COMPLEX ANALYSIS ASSIGNMENT 13

SIDDHARTH KOTHARI

Problem 2
Let Γ be the boundary of the fundamental parallelogram, oriented positively. We may assume that
f is holomorphic on Γ. (We can translate the input if necessary, since that does not change the
nature of the poles or zeros of f .) Let z1 , · · · , zn be the poles of f in the fundamental parallelogram
(Note that f is meromorphic by definition, so no essential singularities). By the residue theorem,
we have that Z n
1 X
f (z) dz = Res(f ; zj ).
2πi Γ
j=1
Note that
Z 1 Z 1 Z 1 Z 1
I = ω1 f (ω1 t) dt + ω2 f (ω1 + ω2 t) − ω1 f (ω1 (1 − t) + ω2 ) dt − ω2 f (ω2 (1 − t)) dt = 0,
0 0 0 0
R1 R1 R1 R1
and since 0 f (ω1 t) dt = − t) + ω2 ) dt and 0 f (ω1 + ω2 t) dt = 0 f (ω2 (1 − t)) dt by the
0 f (ω1 (1
Λ periodicity of f , we have I = 0. Therefore,
Xn
Res(f ; zj ) = 0.
j=1
Problem 4
Proposition 0.1. We have ℘aΛ (z) = a−2 ℘Λ (z/a) for all z ∈ C \ Λ and a ∈ C \ {0}.
Proof. We simply use the definition of the ℘ function to expand the left side:

1 X  1 1

1 X  1 1

℘aΛ (z) = 2 + − = 2+ −
z ∗
(z + ω)2 ω 2 z ∗
(z + aω)2 a2 ω 2
ω∈(aΛ) ω∈Λ

1 X  1 1

= 2+ −
z a2 (z/a + ω)2 a2 ω 2
ω∈Λ∗
 
1 1 X 1 1
= 2+ 2 −
z a (z/a + ω)2 ω 2
ω∈Λ∗
X  !
1 1 1 1
= 2 + −
a (z/a)2 ∗
(z/a + ω)2 ω 2
ω∈Λ
1
= 2 ℘Λ (z/a),
a
as desired. ■
Problem 8
Without loss of generality, we may assume that τ ′ ∈ C is such that R(τ ′ ) ̸= 0 and J(τ ′ ) ̸= 0. To see
this, we apply casework on the signs of the real and imaginary parts of τ = a + bi and τ ′ = a′ + b′ i.
We may assume that τ ′ ̸= ±τ , since if τ ′ = τ then we set a = 1 = d and c = 0 = d, and if τ ′ = −τ ,
then a = −1, d = 1, c = 0 = b.
1
2 SIDDHARTH KOTHARI

• For the sake of contradiction, assume that a = 0 = a′ . Since Λ(1, τ ) = Λ(1, τ ′ ), we must
have that τ ′ ∈ Λ(1, τ ), meaning there exists integers m1 and m2 such that m1 τ + m2 =
τ ′ =⇒ m1 bi + m2 = b′ i. Taking the real part of both sides shows us that m2 = 0, and so
m1 b = b′ . Thus, we have b′ = mb where m = m1 is a non-zero integer such that m ̸= ±1.
On the other hand, τ ∈ Λ(1, τ ′ ) =⇒ bi = n1 b′ i + n2 =⇒ b = nb′ for some integer n.
However, from the previous logic we know that the unique solution to this linear equation
in the complex numbers is n = 1/m, which is not an integer, a contradiction. Thus, at least
one of a, a′ must be non-zero, WLOG assume that a′ ̸= 0.
• Similarly, one can show that at least one of b, b′ must be non-zero. If b′ ̸= 0, then we are
done. Otherwise, assume that b′ = 0 and b ̸= 0. Then Λ(1, τ ′ ) ⊆ R while Λ(1, τ ) contains
imaginary numbers, which is a contradiction as Λ(1, τ ) = Λ(1, τ ′ ).
An important corollary is that if x1 τ ′ + x2 = 0, then x1 = 0 = x2 : x1 (a′ + b′ i) + x2 = (a′ x1 + x2 ) +
x2 b′ i = 0. Equating the real and imaginary parts, we have that x2 b = 0, and since b ̸= 0, we must
have x2 = 0. Next, a′ x1 + x2 = a′ x1 = 0, and since a′ ̸= 0, we must have x1 = 0. Now we begin
with the actual proof.
First, note 1, τ ′ ∈ Λ(1, τ ), and so there exists integers a, b, c, d ∈ Z such that τ ′ = aτ + b and
1 = cτ + d, meaning that
τ′ aτ + b
τ′ = = .
1 cτ + d
Reformulating this as a matrix vector equation yields
     ′
a b τ τ
= ,
c d 1 1
and so
d −b τ ′
    
τ 1
= .
1 ad − bc −c a 1
Extracting the two equations, we get
d b
τ = τ′ − ,
α α
and
c a
τ = − τ′ + ,
α α
where α = ad − bc. If α ̸= ±1, then not all of a/α, b/α, c/α, d/α can be integers.
(Indeed: assume they are all integers, so a = a′ α, b = b′ α, c = c′ α and d = d′ α for some
a′ , b′ , c′ , d′ ∈ Z. In that case, we have ad − bc = α =⇒ α4 (a′ d′ − b′ c′ ) = α, so a′ d′ − b′ c′ = α−3 .
This is a contradiction because the left side is an integer, whereas the right side is in Q \ Z.
Therefore, at least one of a/α, b/α, c/α, d/α must be in Q \ Z; without loss of generality, assume
that d/α ̸∈ Z.)
In that case, since τ ∈ Λ(1, τ ′ ), we must have τ ′ (d/α) − (b/α) = n1 τ ′ + n2 for some n1 , n2 ∈ Z.
Thus, (d/α − n1 )τ ′ − (n2 + b/α) = 0 =⇒ d/α = n1 , meaning d/α ∈ Z, which is a contradiction.
Therefore, our original assumption that α ̸= ±1 must have been false, so α = ad − bc = ±1.