MTH6111 Complex Analysis 2010-2011: Summary of Main Results

WARNING: Just because a result is not in this list does not mean it will not be in the exam. The results
listed below are the key ones to understanding the whole course. Results which follow fairly directly from
definitions have been omitted. When you revise for the examination you should look through ALL results
and ALL proofs, except for those which are labelled ”not examinable”.

SECTION I: Holomorphic Functions

3.2 Theorem Every bounded sequence in C has a convergent subsequence.

Big idea in proof of 3.2. Repeatedly divide a square into four squares of half the side length. At each
step one of the four must contain infinitely many points of the sequence.

3.3 Corollary S ⊆ C is compact (i.e. closed and bounded) if and only if every sequence in S has a
subsequence which converges to a limit in S. (“S is compact if and only if S is sequentially compact.”)

3.4 Theorem let S ⊆ C be compact and f : S → C be continuous. Then f (S) is compact and |f | attains
its bound on S (i.e. if M = supz∈S |f (z)| then ∃a ∈ S such that |f (a)| = M ).

Big ideas in proof of 3.4. Use 3.3 and the fact that since f is continuous f (lim zn ) = lim f (zn ) for
any convergent sequence (zn ).

4.1 Theorem (Cauchy-Riemann equations). Let f : U → C be differentiable at a point z = x + iy ∈

U , and let u(x, y) = Re(f (z)) and v(x, y) = Im(f (z)).
Then the partial derivatives ∂u/∂x, ∂u/∂y, ∂v/∂x, ∂v/∂y exist at (x, y) and satisfy
∂u ∂v ∂u ∂v
= and =−
∂x ∂y ∂y ∂x
there.

Big idea in proof of 4.1. limh→0 (f (z + h) − f (z))/h has to be the same whatever the direction in
which we let h go to zero. Comparing the limits with h real and h imaginary gives the Cauchy-Riemann
equations.

4.2 Proposition Let f ∈ H(D(0, R)) (i.e. f is holomorphic on D(0, R)). Then
(i) if f 0 (z) = 0 for all z ∈ D(0, R) then f is constant on D(0, R);
(ii) if |f (z)| is constant on D(0, R) then f is constant on D(0, R).

Big idea in proof of 4.2. Let f = u + iv.

For (i), between any two points in D(0, R) there is a path made up of a horizontal segment and a vertical
segment. Apply the real mean value theorem to prove that u and v are constant.

1
For (ii), partially differentiate u2 + v 2 = c2 with respect to x and y and apply the Cauchy-Riemann
equations. Deduce that the partial derivatives of u and v with respect to x and y are all zero. Now apply
(i).

P∞
5.4 Proposition Every power series 0 an z n has a radius of convergence, i.e. ∃R, possibly 0 or ∞,
such that the series converges absolutely for all z with |z| < R and diverges for all z with |z| > R.
P∞ P∞
Big idea in proof of 5.4. It suffices to show that if an wn converges, and |z| < |w|, then 0 an z n
0
|z| P n
|an z n | converges.
P
converges absolutely. Let b = |w| . Using the fact that b converges show that

P∞ P∞
5.5 Lemma n=0 an z n and 1 nan z n−1 have the same radius of convergence.
P∞
Big idea in proof of 5.5. Show that if 0 an z n converges absolutely for |z| < R 6= 0, then so does
P∞ n−1
1 nan z . To do this, choose r with |z| < r < R. Now
 n−1
n |z|
|nan z n−1 | = |an rn |.
r r
 n−1  n−1
But there exists M with nr |z|r < M for all n since lim n∞
n |z|
r r = 0 (by the ratio test). Finally
P∞ n−1
P∞ n
show 1 |nan z | converges, by the comparison test, since 0 |an r | converges.

P∞
5.6 Theorem If an z n has radius of convergence R 6= 0, and f (z) is the sum of this series, then f
0
P∞
is a holomorphic function on D(0, R), and its derivative is f 0 (z) = 1 nan z n−1 for all z ∈ D(0, R).
Proof of 5.6 not examinable.

P∞
5.7 Corollary If f (z) = 0 an z n has radius of convergence R 6= 0 then f has derivatives of all orders
on D(0, R) and f (n) (0) = n!an for all n ≥ 0 (where 0! is defined to be 1).

5.8 Proposition
(i) ez is entire (that is, differentiable for all z ∈ C) and the derivative of ez is ez .
(ii) ∀z, w ∈ C, ez+w = ez ew .

Big idea in proof of 5.8. (i) is obvious using 5.6. For (ii), for any ζ ∈ C, differentiate the product
f (z) = ez eζ−z and deduce that f is constant using 4.2(i).

2
SECTION II: Integration and Cauchy’s Theorem

7.1 Fundamental theorem of calculus Let γ be a path [α, β] → C, and let F : U → C be holomorphic,
where U is an open set containing γ ∗ . Then
Z
F 0 (z)dz = F (γ(b)) − F (γ(a))
γ

F 0 (z)dz = 0.
R
In particular, if γ is a closed path (this means γ(α) = γ(β)) then γ

Big idea in proof of 7.1. Prove it for a smooth path γ by applying the fundamental theorem of calculus
for real functions to the real and imaginary parts.

8.1 Estimation Lemma Let f : U → C be continuous (where U is some subset of C), let γ be a path in
U , and suppose |f (z)| < M for all z ∈ γ ∗ . Let length(γ)= L. Then
Z
| f (z)dz| ≤ M L.
γ

Big ideas in proof of 8.1 First prove that for real-valued functions u and v of t ∈ R,
Z b Z b
(∗) | (u(t) + iv(t))dt| ≤ |u(t) + iv(t)|dt
a a
Rb Rb Rb
To do this this, write (u(t) + iv(t))dt = Reiθ and show that |
a a
(u(t) + iv(t))dt| = a
Re(e−iθ (u(t) +
Rβ
iv(t))dt. Now apply (∗) to | γ f | = | α f (γ(t))γ 0 (t)dt|.
R

8.2 Proposition Let γ be a path in U ⊂ C and (Fn )n≥0 be a sequence of continuous functions U → C
converging uniformly on γ ∗ to a continuous function F . Then
Z Z
F = lim Fn
γ n→∞ γ

Big idea in proof of 8.2. Apply the Estimation Lemma toshow that
Z Z Z
| F − Fn | = | (F − Fn )| ≤ L(γ).supz∈γ ∗ |F (z) − Fn (z)|.
γ γ γ

Now take the limit as n tends to infinity.

P∞
8.3 Corollary If a series m=0 fm (z) of continuous functions converges uniformly on γ∗ to a continuous
function, then
∞ ∞ Z
Z ! 
X X
fm (z) dz = fm (z)dz
γ m=0 m=0 γ

3
9.1 Cauchy’s Theorem for a triangle If f is holomorphic on an open set U containing the triangle
R
γ∗, and its interior, then γ f (z)dz = 0.

Big ideas in proof of 9.1. Suppose the triangle has perimeter of length L, and that the integral is
I 6= 0. Subdivide the triangle into four, each of perimeter L/2. One of these has integral of modulus at
least |I|/4. Choose it and repeat. Inductively get a triangle of perimeter L/2n and on which the integral
has modulus at least |I|/4n . These triangles nest down to a point. Using differentiablity of f at this
point, and the Estimation Lemma, we can get a contradiction and hence deduce that I = 0.

9.2 Theorem (Existence of Antiderivatives) Let U be a convex open subset of C and let f be
holomorphic on U . Then there exists a holomorphic function F on U such that F 0 = f .

Big ideas in proof of 9.2. Choose a point a ∈ U , and for each z ∈ U define
Z
F (z) = f (w)dw
[a,z]

Now prove F is differentiable, with derivative f by observing that, by 9.1,

Z
F (z + h) − F (z) = f (w)dw,
[z,z+h]
R
and applying the Estimation Lemma to show that | [z,z+h]
(f (w) − f (z))dw| is small and hence that
R R
[z,z+h]
f (w)dw is close to [z,z+h] f (z)dw = hf (z).

9.3 Corollary (Cauchy’s Theorem for a convex region) Let U be an open convex subset of C and
R
f be holomorphic on U . Then γ f (z)dz = 0 for every closed path γ in U (where γ : [α, β] → U is said
to be a closed path if γ(α) = γ(β)).

Big idea in proof of 9.3. Theorem 9.2 and the Fundamental Theorem of Calculus (7.1).

9.4 Cauchy’s Theorem for a simple closed path If f is holomorphic everywhere on and inside a
R
simple closed path γ, then γ f (z)dz = 0.

Not proved in course.

9.5 Deformation Principle

R R
(i) If γ0 and γ1 are homotopic closed paths in U and f is holomorphic on U then γ1
f (z)dz = γ2
f (z)dz.
(ii) If γ1 and γ2 are paths in U between the same points a and b which are homotopic in U relative to
R R
their end points, and f is holomorphic on U , then γ1 f (z)dz = γ2 f (z)dz.

Not proved in course.

4
9.6 Cauchy’s Theorem for a simply-connected region If f is holomorphic on a simply-connected
R
region U then γ f (z)dz = 0 for every closed path γ in U .

Not proved in course.

9.7 Cauchy’s Theorem for a multiply-connected region

If γ is a simple closed path, and γ1 , . . . , γn are disjoint simple closed paths in the interior of γ, such that
the interiors of γ1 , . . . , γn are disjoint, then if f is holomorphic on γ ∗ , and on γ1∗ , . . . , γn∗ , and on the
region between γ ∗ and γ1∗ ∪ . . . ∪ γn∗ , we have
Z Z Z
f (z)dz = f (z)dz + . . . + f (z)dz
γ γ1 γn

(provided γ and γ1 , . . . , γn are all parametrized in the the same direction e.g. positively, that is, anti-
clockwise.)

Not proved in course.

SECTION III

10.1 Cauchy’s Integral Formula

Let f be holomorphic on and everywhere inside a simple closed path γ. Then for every point a inside γ,
Z
1 f (z)
f (a) = dz
2πi γ z − a

Big idea in proof of 10.1 The deformation principle tells us that

Z Z
f (z) f (z)
dz = dz
γ z − a C(a,r) z −a

But we know that

Z Z Z 2π
f (a) 1 1
dz = f (a) dz = f (a) ireit dt = 2πif (a)
C(a,r) z−a C(a,r) z−a 0 reit

so it suffices to prove that we can make

f (z) − f (a)
Z
dz
C(a,r) z−a

arbitrarily small by taking r sufficiently small. We use the Estimation Lemma to prove this.

10.2 Liouville’s Theorem

If f is entire (that is, holomorphic everywhere on C) and f is bounded on C, then f is constant.

5
Big idea in proof of 10.2 Choose any two points a, b ∈ C, and take R ≥ 2max{|a|, |b|}. Using Cauchy’s
Integral Formula, we have
 
a−b
Z Z
1 1 1 1
f (a) − f (b) = f (z) − dz = f (z) dz
2πi C(0,R) z−a z−b 2πi C(0,R) (z − a)(z − b)

Now use the Estimation Lemma.

10.3 The Fundamental Theorem of Algebra

Let p(z) be a non-constant polynomial (that is, a polynomial of degree ≥ 1) with coefficients in C. Then
there exists w ∈ C such that p(w) = 0.

Big idea in proof of 10.3 Suppose there is no solution in C to the equation p(z) = 0. Then the
function f (z) defined by f (z) = 1/p(z) is entire. Choose R such that for all z with |z| > R the terms
|a0 /z n |, . . . , |an−1 /z| are all < |an |/2n. Now for all z with |z| > r
|an |Rn 2
|p(z)| > and so |f (z)| < .
2 |an |Rn
Hence f (z) is bounded on C, and therfore it is constant by Liouville’s Theorem.

10.4 Cauchy’s formula for derivatives (extended form of Cauchy’s Integral Formula)
Let f be holomorphic on and everywhere inside a simple closed path γ. Then at every point a inside γ,
the nth derivative f (n) (a) exists for n = 0, 1, 2, 3 . . ., and
Z
n! f (z)
f (n) (a) = dz
2πi γ (z − a)n+1

Proof of 10.4 is an immediate consequence of 11.2 (Taylor’s Theorem) below.

11.1 Lemma Let f be holomorphic on D(a, R). Then there exist constants cn ∈ C such that
∞
X
f (a + h) = cn hn
n=0

and this series converges absolutely for all h with |h| < R. Moreover
Z
1 f (z)
cn = dz
2πi C(a,r) (z − a)n+1

for any r with 0 < r < R.

Big idea in proof of 11.1 Cauchy’s Integral Formula gives

Z ∞
1 X f (z)
f (a + h) = n+1
.hn dz.
2πi C(a,r) n=0 (z − a)

6
 RP PR
Expand 1/(z − (a + h)) as a power series in h and apply the fact that = (which follows from
the uniform convergence of the terms in the sum of functions being integrated).

11.2 Taylor’s Theorem

If f is holomorphic on the open set U ⊆ C then all the higher derivatives of f exist everywhere in U , and
for all z in any disc D(a, R) ⊆ U , f (z) is equal to the sum of the Taylor series
∞
X f (n) (a)
(z − a)n
n=0
n!

Moreover for any r with 0 < r < R

Z
(n) n! f (z)
f (a) = dz
2πi C(a,r) (z − a)n+1

Big idea in proof of 11.2 By Lemma 11.1,

∞
X
f (a + h) = cn hn ∀h with |h| < R
n=0

so, writing z = a + h we have

∞
X
f (z) = cn (z − a)n for |z − a| < R
n=0

By 5.6 and 5.7 any power series can be differentiated arbitrarily many times and for the power series
above,
f (n) (a) = n!cn .

Finally by Lemma 11.1 Z

1 f (z)
cn = dz
2πi C(a,r) (z − a)n+1

11.3 Corollary (Cauchy’s Estimate)

If f is differentiable for |z − a| < R, if 0 < r < R and if |f (z)| ≤ M everywhere on the circle |z − a| = r,
then
M.n!
|f (n) (a)| ≤ ∀n ≥ 0
rn

Proof of 11.3 Z
n! f (z) n! M M.n!
|f (n) (a)| = | n+1
dz| ≤ . n+1 .2πr = n
2πi C(a,r) (z − a) 2π r r

7
12.1 Laurent’s Theorem
Let A = {z ∈ C : R1 < |z − a| < R2 }, where R1 and R2 are constants such that 0 ≤ R1 < R2 ≤ ∞, and
let f ∈ H(A) (that is, f is holomorphic on A). Then
∞
X
f (z) = cn (z − a)n ∀z ∈ A
−∞

where Z
1 f (w)
cn = dw ∀n ∈ Z
2πi C(a,r) (w − a)n+1
for any r such that R1 < r < R2 .

Big idea in proof of 12.1 By a change of coordinates assume that a = 0. Choose any z ∈ A and let
r1 and r2 be such that R1 < r1 < |z| < r2 < R2 . By Cauchy’s Integral Formula and the deformation
principle Z Z
1 f (w) 1 f (w)
f (z) = dw − dz.
2πi C(0,r2 ) w−z 2πi C(0,r1 ) w−z
Since |z| < |w| for w on C(0, r2 ) and |z| > |w| for w on C(0, r1 ), expanding 1/(w − z) in the two integrals
gives
∞ ∞
zn −wm
Z Z
1 X 1 X
f (z) = n+1
f (w)dw − f (w)dw.
2πi
C(0,r2 ) n=0 w 2πi C(0,r1 ) m=0 z m+1
R P
Now invoke uniform convergence to interchange with and complete the proof.

12.2 Lemma If f is holomorphic on D0 (a, R) then for any positively oriented simple closed path γ around
a in D0 (a, R), Z
f (z)dz = 2πi.res(f, a)
γ

Proof of 12.2 By Laurent’s Theorem, for any r < R

Z
1
c−1 = f (z)dz
2πi C(a,r)

The result follows by the deformation principle (just deform C(a, r) to γ).

12.3 The Residue Theorem Let γ be a positively oriented simple closed path in C. If f is holomorphic
on γ ∗ and everywhere inside γ ∗ except at a finite number of isolated singularities z1 , . . . , zn inside γ ∗ ,
then Z n
X
f (z)dz = 2πi res(f, zj )
γ j=1

Proof of 12.3 Immediate from 9.7 (Cauchy’s Theorem for a multiply-connected region) and 12.2.

8
13.1 Lemma (i) If f (z) = g(z)/(z − a) where g is holomorphic at a with g(a) 6= 0 then a is a simple
pole of f and Res(f, a) = g(a).
(ii) If f (z) = h(z)/k(z) where h and k are holomorphic at a with h(a) 6= 0, k(a) = 0 and k 0 (a) 6= 0, then
a is a simple pole of f and Res(f, a) = h(a)/k 0 (a).

Big idea in proof of 13.1 Consider the Taylor series for g,h and k and hence compute the first few
terms of the Laurent series for f in each case.

13.2 Lemma if f (z) = g(z)/(z − a)m , where g is holomorphic at a with g(a) 6= 0, then a is a pole of
order m of f and
g (m−1) (a)
Res(f, a) =
(m − 1)!

Proof of 13.2 Immediate from dividing the Taylor series for g at z = a by (z − a)m .

14.1 Proposition If a holomorphic function f (z) on the punctured disc D0 (a, R) is bounded there, then
a is a removable singularity of f .

Idea in Proof of 14.1 The coefficient bn of (z − a)−n in the Laurent series for f around a is given by:
Z
1
bn = f (z)(z − a)n−1 dz
2πi C(a,r)

By taking r arbitrarily small and applying the Estimation Lemma deduce that bn = 0 (for all n > 0).

14.2 Proposition f (z) has a pole of order m at a if and only if 1/f (z) has a zero of order m at a.
(More precisely, if and only if 1/f (z) has a removable singularity at a, which when removed gives a zero
of order m)

Proof of 14.2 Write g(z) = (z − a)m f (z). Consider the Laurent series for f (z) at z = a and the Taylor
series for g(z) at z = a, and observe that:
f (z) has a pole of order m at z = a
⇔ g(z) is holomorphic on D(a, r) for some r > 0, and g(a) 6= 0
(more precisely g(z) has a removable singularity at z = a which when removed gives g(a) 6= 0)
⇔ 1/g(z) is holomorphic on D(a, r0 ) for some r0 > 0, and 1/g(a) 6= 0
⇔ 1/f (z) (=(z − a)m /g(z)) has a zero of order m at z = a.

14.3 Corollary If f has a pole of order m at a, then limz→a |f (z)| = ∞ (that is, given any M > 0 there
exists δ > 0 such that |f (z)| > M whenever |x − a| < δ).

9
14.4 Theorem (Weierstrass-Casorati) If a is an isolated essential singularity of f , then in every
neighbourhood of a the function f takes values arbitraily close to any assigned complex number w (that
is, given any r > 0,  > 0 and w ∈ C, there exists ζ ∈ D0 (a, r) with |f (ζ) − w| < |).
Proof of 14.4 Given any w ∈ C, let φ(z) = f (z) − w. Since its Laurent series differes from that of f
only in the constant term, φ has an essential singularity at a. Thus it suffices to prove that if φ is any
function with an essential singularity at a then for every r > 0 there exists ζ ∈ D0 (a, r) with |ζ| < .
If a is a limit of zeros of φ this is obvious. If not, there exists r0 such that φ(z) 6= 0 for z ∈ D0 (z, r0 ).
Consider the function 1/φ(z). Either there exists ζ ∈ D0 (a, r0 ) for which 1/|φ(ζ)| > 1/ (in which case
we are finished) or 1/φ(z) is bounded on D0 (a, r0 ), in which case 1/φ(z) has a removable singularity at a
by 14.1. But then, by 14.2, φ(z) has (at worst) a pole at a, contradicting our hypothesis.

15.1 Theorem A function f (z) is meromorphic on the extended complex plane C ∪ {∞} if and only if
f (z) is a rational function (that is f (z) = p(z)/q(z) for some polynomials p(z) and q(z)).

Big ideas in proof of 15.1 A rational function p(z)/q(z) is meromorphic on C ∪ {∞} since:
• the only singularities of p(z)/q(z) in C are the zeros of q(z) and they are all poles;
• z = ∞ is a pole or removable singularity of p(z)/q(z) (as z = 0 is a pole or removable singularity of
p(1/z)/q(1/z)).

For the converse, given a function f which is meromorphic on C ∪ {∞}:

• All the singularities of f in C are in some D(0, R), since the singularity at ∞ is isolated. Suppose these
are z1 , . . . , zk and that they are poles of orders n1 , . . . , nk .
• Let ψ(z) = (z − z1 )n1 . . . (z − zk )nk and g(z) = ψ(z)f (z). Then g(z) is holomorphic on C, and has a
pole at ∞.
• The Taylor series for g at z = 0 converges on the whole of C. But g(1/z) can have only finitely many
terms in 1/z since g has a pole at ∞. Hence g(z) is a polynomial.

SECTION IV
16.1 Lemma If f is holomorphic at a and f 0 (a) 6= 0 then f is conformal at a.

Big idea in proof of 16.1 By the chain rule

(f ◦ γ2 )0 (t2 ) f 0 (γ2 (t2 ))γ20 (t2 ) f 0 (a)γ20 (t2 ) γ 0 (t2 )

0
= 0 0 = 0 0 = 20
(f ◦ γ1 ) (t1 ) f (γ1 (t1 ))γ1 (t1 ) f (a)γ1 (t1 ) γ1 (t1 )

17.1 Lemma The automorphisms of Ĉ are the maps of the form:

az + b
φ(z) =
cz + d

where a, b, c, d ∈ C and ad − bc 6= 0 .

10
Big idea in proof of 17.1 If f is a rational map of degee d then f −1 (w) consists of d distinct points,
for almost every w ∈ Ĉ. So for f to be injective we need d = 1.

17.3 Proposition A Mobius transformation is conformal at every point of Ĉ.

Big idea in proof of 17.3 Differentiate and show the derivative is never zero.

17.4 Proposition Every Möbius transformation φ is a composition of transformations of the three types:
z → az a 6= 0, z → z + b, z → 1/z

Big idea in proof of 17.4 Decompose into

z→Z→W →w

where Z = cz + d, W = 1/Z, and w = αW + β. Compute the values of α and β.

17.5 Corollary Möbius transformations send circles to circles (where a ‘circle through ∞’ is a straight
line in C).

17.6 Proposition Let z1 , z2 , z3 be any triple of distinct points in Ĉ and w1 , w2 , w3 also be a triple of
distinct points in Ĉ. Then there exists a unique Möbius transformation φ such that φ(zj ) = wj for
j = 1, 2, 3.

Big idea in proof of 17.6   

z1 − z3 z2 − z
φ:z→
z2 − z3 z1 − z
maps z1 , z2 , z3 to ∞, 0, 1 respectively. Similarly
  
w1 − w3 w2 − w
ψ:w→
w2 − w3 w1 − w

maps w1 , w2 , w3 to ∞, 0, 1. Thus ψ −1 ◦ φ has the required property.

18.1 Proposition Every automorphism of the plane C (that is to say holomorphic bijection C → C) has
the form
φ(z) = az + b

for some a ∈ C with a 6= 0.

Big idea in proof of 18.1: details not examinable. Every automorphism of the plane extends to
an automorphism of the Riemann sphere.

19.1 Theorem Let f be holomorphic in an open set U , with real and imaginary parts u and v. Then
both u and v are harmonic in U .

11
Big idea in proof of 19.1 Differentiate the Cauchy-Riemann equations with respect to x and y.

19.2 Theorem Let D ⊂ R2 be an open disc and suppose that u : D → R is harmonic. Then there exists
a complex function f , holomorphic in D, such that u = Re(f ).

Proof of 19.2 omitted in lectures because of lack of time, so non-examinable.

20.1 The Maximum Modulus Principle If f is holomorphic on an open disc D = D(a, r) and b ∈ D
is a maximum point for |f |, that is |f (b)| ≥ |f (z)| ∀z ∈ D, then f is constant on D.

Big idea in proof of 20.1 Follows from Gauss’ Mean Value Theorem which states that if f is holomorphic
on D(z0 , R), and r < R, then
Z 2π
1
f (z0 ) = f (z0 + reiθ )dθ
2π 0
(Gauss’ Mean Value Theorem follows directly from Cauchy’s integral formula.)

20.2 Schwarz’s Lemma Let f : D → D be a holomorphic function of the unit disc to itself such that
f (0) = 0. Then
(i) |f (z)| ≤ |z| for all z ∈ D.
(ii) If for some z0 6= 0 we have |f (z0 )| = |z0 | then there is some complex number µ of modulus 1 such
that f (z) = µz.

Proof of 20.2 Since f (0) = 0 the Taylor series for f (z) at z = 0 begins f (z) = a1 z + a2 z 2 + . . .. Hence
f (z)/z = a1 + a2 z + a3 z 2 + . . . is holomorphic on the unit disc D. For any r < 1 choose z0 with |z0 | = r.
Now |f (z0 )/z0 | < 1/r so by the Maximum Modulus Principle |f (z)/z| ≤ 1/r for all z with |z| ≤ |z0 | = r.
Letting r tend to 1 gives (i).
For (ii), if ∃z0 ∈ D with |f (z0 )/z0 | = 1 then by the Maximum Modulus Principle f (z)/z is constant on
D. Hence f (z)/z = µ for some constant µ of modulus 1.

20.3 Corollary Let f : D → D be an automorphism of the unit disc. Then f is a Möbius transformation.
Indeed if a ∈ D is the unique point such that f (a) = 0, then
a−z
f (z) = eiθ
1 − āz
for some real number θ.

Proof of 20.3 The Möbius transformation

a−z
g(z) =
1 − āz
sends D to D and it sends the point a to 0. So f ◦ g −1 is an automorphism h of D which has h(0) = 0.
Apply 20.2(i) to h and to h−1 to deduce |h(z)| = |z| for all z ∈ D. Part (ii) of 20.2 now gives h(z) = eiθ z
for some real θ. So f = h ◦ g has the required form.

12
20.4 Corollary The automorphisms of the complex upper half-plane H are the Möbius transformations
of the form
az + b
z→
cz + d
with a, b, c, d ∈ R and ad − bc > 0

Proof of 20.4 Let φ be a Möbius transformation which is a bijection from H to the unit disc D. If f is an
automorphism of H then φ ◦ f ◦ φ−1 is an automorphism of D, and by 20.3 it is a Möbius transformation
ψ. Hence f (= φ−1 ◦ ψ ◦ φ) is a Möbius transformation. As f sends R̂ to R̂, preserving orientation, it has
the stated form.

SECTION V

23.1 Proposition (Schwarz reflection principle) Let U be a connected open set in C which is sym-
metric about the real axis, that is, z ∈ U ⇔ z̄ ∈ U . Let U + = {z ∈ U : Im(z) > 0}, U − = {z ∈ U :
Im(z) < 0} and U 0 = {z ∈ U : Im(z) = 0}. Suppose f : U + ∪ U 0 → C is continuous, that f (z) is real
for all z ∈ U 0 , and that f is holomorphic on U + . Then f can be extended analytically to U − by setting
f (z) = f (z̄) for each z ∈ U − .

Statement and proof of 23.1 omitted from lectures because of lack of time, so not examinable

24.1 Lemma Every double-periodic holomorphic function is constant.

Proof of 24.1 The fundamental domain P is closed and bounded. So f (P ) is bounded. Hence f is
bounded on C, since for any z ∈ C, f (z) = f (z 0 ) for some z 0 ∈ P . Hence, by Liouville’s Theorem, f is
constant.

24.2 Proposition
(i) Every elliptic function of order 0 is constant.
(ii) There are no elliptic functions of order 1.

Proof of 24.2 Assume the fundamental parallelogram P chosen with no poles on its edges. Part (i) is
now Lemma 24.1 (since a meromorphic f is holomorphic if and only if it has no poles). For part(ii), note
that integrating f once around the boundary of P gives zero, since the integrals along opposite sides
cancel. So by the Residue Theorem the sum of the residues of f inside P is zero. Hence f must have at
least two simple poles, or at least one pole of order ≥ 2.

24.3 Proposition For each N ≥ 3

X 1
FN (z) =
(z − ω)N
ω∈Ω

is elliptic of order N with respect to Ω

13
Idea in proof of 24.3 Once we know the sum is convergent (proof not expected), so FN is a well defined
(meromorphic) function, it follows from the expression for FN that it is doubly periodic and of order N .

25.1 Picard’s Little Theorem If f : C → C is an entire function such that there exist at least two
points in C not in the image of f , then f is constant.

Proof of 25.1 not examinable.

14