MA 530 Complex Analysis: Review

Yingwei Wang

Department of Mathematics, Purdue University, West Lafayette, IN, USA
Contents
1 How to prove analytic (holomorphic, complex dierentiable)? 3
1.1 Denition (Dierence Quotient) . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Cauchy-Riemann Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Integration along closed curves equals zeros . . . . . . . . . . . . . . . . . . . . . 3
1.4 Power series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4.1 From holomorphic to power series . . . . . . . . . . . . . . . . . . . . . . 4
1.4.2 From power series to holomorphic . . . . . . . . . . . . . . . . . . . . . . 5
1.4.3 Dierentiate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Integration along curves 6
2.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Fundamental Theorem of Calculus . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Cauchy Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.4 Cauchy Integral Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.5 Liouvilles theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.6 Fundamental Theorem of Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3 Useful properties of holomorphic functions 13
3.1 Isolated zeros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Identity theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3 Averaging property . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Maximum principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.5 Rouches Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.6 Argument Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

This is based on the lecture notes of Prof Buzzard (Fall 2010) and Prof Bell (Spring 2011).
1
Yingwei Wang Complex Analysis
4 Harmonic functions 18
4.1 Denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2 Harmonic conjugate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.3 Poisson integral formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.4 Mean Value Property and Maximum Principle . . . . . . . . . . . . . . . . . . . 20
4.5 Zeros of harmonic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5 Isolated singularity 21
5.1 Denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.1.1 Removable singularity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.1.2 Pole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.1.3 Essential singularity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.1.4 From the view of power series . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2 Riemann removable singularities theorem . . . . . . . . . . . . . . . . . . . . . . 23
5.2.1 Method I: integral formulas + estimates . . . . . . . . . . . . . . . . . . 23
5.2.2 Method II: construct function . . . . . . . . . . . . . . . . . . . . . . . . 24
5.3 Pole Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.4 Casorati-Weierstrass Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.5 Meromorphic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.5.1 Polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.5.2 Rational functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6 Uniform convergence 29
6.1 Hurwitz Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.2 Montel Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.2.1 Normal Family . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.2.2 Uniform boundedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.2.3 Equicontinuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.2.4 Montel Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.3 Open mapping theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7 Univalence 31
7.1 Local univalence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
7.2 Global univalence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
7.3 Limits univalence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
8 Conformal mappings 33
8.1 Schwartz Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
8.2 Automorphism of the disc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
8.3 From upper half plan to the unit disc . . . . . . . . . . . . . . . . . . . . . . . 34
9 Roots of functions 35
2
Yingwei Wang Complex Analysis
1 How to prove analytic (holomorphic, complex dier-
entiable)?
Note: let be an open set in C and f be a complex-valued function on .
1.1 Denition (Dierence Quotient)
Denition 1.1. Say f is complex dierentiable (holomorphic) at z
0
, if
DQ =
f(z
0
+ h) f(z
0
)
h
converges to a limit when h 0. Call the limit f

(z
0
).
If f is complex dierentiable at all points in , then call f holomorphic on .
Remark 1.1. It should be emphasized that in the above limit, h is a complex number that may
approach 0 from any direction.
Remark 1.2. A holomorphic function will actually be innitely many times complex dieren-
tiable, that is, the existence of the rst derivative will guarantee the existence of derivatives of
any order. For proof, see Section 1.4.3 and Theorem 2.4.
For more application of this method, see Lemma 2.3, Theorem 2.2, Theorem 2.4.
1.2 Cauchy-Riemann Equations
Theorem 1.1 (C-R analytic). Suppose u, v C
1
() and satisfy the C-R equations
_
u
x
= v
y
,
u
y
= v
x
,
(1.1)
then f(x, y) = u(xy) + iv(x, y) is analytic on .
Remark 1.3. Actually, by Looman-Moncho Theorem, we just need that u, v are continuous
and all their rst partial derivatives exit (may be not continuous) and satisfy the C-R equations,
then f = u + iv is analytic.
1.3 Integration along closed curves equals zeros
Theorem 1.2 (Morera). Suppose f is continuous on an open set and for any triangle T
contained in ,
_
T
f(z)dz = 0,
then f is holomorphic.
3
Yingwei Wang Complex Analysis
1.4 Power series
1.4.1 From holomorphic to power series
Lemma 1.1. Let
S
N
(z) = 1 + z + + z
N
, (1.2)
E
N
(z) =
z
N+1
1 z
, (1.3)
then
1
1 z
= S
N
(z) + E
N
(z).
Furthermore, if |z| < < 1, then |E
N
(z)|

N+1
1
.
Theorem 1.3 (holomorphic analytic). Suppose f is holomorphic on and D
r
(z
0
) .
Then f has a power series expansion in D
r
(z
0
),
f(z) =

n=0
a
n
(z z
0
)
n
, z D
r
(z
0
),
with a
n
=
f
(n)
(z
0
)
n!
=
1
2i
_
Cr(z
0
)
f(w)
(wz
0
)
n+1
dw.
Proof. Without loss of generality, we can take z
0
= 0 and < r. By Cauchy formula (Thm
2.3) and Lemma 1.1,
f(z) =
1
2i
_
Cr
f(w)
w z
dw
=
1
2i
_
Cr
f(w)
w
1
1
z
w
dw
=
1
2i
_
Cr
f(w)
w
S
N
(z/w) dw +
1
2i
_
Cr
f(w)
w
E
N
(z/w) dw
=
N

n=0
_
1
2i
_
Cr
f(w)
w
n+1
dw
_
z
n
+
N
(z),
where
|
N
(z)|
1
2
sup
Cr
|f|
r
(/r)
N+1
1 /r
(2r) 0.
as N .
Besides,
a
n
=
1
2i
_
Cr
f(w)
w
n+1
dw =
f
(n)
(0)
n!
.
4
Yingwei Wang Complex Analysis
1.4.2 From power series to holomorphic
Denition 1.2. Say {f
n
} converges uniformly on compact subsets of to f, if for any compact
subsect K , and > 0, there is an N such that |f
n
(z) f(z)| < , z K, n > N.
Remark 1.4. Power series converge uniformly on compact subsects inside the circle.
Theorem 1.4 (uniformly limit analytic). Suppose {f
n
} analytic on and converges uni-
formly on compact subsets of to f. Then f is analytic.
Consequently, power series converge to analytic functions.
Proof. Let D be any disc whose closure is contained in and T be any triangle in that disc.
Then, since each f
n
is holomorphic, Goursats theorem (Lemma 2.4) implies
_
T
f
n
(z) dz = 0, n.
By assumption, f
n
f uniformly on

D, so f is continuous and
_
T
f
n
(z)dz
_
T
f(z) dz.
As a result, we nd
_
T
f(z)dz = 0 for T D. By Morera theorem (Thm 1.2), we conclude
that f is holomorphic in D. Since this conclusion is true for every D whose closure is contained
in , we nd that f is holomorphic in all of .
1.4.3 Dierentiate
Theorem 1.5 (Can dierentiate power series term by term). Suppose
f(z) =

n=0
a
n
(z z
0
)
n
with radius of convergence R. Then power series for f

has the same radius of convergence and

f

(z) =

n=0
na
n
(z z
0
)
n1
.
Theorem 1.6 (Derivative convergence). Suppose {f
n
} analytic on and converges uniformly
on compact subsets of to f. Then {f
(k)
n
} converges uniformly on compact subsets of to
f
(k)
.
5
Yingwei Wang Complex Analysis
2 Integration along curves
2.1 Preliminaries
Denition 2.1. Given a curve C with para z(t) : [a, b] C. Suppose f : C is
continuous. Then
_

f(z)dz =
_
b
a
f(z(t))z

(t)dt = lim
|z|0
N1

j=0
f(z
j
)(z
j+1
z
j
),
where z
j
= z
j+1
z
j
, |z| = max
j
|z
j
|.
Lemma 2.1 (Basic estimate).

f(z)dz

length () sup
z
|f(z)|
where length () =
_
b
a
|z

(t)| dt.
Lemma 2.2 (Reverse orientation). If

is with reverse orientation, then

_

f(z)dz =
_

f(z)dz.
Proof. Let be parameterized by z(t) : [a, b] C and

parameterized by z

(t) : [a, b] C.
The relationship between z(t) and z

(t) is z

(t) = z(a + b t).

Let i = 0, , n, x =
ba
n
, x
i
= a+ix, x
0
= a, x
n
= b, and y
i
= a+b x
i
, y = x, y
0
=
b, y
n
= a. By the denition of integration along curves (Def 2.1),
_

f(z)dz =
_
b
a
f(z

(t))(z

(t)dt
= lim
n
n

i=1
f(z

(x
i
))(z

(x
i
)x
= lim
n
n

i=1
f(z(y
i
)) (z

(y
i
)) y
=
_
b
a
f(z(t))z

(t)dt
=
_

f(z).
6
Yingwei Wang Complex Analysis
2.2 Fundamental Theorem of Calculus
Denition 2.2. A primitive for f on is a function F that is holomorphic on and such
that
F

(z) = f(z), z .
Theorem 2.1 (Fundamental Theorem of Calculus #2). Let be an open set in C and be a
curve (or path) in that begins at w
1
and ends at w
2
.
Version I: If f is analytic on , then
_

(z)dz = f(w
2
) f(w
1
).
Version II: If f has a primitive in , then
_

f(z)dz = F(w
2
) F(w
1
).
Proof. Chain Rule + Fundamental Theorem of Calculus # 1.
Corollary 2.1. If is a closed curve and f is holomorphic, then
_

f(z)dz = 0.
Corollary 2.2. Any two primitives of f (if they exist) dier by a constant.
Proof. Suppose both F and G are the primitives of function f. According to Thm 2.1 (version
II), we know that if is a curve in from w
0
to w, then
_

f(z)dz = F(w) F(w

0
) = G(w) G(w
0
).
Fix w
0
, then w , F(w) G(w) = F(w
0
) G(w
0
) = constant.
Corollary 2.3. If is a region (open+connected), f is complex dierentiable at each point in
, and f

(z) = 0 for all z , then f is a constant.

Proof. Method I: Path connected + Theorem 2.1.
Fix a point w
0
. It is suces to show that f(w) = f(w
0
) for all w .
Since is connected, for any w , there exists a curve which joins w
0
to w. By Thm
2.1 (version I),
_

(z)dz = f(w) f(w

0
).
By assumption, f

= 0 so the integral on the left is 0 and we conclude that f(w) = f(w

0
) as
desired.
7
Yingwei Wang Complex Analysis
Method II: Denition + C-R equations.
f

(z) = 0,

f
z
=
1
2
_
f
x
+
1
i
f
y
_
= 0,
(u
x
+ iv
x
) i(u
y
+ iv
y
) = 0, since f
x
= u
x
+ iv
x
, f
y
= u
y
+ iv
y
,
(u
x
+ v
y
) + i(v
x
u
y
) = 0,
u
x
+ v
y
= 0, v
x
u
y
= 0.
Besides, by C-R Eqn.(1.1), we can get that
u
x
= u
y
= v
x
= v
y
= 0
u(x, y) = constant, v(x, y) = constant
f = constant.
2.3 Cauchy Theorem
Lemma 2.3 (Deferential under integral). Suppose that (z) is continuous function on the trace
of a path . Prove that the function
f(z) =
_

(w)
w z
dw,
is analytic on C \ .
Idea: just need to show that
f

(z) =
_

(w)
(w z)
2
dw. (2.1)
Proof. Recall the DQ method, for z
0
C \ ,
f(z) f(z
0
)
z z
0

(w)
(w z)
2
dw
=
_

(w)
_
1
(w z)(w z
0
)

1
(w z
0
)
2
_
dw
=
_

(w)
_
z z
0
(w z)(w z
0
)
2
_
dw
= (z z
0
)
_

_
(w)
(w z)(w z
0
)
2
_
dw.
8
Yingwei Wang Complex Analysis
Let
E(z) =
_

_
(w)
(w z)(w z
0
)
2
_
dw,
D = dist (z
0
, tr()) = min
wtr ()
|z
0
w|,
M = max
wtr()
|(w)|.
Choose a disc D
D/2
(z
0
), z D
D/2
(z
0
), w , we have
|z w| D/2, |z
0
w| D/2.
By Lemma 2.1, we have
|E(z)| M
1
(D/2)(D/2)
2
length (),

f(z) f(z
0
)
z z
0

(w)
(w z)
2
dw

|z z
0
|
M
(D/2)(D/2)
2
length () 0, as z z
0
.
It implies Eq.(2.1).
Lemma 2.4 (Goursat). If is an open set in C, and T a triangle whose interior is also
contained in , then
_
T
f(z)dz = 0,
whenever f is continuous on and analytic on \{p}.
Theorem 2.2 (Cauchys theorem on a convex open set). Suppose is convex and open, p .
If f is continuous on and analytic on \{p}, then
_

fdz = 0,
for any closed in .
Idea: Construct F holomorphic with F

= f, Then f is continuous so F is continuous and

by Thm 2.1,
_

fdz =
_

dz = F(end) F(start) = 0. (2.2)

9
Yingwei Wang Complex Analysis
Proof. Fix a . Let L
z
a
be the line segement from a to z. is convex indicates that L
z
a
is
contained in .
Dene
f(z) =
_
L
z
a
f(w)dw.
Fix z
0
and consider z near z
0
. By Lemma 2.4,
_
_
L
z
0
a
+
_
L
z
z
0
+
_
L
z
a
_
f(w)dw = 0,
F(z
0
) +
_
L
z
z
0
f(w)dw F(z) = 0,

F(z) F(z
0
)
z z
0
=
1
z z
0
_
L
z
z
0
f(w)dw.
Since f is continuous at z
0
, then
f(z) = f(z
0
) + R(z),
with R(z) 0, as z z
0
. So for > 0, > 0 so that if |z z
0
| < then |R(z)| < .
Also,
_
L
z
z
0
f(z
0
)dw =
_
L
z
z
0
d
dw
[f(z
0
)w]dw = f(z
0
)(z z
0
).
Hence,
F(z) F(z
0
)
z z
0
=
1
z z
0
_
L
z
z
0
(f(z
0
) + R(w))dw = f(z
0
) +
1
z z
0
_
L
z
z
0
R(w)dw,

F(z) F(z
0
)
z z
0
f(z
0
)

sup
wL
z
z
0
|R(w)| 0, as z z
0
.
It indicates that F

(z
0
) = f(z
0
). Then F

= f. By Eq.(2.2), we can get the conclusion.

2.4 Cauchy Integral Formula
Theorem 2.3 (Cauchy Integral Formula on a disc). Suppose f is holomorphic on D
R
(z
0
) and
0 < r < R. Then a D
r
(z
0
),
f(a) =
1
2i
_
Cr(z
0
)
f(z)
z a
dz. (2.3)
10
Yingwei Wang Complex Analysis
Proof. Let
G(z) =
_
f(z)f(a)
za
z = a,
f

(a) z = a.
(2.4)
It is easy to know that G(z) is continuous and G(z) is holomorphic on D
r
(z
0
)\{a}. By Thm
2.2, we have
_
Cr(z
0
)
G(z)dz = 0,

_
Cr(z
0
)
_
f(z)
z a

f(a)
z a
_
dz = 0,

_
Cr(z
0
)
f(z)
z a
dz =
_
Cr(z
0
)
f(a)
z a
dz = 2if(a).
Theorem 2.4 (Cauchy Integral Formula with derivatives). If f is holomorphic in an open
set , then f has innitely many complex derivatives in . Moreover, if D
r
(z
0
) , then
z D
r
(z
0
),
f
(n)
(z) =
n!
2i
_
Cr(z
0
)
f(w)
(w z)
n+1
dw. (2.5)
Proof. Here, we give three methods to prove this.
Method I Induction on n and by Def 1.1.
n = 0, by Thm 2.3,
f(z) =
1
2i
_
C
f(w)
w z
dw.
Suppose f has desideratives 0, 1, , n 1 and the formula (2.5) is ture. Then
DQ =
f
(n1)
(z + h) f
(n1)
(z)
h
=
(n 1)!
2i
_
C
f(w)
h
_
1
(w z h)
n

1
(w z)
n
_
dw
Let A =
1
wzh
, B =
1
wz
, then
AB =
h
(w z h)(w z)
,
A
n
B
n
= (AB)(A
n1
+ A
n2
B + + B
n1
).
11
Yingwei Wang Complex Analysis
So we have
lim
h0
DQ
= lim
h0
(n 1)!
2i
_
C
f(w)
(w z h)(w z)
(A
n1
+ A
n2
B + + B
n1
)dw,
=
(n 1)!
2i
_
C
lim
h0
f(w)
(w z h)(w z)
(A
n1
+ A
n2
B + + B
n1
)dw,
=
n!
2i
_
C
f(w)
(w z)
n+1
dw.
Method II Induction on n and dierential under integral (Lemma 2.3).
d
dz
f
(n1)
(z)
=
(n 1)!
2i
_
C
_
d
dz
f(w)
(w z)
n
_
dw
=
n!
2i
_
C
f(w)
(w z)
n+1
dw.
Method III Power series expansion By Eq.(2.3),
f(z) =
1
2i
_
Cr(z
0
)
f(w)
w z
dw.
Do power expansion
1
w z
=
1
(w z
0
) (z z
0
)
=
1
w z
0
1
1
zz
0
wz
0
=
1
w z
0

n=0
_
z z
0
w z
0
_
n
.
Then
f(z) =
1
2i
_
Cr(z
0
)
f(w)
w z
0

n=0
_
z z
0
w z
0
_
n
dw
=

n=0
_
1
2i
_
Cr(z
0
)
f(w)
(w z)
n+1
dw
_
(z z
0
)
n
.
It implies Eq.(2.5).
12
Yingwei Wang Complex Analysis
2.5 Liouvilles theorem
Theorem 2.5 (Cauchy inequality). If f is holomorphic in an open set that contains the closure
of a disc D centered at z
0
and of radius R, then
|f
(n)
(z
0
)|
n!f
C
R
n
,
where f
C
= sup
zC
|f(z)| denotes the supremum of |f| on the boundary circle C.
Theorem 2.6 (Liouvilles theorem). If f is entire and bounded, then f is constant.
2.6 Fundamental Theorem of Algebra
Lemma 2.5 (Basic polynomial estimate). Suppose
p(z) = a
N
z
N
+ a
N1
z
N1
+ + a
1
z + a
0
,
is a complex polynomial of degree N. Then there exist constants 0 < A < B and a radius R
such that
A|z|
N
|p(z)| B|z|
N
, if |z| > R. (2.6)
Remark 2.1. Here, 0 < A < |a
N
| < B and A, B can be as close to a
N
as desired.
Theorem 2.7 (Fundamental Theorem of Algebra). Every non-constant polynomial p(z) with
complex coecients has a root in C.
Proof. Assume that p(z) = 0, z C. By Eq.(2.6), we know that f(z) =
1
p(z)
is bounded
entire. And by Liouville Theorem (Thm 2.6), f is constant, so p(z) is constant, which is a
contradiction.
Corollary 2.4. Every polynomial p(z) of degree n 1 has precisely n roots in C. If these roots
are denoted by w
1
, w
2
, , w
n
, then p(z) can be factored as
p(z) = a
n
(z w
1
)(z w
2
) (z w
n
).
3 Useful properties of holomorphic functions
3.1 Isolated zeros
Theorem 3.1 (Zero theorem). Suppose f is holomorphic on and f(z
0
) = 0, z
0
. If
r > 0 such that D
r
(z
0
) and f(z) is not identically 0 on D
r
(z
0
), then n
0
Z
+
and a
function h(z) which is holomorphic on D
r
(z
0
) so that h(z
0
) = 0 and
f(z) = (z z
0
)
n
0
h(z), z D
r
(z
0
).
13
Yingwei Wang Complex Analysis
Proof. From Theorem 1.3, we know that
f(z) =

n=0
a
n
(z z
0
)
n
, z D
r
(z
0
),
with a
n
=
f
(n)
(z
0
)
n!
.
Since f is not identically 0 on D
r
(z
0
), we know that n such that a
n
= 0. Also, f(z
0
) =
0 a
0
= 0.
Let n
0
= min{n Z
+
: f
(n)
(z
0
) = 0}, then
f(z) = (z z
0
)
n
0

n=n
0
a
n
(z z
0
)
nn
0
.
Let h(z) =

n=n
0
a
n
(z z
0
)
nn
0
which is holomorphic on D
r
(z
0
). Also, h(z
0
) = a
n
0
=
f
n
0 (z
0
)
n!
= 0. Then we get the conclusion.
Corollary 3.1. Suppose f and g are analytic on a domain and that f
2
= g
2
on . Prove
that either f = g or f = g on .
Proof. Choose F = f + g, G = f g. Assume both of F and G do not vanish on . It means
that F and G are not identically zero on . It also means that F and G only have isolated
zeros in .
f
2
= g
2
F(z)G(z) = 0
z
0
such that either F(z
0
) = 0 or G(z
0
) = 0.
Without loss of generality, let F(z
0
) = 0. Since z
0
is an isolated zero of F, r
1
> 0 such that
F(z) = 0 on

D
r
1
(z
0
) = D
r
1
(z
0
) \ {z
0
}. Besides, G also only has isolated zeros, so we can choose
a

D
r
1
(z
0
), such that G(a) = 0. r
2
> 0 such that G(z) = 0 on D
r
2
(a) and D
r
2
(a)

D
r
1
(z
0
).
Then
F(z)G(z) = 0, z D
r
2
(a),
which contradicts with the assumption.
3.2 Identity theorem
Theorem 3.2 (Identity theorem). Suppose f : C is holomorphic and Z
f
= {z :
f(z) = 0}. Then either Z
f
= or Z
f
has no limit points in .
Proof. Step 1: Disc version.
Step 2: Let U be the interior of Z
f
. Then U is open and nonempty.
Step 3: Let V = \U, then V is also open.
Step 4: Since is connected and = U V , U is not empty, we know that V = and
= Z
f
.
14
Yingwei Wang Complex Analysis
Corollary 3.2. Suppose f and g are holomorphic in a region and f(z) = g(z) for all z in
some non-empty open subsets of (or more generally for z in some sequence of distinct points
with limit point in ). Then f(z) = g(z) throughout .
Corollary 3.3. Only one way to extend e
x
and trig functions to C. Besides, any trig identity
holds for complex angles.
Remark 3.1. An analytic function may have innitely many (at most countable) zeros on a
bounded domain as soon as the limit point of these zeros is not in this domain. For example,
f(z) = sin
_
1
1z
_
has innitely many zeros on the open unit disc D
1
(0), i.e. z
k
= 1
1
k
, but
z
k
1

D
1
(0). So each {z
k
} is also isolated zero. Besides, z = 1 is the essential singularity of
f(z).
3.3 Averaging property
Lemma 3.1 (Averaging property). f is holomorphic on and D
r
(z
0
) , then
f(z
0
) =
1
2
_
2
0
f(z
0
+ re
it
)dt (3.1)
Proof. Let C
r
be a circle in , centered at z
0
with radius r. If we parameterize C
r
by z =
z
0
+ re
it
, 0 t 2, then by Cauchy integral formula,
f(z
0
) =
1
2i
_
Cr
f(z)
z z
0
dz
=
1
2i
_
2
0
f(z
0
+ re
it
)
re
it
ire
it
dt
=
1
2
_
2
0
f(z
0
+ re
it
) dt
3.4 Maximum principle
Theorem 3.3 (Maximum principle #1). Suppose f is holomorphic on a domain . If |f|
attains a local maximum at a point in , then f constant.
Proof. Method I By Open mapping theorem
Suppose |f| has a local max at z
0
. Let w
0
= f(z
0
). Then r > 0 such that D
r
(z
0
)
and |f(z)| |f(z
0
)| for z D
r
(z
0
).
By Open Mapping Theorem (Thm 6.3), if f is nonconstant, f(D
r
(z
0
)) contains a disc about
f(z
0
). But there are points in such a disc with modulus bigger than |w
0
|. This is a contradiction.
Method II By averaging property
15
Yingwei Wang Complex Analysis
Suppose z
0
and that for all z , |f(z)| |f(z
0
)|. By Eq.(3.1), we know that
|f(z
0
)|
1
2
_
2
0
|f(z
0
+ re
it
)| dt. (3.2)
However, the assumption |f(z
0
)| |f(z)| for all z implies that
1
2
_
2
0
|f(z
0
+ re
it
)| dt
1
2
_
2
0
|f(z
0
)| dt = |f(z
0
)|. (3.3)
From (3.2)-(3.3), we can get
|f(z
0
)| =
1
2
_
2
0
|f(z
0
+ re
it
)| dt. (3.4)
It follows that
0 = |f(z
0
)|
1
2
_
2
0
|f(z
0
+ re
it
)| dt
=
1
2
_
2
0
(|f(z
0
)| |f(z
0
+ re
it
)|) dt,
which means |f(z
0
)| = |f(z
0
+ re
it
)| for all t [0, 2]. That is to say |f(z)| = |f(z
0
)| for all
z C
r
. Since r is arbitrary, it follows that |f(z)| = |f(z
0
)| for all z . Since f is holomorphic
and |f(z)| is a constant, f is also a constant (by Chapter One, # 13(c), Page 28).
Theorem 3.4 (Maximum principle #2). Suppose is a bounded domain. f is continuous on

and holomorphic on . Then |f| assumes its maximum value on the boundary of .
3.5 Rouches Theorem
Theorem 3.5 (Rouches Theorem). Suppose f and g are meromorphic functions on a connected
open G C and is a piecewise C
1
closed curve in G with
(i) Ind

(w) = 0 for w C \ G;
(ii) no zeros or poles of f or g on ;
(iii) |f(z) g(z)| < |f(z)| for all z (that is, the dierence is strictly smaller than one
of the functions |f| on ).
Then,
N
f
P
f
= N
g
P
g
, (3.5)
where
N
f
=

aG,f(a)=0
mult
a
Ind

(a),
P
f
=

bG,f(b)=
order
b
Ind

(b),
and similarly for N
g
and P
g
.
16
Yingwei Wang Complex Analysis
Here are some applications.
Corollary 3.4. There are no such sequence of polynomials that uniformly converges to f(z) =
1
z
on the circle {z : |z| = 1}.
Proof. Suppose p
n
(z)
1
z
uniformly on C
1
(0), then
|zp
n
(z) 1| = |p
n
(z)
1
z
| 0, z C
1
(0).
Then sucient large N such that
|zp
n
(z) 1| < 1, n > N, z C
1
(0).
By Rouches Theorem, N
zpn(z)
= N
1
. But zp
n
(z) has at least one zero, while 1 has no zeros,
which is a contradiction.
Corollary 3.5. There are no such sequence of polynomials that uniformly converges to f(z) =
( z)
2
on the circle {z : |z| = 1}.
Proof. Suppose p
n
(z) ( z)
2
uniformly on C
1
(0), then
|z
2
p
n
(z)| |z
2
( z)
2
| = |z|
4
= 1, z C
1
(0).
Then do the similar thing as last corollary.
3.6 Argument Principle
Theorem 3.6 (Argument principle). Let C be a simple closed path. Suppose that f(z) is ana-
lytic and nonzero on C and meromorphic inside C. List the zeros of f inside C as z
1
, z
2
, , z
k
with multiplicities N
1
, , N
k
, and Z
C
=

k
i=1
N
i
. List the poles of f inside C as w
1
, w
2
, , w
l
with orders M
1
, , M
l
, and P
C
=

l
j=1
M
j
. Then
Z
C
P
C
=
1
2

C
arg f(z) (3.6)
=
1
2i
_
C
f

()
f()
d. (3.7)
Remark 3.2. From Eqs.(3.6)-Eq.(3.7), we know that

C
arg f(z) =
1
i
_
C
f

()
f()
d. (3.8)
This formula is always true even if the curve C is not a closed path.
17
Yingwei Wang Complex Analysis
Corollary 3.6. How many zeros does the polynomial
f(z) = z
1998
+ z + 2001
have in the rst quadrant?
Proof. Chose a closed curve:
= L
R
0
+ C
+
R
+ L
0
iR
,
where C
+
R
here means the circle in the rst quadrant and R is sucient large.
On one hand, on L
R
0
and L
0
iR
, we know that

L
R
0
arg f(z) = 0, (3.9)

L
0
iR
arg f(z) = . (3.10)
Hint for Eq.(3.10): if z = iR, then f(z) = R
1
998 +iR+2001. And consider R from to
0.
On the other hand, on C
+
R
, we know that
_
C
+
R
f

()
f()
d =
_
C
+
R
1998
1997
+ 1

1998
+ + 2001
d

_
C
+
R
1998

d
= 1998
2i
4
= 999i. (3.11)
By Thm. 3.6 and Eqs.(3.9)-(3.11), we know that the number of zeros of f(z) in the rst
quadrant is
999
2
= 499.
4 Harmonic functions
4.1 Denition
The following denitions of harmonic functions are equivalent:
Denition 4.1. Say u : R is harmonic if u C

and u 0.
Denition 4.2. Say u : R is harmonic if u C
2
and u 0.
Denition 4.3. Say u : R is harmonic if u is locally the real (imaginary) part of a
holomorphic function.
Denition 4.4. Say u : R is harmonic if u is continuous on and u
x
, u
y
, u
xx
, u
yy
exist
and u = 0 on .
Remark 4.1. Here, =

2
x
2
+

2
y
2
is called Laplacian operator and u 0 is called Laplacian
equation.
18
Yingwei Wang Complex Analysis
4.2 Harmonic conjugate
Theorem 4.1. A harmonic function on a simply connected domain has a global harmonic
conjugate.
Remark 4.2. Suppose u be harmonic and f be the analytic function with Ref = u. Then,
f

= u
x
iu
y
is also analytic.
4.3 Poisson integral formula
Denition 4.5 (Poisson kernel).
P(z, t) =
1
2
1 |z|
2
|e
it
z|
2
=
1
2
Re
_
e
it
+ z
e
it
z
_
, (4.1)
P(re
i
, t) =
1
2
1 r
2
1 2r cos( t) + r
2
. (4.2)
Theorem 4.2 (Dirichlet problem). Suppose u is continuous on D
1
(0) and harmonic on D
1
(0).
Then
u(z) =
_
2
0
P(z, t)u(e
it
)dt.
Lemma 4.1. Some properties about Poisson kernel:
(i) P(a, t) > 0 if a D
1
(0), t [0, 2].
(ii)
_
2
0
P(a, t)dt = 1 for any a D
1
(0).
(iii) P(z, t) 0 as |z| 1.
(iv) P(z, t) is harmonic in z.
Theorem 4.3 (Convergence). Suppose {u
m
}

m=1
is a sequence of harmonic functions on such
that u
m
converges uniformly to a function u on each compact subset of . Then u is harmonic
on .
Moreover, for every multi-index , D

u
m
converges uniformly on each compact subset of .
Proof. Given D
r
(a) , we need only show that u is harmonic on D
r
(a). Without loss of
generality, we assume D
r
(a) = D
1
(0).
By Thm 4.2, we know that
u
m
(z) =
_
2
0
P(z, t)u
m
(e
it
)dt,
for z D
1
(0) and m. Taking the limit of both sides, we obtain
u(z) =
_
2
0
P(z, t)u(e
it
)dt,
for z D
1
(0). Thus, u is harmonic on D
1
(0).
19
Yingwei Wang Complex Analysis
4.4 Mean Value Property and Maximum Principle
Theorem 4.4 (Mean Value Property). Since holomorphic f = u + iv satises the averaging
property, take the real part of Eq.(3.1), we can get the mean value property of harmonic u.
More precisely,
u(z) =
1
2
_
2
0
u(z + re
i
)d.
An important sequence of the mean value property is the following maximum principle for
harmonic functions.
Theorem 4.5 (Maximum Principle). Suppose is connected, u is real valued and harmonic
on , and u has a maximum or minimum in . Then u is a constant.
The following corollary is frequently useful. Note that the connectivity of is not needed
here.
Corollary 4.1. Suppose is bounded, u is a continuous real valued function on

that is
harmonic on . Then u attains its maximum and minimum values over

on .
The next corollary is a version of maximum principle for complex valued functions.
Corollary 4.2. Let be connected and u be harmonic on . If |u| has a maximum in , then
u is a constant.
Remark 4.3. For holomorphic cases, see Thm 1.4 and Thm 1.6.
Theorem 4.6 (Converse of the mean value property #1). A continuous function that satises
the mean value property must be harmonic.
Proof. Step I: Poisson Kernel
Let
v(z) =
_
2
0
P(z, t)u(e
it
)dt,
then v(z) is harmonic and v(e
it
) = u(e
it
).
Step II: Maximum Principal
Since u(z) satises the mean value property, then u(z) also satises the maximum principal.
So uv satises the maximum principal. Besides, u v 0 on the boundary, so uv 0 for
all z D
1
(0).
Repeat the argument, we can get v u 0. Hence, u v.
Denition 4.6. Say f satises the weak mean value property if for each z
0
, > 0, such
that
u(z
0
) =
1
2
_
2
0
u(z
0
+ e
it
)dt,
for all with 0 < < .
Theorem 4.7 (Converse of the mean value property #2). A continuous function that satises
the weak mean value property must be harmonic.
20
Yingwei Wang Complex Analysis
4.5 Zeros of harmonic functions
Theorem 4.8. Let u be real-valued and harmonic function in the open set .
(1) Let A = {z : u(z) = 0}. show that A cannot be isolated.
(2) Let B = {z : u = 0}. Show that either B = or B is isolated.
Proof. (1) Suppose z
0
A and D
rn
(z
0
) , r
n
=
1
n
.
By Mean value property (Thm 4.4),
0 = u(z
0
) =
1
2
_
Crn
u(z)dz.
For each n, since u(z) is continuous on C
rn
(z
0
), there exists z
n
C
rn
(z
0
) such that u(z
n
) = 0.
Then we found a sequence {z
n
}

n=1
A such that z
n
z
0
. So A cannot be isolated.
(2) Since u is harmonic, there exists a holomorphic function f = u + iv on .
u = 0 means u
x
= 0, u
y
= 0. So we have f

(z) = u
x
iu
y
= 0, z B.
Let g(z) = f

(z), then g is also holomorphic on , and A is the set of zeros of g. So either

g(z) 0 or z
0
is isolated for z
0
B.
It follows that either B = or B is isolated.
Remark 4.4. Suppose u is a non-constant real-valued function on the whole complex plane.
Then the zero set {z C : u(z) = 0} is an unbounded set.
5 Isolated singularity
Note: let punctured domain

= \{z
0
}, and punctured disk

D
r
(z
0
) = D
r
(z
0
)\{z
0
}, where
is an open and connected domain and D
r
(z
0
) is an open disc centered at z
0
with radius r.
5.1 Denition
A point singularity of a function f is a complex number z
0
such that f is dened in a neigh-
borhood of z
0
but not at the point z
0
itself. We also call such points isolated singularities.
5.1.1 Removable singularity
Let f be holomorphic in

. If we can dene f at z
0
in such a way that f becomes holomorphic
in all of , we say that z
0
is a removable singularity of f.
5.1.2 Pole
We say that a function f dened in

D
r
(z
0
) has a pole at z
0
, if the function
1
f
, dened to be
zero at z
0
, is holomorphic in D
r
(z
0
).
21
Yingwei Wang Complex Analysis
5.1.3 Essential singularity
f oscillates and may grow faster than any power at z
0
, which is called essential singularity.
Remark 5.1. f has an isolated singularity at means f(
1
z
) has an isolated singularity at 0.
Remark 5.2. f is meromorphic on means that at each z , either f is holomorphic or f
has a pole.
5.1.4 From the view of power series
Consider the Laurent expansion of a function in the punctured disk

D
r
(z
0
):
f(z) =

n=
a
n
(z z
0
)
n
where
a
n
=
1
2i
_
Cr(z
0
)
f()
( z
0
)
d.
If
f(z) =

n=0
a
n
(z z
0
)
n
,
then f(z) has a removable singularity at z = z
0
, which means f(z) may be extended by
dening f(z
0
) = a
0
, and the resulting function is analytic in the open disk D
r
(z
0
).
If
f(z) =

n=N
a
n
(z z
0
)
n
, N > 0, a
N
= 0,
then f(z) has a zero of multiply N at z = z
0
. Near z
0
, f(z) = (z z
0
)
N
g(z), where g(z)
is analytic in D
r
(z
0
), g(z
0
) = 0.
If
f(z) =

n=M
a
n
(z z
0
)
n
, M > 0, a
M
= 0,
then f(z) has a pole of order M at z = z
0
. Near z
0
, f(z) = (z z
0
)
M
g(z), where g(z)
is analytic in D
r
(z
0
), g(z
0
) = 0.
If
f(z) =

n=
a
n
(z z
0
)
n
, a
n
= 0 for innitely many negative n,
then f(z) has a essential singularity at z = z
0
.
22
Yingwei Wang Complex Analysis
The coecient of (z z
0
)
1
is called the residue of f(z) at z
0
. Suppose z
0
is the m order
pole of f(z), then
Res
z
0
f(z) = (m1)!
d
m1
dz
m1
[(z z
0
)
m
f(z)] .
Remark 5.3. Here, from this point of view, we make the summary of Thm.1.3, Thm.3.1,
Thm.5.1, Thm.5.2 and Thm.5.3.
5.2 Riemann removable singularities theorem
Theorem 5.1. Suppose f is holomorphic and bounded in

, then f has a removable singularity
at z
0
.
That is to say, if M > 0 such that |f(z)| < M, z

D
r
(z
0
), then h(z) holomorphic in
D
r
(z
0
) and h = f on

D
r
(z
0
).
Proof: Here we give too methods to prove this.
5.2.1 Method I: integral formulas + estimates
We shall prove that for z

D
r
(z
0
), we have
f(z) =
1
2i
_
D
f()
z
d. (5.1)
By Cauchy theorem, we have
_
D
f()
z
d +
_

f()
z
d +
_

f()
z
d = 0, (5.2)
where

and

are small circles of radius with negative orientation and centered at z and z
0
respectively.
On one hand,
_

f()
z
d = 2if(z). (5.3)
On the other hand, since f is bounded and is small, stays away from z, we have

f()
z
d

C. (5.4)
By (5.2)-(5.4) and letting tend to 0, then we can get (5.1).
Now it is OK to choose h(z) =
1
2i
_
D
f()
z
d.
23
Yingwei Wang Complex Analysis
5.2.2 Method II: construct function
Consider the function g(z) dened by
g(z) =
_
0, if z = z
0
,
(z z
0
)
2
f(z), if z = z
0
.
(5.5)
By assumption, g(z) is holomorphic on

D
r
(z
0
). Next to nd g

(z
0
).
On one hand,
g(z) = g(z
0
) + 0(z z
0
) + [(z z
0
)f(z)](z z
0
). (5.6)
Note that |(z z
0
)f(z
0
)| |z z
0
|M 0 as z z
0
.
On the other hand, consider the Taylor series
g(z) = g(z
0
) + g

(z
0
)(z z
0
) + R(z)(z z
0
), (5.7)
where R(z) 0 as z z
0
.
By (5.6) and (5.7), we can know that g

(z
0
) = 0 and g(z) is holomorphic on D
r
(z
0
).
So we have
g(z) =

n=2
g
(n)
(z
0
)
n!
(z z
0
)
n
= (z z
0
)
2

n=2
g
(n)
(z
0
)
n!
(z z
0
)
n2
. (5.8)
Let h(z) =

n=2
g
(n)
(z
0
)
n!
(z z
0
)
n2
, which is holomorphic on D
r
(z
0
). Besides, by (5.5) and
(5.8), we can know that h(z) = f(z) for z

D
r
(z
0
).
5.3 Pole Theorem
Lemma 5.1. z
0
is a pole of f |f(z)| as z z
0
.
Theorem 5.2 (Pole theorem). Suppose f :

C is holomorphic and lim
zz
0
|f(z)| = .
Then n Z
+
and h(z) satisfying h(z
0
) = 0 and f(z) =
h(z)
(zz
0
)
n
Proof. Since |f(z)| = as z z
0
, r > 0 such that D
r
(z
0
) and |f(z)| > 1 on

D
r
(z
0
).
Let g(z) =
1
f(z)
, then g(z) is holomorphic and bounded on

D
r
(z
0
). By Theorem 5.1, g(z)
can extend to be holomorphic on D
r
(z
0
) by dening
g(z
0
) = lim
zz
0
1
f(z)
= 0.
Since z
0
is a isolated zero of g(z), by Theorem 3.1, there exits n Z
+
and H(z) which is
holomorphic on D
r
(z
0
) with H(z
0
) = 0 such that g(z) = (z z
0
)
n
H(z).
Then h(z) =
1
H(z)
is holomorphic in D
r
(z
0
) and
f(z) =
1
g(z)
=
h(z)
(z z
0
)
n
.
24
Yingwei Wang Complex Analysis
5.4 Casorati-Weierstrass Theorem
Theorem 5.3 (Casorati-Weierstrass). Suppose f :

D
R
(z
0
) C has an essential singularity at
z
0
. Then r (0, R), f(

D
r
(z
0
)) is dense in C.
Proof. Assume that r > 0 such that f(

D
r
(z
0
)) is not dense, then there exits a C and > 0
such that
|f(z) a| > , for z

D
r
(z
0
).
Consider g(z) =
1
f(z)a
. Since g(z) is holomorphic and bounded on

D
r
(z
0
), by theorem
(5.1), there exits h(z) which is holomorphic on D
r
(z
0
) and h(z) = g(z) on

D
r
(z
0
).
Then f(z) =
1
h(z)
+ a. If h(z
0
) = 0, then f has a pole at z
0
; if h(z
0
) = 0, f has a removable
singularity at z
0
. This contradicts that f has an essential singularity at z
0
.
Remark 5.4. In order to prove that z
0
is the essential singularity of f(z), just need to show
that two distinguishable sequences {z
n
}

n=1
and {w
n
}

n=1
such that
z
n
z
0
, w
n
z
0
, as n ,
but
f(z
n
) z, f(w
n
) w, as n ,
where z = w.
5.5 Meromorphic functions
5.5.1 Polynomials
Lemma 5.2. Suppose f is an entire function that satises an estimate of the form
|f(z)| C|z|
N
, if |z| > R, (5.9)
for some positive integer N and positive real constants C and R. Then f must be a polynomial
with degree N or less.
Proof. Method I: Cauchy inequality + power series
Let r > R, M
r
= sup
|z|=r
|f(z)| Cr
N
. By Cauchy inequality (Thm 2.5),
|f
(n)
(0)|
n!M
r
r
n
Cn!r
Nn
0,
as r if n > N. It implies that f
(n)
(0) = 0 if n > N.
By Theorem 1.3,
f(z) =
N

n=0
f
(n)
(0)
n!
z
n
,
25
Yingwei Wang Complex Analysis
which is a polynomial with degree N or less.
Method II: Principal parts + Liouville theorem
By Theorem 1.3,
f(z) =

n=0
a
n
z
n
,

f(z)
z
N
=
_
a
0
z
N
+
a
1
z
N1
+ +
a
N1
z
_
+

k=0
a
N+k
z
k
. (5.10)
Let R(z) =
a
0
z
N
+
a
1
z
N1
+ +
a
N1
z
which is called the principal part of f(z)/z
N
at point
z = 0, g(z) =

k=0
a
N+k
z
k
=
f(z)
z
N
R(z) which is entire.
Besides, claim that g(z) is bounded since
lim
z
|g(z)| lim
z

f(z)
z
N

+ lim
z
|R(z)| = C.
By Liouville theorem (Thm 2.6), g(z) = constant. Denote g(z) = a
N
, then by Eq.(5.10),
f(z) =
N

n=0
a
n
z
n
.
Lemma 5.3. Suppose f is an entire function that satises an estimate of the form
|f(z)| C|z|
N
, if |z| > R, (5.11)
for some positive integer N and positive real constants C and R. Then f must be a polynomial
with degree N or more.
Proof. Method I: Cauchy inequality + power series
Consider the function f(1/z). By assumption,
f(1/z)
C
z
N
, if |z| < R,
f(1/z) =
g(z)
z
m
, m N, g(z) entire,
f(z) = z
m
g(1/z).
Since lim
z
g(1/z) = g(0), we can choose r > R such that
sup
|z|=r
|g(z)| |g(0) + 1|.
26
Yingwei Wang Complex Analysis
By Cauchy formula,
|f
n
(0)| =

n!
2i
_
Cr
w
m
g(1/w)
w
n+1
dw

n!
2
2r r
m(n+1)
(|g(0)| + 1)
Cr
mn
0, as r , if n > m N.
By Theorem 1.3,
f(z) =
m

n=0
f
(n)
(0)
n!
z
n
,
which is a polynomial with degree N or more.
Method II: Principal parts + Liouville theorem
Eq.(5.11) implies that f has nitely many zeros.
Consider
g(z) =
1
f(z)

principal parts at nitely zeros of f(z),

then g(z) is bounded entire. By Liouville theorem (Thm 2.6), g(z) is constant. Thus f(z) is a
rational function. Furthermore, since f(z) is entire, f should be a polynomial.
Denote N
f
= degreef(z), by Lemma 2.5, positive constants c
1
, c
2
, R
0
such that
c
1
|z|
N
f
|f(z)| c
2
|z|
N
f
, |z| > R
0
.
Compared with Eq.(5.11), we know that N
f
N.
Theorem 5.4. Suppose f is entire and with a pole of order m at . Then f is a polynomial
of degree m.
Proof. f is entire means
f(z) =

n=0
a
n
z
n
, z C. (5.12)
f(1/z) =

n=0
a
n
z
n
, z = 0. (5.13)
That f(z) has a pole of order m at at means f(1/z) has a pole of order m at 0. Hence,
Eq.(5.13) only has nitely many terms; more precisely, n > m, a
n
= 0. It indicates that f(z)
is a polynomial of order m.
Corollary 5.1. Suppose f is entire and one-to-one, then f must be a linear function
f(z) = az + b, a = 0.
27
Yingwei Wang Complex Analysis
Proof. Step I: Claim that f has a pole at z = .
If removable, then f is bounded entire and must be a constant; if essential, then f can not
be one-to-one since f(z) is dense in any neighborhood of .
Step II: Claim that f must be a polynomial.
See the proof of Theorem 5.4.
Step III: Claim that the degree of f must be one.
If the degree of f is bigger than one, then f

at least has one zero in C, which contradicts

with Theorem 7.3.
5.5.2 Rational functions
Theorem 5.5. Suppose P(z) and Q(z) are functions with no common factors and N
Q
=
deg (Q) > N
p
= deg P. Let {a
k
}
M
k=1
denote the zeroes of Q(z) with order m
k
, then
P(z)
Q(z)
=
M

k=1
R
k
(z),
where
R
k
(z) =
A
0
(z a
k
)
m
k
+ +
A
m
k
1
z a
k
,
is the principal part of
P(z)
Q(z)
at a
k
.
Proof. Step I: By Thm 3.1, near a
k
,
Q(z) = (z a
k
)
m
k
q
k
(z),
where q
k
(z) is a polynomial and q
k
(a
k
) = 0.
Step II:
P(z)
Q(z)
=
1
(z a
k
)
m
k
_
P(z)
q
k
(z)
_
=
1
(z a
k
)
m
k
[A
0
+ A
1
(z a
k
) + ]
=
A
0
(z a
k
)
m
k
+ +
A
m
k
1
z a
k
+ holomorphic function
= R
k
(z) + holomorphic function.
Step III: Consider f(z) =
P(z)
Q(z)

M
k=1
R
k
(z). Claim that f(z) is bounded entire.
First, need to show that a

k
s are removable singularity. Near a
j
,
P(z)
Q(z)

M

k=1
R
k
(z) =
_
P(z)
Q(z)
R
j
(z)
_

k=j
R
k
(z),
28
Yingwei Wang Complex Analysis
which is holomorphic and hence bounded.
Second,
lim
|z|
|R
k
(z)| = 0, lim
|z|

P(z)
Q(z)

= 0,
lim
|z|
|f(z)| = 0. (5.14)
Step IV: By Thm 2.6, f(z) is a constant. Besides, by Eq.(5.14), we know that the constant
should be 0. Then we can get the conclusion.
Theorem 5.6. f is meromorphic f is a rational function.
Proof. is obviously.
: Step I: Claim that f(z) has only nitely many singularities in C.
f(1/z) has either removable singularity or a pole at 0,
r > 0 such that f(1/z) is holomorphic on D
r
(0) \ {0}
f(z) is holomorphic on C \ D
1/r
(0).
Since D
1/r
(0) is compact and each singularity is isolated, so f(z) has only nitely many
singularities in C, say z
1
, z
2
, , z
n
, with order m
1
, m
2
, , m
n
.
Step II: There are two methods.
Method 1: Let Q(z) =

n
k=1
(z z
k
)
m
k
, then near z
j
, a holomorphic function h
j
(z) such
that
f(z) =
h
j
(z)
(z z
j
)
m
j
, h
j
(z
j
) = 0,
f(z)Q(z) = h
j
(z)

k=j
(z z
k
)
m
k
,
which is bounded near z
j
and holomorphic everywhere else, hence extent to be entire.
Let P(z) = f(z)Q(z) which is entire. Besides, P(z) has a pole or removable singularity at
. By Thm 5.4, P(z) is a polynomial and then f(z) =
P(z)
Q(z)
is a rational function.
Method 2: Let R
k
(z) be the principal part of f(z) at z
k
and R

(z) be the principal part

of f(1/z) at 0.
Let H(z) = f(z) R

(z)

n
k=1
R
k
(z). Then H(z) is holomorphic on

C\ {z
1
, , z
m
, }
and has removable singularity at these points. Hence H(z) can be extent to be holomorphic
on

C, which is bounded entire. By Thm 2.6, H(z) is a constant. So f(z) = H(z) + R

(z) +

n
k=1
R
k
(z) is a rational function.
6 Uniform convergence
6.1 Hurwitz Theorem
Theorem 6.1. Suppose f
n
: C is holomorphic with f
n
= 0 on , and f
n
f uniformly
on compact subsets of . Then either f 0 or f(z) = 0 or all z .
29
Yingwei Wang Complex Analysis
Proof. By Thm 1.4, f is holomorphic.
Suppose f not identically 0 but f(z
0
) = 0.
On one hand, by Identity theorem 3.2, r > 0 so that f(z) = 0 for all z

D
r
(z
0
) and
|f(z)| > > 0, z C
r
(z
0
). (6.1)
On the other hand, since f
n
f uniformly on C
r
(z
0
), n
0
such that n > n
0
,
|f
n
(z) f(z)| < , z C
r
(z
0
). (6.2)
By (6.1), (6.2) and Rouches theorem (Thm 3.5), we know that f and f + (f
n
f) = f
n
has
the same number of zeros. But this contradicts with f(z
0
) = 0 and f
n
= 0 on D
r
(z
0
).
6.2 Montel Theorem
6.2.1 Normal Family
Let F be a set of holomorphic functions on . F is a normal family means any sequence
{f
n
} F, a subsequence {f
n
k
} such that compact set K , f
n
k
converges uniformly on
K.
6.2.2 Uniform boundedness
The family F is said to be uniformly bounded on compact subsets of if for each compact set
K , M = M(K) > 0 such that
|f(z)| < M, for z K, f F.
6.2.3 Equicontinuity
The family F is said to be equicontinuous on a compact set K if for > 0, > 0 such that
z, w K with |z w| < , then
|f(z) f(w)| < , f F.
6.2.4 Montel Theorem
Theorem 6.2. Let F be a set of holomorphic functions on . If F is uniformly bounded on
compact subsets of , then
(i) F is equicontinuous on any compact subset of .
(ii) F is a normal family.
Proof:
(i) Use the Cauchy estimates on small circles.
(ii) Use pointwise convergence on a dense set plus equicontinuity and diagonalization.
30
Yingwei Wang Complex Analysis
6.3 Open mapping theorem
Theorem 6.3. Suppose f : C is holomorphic, then f maps open sets to open sets.
Proof. Let w
0
= f(z
0
) for some z
0
. For w near w
0
, let
g(z) = f(z) w,
F(z) = f(z) w
0
,
G(z) = w
0
w,
then g(z) = F(z) + G(z).
Since z
0
is an isolated zero for F and for w near w
0
, |G| is small, we can nd r > 0 such
that D
r
(z
0
) and > 0,
|F(z)| > > |G(z)|, z C
r
(z
0
).
By Rouche Theorem, g = F +G has a zero in D
r
(z
0
), say z
1
D
r
(z
0
) such that g(z
1
) = 0,
i.e. f(z
1
) = w. Hence w f(D
r
(z
0
)) and D

(w
0
) f(D
r
(z
0
)).
7 Univalence
7.1 Local univalence
Theorem 7.1. Suppose f : C is holomorphic and z
0
with f

(z
0
) = 0. Then r > 0
such that f is univalent in D
r
(z
0
).
Proof. Since f

(z
0
) = 0, then f(z) f(z
0
) has a zero of order 1 at z
0
. So
f(z) f(z
0
) = h(z)(z z
0
),
where h(z) is holomorphic in and h(z
0
) = 0.
Then
R > 0, s.t. |h(z)| >
1
2
|h(z
0
)|, for z D
R
(z
0
).
|f(z) f(z
0
)| >
1
2
|h(z
0
)|R, for z C
R
(z
0
).
r > 0, s.t. |f(z) f(a)| >
1
4
|h(z
0
)|R, for z C
R
(z
0
), a D
r
(z
0
).
For xed a D
r
(z
0
), let g
a
(z) = f(z) f(a) and dene
F(a) =
1
2i
_
C
R
(z
0
)
g

a
(z)
g
a
(z)
dz.
On one hand, since g
a
(z) is continuous in (z, a) and uniformly bounded, F should be con-
tinuous. On the other hand, by Rouche Theorem, F(a) Z and F(z
0
) = 1. So F(a) = 1 for
all a D
r
(z
0
). It implies that for a D
r
(z
0
), the equation f(z) = f(a) has unique solution,
which means f is univalent in D
r
(z
0
).
31
Yingwei Wang Complex Analysis
Corollary 7.1. Denition: A holomorphic mapping f : U V is a local bijection on U if for
z U, there exists an open disc D U centered at z such that f : D f(D) is a bijection.
Question: A holomorphic mapping f : U V is a local bijection on U f

(z) = 0 for
z U.
Proof. is obvious by the Theorem 7.1.
: Suppose f

(z
0
) = 0 for some z
0
U. Then r > 0, such that
f(z) f(z
0
) = a(z z
0
)
k
+ G(z), for z D
r
(z
0
), (7.1)
with a = 0, k 2 and G(z) vanishing to order k + 1 at z
0
.
Besides, since f is bijective in D
r
(z
0
), z = z
0
should be an isolated zero of f

(z). Thus, for

suciently small r, we have
f

(z) = 0, for z D
r
(z
0
)\{z
0
}. (7.2)
Lets choose w C suciently small such that for z C
r
(z
0
), we have
|w| < |a(z z
0
)
k
|, (7.3)
|G(z)| < |a(z z
0
)
k
w|. (7.4)
Let F(z) = a(z z
0
)
k
w. By (7.3), (7.4) and Rouches Theorem, in the disk D
r
(z
0
), the
function a(z z
0
)
k
has at least two zeros, then so does F(z) and further does f(z) f(z
0
) =
F(z) + G(z). It contradicts with the fact that f(z) is bijective in D
r
(z
0
).
Theorem 7.2. Suppose f : C is holomorphic and z
0
with f

(z
0
) = f

(z
0
) = =
f
(n1)
(z
0
) = 0, f
(n)
(z
0
) = 0. Then V z
0
with V and holomorphic on V such that
(i) (z
0
) = 0,

(z) = 0, z V .
(ii) f(z) = f(z
0
) + [(z)]
n
, z V .
(iii) is a univalent map from V onto D
r
(0) for some r > 0.
7.2 Global univalence
Theorem 7.3. Suppose f : C is holomorphic and univalent. Then f

(z) = 0, z .
Proof. z
0
, let f(z
0
) = w
0
and
g(z) = f(z) w
0
, z .
The fact that f is univalent implies that z = z
0
is a simple root of g(z), by Theorem 3.1,
h(z) holomorphic on and h(z
0
) = 0, such that
g(z) = (z z
0
)h(z) (7.5)
g

(z) = h(z) + (z z
0
)h

(z) (7.6)
f

(z
0
) = g

(z
0
) = h(z
0
) = 0. (7.7)
32
Yingwei Wang Complex Analysis
7.3 Limits univalence
Theorem 7.4. Suppose f
n
: C is holomorphic and univalent on , and f
n
f uniformly
on compact subsets of . Then f is either constant or univalent in .
Proof. Suppose f is not constant in . Suppose z
1
= z
2
but f(z
1
) = f(z
2
) = w. Let F
n
= f
n
w,
F = f w.
r
1
, r
2
such that D
r
1
(z
1
) D
r
2
(z
2
) = .
N such that F
N
and F have the same number of zeros in D
r
1
(z
1
) and D
r
2
(z
2
).
Then we get two points
N
1
D
r
1
(z
1
) and
N
2
D
r
2
(z
2
) in two discs satisfying f
N
(
N
1
) =
f
N
(
N
2
) = w. It is a contradiction.
8 Conformal mappings
Let D = D
1
(0), Aut(D) denote the set of all automorphism of D; UHP denote the upper half
plane.
8.1 Schwartz Lemma
Theorem 8.1 (Schwartz Lemma). Let f Aut(D) with f(0) = 0. Then
(i) |f(z)| |z| for z D;
(ii) |f

(0)| 1;
(iii) If either the equality in (i) holds for some z = 0 or the equality in (ii) holds, then f(z)
is a rotation.
Proof. Dene
F(z) =
_
f(z)
z
, if z = 0,
f

(0), if z = 0.
(8.1)
Then F(z) is holomorphic on D.
By MMP (Thm 3.3),
max
|z|r
|F(z)| = max
|z|=r
|F(z)| = max
|z|=r

|f(z)|
|z|

<
1
r
.
Let r 1, we can get (i) and (ii).
For (iii), if either the equality in (i) holds for some z = 0 or the equality in (ii) holds, then
F(z) assumes maximum modulus at point inside the disc D, so F(z) constant.
Corollary 8.1. Let f Aut(D). If f(z) has zeros of order N, then
|f(z)| |z|
N
.
33
Yingwei Wang Complex Analysis
Proof. Let f(z) = z
N
g(z), g(0) = 0. Dene
F(z) =
_
f(z)
z
N
, if z = 0,
g(0), if z = 0.
(8.2)
Then do the similar things as we just did.
8.2 Automorphism of the disc
Lemma 8.1 (Blashke factor). Let a D, then

a
=
a z
1 az
Aut (D).
Furthermore,
(i)
1
a
=
a
;
(ii)
a
(0) = a,
a
(a) = 0;
(iii)

a
(z) =
|a|
2
1
(1 az)
2
,
a
(0) = |a|
2
1,
a
(a) =
1
|a|
2
1
.
Theorem 8.2. If f Aut(D), then R and a D such that
f(z) = e
i
a z
1 az
.
Corollary 8.2. If f Aut(D), then |f

(0)| 1 |f(0)|
2
.
Proof. Suppose f(0) = a, then construct
a
=
az
1 az
. Let F(z) = (
a
f)(z), then F(0) = 0.
By Schwartz Lemma (Thm 8.1) and Lemma 8.1,
|F

(0)| 1
|
a
(a)||f

(0)| 1,
|f

(0)| 1 |a|
2
.
8.3 From upper half plan to the unit disc
Theorem 8.3. The conformal mapping from UPH to D has the form
f(z) = e
i
z a
z a
, Im (a) > 0, R.
34
Yingwei Wang Complex Analysis
9 Roots of functions
Theorem 9.1 (Log roots on a convex open set). Given analytic and non-vanishing function
F(z) on a convex open set , there is an analytic function G(z) on such that F(z) = e
G(z)
.
Given a positive integer N, then H(z) = e
G(z)/N
is an N-th root of F(z), i.e. F(z) = H(z)
N
.
Proof. Fix a point a , and dene the function
G(z) =
_
L
z
a
F

(w)
F(w)
dw,
where L
z
a
is the line from a to z. Then we have
G

=
F

F
(9.1)

d
dz
(Fe
G
) = F

e
G
FG

e
G
= 0, (9.2)
Fe
G
c, (9.3)
where c is a constant.
Pick up log(c), then
F(z) = ce
G
= e

e
G
= e
G+
.
Dene

G = G + , and H(z) = e

G/N
, then we have F = H
N
.
Corollary 9.1. Suppose that f is non-vanishing analytic function on the complex plane minus
the origin. Let denote the curve given by z(t) = e
it
where 0 t 2. Suppose that
1
2i
_

(z)
f(z)
dz
is divisible by 3. Prove that f has an analytic cube root on C \ {0}.
Proof. Step I: Fix a and dene the function
F(z) = exp
_
1
3
_

z
a
f

(w)
f(w)
dw
_
,
where
z
a
is a curve from a to z.
Need to show that F(z) is well dened. Choose two curves
z
a
and
z
a
, then
_

z
a
f

(w)
f(w)
dw
_

z
a
f

(w)
f(w)
dw =
_

z
a
(
z
a
)
f

(w)
f(w)
dw = N 3m 2i,
exp
_
1
3
_

z
a
f

(w)
f(w)
dw
_

z
a
f

(w)
f(w)
dw
_
= exp (2Ni) = 1,

(z)
F

(z)
= 1,
35
Yingwei Wang Complex Analysis
which implies that F(z) is independent of the curve .
Step II: Claim that F

=
1
3
f

f
F.
Restrict attention to a disk.
F(z) = exp
_
1
3
_
_

z
0
a
f

(w)
f(w)
dw +
_
L
z
z
0
f

(w)
f(w)
dw
__
.
Use chain rule and the fact that on convex disc that
d
dz
_
L
z
z
0
f

(w)
f(w)
dw =
f

(z)
f(z)
.
Step III: Claim that (f/F
3
)

= 0 f/F
3
C.
_
f
F
3
_

=
f

F
3
f3F
2 1
3
f

f
F
F
6
= 0.

f
F
3
= C.
Choose = log(C), and dene

F(z) = exp
_
1
3
_

z
a
f

(w)
f(w)
dw +
_
.
It is easy to know that f =

F
3
.
Remark 9.1. Let f(z) be holomorphic on \{p
0
}, which has a zero at z = z
0
with multiplicity
n and has a pole at z = p
0
with order m. Choose a \ {p
0
}, dene a new function as
g(z) = exp
__

z
a
f

(w)
f(w)
dw
_
.
Then we know that:
h(z) =
f

(z)
f(z)
has a simple pole at z = z
0
with residue equal to a positive integer.
g(z) is a well dened analytic function on \ {p
0
, z
0
}.
g(z) has a removable singularity at z = z
0
. More precisely, if g(z) is redened at z = z
0
as g(z
0
) = 0, then z = z
0
is also a zero of g(z) with multiplicity n.
g(z) also has a pole at z = p
0
with order m.
36