PMATH352W18 - Complex

Analysis

Classnotes for Winter 2018

by

Johnson Ng

BMath (Hons), Pure Mathematics major, Actuarial Science Minor

University of Waterloo
ë Table of Contents

1 Lecture 1 Jan 3rd 2018 11

1.1 Complex Numbers and Their Properties . . . . . . . . . . 11

2 Lecture 2 Jan 5th 2018 19

2.1 Complex Numbers and Their Properties (Continued) . . 19

3 Lecture 3 Jan 8th 2018 23

3.1 Complex Numbers and Their Properties (Continued 2) . 23
3.1.1 Roots of Complex Numbers . . . . . . . . . . . . . 27

4 Lecture 4 Jan 10th 2018 31

4.1 Examples for nth Roots of Unity . . . . . . . . . . . . . . 31

5 Lecture 5 Jan 12 2018 37

5.1 Complex Functions . . . . . . . . . . . . . . . . . . . . . . 37
5.1.1 Limits . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.1.2 Continuity . . . . . . . . . . . . . . . . . . . . . . . 39

6 Lecture 6 Jan 15th 2018 41

6.1 Continuity (Continued) . . . . . . . . . . . . . . . . . . . 41
6.2 Differentiability . . . . . . . . . . . . . . . . . . . . . . . . 42
6.2.1 Cauchy-Riemann Equations . . . . . . . . . . . . . 44

7 Lecture 7 Jan 17th 2018 47

7.1 Differentiability (Continued) . . . . . . . . . . . . . . . . 47
7.1.1 Cauchy-Riemann Equations (Continued) . . . . . 47
7.1.2 Power Series . . . . . . . . . . . . . . . . . . . . . . 49

8 Lecture 8 Jan 19 2018 51

8.1 Power Series (Continued) . . . . . . . . . . . . . . . . . . 51
8.1.1 Radius of Convergence . . . . . . . . . . . . . . . 51
4 ë TABLE OF CONTENTS - ë TABLE OF CONTENTS

9 Lecture 9 Jan 22nd 2018 55

9.1 Power Series (Continued 2) . . . . . . . . . . . . . . . . . 55
9.1.1 Radius of Convergence (Continued) . . . . . . . . 55

10 Lecture 10 Jan 24th 2018 59

10.1 Power Series (Continued 3) . . . . . . . . . . . . . . . . . 59
10.1.1 Radius of Convergence (Continued 2) . . . . . . . 59
10.2 Integration in C . . . . . . . . . . . . . . . . . . . . . . . . 60
10.2.1 Curves and Paths . . . . . . . . . . . . . . . . . . . 60
10.2.2 Integral . . . . . . . . . . . . . . . . . . . . . . . . . 62

11 Lecture 11 Jan 26th 2018 65

11.1 Integration in C (Continued) . . . . . . . . . . . . . . . . 65
11.1.1 Integral (Continued) . . . . . . . . . . . . . . . . . 65

12 Lecture 12 Jan 29th 2018 71

12.1 Integration in C (Continued 2) . . . . . . . . . . . . . . . 71
12.1.1 Fundamental Theorem of Calculus . . . . . . . . 71

Tutorial 75
12.2 Practice Problems . . . . . . . . . . . . . . . . . . . . . . . 75

13 Lecture 13 Feb 9th 2018 81

13.1 Cauchy’s Integral Formula . . . . . . . . . . . . . . . . . . 81

14 Lecture 14 Feb 12 2018 85

14.1 Cauchy’s Integral Formula (Continued) . . . . . . . . . . 85

15 Lecture 15 Feb 14th 2018 91

15.1 Cauchy’s Integral Formula (Continued 1) . . . . . . . . . 91
15.1.1 Applications of Cauchy’s Integral Formula . . . . 91

16 Lecture 16 Feb 16th 2018 95

16.1 Cauchy’s Integral Formula (Continued 3) . . . . . . . . . 95
16.1.1 Applications of Cauchy’s Integral Formula (Con-
tinued) . . . . . . . . . . . . . . . . . . . . . . . . . 95

17 Lecture 17 Feb 26th 2018 99

17.1 Analytic Continuity . . . . . . . . . . . . . . . . . . . . . . 99
17.2 Morera’s Theorem . . . . . . . . . . . . . . . . . . . . . . 101

18 Lecture 18 Feb 28th 2018 103

18.1 Winding Numbers . . . . . . . . . . . . . . . . . . . . . . 103
 PMATH352W18 - Complex Analysis 5

19 Lecture 19 Mar 2nd 2018 107

19.1 Singularities . . . . . . . . . . . . . . . . . . . . . . . . . . 107

20 Lecture 20 Mar 5th 2018 111

20.1 Singularity (Continued) . . . . . . . . . . . . . . . . . . . 111

21 Lecture 21 Mar 7th 2018 115

21.1 Singularity (Continued 2) . . . . . . . . . . . . . . . . . . 115

22 Lecture 22 Mar 9th 2018 119

22.1 Singularity (Continued 3) . . . . . . . . . . . . . . . . . . 119
22.2 The Residue Theorem . . . . . . . . . . . . . . . . . . . . 119

23 Lecture 23 Mar 12th 2018 123

23.1 The Residue Theorem (Continued) . . . . . . . . . . . . . 123
23.2 Applications of Cauchy’s Residue Theorem . . . . . . . . 126

24 Lecture 24 Mar 14 2018 129

24.1 Application of Cauchy’s Residue Theorem (Continued) . 129

25 Lecture 25 Mar 16 2018 135

25.1 The Argument Principle . . . . . . . . . . . . . . . . . . . 135

26 Lecture 26 Mar 19 2018 139

26.1 The Argument Principle (Continued) . . . . . . . . . . . 139
26.1.1 Alternative Proof for FTA . . . . . . . . . . . . . . 140
26.1.2 Open Mapping Theorem . . . . . . . . . . . . . . 141

27 Lecture 27 Mar 21 2018 143

27.1 Introductory Passage to Log Functions in C . . . . . . . 143
27.2 Simply Connected Domains . . . . . . . . . . . . . . . . . 144

28 Lecture 28 Mar 23 2018 147

28.1 Constructing Logarithm . . . . . . . . . . . . . . . . . . . 147
28.2 Branches of the Logarithm . . . . . . . . . . . . . . . . . . 149

29 Lecture 29 Mar 26 2018 151

29.1 Examples for Analytic Continuation . . . . . . . . . . . . 151
29.2 Characterizing Logarithms . . . . . . . . . . . . . . . . . 153

30 Lecture 30 Mar 28 2018 155

30.1 Characterizing Logarithms . . . . . . . . . . . . . . . . . 155
30.2 Infinite Products . . . . . . . . . . . . . . . . . . . . . . . . 156
6 ë TABLE OF CONTENTS - ë TABLE OF CONTENTS

31 Lecture 31 Apr 02 2018 161

31.1 Infinite Products (Continued) . . . . . . . . . . . . . . . . 161
31.1.1 Application to Riemann Zeta Function . . . . . . 162

32 Lecture 32 Apr 04 2018 165

32.1 Infinite Products (Continued 2) . . . . . . . . . . . . . . . 165
32.1.1 Weierstrass Products . . . . . . . . . . . . . . . . . 165

33 Index 171
 List of Definitions

1 Complex Number, Complex Plane . . . . . . . . . . . 11

2 Sum and Product . . . . . . . . . . . . . . . . . . . . . 12

3 Conjugate . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4 Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5 Argument of a Complex Number . . . . . . . . . . . . 23

6 Convergence . . . . . . . . . . . . . . . . . . . . . . . . 37

7 Convergence for Complex Functions . . . . . . . . . . 38

8 Continuity . . . . . . . . . . . . . . . . . . . . . . . . . 39

9 Neighbourhood . . . . . . . . . . . . . . . . . . . . . . 42

10 Differentiable/Holomorphic . . . . . . . . . . . . . . . 42

11 Power Series . . . . . . . . . . . . . . . . . . . . . . . . 49

12 Entire Function . . . . . . . . . . . . . . . . . . . . . . . 56

13 Curves in C . . . . . . . . . . . . . . . . . . . . . . . . . 61

14 Equivalent Parameterization . . . . . . . . . . . . . . . 61

15 Smooth Curve . . . . . . . . . . . . . . . . . . . . . . . 61

16 Piecewise Smooth . . . . . . . . . . . . . . . . . . . . . 62

17 Contour . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

18 Closed Path . . . . . . . . . . . . . . . . . . . . . . . . . 72

19 Convex Set . . . . . . . . . . . . . . . . . . . . . . . . . 81

20 Analytic Functions . . . . . . . . . . . . . . . . . . . . . 91

21 Winding Numbers . . . . . . . . . . . . . . . . . . . . . 103

8 ë TABLE OF CONTENTS - ë TABLE OF CONTENTS

22 (Isolated) Singularity . . . . . . . . . . . . . . . . . . . 107

23 Removable Singularity, Pole, Essential Singularity . . 108

24 Zero of Order n & Simple Zero . . . . . . . . . . . . . 112

25 Pole of order n & Simple Pole . . . . . . . . . . . . . . 113

26 Principal Part . . . . . . . . . . . . . . . . . . . . . . . . 114

27 Residue . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

28 Meromorphic Functions . . . . . . . . . . . . . . . . . 121

29 Monic Polynomial . . . . . . . . . . . . . . . . . . . . . 140

30 Monomial . . . . . . . . . . . . . . . . . . . . . . . . . . 140

31 Homotopy (Poincaré) . . . . . . . . . . . . . . . . . . . 144

32 Simply Connected Domain . . . . . . . . . . . . . . . . 145

33 Infinite Products . . . . . . . . . . . . . . . . . . . . . . 156

1 List of Theorems

0 Proposition 1 Basic Inequalities . . . . . . . . . . . . . . . . 16

0 Proposition 2 nth Roots of a Complex Number . . . . . . 27

1 Theorem 3 Cauchy-Riemann Equations . . . . . . . . . 45

1 Theorem 4 Conditional Converse of CRE . . . . . . . . 48

1 Theorem 5 Convergence in the Radius of Convergence 51

0 Proposition 6 A Property of limsup . . . . . . . . . . . . . 52

1 Theorem 7 Power function, holomorphic function, region

of convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

 Corollary 8 Corollary of 1 Theorem 7 . . . . . . . . . . 59

0 Proposition 9 Properties of integrals in C . . . . . . . . . . 66

1 Theorem 10 Fundamental Theorem of Calculus . . . . . 71

 Corollary 11 Corollary of FTC . . . . . . . . . . . . . . . . 72

1 Theorem 12 Goursat’s Theorem / Cauchy’s Theorem for

a triangle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

1 Theorem 13 Cauchy’s Theorem for Convex Set . . . . . . 81

1 Theorem 14 Cauchy’s Integral Formula 1 . . . . . . . . . 82

+ Lemma 15 . . . . . . . . . . . . . . . . . . . . . . . . . . 85

0 Proposition 16 Holomorphic Functions can be expressed as

Power series . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

1 Theorem 17 Cauchy’s Integral Formula 2 . . . . . . . . . 88

 Corollary 18 Taylor Expansion of Entire Functions . . . . 89

+ Lemma 19 Principle of Analytic Continuation . . . . . 94

10 ë TABLE OF CONTENTS - ë TABLE OF CONTENTS

 Corollary 21 Uniqueness of a Function . . . . . . . . . . . 100

1 Theorem 22 Morera’s Theorem . . . . . . . . . . . . . . . 101

1 Theorem 23 Winding Number Theorem . . . . . . . . . . 103

1 Theorem 24 Theorem 9 . . . . . . . . . . . . . . . . . . . . 109

1 Theorem 25 Theorem 10 . . . . . . . . . . . . . . . . . . . 111

1 Theorem 26 Theorem 9.1 . . . . . . . . . . . . . . . . . . . 113

1 Theorem 27 Theorem 11 . . . . . . . . . . . . . . . . . . . 113

1 Theorem 28 Casorati-Weierstrass . . . . . . . . . . . . . . 115

 Corollary 29 . . . . . . . . . . . . . . . . . . . . . . . . . . 119

1 Theorem 30 Cauchy’s Residue Theorem . . . . . . . . . . 120

1 Theorem 31 Cauchy’s Residue Theorem - Generalized . 123

1 Theorem 32 Argument Principle . . . . . . . . . . . . . . 136

1 Theorem 33 Rouché’s Theorem . . . . . . . . . . . . . . . 137

 Corollary 34 . . . . . . . . . . . . . . . . . . . . . . . . . . 141

1 Theorem 35 Open Mapping Theorem . . . . . . . . . . . 141

1 Theorem 36 Theorem 17 . . . . . . . . . . . . . . . . . . . 147

1 Theorem 38 Theorem 18 . . . . . . . . . . . . . . . . . . . 155

+ Lemma 39 Bounds of the Partial Product . . . . . . . . 157

1 Theorem 40 Theorem 19 . . . . . . . . . . . . . . . . . . . 158

 Corollary 41 Corollary for Theorem 19 . . . . . . . . . . . 163

1 Theorem 42 Theorem 20 . . . . . . . . . . . . . . . . . . . 166

1 Theorem 43 Theorem 21 . . . . . . . . . . . . . . . . . . . 167

1 Lecture 1 Jan 3rd 2018

1.1 Complex Numbers and Their Properties

 Definition 1 (Complex Number, Complex Plane)

A complex number is a vector in R2 . The complex plane, denoted by

C, is a set of complex numbers,
( ! )
2 x
C=R = : x, y ∈ R
y

In C, we usually write
! !
0 1
0= 1=
0 0
! !
0 x
i= x=
1 0
!
0
iy =
y

where x, y ∈ R. Consequently, we have that

!
x
x + iy = x + yi =
y

If for x, y ∈ R, z = x + iy, then x is called the real part of z and y is

called the imaginary part of z, and we write

Re(z) = x Im(z) = y.
12 Lecture 1 Jan 3rd 2018 - Complex Numbers and Their Properties

å Note
• It is easy to see how R is a subset of C.

0
where y ∈ R are called purely


• Complex Numbers of the form y
imaginary numbers.

0


• Certain authors may prefer to denote i = 1 .

 Definition 2 (Sum and Product)

We define the sum of two complex numbers to be the usual vector sum,
i.e.
! !
a c
( a + ib) + (c + id) = +
b d
!
a+c
=
b+d
= ( a + c) + i (b + d)

where a, b, c, d ∈ R.

We define the product of two complex numbers by setting i2 = −1,

and by requiring the product to be commutative, associative, and
distributive over the sum. In this setup, we have that

( a + ib)(c + id) = ac + iad + ibc + i2 bd

= ( ac − bd) + i ( ad + bc) (1.1)

å Note
It is interesting to note that any complex number times zero is zero,
just like what we have with real numbers.

∀z = x + iy ∈ C x, y ∈ R 0 ∈ C
z · 0 = ( x + iy)(0 + i0) = 0 + i0 = 0

Example 1.1.1
 PMATH352W18 - Complex Analysis 13

Let z = 2 + i, w = 1 + 3i. Find z + w and zw.

z + w = (2 + i ) + (1 + 3i )
= 3 + 4i

zw = (2 + i )(1 + 3i )
= (2 − 3) + i (6 + 1) By Equation (1.1)
= −1 + 7i

Example 1.1.2

Show that every non-zero complex number has a multiplicative inverse,

z−1 , and find a formula for this inverse.

Let z = a + ib where a, b ∈ R with a2 + b2 6= 0. Then

z( x + iy) = 1
⇐⇒ ( ax − by) + i ( ay + bx ) = 1
! !
ax − by 1
⇐⇒ =
ay + bx 0
! ! !
a −b x 1
⇐⇒ =
b a y 0
! ! !
x 1 a b 1
⇐⇒ = 2
y a + b2 − b a 0
! !
x 1 a
⇐⇒ = 2 2
y a +b −b
a b
⇐⇒ x + iy = −i 2
a2 + b2 a + b2
Therefore, we have that the formula for the inverse is

a b
( a + ib)−1 = −i 2 (1.2)
a2 + b2 a + b2

 Notation
For z, w ∈ C, we write

−z = −1z w − z = w + (−z)
1 −1 w
= wz−1
z =z z
14 Lecture 1 Jan 3rd 2018 - Complex Numbers and Their Properties

Example 1.1.3

(4−i )−(1−2i )
Find 1+2i .

(4 − i ) − (1 − 2i ) 3+i
=
1 + 2i 1 + 2i
1 2
= (3 + i )( − i )
5 5
= 1−i

å Note
The set of complex numbers is a field under the operations of addition and
multiplication. This means that ∀u, v, w ∈ C,

u+v=v+u uv = vu
(u + v) + w = u + (v + w) (uv)w = u(vw)
0+u=u 1u = u
u + (-u) = 0 uu−1 = 1, u 6= 0
u(v + w) = uv + uw

Since the distributive law holds for complex numbers, note that the
binomial expansion works for (w + z)n where w, z ∈ C and n ∈ N. (I
did not verify if this is still true for when n ∈ R.)

 Definition 3 (Conjugate)

If z = x + iy where x, y ∈ R, then the conjugate of z is given by

z = x − iy

Example 1.1.4

Let z = 3 + 4i. Then the z = 3 − 4i. Represented in the complex plane, we

have the following:
 PMATH352W18 - Complex Analysis 15

5
Im
4 z
3
2
1
Re
−4 −3 −2 −1 1 2 3 4
−1
−2
−3
−4 z
−5

We observe that on the complex plane, the conjugate of a complex number

is simply its reflection on the real axis.

 Definition 4 (Modulus)

We define the modulus (length, magnitude) of z = x + iy ∈ C, x, y ∈ R,

to be q
|z| = x2 + y2 ∈ R. (1.3)

å Note
Note that this definition is consistent with the notion of the absolute value
in real numbers when z is a real number, since if y = 0, |z| = | x + i0| =
√
x2 = ± x.

å Note
For z, w ∈ C and n ∈ N, we have

z=z z + z = 2 Re(z) z − z = 2i Im(z)

2
zz = |z| |z| = |z| z±w = z±w
zw = zw |zw| = |z| |w| zn = zn

but note that |z + w| 6= |z| + |w|.

Also, note that the last equation is a generalization of the highlighted

equation.
16 Lecture 1 Jan 3rd 2018 - Complex Numbers and Their Properties

å Note
While inequalities such as z1 < z2 , where z1 , z2 ∈ C, are meaningless
unless if both of them are real, |z1 | < |z2 | means that the point z1 in the
complex plane is closer to the origin than the point z2 .

0 Proposition 1 (Basic Inequalities)

1. |Re(z)| ≤ |z|

2. |Im(z)| ≤ |z|

3. |z + w| ≤ |z| + |w| Triangle Inequality

4. |z + w| ≥ | |z| − |w| | Inverse Triangle Inequality

Ò Proof
√
Note that |z|2 = Re(z)2 + Im(z)2 and that we can express | x | = x2 for
any x ∈ R. 1 and 2 immediately follows from that.

To prove 3, we have that

|z + w|2 = (z + w)(z + w)
= |z|2 + |w|2 + (wz + wz)
= |z|2 + |w|2 + 2 Re(wz)
≤ |z|2 + |w|2 + 2 |wz| by 1
= |z|2 + |w|2 + 2 |wz| since |wz| = |w| |z| and |z| = |z|
= (|z| + |w|)2

To prove 4, note that

|z| = |z + w − w| ≤ |z + w| + |w| (1.4)

|w| = |w + z − z| ≤ |z + w| + |z| (1.5)
 PMATH352W18 - Complex Analysis 17

Observe that

Equation (1.4) =⇒ |z| − |w| ≤ |z + w|

Equation (1.5) =⇒ |w| − |z| ≤ |z + w|

Thus, we have that

|z + w| ≥ | |z| − |w| |

as required. 

Item 3 in 0 Proposition 1 can be generalized by the means of

mathematical induction to sums involving any finite number of
terms, as:

| z1 + z2 + . . . + z n | ≤ | z1 | + | z2 | + . . . + | z n | (1.6)

where n ∈ N \ {0, 1}.

To note the induction proof, when n = 2, Equation (1.6) is just

Item 3. If Equation (1.6) is true for when n = m where m ∈ N \ {0, 1},
n = m + 1 is also true since by Item 3,

|(z1 + z2 + . . . + zm ) + zm+1 | ≤ |z1 + z2 + . . . + zm | + |zm+1 |

≤ (|z1 | + |z2 | + . . . + |zm |) + |zm+1 | .

The distance between two points z1 = x1 + iy1 , z2 = x2 + iy2 ∈

C, x1 , x2 , y1 , y2 ∈ R is |z1 − z2 |, since |z1 − z2 | = ( x1 − x2 )2 (y1 − y2 )2
p

is our usual notion of the Euclidean distance of two points on a

plane.

Also, note that

z1 − z2 = z1 + (−z2 )

and thus if we apply our knowledge of vector representation, z1 − z2

is the directed line segment from the point z2 to z1 .

With the notion of a “distance” set on the complex plane, we can

now explore upon points lying on a circle with a center z0 and radius
R, which satisfies the equation

|z − z0 | = R.
18 Lecture 1 Jan 3rd 2018 - Complex Numbers and Their Properties

We may simply refer to this set of points as the circle |z − z0 | = R.

Example 1.1.5

We may describe a set {z ∈ C : |z − i | = 1} as follows:

Im
3

Re
−3 −2 −1 1 2 3

−1

Let a, b ∈ C describe the set {z ∈ C : |z − a| < |z − b|}.

Suppose the following coordinates for a and b are arbitrary,

f Im
10
b
g
a Re
−15 −10 −5 5 10 15

−10

In the above, g is the line segment that connects the points a and b on the
complex plane, while f is the perpendicular bisector of the line segment g.
The area described by the set {z ∈ C : |z − a| < |z − b|} is the shaded area
which is below f .
2 Lecture 2 Jan 5th 2018

2.1 Complex Numbers and Their Properties (Continued)

Example 2.1.1

Let a ∈ C. Describe the set {z ∈ C : 1 < |z − a| < 2}.

Im
6

4
a
2

Re
−2 2 4 6

−2

Example 2.1.2

Show that every non-zero complex number has exactly two complex square
roots, and find a formula for the square roots.

Let z = x + iy ∈ C, x, y ∈ R, and let w = u + iv, u, v ∈ R. Then

w2 = z ⇐⇒ (u + iv)2 = x + iy
⇐⇒ (u2 − v2 ) + i (2uv) = x + iy
⇐⇒ x = u2 + v2 and (2.1)
y = 2uv (2.2)
20 Lecture 2 Jan 5th 2018 - Complex Numbers and Their Properties (Continued)

Square both sides of Equation (2.2), and thus we have y2 = 4u2 v2 .

Multiply Equation (2.1) by 4u2 , and we get

4u2 x = 4u4 − 4u2 v2 = 4u4 − y2

⇐⇒ 0 = 4u4 − 4u2 x − y2
p
4x ± 16x2 + 16y2
⇐⇒ u2 =
p 8
x ± x 2 + y2
=
2

√
p x+ x 2 + y2
Suppose y 6= 0. Note that x < x2 + y2 . Thus u2 = 2 =⇒
 √ 1
x + x 2 + y2 2
u= 2 .

Similarly, we can get

!1
2
p
−x + x 2 + y2
v=±
2

Note that all four choices of signs satisfy Equation (2.1). If y > 0, then u
and v are either both positive or both negative by Equation (2.2).

Suppose y = 0. Then we have

w2 = z = x

Therefore, we get
 " √ 1  √ 2 2  12 #
2 + y2 2
 x + x − x + x +y



 ± 2 +i 2 y>0


 " √ 2 2  12 √ 1 #

 
 x + x +y − x + x 2 + y2 2
w= ± 2 −i 2 y<0

√





 ± x y = 0, x > 0
 √



±i x y = 0, x < 0

Remark
√
Let z ∈ C. The notation z may represent either one of the square roots of z
√
or both of the square roots, i.e. it is possible that z represents a set.

Exercise 2.1.1
√ √ √
Is it always okay for complex numbers such that zw = z w, for
 PMATH352W18 - Complex Analysis 21

z, w ∈ C?

No. For example, consider z = w = −1. Then we have

√ √
zw = 1 = ±1

while
√ √
z w = i · i = −1

and thus
√ √ √
zw 6= z w.

Example 2.1.3

√
Find the values of 3 − 4i.

By Example 2.1.2,

s
√ s √ 
√ 3+ 9 + 16 −3 + 9 + 16 
3 − 4i = ±  −i
2 2

= ±(2 − i )

Remark
The quadratic formula holds for complex polynomials, i.e.

∀ a, b, c ∈ C a 6= 0 ∀z ∈ C az2 + bz + c = 0,

the solution for z is given by

√
−b + b2 − 4ac
z1,2 = (2.3)
b

The following is a short proof.

22 Lecture 2 Jan 5th 2018 - Complex Numbers and Their Properties (Continued)

Ò Proof

b c
az2 + bz + c = 0 ⇐⇒ z2 + z + = 0
a a
 2  2
b b b c
⇐⇒ z2 + z + − + =0
a 2a 2a a
2 2 2
b − 4ac

b b c
⇐⇒ z + = 2− =
2a 4a a 4a2
√
−b + b2 − 4ac
⇐⇒ z =
2a

(Personal Note: where did the − for the supposed ± go? Or should it
really be ±?)

Example 2.1.4

Solve iz2 − (2 + 3i )z + 5(1 + i ) = 0.

p
2 + 3i + (2 + 3i )2 − 4i [5(1 + i )]
z=
√ 2i
2 + 3i + −5 + 12i − 20i + 20
=
√ 2i
2 + 3i + 15 + 8i
=
2i
Note that by Example 2.1.2,
s
√ s √ 
√ 15 +
225 + 64 −15 + 225 + 64 
15 − 8i = ±  −i
2 2
"r r #
15 + 17 −15 + 17
=± −i
2 2

= ±(4 − i )

Thus we have
√
2 + 3i + 15 + 8i
z=
2i
2 + 3i ± (4 − i )
=
2i   
1 1
= (6 + 2i ) − i or (−2 + 4i ) − i by Example 1.1.2
2 2
= (1 − 3i ) or (2 + i )
3 Lecture 3 Jan 8th 2018

3.1 Complex Numbers and Their Properties (Continued 2)

 Definition 5 (Argument of a Complex Number)

Let z ∈ C \ {0}. The argument (or the angle) of z, denoted by arg z,

Arg z, or simply θ = θ (z), is the angle modulo 2π (i.e. 0 ≤ θ < 2π)
between the vector defining z and the positive real axis (in the counter-
clockwise direction).

Im
z1
θ1 z

θ
Re

 Notation
Let eiθ := cos θ + i sin θ. Note that this definition, called Euler’s for-
mula, can be derived by the extending the Taylor expansion of e x =
∑∞
n
n=0 n! for when x ∈ C (the sum of the real parts of the expansion is the
x

Taylor expansion of cosine while the imaginary part for sine).

Now eiθ is on the unit circle.

24 Lecture 3 Jan 8th 2018 - Complex Numbers and Their Properties (Continued 2)

Im
etθ
sin θ

cos θ Re

Remark
If z = 0, the coordinate θ is undefined, and so it is implied that z 6= 0
whenever we use the polar form.

Example 3.1.1

Some examples of θ ∈ [0, 2π ):

π
√ √ π
2 2
ei 4 = 2 √+ i 2 √ ei 2 = i
3π
ei 4 = − 22 + i 22 eiπ + 1 = 0

Remark

∀k ∈ Z ∀θ ∈ R eiθ = ei(θ +2πk)

Remark
The complex number reiθ , where r > 0, θ ∈ [0, 2π ), represents the complex
number with modulus r and argument θ.

Im
z

θ
r
Re

Therefore, ∀z ∈ C, we can express

z := |z| ei Arg z . (3.1)

With that, we now have two representations of a complex number:

 PMATH352W18 - Complex Analysis 25

• Cartesian representation: z = x + iy where x = Re(z) and

y = Im(z)

• Polar representation: z = reiθ where r = |z| and θ = Arg z ∈

[0, 2π )

To convert between the two representations, we have the following

equations:

Polar → Cartesian:

x = r cos θ y = r sin θ (3.2)

Cartesian → Polar:

r = |z|
y
x 6= 0 =⇒ tan θ = (3.3)
x
π 3π
x = 0 =⇒ θ = or
2 2

On another note,

z = reiθ =⇒ z = re−iθ

and
1 1
z 6= 0 =⇒ = e−iθ (3.4)
z r
Remark

∀r1 , r2 ∈ R ∀θ1 , θ2 ∈ [0, 2π )

z1 := r1 eiθ1 z2 := r2 eiθ2

Then
z1 z2 = r1 r2 eiθ1 eiθ2 = r1 r2 ei(θ1 +θ2 )
26 Lecture 3 Jan 8th 2018 - Complex Numbers and Their Properties (Continued 2)

Note that eix eiy = ei( x+y) is true for all x, y ∈ R since

eix eiy = (cos x + i sin x )(cos y + i sin y)

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

Generalizing the above, we get that

∀n ∈ Z z = (rein ) = r n einθ (3.5)

which is commonly known as deMoivre’s Law. Note that by simply gener-

alizing the above, all we have is that n ∈ Z+ . But by Equation (3.4), we can
have that for n ∈ Z− , let m = −n, and thus
 m  m  −n
1 i(−θ ) 1 1
zn = e = eim(−θ ) = ei(−n)(−θ ) = r n eiθ
r r r

This proves that deMoivre’s Law also holds for when n ∈ Z− .

Observe that if r = 1, Equation (3.5) becomes

(eiθ )n = einθ for all n ∈ Z \ {0} (3.6)

When written in the form

(cos θ + i sin θ )n = cos nθ + i sin nθ (n ∈ Z \ {0}) (3.7)

this is known as deMoivre’s formula.

Example 3.1.2

Equation (3.7) with n = 2 tells us that

(cos θ + i sin θ )n = cos 2θ + i sin 2θ

or we can express the equation as

cos2 θ − sin2 θ + i2 sin θ cos θ = cos 2θ + i sin 2θ

Equating real and imaginary parts, we have the familiar double angle
trigonometric identities

cos 2θ = cos2 θ − sin2 θ, sin 2θ = 2 sin θ cos θ.

 PMATH352W18 - Complex Analysis 27

3.1.1 Roots of Complex Numbers

0 Proposition 2 (nth Roots of a Complex Number)

∀z = reiθ ∈ C r = |z| ∈ R θ ∈ [0, 2π )

∃w = seiτ ∈ C s ∈ R τ ∈ [0, 2π )
∀n ∈ Z
n

wn = seiτ = z = reiθ

The nth roots of z is described by the set

θ +2πk
n 1
o
r n ei ( n ) : k = 0, 1, ..., n − 1 (3.8)

Ò Proof

1
sn = r ⇐⇒ s = r n
τ + 2πk
einθ = eiτ ⇐⇒ θ =
n

Therefore, the set that describes the nth roots of z is

θ +2πk
n 1
o
w = r n ei ( n ) : k = 0, 1, ..., n − 1

Remark (nth Roots of Unity)

The nth roots of unity is a direct consequence of 0 Proposition 2 where we
solve for the equation zn = 1 for any z ∈ C, n ∈ Z.

The set that describes the nth roots of unity is

 
iθ 2πk
e :θ= , k = 0, 1, ..., n − 1 (3.9)
n

It is easy to see how the nth roots of unity partitions the unit
circle into n parts.
28 Lecture 3 Jan 8th 2018 - Complex Numbers and Their Properties (Continued 2)

Example 3.1.3

Find the cubic roots of −2 + 2i.

√ 3π
Let z = −2 + 2i. Note that |z| = 2 2 and Arg z = 4 .

√ 3π
Therefore, in polar form, z = 2 2ei 4 .

Let w = reiθ , where θ ∈ [0, 2π ), and w3 = z. Then

√ 1
r = (2 2) 3
3π
4 + 2πk
θ= , k = 0, 1, 2
3

The set that describes the cubic root of −2 + 2i is thus

( )
√ 1
3π
+ 2πk
(2 2) eiθ : θ =
3 4
, k = 0, 1, 2
3

Example 3.1.4

Describe the set {z ∈ C : Arg z − π

2 < 2 }.
π
(Note: Arg z ∈ [0, 2π ))

2 Im

Re
−2 −1 1 2

−1

Exercise 3.1.1

Solve

1. z4 = −1

Let z = reiθ
π + 2πk (2k + 1)π
r = |−1| = 1 θ= = , k = 0, 1, 2, 3
4 4
 PMATH352W18 - Complex Analysis 29

√
2. z4 = −1 + 3i

Let z = reiθ
√ q √
r = −1 + 3i = (−1)2 + 32 = 10
2π
3 + 2πk (2k + 23 )π
θ= = , k = 0, 1, 2, 3
4 4
4 Lecture 4 Jan 10th 2018

4.1 Examples for nth Roots of Unity

2πk
Recall that the nth roots of unity are given by ei n , k = 0, 1, ..., n − 1.

Exercise 4.1.1

Let z be any nth root of unity other than 1. Show that

z n −1 + z n −2 + . . . + z + 1 = 0 (4.1)

Ò Proof
By the Sum of Finite Geometric Terms,

1 − zn
z n −1 + z n −2 + . . . + z + 1 = .
1−z
Since zn = 1, RHS is thus zero, which in turn completes the proof.

As an aside, if we wish to remove the restriction that z can also be

1, we may consider that

zn − 1 = (z − 1)(1 + z + . . . + zn−1 )

Since zn = 1, LHS is zero. Then either z = 1 or (1 + z + . . . + zn−1 ) =

0.

Exercise 4.1.2

Consider the n − 1 diagonals of a regular n-gon, inscribed in a circle of

radius 1, obtained by connecting one vertex on the n-gon to all its other
32 Lecture 4 Jan 10th 2018 - Examples for nth Roots of Unity

vertices.

For example, if we are given n = 6, we obtain the following diagram.

Figure 4.1: n = 6, where a is an arbi-

trary vertex on the hexagon
z1 z2

z3
a
z5 z4

Show that the product of the lengths of these diagonals is equal to n.

Ò Proof
Note that Figure 4.1 can be translated into Figure 4.2.

Figure 4.2: A regular n-gon with the

roots of unity on its vertices
z4 z3

z2

a = z1

Thus the equation that we wish to prove becomes

|1 − z2 | |1 − z3 | . . . |1 − z n | = n (4.2)

Note that z2 , ..., zn are the nth roots of unity other than 1.

Let z be a variable and consider the polynomial

P ( z ) : = 1 + z + z 2 + . . . + z n −1 (4.3)

Since the roots of P(z) are the nth roots of unity other than 1, we can
 PMATH352W18 - Complex Analysis 33

factorize Equation (4.3) into

P(z) = (z − z2 )(z − z3 ) . . . (z − zn )

Now let z = 1 and take the modulus of P(z), and we get Equation (4.2).

Exercise 4.1.3

23n +2(−1)n
Let n ∈ N. Show that ∑nj=0 (3n
3j ) = 3 .

Ò Proof
2π
Let α = ei 3 . Then α is a cubic root of unity, i.e. α3 = 1, and from
Exercise 4.1.1, 1 + α + α2 = 0.

Consider
         
3n 3n 3n 3n 3n 3n
(1 + 1) = + + + +
0 1 2 3 4
      (4.4)
3n 3n 3n
+ + +...+
5 6 3n
         
3n 3n 3n 2 3n 3n
(1 + α)3n = + α+ α + + α
0 1 2 3 4
      (4.5)
3n 2 3n 3n
+ α + +...+
5 6 3n
         
3n 3n 2 3n 3n 3n 2
(1 + α2 )3n = + α + α+ + α
0 1 2 3 4
      (4.6)
3n 3n 3n
+ α+ +...+
5 6 3n

Adding Equation (4.4), Equation (4.5) and Equation (4.6), we observe

that the terms with coefficients (3n
k ) where k is not a multiple of 3 sums to
34 Lecture 4 Jan 10th 2018 - Examples for nth Roots of Unity

0 as given by 1 + α + α2 = 0, and therefore we obtain

n 

3n
2 3n
+ (1 + α ) 3n
+ (1 + α )2 3n
=3∑
j =0
3j
n  
1 h 3n i 3n
2 + (1 + α)3n + (1 + α2 )3n =∑
3 j =0
3j
n  
1 h 3n i 3n
2 + (−α2 )3n + (−α)3n =∑ since 1 + α + α2 = 0
3 j =0
3j
n  
1 h 3n i 3n
2 + (−1)n + (−1)n =∑ since α3 = 1
3 j =0
3j
n  
23n + 2(−1)n 3n
=∑
3 j =0
3j

as required.

Exercise 4.1.4

Note that we can define Arg z in any interval of length 2π, i.e. it is not
necessary that Arg z ∈ [0, 2π ).

For example, if we restrict Arg z ∈ [−π, π ], then we can write

 
1 1 3π
Arg − √ − √ i =−
2 2 4

/ R, i.e.
Let z be on the unit circle and Arg z ∈ [−π, π ]. Suppose that z ∈
z 6= 1, z 6= −1. Show that


z−1
 π Im z > 0
2
Arg =
z+1 − π Im z < 0
2

Ò Proof
Note that ∀w1 , w2 ∈ C, where Arg w1 = τ1 , Arg w2 = τ2 for τ1 , τ2 in
the same 2π-interval,

w1 eiτ1
Arg = iτ ≡ ei(τ1 −τ2 ) = Arg w1 − Arg w2 (4.7)
w2 e 2

in modulo 2π.

Suppose Im z > 0. Let θ1 = Arg(z − 1) and θ2 = Arg(z + 1).

Consider Figure 4.3. Note that since both θ1 , θ2 ∈ [0, π ], we have that
 PMATH352W18 - Complex Analysis 35

θ1 − θ2 ∈ [−π, π ], and thus Equation (4.7) holds true without the need
of the condition of being in modulo 2π. We observe that
π
= θ2 + π − θ1
2
π
θ1 − θ2 =
2
as desired.

Figure 4.3: (Right) Depicted question,

(Left) Translated Angles
z−1 z z+1

θ1 θ2

z−1 z z+1

θ1
θ2

Similarly, we can obtain θ1 − θ2 = − π2 for when Im z < 0. This

completes the proof.

Exercise 4.1.5

Let f (z) = ez for z ∈ C. Let A = {z = x + iy ∈ C : x ≤ 1, y ∈ [0, π ]}.

Describe the image of f ( A).

Ò Solution
Firstly, note that

ez = e x+iy
e x ∈ (0, e]
y ∈ [0, π ]

It is clear that the image will be in on the positive side of the imaginary-
axis. Also, since e x ∈ (0, e], we get the right graph represented in Fig-
36 Lecture 4 Jan 10th 2018 - Examples for nth Roots of Unity

Im Figure 4.4: (Right) Domain of f ( A),

π (Left) Image of f ( A)

Re
1

Im
e

Re
−e e

ure 4.4. The image of f ( A) is described in the left image of Figure 4.4.
 5 Lecture 5 Jan 12 2018

5.1 Complex Functions

5.1.1 Limits

 Definition 6 (Convergence)

A sequence of complex numbers z1 , z2 , z3 , ... converges to z ∈ C if

lim |zn − z| = 0 (5.1)

n→∞

or we may say

∀ε > 0 ∃ N ∈ N ∀n > N |zn − z| < ε (5.2)

å Note
If {zn }n∈N converges to z, we may write limn→∞ zn = z or zn → z (as
n → ∞).

Example 5.1.1

For |z| > 1, does { z1n }∞

n=1 converge? Explain.
38 Lecture 5 Jan 12 2018 - Complex Functions

Ò Solution
We claim that the limit is 0. Since |z| > 1, we have that

n
1 1
lim − 0 = lim
n→∞ zn n→∞ z

=0

1
Another way to prove this, since |z| > 1 =⇒ 0 < z < 1,

1
∀ε = >0
z
n
1 1 1
−0 = < =ε
zn z z

 Definition 7 (Convergence for Complex Functions)

∀Ω ⊆ C, let f : Ω → C. We say that

lim f (z) = L (5.3)

z → z0

for some L ∈ C if for every sequence {zn }n ⊆ Ω (not including z0 if it is

in Ω), we have that

zn → z0 =⇒ f (zn ) → L (5.4)

Note that L need not be in Ω.

Example 5.1.2

Let f (z) = zz , z ∈ C \ {0}. Find limz→0 f (z).

Ò Solution
Suppose z = x ∈ R \ {0}. Then f (z) = f ( x ) = x
x = 1.
−iy
Suppose z = iy, y ∈ R \ {0}. Then f (z) = f (iy) = iy = −1.

Therefore, the limit limz→0 f (z) does not exist.

Exercise 5.1.1

Show that zn → z ⇐⇒ Re(zn ) → Re(z) ∧ Im(zn ) → Im(z).

(Hint: |Re(z)| , |Im(z)| ≤ |z| ≤ |Re(z)| + |Im(z)|)
 PMATH352W18 - Complex Analysis 39

Ò Solution
Suppose zn → z. Then ∀ε 0 > 0 ∃ N ∈ N ∀n > N |zn − z| < ε. Note once
and for all that

Re(zn − z) = Re(zn ) − Re(z)

Im(zn − z) = Im(zn ) − Im(z).

Thus

|Re(zn ) − Re(z)| = |Re(zn − z)|

≤ |zn − z| < ε
|Im(zn ) − Im(z)| = |Im(zn − z)|
≤ |zn − z| < ε

For the other direction,

ε ε
∀ > 0 ∃ N0 ∈ N ∀n > N0 |Re(zn ) − Re(z)| <
2 2
ε ε
∀ > 0 ∃ N1 ∈ N ∀n > N1 |Im(zn ) − Im(z)| < .
2 2
Therefore,

|zn − z| = |Re(zn ) + Im(zn ) − Re(z) − Im(z)|

≤ |Re(zn ) − Re(z)| + |Im(zn ) − Im(z)|
≤ε

5.1.2 Continuity

 Definition 8 (Continuity)

∀Ω ⊆ C, let f : Ω → C. We say that f is continuous at z0 ∈ Ω if

1. ∀{zn }n∈N
zn → z0 =⇒ f (zn ) → f (z0 )

2. ∀ε > 0 ∃δ > 0
|z − z0 | < δ =⇒ | f (z) − f (z0 )| < ε
40 Lecture 5 Jan 12 2018 - Complex Functions

Remark
1. f is continuous on Ω if it is continuous on every point in Ω.

2. We may split f into its feal and imaginary parts, i.e.

f (z) = f ( x, y) = u( x, y) + iv( x, y) (5.5)

where u, v : R2 → R.

Example 5.1.3

Let f : C → C and for z ∈ C, f (z) = zz . To split f into real and imaginary

parts:

z
f (z) =
z  
x y
= ( x + iy) 2 2
−i 2
x +y x + y2
x 2 − y2 (−2xy)
= +i 2
x 2 + y2 x + y2

and we get

x 2 − y2
u( x, y) =
x 2 + y2
2xy
v( x, y) = − 2
x + y2
6 Lecture 6 Jan 15th 2018

6.1 Continuity (Continued)

Exercise 6.1.1

Let f : Ω → C. Prove that f (z) is continuous at z0 = x0 + iy0 ∈ C ⇐⇒

functions u, v : R2 → R, such that f (z) = u( x, y) + iv( x, y) are both
continuous at ( x0 , y0 ).

Ò Solution
We shall first prove the forward direction. Suppose that f (z) is continuous
at z0 = x0 + iy0 ∈ C. By  Definition 8, ∀{zn }n∈N ⊆ Ω, zn → z0 =⇒
f (zn ) → f (z0 ). By Exercise 5.1.1,

zn → z0 ⇐⇒ Re zn → Re z0 ∧ Im zn → Im z0
⇐⇒ xn → x0 ∧ yn → y0 (6.1)

where zn = xn + iyn for xn , yn ∈ R.

Similarly so, and by Equation (5.5),

f (zn ) + f (z0 ) ⇐⇒ u( xn , yn ) → u( x0 , y0 ) ∧ v( xn , yn ) → v( x0 , y0 )
(6.2)

Putting together Equation (6.1) and Equation (6.2), we get

( xn , yn ) → ( x0 , y0 ) =⇒ u( xn , yn ) → u( x0 , y0 ) ∧ v( xn , yn ) → v( x0 , y0 )

as desired.

The proof of the other direction is simply a reversed process of the above.

 42 Lecture 6 Jan 15th 2018 - Differentiability

6.2 Differentiability

 Definition 9 (Neighbourhood)

For z0 ∈ C, r ∈ R, let

D ( z0 , r ) : = { z ∈ C : | z − z0 | < r }. (6.3)

On the complex plane, this is seen as a open disk centered around the
point z0 with radius r, as shown below. This open disk is called a neigh-

Im Figure 6.1: Open disk centered around

z0 with radius r

r
z0

Re

bourhood of z0 .

 Definition 10 (Differentiable/Holomorphic)

Let f (z) be defined in a neighbourhood of z0 ∈ C. We say f is differen-

tiable/holomorphic at z0 if for some h ∈ C,

f ( z0 + h ) − f ( z0 )
lim (6.4)
h →0 h

exists. If such a limit exists, we denote the limit by f 0 (z0 ).

Remark
h ∈ C : h need not necessarily be real. In this sense, h approaches 0 from
any direction around 0 ∈ C.

Example 6.2.1
 PMATH352W18 - Complex Analysis 43

For z ∈ C \ {0}, let f (z) = 1z . Let z0 ∈ C \ {0}. Note that

1 1
z0 + h − z0 1

−h

1
lim = lim =− 2
h →0 h h →0 h ( z0 + h ) z0 z0

Thus f is holomorphic at any z ∈ C \ {0}, and hence f 0 (z) = − 1z .

Example 6.2.2

For z ∈ C, let f (z) = z. Let z0 ∈ C. Notice that

z0 + h − z h
lim = lim .
h →0 h h →0 h

From Example 5.1.2, we know that such a limit does not exist. Thus f is not
holomorphic on any z ∈ C.

Exercise 6.2.1 (Holomorphic Functions Properties)

If f , g are holomorphic at z ∈ C, prove that

1. f + g is holomorphic and ( f + g)0 = f 0 + g0 .

2. f g is holomorphic and ( f g)0 = f 0 g + f g0 .

f f f 0 g− f g0
3. if g(z) 6= 0, g is holomorphic and ( g )0 = g2
.

Ò Solution
1. For f + g,

f (z + h) + g(z + h) − f (z) − g(z)

lim
h →0 h
f (z + h) − f (z) g(z + h) − g(z)
 
= lim +
h →0 h g
= f 0 (z) + g0 (z)

Thus ( f + g)0 = f 0 + g0 .

2. For f g,

f (z + h) g(z + h) − f (z) g(z)

lim
h →0 h
f (z + h) g(z + h) + f (z) g(z + h) − f (z) g(z + h) − f (z) g(z)
= lim
h →0 h
f (z + h) − f (z) g(z + h) − g(z)
 
= lim g(z + h) + f (z)
h →0 h h
= f 0 (z) g(z) + f (z) g0 (z)
 44 Lecture 6 Jan 15th 2018 - Differentiability

Therefore, ( f g)0 = f 0 g + f g0 .

f
3. When ∀z ∈ C g(z) 6= 0, for g ,

f (z+h) f (z)
g(z+h)
− g(z)
lim
h →0 h
f (z + h) g(z) − f (z) g(z + h)
 
1
= lim
h →0 h g(z + h) g(z)
f (z + h) g(z) + f (z) g(z) − f (z) g(z) − f (z) g(z + h)
 
1
= lim
h →0 g ( z + h ) g ( z ) g
[ f (z + h) − f (z)] g(z) − f (z)[ g(z + h) − g(z)]
 
1
= lim
h →0 g ( z + h ) g ( z ) h
f 0 (z) g(z) − f (z) g0 (z)
=
g2 ( z )

f f 0 g− f g0
Hence, g = g2

å Note
If we look at the example above from the perspective of f being treated as a
real-valued function, i.e. f (z) = u( x, y) + iv( x, y) where u, v : R2 → R
and z = x + iy, observe that ∀( x, y) ∈ R2 , ( x, y) 7→ ( x, −y), which we
see that u and v are partially differentiable in R2 .

We will now look into this “discrepancy”.

6.2.1 Cauchy-Riemann Equations

Consider the following function taken from Equation (6.4),

f ( z0 + h ) − f ( z0 )
f 0 (z0 ) = lim (6.5)
h →0 h

While h may approach 0 ∈ C from infinitely many sides on the

complex plane, we will consider 2 cases.

Case 1: h → 0 via the real axis

 PMATH352W18 - Complex Analysis 45

In this case, h = x + i (0) and x → 0 ∈ R. Then Equation (6.5) gives

u( x0 + x, y0 ) + iv( x0 + x, y0 ) − u( x0 , y0 ) − iv( x0 , y0 )
f 0 (z0 ) = lim
x →0 x
u( x0 + x, y0 ) − u( x0 , y0 ) v( x0 + x, y0 ) − v( x0 , y0 )
 
= lim +i
x →0 x x
∂u ∂v
= +i (6.6)
∂x ( x0 ,y0 ) ∂x ( x0 ,y0 )

Case 2: h → 0 via the imaginary axis

In this case, h = 0 + iy and y → 0 ∈ R. In a similar fashion,

Equation (6.5) becomes

u ( x0 , y0 + y ) − u ( x0 , y0 ) v ( x0 , y0 + y ) − v ( x0 , y0 )
 
f 0 (z0 ) = lim +
y →0 iy y
1 ∂u ∂v
= · + (6.7)
i ∂y ( x0 ,y0 ) ∂y ( x0 ,y0 )

Note that since f 0 (z0 ) exists, the real and imaginary part of Equa-
1
tion (6.6) and Equation (6.7) must equate. Also note that i = −i.
With that, we obtain the following theorem.

1 Theorem 3 (Cauchy-Riemann Equations)

If f (z) is holomorphic at z0 = x0 + iy0 ∈ C where x0 , y0 ∈ R, then, at

( x0 , y0 ),
∂u ∂v ∂v ∂u
= and =− . (6.8)
∂x ∂y ∂x ∂y
 7 Lecture 7 Jan 17th 2018

7.1 Differentiability (Continued)

7.1.1 Cauchy-Riemann Equations (Continued)

It is natural to wonder if the converse of 1 Theorem 3 is true. We

present the following example.

Example 7.1.1

Let 
 z2 if z 6= 0
z
f (z) =
0 if z = 0

Check if

1. f is holomorphic at 0.

2. 1 Theorem 3 holds at (0, 0).

Ò Proof
1. Observe that by letting h = xh + iyh where xh , yh ∈ R,

0+ h2 2
−0 h2 xh − iyh

0+ h
lim = lim = lim
h →0 h h →0 h xh +iyh →0 xh + iyh

Consider yh = kxh , for k ∈ R \ {0}. Then

2 2
xh − ikxh 1 − ik
 
lim = ,
x h →0 xh + ikxh 1 + ik

where we see that the limit depends on the value of k. Therefore, the
limit DNE. Hence f is not holomorphic at 0.
48 Lecture 7 Jan 17th 2018 - Differentiability (Continued)

2. Let z = x + iy for x, y ∈ R. Then

z2 ( x − iy)2 ( x − iy)3 x3 − 3xy2 (−3x2 y + y3 )

= = 2 = + i
z x + iy x + y2 x 2 + y2 x 2 + y2

Therefore, we obtain
 3
 x −3xy2 ( x, y) 6= (0, 0)
x 2 + y2
u( x, y) =
0 ( x, y) = (0, 0)
 3 2
 y −3x y ( x, y) 6= (0, 0)
x 2 + y2
v( x, y) =
0 ( x, y) = (0, 0)

Observe that
∂u u( x, 0) − u(0, 0)
= lim =1
∂x (0,0) x →0 x
∂v v(0, y) − v(0, 0)
= lim =1
∂y (0,0) y →0 y
and
∂u u(0, y) − u(0, 0)
= lim =0
∂y (0,0) y →0 y
∂v v( x, 0) − v(0, 0)
= lim =0
∂x (0,0) x →0 x

satisfies Equation (6.8).

This illustrates that the converse of 1 Theorem 3 is not true. We will,

however, show that the converse will be true given an extra condition.

1 Theorem 4 (Conditional Converse of CRE)

Let z0 = x0 + iy0 ∈ Ω ⊆ C, x0 , y0 ∈ R, and u, v : R2 → R, f = u + iv :

Ω → C. If

1. the partials of u, v exist in a neighbourhood of ( x0 , y0 ),

2. the partials of u, v are continuous at ( x0 , y0 ), and

3. ∂u
∂x = ∂v
∂y and ∂v
∂x = − ∂u
∂y at ( x0 , y0 ),

then f is holomorphic at z0 .
 PMATH352W18 - Complex Analysis 49

A proof of the theorem is in page 36 of Newman and Bak (recom-

mended text of PMATH352W18). I may include the proof whenever I
am free.

7.1.2 Power Series

 Definition 11 (Power Series)

A power series in C is an infinite series of the form

∑ cn zn , (7.1)
n ∈N

where each cn ∈ C is the coefficient of z of the n-th power.

In this subsection, we are interested to see if Equation (7.1) con-

verges.

Recall the notion of convergence in series from R. Equation (7.1)

converges if the sequence of partial sums {S N } converges as N → ∞,
where
N
S N := ∑ cn zn
n =0

In other words, using the same definition of S N ,

∀ ε > 0 ∃ N ∈ N \ {0} ∀ n > N

| Sn − L | < ε

where L ∈ C is the limit that the sequence converges to.

We also know that Equation (7.1) converges absolutely if ∑∞ n =0 | c n | | z |

n

converges. This is a stronger statement (i.e. absolute convergence

=⇒ convergence)

N N
∵ ∑ cn zn ≤ ∑ |cn | |z|n for each N ∈ N
n =0 n =0

Example 7.1.2

∑∞ n
n=0 z converges absolutely for | z | < 1.
50 Lecture 7 Jan 17th 2018 - Differentiability (Continued)

Note that the partial sum of a geometric series is

N
1 − r N +1
∑ rn = 1−r
n =0

and so the limit as N → ∞ exists if |r | < 1, and hence we see that

N
1
∑ rn → 1 − r
n =0

if |r | < 1 as N → ∞.

However, if |z| = 1, the power series diverges.

Another note that we shall point out is that if Equation (7.1) con-
verges absolutely for some z0 ∈ C, then it converges absolutely for
any z where |z| < |z0 |.

These notions, in turn, begs the question of what is the largest

possible |z0 | for the series to converge absolutely.
 8 Lecture 8 Jan 19 2018

8.1 Power Series (Continued)

8.1.1 Radius of Convergence

1 Theorem 5 (Convergence in the Radius of Convergence)

For any power series ∑n∈N cn zn , ∃0 ≤ R < ∞, such that

1. |z| < R =⇒ series converges absolutely.

2. |z| > R =⇒ series diverges.

Moreover, R is given by Hadamard’s Formula:

1 1
:= lim sup |cn | n (8.1)
R n→∞

Remark
1. R is called the radius of convergence of the series. {z ∈ C : |z| < R} is
called the disk of convergence of the series.

2. Recall the definition of the limit supremum

lim sup an := lim ( sup am ) (8.2)

n→∞ n→∞ m≥n

which we may colloquially say as the “highest peak ‘reached’ by an ’s as

n → ∞”

0 Proposition 6 (A Property of limsup)

52 Lecture 8 Jan 19 2018 - Power Series (Continued)

∀{ an }n∈N L := lim sup an =⇒

n→∞

∀ε > 0 ∃ N > 0 ∀n > N

L − ε < an < L + ε

(Proof to be included)

Ò Proof (1 Theorem 5)
1
1
Let L := R = lim supn→∞ |cn | n . Clearly, L ≥ 0.

1. Suppose |z| < R. ∃ε > 0, r := |z| ( L + ε) such that 0 < r < 1. By

1
0 Proposition 6, ∃ N ∈ N, ∀n > N, |cn | n < L + ε.

1
Now since L = R,

∞ ∞ 1
∞
∑ |cn | |z|n = ∑ (|cn | n |z|)n < ∑ rn
n= N n= N n= N

and since 0 < r < 1, the final summation converges (as it is a

geometric sum). Thus by comparison test, ∑∞ n
n= N | cn | | z | converges.

We may also proceed with noticing that the partial sum of ∑∞

n= N |cn | |z |
n

is bounded and monotonic, which shows that the series converges.

2. Suppose |z| > R. ∃ε > 0, r := |z| ( L − ε) such that r > 1. By

1
0 Proposition 6, ∃ N ∈ N, ∀n > N, |cn | n > L − ε. Then analogous to
the proof above,
∞ ∞ 1
∞
∑ |cn | |z|n = ∑ (|cn | n |z|)n > ∑ rn
n= N n= N n= N

where the final summation diverges, and thus implying that ∑∞

n= N |cn | |z |
n

diverges.

1 Theorem 7 (Power function, holomorphic function, region of

convergence)

Suppose f (z) = ∑n∈N cn zn has a radius of convergence R ∈ R. Then

 PMATH352W18 - Complex Analysis 53

f 0 (z) exists and equals

∞
∑ ncn zn−1
n =1

throughout |z| < R.

Moreover, f 0 has the same radius of convergence as f .

Ò Proof
Note that f 0 has the same radius of convergence as f since

1 1 1 1
lim sup |ncn | n = lim sup |n| n |cn | n = lim sup |cn | n
n→∞ n→∞ n→∞

1
where note that limn→∞ |n| n = 1.

Let |z0 | ≤ r < R and g(z0 ) := ∑∞ n −1

n=1 ncn z0 .

WTS
f ( z0 + h ) − f ( z0 )
lim = g ( z0 )
h →0 h
OR

∀ε > 0 ∃δ > 0 (†)

f ( z0 + h ) − f ( z0 )
|h| < δ =⇒ − g ( z0 ) < ε
h

(Note that we would want δ > 0 such that |z0 + h| ≤ r < R)

WTP (†). ∀ε > 0, choose δ > 0. Suppose |z0 + h| < |z0 + δ| ≤ r <
R. Write
N ∞
f (z) = ∑ cn zn + ∑ cn zn
n =0 n= N

and let S N (z) and EN (z) be the first and second terms respectively. Then
we have that
N
S0N (z) = ∑ ncn zn−1
n =1

Consider
S N ( z0 + h ) − S N ( z0 ) E ( z + h ) − E N ( z0 )
+ N 0 − g(z0 ) + S0N (z0 ) − S0N (z0 )
h h
S ( z + h ) − S N ( z0 ) E ( z + h ) − E N ( z0 )
≤ N 0 − S0N (z0 ) + N 0 + S0N (z0 ) − g(z0 )
h h
(8.3)
54 Lecture 8 Jan 19 2018 - Power Series (Continued)

For the second term, since a2 − b2 = ( a − b)( an−1 + an−2 b + . . . abn−2 +

bn−1 ) and |z0 | , |z0 + h| < r ≤ R, we obtain (z0 + h)n − z0n ≤ hnr n−1 .
Thus
∞ ∞
E N ( z0 + h ) − E N ( z0 ) 1
h
=
h ∑ cn [(z0 + h)n − z0n ] ≤ ∑ ncn r n−1 .
n = N +1 n = N +1

Note that
∞
∑ ncn rn−1 = g(r), (8.4)
n =1

and since r < R, Equation (8.4) converges absolutely by 1 Theorem 5,

thus the tail ∑∞
n= N +1 ncn r
n−1 converges absolutely. Therefore, by com-

parison, we can pick ε

3 > 0 so that ∃ N1 ∈ N, ∀n > N1

E N ( z0 + h ) − E N ( z0 ) ε
< .
h 3

For the third term in Equation (8.3), we observe that by definition,

S0N (z0 ) = ∑nN=1 ncn zn−1 . Since

N ∞
lim S0N (z0 ) = lim
N →∞ N →∞
∑ ncn zn−1 = ∑ ncn zn−1 = g(z0 )
n =1 n =1

we know that we can pick ε

3 > 0, ∃ N2 ∈ N, ∀n > N2 , we have
ε
S0N (z0 ) − g(z0 ) <
3
.

For the first term in Equation (8.3), note that (S N (z0 ))0 = S0N (z0 ).
Let ε
3 > 0, ∃δ > 0, ∃ N > max{ N1 , N2 }, ∀n > N, since |h| < δ, we have
that
S N ( z0 + h ) − S N ( z0 ) ε
− S0N (z0 ) <
h 3

This completes the proof. 

 9 Lecture 9 Jan 22nd 2018

9.1 Power Series (Continued 2)

9.1.1 Radius of Convergence (Continued)

Example 9.1.1

Let f (z) = ∑∞ z n
n=1 n . To find the radius of convergence, we use Hadamard’s
Formula:
 1
1 1 n 1
= lim sup =1 ∵ lim n n = 1
R n→∞ n n → ∞

Therefore R = 1. Thus, by 1 Theorem 5, f converges absolutely when

|z| < 1 and diverges when |z| > 1. As for the boundary, i.e. |z| = 1,
consider the following two cases:

1. If z = 1, then f (1) = ∑∞
n =1
1
n is a harmonic series, and hence f
diverges.

2. If z = i, then
∞
in
f (i ) = ∑ n
n =1
1 −i 1 i 1
= i− + + + −
 2 3 4 5  6  
1 1 1 1 1
= − + − +... +i 1− + +... .
2 4 6 3 5

Observe that both the real and imaginary parts are alternating series
where the absolute values of each term is decreasing, which, by the alter-
nating series test, converge. Thus in this case, f converges.

Therefore, we observe that both convergence and divergence may occur

on the boundary, depending on the value of z.
56 Lecture 9 Jan 22nd 2018 - Power Series (Continued 2)

å Note
We may not always exchange the position of lim and ∑ba=1 when we
consider an infinite sum (i.e. b = ∞). Here’s an example why this is true.
Consider the function f ( x ) = ∑∞ n
n =1 ( x − x
n−1 ) for | x | < 1. Is

∞ ∞
lim
x →1 n =1
∑ ( x n − x n −1 ) = ∑ xlim
→1
( x n − x n +1 )
n =1

true?

Clearly, RHS is 0. For LHS, note that

N
f ( x ) = lim
N →∞
∑ ( x n − x n +1 )
n =1

= lim ( x − x2 + x2 − x3 + . . . + x N − x N +1 )
N →∞

= lim ( x − x N +1 ) = x.
N →∞

So,
LHS = lim x = 1
x →1
And we see that RHS 6= LHS.

 Definition 12 (Entire Function)

A function f is said to be entire if f is holomorphic in the entire com-

plex plane.

Exercise 9.1.1

Define ez = ∑∞
n =0
zn
n! . Show that

1. the radius of convergence of this series is ∞, and hence that ez is an entire

√
function. (Hint: Use Stirling’s formula: n! ∼ ( ne )n 2πn)

2. (ez )0 = ez

Ò Solution
1. Using Stirling’s formula, note that we have

∞  ez n
1
ez = ∑ √
2πn n
n =0
 PMATH352W18 - Complex Analysis 57

To find R, we have
1
1 1  e n n
= lim sup √
R n→∞ 2πn n
1
e 1 n
= lim sup lim sup √
n→∞ n n→∞ 2πn
=0
1
n
since lim supn→∞ e
n = 0 and lim supn→∞ √1 = 1. Thus R = ∞.
2πn
By 1 Theorem 5, ez is an entire function.

2. Note that

ez+h − ez eh − 1
lim = ez lim
h →0 h h →0 h
h2 3
z 1+h+ 2 + h3 + . . . − 1
= e lim
h →0 h
= ez

Thus (ez )0 = ez as required.

 10 Lecture 10 Jan 24th 2018

10.1 Power Series (Continued 3)

10.1.1 Radius of Convergence (Continued 2)

A power series is infinitely C-differentiable in its radius of conver-

gence. All its derivatives are also power series, obtained by term-wise
differentiation.

E.g.

∞ ∞
f (z) − ∑ cn zn then f (2) (z) = ∑ n ( n − 1 ) c n z n −2
n =0 n =0

In general, we may have ∑∞ n

n=0 cn ( z − z0 ) , which is a power series
centered at z0 ∈ C. Then, as before, the radius of convergence of this
power series is given by

1 1
= lim sup |cn | n
R n→∞

So instead of having the disc of convergence centered around 0, we

now have one that is centered around z0 .

 Corollary 8 (Corollary of 1 Theorem 7)

From 1 Theorem 7, we have shown that

f (z) has a power series expansion at z0 (i.e.

f (z) = ∑∞ n
n=0 cn ( z − z0 ) in some
=⇒ f is holomorphic at z0
neighbourhood of z0 ) with radius of convergence
R>0
 60 Lecture 10 Jan 24th 2018 - Integration in C

The converse of the statement above is true, i.e.

f (z) has a power series expansion at z0 (i.e.

f (z) = ∑∞ n
n=0 cn ( z − z0 ) in some
f is holomorphic at z0 =⇒
neighbourhood of z0 ) with radius of
convergence R > 0

This converse, however, is not possible to be proven given the cur-

rent tools on our belt. And so we now have to venture into integrals
in C.

10.2 Integration in C

10.2.1 Curves and Paths

Before we begin with the definition of a curve in C, let us consider

how a straight line should be described as a vector-valued function in
the complex plane. For instance, if we have two points α, β ∈ C, and
we want to describe the straight line connecting the two.

β Let γ be the function that describes this line. We may then

define γ : [0, 1] → C to be either

γ(t) = α + ( β − α)t or γ = α(1 − t) + βt.

α We would then have the following mapping:

Figure 10.1: Mapping from R → C with

γ, which is called the curve γ

γ b

c
R [ ] C
a b

c
a
 PMATH352W18 - Complex Analysis 61

 Definition 13 (Curves in C)

A curve in C is a continuous function, γ(t) : [ a, b] → C, where a, b ∈ R.

The image of γ in C is called γ∗ .

Example 10.2.1

Let z0 ∈ C, r > 0.

π
t= 2

1. Let γ : [0, 2π ] → C, such that γ(t) = z0 + reit .

r
2. Let γ0 : [0, 1] → C, such that γ0 (t) = z0 + re2πit . t=π z0 t = 0 or 2π
The two functions above describe a circle centered at z0 with
radius r, anticlockwise-oriented. 3π
t= 2

We say that γ and γ0 are equivalent parameterizations for the same

oriented path.

 Definition 14 (Equivalent Parameterization)

Let γ1 : [ a, b] → C, γ2 : [c, d] → C where a, b, c, d ∈ C describe the path

γ∗ . The two parameterizations are said to be equivalent parameteriza-
tions if ∃h : [ a, b] → [c, d] that is a bijection and a continuous function
such that
γ1 (t) = γ2 (h(t))

where t ∈ [ a, b].

å Note
We will not look at functions like the Weierstrass function in this course.

 Definition 15 (Smooth Curve)

Let γ : [ a, b] → C, a, b, ∈ C. γ is said to be smooth if its derivative γ0

exists and is continuous on [ a, b] and ∀t ∈ [ a, b], γ0 (t) 6= 0.
 62 Lecture 10 Jan 24th 2018 - Integration in C

 Definition 16 (Piecewise Smooth)

Let γ : [ a, b] → C. γ is said to be piecewise smooth if it is smooth on

[ a, b] except on finitely many points in [ a, b].

Remark
Piecewise smooth curves shall be called paths.

10.2.2 Integral

 Definition 17 (Contour)

Given a path γ : [ a, b] → C and f : C → C, a function continuous on γ.

We define the integral f along γ, called a contour, as
Z Z b
f (z) dz := f (γ(t))γ0 (t)dt (10.1)
γ a

where we let z = γ(t) and hence dz = γ0 (t)dt.

Remark
1. Suppose g is a complex-valued function, then
Z b Z b Z b
g(t)dt = Re( g(t))dt + i Im( g(t))dt
a a a

2. The integral of f along γ can be shown to be independent of the chosen

parameterization for γ∗ .

Ò Proof
Let a, b, c, d ∈ R, γ1 : [ a, b] → C, γ2 : [c, d] → C describe the same
path γ∗ . By  Definition 14, define a bijection h : [ a, b] → [c, d] that
is a continuous function such that t 7→ τ, so that

γ1 (t) = γ2 (h(t)) = γ(τ ).

 PMATH352W18 - Complex Analysis 63

Note that

γ10 (t) = h0 (t)γ20 (h(t)) and

h(t) = τ =⇒ h0 (t)dt = dτ.

Now since h is a bijection, we claim that h( a) = c while h(b) = d.

We know that h cannot be a constant function. Suppose h is an in-

creasing function, then since a ≤ b and c ≤ d, it is clear that h( a) = c
and h(b) = d. Similarly, if h is a decreasing function, then h( a) = d
and h(b) = c. But this is a contradiction to our supposition that γ1
and γ2 describe the same orientation. Thus h must be an increasing
function, and hence we have h( a) = c and h(b) = d.

(This can be more rigorous but that is an easy proof, and we

may use perhaps the Approximation Property of R to that end,
which is a fun exercise that shall not be included within these
covers.)

Now
Z Z b
f γ1 (t) γ10 (t)dt


f (z) dz =
γ1 a
Z b 

= f γ2 h(t) h0 (t)γ20 (h(t))dt
a
Z d
f γ2 (τ ) γ20 (τ )dτ


=
Zc
= f (z) dz
γ2

This completes the proof. 

 11 Lecture 11 Jan 26th 2018

11.1 Integration in C (Continued)

11.1.1 Integral (Continued)

å Note (Recall)
Let γ : [ a, b] → C be a piecewise smooth curve. For a function f that is
continuous on γ, we defined
Z Z b
f γ(t) γ0 (t)dt


f (z) dz =
γ a
Z b   Z b  
Re f γ(t) γ0 (t) dt + i Im f γ(t) γ0 (t) dt



=
a a

and have

γ0 (t) = u0 (t) + iv0 (t)

if γ(t) = u(t) + iv(t)

Example 11.1.1

Let f (z) = f ( x + iy) = x2 + y2 be continuous along γ : [0, 1] → C t 7→

R
t + it. Evaluate γ f (z) dz.
66 Lecture 11 Jan 26th 2018 - Integration in C (Continued)

Ò Solution

Z Z 1
f (z) dz = f (t + it)(1 + i )dt
γ 0
Z 1
= (1 + i )2 t2 dt
0
1 1
= (1 + i )2 · t3
3 0
2i
=
3
Example 11.1.2

∀n ∈ Z, evaluate γ zn dz that is continue on the path γ that describes any

R

circle centered at origin oriented anticlockwise.

Ò Solution
Let R ∈ R, and define

γ : [0, 1] → C t 7→ Re2πit
γ0 (t) = 2Rπie2πit = 2πiγ(t)

Then
Z Z 1
zn dz = Rn e2πint · 2πi · Re2πit dt
γ 0
Z 1
n +1
= 2πiR e2πi(n+1)t dt
0
1

n +1
 R e2πi(n+1)t

n +1 if n ∈ Z \ {−1}
= 0
1
2πit if n = −1

0
  
 Rn+1 e2πi(n+1) − 1 if n ∈ Z \ {−1}
n +1
= ∵ e2πki ≡ 1 mod 2π
2πi if n = −1

0 if n ∈ Z \ {−1}
=
2πi if n = −1

Note that our final answer does not depend on R, the radius of the circle.

0 Proposition 9 (Properties of integrals in C)

1. (Linearity) Let α, β ∈ C.
R R
γ (α f (z) + βg(z)) dz = α γ f (z) dz +
R
β γ g(z) dz.
 PMATH352W18 - Complex Analysis 67

2.(a) For any complex-valued function g, and b ≥ a,

Z b Z b
g(t)dt ≤ | g(t)| dt
a a

(b) For any function f (z) that is continuous on a path γ : [ a, b] → C,

Z Z b
f (z) dz ≤ sup | f (z)| · γ0 (t) dt
γ z∈γ a
| {z }
length of the path

3. If γ− is the path γ : [ a, b] → C with a reversed direction, then

Z Z
f (z) dz = − f (z) dz
γ− γ

Ò Proof
1.
Z
LHS = α f (z) + βg(z) dz
γ
Z    
= α Re( f (z)) + i Im( f (z)) + β Re( g(z)) + i Im( g(z)) dz
γ
Z b 
α Re( f (z)) + i Im( f (z)) dt
a
= Z b  (†)
+β Re( g(z)) + i Im( g(z)) dt
a
Z b Z b
=α f (z) dz + β g(z) dz
a a

where (†) is because of Item 1 from an earlier remark and Linearity of

Integrals in R. 

Rb
2.(a) Let R ∈ R. Suppose a g(t)dt = Reiθ . Then
Z b
LHS = R = e−iθ g(t) dt
a | {z }
u(t)+iv(t)
Z b Z b
= u(t)dt + i v(t)dt
a a
68 Lecture 11 Jan 26th 2018 - Integration in C (Continued)

Rb
Since R ∈ R, we have a v(t)dt = 0, so
Z b
R= u(t)dt
a
Z b
≤ |u(t)| dt ∵ a ≤ b ∧ u(t) ≤ |u(t)| ∈ R
a
Z b
≤ |u(t) + iv(t)| dt
a
Z b
= e−iθ g(t) dt
a
Z b
= | g(t)| dt = RHS
a

(b)
Z
LHS = f (z) dz
γ
Z b
f γ(t) γ0 (t)dt


=
a
Z b
f γ(t) γ0 (t) dt


≤ by Item 2a
a
Z b
≤ sup | f (z)| γ0 (t) dt since | f (z)| ≤ sup | f (z)|
a z∈γ z∈γ
Z b
= sup | f (z)| · γ0 (t) dt = RHS
z∈γ a

3. Let γ− : [b, a] → C such that γ− = γ(b − t + a). Let k = b − t + a

so that dk = −dt, and t = b =⇒ k = a ∧ t = a =⇒ k = b. Note
that k ∈ [ a, b] with this definition. Then
Z
LHS = f (z) dz
γ−
Z a
0
= f γ− (t) γ− (t)dt
b
Z a
f γ(b − t + a) γ0 (b − t + a)dt


=
b
Z b
f γ(k) γ0 (k)dk


=−
Za
=− f (z) dz = RHS
γ

as required. 
 PMATH352W18 - Complex Analysis 69

We are now in a position to generalize the Fundamental Theorem

of Calclus for C.
 12 Lecture 12 Jan 29th 2018

12.1 Integration in C (Continued 2)

12.1.1 Fundamental Theorem of Calculus

To simplify statements from hereon, we shall use the following nota-

tions.

 Notation
Let Ω ⊆ C be an open set in C. We denote f ∈ H (Ω) ⇐⇒ f is
holomorphic on Ω.

1 Theorem 10 (Fundamental Theorem of Calculus)

Let γ : [ a, b] → C be a path inside an open set Ω ⊆ C. Suppose f (z) is

continuous on γ, and has an antiderivative F ∈ Ω. Then
Z


f (z) dz = F γ(b) − F γ( a) (12.1)
γ

Ò Proof
Let G = F ◦ γ and suppose γ is a smooth function. Since γ is smooth,
γ0 exists and is continuous on [ a, b] and γ0 (t) 6= 0 for all t ∈ [ a, b], and
since f is continuous on [ a, b], G (t) = F 0 (γ(t))γ0 (t) is continuous as
well.
72 Lecture 12 Jan 29th 2018 - Integration in C (Continued 2)

Now
Z Z b
f γ(t) γ0 (t)dt


f (z) dz =
γ a
Z b
F 0 γ(t) γ0 (t)dt


=
a
Z b
= G 0 (t)dt
a
= G (b) − G ( a) by applying FTC in R to real and imaginary parts
= F (γ(b)) − F (γ( a))

If γ is piecewise smooth, then we can simply apply the above to each of

the smooth paths separately and sum up all of the integrals. 

 Definition 18 (Closed Path)

A path γ : [ a, b] → C is said to be closed if γ( a) = γ(b).

 Corollary 11 (Corollary of FTC)

If F ∈ H (Ω), Ω ⊆ C (hence F 0 is continuous on Ω), then

Z
F 0 (z) dz = 0
γ

on any closed path γ on Ω.

Ò Proof
A closed path γ : [ a, b] → C has γ( a) = γ(b). By 1 Theorem 10,
R 0
γ F ( z ) dz = F ( γ ( b )) − F ( γ ( a )) = 0 as required. 

Example 12.1.1

Take f (z) = zn where n ∈ Z \ {−1} as in Example 11.1.2. Then f is

continuous on C \ {0} (not sure why this would be problematic when
n +1
we’ve already excluded -1 for n). Then f = F 0 for F (z) = zn+1 and
F ∈ H (C \ {0}). Therefore by  Corollary 11, γ zn dz = 0 for any closed
R

path γ not passing through 0.

 PMATH352W18 - Complex Analysis 73

If we do include −1 for n, note that F 0 would not be continuous on 0,

and thus the corollary would not apply. We have also shown in the earlier
example that γ 1z dz = 2πi.
R

å Note (Recall)
The interior of a set Ω is defined as {z ∈ Ω : ∀ε > 0 B(z, ε) ⊆ Ω}, and
denoted as Ω0 .

1 Theorem 12 (Goursat’s Theorem / Cauchy’s Theorem for a

triangle)

Let Ω ⊆ C be an open set. Suppose ∆ ⊆ Ω is a closed triangle whose

interior is also contained in Ω. Let f ∈ H (Ω). Then
Z
f (z) dz = 0
∆

This theorem holds more meaning than the presented statement,

as it implies that, essentially, given any two points connected by two
different paths in an open set in C, and a function that is holomor-
phic over the two paths, the two path integrals of the function will
yield the same result!

Ò Proof
(1) (1) (1) (1)
Let ∆1 , ∆2 , ∆3 , ∆4 be smaller triangles by bisecting each side of ∆.
(1)
∀i ∈ {1, 2, 3, 4}, orient ∆i anticlockwise. Then we have

Z 4 Z
J :=
∆
f (z) dz = ∑ (1)
f (z) dz (12.2)
i =1 ∆ i

(1) | J|
Note that there must at least one of the ∆i
R
such that (1) ≥ 4 , since
∆i
R | J|
∀i ∈ {1, 2, 3, 4}, (1) < 4 would contradict Equation (12.2). Without
∆i
(1)
loss of generality, let ∆1 be the largest triangle of the four.
(1)
Now note that each of the perimeter of ∆i is half of the perimeter
of ∆. Let `( x ) be the perimeter of x. Continue with taking bisectors of
74 Lecture 12 Jan 29th 2018 - Integration in C (Continued 2)

(1) (2)
∆1 , ∆1 , . . . such that

(1) (2)
∆ ⊇ ∆1 ⊇ ∆1 ⊇ . . . ,

( j)
then we have that for each j ∈ N \ {0}, ∆i is such that

| J|
Z
f (z) dz ≥
( j)
∆i 4j

( j)
and `(∆i ) = 1
2j
`(∆). By the Nested Rectangle Theorem from Real
( j)
Analysis, ∃z0 ∈ C such that z0 ∈ ∆i for all j ∈ N \ {0} that is a limit
point. Since z0 ∈ Ω ∧ f ∈ H (Ω), we have that

∀z ∈ Ω ∀ε > 0 ∃δ > 0
f ( z ) − f ( z0 )
0 < |z − z0 | < δ =⇒ − f 0 ( z0 ) < ε
z − z0

(n)
Consider D (z0 , δ), ∃n ∈ N \ {0}, ∆1 ⊆ D (z0 , δ). Consider
Z Z
f (z) − f (z0 ) − f 0 (z0 )(z − z0 ) dz,


(n)
f (z) dz = (n)
∆1 ∆1

f 0 (z0 )(z − z0 ) dz = 0 by
R R
where we note that (n) f (z0 ) dz = (n)
∆1 ∆1
 Corollary 11.

By Item 2a in 0 Proposition 9,
Z Z
(n)
f (z) dz ≤ (n)
f (z) − f (z0 ) − f 0 (z0 )(z − z0 ) dz
∆1 ∆1
Z Z
(n) (n)
≤ (n)
ε |z − z0 | dz ≤ ε`(∆1 ) (n)
dz = ε`(∆1 )2
∆1 ∆1

(n) (n)
where we note that |z − z0 | ≤ `(∆1 since z ∈ ∆1 . This implies that

| J| (n) 2 `(∆)2
≤ ε `( ∆ 1 ) ≤ ε = | J| ≤ ε
4n 4n
and therefore, | J | = 0!
 Tutorial Jan 31 2018

å Note
1
Consider the power series ∑n≥0 an (z − z0 )n and let
p
n
R := lim supn→∞ sup | an | ∈
[0, ∞).

• If |z − z0 | < R, ∑n≥0 an (z − z0 )n converges absolutely.

• If |z − z0 | > R, ∑n≥0 an (z − z0 )n diverges.

• If 0 < r < R, then ∑n≥0 a0 (z − z0 )n converges uniformly on

{ z : | z − z0 | < r }.

12.2 Practice Problems

1. Parameterize the semicircle |z − 4 − 5i | = 3 clockwise, starting

from z = 4 + 8i to z = 4 + 2i.
Ò Solution
Let γ : [− π2 , π2 ] → C such that γ(t) = 3e−it + 4 + 5i. Note that γ
parameterizes the given semicircle:
 π
γ − = 4 + 8i
2
γ(0) = 7 + 5i
π
γ = 4 + 2i
2

2. If the power series f (z) = ∑∞ n

n=0 cn ( z − z0 ) centered at z0 has a
76 Lecture 12 Jan 29th 2018 - Practice Problems

Im
Figure 12.1: Semicircle |z − 4 − 5i | = 3
γ(− π2 ) oriented clockwise, parameterized by γ

γ (0)
4 − 5i

γ( π2 )

Re

non-zero radius of convergence, then show that

f ( m ) ( z0 )
cm =
m!

for any m ∈ Z, m ≥ 0, where f (m) (z0 ) denotes the mth derivative

of f at z0 .

Ò Solution
Since f (z) is a power series and the radius of convergence R 6= 0, by
1 Theorem 7, f (z) is C-differentiable and each derivative has the same
radius of convergence. By induction, it can be shown that
∞
n!
f (m) ( z ) = ∑ ( n − m ) !
c n ( z − z0 ) n − m
n=m

Evaluating f (m) at z0 , we have

∞
n!
f ( m ) ( z0 ) = ∑ (n − m)!
c n ( z0 − z0 ) n − m
n=m

= m!cm

where all terms above m are 0. Then we obtain

f ( m ) ( z0 )
cm =
m!
as desired. 

3. Let γ be the arc of the unit circle centered at the origin in the first
quadrant oriented clockwise (from i to 1). Evaluate the integral
Z
z2 dz
γ
 PMATH352W18 - Complex Analysis 77

by parameterizing the curve.

Ò Solution
Consider the parameterization γ : [− π2 , 0] → C given by γ(t) = e−it .
Note that e−it = eit . Then
Z Z 0  
z2 dz = e2it · −ie−it dt
γ − π2
Z 0
= −i eit dt
− π2
0
= −eit π
2

= −1 − i

4. Evaluate the above integral by finding an antiderivative. (Hint:

2
Use zz
z )

Ò Solution
Note that zz = |z|2 , so on the circle, we have z = 1z . Thus the integral is
equivalent to
1
Z
dz
γ z2
1
Note that the antiderivative of z2
is − 1z . Thus by 1 Theorem 10,

1 1 1
Z Z   π 
z2 dz = 2
= F ( γ ( 0 )) − F γ − = − −i(0) + −i(−π/2) = −1 − i
γ γ z 2 e e

5. Let {cn }∞
n=0 be a sequence of positive real numbers such that

c n +1
L = lim
n→∞ cn

exists. Then show that

1
lim cnn = L
n→∞

This shows that, when applicable, the ratio test can be used in-
stead of the root test to calculate the radius of convergence of a
power series.

Ò Solution
Suppose that
c n +1
L = lim
n→∞ cn
78 Lecture 12 Jan 29th 2018 - Practice Problems

exists. By definition, we have

∀ε > 0 ∃ N ∈ N ∀n > N
cn
−L <ε
c n −1

Thus for n ≥ N,
 1
1
n
cn
c n −1 cN n
cn = · ... · c N −1
c n −1 c n −2 c N −1
 1  1  1 1
cn n c n −1 n cN n
= ... c Nn −1
c n −1 c n −2 c N −1

Now
1 1 1 1 1 1 1 1 1
( L − ε) n ( L − ε) n . . . ( L − ε) n c Nn −1 ≤ cnn ≤ ( L + ε) n ( L + ε) n . . . ( L + ε) n c Nn −1
n − N +1 1 1 n − N +1 1
( L − ε) n c Nn −1 ≤ cnn ≤ ( L + ε) n c Nn −1

Note that
n − N +1 1
lim ( L − ε) n c Nn −1 = L − ε
n→∞
n − N +1 1
lim ( L + ε) n c Nn −1 = L + ε
n→∞

Thus we have
1
L − ε ≤ cnn ≤ L + ε
1
cnn − L ≤ ε

as desired. 

6. Find the radius of convergence of

(a) ∑∞
n =0
nn zn
n!

(b) ∑∞ 2 n
n =0 z

Ò Solution
√
(a) By Stirling’s Approximation, i.e. n! ∼ ( ne )n 2πn, we have that
 PMATH352W18 - Complex Analysis 79

Hadamard’s formula is
1
1 nn n
= lim sup
R n→∞ n!
1
n
nn
= lim sup
√
n n
n→∞ e 2πn
1
en n
= lim sup √
n→∞ 2πn
1
1 n
= e lim sup √ =e
n→∞ 2πn

Therefore, R = 1e .

(b) no solution yet: current problem, not being able to express the
sum as a power series, in turn failing to get cn which is needed
1
for R.

7. Show that for any path γ : [ a, b] → C and f (z) continuous on γ, we

have Z Z b
f (z) dz ≤ sup | f (z)| γ0 (t) dt
γ z∈γ a

Ò Solution

Z
LHS = f (z) dz
γ
Z b
f γ(t) γ0 (t)dt by definition


=
a
Z b
f γ(t) γ0 (t) dt


≤ by Item 2a of 0 Proposition 9
a
Z b
≤ sup | f (z)| γ0 (t) dt since | f (z)| ≤ sup | f (z)|
a z∈γ z∈γ
Z b
= sup | f (z)| · γ0 (t) dt = RHS
z∈γ a
13 Lecture 13 Feb 9th 2018

13.1 Cauchy’s Integral Formula

 Definition 19 (Convex Set)

A set S ⊆ C is called a convex set if the line segment joining any pair of
points in S lies entirely in S.

1 Theorem 13 (Cauchy’s Theorem for Convex Set)

Let Ω ⊆ C be a convex open set, and f ∈ H (Ω). Then

1. f = F 0 for some F ∈ H (Ω).

f (z) dz = 0 for any closed path γ ∈ Ω.

R
2. γ

Ò Proof
Note that it is sufficient to prove 1 since 1 =⇒ 2 by 1 Theorem 10.

Let a ∈ Ω, and let [ a, z] denote the straight line from a to z. Since Ω is

a convex set, [ a, z] is in Ω. Define F (z)1 = [ a,z] f (z) dz2 .
R
1
It can be verified that F is continuous.
2
This is a keys step: defining an “an-
tiderivative” as how we would expect it
WTS F ∈ H (Ω), F 0 (z0 ) = f (z0 ) for any z0 ∈ Ω. to be.

Now by 1 Theorem 12,

82 Lecture 13 Feb 9th 2018 - Cauchy’s Integral Formula

Z
0= f (z) dz
Z∆ Z Z
= f (z) dz + f (z) dz + f (z) dz Ω
[ a,z] [z,z0 ] [z0 ,a]
Z a
∆
= F (z) + f (z) dz + (− F (z0 ))
[z,z0 ]
z0

This implies that z

Z
F ( z ) − F ( z0 ) = f (z) dz.
[z0 ,z]

Divide both sides by z − z0 , then

F ( z ) − F ( z0 ) 1
Z
− f ( z0 ) = f (z) dz − f (z0 )
z − z0 z − z0 [z0 ,z]
1
Z Z
= f (z) − f (z0 ) dz since dz = z − z0
z − z0 [z0 ,z] [z0 ,z]

Since f ∈ H (Ω) and is hence continuous, we have that

∀ε > 0 ∃δ > 0
|z − z0 | < δ =⇒ | f (z) − f (z0 )| < ε

which in turn implies that

F ( z ) − F ( z0 ) 1 1
Z Z
− f ( z0 ) = [ f (z) − f (z0 )] dz ≤ ε dz = ε
z − z0 z − z0 [z0 ,z] | z − z0 | [z0 ,z]

Hence, by first principle, F 0 (z0 ) = f (z0 ). 

1 Theorem 14 (Cauchy’s Integral Formula 1)

Let Ω ⊆ C be a convex open set, and C be a closed circle path in Ω. If

w ∈ Ω \ ∂C, where ∂C is the boundary of C, and f ∈ H (Ω), then

1 f (z)
Z
f (w) IndC (w) = dz
2πi C z−w
Ω
where
1 dz
Z
IndC (w) =
2πi γ z−w
C

denotes the number of times the countour C winds around the point w.
w

is called the index of w with respect to C, or the winding number

 PMATH352W18 - Complex Analysis 83

of C around w.

Ò Proof
Let w ∈ Ω \ ∂C. Define

 f (z)− f (w) if z 6= w
z−w
g(w) =
 f 0 (w) if z = w
Ω

By the construction of g, g is continuous on Ω, and g ∈ H (Ω \ {w}).

We need to construct a convex set Ω0 ⊆ Ω that contains γδ,ε such δ

that g ∈ H ( Ω 0 ). ε
w
C
Cε
We now follow a similar argument as in the proof for 1 Theorem 13. γδ,ε

Let ε > 0 such that ∃δ > 0, so that we can define the “keyhole”
γδ,ε which omits w. Consider D (w, ε), call the image of the border of
D (w, ε) as Cε , let δ be the width of the “corridor”, and the two paths
that are the “sides of the corridor” be called Cδ1 , Cδ2 respectively. Define
R
G (z) = [ a,z] g(z) dz, where a and z are in the interior of C but not in
the interior of Cε . Then if we define a set Ω0 such that it contains the in-
terior of γδ,ε , we have that Ω0 is a convex open set, and G ∈ H (Ω0 ). By
1 Theorem 13, G 0 = g.
R
Also from 1 Theorem 13, we have that γδ,ε g(z) dz = 0 for any
ε, δ > 0. As δ → 0+ , we have that the integrals over Cδ1 and Cδ2 cancel
out. Hence, we are left with

Z Z
g(z) dz + g(z) dz = 0
C Cε

Let’s put our focus on the smaller circle, Cε . Now as ε → 0+ ,

f (z)− f (w)
z−w → 0, and thus

f (z) − f (w)
Z Z
g(z) dz = dz → 0
Cε Cε z−w

Therefore, Z
g(z) dz = 0
C
which implies, in the limit, that

f (z) f (w) dz
Z Z Z
dz = dz = f (w)
C z−w C z−w C z−w
84 Lecture 13 Feb 9th 2018 - Cauchy’s Integral Formula

dz
R
We now require C z−w = 2πi, but we shall prove for a more general
case as a lemma.
14 Lecture 14 Feb 12 2018

14.1 Cauchy’s Integral Formula (Continued)

+ Lemma 15

(Lemma and proof from Newman & Bak on Complex Analysis, 3rd
Ed.)

Suppose a ∈ Cρ0 such that ∃α ∈ Cρ that is the center of the circle Cρ ,

where ρ is the radius of Cρ , and hence | a − α| < ρ. Then

dz
Z
= 2πi
Cρ z−a

Ò Proof
Let z ≡ α + ρeiθ , then dz = iρeiθ dθ. Thus

iρeiθ
Z 2π
dz
Z
= dθ = 2πi
Cρ z−α 0 ρeiθ

while
dz
Z
= 0 for k = 1, 2, 3, ... . (14.1)
Cρ ( z − α ) k +1
The Equation (14.1) follows not only from a direct evaluation of the inte-
gral

iρeiθ
Z 2π Z 2π
dz i
Z
= dθ = k e−ikθ dθ = 0
Cρ ( z − α ) k +1 0 iθ
(ρe ) k + 1 ρ 0

1
but also the fact that ( z − α ) k +1
is the derivative of − k(z−1 α)k , which can
be verified to be holomorphic on Cρ , which simply makes Equation (14.1)
86 Lecture 14 Feb 12 2018 - Cauchy’s Integral Formula (Continued)

true by 1 Theorem 10.

To evaluate Cρ zdz
R
− a , write

1 1 1
= = a−α
z−a (z − α) − ( a − α) (z − α)[1 − z−α ]
1 1
= ·
z−α 1−ω
where
a−α | a − α|
ω= has fixed modulus < 1 throughout Cρ (14.2)
z−α ρ

1
By Equation (14.2) and by the Infinite Geometric Sum that 1− ω =
1+ω + ω2 + . . ., we get

a − α ( a − α )2
 
1 1
= 1+ + + . . .
z−a z−α z−α ( z − α )2
1 a−α ( a − α )2
= + + +....
z − α ( z − α )2 ( z − α )3

Since the convergence is uniform throughout Cρ ,

∞
1 1 ( a − α)k
Z Z Z
dz = dz + ∑ dz = 2πi
Cρ z−a Cρ z−α k =1 Cρ ( z − α ) k +1

We may now continue with completing the previous proof.

Ò Proof (Continued - 1 Theorem 14)

dz
R
Lemma 15 completes the part where we required C z−w = 2πi.

We now have
1 f (z)
Z
f (w) = dz
2πi C z−w

Now note that if we further generalize the number of times the contour
Cρ made around a, where in this case Cρ is a closed path instead of a
simple circle in Ω, in Lemma 15, we would get
R dz
Cρ z− a = 2kπi where k
would represent that number.

In this case, we would get

1 f (z)
Z
f (w)k = dz
2πi C z−w
 PMATH352W18 - Complex Analysis 87

1 dz
R
where k = IndC (w) = 2πi C z−w which represents the number of times
the contour C winds around w. 

Remark
As noted, 1 Theorem 14 holds for any closed path γ ∈ Ω instead of a
simple circle C. If w ∈ Ω \ γ∗ , we get

1 f (z)
Z
dz = f (w) Indγ (w)
2πi γ z−w

0 Proposition 16 (Holomorphic Functions can be expressed as

Power series)

Let Ω ⊆ C be an open set, f ∈ H (Ω). Then f can be expressed as a

power series.

Ò Proof
∀w ∈ Ω, ∃C ⊆ Ω that is a closed circle path with w ∈ C0 . By
1 Theorem 14, and since C is a circle, i.e. the contour winds around
w only once, we have

1 f (z)
Z
f (w) = dz.
2πi C z−w

Let w0 ∈ Ω be the center of C. Then ∀z ∈ ∂C, 0 < |w − w0 | <

|z − w0 |1 . This implies that 1
This is the key step

| w − w0 |
0< <1
| z − w0 |
∞ 
w − w0 n z − w0

1
=⇒ ∑ = w − w = by the Infinite Geometric Sum
n =0
z − w0 1 − z − w00 z−w
f ( z ) z − w0 f ( z ) ∞ w − w0 n
 
1 f (z) 1 1
Z Z Z

2πi C z − w0 n∑
=⇒ dz = dz = dz
2πi C z − w 2πi C z − w0 z − w =0 z − w0

Note that each of the terms in the integrand of the last expression are
absolutely convergent, thus by Fubini’s Theorem, we can interchange the
summation and integral sign to get
∞  
1 f (z)
Z
f (w) = ∑ 2πi C ( z − w0 ) n +1
dz ( w − w0 ) n
n =0 | {z }
an
88 Lecture 14 Feb 12 2018 - Cauchy’s Integral Formula (Continued)

which is a power series centered at w0 with coefficient an .

å Note (Recall)
Consider the power series f (w) = ∑∞ n
n=0 an ( w − w0 ) . Recall Item 2 from
Section 12.2 that
f ( n ) ( w0 )
an =
n!

Applying this to 0 Proposition 16, we get

f ( n ) ( w0 ) 1 f (z)
Z
= dz
n! 2πi C ( z − w0 ) n +1

which holds for any w0 ∈ Ω by having C ⊆ Ω centered at w0 .

1 Theorem 17 (Cauchy’s Integral Formula 2)

Let Ω ⊆ C be open, f ∈ H (Ω). Then

1. ∀w ∈ Ω, f has a power series expansion at w.

2. f is differentiable infinitely many times in Ω.

3. ∀C ⊆ Ω that is a closed circle oriented anticlockwise, we have that

∀w ∈ C0 ,
n! f (z)
Z
f (n) ( w ) = dz (14.3)
2πi C ( z − w ) n +1

Remark
Item 3 is the actual Cauchy’s Integral Formula in the theorem.

Ò Proof
We have shown 1 from 0 Proposition 16 and 2 from 1 Theorem 7. It
remains to prove 3, which we shall prove by induction.

When n = 0, it is simply 1 Theorem 14. Suppose f has up to n − 1

complex derivatives and that

( n − 1) ! f (z)
Z
f ( n −1) ( w ) = dz.
2πi C (z − w)n
 PMATH352W18 - Complex Analysis 89

Consider h > 0, the difference of the quotient for f (n−1) is

f ( n −1) ( w − h ) − f ( n −1) ( w ) ( n − 1) !
 
1 1 1
Z
= f (z) − dz
h 2πi C h z−w−h z−w
(14.4)
Note that

An − Bn = ( A − B)( An−1 + An−2 B + . . . + ABn−2 + Bn−1 )

1 1 2
Let A = z−w−h , B = z−w , then the term in square brackets in Equa- 2
Key step
tion (14.4) becomes

h h i
An−1 + An−2 B + . . . + ABn−2 + Bn−1
(z − w − h)(z − w)

Thus as h → 0, we have

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

which completes the induction proof and proves 3. 

 Corollary 18 (Taylor Expansion of Entire Functions)

If f is an entire function, then ∀z0 ∈ C, we have

f 00 (z0 )
f (z) = f (z0 ) + f 0 (z0 )(z − z0 ) + ( z − z0 )2 + . . .
2!
which is a Taylor Expansion of f around z0 .

Ò Proof
By 0 Proposition 16, we have that

∞  
1 f (w)
Z
f (z) = ∑ n +1
dw (z − z0 )n
n =0
2πi
C ( w − z 0 )
 
1 f (w) 1 f (w)
Z Z
= dw + ( z − z0 ) (14.5)
2πi C w − z0 2πi C (w − z0 )2
 
1 f (w)
Z
+ dw ( z − z0 )2 + . . .
2πi C (w − z0 )3
 
1 f (w)
Z
+ dw ( z − z0 ) k + . . .
2πi C (w − z0 )k+1
90 Lecture 14 Feb 12 2018 - Cauchy’s Integral Formula (Continued)

Now by 1 Theorem 17, we have

0! f (w) 1 f (w)
Z Z
f ( z 0 ) = f (0) ( z 0 ) = dw = dw
2πi C w − z0 2πi C w − z0
1! f (w)
Z
f (1) ( z 0 ) = dw
2πi C (w − z0 )2
2! f (w)
Z
f (2) ( z 0 ) = dw
2πi C (w − z0 )3
..
.
k! f (w)
Z
f ( k ) ( z0 ) = dw
2πi C ( w − z 0 ) k +1
..
.

Thus Equation (14.5) becomes

f (2) ( z 0 ) f ( k ) ( z0 )
f (z) = f (z0 ) + f (1) (z0 )(z − z0 ) + ( z − z0 )2 + . . . + ( z − z0 ) k + . . .
2! k!
as required. 
 15 Lecture 15 Feb 14th 2018

15.1 Cauchy’s Integral Formula (Continued 1)

At this point, it is important that we provide the following definition:

 Definition 20 (Analytic Functions)

We say that f is analytic in Ω if f has a power series expansion at every

z ∈ Ω.

Remark
1. We have proven, in the previous lecture, that Holomorphicity =⇒
Analyticity

2. Should we have defined, in 1 Theorem 17, that the closed circle orients
clockwise, then we would have a negative equation for Equation (14.3).

15.1.1 Applications of Cauchy’s Integral Formula

Exercise 15.1.1

1. (Cauchy’s Inequality)1 Prove that ∀z0 ∈ C ∀ R > 0 ∈ R ∀ f ∈ H (C = 1

In a sense, this inequality implies that
as we take higher derivatives, the value
D (z0 , R)) of the derivatives become smaller.
n!
f ( n ) ( z0 ) ≤ · sup | f (z)|
R n z ∈C
92 Lecture 15 Feb 14th 2018 - Cauchy’s Integral Formula (Continued 1)

Ò Proof
From Equation (14.3), we have

n! f (z)
Z
f ( n ) ( z0 ) = dz
2πi C ( z − z 0 ) n +1

Parameterize C with γ : [0, 2π ] → C, where t 7→ z0 + Reit . Then

f (z0 + Reit )
Z 2π
n!
f ( n ) ( z0 ) = Rieit dt
2πi 0 ( Reit )n+1
n! 2π f (z0 + Reit )
Z
f ( n ) ( z0 ) ≤ dt ∵ Reit = R
2π 0 Rn
Z 2π
n!
≤ sup | f (z)| dt
2πRn z∈C 0
n!
= sup | f (z)|
R n z∈C

This completes the proof. 

2. (Liouville’s Theorem) A bounded entire function f : C → C is a

constant2 3 . 2
The theorem is not true in R, since
sin x is a bounded function differen-
tiable everywhere, but is not a constant.
3
The theorem also implies that
“trigonometry” in C is unbounded,
whatever the definition of “trigonome-
Ò Proof try” may be.

Since f is entire, we may take R, in Item 1, to be any large value.

Let M be the bound of f , i.e. ∃ M ∈ C, ∀z0 ∈ C, f (n) (z0 ) ≤
n!
Rn supz∈C | f (z)| = n!
Rn supz∈C M. Let n = 1, then | f 0 (z0 )| = M
R.
Thus we observe that R → ∞ =⇒ f (z0 ) → 0 for any z0 ∈ C. By
A2Q5(a), f is a constant.

3. (Parseval’s Theorem) Let Ω ⊆ C be open, f ∈ H (Ω), D (z0 , R) ⊆ Ω.

Then ∀z ∈ D (z0 , R), f (z) = ∑∞ n
n=0 cn ( z − z0 ) , which in turn implies
that 4 4
This is why the L2 -norm is perserved,
∞
as seen in AMATH231.
∀ z ∈ D ( z0 , R ) f (z0 + reiθ ) = ∑ cn (reiθ )n (†)
n =0
 PMATH352W18 - Complex Analysis 93

Consider (the L2 norm)

Z 2π 2
1
f (z0 + reiθ ) dθ
2π 0
Z 2π ∞ 2
1
=
2π 0
∑ cn (re iθ n
) dθ
n =0
∞ ∞
Z 2π
" #" #
1
=
2π 0
∑
cn r n inθ
e ∑ m −inθ
cm r e dθ
n =0 m =0
Z 2π ∞ ∞
1
=
2π 0
∑ ∑ cn cm r n+m ei(n−m)θ dθ
n =0 m =0

Since the series are absolutely convergent, use may use Fubini’s Theorem,
and thus

1 ∞ 2π Z
= ∑
2π n,m=0
cn cm r n+m
0
ei(n−m)θ dθ

 1 ∑∞

n,m=0 cn cm r
n+m 2π if n = m
2π
= 2π
n + m ei (n−m)θ
 1
 2π ∑∞
n,m=0 cn cm r i (n−m)
= 0 if n 6= m
0
∞
= ∑ |cn |2 r2n if n = m
n =0

Therefore, we have what is known as Parseval’s Identity:

Z 2π 2 ∞
1
2π 0
f (z0 + reiθ ) dθ = ∑ |cn |2 r2n (15.1)
n =0

Parseval’s Theorem states that:

L2 -norm of LHS in Equation (15.1) = L2 -norm of RHS of Equation (†)

Before going into the next application, please see Lemma 19.

4. (Maximum Modulus Principle) Let Ω ⊆ C be open and connected, and

f ∈ H (Ω). Then
sup | f (z)| = max | f (z)| .
z∈Ω z∈∂Ω

This implies that f cannot attain its maximum value in Ω0 .

Ò Proof
Suppose not, i.e. ∃z0 ∈ Ω0 , ∀z ∈ Ω such that | f (z0 )| = maxz∈Ω | f (z)| ≥
94 Lecture 15 Feb 14th 2018 - Cauchy’s Integral Formula (Continued 1)

| f (z)|

=⇒ ∃r > 0 D (z0 , r ) ⊆ Ω
∞
=⇒ ∀z ∈ D (z0 , r ) f (z) = ∑ c n ( z − z0 ) n
n =0

f (0) ( z 0 )
Note that c0 = 0! = f (z0 ). By Item 3,
∞ Z 2π 2
1
∑ |cn |2 r2n =
2π 0
f (z0 + reiθ ) dθ
n =0
∞ 2
1
=⇒ f (z0 )2 + ∑ |cn |2 r2n = 2π f (z0 + reiθ ) dθ
n =1
1
≤ | f (z0 )|2 (2π ) ∵ f (z0 ) = max f (z)
2π z∈Ω
∞
=⇒ f (z0 )2 + ∑ |cn |2 r2n ≤ | f (z0 )|2
n =1
∞
=⇒ ∑ |cn |2 r2n ≤ 0
n =1

=⇒ c1 , c2 , ... = 0
=⇒ f is a constant in D (z0 , r )
=⇒ f is a constant in Ω by Lemma 19

which is a contradiction. 

+ Lemma 19 (Principle of Analytic Continuation)

Let Ω ⊆ C be open and connected, and f ∈ H (Ω). Let Z ( f ) = { a ∈ Ω :

f ( a) = 0}. Then either

• Z ( f ) = Ω, i.e. ∀z ∈ Ω, f (z) = 0; or

• Z ( f ) has no limit point, i.e. points where f = 0 are isolated

This is a powerful result, since if we can find a small region for

where f is 0 in Ω, then f would be 0 in the entirety of Ω. If not, then
f is only 0 at isolated points, i.e. points where f = 0 are all apart
from each other.
 16 Lecture 16 Feb 16th 2018

16.1 Cauchy’s Integral Formula (Continued 3)

16.1.1 Applications of Cauchy’s Integral Formula (Continued)

Exercise 15.1.1 (Continued)

We shall restate the Item 4 in the following manner.

4. Maximum Modulus Principle (MMP) Let Ω ⊆ C, f ∈ H (Ω), Dz0 =

D (z0 , r ) ⊆ Ω. Then | f (z0 )| ≤ maxz∈∂Dz | f (z)| with
0

| f (z0 )| = max | f (z)| ⇐⇒ f is a constant on Ω

z∈∂Dz0

Remark
(a) This implies that for a non-constant analytic function f , ∀z ∈
Ω0 , f (z) 6= maxw∈Ω f (w).

(b) Since a global maximum is also a local maximum, we observe that for
any smaller region Ω0 ⊆ Ω, f cannot attain its maximum value for
any point in Ω00 . This is a stronger statement than the our previous
statement about the MMP.

Ò Proof
Suppose for Ethat f has a maximum in Ω0 , say at z0 . Hence ∃r >
0, Dz0 = D (z0 , r ) where

| f (z0 )| ≥ max | f (z)|

z ∈ Dz 0

On Dz0 , we have
∞
f (z) = ∑ c n ( z − z0 ) n (16.1)
n =0
96 Lecture 16 Feb 16th 2018 - Cauchy’s Integral Formula (Continued 3)

Note that c0 = f (z0 ). By Item 3, on Dz0 ,

∞ Z 2π 2
1
∑ |cn | 2 2n
r =
2π 0
f (z0 + reiθ ) dθ by Equation (15.1)
n =0
Z 2π
1
≤ | f (z0 )|2 dθ by Equation (16.1)
2π 0
= | f (z0 )|2 .

Then we have
∞
| c0 |2 + ∑ |cn |2 r2n = | f (z0 )|2
n =1
∞
| f (z0 )|2 + ∑ |cn |2 r2n = | f (z0 )|2
n =1
∞
∑ |cn |2 r2n = 0
n =1

which =⇒ c1 = c2 = . . . = 0. Thus ∀z ∈ Dz0 , f (z) ≡ c0

mod 2π. Then by Lemma 19, since f (z0 ) − c0 , as f (z) − c0 ) contains
D (z − 0, r ), we see that f (z) − c0 ≡ 0 in Ω, which implies the
equality of Item 4. 

5. Fundamental Theorem of Algebra (FTA) Any polynomial P(z) ∈

C[z] of degree greater than 1 has precisely n roots in C, given by
α1 , α2 , ..., αn . P(z) can be factored as P(z) = A(z − α1 ) . . . (z − αn )
for some A ∈ C.

Ò Proof
We may write P(z) = A(zn + an−1 zn−1 + . . . a1 z + a0 ), which then

P(z)  a n +1 a1 a0 
= A 1 + + . . . +
zn z z n −1 zn
which then, by the Reverse Triangle Inequality,
 
P(z) | a n −1 | | a1 | | a0 |
=⇒ ≥ | A | 1 − − . . . − − (16.2)
zn |z| | z n −1 | | z n |

P(z)
So as |z| → ∞, → | A|, from Equation (16.2). Since |z| →
zn
∞, ∃ R > 0, ∀ |z| > R, then ∀θ ∈ [0, 2π ],

| A| n | A| n
P( Reiθ ) = | P(z)| ≥ |z| ≥ R
2 2
 PMATH352W18 - Complex Analysis 97

Taking R to be even larger if necessary, we can get

P( Reiθ ) ≥ | P(0)| (†)

Suppose, for contradiction, P(z) has no root in C. Then g(z) = 1

P(z)
is an entire function. By Equation (†), we have that g( Reiθ ) ≤
| g(0)| for all θ ∈ [0, 2π ]. But this contradicts Item 4 unless if g(z)
is constant on C, which in turn implies that P is a constant, but that
contradicts that P has degree greater than 1.

∴ P(z) has to have a zero in C, say α1 . This implies that

P(z) = A(z − α1 ) P1 (z)

where P1 (z) ∈ C[z]. By repeatedly taking the above steps, inductively

so, for P1 , P2 , ..., the proof is completed. 
17 Lecture 17 Feb 26th 2018

17.1 Analytic Continuity

We shall restate the important lemma that we have been using in the
last two lectures, and proceed to prove this lemma.

+ Lemma 20 (Principle of Analytic Continuity)

Let Ω ⊆ C be open and connected, and f ∈ H (Ω). Let Z ( f ) = { ainΩ :

f ( a) = 0}. Then either

• Z ( f ) = Ω, i.e. ∀z ∈ Ω, f (z) = 0; or

• Z ( f ) has no limit point, i.e. points where f = 0 are isolated

Ò Proof
Let z0 ∈ Z ( f )∗ .

Step 1: Show that z0 ∈ Z ( f )0 , i.e. f is identically 0 on some

D (z0 , r ) ⊆ Ω for r > 0.

On D (z0 , r ), f (z) = ∑∞ n
n=0 cn ( z − z0 ) . Suppose f is not identically
0 on D (z0 , r ). Then ∃m ∈ N, cm 6= 0, ∀ j < m, c j = 0, i.e. f (z) =
cm (z − z0 )m + cm+1 (z − z0 )m+1 + . . ..

Define, in Ω,

f (z)

( z − z0 ) m
z ∈ Ω \ { z0 }
g(z) =
c
m z = z0
100 Lecture 17 Feb 26th 2018 - Analytic Continuity

Clearly, g ∈ H (Ω \ {z0 }). But on D (z0 , r ),

g ( z ) = c m + c m +1 ( z − z 0 ) + c m +2 ( z − z 0 )2 + . . .

which implies g ∈ H (Ω). Now g(z0 ) = cm 6= 0, so there exists a

neighbourhood Uz0 of z0 , such that g 6= 0 on Uz0 .

∀ a 6= z0 ∈ Z ( f ), we have that g( a) = 0 by defintion of Z ( f ), which

implies that a ∈/ Uz0 , which contradicts that z0 ∈ Z ( f )∗ . This implies
f ≡ 0 in D (z0 , r ).

Step 2: Z ( f )0 is both open and closed.

Note that
n o
Z ( f )0 : = a ∈ Z ( f ) : ∃r > 0, D ( a, r ) ⊆ Z ( f )

is open by definition.

WTP [ Z ( f )0 ]∗ ⊆ [ Z ( f )]∗ .

From Step 1, we know that [ Z ( f )0 ]∗ ⊆ Z ( f )0 . Thus Z ( f )0 contains

its limit points and is hence closed by definition.

Step 3: Z ( f ) = ∅ or Ω.

Ω is connected
 c
=⇒ Ω = Z ( f )0 t Z ( f )0
 c
=⇒ Z ( f )0 is open and closed by Step 2

A connected set cannot be expressed as a disjoint union of non-trivial

open sets. Therefore, either Z ( f )0 = ∅ or Z ( f )0 = Ω.

Z ( f )0 = ∅ =⇒ Z ( f )∗ = ∅ by Step 1 =⇒ Z ( f ) = ∅
Z ( f )0 = Ω =⇒ Z ( f ) = Ω by Step 1

 Corollary 21 (Uniqueness of a Function)

Let Ω ⊆ C be open and connected. ∀ f , g ∈ H (Ω) with f (z) = g(z) for

z ∈ Ω1 ⊆ Ω where Ω1 has limit points. Then ∀z ∈ Ω, f (z) = g(z).
 PMATH352W18 - Complex Analysis 101

Ò Proof
Apply Lemma 19 to the function f − g.

Remark
1. In C, we cannot have two functions sharing a region of points in their
images. (But this is possible in R)

2. Suppose f ∈ H (Ω), Ω ∈ C is open and connected, F ∈ H (Ω0 ) with

Ω ⊆ Ω0 . If f , F agree on Ω, then F is called an analytic continuation
of f in Ω0 (i.e. F ‘extends’ f in Ω0 ). Lemma 19 states that F is uniquely
determined by f , i.e. there is a unique way to analytically ‘continue’ f .

17.2 Morera’s Theorem

Remark (Recall)
From Cauchy’s Theorem, we know that ∀ f ∈ H (Ω) =⇒ ∀γ ∈ Ω
R
f =
0. We used Goursat’s Theorem, i.e. ∀∆ ∈ Ω ∆ f = 0 to proof this, and
R

in the process we constructed an antiderivative. Now, our question is, is the

converse of the said Cauchy’s Theorem true?

Unfortunately for us, that is not true (example needed). But a “par-
tial” converse exists.

1 Theorem 22 (Morera’s Theorem)

Let f be continuous on Ω ⊆ C, which is an open set, and ∀∆ ∈

Ω, ∆ f = 0, where ∆ is a triangular path. Then f ∈ H (Ω).
R

Ò Proof
Use the same construction as in Cauchy’s Theorem for Convex Sets to get
an antiderivative F for f , where F ∈ H (Ω), i.e.
Z
F (z) := f (z) dz
[ a,z]

Then F 0 (z) = f (z), which in turn implies that f ∈ H (Ω) since F is

C-differentiable on Ω by 1 Theorem 17.
102 Lecture 17 Feb 26th 2018 - Morera’s Theorem
18 Lecture 18 Feb 28th 2018

18.1 Winding Numbers

Recall Cauchy’s Integral Formula. We claimed that


1 w ∈ C0
IndC (w) =
0 w∈
/C

We will now formally define this index.

 Definition 21 (Winding Numbers)

Let γ : [ a, b] → C be a closed and oriented anti-clockwise, and γ∗ be the

image of γ in C. Let Ω = C \ γ∗ . ∀w ∈ Ω, define the index of w with
respect to γ as
1 dz
Z

2πi γ z−w
in which shall be called the winding number of γ around w.

1 Theorem 23 (Winding Number Theorem)

We shall use notation as the definition above. Indγ (w) is

1. always an integer;

2. constant on any connected component of Ω; and

3. zero on the unbounded component of Ω.

104 Lecture 18 Feb 28th 2018 - Winding Numbers

å Note
γ is compact in C (since it creates a ring from [ a, b] under γ). So for some
disc D, γ∗ ⊆ D. Let Ω ⊃ C \ D, where we note that the contained set
is connected and unbounded. Then Ω contains one unbounded compo-
nent, while other components of Ω are inside D. Therefore, we know that
components in D are bounded.

Ò Proof
1. By definition,

1 dz
Z
Indγ (w) =
2πi γ z−w
Z b 0
1 γ (t) dt
=
2πi a γ(t) − w
Rb γ0 (t) dt
WTS Indγ (w) ∈ Z ≡ a γ(t)−w ∈ 2πiZ.

Note that z ∈ 2πiZ ⇐⇒ ez = 1. Thus it suffices to show that

Rbγ0 (t) dt
a γ(t)−w
e =1

u γ0 (t) dt
R 
Idea: Think of exp a γ(t)−w as a function of u, call it φ(u).
Then we just need to show that φ(b) = 1. We know that φ( a) =
Ra

exp a . . . = 1. This motivates us to find the derivative of φ.

Define φ accordingly, and then since (e f (u) )0 = e f (u) · f 0 (u),

d u γ0 ( t ) dt
Z
φ0 (u) = φ(u) ·
du a γ(t) − w
φ0 (u) γ0 (u)
by FTC =⇒ =
φ(u) γ(u) − w
=⇒ φ0 (u) (γ(u) − w) − γ0 (u)φ(u) = 0
 
d φ(u)
=⇒ = 0 by quotient rule
du γ(u) − w
φ(b) φ( a) φ(u)
=⇒ = since is a constant function of u
γ(b) − w γ( a) − w γ(u) − w
=⇒ φ(b) = φ( a) = 1 ∵ γ is closed.
 PMATH352W18 - Complex Analysis 105

We will prove that Indγ (w) is continuous.

∀w ∈ Ω ∀z ∈ γ∗ ∃ M > 0 |w − z| > M
M2 πε
∀ε > 0 ∃δ = R > 0 ∀ w0 ∈ Ω
γ dz
M
| w − w0 | < δ ∧ | w0 − z | >
2
then
1 dz 1 dz
Z Z
|Indγ (w) − Indγ (w0 )| = −
2πi γ z − w 2πi γ z − w0
1 w − w0
Z
= dz
2π γ (z − w)(z − w0 )
1 w − w0
Z
≤ dz
2π γ (z − w)(z − w0 )
1 2
Z
< δ dz
2π γ M · M
1
Z
= δ dz = ε
M2 π γ

2. Also Indγ (w) takes only integer values, thus it must be constant on
each open connected component1 (why?). 1
We may invoke Lemma 19 but it is,
to an extent, unnecessary for such a
powerful statement.
3. Note that
Z b 0
1 γ (t) dt
|Indγ (w)| =
2π a γ(t) − w
Let w be in the unbounded component in the complement of γ such
that |w| → ∞. Then ∀t ∈ [ a, b], ∃ M > 0 such that

1 1
≤
|γ(t) − w| M

which implies that

Z b
1 1
|Indγ (w)| ≤ · γ0 (t) dt
2π M a
| {z }
is a fixed constant as
γ is a fixed path
=⇒ (|w| → ∞ =⇒ M → ∞ =⇒ |Indγ (w)| → 0)

Then by parts 1 and 2, the proof is completed. 

106 Lecture 18 Feb 28th 2018 - Winding Numbers

Remark
Note that by 2, we have that ∀w ∈ C0 ,

Rieiθ
Z 2π
1 dz 1 dz 1
Z Z
= = dθ = 1
2πi C z−w 2πi C z − z0 2πi 0 Reiθ

where z0 is the center of the circle path C.

19 Lecture 19 Mar 2nd 2018

19.1 Singularities

Exercise 19.1.1

Let C : [0, 2π ] → C such that ∀t ∈ [0, 2π ], t → eit . Suppose f ∈ H (Ω),

then by Cauchy Z
f (z) dz = 0
C

Let f (z) = 1z , then 1

R
C z dz = 2πi IndC (0) = 2πi when it is “supposed”
/ H (Ω). In fact, f is
to be 0 by the argument above. Then in this case, f ∈
undefined at 0.

The example above introduces us to the study of such exceptional

points.

 Definition 22 ((Isolated) Singularity)

∀ a ∈ C, ∃r > 0, ∃ D = D ( a, r ).

f ∈ H ( D \ { a}) ∧ f ( a) is undefined ⇐⇒

f has a(n) point/isolated singularity at z = a.

Example 19.1.1
e z −1
1. Given f ∈ H (C \ {0}), define f (z) = z . Clearly, z is a singularity.
Consider the function − 1) ∈ H (C). Then we have that the function
(ez
has a power series expansion around z = 0. So ∀z ∈ C,

z2 z3
ez − 1 = z + + +...
2! 3!
108 Lecture 19 Mar 2nd 2018 - Singularities

And for z 6= 0, we have

ez − 1 z z2
= 1+ + +... (19.1)
z 2! 3!
This motivates us to define

 e z −1 z ∈ C \ {0}
z
g(z) =
1 z=0

Clearly then g ∈ H (C), where in C \ {0} its holomorphicity is given

by f , and in a neighbourhood of 0, from Equation (19.1). Therefore, y
assigning f the value of 1 at z = 0, we can make f “entire”.

We call such a point z as a removable singularity for f .

2. Given f ∈ H (C \ {0}), define f (z) = 1

z. Is the singularity at 0
removable?

Suppose ∃ g ∈ H (C) such that

∀ z ∈ C \ {0} g(z) = f (z) (19.2)

∴ ∃r > 0 ∀z ∈ D (0, r )

g ( z ) = c0 + c1 z + c2 z2 + . . . (19.3)

Consider the function zg(z). By Equation (19.2),

∀z ∈ C \ {0} zg(z) = 1

By Equation (19.3), z = 0 =⇒ zg(z) = 0. But this cannot happen

since zg(z) ∈ H (C) (if we pick an open ball of, say, 1
2 around 0,
then there are no points in the entirety of C that is close to 0).
Therefore z = 0 is not a removable singularity for f .

 Definition 23 (Removable Singularity, Pole, Essential Singu-

larity)

Let f have a singularity at z0 ∈ C.

1. ∃r > 0 ∀z ∈ D = D (z0 , r ) ∃ g(z) ∈ H ( D ) ∀z ∈ D \ {z0 } g(z) =

f (z)
=⇒ f has a removable singularity at z0 1 . 1
For the laymen, "the value of f at z0
can be corrected or defined to make it
holomorphic in its designated region."
 PMATH352W18 - Complex Analysis 109

2. ∃r > 0 ∀z ∈ D = D ∗ (z0 , r ) ∃ A, B ∈ H ( D ) A(z0 ) 6= 0 ∧ B(z0 ) =

A(z)
0 f (z) = B(z)
=⇒ f has a pole at z0 (a non-removable singularity)2 2
For the laymen, "the singularity of f
comes from a zero of its denominator."
3. f has a singularity at z0 which is neither removeable nor a pole =⇒
f has an essential singularity at z0 .

Example 19.1.2

To show an example of an essential singularity, consider the function

1
f (z) = e z . If we attempt to do a “Taylor expansion” on the function (which
is invalid at z = 0), we have
1 1 1
f (z) = 1 + + 2
+ +...
z 2!z 3!z3
The point 0 for f is said to be a “pole of infinite order” (this shall be defined
later on)

While removable singularities are nice to have, they are not as

interesting to us. On the other hand, we are more interested in their
non-removable counterpart, the poles. This motivates the study of
zeros of holomorphic functions.

1 Theorem 24 (Theorem 9)

Let Ω ⊆ C be open and connected. Suppose that f ∈ H (Ω) with f 6≡ 0

on Ω and that f has a zero at z0 ∈ Ω. Then

∃r > 0 ∀z ∈ D = D (z0 , r ) ∃ g ∈ H ( D ) g(z0 ) 6= 0 ∃!n ∈ N

f ( z ) = ( z − z0 ) n · g ( z ) (19.4)

Ò Proof
By Analytic Continuation, zeros of f are isolated since f 6≡ 0. So ∃r > 0
such that ∃ D = D (z0 , r ), in which ∀z ∈ D \ {z0 }, f (z) 6= 0.

Since f ∈ H (Ω), ∀z ∈ D,
∞
f (z) = ∑ c k ( z − z0 ) k
k =0
110 Lecture 19 Mar 2nd 2018 - Singularities

As f 6≡ 0 in D, ∃n ∈ N \ {0} that is the smallest such that cn 6= 03 . 3

n 6= 0 since we have f (z0 ) = 0 which
implies c0 = 0.

∴ f ( z ) = c n ( z − z 0 ) n + c n +1 ( z − z 0 ) n +1 + . . .
= ( z − z 0 ) n [ c n + c n +1 ( z − z 0 ) + . . . ]
| {z }
call this g(z)

Note that g(z0 ) 6= 0 since cn 6= 0. Thus g(z) ∈ H ( D ) since it has the

same radius of convergence as f .

To prove uniqueness, suppose that we may write

∞
f (z) = ∑ ( z − z0 ) n · g ( z ) = ( z − z0 ) m · h ( z )
k =0

for some m ∈ N and that h(z) 6= 0. If m > n, dividing both sides by

( z − z0 ) n ,
g ( z ) = ( z − z0 ) m − n h ( z ).

As z → z0 , we would have g(z0 ) = 0, which is a contradiction. If m < n,

we can perform a similar argument and have that as z → z0 , h(z0 ) = 0,
also a contradiction. Therefore, m = n and h = g. 

We say that f has a zero of order n at z0 if Equation (19.4) holds.

20 Lecture 20 Mar 5th 2018

20.1 Singularity (Continued)

Recall the definition of a removable singularity from  Definition 23.

1 Theorem 25 (Theorem 10)

If f ∈ H (Ω \ {z0 }) has an isolated singularity at z0 and limz→z0 (z −

z0 ) f (z) = 0, then the singularity at z0 is removable.

Ò Proof
Since f (z0 ) is undefined, set

 ( z − z )2 f ( z )
0 ∀ z ∈ Ω \ { z0 }
h(z) =
0 z = z0

Clearly h ∈ H (Ω \ {z0 }). At z0 ,

h ( z ) − h ( z0 ) ( z − z0 )2 f ( z ) 1
lim = lim
z → z0 z − z0 z → z0 z − z0
= 0 by assumption
1
Goes to show that the definition of h is
no foresight.
∴ h0 (z0 ) exists and equals 0. Clearly then that h ∈ H (Ω). So ∃r > 0
such that ∃ D = D (z0 , r ), so that ∀z ∈ D,

h ( z ) = c0 + c1 ( z − z0 ) + c2 ( z − z0 )2 + . . .

But c0 = h(z0 ) = 0 and c1 = h0 (z0 ) = 0. Thus the power series can be

112 Lecture 20 Mar 5th 2018 - Singularity (Continued)

written as

h ( z ) = c2 ( z − z0 )2 + c3 ( z − z0 )3 + . . .
= ( z − z0 )2 [ c2 + c3 ( z − z0 ) + . . . ]

Hence by the definition of h, ∀z ∈ Ω \ {z0 }, f (z) = c2 + c3 (z − z0 ) +

. . .. Therefore, by redefining f (z0 ) = c2 , we see that the singularity at z0
is removable.

We may also complete the proof by defining a function g as, ∀z ∈ Ω,


 f (z) z 6 = z0
g(z) =
c
2 z = z0

Recall 1 Theorem 24

Let Ω ⊆ C be open and connected, and f ∈ H (Ω) where ∀z ∈

Ω, f (z) 6= 0.

f (z0 ) = 0 =⇒

∃r > 0 ∃ D = D (z0 , r ) ∀z ∈ D ∃!n ∈ N

∃!g ∈ H ( D ) g(z0 ) 6= 0
f ( z ) = ( z − z0 ) n g ( z )

 Definition 24 (Zero of Order n & Simple Zero)

By the above setting, we say that f has a zero of order n at z0 .2 2

In laymen terms, "Rate at which the
function vanishes at z0 . The greater n is,
If n = 1, we say that z0 is a simple zero. the greater the rate."

Recall definition of a pole from  Definition 23

Suppose f has an isolated singularity at z0 , and that there exists a

neighbourhood D around z0 where A, B ∈ H ( D ), in which A and B
are defined such that ∀z 6= z0 ∈ D, A(z0 ) 6= 0 ∧ B(z0 ) = 0, so that we
A(z)
can let f (z) = B(z)
. Then f has a pole at z0 .
 PMATH352W18 - Complex Analysis 113

1 Theorem 26 (Theorem 9.1)

Stein & Shakarchi - Complex Analysis (pg. 74)

If f has a pole at z0 ∈ Ω, then in a neighbourhood of that point there

exists a non-vanishing holomorphic function h and a unique positive
integer n such that
f ( z ) = ( z − z0 ) − n h ( z )

Ò Proof
1
By 1 Theorem 24, we have = (z − z0 )n g(z), where g is holomor-
f (z)
phic and non-vanishing in a neighbourhood of z0 , so the result follows
with h(z) = g(1z) . 

 Definition 25 (Pole of order n & Simple Pole)

With the above setting, we say that f has a pole of order n at z0 if the
function B has a zero of order n3 3
In laymen terms, "Rate at which f
‘grows’ near z0 ."
If n = 1, then z0 is a simple pole.

1 Theorem 27 (Theorem 11)

Let f have a pole of order n at z0 . Then ∃r > 0, ∃ D = D (z0 , r ), such

that ∀z ∈ D \ {z0 },

c−n c−(n−1) c
f (z) = n
+ n − 1
+ . . . + −1 + G ( z )
( z − z0 ) ( z − z0 ) z − z0

for some G ∈ H ( D ).

Ò Proof
By 1 Theorem 26, write the holomorphic function h as h(z) = a0 +
a1 (z − z0 ) + a2 (z − z0 )2 + . . ., then

1 h
2
i
f (z) = a 0 + a 1 ( z − z 0 ) + a 2 ( z − z 0 ) + . . . .
( z − z0 ) n
114 Lecture 20 Mar 5th 2018 - Singularity (Continued)

The proof is complete by expanding the equation. 

 Definition 26 (Principal Part)

c− j
In 1 Theorem 27, the sum ∑nj=1 = ( z − z0 ) j
is called the principal part
of f at the pole z0 .

 Definition 27 (Residue)

In 1 Theorem 27, the coefficient c−1 is called the residue of f at the pole
z0 , denoted Res f (z).
z = z0

The residue shall be more carefully studied later on.

21 Lecture 21 Mar 7th 2018

21.1 Singularity (Continued 2)

1 Theorem 28 (Casorati-Weierstrass)

Let z0 ∈ Ω and f ∈ H (Ω \ {z0 }). Suppose f has a singularity at z0 .

Then one of the following occurs:

1. f is a removable singularity at z0 ;

2. ∃m ∈ N, {c j }m m −
j=1 ⊆ C, f ( z ) − ∑ j=1 c j ( z − z0 ) j has a removable
singularity at z0 ; or

3. ∀r > 0, B(z0 , r ) ⊆ Ω such that f ( B0 (z0 , r )) is dense in C.1 1

B0 (z0 , r ) is the punctured ball.

Ò Proof
Suppose 3. does not hold, i.e. f ( B0 (z0 , r )) is not dense in C for some
r > 0. Then ∃w ∈ C, ∃δ > 0, such that

f B0 (z0 , r ) ∩ B(w, δ) = ∅

=⇒ ∀z ∈ B0 (z0 , r ) | f (z) − w| > δ

1
Consider g(z) = f (z)−w
for z ∈ B0 (z0 , r ), in which g ∈ H ( B0 (z0 , r )). Plausible since f (z) − w 6= 0.
1
Then | g(z)| ≤ δ for all z ∈ B0 ( z 0 , r ), which implies that

lim (z − z0 ) g(z) = 0
z → z0
Squeeze Theorem
By 1 Theorem 25, g has a removable singularity at z0 , thus we can
extend the function to a function g̃ ∈ H ( B(z0 , r )). From here, we try to
construct a function that extends on f onto the singularity z0 , say, f˜. The
116 Lecture 21 Mar 7th 2018 - Singularity (Continued 2)

1
construction of g̃ satisfies the equation g̃(z)
+ w = f (z) except, possibly,
at z0 .

Case 1: Suppose g̃(z0 ) 6= 0.

We can simply define


1  f (z) − w z ∈ B0 ( z0 , r )
=
g̃(z)  1 z = z0
g̃(z0 )

1
Clearly then g̃ ∈ H ( B0 (z0 , r )). At z0 ,

1
g̃(z0 ) 6= 0 =⇒ ∃r1 > 0 D = D (z0 , r1 ) ∀z ∈ D g̃(z) 6= 0 =⇒ ∈ H (D)
g̃

1
Therefore, g̃ ∈ H ( B(z0 , r ))2 2 1
is the inverse of a non-zero holo-
g̃
morphic function.
1
Since ∀z ∈ B0 (z0 , r ) f (z) = g̃(z)
+ w by construction, we may define

1
f˜(z) = +w
g̃(z)

1
such that f (z0 ) is defined as g̃(z0 )
+ w. By this construction, 1. holds.

Case 2: g̃(z0 ) = 0.

∵ g̃ ∈ H ( B0 (z0 , r )) ∧ g̃(z0 ) = 0 ∧ (∀z ∈ B0 (z0 , r ) g̃(z0 ) 6= 0)

By 1 Theorem 24,

∃!m ∈ N ∃0 < r1 < r ∃ g1 ∈ H B(z0 , r1 )

∀z ∈ B(z0 , r1 ) g̃(z) = (z − z0 )m g1 (z) (21.1)

∵ g1 ∈ H B(z0 , r1 ) ∧ g(z0 ) 6= 0, we can repeat the argument as in
Case 1 for g1 to get that g1 ∈ H B(z0 , r1 ) , which implies


1

1
∀ z ∈ B ( z0 , r1 ) = a0 + a1 ( z − z0 ) + . . .
g1 ( z )

By construction given by Equation (21.1), ∀z ∈ B0 (z0 , r1 )

1
f (z) − w =
g1 (z)(z − z0 )m
a0 a1 a
= + + . . . + m −1 + a m + a m +1 ( z − z 0 ) + . . .
( z − z 0 ) m ( z − z 0 ) m −1 z − z0
 PMATH352W18 - Complex Analysis 117

∴ ∀ z ∈ B0 ( z0 , r1 ),
a0 a
f (z) − − . . . − m −1 = w + a m + a m +1 ( z − z 0 ) . . .
( z − z0 ) m z − z0

Thus we may define an “extended” function of f , f˜ to be w + am at the

singularity z0 . Therefore, f (z) − ∑m − j has a removable
j =1 a j ( z − z 0 )
singularity at z0 , i.e. 2. holds. 
22 Lecture 22 Mar 9th 2018

22.1 Singularity (Continued 3)

 Corollary 29

If f has an essential singularity at z0 and is holomorphic in some B0 (z0 , r )

where r > 0, then f ( B0 (z0 , r )) is dense in C.

Ò Proof
Suppose not, i.e. 3. of 1 Theorem 28 does not hold. Then either 1., which
implies that z0 is removable, or 2., which implies that z0 is a pole, is true.
This contradicts the assumption that z0 is an essential singularity. 

Remark
There are a lot more that are actually true from 1 Theorem 28! Picard
showed that in any such punctured ball B0 (z0 , r ) around the essential sin-
gularity z0 , f takes on every complex value (except possibly one value)
infinitely often.

22.2 The Residue Theorem

å Note (Recall)
If f has a pole at z0 , f ∈ H (Ω \ {z0 }), then in some open neighbourhood
120 Lecture 22 Mar 9th 2018 - The Residue Theorem

D of z0 , we can write ∀z ∈ D \ {z0 }

c−k c −1
f (z) = +...+ + c + c1 ( z − z0 ) + . . . (22.1)
( z − z0 ) k (z − z0 ) |0 {z }
| {z } G (z)
Principal Part

with G ∈ H ( D ).

1 Theorem 30 (Cauchy’s Residue Theorem)

Let Ω ⊆ C be open, f ∈ H (Ω \ {z0 }) where z0 ∈ Ω is a pole. If γ is a

closed path in Ω \ {z0 } such that ∀w 6∈ Ω, Indγ (w) = 0. Then
 
1
Z
f (z) d(z) = Res f (z) Indγ (z0 )
2πi γ z = z0

1 1
R
where Indγ (z0 ) := 2πi γ z − z0 dz.

Ò Proof
Using notation of Equation (21.1), define g(z) such that

 f (z) − ∑k c− j
j =1 ( z − z0 ) j z ∈ Ω \ { z0 }
g(z) :=
c
0 z = z0

Clearly, g ∈ H (Ω \ {z0 }), since f (z) minus finitely many polynomials

with non-zero denominators is still a holomorphic function. At z0 , with a
neighbourhood D around the point, we have, from Equation (21.1),

g ( z ) = c0 + c1 ( z − z0 ) + . . .

which g(z0 ) agrees with c0 and for any point z ∈ D \ {z0 }, by definition
of g using Equation (21.1). This implies that g ∈ H ( D ) =⇒ g ∈
H ( Ω ).

Thus, by Cauchy’s Theorem,

Z
g(z) dz = 0
γ
 PMATH352W18 - Complex Analysis 121

Then ∀z ∈ γ and since z0 6∈ γ, we get

Z Z k c− j
f (z) dz = ∑ (z − z j
dz
γ γj =1 0)

Consider each term of RHS in turn. Note that for m ≥ 2, since

−1 is the antiderivative of 1
,
(m−1)(z−z0 )m−1 ( z − z0 ) m

1
Z


m
dz = F γ(b) − F γ( a) by FTC
γ ( z − z0 )
= 0 since γ is closed.

If m = 1, then

1 1
Z
dz = Indγ (z0 ) by definition
2πi γ z − z0

1
Z
∴ f (z) dz = c−1 Indγ (z0 )
2πi γ
 
= Res f (z) Indγ (z0 )
z = z0

 Definition 28 (Meromorphic Functions)

A function f is said to be meromorphic on Ω if ∃A ⊆ Ω such that

1. A ∗ = ∅

2. f ∈ H (Ω \ A )

3. ∀z ∈ A f has a pole of finite order on z.

Remark
Holomorphicity ⊆ Meromorphicity (let A = ∅)
23 Lecture 23 Mar 12th 2018

23.1 The Residue Theorem (Continued)

We can generalize 1 Theorem 30 for when there are more than one
pole.

1 Theorem 31 (Cauchy’s Residue Theorem - Generalized)

Let Ω ⊆ C be open, f be meromorphic on Ω, A be a set of poles. If γ is a

closed path in Ω \ A such that ∀w 6∈ Ω Indγ (w) = 0, then

1
Z
∑


f (z) dz = Res f (z) Indγ ( a) (23.1)
2πi γ a ∈A
z= a
| {z }
this is a finite sum

Ò Proof
We will first need to prove that A has only finitely many points.

Since f is meromorphic, the set of poles, A , must not contain its limit
points by definition of meromorphisms, i.e. each of the poles are isolated
singularities and A ∗ = ∅. Let A 0 := { a ∈ A : Indγ ( a) 6= 0}.
Suppose, for contradiction, that A 0 has infinitely many points. Since
the union of γ and its interior is compact, by Bolzano-Weierstrass, any
infinite subset (subsequence, to be exact, but we may simply index the
points in the subset) must converge to a point in the union of γ and its
interior, call it w. Since A 0 is a subset (subsequence) of the union of γ
and its interior, A 0 converges to w. Now since | f (z)| → ∞ and z → a
for all a ∈ A 0 , and since f is continuous since it is meromorphic and
hence holomorphic, we have that | f (z)| → ∞ as z → w. Hence w is a
pole in the interior of γ and is hence in A 0 . In other words, A 0 contains
124 Lecture 23 Mar 12th 2018 - The Residue Theorem (Continued)

its limit points, which contradicts the fact that A 0 ⊆ A does not contain
its limit points. Therefore, there are only finitely many points in A 0 .

It remains to prove Equation (23.1).

Since A is finite, let |A | = n for some n ∈ N. For any a, b ∈ N,

∀z a , zb ∈ A , ∃r a , rb > 0 such that zb 6∈ D (z a , r a ) ∧ z a 6∈ D (zb , rb ).
Therefore, ∀l ∈ N, ∀zl ∈ A , we can write, ∀z ∈ D (zl , rl ), that
cl,−kl c
f (z) = + . . . + l,−1 + cl,0 + cl,1 (z − zl ) + . . . , (23.2)
(z − zl )k l z − zl

where k l is the order of the pole zl . Since we are only concerned with the
poles in the interior of γ, let A 0 := { a ∈ A : Indγ ( a) = 1} =
{z1 , z2 , ..., zn }.

Consider the function

n kl

cl,− j
− ∑ ∑ (z − z ) j ∀z ∈ Ω \ A 0



 f ( z )
l

 l =1 j =1
g(z) = n km cm,− j
cl,0 − ∑ ∑ z∈A0


j

m =1 j =1 ( z − z m )



m6=l

At the neighbourhood of each zl ∈ A 0 save zl itself, that is in

D (zl , rl ) \ {zl } where D (zl , rl ) is as defined above, we have that

kl cl,− j n km cm,− j
g(z) = f (z) − ∑ (z − z ) j − ∑ ∑ (z − zm ) j (23.3)
j =1 l m =1 j =1
| {z } m6=l
Az,l | {z }
Bz,l

in which we observe that Az,l is holomorphic by Casorati-Weierstrass, and

Bz,l is holomorphic in D (zl , rl ) since ∀z ∈ D (zl , rl ) \ {zl }, z 6∈ A 0


and hence the denominators cannot be zero. Thus g ∈ H D (zl , rl ) .
At zl , g(zl ) agrees with Equation (23.3) by Equation (23.2). Thus g ∈


H D (zl , rl ) .

Then by Cauchy’s Theorem,

Z
g(z) dz = 0.
γ

Then, ∀z ∈ γ∗ (image of γ), since A ∩ γ∗ = ∅ (since γ∗ ∈ Ω \ A ), we

get
Z Z n kl cl,− j
f (z) dz = ∑ ∑ (z − z j
dz
γ γl =1 j =1 0 ))
 PMATH352W18 - Complex Analysis 125

k cl,− j
For each of the terms in ∑ j=
l
1 (z−zl ) j
, for j ≥ 2, we have that the
1 1
antiderivative of (z−zl ) j
is − ( j−1)(z−z ) j−1 . Thus by the Fundamental
0
Theorem of Calculus, since we are integrating over the closed path γ, the
1
R 1
integral of these terms are 0. When j = 1, since 2πi γ z−z = Indγ ( zl ), l
we have that
n
1
Z

2πi γ
f (z) dz = ∑ cl,−1 Indγ (zl )
l =1
n  
= ∑ Res f (z) Indγ (zl )
z=zl
l =1

as required.

Another proof that we can be shown is as follows:

Using similar notations as above, on each of the neighbourhood of the

points in A , we define a keyhole-like path around the points, with a circu-
lar path γl , oriented in the opposite direction of γ, of radius rl ≥ ε l > 0
centered around each zl , connected to the path γ through a corridor of
width δl . This construction is similar to the proof of Cauchy’s Integral
Formula in our lectures, where we extend the path of γ inwards so as to
avoid the poles. As each δl → 0, we have
Z n Z n Z ∞

γ
f (z) dz = ∑ f (z) dz = ∑ ∑ cl,j (z − zl ) j dz
l = 1 γl l =1 γl j=−k l

Now for −k l ≤ j ≤ 2, the term cl,j (z − z0 ) j have antiderivatives as

described above, and are thus 0 by FTC, on the path γl . When j ≥ 0,
∑∞ j
j=0 cl,j ( z − zl ) is a holomorphic function by construction, and hence
their integrals are 0 by Cauchy’s Theorem. We are, therefore, left with
Z n Z cl,−1
γ
f (z) = ∑ z − zl
dz
l = 1 γl

which in turn implies

n
1
Z  
2πi γ
f (z) dz = ∑ Res f (z) Indγl (zl )
z=zl
l =1

This implies the desired result (through a similar proof given in 1 Theorem 30).


Remark
We need Ω to be connected with the interior of γ contained in Ω, i.e. Ω is
 126 Lecture 23 Mar 12th 2018 - Applications of Cauchy’s Residue Theorem

simply connected.

Now all the above begs the question: how exactly do we find the
residue of a pole?

Suppose that f has a pole of order k at z0 . Then in some neigh-

bourhood D of z0 , we have the Laurent expansion
a−k a
f (z) = + . . . + −1 + a 0 + a 1 ( z − z 0 ) + . . .
( z − z0 ) k z − z0

which implies

f (z)(z − z0 )k = a−k + a−k+1 (z − z0 ) + . . . + a−1 (z − z0 )k−1 + . . .

So a−1 is the (k − 1)th coefficient for f (z)(z − z0 )k , i.e. we can get

1 d k −1
Res f (z) = a−1 = lim f (z)(z − z0 )k
z = z0 z → z0 (k − 1)! dzk−1

23.2 Applications of Cauchy’s Residue Theorem

Exercise 23.2.1
R∞ 1
Evaluate − ∞ 1+ x 4 dx.

The typical approach (from a complex analysis standpoint) is:

1. Choose a complex function and integrate along some path / contour γ. By

the Residue Theorem, we can get our answer in a straightforward way.

2. Break the contour into different parts

• the needed real integral

• use symmetry, decay of function, etc., in the limit (we shall see more
about this later on)

1
Let f (z) = 1+ z4
. The singularities are

π 3π 5π 7π
z4 = −1 =⇒ z = ei 4 , ei 4 , ei 4 , ei 4

(Note: These are all simple poles)

Let R > 0, and let Γ R be the semi-circular, anti-clockwise contour,

centered at zero, sitting in the positive side of the imaginary axis on the
 PMATH352W18 - Complex Analysis 127

complex plane. 1 Theorem 31 gives that

1
Z
f (z) dz = Resπ f (z) + Res f (z)
2πi γ z=e
i
4 i 3π
z=e 4
24 Lecture 24 Mar 14 2018

24.1 Application of Cauchy’s Residue Theorem (Continued)

We will continue with the previous example.

Exercise 24.1.1
R∞ 1
Evaluate I = − ∞ 1+ x 4 dx.

1
Consider the function f (z) = 1+ z4
. Then f has simple poles at α1 =
i π4 i 3π i 5π 7π
e , α2 = e 4 , α3 = e 4 , α4 = e i 4
. Consider the contour Γ R , where R is
large, that consists of an anticlockwise semi-circle CR going from R to − R,
and a straight line from − R to R on the real axis.

By the Residue Theorem,

1 1
Z
dz = Res f (z) + Res f (z) (24.1)
2πi γ 1 + z4 z = α1 z = α2

Note that for Equation (24.1),

R
Z 
1 1 1
Z
LHS = dx + dz
2πi −R 1 + x4 CR 1 + z4

On CR , we have that |z| = R, so 1 + z4 ≥ |1| − |z|4 = R4 − 1, and

therefore

1 1
Z Z
dz ≤ dz
CR 1 + z4 CR 1 + z4
1
Z
≤ dz
CR R4 − 1
1
Z
= 4
|dz|
R − 1 CR
1
= 4 · πR
R −1
130 Lecture 24 Mar 14 2018 - Application of Cauchy’s Residue Theorem (Continued)

As R → ∞, we have 1
R
CR 1+ z4 dz → 0, since it is bounded above by πR
R4 −1
that goes to 0.

Therefore, taking the limit of LHS (as well as RHS) as R → ∞ in Equa-

tion (24.1), we have
Z ∞
1 1
= Res f (z) + Res f (z)
2πi −∞ 1 + x4 z = α1 z = α2

Next, we compute the residues:

Res f (z) = lim f (z)(z − α1 )

z = α1 z → α1
z − α1
= lim where g(z) = 1 + z4
g(z)
z → α1
z − α1
= lim ∵ g ( α1 ) = 0
z → α1 g ( z ) − g ( α 1 )
1 1 1
= 0 = 3 = 3
g ( z ) z = α1 4z α1 4α1
1 1
Res f (z) = =
z = α2 4z3 α2 4α32

So RHS of Equation (24.1) is

 
1 1 1 1  −i 3π π
 i π i
RHS = π + π = e 4 + ei 4 = sin = − √
4 e3i 4 e9i 4 4 2 4 2 2

Therefore,
Z ∞  
1 i π
4
dx = 2πi − √ =
−∞ 1+x 2 2 2

Exercise 24.1.2
R∞ sin x
Show that −∞ x dx = π
(Note: This integrand is not absolutely convergent)

sin z 1
If we try f (z) = z on some semi-circle arc CR with | f (z)| ≤ R, then

1 length of CR
Z Z
f (z) dz ≤ |dz| = ≈π
CR CR R R

which means that the decay of the f is insufficient to help us compute our
desired result.

eix −e−ix
Consider sin x = 2i . We then need to show
Z ∞ ix
e − e−ix
I= dx = π
−∞ 2ix
 PMATH352W18 - Complex Analysis 131

eiz (iz)2
Let f (z) = z= 1z (1 + iz + 2 + . . .). Thus F has a simple poles at
z = 0 with residue 1.

Consider the countour Γ R :

Im
Figure 24.1: Contour Γ R
CR

Cε

Re
−R −ε ε R

The Residue Theorem gives

1 eiz
Z
dz = 0
2πi ΓR z
Z −ε ix
eiz
Z R ix
e e eiz
Z Z
=⇒ 0 = dx + dz + dx + dz
−R x Cε z ε x CR z

Note that

eiz
Z π iR(cos θ +i sin θ )
e
Z
dz = Rieiθ dθ
CR z 0 Reiθ
Z π
≤ e− R sin θ dθ
0
Z π Z π
2 − R sin θ
= e dθ + π
e− R sin θ dθ
0 2
Z π
2
=2 e− R sin θ dθ by symmetry
0

By a similar argument as in A3Q1(c), on [0, π2 ],

2
sin θ ≥ θ
π
2
e− sin θ ≤ e− π θ

Thus

eiz
Z Z π
2 2Rθ
dz ≤ 2 e− π dθ
CR z 0
 π  2Rθ π
2 π
=2 − e− π = − e − R −1
2R 0 R
Thus we observe that as R → ∞, RHS → 0.
132 Lecture 24 Mar 14 2018 - Application of Cauchy’s Residue Theorem (Continued)

By A4Q4, on Cε , as ε → 0, we pick up half of the residue at z = 0, i.e.

1 eiz 1
Z
dz = − ,
2πi Cε z 2

in which we note that the value is negative since Cε is clockwise.

Therefore,

−ε eix Z R ix ∞ eix
Z  Z
e
lim dx + dx = dx = πi
R→∞ −R x ε x −∞ x
ε →0

Using a similar argument, it can be shown that Work out the similar case as an exercise

Z ∞ −ix Exercise 24.1.3

e
dx = −πi Show that
−∞ x Z ∞ −ix
e
dx = −πi
−∞ x
And with that, we obtain the final solution of

πi − (−πi )
=π
2i
as required.

Refer to Stein & Shakarchi, Section 2.1, for more examples.

Interesting Examples from Stein & Shakarchi

Example 24.1.1

Prove that for 0 < a < 1,

Z ∞
e ax π
dx =
−∞ 1 + ex sin πa

e az
Choose f (z) = 1+ e z . Note that f has one pole at z = iπ. Consider a
rectangle lying on the upper-half side of the plane with its base lying on the
real axis and its top the line 2π.

Im
Figure 24.2: Contour Γ R
2π ΓR

Re
−R R
 PMATH352W18 - Complex Analysis 133

By the Residue Theorem, we have that

1
Z
f (z) dz = Res f (z)
2πi ΓR z=iπ

To evaluate the residue, note that

e az z − iπ
(z − iπ ) · z
= e az · z
1+e e − eiπ
z−iπ
and limz→iπ ez −eiπ
is the inverse of the difference quotient (of first princi-
ples), and is hence eiπ = −1. Therefore

e az
Res f (z) = lim (z − iπ ) · = − lim e az = −e aiπ .
z=iπ z→∞ 1 + ez z→iπ

Let I be the desired integral and

Z R
IR : = f (z) dz
−R

so that we may have IR → I as R → ∞. This will describe the bottom part

of the rectangle. For the top part of the integral, notice that it is simply

e az e a( x+2πi)
Z −R Z R
dz = − dx
R 1 + ez − R 1 + e x +2πi
e ax e2πia
Z R
=− dx
− R 1 + e x e2πi
e ax
Z R
= −e2πia dx ∵ e2πi = 1
−R 1 + ex
= −e2πia IR .

Let γr parameterize the vertical path on the right side of the rectangle, i.e.
γr : [0, 2π ] → C such that γr (t) = R + it. Then

e az
Z 2π a( R+ti)
e
Z
dz = dt
γr 1 + ez 0 1 + e R+ti
e aR
Z 2π
≤ dt
1 + eR 0

which goes to 0 as R → ∞ since aR < R for 0 < a < 1. Thus the integral
of the vertical path evaluates to 0. We can perform a similar procedure for
the path on the left and end up with the same result.

With that, we have that

IR − e2πia IR = −2πieπia
134 Lecture 24 Mar 14 2018 - Application of Cauchy’s Residue Theorem (Continued)

and so
−2πieπia 2πi π
IR = = πia =
1 − e2πia e − e−πia sin πa
and hence as R → ∞, we have
Z ∞
e ax π
dx =
−∞ 1 + ex sin πa

as required.
25 Lecture 25 Mar 16 2018

We are now in a position to look into how we can define “loga-

rithms” for C.

25.1 The Argument Principle

Since we may express z = Rei(θ +2kπ ) for some k ∈ Z, we would

expect a logarithm to be of the form

log z = log R + i (θ + 2kπ )

So in general,
log f (z) = log | f (z)| + i arg f (z)
f 0 (z)
The derivative of log z is f (z)
, should we expect the same idea ex-
tending from the reals, which is single-valued. Then the integral

f 0 (z)
Z
dz
γ f (z)

can be interpreted as the change in the argument of f as z traverses

the curve γ. Moreover, assuming that γ is a closed path, this change
of argument is determined entirely by the zeros and poles of f in γ.

å Note (Stein & Shakarchi, pg. 89)

The addivitity formula for log,

log( f 1 f 2 ) = log f 1 + log f 2

fails in general.
136 Lecture 25 Mar 16 2018 - The Argument Principle

1 Theorem 32 (Argument Principle)

Suppose f is meromorphic on a region (open & connected) Ω ⊆ C, γ a

closed path such that γ∗ ∈ Ω \ (A ∪ Z ( f )) such that

• ∀w 6∈ Ω Indγ (w) = 0

• ∀w ∈ Ω \ γ∗ Indγ (w) = 0 or 1

Then
1 f 0 (z)
Z
dz = Z ( f ) ∩ γ0 − A ∩ γ0
2πi γ f (z)
where the zeros and poles are counted by multiplicity.

Ò Proof
(include proof here: use 1 Theorem 31 to CTP).

Example 25.1.1
f0
What are the poles of f ?

Suppose that f has a zero of order k at z0 . Then ∃r > 0, ∀z ∈ D (z0 , r ), f (z) =

(z − z0 )k g(z) where g ∈ H D (z0 , r ) and g 6≡ 0 on D (z0 , r ). So

f 0 (z) = k(z − z0 )k−1 g(z) + (z − z0 )k g0 (z) =⇒

f 0 (z) k g0 (z)
= + =⇒
f (z) z − z0 g(z)

f0
f has a simple pole at z0 with residue k.

Suppose f has a pole of order k. Then ∃r > 0, ∀z ∈ D (z0 , r ), ∃h ∈

H D (z0 , r ) h 6≡ 0, f (z) = (z − z0 )−k h(z). Then

f 0 (z) = −k (z − z0 )−k−1 h(z) + (z − z0 )−k h0 (z) =⇒

f 0 (z) −k h0 (z)
= + =⇒
f (z) z − z0 h(z)

f0
f has a simple pole at z0 with residue −k.
0
∴ f is meromorphic on Ω =⇒ ff has simple zeros and poles at exactly
the zeros and poles of f with residue equals to the order of zeros of f and
negative of the order of poles of f , respectively.
 PMATH352W18 - Complex Analysis 137

1 Theorem 33 (Rouché’s Theorem)

Let Ω ⊆ C be a region, f , g ∈ H (Ω), γ a closed path on Ω with

• ∀w 6∈ Ω Indγ (w) = 0,

• ∀w ∈ Ω \ γ∗ Indγ (w) = 0 or 1.

If f , g satisfy
∀z ∈ γ∗ | f (z) − g(z)| < | f (z)| ,

then f and g have the same number of zeros on γ∗ (counted with multi-
plicity).

Ò Proof (Idea)
1
R f0 1
R g0
WTS 2πi γ f = 2πi since the result would follow by 1 Theorem 32.
γ g,
1
R h f 0 g0 i
This is equivalent to = γ f − g = 0. By our small argument
2πi
about how a log function should behave in C, observe that we thus have
 0
0 0 f
(log f ) − (log g) = log
g

g f
So we wish to get either f or g .

Ò Proof
Given that ∀z ∈ γ∗ , | f − g| < | f |. Note that f 6≡ 0; otherwise we would
have the impossible case of |0 − g| < |0|.

Divide both sides by | f |, then

g
∀z ∈ γ∗ 1− <1
f
g
Therefore, ∀z ∈ γ∗ , F = ∈ B(1, 1). Let γ : [ a, b] → C be a
f
parameterization of γ∗ .Consider the function F ◦ γ, which is a closed
path that is contained in B(1, 1). ∵ z = 0 lies outside of F ◦ γ, we have
138 Lecture 25 Mar 16 2018 - The Argument Principle

IndF◦γ (0) = 0. Then

1 1
Z γ
dz = 0
2πi zF◦
Z b 0

1 F γ(t) γ(t) dt
=⇒
=0
2πi a F γ(t)
1 F 0 (z)
Z
=⇒ dz = 0 by lettingγ(t) = z
2πi γ F (z)
g0 f0
 
1
Z
=⇒ − dz = 0 (†)
2πi γ g f

where for Equation (†), note that

g0 f − g f 0
F0 = by Quotient Rule
f2
F0 g0 f − g f 0 f g0 f0
= · = −
F f2 g g f

The proof is complete by 1 Theorem 32.

26 Lecture 26 Mar 19 2018

26.1 The Argument Principle (Continued)

å Note (Notation)
Let f be a function meromorphic on a region Ω ⊆ C. We write

N f := #zeros of f inside γ∗ − #poles of f inside γ∗

= Z ( f ) ∩ γ0 − A ∩ γ0

Remark
If all conditions of Rouché’s Theorem hold except that, instead, f & g are
meromorphic on Ω, then if γ∗ contains no poles of f & g then we can con-
clude that N f = Ng

Exercise 26.1.1

Find the number of roots of P(z) = z8 − 5z3 + z − 2 lying in |z| ≤ 1.

Ò Solution
Let γ be the circle |z| = 1, oriented anticlockwise. Let g(z) = P(z), f (z) =
−5z3 1 . Then | f (z)| = 5z3 = 5, and 1
We pick the dominant term in P for f

| f (z) − g(z)| = z8 + z − 2

≤ 1 + 1 + 2 by Triangle Inequality, and on γ

= 4 < 5 = | f (z)|

So the inequality in Rouché’s Theorem holds. Hence by Rouché, P(z) =

g(z) has 3 roots (at z = 0, counted thrice since it has order 3) in |z| < 1.

To get the zeros for |z| ≤ 1, change γ to be on |z| = 1 + ε for some ε > 0
 140 Lecture 26 Mar 19 2018 - The Argument Principle (Continued)

and proceed from there.

You should try more of these problems from the recommended

texts.

26.1.1 Alternative Proof for FTA

Before proceeding with providing with alternative proof, note the

following two definitions about polynomials.

 Definition 29 (Monic Polynomial)

A monic polynomial is a polynomial with a leading coefficient of 1.

 Definition 30 (Monomial)

A monomial is a polynomial with only one term.

Recall the statement of the Fundamental Theorem of Algebra

(FTA)

∀ P ∈ C [z] with deg P = n for some n ∈ N, P has n roots in C.

Ò Proof
Without loss of generality, assume that the polynomial is monic (divide
the polynomial by the leading coefficient if necessary). Take

g ( z ) = z n + a n −1 z n −1 + . . . + a 1 z + a 0

a1 ∈ C for i ∈
with n o [1, n − 1] ⊂ N. Let γ be the circle |z| = R >
n −1
max ∑ j=0 a j , 1 2 , oriented anticlockwise. Let f (z) = zn . Then 2
This is chosen from the later part of
the proof
| f (z)| = Rn on γ. We also have

| g(z) − f (z)| = an−1 zn−1 + . . . + a1 z + a0

≤ | a n −1 | R n −1 + . . . + | a 1 | R + | a 0 |
≤ (| an−1 | + . . . + | a1 | + | a0 |) Rn−1
< Rn
 PMATH352W18 - Complex Analysis 141

Hence, the inequality for Rouché’s Theorem holds. Hence by Rouché,

N f = Ng and N f = n.

Exercise: Show that these are the only zeros of g(z), using factoriza-
tion of polynomials in the ring C[z].

Suppose not, i.e. say g has m 6= n zeros. If m > n, then that would
imply that deg g = m, which Eassumption. If m < n, then we can write

g(z) = (z − α1 )(z − α2 ) . . . (z − αm ) P1 (z)

where each α j ∈ C is a root of g and P1 ∈ C[z] has deg P1 = n − m

and that P1 has no roots (otherwise we would have m + 1 roots). Then P1
must be a constant polynomial, but that would imply that deg g = m 6=
n, which is yet another E. 

The above proof leads to the following result:

 Corollary 34

All the zeros

n of a monic polynomial
o lie inside the disc |z| ≤ R with
R = max ∑nj=−01 a j , a where { a j }nj=−01 ⊂ C are the coefficients of the
monic polynomial.

26.1.2 Open Mapping Theorem

1 Theorem 35 (Open Mapping Theorem)

If f is holomorphic and non-constant in a region in C, then f maps open

sets to open sets.

Ò Proof
Let w0 = f (z0 ) for some z0 ∈ Ω ⊆ C. Let d > 0.

0
WTS w0 ∈ f B(z0 , δ) .
142 Lecture 26 Mar 19 2018 - The Argument Principle (Continued)

Let γ = ∂B(z0 , δ) (i.e. |z − z0 | = δ), oriented anticlockwise. ∀z ∈

B(z0 , δ), let F (z) := f (z) − w0 . Then F has at least one zero inside γ (in


particular, z0 ). Let G (z) := f (z) − w for some w ∈ f B(z0 , δ) .

Want to have G (z) having a zero inside γ for w “close enough” to

w0 .

Our setup satisfies Rouché’s inequality:

∀z ∈ γ∗ | F (z) − G (z)| < | F (z)|

or |w − w0 | < | f (z) − w0 | on γ∗

We want f (z) 6= w0 on γ. Now we can choose a δ > 0 such that

B(z0 , δ) ⊆ Ω and ∀z ∈ ∂B(z0 , δ), f (z) 6= w0 .

Let ε = maxz∈γ∗ | f (z) − w0 | > 0. Observe that

|w − w0 | < ε =⇒ Z ( G ) ∩ γ0 6= ∅


=⇒ w ∈ f B(z0 , δ)

0
=⇒ w0 ∈ f B(z0 , δ)

This completes the proof. 

27 Lecture 27 Mar 21 2018

27.1 Introductory Passage to Log Functions in C

We have dealt with integrals of real numbers using our approach

from complex analysis. But what would we do if we come across a
problem of the form
Z ∞
f ( x ) x a dx for some a ∈ R?
−∞

If we try to apply residue integrals to the problem, we would need to

consider f (z)z a . But what is z a , since a ∈ R and not simply ∈ Z!?

When a ∈ N, we know that z a = z| .{z

. . }z. When a ∈ R, we want to
a times
be able to interpret z a as e a log z just as we can do so in R. This leads
to the study of log functions as complex variables.

We shall try to approach the problem via analytic continuation.

Exercise 27.1.1 (A simple problem in analytic continuation)

Let f (z) = ∑∞ n
n=0 z for | z | < 1. We want to analytically continue f onto
C if possible.

For |z| < 1, we know that ∑∞ n

n =0 z =
1
1− z . So let g(z) = 1
1− z . Then we
have that g ∈ H (C \ {1}), where z = 1 is a simple pole, and g agrees with
f on |z| < 1.

∴ g is an analytic continuation of f to C \ {1}, or we say that f can be

analytically continued except at z = 1, which is a simple pole.

In real analysis, log is the inverse of e1 . But on C, the exponential 1

e in R is 1-1, and goes from R → R+
function is not 1-1, e.g.

ez = 1 ⇐⇒ z ∈ 2πiZ
 144 Lecture 27 Mar 21 2018 - Simply Connected Domains

As such, we would like to restrict the domain (why?) for the expo-
nential function. That begs the question: what is the natural domain
on which log z lives for z ∈ C?

• Globally, we would require the notion of Riemann Surfaces

• Locally, we would require the notion of Simply Connected Do-

mains

(What does local and global mean here?)

27.2 Simply Connected Domains

 Definition 31 (Homotopy (Poincaré))

Let X be a topological space2 . Recall that a curve in X is a continuous 2

which we did not define
map γ : I → X where I = [0, 1], and γ is said to be closed if γ(0) =
γ (1).

Two closed curves γ0 and γ1 are said to be homotopic if ∃ H : I ×

I → X with
H (s, 0) = γ0 (s) H (s, 1) = γ1 (s)

and H (s, t) be continuous with respect to s and t.

Alternative Definition from Stein-Shakarchi - Complex Analy-

sis3 3
I preferred this definition cause it’s
easier to read, but I shall be using the
definition from the lecture for the class
Let γ0 and γ1 be two curves in an open set Ω with common end-
itself unless stated otherwise
points. So if γ0 and γ1 are two parameterizations on [ a, b], then

γ0 ( a) = γ1 ( a) = α and γ0 (b) = γ1 (b) = β

where α, β ∈ Ω. The two curves are said to be homotopic in Ω if for

each 0 ≤ s ≤ 1, ∃γs ⊂ Ω parameterized by γs (t) defined on [ a, b], such
that ∀s,
γs ( a) = α and γs (b) = β,

and ∀t ∈ [ a, b],

γs ( t ) = γ0 (t) and γs (t) = γ1 (t).

s =0 s =1

Moreover, γs (t) should be jointly continuous in s ∈ [0, 1] and t ∈ [ a, b].

 PMATH352W18 - Complex Analysis 145

Loosely speaking, γ0 , γ1 are homotopic if we can continuously

deform γ0 to γ1 (wlog) without any obstruction in X.

 Definition 32 (Simply Connected Domain)

Let Ω ⊆ C be open. We say Ω is simply connected if Ω is connected,

and ∀γ that is closed in Ω is homotopic to a point (i.e. a constant map
γ : I → X).

Exercise 27.2.1

1. C is simply connected.

2. C \ {z = x + iy : x ≤ 0, y = 0} is simply connected.

3. C \ {0} is not simply connected.

å Note
I will temporarily use ∼ to represent homotopy, since it is an equivalence
relation.

Here’s a quick proof of that:

1. (Reflexive) Define H : I × I → X, where I = [ a, b] ⊆ R, with

H (s, t) = γt (s), where, in this case, t = 0. This shows reflexivity.

2. (Symmetric) Suppose γ0 ∼ γ1 . Then ∃ H as above such that, this

time, t ∈ [0, 1]. Choose G : I × I → X with G (s, t) = γ−t (s) with
t ∈ [0, 1]. Then γ1 ∼ γ0 .

3. (Transitive) Suppose γ0 ∼ γ1 and γ1 ∼ γ2 . Then ∃ H1 , H, 2 : I × I →

X, I as above, with

H1 (s, t) = γt (s)
H2 (s, q) = γq (s)

with t ∈ [0, 1] and q ∈ [1, 2]. Then we can simply create G : I × I →

X, now with the 2nd argument, say, p ∈ [0, 2]. such that

 H (s, p) = γ p (s)
1 p ∈ [0, 1]
G (s, p) =
 H (s, p) = γ (s)
2 p p ∈ (1, 2]
146 Lecture 27 Mar 21 2018 - Simply Connected Domains

Then γ0 ∼ γ2 .

One of the key facts about simply connected domains is that, if f ∈

H (Ω), then whenever γ0 ∼ γ1 in Ω
Z Z
f = f
γ0 γ1

i.e. in a simply connected domain, the integral does not depend on

the chosen path4 . 4
This should remind you of Goursat’s
Theorem and Cauchy’s Theorem
28 Lecture 28 Mar 23 2018

28.1 Constructing Logarithm

1 Theorem 36 (Theorem 17)

Suppose Ω is simply connected with 0 6∈ Ω. Then in Ω, we can define a

function, call it Log z1 , such that 1
This is called a branch of the loga-
rithm
1. Log z ∈ H (Ω) ∧ (Log z)0 = 1
z

2. eLog z = z for all z ∈ Ω

3. ∀r ∈ R+ [1, r ] ⊆ Ω =⇒ Log r − Log 1 = log r where log denotes

the usual natural logarithm on R+ .

Ò Proof
1. The proof can be completed using the method used in proving Cauchy’s
Theorem for Convex Sets if we define Log as follows:

∀ z ∈ Ω ∃ w 0 ∈ C e w0 = z 0

(If we let z0 = Reiθ , then we choose w0 = log R + iθ) Define

Z z
1
Log z = w0 + dw (†)
z0 w

where the integral is over any path between the points z0 and z in Ω.
From here, use the proof provided in Cauchy’s Theorem for Convex
Sets to complete the proof.

2. Let G (z) = e− Log z · z. WTS G (z) = 1.

148 Lecture 28 Mar 23 2018 - Constructing Logarithm

Note that by part 1,

1 − Log z
∀z ∈ Ω G 0 (z) = e− Log z − z · ·e =0
z
∴ G 0 ≡ 0 in Ω. ∵ G ∈ H (Ω), we may write G as a power series, and
since G 0 ≡ 0 on Ω, we have that G (z) = G (z0 ) in a neighbourhood of
a chosen center z0 ∈ Ω, say with radius r0 > 0. Therefore

∃ c ∈ C ∀ z ∈ B ( z0 , r0 ) G ( z ) = c

Thus by Analytic Continuation, since Ω is connected, we have that

∀z ∈ Ω, G (z) = c.

It is therefore sufficient to show that G (z0 ) = 1, and this is true by the

following:

G (z0 ) = e− Log z0 · c0
= e−w0 · z0 ∵ Equation (†)
z
= w0 = 1 ∵ ew0 = z0
e 0

Thus we have G ≡ 1 on Ω and hence ∀z ∈ Ω, eLog z = z.

3. Suppose r ∈ R+ and [1, r ] ⊆ Ω. By Equation (†),

z−0 1
Log r = w0 + dw
r w
Z 1 Z r
1 1
= w0 + dw + dw
z0 w 1 w
Z 1 Z r
1 1
= w0 + dw + dt
z w t
| {z0 } | 1 {z }
Log 1 by Equation (†) log r −log 1=log r

where we choose the straight line [1, r ] as the path for the 3rd term in
the last line. Therefore we have

Log r − Log 1 = log r

as required. 

å Note
If we choose z0 = 1 and w0 = 0, then Log 1 = 0, and hence Log r =
log r for any r ∈ R+ with [1, r ] ⊆ Ω.
 PMATH352W18 - Complex Analysis 149

28.2 Branches of the Logarithm

1. (Principal Branch) Let Ω1 = C \ (−∞, 0]. We will write z ∈ Ω1 as

z = reiθ with r > 0 and θ ∈ (−π, π ). Pick z0 = 1 ∧ w0 = 0> By
Rz
Equation (†), Log z = 1 w1 dw. Then in this case,

Log z = log r + iθ when z = reiθ with θ ∈ (−π, π )

To see this, pick the straight line path from 1 to r, and then any
path from r to z = reiθ

Im

Re
0 1 r

Then

ireit
Z z Z r
1 1
Z π
dw = dt + dt
1 w 1 t 0 reit
= log r + iθ

Exercise 28.2.1
2πi
Let z1 = e 3 , then, using the Principal Branch, Log z1 = i 2π
3 . But note
that Log(z21 ) 6= i 4π
3 . Instead, since

4πi 2πi
z21 = e 3 = e− 3

(∵ the region in consideration is (−π, π )), we have that

2π
Log(z21 ) = −i
3

2. (a different branch) Let Ω2 = C \ [0, ∞). Write z ∈ Ω2 as z = reiθ

with r > 0 ∧ θ ∈ (0, 2π ). Now we can pick some function so that

Log z = log r + iθ with z = reiθ ∧ θ ∈ (0, 2π )

150 Lecture 28 Mar 23 2018 - Branches of the Logarithm

In this case, we have that Log(z21 ) = 2 Log z1 does hold.

With that established, we may now use z a = e a Log z if we fix a

branch (and a simply connected domain) and stick with it till the end
of the problem.

Remark
For the Principal Branch of the logarithm, the following Taylor expansion
holds:
For |z| < 1,

z2 z3
log(1 + z) = z − + −...
2 3
∞
(−1)n+1 zn
= ∑
n =1
n! n

(To remember this: note − log(1 − z) = ∑∞

n =1
zn
n) Let

F (z) = − log(1 − z)
∞
zn
G (z) = ∑ n.
n =1

It can be shown that for |z| < 1,

1
F 0 (z) = = G 0 (z)
1−z
which hence implies that

( F − G )0 ≡ 0
=⇒ F − C = c for some c ∈ C

Plug in z = 0 and we will get c = 0.

29 Lecture 29 Mar 26 2018

29.1 Examples for Analytic Continuation

Gamma Function

For s ∈ R+ , we define
Z ∞
dt
Γ(s) = e−t ts
0 t
where
Z ∞
: the integral over a locally compact topological group R+
0
−t
e : additive character of R+ (homomorphism from (R+ , +) to R)
ts : multiplicative character of R+ (homomorphism from (R+ , ·) to R)
dt
: Haar measure for R+ (invariant under multiplication)
t
Exercise 29.1.1

R∞
The integral 0 e−t ts dtt converges for s > 0. Prove this.
152 Lecture 29 Mar 26 2018 - Examples for Analytic Continuation

å Note (Euler)
Euler observed that
Z ∞
dt
Γ ( n + 1) = e − t t n +1
0 t
Z ∞
= e−t tn dt
0
∞ Z ∞
= − tn e−t +n e−t tn−1 dt by IBP
0 0
= nΓ(n)
..
.
= n ( n − 1) . . . 2 · 1 · Γ (1)
R∞
and since Γ(1) = 0 e−t dt = 1, we have that

Γ(n + 1) = n!

Remark
Euler observed that Γ(s) is a continuous and differentiable function of s that
interpolates the factorials.

We can extend Γ(s) to complex numbers s as follows:

Z ∞
dt
∀s ∈ C Re s > 0 Γ(s) = e−t ts
0 t

å Note
1. Γ(s) is holomorphic for Re s > 0
R∞
• It can be shown that 0 e−t ts dtt converges for Re s > 0

• It can also show that this is C-differentiable

2. Γ is a Functional Equation: We can repeat Euler’s calculation to

show that
∀s ∈ C Re s > 0 Γ(s + 1) = sΓ(s)

which implies that, if s 6= 0,

Γ ( s + 1)
Γ(s) =
|{z} s }
defined for Re s>0
| {z
defined for Re s>−1
 PMATH352W18 - Complex Analysis 153

because RHS makes sense for Re s > −1, in which we may do

−1 < Re s < 0 =⇒ 0 < Re (s + 1) < 1.

Thus, we can define, for −1 < Re s < 0, that, if s 6= 0,

Γ ( s + 1)
Γ(s) =
s
It is noteworthy that this definition agrees with our original definition
of Γ due to Equation (†).

Q: What happens at s = 0?
Consider Equation (†), with s → 0+ . Then

lim [sΓ(s)] = Γ(1) = 0! = 1

s →0+

∴ Γ(s) behaves like 1

s near s = 0, i.e. Γ has a simple poles at s = 0.

Q: Can we continue the procedure above and go beyond Re s >

−1?
Yes. Equation (†) holes for Γ(s + 2) as well, which then we have,
for Re s > 0,

Γ(s + 2) = (s + 1)Γ(s + 1) = (s + 1)(s)Γ(s)

And thus for Re s > −2 and s 6= 0, −1,

Γ ( s + 2)
Γ(s) =
s ( s + 1)

We can proceed with this procedure inductively so and analytically

continue Γ to C \ {0, −1, −2, ...}.

29.2 Characterizing Logarithms

1 Theorem 37 (Theorem 18)

Any entire function f (z) without any zeros has the form Ae g(z) where g
is some entire function and A ∈ C is some constant.
154 Lecture 29 Mar 26 2018 - Characterizing Logarithms

This is a characterization of the function f that has no zeros or

poles.
30 Lecture 30 Mar 28 2018

30.1 Characterizing Logarithms

1 Theorem 38 (Theorem 18)

∀ f that is entire with Z ( f ) = ∅,

f (z) = Ae g(z)

where g is some entire function and A ∈ C is a constant.

Ò Proof
Note

f0
f ∈ H (C) =⇒ f 0 ∈ H (C) ∧ Z ( f ) = ∅ =⇒ ∈ H (C)


f

Choose
f 0 (z)
g0 (z) = = c0 + c1 z + c2 z2 + . . .
f (z)
where {c j } j∈Z≥0 ⊆ C1 . 1
g can be obtained by term-wise inte-
f0
gration of the Taylor series for f
Consider F (z) = f ( z ) e− g(z) . Then ∀z ∈ C,

F 0 ( z ) = f 0 ( z ) e− g(z) − f ( z ) g0 ( z ) e− g(z)
f 0 ( z ) − g(z)
= f 0 ( z ) e− g(z) − f ( z ) e
f (z)
=0

∴ ∀z ∈ C F 0 (z) ≡ 0. Now because of that and F ∈ H (C), ∃ A ∈

 156 Lecture 30 Mar 28 2018 - Infinite Products

C ∀ z ∈ C F ( z ) ≡ A2 . 2
By considering the Taylor series for F

∴ ∀z ∈ C f (z) = Ae g(z) . 

This characterizes any function f that has Z ( f ) = ∅. Suppose

f ∈ H (C) with A f = { a1 , a2 , a3 , ...} for some { a j } j∈N ⊆ C. Construct
some function h ∈ H (C) with zeros at exactly every point in A f . For
example,
∞
!
z
h(z) = ∏ 1−
aj
j =1

f
∈ H (C)
is an entire function and has zeros at exactly A f . Then h
f
with Z( h )= ∅ on C. Then by 1 Theorem 38, ∃ g, some entire
function, and A ∈ C some constant, such that,

f
= Ae g =⇒ f (z) = Ah(z)e g(z)
h

The construction of h motivates us to study our next topic: infinite

products.

30.2 Infinite Products

 Definition 33 (Infinite Products)

Let u1 , u2 , ... be a sequence in C. Let

N
∏


PN = 1 + uj
j =1

be the N th partial product. If lim N →∞ PN exists, then we say that the

infinite product, ∏∞


j=1 1 + u j , converges, and write

∞
∏


lim PN = 1 + uj
N →∞
j =1

Before proceeding with an important result about infinite prod-

ucts, consider the following lemma.
 PMATH352W18 - Complex Analysis 157

+ Lemma 39 (Bounds of the Partial Product)

With {u j }∞
j=1 being a sequence in C, let

N
∗
∏


PN = 1 + uj .
j =1

Then
 
∗ ≤ exp
1. PN ∑N
j =1 u j

∗ − 13
2. ∀ N ∈ N | PN − 1| ≤ PN 3 ∗ ≥1
Note that PN

Ò Proof
1. Note that ∀ x > 0 1 + x ≤ e x . Thus

N
∗
∏


PN = 1 + uj
j =1
N
≤ ∏ exp ( u j )
j =1
!
N
= exp ∑ uj
j =1

2. Using induction, note that N = 1 holds since,

PN − 1 = 1 + u1 − 1 = u1
∗
PN − 1 = 1 + | u1 | − 1 = | u1 | .

For N + 1, we have

| PN +1 − 1| = | PN (1 + u N +1 ) − 1|
= | PN (1 + u N +1 ) − (1 + u N +1 ) + (1 + u N +1 ) − 1|
= |( PN − 1)(1 + u N +1 ) + u N +1 |
≤ | PN − 1| |1 + u N +1 | + |u N +1 | By Triangle Inequality
≤ ( PN∗ − 1)(1 + |u N +1 |) + |u N +1 | By IH
= PN∗ (1 + |u N +1 |) − (1 + |u N +1 |) + |u N +1 |
= PN∗ +1 − 1

and the induction is complete. 

158 Lecture 30 Mar 28 2018 - Infinite Products

Remark
∗ are replaced by
Lemma 39 continues to hold if PN , PN

N
PN,M = ∏ (1 + u j )
j = M +1
N
∗
PN,M = ∏ (1 + u j )
j = M +1

respectively, where N ≥ M + 1.

1 Theorem 40 (Theorem 19)

Suppose that ∑∞ ∞
n=1 | un | converges. Then ∏n=1 (1 + un ) converges.
Moreover, ∏∞
n=1 (1 + un ) converges to zero iff un0 = −1 for some
n0 ∈ N.

Ò Proof
Let PN = ∏nN=1 (1 + un ).

WTS { PN } is convergent. It suffices to show that { PN } is Cauchy,

i.e.4 4
Cauchy convergence from real analysis
∀ε > 0 ∃ N0 > 0 ∀ M, N > N0 | PN − PM | < ε

!
M
| PM − PN | = PN ∏ (1 + u j ) − 1
j = N +1
M
= | PN | ∏ (1 + u j ) − 1
j = N +1
M
≤ | PN | ∏ (1 + u j ) − 1 By 2 of Lemma 39
j = N +1
!
M
≤ | PN | exp ∑ uj − 1 By 1 of Lemma 39
j = N +1
! !
M
≤ ∗
PN exp ∑ uj −1 .
j = N +1

Thus ∀ε > 0, ∵ ∑∞
j=1 u j < ∞ by assumption, we can choose an
 PMATH352W18 - Complex Analysis 159

appropriate N0 (ε) > 0 such that ∀ M > N ≥ N0 ,

M
∑ u j < ε5
j = N +1

So by choosing the appropriate N0 we can have 5

Since it is convergent and hence
Cauchy.

| PM − PN | ≤ PN∗ (eε − 1)

Note that
!
N
PN ≤ exp ∑ uj By 1 of Lemma 39
j =1

≤ exp c For some c ∈ C

= c0 For some c0 ∈ C

Therefore, for M > N ≥ N0 , we have | PM − PN | ≤ c0 ε and PN is

hence Cauchy.

(Proof to be continued in next lecture)

31 Lecture 31 Apr 02 2018

31.1 Infinite Products (Continued)

Ò Proof (Continued)
Note that in the earlier part of the proof, we showed that
! !
M
| PM − PN | ≤ | PN | exp ∑ uj −1 (31.1)
j = N +1

Notice that by the Reverse Triangle Inequality,

| PM | = | PM − PN + PN | ≥ || PM − PN | − | PN || .

So for large enough M, N,

| PN − PM | ≤ | PN | (eε − 1) by Equation (31.1) and the earlier part.

Thus

| PN | − PM − PN ≥ | PN | (1 − (eε − 1))
= | PN | (2 − eε ).

Therefore, for sufficiently large M, N,

| PM | ≥ || PN | − | PM − PN || ≥ | PN | (2 − eε ) (31.2)

Now to prove the iff statement: Suppose that the infinite product con-
verges to 0. Let M → ∞ and fix N0 from above to be sufficiently large.
Then for Equation (31.2), LHS → 0 as M → ∞. Thus RHS → 0 as
well, and we thus have that, in the limit, PN0 (2 − eε ) = 0 and hence
PN0 = 0. But since PN0 is a finite product, there must ∃n0 ∈ N such
 162 Lecture 31 Apr 02 2018 - Infinite Products (Continued)

that un0 = −1.

The converse is trivially true: suppose that ∃n0 ∈ N such that un0 =
−1. Then we have that (1 − un0 ) = 0 and hence the product is 0. 

Remark
To apply 1 Theorem 40 to a sequence of functions {un (z)} in some region
Ω ⊆ C, we need ∑ un (z) to converge absolutely and uniformly1 . 1
No dependence on z, which is part of
the definition of uniform convergence.

31.1.1 Application to Riemann Zeta Function

We define ζ (s) = ∑∞ 1
n=1 ns for Re ( s ) > 1. This function is the well-
known Riemann Zeta Function
Remark
1. The series ∑∞
n =1
1
ns is absolutely convergent for Re (s) > 1.

2. By the construction of the function, it is holomorphic/analytic for Re (s) >

12 2
Requires the Weierstrass’ M-test.

(History) Euler looked at the series with real numbers first. It was
not until Riemann extended the function to become a function with
complex variables that the series became well-known, and hence
Riemann’s name is prepended to the function instead of Euler.

The series can be analytically continued to the entire complex

plane (using the functional equation3 ), except for a simple pole at 3
This is similar to what we did for the
Gamma function.
s = 1, i.e.
lim (1 − s)ζ (s) = 1.4
s →1+
4
Cauchy’s Residue Theorem

Euler showed that for Re (s) > 1, ∑∞ 1 1

n=1 ns = ∏ pprime (1 + ps +
1
p2s
+ . . .). Observe that RHS converges absolutely for Re (s) > 1. This
identity is known as Euler’s Identity and it is simply a statement
about the unique factorization of integers into primes5 . 5
This is the Fundamental Theorem of
Arithmetic
 PMATH352W18 - Complex Analysis 163

Note that for Re (s) > 1, we can write

 
1 1
ζ (s) = ∏ 1+
ps
+ 2s + . . .
p
pprime
!
1
= ∏ 1 − p1s
(Infinite Geometric Sum)
pprime
  −1
1
= ∏ 1− s
p
pprime

This will be useful for the next statement.

 Corollary 41 (Corollary for Theorem 19)

ζ (s) 6= 0 for Re (s) > 1.

Ò Proof
Fix s with Re (s) > 1. Then ζ (s) = ∏∞
n=1 (1 + un ) with

0 if n 6= p prime
un =
1 1

ps + p2s
+ . . . if n = p prime

1
For each s, we have that each of the sums + p12s + . . . converges
ps
absolutely for Re (s) > 1. Also, ∑∞
n=1 un converges absolutely and
uniformly for Re (s) > 1. 6 6
These two statements are not too hard
to make reliable heuristics to make
Basically, we can apply 1 Theorem 40. So sense that they are true.

∀s ∈ C Re (s) > 1 ζ (s) = 0 ⇐⇒ ∃n ∈ N un = −1

1 1
⇐⇒ 1 + s + 2s + . . . for p prime
p p
ps
⇐⇒ s by the Infinite Geometric Sum
p −1
⇐⇒ ps = 0 ⇐⇒ es log p = 0
E ∀ x ∈ Re x 6= 0.

This completes the proof. 

 32 Lecture 32 Apr 04 2018

32.1 Infinite Products (Continued 2)

32.1.1 Weierstrass Products

Problem: Given a sequence { an }n∈N ⊆ C, where ∀n ∈ N, an 6= 0,

construct an entire function f with zeros precisely at each of the an ’s.
 
It is tempting to consider an infinite product such as ∏∞ z
n =1 1 − a n .
But this may not converge all the time except for some specific se-
quence { an }.

The idea to approach this problem is to take the product of expo-

nential factors so that it takes care of the convergence problem.

Define: ∀n ∈ N

z2 z3 zn
Pn (z) := z + + +...+
2 3 n
En (z) := (1 − z) exp ( Pn (z))

and for n = 0, define

E0 (z) := 1 − z

Observe that

• En is entire and has a simple zero at exactly z = 1.

• En (0) = 1 and En (1) = 0

z n −1 1
• Pn0 (z) = 1 + z + z2 + . . . + zn−1 = z −1
1
Finite geometric sum
166 Lecture 32 Apr 04 2018 - Infinite Products (Continued 2)

1 Theorem 42 (Theorem 20)

For |z| ≤ 1, |1 − En (z)| ≤ |z|n+1 .

Ò Proof
Note that

zn − 1
 
En0 (z) = (1 − z) exp ( Pn (z)) · − exp ( Pn (z))
z−1
= −zn exp ( Pn (z))

and note that by the Taylor expansion of the exponent,

( Pn (z))2
exp ( Pn (z)) = 1 + Pn (z) + +....
2!
Since the power/Taylor expansion of Pn (z) has non-negative and real
coefficients, so does exp ( Pn (z)). Therefore, we have

− En0 (z) = zn exp ( Pn (z))

∞
= zn · ∑ am zm a0 6= 02 , ∀m ∈ N am ≥ 0
m =0
∞
= ∑ cm zm c0 6= 0, ∀m > n cm ≥ 0, (?)
m=n

which holds ∀z ∈ C since En0 is entire3 . 2

This is because the exponential func-
tion is non-zero.
3
This is true by 1 Theorem 5, since En
is entire.
Now note that
Z z
− En0 (w) dw = − En (z) + En (0) by FTC
0
= 1 − En (z).

Also, by Equation (?),

Z z ∞
c m m +1
0
− En0 (w) dw = ∑ m+1
z .
m=n

Therefore,
∞
c m m +1 c n n +1
1 − En (z) = ∑ m+1
z =
n+1
z +...,
m=n
 PMATH352W18 - Complex Analysis 167

which implies that

∞
1 − En (z) cm m−n
z n +1
= ∑ m +1
z ∵ cn 6= 0
m=n

Let φn (z) := = ∑∞
1− En (z) k
k =0 bk z , where as defined, we have that
z n +1
b0 6= 0 and ∀k ∈ N bk ≥ 0. By the Triangle Inequality, and since
|z| ≤ 1, we have that

∞ ∞ ∞
1 − En (z)
z n +1
= ∑ bk z k ≤ ∑ bk z k ≤ ∑ (|bk | · 1)
k =0 k =0 k =0
∞
1 − En (1)
= ∑ bk = φn (1) = 1
= 1.
k =0

Hence for |z| ≤ 1, we have the desired inequality,

|1 − En (z)| ≤ |z|n+1

1 Theorem 43 (Theorem 21)

Let { an }n∈N ⊆ C with an → ∞ as n → ∞, and ∀n ∈ N, an 6= 0. Then

the Weierstrass Product
∞  
z
f (z) = ∏ En
an
(32.1)
n =1

is entire and has zeros at exactly { an } and nowhere else.

Ò Proof
Let R > 0 and Ω = D (0, R) ⊆ C. Let z ∈ Ω. ∵ | an | → ∞,
∃ N0 > 0 ∀n ≥ N0 | an | ≥ 2R. Note
 
∞   ∞     
z z
∏ E = ∏  1 + E − 1
 
n n
a a
 
n= N0 n n= N0 n 
| {z }
call this un (z)

On Ω, since |z| ≤ R ∧ ∀n ≥ N0 | an | ≥ 2R, by 1 Theorem 42 we

168 Lecture 32 Apr 04 2018 - Infinite Products (Continued 2)

have that ∀n ≥ N0 ∧ ∀z ∈ Ω,
  n +1
z z z 1
≤ 1 =⇒ En −1 ≤ ≤
an an an 2n +1

Then by 1 Theorem 40 and its remark, since ∑∞

n=1 un ( z ) converges
absolutely and uniformly on Ω, the infinite product ∏∞
n= N0 (1 + un ( z ))
converges.
Exercise To prove that the infinite product converges uniformly on
Ω, it suffices to show that it satisfies the Cauchy criterion: Define f n :=
∑∞
j=n (1 + u j ( z )). Note that for m > n ≥ N0 ,

m −1
∑


| fm − fn | = 1 + u j (z)
j=n
m −1
∑


≤ 1 + u j (z) by 2 of Lemma 39
j=n
m −1
≤ exp ∑ u j (z) by 1 of Lemma 39
j=n

So ∀ε > 0, since ∑∞
j=1 | un ( z )| converges, ∃ N0 > 0 ∀ m > n ≥ N0 , such
4 4
choose an N0 different from the earlier
N0 if necessary
that
m −1
∑ |un (z)| ≤ ε
j=n

and hence
| f m − f n | ≤ eε ,

thus satisfying the Cauchy criterion and hence showing that the infinite
product is uniformly convergent.

Exercise We now need to show that the infinite product is a holomor- A more complete statement and proof
of this statement is available in Stein &
phic function on Ω. Note that by construction of En , it has no poles, and
Shakarchi’s Complex Analysis (Chapter
 in C that is
is entire. Note that we have that the { f n }n∈N is a sequence 2, Section 5.2, Theorem 5.2).
uniformly convergent to the function f (z) = ∏∞ n=1 En an in Ω. Let
z

D be any disc whose closure5 is contained in Ω, and T any triangle in D. 5

D := {K ∈ C : K closed ∧ D ⊆ K }
Then by Goursat’s Theorem,
Z
f n = 0.
T

Since { f n }n∈N converges uniformly to f , f is continuous6 and 6

This is a relatively easy proof that can
Z Z be done using techniques from Real
Analysis.
fn → f.
T T
Exercise 32.1.1
R Prove that f is continuous under the
Thus T f = 0. Then by Morera’s Theorem, we have that f ∈ H ( D ) and
supposition that { f n }n∈N is uniformly
convergent.
 PMATH352W18 - Complex Analysis 169

since D is arbitrary, f ∈ H (Ω) as required.

Now since R is arbitrary, we have that f ∈ H (C), i.e. f is entire.

To find the zeros of f , let z ∈ Ω. By 1 Theorem 40 and in particular

its remark, we have that

∞
!  
z z
∏ Ej a j = 0 ⇐⇒ En an = 0 for some n ≥ N0 ,
j= N 0

Therefore,

∞
!  
z z
∏ EN aj
= 0 ⇐⇒ En
an
= 0 ⇐⇒ z = an
j =1

for some n ∈ N. This shows that

Z ( f ) = { an } ∩ Ω,

and since R > 0 is arbitrary, we have that Z ( f ) = { an : n ∈ N}. This

completes the proof. 

Example 32.1.1

Construct an entire function with simple zeros at z = 0, ±1, ±2, ..., i.e.
z ∈ Z.

Solution 1: Define f 1 (z) = ∏∞ z

n=1 En n , which is entire and has zeros
at exactly N by 1 Theorem 43, and f 2 (z) = f 1 (−z), which would have
zeros at −N. Then simply choose the function

f (z) = z · f 1 (z) · f 2 (z)

so that we include the root z = 0.

Solution 2: We can also choose

∞   z 2  Exercise 32.1.2
f (z) = z · ∏ 1−
n Check that f indeed has its zeros at all
n =1
integers.

Note that sin(πz) is entire and has simple zeros at exactly z ∈ Z. Then

sin(πz)
has no zeros or poles ∀z,
f (z)
and therefore by 1 Theorem 38, ∃ g ∈ H (C), such that

sin(πz)
= e g(z) .
f (z)

By the Product Formula of Sine7 , 8 , we have that 7

First proven by Euler, although incom-
plete.
∞
z2 It turns out g(z) = log π.
  8

sin(πz) = πz ∏ 1− 2
n =1 n

and hence
∞
z2
 
sin(πz)
f (z) = ∏ 1−
n2
=
π
.
n =1
33 Index

analytic, 91 Euler’s Identity, 162 Open Mapping Theorem, 10, 141

analytic continuation, 143
argument, 23, 135 Functional Equation, 152 partial product, 156
Argument Principle, 10, 136 Fundamental Theorem of Algebra, 96, point/isolated singularity, 107
140 pole, 109
boundary of C, 82 Fundamental Theorem of Arithmetic, pole of order n, 113
162 poles, 109
Casorati-Weierstrass, 10, 115 Fundamental Theorem of Calculus, 9, power series, 49
Cauchy’s Integral Formula, 9, 82, 88 71 Principal Branch, 149
Cauchy’s Residue Theorem, 10, 120 principal part, 114
Cauchy’s Residue Theorem - General- Gamma Function, 151 Product Formula of Sine, 170
ized, 10, 123 Goursat’s Theorem, 9, 73
Cauchy’s Theorem for Convex Set, 9, radius of convergence, 51
81 Haar measure, 151 real part, 11
Cauchy-Riemann Equations, 9, 45 homotopic, 144 removable singularities, 109
closed, 72 Homotopy, 8, 144 removable singularity, 108, 111
complex number, 11 residue, 114
complex plane, 11 imaginary part, 11 Riemann Zeta Function, 162
conjugate of z, 14 infinite product, 156 Rouché’s Theorem, 10, 137
continuous, 39 Infinite Products, 8, 156
contour, 62 infinite products, 156
simple zero, 112
converges, 37 interior, 73
simply connected, 145
convex set, 81
Simply Connected Domain, 8, 145
limit supremum, 51
deMoivre’s formula, 26
deMoivre’s Law, 26 modulus, 15 uniform convergence, 162
differentiable/holomorphic, 42 Monic Polynomial, 8, 140
Monomial, 8, 140 Weierstrass Product, 167
entire, 56 Morera’s Theorem, 10, 101 winding number, 82
equivalent parameterizations, 61 multiplicative inverse, 13 Winding Number Theorem, 10, 103
essential singularity, 109
Euler’s formula, 23 neighbourhood, 42 zero of order n, 112