f has a simple pole at z0 , then

g(z) z − z0 z − z0 g(z0 )
resz0 ( f ) = lim (z − z0 ) f (z) = lim (z − z0 ) = g(z0 ) lim = g(z0 ) lim = 0
z→z0 z→z0 h(z) z→z0 h(z) z→z0 h(z) − h(z0 ) h (z0 )
44 Chapter 3. Integration and Series

2πi 4πi
n o
 Example 3.32 Find residues of all poles of f (z) = z3 1−1 , note that z3 − 1 = 0 ⇐⇒ z ∈ 1, e 3 , e 3 , then
1
f has 3 simple poles. Then residue at simple pole z is 3z2
. This gives us that

1 1 2πi 1 4πi
res1 ( f ) = res 2πi (f) = e 3 res 4πi (f) = e 3
3 e 3 3 e 3 3


 Example 3.33 Consider the following:

1 1
Z ∞ Z ∞
4
dx and I= dx
0 x +1 0 x4 + 1

Note that
1
Z ∞
2I = dx
−∞ x4 + 1
Let ΓR be the line segment running from −R to R in R, then

1
Z
2I = lim dz
R→∞ ΓR z4 + 1

Let CR be the upper semicircle running from R to −R, note ΓR +CR is simple closed positive oriented tour. so
we can sue residue theorem. Consider

1 1 1
Z Z
π
4
dz ≤ 4
dz ≤ |πiR| · 4 ≤ 3 → 0 as R → ∞
CR z +1 CR R R R
n kπi o
1
Next we locate the poles of z4 +1
and find their residues. Note that z4 + 1 = 0 ⇐⇒ z ∈ e 4 : k = 1, 3, 5, 7 .
Then
1 n kπi o
resz0 ( f ) = where z0 ∈ e 4 : k = 1, 3, 5, 7
4z3
Then we have
1
I
π
dz = √
CR +ΓR z4 + 1 2
so Z Z Z 
π π
2I = lim = lim f (z)dz − f (z)dz = √ − 0 = √
R→∞ ΓR R→∞ ΓR +CR CR 2 2


Definition 3.6.1
Extended complex plane C ∪ {∞} = Ĉ

Definition 3.6.2
Define the behavior of f (z) at ∞ to behavior of f ( 1z ) at 0.
3.6 Integration II 45

 Example 3.34 For example, let f (z) = z2 + 1, so that f ( 1z ) = z12 + 1 = z12 + 0z + 1, so it has order 2 and
residue 0 and Z
lim z2 + 1dz = 0
R→∞ CR (0)

also Z
f (z)dz = −2πires∞ ( f )
−C∞ (0)


1
1 z +1 1+z 1
 Example 3.35 Let f (z) = z+1
z−i , so that f ( z ) = 1 −i = 1−iz at z = 0 and f ( z ) = 1 so f is analytic at ∞. 
z

 Example 3.36 Let f (z) = sin(z) so f ( 1z ) = sin 1z does not converge as z → 0, so sin(z) has essential singu-
larity at ∞

At an isolated singularity z0

lim f (z) = c ∈ C =⇒ f analytic at z0 (or removeable singularity)

z→z0

lim | f (z)| = ∞ =⇒ f has a pole at z0

z→z0

lim f (z) does not exist in Ĉ =⇒ f has an essential singularity at z0

z→z0


 Example 3.37 Consider the following

1
Z ∞
dx
0 x3 + 1
1 iπ 5iπ
Let f (z) = z3 +1
so f has poles at z = −1, e , e 3 3 . We define

1
Z ∞ Z
I= dx = lim f (z)dz
0 x3 + 1 R→∞ Γ1

and we can see that

1 1
Z
dz ∼ R · 3 = R−2 → 0 as R → ∞
CR z3 + 1 R
2πi
Let Γ2 : r2 (t) = t · e 3 for t ∈ [0, R],
Z R Z R 2πi
1 1 e3
Z Z ∞
2πi 2πi
dz = ·e 3 dt = dt = e 3 f
3
z +1 2πi 3 t3 + 1
 
Γ2 0 0 0
t ·e 3 +1

Now
2πi − 2πi
Z
f = 2πires iπ (f) = e 3
Γ1 +CR −Γ2 e 3 3
This gives us that √
2πi − 2πi 2πi 2 3π
e 3 = I +0−e 3 I =⇒ I=
3 9

46 Chapter 3. Integration and Series
Definition 3.6.3 — Cauchy Principal Value. Z ∞
Given a continuous function f : R → R, define the Cauchy Principal Value of f (x)dx is
−∞
Z ∞ Z R
p.v. f (x)dx = lim f (x)dx
−∞ R→∞ −R
Z ∞
Note that f (x)dx exists, then
−∞
Z ∞ Z ∞
p.v. f (x)dx = f (x)dx
−∞ −∞

 Example 3.38 Find p.v. for

cos(x)
Z ∞
dx
−∞ 1 + x2
cos(z)
Let f (z) = 1+z2
, then
1 iz −iz )
cos(z) 2 (e + e
Z Z
dz = dz
CR 1 + z2 CR 1 + z2

but consider e−iz at z is iR, as R → ∞ e−i(iR) = eR → ∞

Consider
∞ eiz
Z
I = p.v. 2
dz
−∞ 1 + z

then
eiz 1 1
Z
2
dz ∼ 2 · R = → ∞
CR 1+z R R
and
eiz e−1 π
Z  
f (z)dz = 2πiresi ( f ) = 2πi = 2πi =
CR +Γ 2z z=i 2i e
π π
so I = e −0 = e

Now we consider Z ∞ −iz

e
I2 = p.v. dz
−∞ 1 + z2
then
eiz 1 1
Z
2
dz ∼ 2 R ∼ → 0
CR 1+z R R
then similarly we have
e−iz
Z  
π
f = −2πires−i ( f ) = −2πi =
CR +Γ 2z z=−i e
π π
so I2 = − 0 =
e e.
Then we have
cos(x) 1 1
Z ∞
π
dx = I + I2 =
−∞ 1 + x2 2 2 e

3.6 Integration II 47

 Example 3.39 Consider

Z 2π
sin2 θ dθ
0

1 1
Let z = eiθ = cos(θ ) + i sin(θ ), then sin(θ ) = 2i (z + z ), then

1 2 dθ 1 2 1
Z 2π    
1 1
Z Z
sin2 θ dθ = (z + ) dz = (z + ) · dz
0 C1 (0) 2i z dz C1 (0) 2i z iz
1 2 1
Z
=− z + + 3 dz
4i C1 (0) z z
 
1 2 1 1
= − res0 z + + 3 = − (2πi)(−2) = π
4i z z 4i

 Example 3.40 Let f be continuous on [a, b] except at c with a < c < b, then

Z b Z c−ε Z b 
p.v. f (x)dz = lim f (x)dx + f (x)dx
a ε→0 a c+ε

 Example 3.41 Consider

cos2 θ
Z 2π
p.v. dθ
0 1 − 3 sin θ

Let z = eiθ , then cos θ = 21 (z + 1z ) and sin θ = 1 1 dθ

2i (z − z ), dz = iz1 . Now we can rewrite it as

 2
z+ 1z
2 1 1 z4 + 2z2 + 1
Z
3 1
dz = dz
C1 (0) 1 − 2i (z − z ) iz 4i z2 (− 2i3 z2 + z + 2i3 )

Note that
√
3 3 −i ± 2 2
− z2 + z + = 0 =⇒ z=
2i 2i −3


Proposition 3.6.2 Let p(z), q(z) be polynomial with deg(p) ≤ deg(q) − 2, then for any arc CR of CR (0),

p(z)
Z
lim dz = 0
R→∞ CR q(z)

This is because
p(z) Rdeg(p) 1
Z
∼ R · deg(q) = R · R−2 = → ∞ as R → ∞
CR q(z) R R
48 Chapter 3. Integration and Series

Lemma 3.6.3
Let a > 0 a d deg(q) ≥ 1 + deg(p), let CR be the upper half of CR (0), then

p(z)
lim eiaz dz = 0
R→∞ q(z)

Proof: Parameterize CR by Reit with t ∈ [0, π], now

p(z) p(Reit )
Z Z π
iaz it
e dz = eiaRe · · Rieit dt
CR q(z) 0 q(Reit )

Note that
it
eiaRe = eiaR(cos(t)+i sin(t)) = e−aR sin(t)

Then for large enough R, exists K ∈ R such that

p(Reit ) K
it
≤
q(Re ) R

so that
π
p(Reit ) K
Z π Z π Z π Z
it 2
eiaRe · · Rieit dt ≤ e−aR sin(t) Rdt = K e−aR sin(t) dt = 2K e−aR sin(t) dt
0 q(Reit ) 0 R 0 0

2t
Since sin(t) ≥ 2t
π on [0, π2 ], then e−R sin(t) ≤ e−aR π , so
Z π Z π Z π Z π
2 2 2t
−aR sin(t) −aR sin(t) −aR sin(t)
K e dt = 2K e dt ≤ K e dt = 2K e−aR π dt
0 0 0 0
 π  πK
= 2K · − (e−aR − 1) → → 0 as R → ∞
2aR aR

 Remark 3.8
p(z)
Z ∞
dz Need deg(q) ≥ 2 + deg(p)
−∞ q(z)
p(z)
Z ∞
cos(z) dz Need deg(q) ≥ 1 + deg(p)
−∞ q(z)

Lemma 3.6.4 — Jordan’s Lemma.

Let f be meromorphic with a simple pole at z0 , and Γr be parametrized by r(t) = z0 + reiθ with θ1 < θ < θ2 ,
then Z
lim r → 0+ f (z)dz = i(θ2 − θ1 )resz0 ( f )
Γr

Proof: ∞
a−1 a−1
f (z) = + ∑ an (z − z0 )n = + g(z)
z − z0 n=0 z − z0
3.6 Integration II 49

where g is analytic so g is continuous then ∃R such that for 0 < r ≤ R, ∃M > 0 s.t. |g(z)| ≤ M, so that
Z
g(z)dz ≤ M · length(Γr ) = M · (θ2 − θ1 )r → 0 as r → 0+
Γr

Then
a−1 1
Z Z Z θ2 Z θ2
iθ
f (z)dz = dz + 0 = a−1 rie dθ = a−1 idθ i · resz ( f )
Γr Γr z − z0 θ1 reiθ θ1

 Example 3.42 Consider

1
x3
Z ∞
dx
0 1 + x2
1
z3
Let f (z) = 1+z2
with branch cut along the positive real axis, then
1 1
z3 R3
Z
2
dz ∼ ∼ R− 3 → 0 as R → ∞
CR 1 + z2 R2

Similarly,
1 1
z3 r3
Z
4

2
dz ∼ r ∼ r 3 → 0 as r → 0+
CR 1+z 1
Z Z ∞
Let f→ f (z)dz = I, then
Γ 0
1 1
z3 (ze2πi ) 3
Z Z Z
2πi
f (z)dz = dz = dz = I · e 3
Γ2 Γ2 1 + z2 Γ1 1+z 2

1
Let f has simple pole at z = ±i with residues, so that resz ( f ) = z2z3 , then
√
3 1
resi ( f ) = −i res−i ( f ) = −
4 2
Then we have
√ ! 2πi
1 3 1 πie−
I
2πi
 2πi
 3 π
f = 2πi −i − = 0 + 0 + I − Ie = I 1 − e
3 3 =⇒ I= 2πi =√
CR +Cr +Γ1 −Γ2 4 4 2 1−e 3 3


 Example 3.43
Z ∞ ix Z −r eix Z R ix 
e e
p.v. dx = lim dx + dx
−∞ x R→∞,r→0+ −R x r x
then by Jordan’s Lemma
eiz
Z
dz = i(0 − π)res0 ( f ) = −πi
Cr z
Let Γ : [−R, R], then
Z I Z Z
lim f (z)dz = f− f− f = 0 − 0 − (−πi) = πi
R→∞,r→0+ Γ CR +Cr +Γ CR Cr

50 Chapter 3. Integration and Series

 Remark 3.9
Two techniques, either:
1. convert everything to rectangular, clear denominator.
2. convert to a trig function

 Example 3.44 Consider

1
Z ∞
I= dx
0 1 + x3
log(z)
we let f (z) = 1+z3
, branch cut along the positive real axis., then
Z R
log(z) ln(x)
Z
dz = dx
Γ1 1 + z3 r 1 + x3
and
log xe2πi
Z R
Z R
log(z) ln(x) + 2πi
Z
dz = dx = dx
Γ2 1 + z3 r 1 + x3 r 1 + x3
then we have Z R Z R
ln(x) ln(x) + 2πi
Z Z
f− f= 3
dx − = 2πi · I
Γ1 Γ2 r 1+x r 1 + x3


 Remark 3.10 Let f 6= 0 be meromorphic on D and let Γ be a simple, positively oriented closed contour
f0
with Γ and its interior is in D. Consider f is meromorphic and its poles can only lie at poles and zeros of f .
Let z0 be an order-m zeros of f , then

f (z) = (z − z0 )m g(z) g(z0 ) 6= 0 g is analytic

Now we have
f 0 (z) = m(z − z0 )m−1 g(z) + (z − z0 )m g0 (z)
so that
f 0 (z) m(z − z0 )m−1 g(z) + (z − z0 )m g0 (z) m g0 (z)
= = +
f (z) (z − z0 )m g(z) z − z0 g(z)
Let z0 be an order-m pole of f , then
h(z)
f (z) = h(z0 ) 6= 0 h is analytic
(z − z0 )n
so that
−m(z − z0 )m−1 h(z) + (z − z0 )m h0 (z) m h0 (z)
f0 = = − +
(z − z0 )2m z − z0 h(z)

Theorem 3.6.5 — The Argument Principle. Let f be meromorphic and inside a simple, close, positively
oriented contour Γ. Let N0 ( f ) be the number of zeros in Γ and Np ( f ) be the number of poles in Γ (both
couted with multiplicity), then Z 0
1 f (z)
dz = N0 ( f ) − Np ( f )
2πi Γ f (z)
3.6 Integration II 51
Definition 3.6.4 — Curling Number.
Let Γ be a closed contour and let z0 6= Γ. The curling number of Γ about z0 , denoted n(Γ, z0 ) is the unique
integer n such that Γ is homeomorphic to

C1 (z0 ) +C1 (z0 ) + ... +C1 (z0 )

| {z }
n in total

in C \ {z0 }

Lemma 3.6.6 For z0 ∈ Γ,

1
I
dz = 2πi · n(Γ, z0 )
Γ z − z0

Proposition 3.6.7 Let f (Γ) be which Γ : γ(t) : [a, b] → Γ, then

Z b
1 1
Z
dz = γ 0 (t)dt
Γ z − z0 a γ(t) − z0

so that
f 0 (z)
Z b
1 1
Z Z
dz = f 0 (γ(t))γ 0 (t)dt = dz
Γ f (z) − z0 a f (γ(t)) − z0 f (Γ) f (z) − z0
This gives us that
1 f 0 (z)
Z
dz = n( f (Γ), z0 )
2πi Γ f (z) − z0
Moreover, for z0 = 0 we have that

1 f 0 (z)
Z
dz = n( f (Γ), z0 )
2πi Γ f (z) − z0

 Example 3.45 Note that

d f 0 (z)
log( f (z)) =
dz f (z)
then
f 0 (z)
I
dz = [log( f (z))]zz10 = 2πi · n( f (Γ), 0)
Γiθ f (z)
Let f = reiθ we have
log( f (z)) = +iθ


Theorem 3.6.8 — The Dog-walking Theorem.

Let Γ1 , Γ2 be parametrized by γ1 , γ2 : [a, b] → C and ∀t ∈ [a, b] with |γ1 (t) − γ2 (t)| < |γ1 (t)|. Then
n(Γ1 , 0) = n(Γ2 , 0)
52 Chapter 3. Integration and Series
γ2 (t)
Proof: NOte that γ1 , γ2 6= 0, consider Γ : γ(t) = γ1 (t) , then

γ2 (t) γ1 (t) − γ2 (t)

|1 − γ(t)| = 1 − = <1
γ1 (t) γ1 (t)

so Γ lies in D1 (1) so n(Γ, 0) = 0. Let γ1 = r1 eiθ1 and γ2 = r1 eiθ2 where r1 , r2 , θ1 , θ2 are functions of t, then
γ2 r2
γ= =
γ1 r1 e 1 −θ2 )
i(θ

and
n(Γ1 , 0) = θ1 (b) − θ1 (a) and n(Γ2 , 0) = θ2 (b) − θ2 (a)
so that
0 = n(Γ, 0) = θ2 (b) − θ2 (a) − (θ1 (b) − θ1 (a)) = n(Γ2 , 0) − n(Γ1 , 0)
which is n(Γ2 , 0) = n(Γ1 , 0)

Theorem 3.6.9 — The Generalized Dog-walking Theorem.

Let Γ1 , Γ2 be parametrized by γ1 , γ2 : [a, b] → C and ∀t ∈ [a, b] with

|γ1 (t) − γ2 (t)| < |γ1 (t)| + |γ2 (t)|

then n(Γ2 , 0) = n(Γ1 , 0)

Proof: Let γ(t) = γγ12 (t)

(t) , assume for contradiction that exist c > 0 abd t ∈ [a, b] such that γ(t) = −c.
Then γ1 (t) = −cγ2 (t), so that

|γ1 (t) − γ2 (t)| = |(−c − 1)γ2 (t)| = (c + 1)|γ2 (t)|

but |γ2 (t)| + |γ1 (t)| = |γ2 (t)| + |−cγ2 (t)| = (1 + c)|γ2 (t)|, which contradicts the Dog-walking Theorem,
so there is no such c exists. Then Γ : γ(t) lies in the C \ R≤0 , so n(Γ, 0) = 0, that is n(Γ2 , 0) = n(Γ1 , 0).

Theorem 3.6.10 — Rouche’s Theorem.

Let f , g be analytic on and inside a simple closed contour Γ. Let |g(z)| < | f (z)| for all z ∈ Γ, then f + g
and f have the same number of zeros (connted with multiplicity inside)

Proof: Let h = f + g then

|h(z) + (− f (z))| = |g(z)| < |− f (z)|
on Γ. Then
n(h(Γ), 0) = n( f (γ), 0) that is N0 (h) = N0 ( f )

 Example 3.46 All 5 zeros of h(z) = z5 + 3z + 1 lie inside |z| < 2. Let f (z) = z5 and g(z) = 3z + 1 on C2 (0),
so | f (z)| = 32 and |g(z)| = 7 < | f (z)|, so by Rouche’s Theorem h and f have same number of zeros inside
C2 (0) 
3.6 Integration II 53

 Example 3.47 How many zeros does z + 3 + 2ez have in the left half-plane Re(z) < 0? Let ΓR be the
contour as "D" reflect by y-axis. Let f (z) = z + 3 and g(z) = 2ez , so |g(z)| = 2eRe(z) so |g(z)| ≤ 2 on ΓR for
all R and
( (
|3 + iy| z = iy 3 z = iy
| f (z)| ≥ ≥
R−3 |z| = R R − 3 |z| = R

so for all R > 5 we have | f (z)| > |g(z)| on ΓR , then f has the same number of zeros inside ΓR as z + 3 + 2ez ,
f (z) = z + 3 has one zero inside ΓR namely −3, so z + 3 + 2ez has exactly one zero in the left half-plane. 

Definition 3.6.5
A point z is a limit point of a set if there exists a sequence {zn } ⊆ S with zn 6= z but lim zn = z
n→∞

Theorem 3.6.11 Let f be holomorphic on a domain D, let Z ⊆ D be the set of zeros of D if Z has a limit
point in D, f is identically zero on D

Proof: Let z0 be the limit of {wn } ⊆ Z and z0 6= wn for all n, conside Dε (z0 ) for some sufficiently
small ε > 0, that is
∞
f (z) = ∑ zn (z − z0 )n
n=0

on Dε (z0 ). If f is not identically 0 on Dε (z0 ), then there exists a minimal m ≥ 0 such that am 6= 0, write

f (z) = am (z − z0 )m (1 + g(z − z0 ))

where g(z − z0 ) → 0 as z → z0 . Let k be sufficiently large that wk ∈ Dε (z0 ), wk ∈ Dε (z0 ) for all K ≥ k. Now
f (wk ) = 0 but
0 = f (wk ) = am (wk − z0 )n (1 + g(wk − z0 ))
and am 6= 0, (wk − z0 )n 6= 0 and g(wk − z0 ) → 0 as wk0 as k → ∞. so for large enough k, |g(wk − z0 )| < 1, so
1 + g(wk − z0 ) 6= 0, which is a contradiction, so f = 0 on Dε (z0 ). Let U be the interior of Z, we just showed
that U is non-empty, U is open by definition, let {zn } ⊆ U converging zn → z, f is continuous so f (z) = 0,
by earlier argument, z ∈ U. Then U is closed, so V = D \U is open we have D = U ∪V and U ∩V , U,V
are open and D is connected, so one of U,V is empty. U is non-empty, so V = 0/ so U = D, then f is 0 on D

Corollary 3.6.12 Let f , g analytic on D and f (z) = g(z) on S ⊆ D where S has limit point in D, then
f (z) = g(z) on D

Proof: apply the above theorem to f − g.

Corollary 3.6.13 Let f be analytic and non-constant on a domain D and let z0 ∈ D, f (z0 ) = w0 , then there
exist ε > 0 such that Dε (z0 ) ⊆ D and f (z) − w0 has zero in Dε (z0 ) \ {z0 }

Proof: Let f (z) − w0 is a non-constant analytic function, so its zero cannot have a limit point, done.
54 Chapter 3. Integration and Series

Theorem 3.6.14 — Open Mapping Theorem.

If f is holomorphic on a domain D, then f is a open map on D (map open set to to open set)

Proof: It suffices to show f (D) is open. Let z0 ∈ D and f (z0 ) = w0 , let w ∈ C and

g(z) = f (z) − w = f (z) − w0 + w0 − w

Choose δ > 0 such that Dδ (z0 ) ⊆ D and such that f (z) 6= w0 on the circle |z − z0 | = δ which exists by the
previous corollary. Now we choose ε > 0 such that | f (z) − w0 | ≥ ε on |z − z0 | = δ , so for all w ∈ Dε (w0 )
we have | f (z) − w0 | ≥ ε > |w − w0 | on the circle |z − z0 | = δ , so by Rouche’s Theorem g and f (z) − w0
have the same number of zeros in Dδ (z0 ), namely one. Then ∃z ∈ Dδ ⊆ D, g(z) = 0 = f (z) − w =⇒
f (z) = w =⇒ w ∈ f (D), then Dε (w0 ) ⊆ f (D)

 Example 3.48 Let f be analytic on a domain D and Re( f (z)) is constant , then f is constant, Re( f (z)) = K
contains no open set, so f must be constant by the contrapositive of open mapping theorem. 

Definition 3.6.6 — Gamma Function.

The gamma function is defined for s > 0 in R by
Z ∞
Γ(s) = e−t t s−1 dt
0

Lemma 3.6.15 Γ extends to an analytic function on Re(s) > 0 and

Z ∞
Γ(s) = e−t t s−1 dt
0

still holds there

Proof: It suffices to show lemma on

S = {z ∈ C : δ < Re(s) < M}

for any 0 < δ < M < ∞. Let Re(s) = σ , now

Z ∞ Z 1
ε
e−t t s−1 dt = lim e−t t s−1 dt
0 ε>0 ε

Consider Z 1
ε
Fε (s) = e−t t s−1 dt
ε
Note that Fε (s) is analytic with
Z 1
ε
Fε0 (s) = e−t (s − 1)t s−2 dt
ε
3.6 Integration II 55

Recall that the limit of a uniformly convergent sequence of analytic function is analytic. Consider
Z ∞ Z 1
ε
−t s−1
|Γ(s) − Fε (s)| = e t dt − e−t t s−1 dt
0 ε
Z ε Z ∞
= e−t t s−1 dt + 1
e−t t s−1 dt
0 ε
Z ε Z ∞
−t σ −1
≤ e t dt + 1
e−t t σ −1 dt
0 ε

Now for ε < 1:

1 δ −1 ε δ
Z ε
e−t t σ −1 dt ≤ ε · ·ε =
0 δ δ
Similarly we have
Z ∞ Z ∞
1
e−t t σ −1 dt ≤ 1
e−t t M−1 dt → 0
ε ε

as ε1 → 0.
Then Fε (s) → Γ(s) uniformly, so Γ is analytic on S so Γ is analytic on Re(s) > 0

Proposition 3.6.16
Let n ∈ Z≥0 , then Γ(n + 1) = n!

Lemma 3.6.17 For 0 < Re(a) < 1, then

Z ∞ a−1
v π
dv =
0 1+v sin(πa)

Proof: Let v = ex , then

Z ∞ a−1 Z ∞ (a−1)x
v e eax
Z ∞
dv = dx = dx
0 1+v −∞ 1 + ex −∞ 1 + e
x

eaz
Let f (z) = 1+ez and integrate over a region. That is

Z 2π a(R+it)
e eaR
Z
f (z)dz = dt ≤ C · ∼ Ce(a−1)R → 0 as R → ∞
Γ2 0 1 + eR+it eR
Z
f (z)dz ≤ Ce−aR → 0 as R → ∞
Γ4

eax
Z Z R
f (z)dz = x
dx
Γ1 −R 1 + e

eax
Z Z R
2πia
f (z)dz = −e x
dx
Γ3 −R 1 + e
56 Chapter 3. Integration and Series

Note that f has a pole at z = πi, then

eaz
 
az z − πi
lim (z − πi) f (z) = lim (z − πi) = lim e
z→πi z→πi 1 + ez z→πi ez − eπi
−1
ez − eπi

aπi
=e lim
z→πi z − πi

= eaπi · (eπi )−1

= −eaπi
= resπi ( f )

Then we have Z
f (z)dz = 2πi(−eaπi ) = (1 − e2πia ) = I
Γ
so that
eaπi π
I = −2πi 2πi
=
1−e sin(πa)

Theorem 3.6.18
π
Γ(s)Γ(1 − s) =
sin(πs)
Proof: IT suffices to show this on 0 < Re(s) < 1:
Z ∞ Z ∞
−u (1−s)−1
Γ(1 − s) = e u du = e−u u−s du
0 0

Let u = vt, v > 0, then Z ∞

Γ(1 − s) = t e−vt (vt)−s dv
0
This give us
Z  Z ∞ −s
v
Z ∞ ∞ Z ∞Z ∞
−t s−1 −vt −s
Γ(s)Γ(1 − s) = e t te (vt) dv dt = e−t(v+1) v−s dvdt = dv
0 0 0 0 0 v+1

By the lemma from above we have

Z ∞ −s
v π
Γ(s)Γ(1 − s) = dv =
0 v+1 sin(πs)

Definition 3.6.7 — Riemann Zeta Function.

For real s > 1 as
∞
1
ζ (s) = ∑ s
i=1 n

so ζ immediately has an analytic continuation to Re(s) > 1 and the formula

∞
1
ζ (s) = ∑ s
i=1 n
3.6 Integration II 57

is still valid. If s = σ + it for σ ,t ∈ R, if σ > 1+ > 1, then

∞ ∞ ∞ ∞ ∞ ∞
1 1 1 1 1 1
∑ ns ≤∑ = ∑ = ∑ = ∑ ≤ ∑
i=1 i=1 ns i=1 e
s log(n)
i=1 e
σ log(n)
i=1 n σ
i=1 n
1+δ

then ζ (s) is analytic on Re(s) > 1. Consider the Euler product:

1
∏ 1 − ps
p price

for Re(s) > 1, then we see that

  ∞
1 1 1 1
= 1 + + + ..... = ∑
1 − p−s ps p2s i=1 pns

then    
1 1 1 1 1 1 1 1
∏ 1 − p−s = 1 + 2s + 4s + 8s 1 + 3s + 9s + ... 1 + 5s + 25s + ...
s

but we have unique factorization of positive integers:

 s ∞
1 1
formula above = ∑ j j j
= ∑ s
= ζ (s)
j1 , j2 ,... 2 3 5 ... i=1 n
1 2 3

1
Theorem 3.6.19 ζ (s) − s−1 has an analytic continuation to Re(s) > 0. Then ζ (s) is rercmorphic on
Re(s) > 0 with a simple pole of residu 1 at s = 1

Proof: Consider Z n
1 1
∑ ns − 1 xs
ds
1≤n≤N

Let Z n+1
1 1
n (s) = − ds
n ns xs
By the mean value theorem
1 1 |s|
s
− s ≤ σ +1
n x n
on n ≤ x ≤ n + 1. Then we have uniform convergence of n (s) on 1 + σ > 1 ⇐⇒ Re(s) > 0. Then
∞
∑ δn (s)
i=1

is analyticiction on Re(s) > 0. Now

N Z N
1 1
∑ s ∑ δn (s) +
=
xs
ds
1≤n≤N n n=1 1
58 Chapter 3. Integration and Series

Now Z N  ∞
1 1 1 1−s
Z ∞
lim dx = dx = x
N→∞ 1 xs 1 xs 1−s 0
1
on Re(s) > 1, this = s−1 . Thus,
N Z N
1
∑ n (s) + 1 xs
ds
i=1

converges uniformly and matches ζ (s) on Re(s) > 1, so

∞
1
ζ (s) − = ∑ n (s)
s − 1 i=1

is analytic on Re(s) > 0

Theorem 3.6.20 ζ (s) has no zeros on the line Re(s) = 1

Proof: Let x, y ∈ R, y 6= 0 and define

h(x) = ζ 3 (x)ζ 4 (x + iy)ζ (x + 2yi)

Now
1
ζ (s) = ∏
s 1 − p−s
so
1 −s −s


ln |ζ (s)| = ln ∏ = − ∑ ln 1 − p = − Re ∑ log 1 − p
p 1 − p−s p p

Now ∞
wn
− log(1 − w) = ∑
i=1 n

for |w| < 1, so

1
ln |ζ (s)| = Re ∑ ∑ p−sn
p n n
and then

ln |h(x)| = 3 ln |ζ (x)| + 4 ln |ζ (x + iy)| + ln |ζ (x + 2iy)|

1 1 1
= 3 Re ∑ ∑ p−nx + 4 Re ∑ ∑ p−ns−iny + Re ∑ ∑ p−nx−2iny
p p n p n n p n n
1
= ∑ ∑ p−nx Re 3 + 4p−iny + p−2iny


p n n

Note that
p−iny = enyi ln(p) has Re(piny ) = cos(−ny ln(p))
and
Re(p−2iny ) = cos(−2ny ln(p))
3.6 Integration II 59

so that
Re 3 + 4p−iny + p−2iny = 3 + 4 cos(−ny ln(p)) + cos(−2ny ln(p))

Let θ = −ny ln(p), so this is

Re 3 + 4p−iny + p−2iny = 3 + 4 cos θ + cos 2θ = 2(1 + cos θ )2 ≥ 0

Then we have ln |h(x)| ≥ 0, so |h(x)| ≥ 1 we have

|h(x)| ζ (x + iy) 4 1
= |(x − 1)ζ (x)|3 |ζ (x + 2iy)| ≥
x−1 x−1 x−1

As x → 1+ we have |ζ (x + 2iy)| → |ζ (1 + 2iy)|

lim |(x − 1)ζ (x)| = 1

x→1

if ζ (1 + iy) = 0, but
ζ (x + iy)
lim+ = ζ 0 (1 + iy)
x→1 x−1
|h(x)| 1 1
so lim+ converges to some finite value. but this is ≥ x−1 and lim+ = ∞, this is contradiction so
x→1 x − 1 x→1 x−1
ζ (1 + iy) 6= 0