18.112: Functions of A Complex Variable: Lecturer: Professor Wei Zhang Notes By: Andrew Lin Fall 2019
18.112: Functions of A Complex Variable: Lecturer: Professor Wei Zhang Notes By: Andrew Lin Fall 2019
18.112: Functions of A Complex Variable: Lecturer: Professor Wei Zhang Notes By: Andrew Lin Fall 2019
Fall 2019
1 Introduction
The course syllabus can be found on Stellar, and we were also given a paper copy in class.
This class will assign nine problem sets, roughly due every week except for the three weeks with exams. The two
midterms will be quiz-style – they’re only going to be 80 minutes long, since that’s how much time we have for class,
and they’ll be on October 3rd and November 5th. (The questions will be pretty basic for the quizzes.) The final exam
will take place during finals week, date and location to be determined. Grading is 30% homework (one problem set
dropped), 30% final, 20% for each midterm.
Collaboration is allowed, but we’re encouraged to solve the problems ourselves first. If we do talk with others, we
should indicate all collaborators on the homework itself. As per policy, late homework won’t be accepted, but certain
situations can be accommodate e.g. from S3 .
2 September 5, 2019
Let’s start with some motivation for why we care about functions of a complex variable.
Fact 1 (Notation)
R will denote the reals, Z will denote the integers, and N will denote the natural (positive) integers throughout
this class.
Here, R ⊃ Z ⊃ N, and we’ll be dealing with a superset of the real numbers: C, the complex numbers. We
can identify elements of C naturally as ordered pairs of real numbers: in other words, C can be thought of as a
2-dimensional vector space over R. To review, any complex number z can be written as x + y i for real numbers x and
√
y , where i = −1.
Remark 2. Mathematics defies English here: we say "I is" instead of "I am."
But why do we even care about C in the first place? Are there any applications for this field of study?
Definition 3
Prime numbers are the set of natural numbers whose only divisors are 1 and themselves. The prime counting
function
π(x) = number of primes p : p < x
1
One important result in number theory is the following result:
π2
(For example, we know that ζ(2) = 6 .) The important change of perspective here is that we can replace s with
an arbitrary complex number, and there are a lot of interesting things that come out of this! Obviously, ζ(s) isn’t
convergent everywhere, but it will be convergent for Re(s) > 1, and we’ll study this a bit more later: it turns out we
can prove the prime number theorem using the complex Riemann zeta function.
One good way to start studying functions of a complex variable is to restrict our attention to some “nice” functions.
When we have an arbitrary continuous function f : R → R, we generally like it a lot when we can consider the derivative
f (x + h) − f (x)
f 0 (x) = lim .
h→0 h
If this quantity f 0 exists, then we call our function differentiable. Let’s see if we can copy this over to complex
numbers:
Definition 5
Fix a point z ∈ C. Given a function f : C → C, define the derivative
f (z + h) − f (z)
f 0 (z) = lim ,
h→0 h
if it exists.
We should be a bit more specific here with what we want: since h is not just a real number anymore, what does it
mean for h to go to 0?
To answer this, we need to think about the idea of convergence.
Definition 6
A sequence of complex numbers {zn } converges to z if
lim |zn − z| = 0.
n→∞
(Note that this last statement can be interpreted as an expression in the real numbers.) Let’s try to introduce a
few more real analysis concepts into our language here:
2
Definition 7
{zn } is a Cauchy sequence if
|zn − zm | → 0 as n, m → ∞.
Definition 8
A set Ω is open if every point has an open neighborhood contained in Ω. A set is closed if its complement is open.
Definition 9
Ω is compact if it is closed and bounded.
Example 10
D(z0 , r ), the closed disk centered at z0 with radius r , is compact. However, D(z0 , r ), the open disk, is not
compact, though it is bounded.
The reason we care about compact sets is that every sequence whose elements are in a compact set has a
convergent subsequence.
Definition 11
Ω is connected if we cannot write it as a disjoint union Ω1 t Ω2 of open sets. Ω is path-connected if any two
points x, y in Ω can be connected by a path that lies inside Ω. (We’ll call open connected sets regions.)
In Rn , being open and connected is the same as being path-connected, but this isn’t true in a general topological
space.
Definition 12
Let Ω be an open subset of C (which can be C itself). A function f : Ω → C is holomorphic at z if the derivative
exists at z, and f is holomorphic on Ω if f is holomorphic at every z ∈ Ω.
Demanding that f be holomorphic is actually pretty strong: since h can approach 0 from basically any angle in the
complex plane, and we need to arrive at the same answer in all directions.
Example 13
The function f (z) = z n is holomorphic for all n: expanding out (z + h)n with the binomial theorem is still valid,
so the power rule still holds and the derivative exists: f 0 (z) = nz n−1 . However, g(z) = z is not holomorphic:
g(z + h) − g(z) h
= ,
h h
and notice that h and h will have the same norm as h approaches zero, but the arguments will be different (and
opposite). Thus, g 0 (z) does not exist.
3
Somehow, then, we want functions that involve only z, not z: how do we quantify this? Notice that when we have
a function f : C → C, we can think of this as taking ordered pairs (x, y ) ∈ R2 to other ordered pairs (u, v ) ∈ R2 . So
let’s try to study f in this new context!
Notice that u and v are both functions of x and y , so we have four partial derivatives: all four of them are being
used here!
Proof. We can get expressions in these partial derivatives by looking at specific limits: for example,
f (x + i y + h) − f (x + i y ) ∂f ∂u ∂v
lim = = +i .
h∈R h ∂x ∂x ∂x
h→0
f (x + i (y + h)) − f (x + i y ) 1 ∂f 1 ∂u ∂v
lim = = + .
h∈R ih i ∂y i ∂y ∂y
h→0
But if the limit exists, these two are both equal to f 0 (x + i y ), so we can equate real and imaginary parts:
∂u ∂v 1 ∂u ∂v
= and =i .
∂x ∂y i ∂y ∂x
Proposition 15
A function f is holomorphic if and only if
∂f 1 ∂f
= .
∂x i ∂y
where φ(h) → 0 as h → 0. (This can also be written as df = f 0 (z)dz .) We can (intuitively) write something that
looks like
∂f ∂f
df = dz + dz
∂z ∂z
with the multivariable chain rule. So now
df = d(u + i v ) = du + i dv = ux dx + uy dy + i vx dx + i vy dy ,
4
which yields the differentials
1 1 1 1
dx = dz + dz , dy = dz − dz .
2 2 2i 2i
So now looking at the right side of the boxed expression,
∂f ∂f ∂x ∂f ∂y
= +
∂z ∂x ∂z ∂y ∂z
again by the multivariable chain rule, which simplifies to
1 ∂f ∂f
+ .
2 ∂x ∂y