Analytic Functions

“Differentiable Functions of a Complex Variable”

In this chapter, we shall generalize the ideas for polynomials and power
series of a complex variable we developed in the previous chapter to
general functions of a complex variable. Once we have proved results to
determine whether or not a function is analytic, we shall then consider
generalizations of some of the more common single variable functions
which are not polynomials - namely trigonometric functions and expo-
nential functions.

1. Analyticity and the Cauchy-Riemann Equations

Recall that for a polynomial P (x, y) = u(x, y) + iv(x, y) with complex
coefficients, we showed that it was analytic, or differentiable as a func-
tion of a complex variable if and only if Py = iPx , or equivalently if
it satisfied the Cauchy-Riemann equations ux = vy and uy = −vx . In
this section, we shall show that this result can be partially extended to
any function of a complex variable.
1.1. Determining whether a Function is Analytic. First we shall
show that if a function of a complex variable is differentiable, then
it must satisfy the Cauchy-Riemann equations (so it is a necessary
condition to satisfy CR).
Proposition 1.1. If f = u + iv is differentiable at z, then fx and fy
exist and satisfy the CR equations i.e.
fy = ifx
or
ux = vy ; uy = −vx .
Proof. Before we prove the result, we make a couple of observations.
First note that since we are assuming that f is differentiable,
f (z + h) − f (z)
lim
h→0 h
exists regardless of the way h approaches 0. Secondly, we observe that
by definition,
f (x + h, y) − f (x, y)
fx = lim
h→0 h
and
f (x, y + h) − f (x, y)
fy = lim .
h→0 h
1
2

Now if we take h to be real, then we have

f (z + h) − f (z) f (x + iy + h) − f (x, y)
lim = lim
h→0 h h→0 h
f ((x + h) + iy) − f (x, y) f (x + h, y) − f (x, y)
= lim = lim = fx (z).
h→0 h h→0 h
Likewise, taking h = iη purely imaginary, we have
f (z + h) − f (z) f (x, y + η) − f (x, y) fy
lim = lim = .
h→0 ih η→0 iη i
Since the limit
f (z + h) − f (z)
lim
h→0 h
exists independent of direction, it follows that fx = fy /i or fy = ifx .
The CR equations follow.

Unfortunately, just because the CR equations exist does not mean that
a function is differentiable as the following example shows.
Example 1.2. Show that the function
p
f (x, y) = |xy|
satisfies the CR equations but is not differentiable at (0, 0).
Here we have
f (0 + h) − f (0, 0) 0−0
fx (0, 0) = lim = lim =0
h→0 h h→0 h
for h real, and
f (0 + h) − f (0, 0) 0−0
fx (0, 0) = lim = lim =0
h→0 h h→0 h
for h imaginary, so 0 = fy (0, 0) = ifx = 0. However, if we approach
along the line y = x, we have
√
f (0 + h(1 + i)) − f (0, 0) h2 − 0 1
lim = lim =±
h→0 h(1 + i) h→0 h(1 + i) 1+i
for h real. In particular, this limit doesn’t exist and so f is not differ-
entiable.
As the last example illustrates, Satisfying the CR equations are not a
sufficient condition for analyticity (unlike with polynomials and power
series). Under certain stronger assumptions however, the CR equations
are enough.
Proposition 1.3. Suppose fx and fy exist in a neighbourhood of z.
Then if fx and fy are continuous at z and fy (z) = ifx (z), then f is
differentiable at z.
 3

Proof. We need to show that the limit

f (z + h) − f (z)
lim
h→0 h
exists independent of how h approaches 0. We shall do this by showing
that
f (z + h) − f (z)
lim = fx (z) = ux (z) + ivx (z)
h→0 h
where f = u(x, y) + iv(x, y).
Let h = ζ + iη where ζ and η are real numbers. We consider u(x, y)
and v(x, y) separately. Observe that
u(z + h) − u(z) u(x + ζ, y + η) − u(x, y)
=
h ζ + iη
1
= ([u(x + ζ, y + η) − u(x + ζ, y)] + [u(x + ζ, y) − u(x, y)]).
ζ + iη
Note that u(x+ζ, y +η)−u(x+ζ, y) is the change of the single variable
function f (y) = u(x + ζ, y) over the interval [y, y + η]. Therefore,
since fy exists and is continuous (by assumption), so is uy , so the mean
value theorem for single real variable functions implies there exists some
number, say y + ϑ1 η (for some 0 6 ϑ1 6 1) in the interval [y, y + η]
such that
u(x + ζ, y + η) − u(x + ζ, y) = ((y + η) − y)uy (x + ζ, y + ϑ1 η).
We get similar results for [u(x+ζ, y)−u(x, y)]), and for the components
of the function v(x, y) giving
u(z + h) − u(z) η ζ
= uy (x + ζ, y + ϑ1 η) + ux (x + ϑ2 ζ, y)
h ζ + iη ζ + iη
and
v(z + h) − v(z) η ζ
= vy (x + ζ, y + ϑ3 η) + vx (x + ϑ4 ζ, y).
h ζ + iη ζ + iη
Therefore, we get
   
f (z + h) − f (z) η ζ
= uy (z1 )+ivy (z2 ) + ux (z3 )+ivx (z4 )
h ζ + iη ζ + iη
where zk → z for k = 1, 2, 3, 4 as h → 0.
Next we observe that since ifx = fy , we have
ηifx + ζfx η ζ
fx = = fy + fx .
ζ + iη ζ + iη ζ + iη
Using these equalities, subtracting fx from
f (z + h) − f (z)
h
4

we get
f (z + h) − f (z)
− fx (z)
h
     
η ζ η ζ
= uy (z1 )+ivy (z2 ) + ux (z3 )+ivx (z4 ) − fy + fx
ζ + iη ζ + iη ζ + iη ζ + iη
   
η ζ
= uy (z1 ) + ivy (z2 ) − fy (z) + ux (z3 ) + ivx (z4 ) − fx (z) .
ζ + iη ζ + iη
Now note that
η ζ
| |, | |61
ζ + iη ζ + iη
for all h and
uy (z1 ) + ivy (z2 ) − fy (z), ux (z3 ) + ivx (z4 ) − fx (z) → 0
as h → 0, so it follows that
f (z + h) − f (z)
− fx (z) → 0
h
or
f (z + h) − f (z)
→ fx (z).
h
Thus
f (z + h) − f (z)
lim = fx (z),
h→0 h
and in particular, f (z) is differentiable at z.

Usually we consider a function to be differentiable if it is differentiable
in some interval (we would not usually consider a function to be differ-
entiable only at a point, see example below).
Example 1.4. Consider f (x, y) = x2 + y 2. Here we have fx = 2x
and fy = 2y, so both partial derivatives are continuous. By the last
result, this means fx is differentiable everywhere the CR equations are
satisfied, so just at the point (0, 0).
For this reason, when we define analyticity of a function at a point,
rather than just requiring differentiability at a point, we want it to
describe the local behaviour at and near the point. Rg
Definition 1.5. We say f is analytic at z if f is differentiable in a
neighbourhood of z. Similarly, f is analytic on a set S if it is differen-
tiable on some open set containing S.
Definition 1.6. We call a function which is differentiable everywhere
an entire function.
 5

1.2. Generalizing Results from Real Variable Calculus. Now

we have a formal definition for an analytic function, we shall consider
a small number of results which we can generalized from their real
counterparts. We start by considering how to differentiate an inverse
function of a complex analytic function. In order to do this, we need
the following definition.
Definition 1.7. Suppose that S and T are open sets and that f is
1 − 1 on S with F (S) = T . g is called the inverse function of f on T
if f (g(z)) = z for z ∈ T . g is said to be the inverse of f at a point z0
if it is the inverse in some neighbourhood of z0 .
Proposition 1.8. Suppose that g is the inverse of f at z0 and that g
is continuous at z0 . If f is differentiable at g(z0 ) and if f ′ (g(z0 )) 6= 0,
then g is differentiable at z0 and
1
g ′ (z0 ) = ′ .
f (g(z0 ))
Proof. For any z 6= z0 in a neighbourhood of z0 , we have
g(z) − g(z0 ) 1
= f (g(z))−f (g(z0 )) .
z − z0
g(z)−g(z0 )

Since g is continuous at z0 , g(z) → g(z0 ) as z → z0 , so by the differen-

tiability of f
g(z) − g(z0 ) 1
lim = ′ ,
z→z0 z − z0 f (g(z0 ))
since by definition
f (g(z)) − f (g(z0 ))
f ′ (g(z0 )) = lim .
z→z0 z − z0

Other results which generalize from single variable are the following.
Proposition 1.9. If f = u+iv is analytic in a region and u is constant,
then f is constant.
Proof. Since u is constant, ux = uy = 0. Since f is analytic, it satisfies
the CR equations, so it follows that vx = vy = 0. Applying Proposition
3.17, it follows that u and v are constant and hence f is constant. 
Proposition 1.10. If f is analytic in a region D and if |f | is constant
on D, then f is constant on D.
Proof. If |f | = 0, the proof is obvious. Else we have
u2 + v 2 = C 6= 0.
Taking partial derivatives, we have
uux + vvx = 0
6

and
uuy + vvy = 0
(using the chain rule). Using the Cauchy Riemann equations, this can
be modified to get
uux − vuy = 0
and
vux + uuy = 0
which is equivalent to
u2 ux − uvuy = 0
and
v 2 ux + vuuy = 0.
Taking the difference, we get
(u2 + v 2 )ux = 0
and since u2 + v 2 6= 0, it follows that ux = vy = 0. We get a similar
result for uy and vx , so the result follows.


2. Generalizing Functions from Real Variables

We now consider generalizing some of the functions we know from real
variable calculus (by generalizing, we mean that we define a function
of a complex variable which agrees with the original real function when
evaluated at purely real numbers).
2.1. A Complex Exponential Function. We want to define a gener-
alization of the real exponential function to complex variables. Specif-
ically, we want to define a function f (z) satisfying
(i ) f (z1 + z2 ) = f (z1 )f (z2 ) for any z1 , z2 ∈ C
(ii ) f (x) = ex for any real x.
(i ) From the book, pages 41-42: Questions
Suppose that f (z) is a function which satisfies these two conditions.
Then it follows that
f (z) = f (x + iy) = f (x)f (iy) = ex f (iy),
so we need to determine what conditions will be imposed on the purely
imaginary part of a complex number. Therefore, suppose that f (iy) =
A(y) + iB(y), so we have f (x + iy) = ex A(y)iex B(y). In order for f to
be analytic, it needs to satisfy the Cauchy Riemann equations, so we
must have ux = vy and uy = −vx or
ex A(y) = ex B ′ (y)
and
ex A′ (y) = ex B(y).
 7

This means that

A′′ (y) = −A(y).
A general solution to this differential equation is
A(y) = α cos (y) + β sin (y)
for some real numbers α and β and
B(y) = −A′ (y) = −β cos (y) + α sin (y),
so we get
f (z) = ex (α cos (y) + β sin (y)) + iex (−β cos (y) + α sin (y))

= αex (cos (y) + i sin (y)) + βex (sin (y) − i cos (y)).
If f (z) agrees with ex for real numbers, we must have
1 = f (0) = α − iβ,
so α = 1 and β = 0. Putting all this together, we have the following:
Definition 2.1. We define the complex exponential function f (z) = ez
as
f (z) = f (x + iy) = ex (cos (y) + i sin (y)).
Proposition 2.2. The exponential function satisfies the following:
(i ) |ez | = ex
(ii ) ez 6= 0 for any value of z
(iii ) eiy = cis(y)
(iv ) ez = α has infinitely many solutions for α 6= 0.
(v ) (ez )′ = ez
Proof. Most of these results are fairly trivial to prove.
(i )
|ez | = ex | cos (y) + i sin (y))| = ex (cos2 (y) + sin2 (y)) = ex .
(ii ) This simply follows because ex 6= 0 for any x and cos (y) +
i sin (y)) 6= 0 for any y.
(iii )
ez = ex+iy = xx eiy = ex (cos (y) + i sin (y))
so eiy = cos (y) + i sin (y)).
(iv ) Suppose α is some non-zero complex number. Then in polar
form, we have α = reiϑ = rcis(ϑ) for some r > 0 and some
angle 0 6 ϑ < 2π. It follows that solutions to ez = ex eiy = α
will be x = ln (r) and y = ϑ + 2kπ for any integer k.
(v ) Recall that f ′ (z) = fx (z), so (ez )′ = (ex eiy )′ = ex eiy = ez .

8

2.2. Defining the Trigonometric Functions. Now we have defined

the exponential function, we can define trigonometric functions. First
we observe that for real values of y, we have
eiy = cos (y) + i sin (y))
and
e−iy = cos (y) − i sin (y)).
Thus for real values of y, we can define
 
1 iy −iy
sin (y) = e −e
2i
and  
1 iy −iy
cos (y) = e +e .
2
To define complex trigonometric, we simply extend these definitions to
complex variables. Specifically, we define them as follows:
Definition 2.3. We define the complex functions sin and cos as
 
1 iz −iz
sin (z) = e −e
2i
and  
1 iz −iz
cos (z) = e +e .
2
The complex trig functions share many of the identities with their real
counterparts. We illustrate with an example.
Example 2.4. We shall show that the indentity cos2 (z) + sin2 (z) = 1
holds for complex numbers.
Using the definitions, we have
  2   2
2 2 1 iz −iz 1 iz −iz
cos (z) + sin (z) = e +e + e −e
2 2i
   
1 2iz −2iz 1 2iz −2iz
= e +2+e − e −2+e =1
4 4
Of course, they also have many differences (most notably, complex trig
functions are not bounded).

Homework: (Pages 41 & 42) 2,4,6,8,11,13,14,16,17,19,21