Complex Functions

[Problem Set 5]
October 16, 2025
Dakota Nobuyuki Goto

1. Let M ⊂ C be a region, and let f : M → C be an arbitrary function. Show that any two
antiderivative of f differ by a constant.

Solution:

Proof. Let F and G be two antiderivatives of the same function f on the region M , i.e.

F ′ (z) = f (z) = G′ (z) for all z ∈ M.

Define H := F − G. Then H is holomorphic on M and

H ′ (z) = F ′ (z) − G′ (z) = 0 for all z ∈ M.

Write H = u + iv where u, v : M → R are the real and imaginary parts of H. Since H

is complex-differentiable and H ′ (z) ≡ 0, the Cauchy–Riemann equations imply that all first
partial derivatives of u and v vanish:
∂u ∂u ∂v ∂v
= = = =0 on M.
∂x ∂y ∂x ∂y
Hence u and v are locally constant: for every z ∈ M there is a neighborhood Uz ⊂ M on which
u and v are constant functions.
Fix an arbitrary point z0 ∈ M and consider the level set

S := {z ∈ M : H(z) = H(z0 )} = {z ∈ M : u(z) = u(z0 ) and v(z) = v(z0 )}.

We show S is nonempty, open and closed in M .

• S is nonempty because z0 ∈ S.
• S is open in M : if z ∈ S, then u and v are constant on some neighborhood Uz ⊂ M , so
every point of Uz has the same (u, v)-values as z, hence Uz ⊂ S. Thus S is a union of
neighborhoods and therefore open.
• S is closed in M : S is the preimage H −1 ({H(z0 )}) of a closed singleton under the contin-
uous map H, so S is closed.

Since M is a region (nonempty and connected) and S ⊂ M is nonempty, open and closed, we
must have S = M . Therefore H(z) = H(z0 ) for every z ∈ M , i.e. H is constant on M .
Thus F − G is constant on M , and any two antiderivatives of f differ by a constant.

2. Consider the parametrized smooth curve u : [0, 44] → C given by u(t) := t + t3/2 i. Show that the
 length of this curve equals 296.

Solution:

Proof. The curve is

u(t) = t + i t3/2 , t ∈ [0, 44].
Differentiate:
3
u′ (t) = 1 + i · t1/2 ,
2
so r 3 2 r 9
′ 1/2
|u (t)| = 1 + t = 1 + t.
2 4
The length of the curve is
Z 44 Z 44
r
′ 9
L= |u (t)| dt = 1 + t dt.
0 0 4

Make the substitution s = 1 + 94 t, so ds = 94 dt and dt = 94 ds. When t = 0 we have s = 1, and

when t = 44 we have s = 1 + 49 · 44 = 1 + 99 = 100. Hence
Z 100
4 4 2  3/2 100 8
s1/2 ds = 1003/2 − 1 . = 296.


L= · s 1
=
9 1 9 3 27

Therefore the length of the curve is 296.

3. Let γ be the negatively oriented circle |z| = 1. Show that

Z
sin(z) πi
4
dz = .
γ z 3

Solution:

Proof. Recall the Taylor series for sin z about 0,

∞
X z 2k+1
sin z = (−1)k ,
(2k + 1)!
k=0

which converges for every z ∈ C. We first show this series converges uniformly on the unit circle
|z| = 1. For |z| = 1 we have

z 2k+1 |z|2k+1 1
= = .
(2k + 1)! (2k + 1)! (2k + 1)!
P∞ 1
Since the numerical series k=0 (2k+1)! converges: indeed it is bounded by

∞
X 1
=e
n=0
n!

Page 2
 Dividing termwise by z 4 (which is legitimate on |z| = 1 because z ̸= 0 there) produces the
Laurent series
∞ 2k−3
sin z X
k z
= (−1) ,
z4 (2k + 1)!
k=0

and this series also converges uniformly on |z| = 1 because

z 2k−3 1
=
(2k + 1)! (2k + 1)!

for |z| = 1, so the same bound applies

Parameterize the negatively oriented (clockwise) unit circle by

z(t) = e−it , 0 ≤ t ≤ 2π

Then
dz = −ie−it dt = −iz(t) dt
Substituting this parameterization we obtain
∞
2π X −it 2k−3 ∞ 2π
−i
Z Z Z
sin z k (e ) −it
X
dz = (−1) (−ie ) dt = (−1)k
e−i(2k−2)t dt.
γ z4 0 (2k + 1)! (2k + 1)! 0
k=0 k=0

For any integer m, (

Z 2π 2π, m = 0,
e−imt dt =
0 0, m ̸= 0,
because e−imt has mean zero over a full period unless it is the constant function 1.
In the present sum m = 2k − 2, so the only term that survives is the one with 2k − 2 = 0, i.e.
k = 1.
For k = 1 the coefficient equals
(−1)1 /(3!) = −1/6
and the corresponding integral equals 2π. Therefore the entire sum reduces to
 
1 πi
− (−i)(2π) = .
6 3

Thus Z
sin z πi
dz = ,
γ z4 3
as required.

4. Determine the domain and range of the following complex functions:

(a) f (z) = ez
(b) f (z) = sin(z)/ cos(z)

Page 3
Solution:

Proof.
(a) For any complex number z = x + iy, we have

ez = ex+iy = ex (cos y + i sin y).

The modulus of ez is |ez | = ex , which can take any positive real value since x ∈ R, and its
argument is arg(ez ) = y (mod 2π). Therefore:
• The domain of f is all of C, because the power series for ez converges everywhere.
• The range of f is C \ {0}, since ez never vanishes and every nonzero complex number
can be written as reiθ for some r > 0 and θ ∈ R.
(b) Using the exponential representations

eiz − e−iz eiz + e−iz

sin z = , cos z = ,
2i 2
we find
sin z 1 e2iz − 1
f (z) = = · 2iz .
cos z i e +1
This shows that f is defined precisely when e2iz + 1 ̸= 0. The equality e2iz = −1 holds
when
π
2iz = i(2k + 1)π ⇐⇒ z = + kπ, k ∈ Z.
2
Hence the domain of f is nπ o
C\ + kπ : k ∈ Z .
2
For the range, note that for any w ∈ C, we can solve

1 e2iz − 1
w= ·
i e2iz + 1

for e2iz , obtaining

1 + iw
e2iz = .
1 − iw
This value is defined for every w ̸= −i, so on this domain there always exists some z ∈ C
satisfying the equation. Therefore the range of f is C \ {−i}.

Page 4