2 Complex Functions and the Cauchy-Riemann

Equations
2.1 Complex functions
In one-variable calculus, we study functions f(x) of a real variable x. Like-
wise, in complex analysis, we study functions f(z) of a complex variable
z C (or in some region of C). Here we expect that f(z) will in general
take values in C as well. However, it will turn out that some functions are
better than others. Basic examples of functions f(z) that we have already
seen are: f(z) = c, where c is a constant (allowed to be complex), f(z) = z,
f(z) = z, f(z) = Re z, f(z) = Imz, f(z) = |z|, f(z) = e
z
. The func-
tions f(z) = arg z, f(z) =

z, and f(z) = log z are also quite interesting,
but they are not well-dened (single-valued, in the terminology of complex
analysis).
What is a complex valued function of a complex variable? If z = x +iy,
then a function f(z) is simply a function F(x, y) = u(x, y) + iv(x, y) of the
two real variables x and y. As such, it is a function (mapping) from R
2
to
R
2
. Here are some examples:
1. f(z) = z corresponds to F(x, y) = x + iy (u = x, v = y);
2. f(z) = z, with F(x, y) = x iy (u = x, v = y);
3. f(z) = Re z, with F(x, y) = x (u = x, v = 0, taking values just along
the real axis);
4. f(z) = |z|, with F(x, y) =

x
2
+ y
2
(u =

x
2
+ y
2
, v = 0, taking
values just along the real axis);
5. f(z) = z
2
, with F(x, y) = (x
2
y
2
) + i(2xy) (u = x
2
y
2
, v = 2xy);
6. f(z) = e
z
, with F(x, y) = e
x
cos y + i(e
x
siny) (u = e
x
cos y, v =
e
x
siny).
If f(z) = u + iv, then the function u(x, y) is called the real part of f and
v(x, y) is called the imaginary part of f. Of course, it will not in general be
possible to plot the graph of f(z), which will lie in C
2
, the set of ordered
pairs of complex numbers, but it is the set {(z, w) C
2
: w = f(z)}. The
graph can also be viewed as the subset of R
4
given by {(x, y, s, t) : s =
u(x, y), t = v(x, y)}. In particular, it lies in a four-dimensional space.
The usual operations on complex numbers extend to complex functions:
given a complex function f(z) = u+iv, we can dene functions Re f(z) = u,
1
Imf(z) = v, f(z) = u iv, |f(z)| =

u
2
+ v
2
. Likewise, if g(z) is another
complex function, we can dene f(z)g(z) and f(z)/g(z) for those z for which
g(z) = 0.
Some of the most interesting examples come by using the algebraic op-
erations of C. For example, a polynomial is an expression of the form
P(z) = a
n
z
n
+ a
n1
z
n1
+ + a
0
,
where the a
i
are complex numbers, and it denes a function in the usual
way. It is easy to see that the real and imaginary parts of a polynomial P(z)
are polynomials in x and y. For example,
P(z) = (1 + i)z
2
3iz = (x
2
y
2
2xy + 3y) + (x
2
y
2
+ 2xy 3x)i,
and the real and imaginary parts of P(z) are polynomials in x and y. But
given two (real) polynomial functions u(x, y) and z(x, y), it is very rarely the
case that there exists a complex polynomial P(z) such that P(z) = u + iv.
For example, it is not hard to see that x cannot be of the form P(z), nor can
z. As we shall see later, no polynomial in x and y taking only real values for
every z (i.e. v = 0) can be of the form P(z). Of course, since x =
1
2
(z + z)
and y =
1
2i
(z z), every polynomial F(x, y) in x and y is also a polynomial
in z and z, i.e.
F(x, y) = Q(z, z) =

i,j0
c
ij
z
i
z
j
,
where c
ij
are complex coecients.
Finally, while on the subject of polynomials, let us mention the
Fundamental Theorem of Algebra (rst proved by Gauss in 1799): If
P(z) is a nonconstant polynomial, then P(z) has a complex root. In other
words, there exists a complex number c such that P(c) = 0. From this, it is
easy to deduce the following corollaries:
1. If P(z) is a polynomial of degree n > 0, then P(z) can be factored
into linear factors:
P(z) = a(z c
1
) (z c
n
),
for complex numbers a and c
1
, . . . , c
n
.
2. Every nonconstant polynomial p(x) with real coecients can be fac-
tored into (real) polynomials of degree one or two.
2
Here the rst statement is a consequence of the fact that c is a root of P(z) if
and only if (zc) divides P(z), plus induction. The second statement follows
from the rst and the fact that, for a polynomial with real coecients,
complex roots occur in conjugate pairs.
One consequence of the Fundamental Theorem of Algebra is that, having
enlarged the real numbers so as to have a root of the polynomial equation
x
2
+1 = 0, we are now miraculously able to nd roots of every polynomial
equation, including the ones where the coecients are allowed to be complex.
This suggests that it is very hard to further enlarge the complex numbers in
such a way as to have any reasonable algebraic properties. Finally, we should
mention that, despite its name, the Fundamental Theorem of Algebra is not
really a theorem in algebra, and in fact some of the most natural proofs of
this theorem are by using methods of complex function theory.
We can dene a broader class of complex functions by dividing polynomi-
als. By denition, a rational function R(z) is a quotient of two polynomials:
R(z) = P(z)/Q(z),
where P(z) and Q(z) are polynomials and Q(z) is not identically zero. Using
the factorization (1) above, it is not hard to see that, if R(z) is not actually
a polynomial, then it fails to be dened, roughly speaking, at the roots of
Q(z) which are not also roots of P(z), and thus at nitely many points in
C. (We have to be a little careful if there are multiple roots.)
For functions of a real variable, the next class of functions we would
dene might be the algebraic functions, such as

x or
5

1 + x
2
2

1 + x
4
.
However, in the case of complex functions, it turns out to be fairly involved
to keep track of how to make sure these functions are well-dened (single-
valued) and we shall therefore not try to discuss them here.
Finally, there are complex functions which can be dened by power series.
We have already seen the most important example of such a function, e
z
=

n=0
z
n
/n!, which is dened for all z. Other examples are, for instance,
1
1 z
=

n=0
z
n
, |z| < 1.
However, to make sense of such expressions, we would have to discuss con-
vergence of sequences and series for complex numbers. We will not do so
here, but will give a brief discussion below of limits and continuity for com-
plex functions. (It turns out that, once things are set up correctly, the
comparison and ratio tests work for complex power series.)
3
2.2 Limits and continuity
The absolute value measures the distance between two complex numbers.
Thus, z
1
and z
2
are close when |z
1
z
2
| is small. We can then dene the
limit of a complex function f(z) as follows: we write
lim
zc
f(z) = L,
where c and L are understood to be complex numbers, if the distance from
f(z) to L, |f(z) L|, is small whenever |z c| is small. More precisely,
if we want |f(z) L| to be less than some small specied positive real
number , then there should exist a positive real number such that, if
|z c| < , then |f(z) L| < . Note that, as with real functions, it does
not matter if f(c) = L or even that f(z) be dened at c. It is easy to see
that, if c = (c
1
, c
2
), L = a + bi and f(z) = u + iv is written as a real and
an part, then lim
zc
f(z) = L if and only if lim
(x,y)(c
1
,c
2
)
u(x, y) = a and
lim
(x,y)(c
1
,c
2
)
v(x, y) = b. Thus the story for limits of functions of a complex
variable is the same as the story for limits of real valued functions of the
variables x, y. However, a real variable x can approach a real number c only
from above or below (or from the left or right, depending on your point of
view), whereas there are many ways for a complex variable to approach a
complex number c.
Sequences, limits of sequences, convergent series and power series can be
dened similarly.
As for functions of a real variable, a function f(z) is continuous at c if
lim
zc
f(z) = f(c).
In other words: 1) the limit exists; 2) f(z) is dened at c; 3) its value at c
is the limiting value. A function f(z) is continuous if it is continuous at all
points where it is dened. It is easy to see that a function f(z) = u + iv
is continuous if and only if its real and imaginary parts are continuous, and
that the usual functions z, z, Re z, Imz, |z|, e
z
are continuous. (We have to
be careful, though, about functions such as arg z or log z which are not
well-dened.) All polynomials P(z) are continuous, as are all two-variable
polynomial functions in x and y. A rational function R(z) = P(z)/Q(z) with
Q(z) not identically zero is continuous where it is dened, i.e. at the nitely
many points where the denominator Q(z) is not zero. More generally, if
f(z) and g(z) are continuous, then so are:
1. cf(z), where c is a constant;
4
2. f(z) + g(z);
3. f(z) g(z);
4. f(z)/g(z), where dened (i.e. where g(z) = 0).
5. (g f)(z) = g(f(z)), the composition of g(z) and f(z), where dened.
2.3 Complex derivatives
Having discussed some of the basic properties of functions, we ask now
what it means for a function to have a complex derivative. Here we will
see something quite new: this is very dierent from asking that its real and
imaginary parts have partial derivatives with respect to x and y. We will
not worry about the meaning of the derivative in terms of slope, but only
ask that the usual dierence quotient exists.
Denition A function f(z) is complex dierentiable at c if
lim
zc
f(z) f(c)
z c
exists. In this case, the limit is denoted by f

(c). Making the change of

variable z = c +h, f(z) is complex dierentiable at c if and only if the limit
lim
h0
f(c + h) f(c)
h
exists, in which case the limit is again f

(c). A function is complex dieren-

tiable if it is complex dierentiable at every point where it is dened. For
such a function f(z), the derivative denes a new function which we write
as f

(z) or
d
dz
f(z).
For example, a constant function f(z) = C is everywhere complex dier-
entiable and its derivative f

(z) = 0. The function f(z) = z is also complex

dierentiable, since in this case
f(z) f(c)
z c
=
z c
z c
= 1.
Thus (z)

= 1. But many simple functions do not have complex derivatives.

For example, consider f(z) = Re z = x. We show that the limit
lim
h0
f(c + h) f(c)
h
5
does not exist for any c. Let c = a + bi, so that f(c) = a. First consider
h = t a real number. Then f(c + t) = a + t and so
f(c + h) f(c)
h
=
a + t a
t
= 1.
So if the limit exists, it must be 1. On the other hand, we could use h = it.
In this case, f(c + it) = f(c) = a, and
f(c + h) f(c)
h
=
a a
it
= 0.
Thus approaching c along horizontal and vertical directions has given two
dierent answers, and so the limit cannot exist. Other simple functions
which can be shown not to have complex derivatives are Imz, z, and |z|.
On the bright side, the usual rules for derivatives can be checked to hold:
1. If f(z) is complex dierentiable, then so is cf(z), where c is a constant,
and (cf(z))

= cf

(z);
2. (Sum rule) If f(z) and g(z) are complex dierentiable, then so is f(z)+
g(z), and (f(z) + g(z))

= f

(z) + g

(z);
3. (Product rule) If f(z) and g(z) are complex dierentiable, then so is
f(z) g(z) and (f(z) g(z))

= f

(z)g(z) + f(z)g

(z);
4. (Quotient rule) If f(z) and g(z) are complex dierentiable, then so is
f(z)/g(z), where dened (i.e. where g(z) = 0), and

f(z)
g(z)

=
f

(z)g(z) f(z)g

(z)
g(z)
2
;
5. (Chain rule) If f(z) and g(z) are complex dierentiable, then so is
f(g(z)) where dened, and (f(g(z)))

= f

(g(z)) g

(z).
6. (Inverse functions) If f(z) is complex dierentiable and one-to-one,
with nonzero derivative, then the inverse function f
1
(z) is also dif-
ferentiable, and
(f
1
(z))

= 1/f

(f
1
(z)).
Thus for example we have the power rule (z
n
)

= nz
n1
, every polyno-
mial P(z) = a
n
z
n
+ a
n1
z
n1
+ + a
0
is complex dierentiable, with
P

(z) = na
n
z
n1
+ (n 1)a
n1
z
n2
+ a
1
,
and every rational function is also complex dierentiable. It follows that a
function which is not complex dierentiable, such as Re z or z cannot be
written as a complex polynomial or rational function.
6
2.4 The Cauchy-Riemann equations
We now turn systematically to the question of deciding when a complex
function f(z) = u + iv is complex dierentiable. If the complex derivative
f

(z) is to exist, then we should be able to compute it by approaching z

along either horizontal or vertical lines. Thus we must have
f

(z) = lim
t0
f(z + t) f(z)
t
= lim
t0
f(z + it) f(z)
it
,
where t is a real number. In terms of u and v,
lim
t0
f(z + t) f(z)
t
= lim
t0
u(x + t, y) + iv(x + t, y) u(x, y) v(x, y)
t
= lim
t0
u(x + t, y) u(x, y)
t
+ i lim
t0
v(x + t, y) v(x, y)
t
=
u
x
+ i
v
x
.
Taking the derivative along a vertical line gives
lim
t0
f(z + it) f(z)
it
= i lim
t0
u(x, y + t) + iv(x, y + t) u(x, y) v(x, y)
t
= i lim
t0
u(x, y + t) u(x, y)
t
+ lim
t0
v(x, y + t) v(x, y)
t
= i
u
y
+
v
y
.
Equating real and imaginary parts, we see that: If a function f(z) =
u+iv is complex dierentiable, then its real and imaginary parts satisfy the
Cauchy-Riemann equations:
u
x
=
v
y
;
v
x
=
u
y
.
Moreover, the complex derivative f

(z) is then given by

f

(z) =
u
x
+ i
v
x
=
v
y
i
u
y
.
Examples: the function z
2
= (x
2
y
2
) + 2xyi satises the Cauchy-
Riemann equations, since

x
(x
2
y
2
) = 2x =

y
(2xy) and

x
(2xy) = 2y =

y
(x
2
y
2
).
7
Likewise, e
z
= e
x
cos y + ie
x
siny satises the Cauchy-Riemann equations,
since

x
(e
x
cos y) = e
x
cos y =

y
(e
x
siny) and

x
(e
x
siny) = e
x
siny =

y
(e
x
cos y).
Moreover, e
z
is in fact complex dierentiable, and its complex derivative is
d
dz
e
z
=

x
(e
x
cos y) +

x
(e
x
siny) = e
x
cos y + e
x
siny = e
z
.
The chain rule then implies that, for a complex number ,
d
dz
e
z
= e
z
.
One can dene cos z and sinz in terms of e
iz
and e
iz
(see the homework).
From the sum rule and the expressions for cos z and sinz in terms of e
iz
and
e
iz
, it is easy to check that cos z and sinz are analytic and that the usual
rules hold:
d
dz
cos z = sinz;
d
dz
sinz = cos z.
On the other hand, z does not satisfy the Cauchy-Riemann equations, since

x
(x) = 1 =

y
(y).
Likewise, f(z) = x
2
+iy
2
does not. Note that the Cauchy-Riemann equations
are two equations for the partial derivatives of u and v, and both must be
satised if the function f(z) is to have a complex derivative.
We have seen that a function with a complex derivative satises the
Cauchy-Riemann equations. In fact, the converse is true:
Theorem: Let f(z) = u + iv be a complex function dened in a region
(open subset) D of C, and suppose that u and v have continuous rst partial
derivatives with respect to x and y. If u and v satisfy the Cauchy-Riemann
equations, then f(z) has a complex derivative.
The proof of this theorem is not dicult, but involves a more careful
understanding of the meaning of the partial derivatives and linear approxi-
mation in two variables.
Thus we see that the Cauchy-Riemann equations give a complete cri-
terion for deciding if a function has a complex derivative. There is also
a geometric interpretation of the Cauchy-Riemann equations. Recall that
u =

u
x
,
u
y

and that v =

v
x
,
v
y

. Then u and v satisfy the

Cauchy-Riemann equations if and only if
v =

v
x
,
v
y

u
y
,
u
x

.
8
If this holds, then the level curves u = c
1
and v = c
2
are orthogonal where
they intersect.
Instead of saying that a function f(z) has a complex derivative, or equiv-
alently satises the Cauchy-Riemann equations, we shall call f(z) analytic
or holomorphic. Here are some basic properties of analytic functions, which
are easy consequences of the Cauchy-Riemann equations:
Theorem: Let f(z) = u + iv be an analytic function.
1. If f

(z) is identically zero, then f(z) is a constant.

2. If either Re f(z) = u or Imf(z) = v is constant, then f(z) is constant.
In particular, a nonconstant analytic function cannot take only real or
only pure imaginary values.
3. If |f(z)| is constant or arg f(z) is constant, then f(z) is constant.
For example, if f

(z) = 0, then
0 = f

(z) =
u
x
+ i
v
x
.
Thus
u
x
=
v
x
= 0. By the Cauchy-Riemann equations,
v
y
=
u
y
= 0
as well. Hence f(z) is a constant. This proves (1). To see (2), assume
for instance that u is constant. Then
u
x
=
u
y
= 0, and, as above, the
Cauchy-Riemann equations then imply that
v
x
=
v
y
= 0. Again, f(z) is
constant. Part (3) can be proved along similar but more complicated lines.
2.5 Harmonic functions
Let f(z) = u + iv be an analytic function, and assume that u and v have
partial derivatives of order 2 (in fact, this turns out to be automatic). Then,
using the Cauchy-Riemann equations and the equality of mixed partials, we
have:

2
u
x
2
=

x
u
x
=

x
v
y
=

y
v
x
=

y
u
y
=

2
u
y
2
.
In other words, u satises:

2
u
x
2
+

2
u
y
2
= 0.
9
The above equation is a very important second order partial dierential
equation, and solutions of it are called harmonic functions. Thus, the real
part of an analytic function is harmonic. A similar argument shows that
v is also harmonic, i.e. the imaginary part of an analytic function is har-
monic. Essentially, all harmonic functions arise as the real parts of analytic
functions.
Theorem: Let D be a simply connected region in C and let u(x, y) be a real-
valued, harmonic function in D. Then there exists a real-valued function
v(x, y) such that f(z) = u + iv is an analytic function.
We will discuss the meaning of the simply connected condition in the
exercises in the next handout. The problem is that, if D is not simply con-
nected, then it is possible that u can be completed to an analytic function
f(z) = u+iv which is not single-valued, even if u is single valued. The basic
example is Re log z =
1
2
ln(x
2
+y
2
). A calculation (left as homework) shows
that this function is harmonic. But an analytic function whose real part is
the same as that of log z must agree with log z up to an imaginary constant,
and so cannot be single-valued.
The point to keep in mind is that we can generate lots of harmonic
functions, in fact essentially all of them, by taking real or imaginary parts of
analytic functions. Harmonic functions are very important in mathematical
physics, and one reason for the importance of analytic functions is their
connection to harmonic functions.
2.6 Homework
1. Write the function f(z) in the form u + iv:
(a) z + iz
2
; (b) 1/z; (c) z/z.
2. If f(z) = e
z
, describe the images under f(z) of horizontal and vertical
lines, i.e. what are the sets f(a + it) and f(t + ib), where a and b are
constants and t runs through all real numbers?
3. Is the function z/z continuous at 0? Why or why not? Is the function
z/z analytic where it is dened? Why or why not?
4. Compute the derivatives of the following analytic functions, and be
prepared to justify your answers:
(a)
iz + 3
z
2
(2 + i)z + (4 3i)
; (b) e
z
2
; (c)
1
e
z
+ e
z
.
10
5. Let f(z) be a complex function. Is it possible for both f(z) and f(z)
to be analytic?
6. Let f(z) = u+iv be analytic. Recall that the Jacobian is the function
given by the following determinant:
(u, v)
(x, y)
=

u/x u/y
v/x v/y

.
Using the Cauchy-Riemann equations, show that this is the same as
|f

(z)|
2
.
7. Dene the complex sine and cosine functions as follows:
cos z =
1
2
(e
iz
+ e
iz
); sinz =
1
2i
(e
iz
e
iz
).
Note that, if t is real, then this denition of cos t and sint agree with
the usual ones, and that (for those who remember hyperbolic func-
tions) cos z = coshiz and sinz = i sinhiz. Verify that cos z and sinz
are analytic and that (cos z)

= sinz and that (sinz)

= cos z. Write
cos z as u+iv, where u and v are real-valued functions of x and y, and
similarly for sinz.
8. Verify that Re 1/z, Im1/z, and Re log z =
1
2
ln(x
2
+y
2
) are harmonic.
9. Which of the following are harmonic?
(a) x
3
y
3
; (b) x
3
y xy
3
; (c) x
2
2xy.
10. If f(z) = u + iv is a complex function such that u and v are both
harmonic, is f(z) necessarily analytic?
11