2.

Functions and limits: Analyticity and Harmonic Functions

Let S be a set of complex numbers in the complex plane. For every
point z = x +iy S, we specic the rule to assign a corresponding
complex number w = u+iv. This denes a function of the complex
variable z, and the function is denoted by
w = f(z).
The set S is called the domain of denition of the function f and
the collection of all values of w is called the range of f.
1. f(z) = Arg z is dened everywhere except at z = 0, and Arg z
can assume all possible real values in the interval (, ].
2. The domain of denition of f(z) =
z +3
z
2
+1
is C\{i, i}. What is
the range of this function?
1
Real and imaginary parts of a complex function
Let z = x + iy. A complex function of the complex variable z may
be visualized as a pair of real functions of the two real variables x
and y. Let u(x, y) and v(x, y) be the real and imaginary parts of
f(z), respectively.
f(z) = u(x, y) +iv(x, y), z = x +iy.
Consider the function
f(z) = z
2
= (x +iy)
2
= x
2
y
2
+2ixy;
its real and imaginary parts are the real functions
u(x, y) = x
2
y
2
and v(x, y) = 2xy.
2
Example
Determine whether
w(z) =
iz +4
2z +3i
is one-to-one.
Solution
For any two complex numbers z
1
and z
2
, we have
w(z
1
) = w(z
2
)

iz
1
+4
2z
1
+3i
=
iz
2
+4
2z
2
+3i
2iz
1
z
2
+8z
2
3z
1
+12i = 2iz
1
z
2
+8z
1
3z
2
+12i
z
1
= z
2
.
Therefore, the function is one-to-one.
3
Complex velocity of a uid source
Find a complex function v(z) which gives the velocity of the ow at
any point z due to a uid source at the origin.
The direction of the velocity is radially outward from the uid source
and the magnitude of velocity is inversely proportional to the dis-
tance from the source. We observe
Arg v = Arg z and |v| =
k
|z|
,
where k is some real positive constant. If we write z = re
i
, the
velocity function is given by
v(z) = |v|e
iArg v
=
k
|z|
e
i
=
k
z
, z = 0.
The constant k is the strength of the source, which is related to
the amount of uid owing out from the source per unit time.
4
Let U(r) be the speed at any point on the circle |z| = r. From
physics, the speed at any point in the ow eld depends on the
radial distance from the uid source. We then have U(r) = k/r.
Write m as the amount of uid owing out from the source per unit
time so that
m = 2rU(r) = 2k
giving
k =
m
2
.
5
Example
Consider the function f(z) = z
2
and write it as
f(z) = u(x, y) +iv(x, y) where z = x +iy.
1. Find the curves in the x-y plane such that u(x, y) = and
v(x, y) = .
2. Find the curves in the u-v plane such that the preimage curves
in the x-y plane are x = a and y = b.
3. What is the image curve in the u-v plane of the closed curve:
r = 2(1 +cos ) in the x-y plane?
6
Solution
For z = x +iy,
f(z) = (x +iy)
2
= x
2
y
2
+2ixy
so that
u(x, y) = x
2
y
2
and v(x, y) = 2xy.
For all points on the hyperbola: x
2
y
2
= in the x-y plane, the
image points in the w plane are on the coordinate curve u = .
Similarly, the points on the hyperbola: 2xy = are mapped onto
the coordinate curve v = .
7
Recall the result:
(u +iv)
1/2
=
_

_u +
_
u
2
+v
2
2
+i
v
|v|

_
_
u
2
+v
2
u
2
_

_
.
The image curve in the w-plane corresponding to x = a is given by
u +
_
u
2
+v
2
2
= a
2
4a
2
(a
2
u) = v
2
.
Similarly, the image curve in the w-plane corresponding to y = b is
given by
_
u
2
+v
2
u
2
= b
2
4b
2
(b
2
+u) = v
2
.
Both image curves are parabolas in the w-plane.
8
Mapping properties of the complex function w = z
2
(i) For z = iy, y > 0, we have w = z
2
= y
2
so that the line:
x = 0, y > 0 is mapped onto the line: u < 0, v = 0.
(ii) For z = x, we have w = z
2
= x
2
so that the line: 0 x a, y = 0
is mapped onto the line: 0 u a
2
, v = 0.
(iii) For z = a +iy, y > 0, we have u = a
2
y
2
and v = 2aiy so that
the line: x = a, a > 0, y > 0 is mapped onto the upper portion
of parabola: 4a
2
(a
2
u) = v
2
, v > 0.
Hence, the semi-innite strip: {(x, y) : 0 x a, y 0} in the z
plane is mapped onto the semi-innite parabolic wedge: {(u, v) :
4a
2
(a
2
u) v
2
, v 0} in the w-plane.
9
The complex function w = z
2
maps a semi-innite strip onto a
semi-innite parabolic wedge.
10
Let Re
i
be the polar representation of w. Since w = z
2
, and
z = re
i
in polar form, we have Re
i
= r
2
e
2i
. Comparing, we
deduce that
= 2 and R = r
2
.
Consider a point on the curve whose polar form is r = 2(1+cos ) =
4cos
2

2
, we see that
r
2
= 16cos
4

2
.
Hence, we obtain
R = r
2
= 16cos
4

2
= 16cos
4

4
so that the polar form of the image curve in the w-plane is
R = 16cos
4

4
.
11
Limit of a complex function
Let w = f(z) be dened in the point set S and z
0
be a limit point
of S. The mathematical statement
lim
zz
0
f(z) = L, z S,
means that the value w = f(z) can be made arbitrarily close to L if
we choose z to be close enough, but not equal, to z
0
. The formal
denition of the limit of a function is stated as:
For any > 0, there exists > 0 (usually dependent on ) such that
|f(z) L| < if 0 < |z z
0
| < .
Remark
We require z
0
to be a limit point of S so that it would not occur
that in some neighborhood of z
0
inside which f(z) is not dened.
12
The circle |z z
0
| = in the z-plane is mapped onto the closed
curve in the w-plane. The annulus 0 < |z z
0
| < in the z-plane
is mapped onto the region enclosed by the curve in the w-plane.
The curve lies completely inside the annulus 0 < |w L| < .
13
The function f(z) needs not be dened at z
0
in order for the
function to have a limit at z
0
.
The limit L, if it exists, must be unique.
The value of L is independent of the direction along which z
approaches z
0
.
Example
Let S be the point set where
sinz
z
is dened. It can be shown that
lim
z0
sinz
z
= 1
though the function
sinz
z
is not dened at z = 0. Note that z = 0
is a limit point of S. Here, S is actually C\{0}.
14
Example
Prove that lim
zi
z
2
= 1.
We show that for given > 0 there is a positive number such that
|z
2
(1)| < whenever 0 < |z i| < .
Since
z
2
(1) = (z i)(z +i) = (z i)(z i +2i),
it follows from the the triangle inequality that
|z
2
(1)| = |z i| |z i +2i| |z i|(|z i| +2). (1)
To ensure that the left-hand side of (1) is less than , we merely have
to insist that z lies in a -neighborhood of i, where = min(1, /3).
If so, then
|z i|(|z i| +2)

3
(1 +2) = .
15
Some properties of the limit of a function
If L = +i, f(z) = u(x, y) +iv(x, y), z = x +iy and z
0
= x
0
+iy
0
,
then
|u(x, y) | |f(z) L| |u(x, y) | +|v(x, y) |,
|v(x, y) | |f(z) L| |u(x, y) | +|v(x, y) |.
It is obvious that lim
zz
0
f(z) = L is equivalent to the following pair of
limits
lim
(x,y)(x
0
,y
0
)
u(x, y) =
lim
(x,y)(x
0
,y
0
)
v(x, y) = .
Therefore, the study of the limiting behavior of f(z) is equivalent
to that of a pair of real functions u(x, y) and v(x, y).
16
Consequently, theorems concerning the limit and continuity of the
sum, dierence, product and quotient of complex functions can be
inferred from those for real functions.
Suppose that
lim
zz
0
f
1
(z) = L
1
and lim
zz
0
f
2
(z) = L
2
,
then
lim
zz
0
[f
1
(z) f
2
(z)] = L
1
L
2
,
lim
zz
0
f
1
(z)f
2
(z) = L
1
L
2
,
lim
zz
0
f
1
(z)
f
2
(z)
=
L
1
L
2
, L
2
= 0.
17
Limit at innity
The denition of limit holds even when z
0
or L is the point at innity.
We can simply replace the corresponding neighborhood of z
0
or L
by the neighborhood of . The statement
lim
z
f(z) = L
can be understood as
For any > 0, there exists () > 0 such that
|f(z) L| < whenever |z| >
1

.
As for convention, z refers to a point in the nite complex plane.
Hence, |z| >
1

is a deleted neighborhood of .
18
Theorem
If z
0
and w
0
are points in the z-plane and w-plane, respectively, then
(i) lim
zz
0
f(z) = if and only if lim
zz
0
1
f(z)
= 0;
(ii) lim
z
f(z) = w
0
if and only if lim
z0
f
_
1
z
_
= w
0
.
Proof
(i) lim
zz
0
f(z) = implies that for any > 0, there is a positive
number such that
|f(z)| >
1

whenever 0 < |z z
0
| < .
We may rewrite as

1
f(z)
0

< whenever 0 < |z z

0
| < .
Hence, lim
zz
0
1
f(z)
= 0.
19
(ii) lim
z
f(z) = w
0
implies that for any > 0, there exists > 0 such
that
|f(z) w
0
| < whenever |z| >
1

.
Replacing z by 1/z, we obtain

f
_
1
z
_
w
0

< whenever 0 < |z 0| < ,

hence lim
z0
f
_
1
z
_
= w
0
.
Example
lim
z1
iz +3
z +1
= since lim
z1
z +1
iz +3
= 0
and
lim
z
2z +i
z +1
= 2 since lim
z0
2
z
+i
1
z
+1
= lim
z0
2 +iz
1 +z
= 2.
20
Continuity of a complex function
Continuity of a complex function is dened in the same manner
as for a real function. The complex function f(z) is said to be
continuous at z
0
if
lim
zz
0
f(z) = f(z
0
).
The statement implicitly implies the existence of both lim
zz
0
f(z) and
f(z
0
). Alternatively, the statement can be understood as:
For any > 0, there exists () > 0 such that
|f(z) f(z
0
)| < whenever |z z
0
| < .
21
Example
Discuss the continuity of the following functions at z = 0.
(a) f(z) =
Im z
1+|z|
;
(b) f(z) =
_
_
_
0 z = 0
Re z
|z|
z = 0
.
Solution
(a) Let z = x +iy, then
Im z
1+|z|
=
y
1+

x
2
+y
2
. Now,
lim
z0
f(z) = lim
(x,y)(0,0)
y
1 +
_
x
2
+y
2
= 0 = f(0).
Therefore, f(z) is continuous at z = 0.
22
(b) Let z = x +iy, then
Re z
|z|
=
x
_
x
2
+y
2
.
Suppose z approaches 0 along the half straight line y = mx (x >
0), then
lim
z 0,
y = mx, x > 0
Re z
|z|
= lim
x0
+
x
_
x
2
+m
2
x
2
= lim
x0
+
x
x
_
1 +m
2
=
1
_
1 +m
2
.
Since the limit depends on m, lim
z0
f(z) does not exist. There-
fore, f(z) cannot be continuous at z = 0.
23
Properties of continuous functions
A complex function is said to be continuous in a region R if it is
continuous at every point in R.
Since the continuity of a complex function is dened using the con-
cept of limits, it can be shown that lim
zz
0
f(z) = f(z
0
) is equivalent
to
lim
(x,y)(x
0
,y
0
)
u(x, y) = u(x
0
, y
0
),
lim
(x,y)(x
0
,y
0
)
v(x, y) = v(x
0
, y
0
).
For example, consider f(z) = e
z
; its real and imaginary parts are,
respectively, u(x, y) = e
x
cos y and v(x, y) = e
x
siny. Since both
u(x, y) and v(x, y) are continuous at any point (x
0
, y
0
) in the nite x-
y plane, we conclude that e
z
is continuous at any point z
0
= x
0
+iy
0
in C.
24
Examples of continuous functions in C are polynomials, exponential
functions and trigonometric functions.
Sum, dierence, product and quotient of continuous functions re-
main to be continuous.
Uniform continuity
Suppose f(z) is continuous in a region R, then by hypothesis, at
each point z
0
inside R and for any > 0, we can nd > 0 such
that |f(z) f(z
0
)| < whenever |z z
0
| < .
Usually depends on and z
0
together. However, if we can nd a
single value of for each , independent of z
0
chosen in R, we say
that f(z) is uniformly continuous in the region R.
25
Example
Prove that f(z) = z
2
is uniformly continuous in the region |z| R,
R is nite.
Solution
We must show that given any > 0, we can nd > 0 such that
|z
2
z
2
0
| < when |z z
0
| < , where depends only on and not
on the particular point z
0
of the region.
If z and z
0
are any points in |z| R, then
|z
2
z
2
0
| = |z +z
0
| |z z
0
| {|z| +|z
0
|}|z z
0
| < 2R|z z
0
|.
Thus if |z z
0
| < , it follows that |z
2
z
2
0
| < 2R. Choosing =

2R
,
we see that |z
2
z
2
0
| < when |z z
0
| < , where depends only on
but not on z
0
. Hence, f(z) = z
2
is uniformly continuous in the
region.
26
Example
Prove that f(z) = 1/z is not uniformly continuous in the region
0 < |z| < 1.
Solution
Given > 0, let z
0
and z
0
+ be any two points of the region such
that
|z
0
+ z
0
| = || = .
Then
|f(z
0
+) f(z
0
)| =

1
z
0
+

1
z
0

=
||
|z
0
||z
0
+|
=

|z
0
||z
0
+|
can be made larger than any positive number by choosing z
0
su-
ciently close to 0. Hence, the function cannot be uniformly contin-
uous in the region.
27
Dierentiation of complex functions
Let the complex function f(z) be single-valued in a neighborhood
of a point z
0
. The derivative of f(z) at z
0
is dened by
df
dz
(z
0
) = lim
zz
0
f(z) f(z
0
)
z z
0
= lim
z0
f(z
0
+z) f(z
0
)
z
, z = z z
0
,
provided that the above limit exists. The value of the limit must be
independent of the path of z approaching z
0
.
Many formulas for the computation of derivatives of complex func-
tions are the same as those for the real counterparts.
28
The existence of the derivative of a complex function at a point
implies the continuity of the function at the same point. Suppose
f

(z
0
) exists, we can write
lim
zz
0
[f(z) f(z
0
)] = lim
zz
0
f(z) f(z
0
)
z z
0
lim
zz
0
(z z
0
) = 0,
so that
lim
zz
0
f(z) = f(z
0
).
This shows that f(z) is continuous at z
0
. However, continuity of
f(z) may not imply the dierentiability of f(z) at the same point.
It may occur that a complex function can be dierentiable at a given
point but not so in any neighborhood of that point.
29
Example
Show that the functions z and Re z are nowhere dierentiable, while
|z|
2
is dierentiable only at z = 0.
Solution
The derivative of z is given by
d
dz
z = lim
z0
z +z z
z
= lim
z0
z
z
= lim
z0
[e
2i Arg z
].
The value of the limit depends on the path approaching z. There-
fore, z is nowhere dierentiable.
30
Similarly,
d
dz
Re z =
d
dz
1
2
(z +z)
=
1
2
lim
z0
(z +z +z +z) (z +z)
z
=
1
2
lim
z0
z +z
z
=
1
2
+
1
2
lim
z0
z
z
.
Again, Re z is shown to be nowhere dierentiable.
Lastly, the derivative of |z|
2
is given by
d
dz
|z|
2
= lim
z0
|z +z|
2
|z|
2
z
= lim
z0
_
z +z
z
z
+z
_
.
The above limit exists only when z = 0, that is, |z|
2
is dierentiable
only at z = 0.
31
Complex velocity and acceleration
We treat z(t) as the position vector of a particle with time parameter
t. Then the velocity of the particle is
dz(t)
dt
= lim
t0
z(t +t) z(t)
t
.
32
Decomposition into various components
Write z(t) in its polar form
z(t) = r(t)e
i(t)
u = z = re
i
+r

(ie
i
)
u
r
= r = radial component of complex velocity
u

= r

= tangential component of complex velocity

33
Note that e
i
and ie
i
represent the unit vector along the radial and
tangential direction, respectively.
a =
du
dt
= (r r

2
)e
i
. .
radial component
of acc = a
r
+ (2 r

+r

)ie
i
. .
tangential component
of acc = a

a
r
= r r

2
, a

= 2 r

+r

Each term has its individual terminology

r = radial acceleration
r

2
= centripetal acceleration
r

= tangential acceleration
2 r

= Coriolis acceleration
34
Example
Consider the motion along the ellipse
z(t) = a cos t +ib sint
d
2
z(t)
dt
2
= a
2
cos t ib
2
sint =
2
z(t) =
2
|z(t)|e
i
.
The acceleration at any point is always directed toward the origin.
The radial component of acceleration =
2
|z|.
Note that x(t) and y(t) are in simple harmonic motions, since
d
2
x
dt
2
=
2
x and
d
2
y
dt
2
=
2
y.
35
The periodic trajectory is an ellipse whose equation is
x
2
a
2
+
y
2
b
2
= 1.
In particular, when a = b, the trajectory becomes a circle of radius
a (= b). In this case, r = a so that r = 0. Hence, the radial
component of acceleration consists of only one term, namely, the
centripetal term whose value is a
2
. Here, z(t) = ae
it
so that
(t) = t. This gives

= , which is the angular speed.
36
Cauchy-Riemann relations
What are the necessary and sucient conditions for the existence
of the derivative of a complex function?
The necessary conditions are given by the Cauchy-Riemann equa-
tions.
The sucient conditions require, in addition to the Cauchy-
Riemann relations, the continuity of all rst order partial deriva-
tives of u and v.
Let f(z) be single-valued in a neighborhood of the point z
0
= x
0
+
iy
0
, and it is dierentiable at z
0
, that is, the limit
f

(z
0
) = lim
z0
f(z
0
+z) f(z
0
)
z
exists. This limit is independent of the direction along which z
approaches 0.
37
(i) First, we take z 0 in the direction parallel to the x-axis, that
is, z = x. We then have
f(z
0
+z) f(z
0
) = u(x
0
+x, y
0
) +iv(x
0
+x, y
0
)
u(x
0
, y
0
) iv(x
0
, y
0
),
so that
f

(z
0
) = lim
x0
u(x
0
+x, y
0
) u(x
0
, y
0
)
x
+ i lim
x0
v(x
0
+x, y
0
) v(x
0
, y
0
)
x
=
u
x
(x
0
, y
0
) +i
v
x
(x
0
, y
0
).
38
(ii) Next, we let z 0 in the direction parallel to the y-axis, that
is z = iy. Now, we have
f(z
0
+z) f(z
0
) = u(x
0
, y
0
+y) +iv(x
0
, y
0
+y)
u(x
0
, y
0
) iv(x
0
, y
0
),
so that
f

(z
0
) = lim
y0
u(x
0
, y
0
+y) u(x
0
, y
0
)
iy
+ i lim
y0
v(x
0
, y
0
+y) v(x
0
, y
0
)
iy
=
1
i
u
y
(x
0
, y
0
) +
v
y
(x
0
, y
0
).
39
Combining the above two equations, we obtain
f

=
u
x
+i
v
x
=
v
y
i
u
y
.
Equating the respective real and imaginary parts gives
u
x
=
v
y
and
v
x
=
u
y
.
The results are called the Cauchy-Riemann relations. They are the
necessary conditions for the existence of the derivative of a complex
function.
40
Theorem
Given f(z) = u(x, y) +iv(x, y), z = x +iy, and assume that
1. Cauchy-Riemann relations hold at a point z
0
= x
0
+iy
0
,
2. u
x
, u
y
, v
x
, v
y
are all continuous at (x
0
, y
0
).
Then, f

(z
0
) exists and it is given by
f

(z
0
) = u
x
(x
0
, y
0
) +iv
x
(x
0
, y
0
) = v
y
(x
0
, y
0
) iu
y
(x
0
, y
0
).
41
Proof
Since u(x, y) and v(x, y) have continuous rst order partials at (x
0
, y
0
)
and satisfy the Cauchy-Riemann relations at the same point, we
have
u(x, y) u(x
0
, y
0
) = u
x
(x
0
, y
0
)(x x
0
) v
x
(x
0
, y
0
)(y y
0
) +
1
(|z|)
(i)
v(x, y) v(x
0
, y
0
) = v
x
(x
0
, y
0
)(x x
0
) +u
x
(x
0
, y
0
)(y y
0
) +
2
(|z|),
(ii)
where z = (x + iy) (x
0
+ iy
0
),
1
and
2
represent higher order
terms that satisfy
lim
|z|0

1
(|z|)
|z|
= lim
|z|0

2
(|z|)
|z|
= 0, |z| =
_
(x x
0
)
2
+(y y
0
)
2
.
(iii)
42
Adding (i) and i times (ii) together, we obtain
f(z)f(z
0
) = [u
x
(x
0
, y
0
)+iv
x
(x
0
, y
0
)](z z
0
)+
1
(|z|)+i
2
(|z|),
and subsequently,
f(z) f(z
0
)
z z
0
[u
x
(x
0
, y
0
) +iv
x
(x
0
, y
0
)] =

1
(|z|) +i
2
(|z|)
z z
0
.
Note that

1
(|z|) +i
2
(|z|)
z z
0

1
(|z|)
|z|
+

2
(|z|)
|z|
;
and as z 0, also |z| 0.
43
It then follows from the results in (iii) that
lim
z0

1
(|z|) +i
2
(|z|)
z z
0
= 0.
Hence, we obtain
f

(z
0
)
= lim
zz
0
f(z) f(z
0
)
z z
0
= u
x
(x
0
, y
0
) +iv
x
(x
0
, y
0
) = v
y
(x
0
, y
0
) iu
y
(x
0
, y
0
).
44
Example
Discuss the dierentiability of the function
f(z) = f(x +iy) =
_
|xy|
at z = 0.
Solution
Write f(x +iy) = u(x, y) +iv(x, y) so that
u(x, y) =
_
|xy| and v(x, y) = 0.
Since u(x, 0) and u(0, y) are identically equal to zero, and so
u
x
(0, 0) = u
y
(0, 0) = 0.
Also, obviously
v
x
(0, 0) = v
y
(0, 0) = 0,
the Cauchy-Riemann relations are satised at the point (0, 0).
45
However, suppose z approaches the origin along the ray: x = t, y =
t, assuming that and cannot be zero simultaneously.
For z = t +it, we then have
f(z) f(0)
z 0
=
f(z)
z
=
_
||
+i
.
The limit of the above quantity as z 0 depends on the choices of
and , and so the value is non-unique.
Therefore, f(z) is not dierentiable at z = 0 though the Cauchy-
Riemann relations are satised at z = 0. Note that the rst order
partial derivatives of u(x, y) is NOT continuous at (0, 0).
46
Example
Show that the function f(z) = zRe (z) is nowhere dierentiable
except at the origin; hence nd f

(0). Is f(z) continuous at z = 0?

Explain why or why not?
Solution
When z
0
= 0, we have
f

(0) = lim
z0
f
z
= lim
z0
zRe z
z
= 0.
47
When z
0
= 0, we write z
0
= x
0
+iy
0
. Let z = x +iy, we then have
lim
zz
0
f
z
= lim
zz
0
zRe z z
0
Re z
0
z z
0
= lim
zz
0
_
(z z
0
)Re z
z z
0
+
z
0
(Re z Re z
0
)
z z
0
_
= lim
zz
0
_
x +z
0
x x
0
z z
0
_
.
Suppose we approach z
0
along the direction parallel to the y-axis
lim
z z
0
x = x
0
f
z
= x
0
. (i)
On the other hand, if we approach z
0
along the direction parallel to
the x-axis, we obtain
48
lim
z z
0
z = x +iy
0
f
z
= lim
xx
0
(x +iy
0
)x (x
0
+iy
0
)x
0
x x
0
(ii)
= lim
xx
0
x
2
x
2
0
+iy
0
(x x
0
)
x x
0
= lim
xx
0
x +x
0
+iy
0
= 2x
0
+iy
0
.
Since the two limits in Eqs. (i) and (ii) are not equal, the limit
lim
zz
0
f
z
does not exist. Therefore, the function is not dierentiable
at any z
0
= 0.
Recall that dierentiability of a complex function at a point implies
continuity of the function at the same point. Since f

(0) exists, so
f(z) is continuous at z = 0.
49
Analyticity
A function f(z) is said to be analytic at some point z
0
if it is
dierentiable at every point of a certain neighborhood of z
0
. In
other words, f(z) is analytic if and only if there exists a neighborhood
N(z
0
; ), > 0, such that f

(z) exists for all z N(z

0
; ).
Since z
0
N(z
0
; ), analyticity at z
0
implies dierentiability at z
0
.
The converse statement is not true, that is, dierentiability of f(z)
at z
0
does not guarantee the analyticity of f(z) at z
0
.
For example, the function f(z) = |z|
2
is nowhere dierentiable ex-
cept at the origin, hence f(z) = |z|
2
is not analytic at z = 0.
50
Entire functions
If a function is analytic in the entire complex plane, then the function
is called an entire function.
To show that f(z) is analytic in an open region or domain D, we
may either show
(i) f

(z) exists for all z in D, or

(ii) the real and imaginary parts of f(z) have continuous rst or-
der partials and their derivatives satisfy the Cauchy-Riemann
relations at every point inside D.
Remark
Since every point in an open region is an interior point, so if f

(z)
exists for all z in D, then f is analytic everywhere inside D.
51
Example
Find the domains in which the function
f(z) = |x
2
y
2
| +2i|xy|, z = x +iy,
is analytic.
Domain of analyticity (shown in shaded areas) of f(z) = |x
2
y
2
| +
2i|xy|.
52
The functional values of f(z) depend on the signs of x
2
y
2
and
xy:
(i) x
2
y
2
changes sign when (x, y) crosses the lines x = y or x = y;
(ii) xy changes sign when (x, y) crosses the x-axis or y-axis.
When x
2
y
2
> 0 and xy > 0, f(z) = z
2
. When x
2
y
2
< 0 and
xy < 0, f(z) = z
2
. Both functions are known to be analytic.
When x
2
y
2
> 0 and xy < 0, f(z) = x
2
y
2
2ixy. Inside these
domains, the Cauchy-Riemann relations are not satised, and so
f(z) fails to be analytic. Inside the domains dened by x
2
y
2
< 0
and xy > 0, the function becomes f(z) = (x
2
y
2
)+2ixy, which
is non-analytic.
The function is analytic within the following domains:
0 < Arg z <

4
,

2
< Arg z <
3
4
< Arg z <
3
4
and

2
< Arg z <

4
.
53
Example
Show that there is no entire function f such that f

(z) = xy
2
for all
z C.
Solution
Suppose that f is entire with f

(z) = xy
2
= u
x
+ iv
x
at all points.
This gives u
x
= xy
2
and v
x
= 0. Then u(x, y) =
1
2
x
2
y
2
+ F(y) and
v(x, y) = G(y), where F and G are arbitrary functions of y. On the
other hand, given that f is entire, the Cauchy-Riemann equations
are satised at all points, so that v
y
= xy
2
and u
y
= 0. This leads
to a contradiction.
54
Example
Suppose f(z) and g(z) are analytic inside the domain D. Show
that f(z) is constant inside D if |f(z)| is constant inside D. Is g(z)
constant inside D if Re (g(z)) is constant inside D? Explain why or
why not.
Solution
Write f(z) = u(x, y) +iv(x, y), z = x +iy. Consider
|f(z)|
2
= u
2
+v
2
= constant
so that
uu
x
+vv
x
= 0 and uu
y
+vv
y
= 0
_
u
x
v
x
u
y
v
y
__
u
v
_
=
_
0
0
_
.
Note that |f(z)| = 0 f(z) = 0. Now, consider the case where
f(z) = u(x, y) +iv(x, y) = 0.
55
Since
_
u
v
_
=
_
0
0
_
, so we must have
det
_
u
x
v
x
u
y
v
y
_
= u
x
v
y
u
y
v
x
= 0.
From the Cauchy-Riemann relations, u
x
= v
y
, v
x
= u
y
; we obtain
u
2
x
+u
2
y
= 0 so u
x
= u
y
= 0. that is, u = constant.
Similarly, v
x
= v
y
= 0, so v = constant. Hence, f(z) is constant in
D.
Write g(z) = (x, y) + i(x, y), z = x + iy. When = constant,

x
=
y
= 0. From the Cauchy-Riemann relations,
x
=
y
= 0 so
that = constant. Hence, g(z) in constant in D.
56
Example
Suppose f(z) and f(z) are analytic in a domain D. Show that f(z)
is constant in D.
Solution
Write f = u +iv and f = u iv.
Since f is analytic, we have u
x
= v
y
and u
y
= v
x
. Also, since f is
analytic, we have u
x
= v
y
and u
y
= v
x
. Combining the results, we
obtain
v
x
= v
y
= 0 and u
x
= u
y
= 0.
Hence, f is a constant.
57
Harmonic functions
A real-valued function (x, y) of two real variables x and y is said to
be harmonic in a given domain D in the x-y plane if has continuous
partial derivatives up to the second order in D and satises the
Laplace equation

xx
(x, y) +
yy
(x, y) = 0.
Analytic functions are closely related to harmonic functions. Sup-
pose f(z) = u(x, y) + iv(x, y) is an analytic function in D, we will
show that both the component functions u(x, y) and v(x, y) are har-
monic in D.
58
We state without proof the following result:
If a complex function is analytic at a point, then its real and imag-
inary parts have continuous partial derivatives of all orders at that
point.
Suppose f(z) is analytic in D, then
u
x
= v
y
and v
x
= u
y
.
Dierentiating both sides of the equations with respect to x, we
obtain
u
xx
= v
yx
and v
xx
= u
yx
in D.
Similarly,
u
xy
= v
yy
and v
xy
= u
yy
.
59
Since the above partial derivatives are all continuous, it is guaran-
teed that
u
xy
= u
yx
and v
xy
= v
yx
.
Combining the results,
v
yy
= u
xy
= u
yx
= v
xx
and so v
xx
+v
yy
= 0,
and
u
yy
= v
xy
= v
yx
= u
xx
and so u
xx
+u
yy
= 0.
Therefore, both u(y, y) and v(x, y) are harmonic functions.
60
Take any two harmonic functions, they normally do not form a
complex analytic function.
For example, = e
x
cos y, = 2xy
f = +i = e
x
cos y +2ixy is NOT analytic.
Suppose is changed to e
x
siny, or is changed to x
2
y
2
, then
f
1
= e
x
cos y +ie
x
siny = e
z
or
f
2
= x
2
y
2
+2ixy = z
2
.
Trick: As necessary conditions, we require
x
=
y
and
y
=

x
.
61
Harmonic conjugate
Given two harmonic functions (x, y) and (x, y) and if they satisfy
the Cauchy-Riemann relations throughout a domain D, with

x
=
y
and
y
=
x
.
We call a harmonic conjugate of in D.
Note that harmonic conjugacy is not a symmetric relation because
of the minus sign in the second Cauchy-Riemann relation. While
is a harmonic conjugate of , is a harmonic conjugate of .
For example, e
x
siny is a harmonic conjugate of e
x
cos y while e
x
cos y
is a harmonic conjugate of e
x
siny.
62
Theorem
A complex function f(z) = u(x, y) + iv(x, y), z = x + iy, is analytic
in a domain D if and only if v is a harmonic conjugate to u in D.
Proof
Given that f = u+iv is analytic, then u and v are harmonic and
Cauchy-Riemann relations are satised. Hence, v is a harmonic
conjugate of u.
Given that v is a harmonic conjugate of u in D, we have the
satisfaction of the Cauchy-Riemann relations and the continuity
of the rst order partials of u and v in D. Hence, f = u + iv is
dierentiable for all points in D. Since D is an open set, every
point in D is an interior point, so f is analytic in D..
63
Exact dierentials
The dierential M(x, y) dx+N(x, y) dy is an exact dierential if and
only if M and N observe
M
y
=
N
x
.
Under such scenario, then
M(x, y) dx +N(x, y) dy = dF
for some F, that is,
F
x
= M and
F
y
= N.
The line integral of the dierential along any path joining (x
0
, y
0
)
and (x
1
, y
1
) is given by
_
(x
1
,y
1
)
(x
0
,y
0
)
M dx +N dy = F(x
1
, y
1
) F(x
0
, y
0
).
64
The integral value is path independent provided that there is no
singular points enclosed inside the closed curve represented by the
two paths of integration.
y
x
>
>
( , ) x y
1 1
( , ) x y
0 0
65
Given that (x, y) is harmonic in a simply connected domain D,
it can be shown that it is always possible to obtain its harmonic
conjugate (x, y) by integration. Starting with the dierential form:
d =
x
dx +
y
dy,
and using the Cauchy-Riemann relations, we have
d =
y
dx +
x
dy.
To obtain , we integrate along some path joining a xed point
(x
0
, y
0
) to (x, y), that is,
(x, y) =
_

y
dx +
x
dy.
The above integral is an exact dierential provided that
(
y
)
y
+(
x
)
x
= 0,
that is, is harmonic.
66
To ease the computation, we choose the path which consists of
horizontal and vertical line segments as shown in the Figure.
(x, y) =
_
x
x
0

y
(x, y
0
) dx +
_
y
y
0

x
(x, y) dy.
>
<
y
x
(x
0
, y
0
)
(x, y
0
)
(x, y )
The choice of a dierent starting point (x
0
, y
0
) of the integration
path simply leads to a dierent additive constant in (x, y). Recall
that (x, y) is unique up to an additive constant.
67
Example
Find a harmonic conjugate of the harmonic function
u(x, y) = e
x
cos y +xy.
Solution
1. Take (x
0
, y
0
) = (0, 0), u
y
(x, 0) = x and u
x
(x, y) = e
x
cos y +y.
v(x, y) =
_
x
0
xdx +
_
y
0
(e
x
cos y +y) dy
=
x
2
2
e
x
siny +
y
2
2
.
68
2. From the rst Cauchy-Riemann relation, we have
v
y
=
u
x
= e
x
cos y +y.
Integrating with respect to y, we obtain
v(x, y) = e
x
siny +
y
2
2
+(x),
where (x) is an arbitrary function arising from integration.
Using the second Cauchy-Riemann relation, we have
v
x
= e
x
siny +

(x) =
u
y
= e
x
siny x.
69
Comparing like terms, we obtain

(x) = x,
and subsequently,
(x) =
x
2
2
+C, where C is an arbitrary constant.
Hence, a harmonic conjugate is found to be (taking C to be zero
for convenience)
v(x, y) = e
x
siny +
y
2
x
2
2
.
The corresponding analytic function, f = u +iv, is seen to be
f(z) = e
z

iz
2
2
, z = x +iy,
which is an entire function.
70
3. It is readily seen that
e
x
cos y = Re e
z
and xy =
1
2
Im z
2
.
A harmonic conjugate of Re e
z
is Im e
z
, while that of
1
2
Im z
2
is
1
2
Re z
2
. Therefore, a harmonic conjugate of u(x, y) can be
taken to be
v(x, y) = Im e
z

1
2
Re z
2
= e
x
siny +
y
2
x
2
2
.
71
Example
Show that f

(z) =
u
x
(z, 0) i
u
y
(z, 0). Use the result to nd a
harmonic conjugate of
u(x, y) = e
x
(xsiny y cos y).
Solution
Observe that f

(z) =
u
x
(x, y) i
u
y
(x, y). Putting y = 0, we obtain
f

(x) =
u
x
(x, 0) i
u
y
(x, 0).
Replacing x by z, we obtain f

(z) =
u
x
(z, 0) i
u
y
(z, 0).
72
Now, for u(x, y) = e
x
(xsiny y cos y), we have
u
x
= e
x
siny xe
x
siny +ye
x
cos y
u
y
= xe
x
cos y +ye
x
siny e
x
cos y.
We then have
f

(z) =
u
x
(z, 0) i
u
y
(z, 0) = i(ze
z
e
z
).
Integrating with respect to z, we obtain f(z) = ize
z
.
The imaginary part of f(z) = v(x, y) = e
x
(y siny +xcos y).
73
Theorem
If is a harmonic conjugate of , then the two families of curves
(x, y) = and (x, y) =
are mutually orthogonal to each other.
Proof
Consider a particular member from the rst family
(x, y) =
1
,
the slope of the tangent to the curve at (x, y) is given by
dy
dx
where

x
+

y
dy
dx
= 0,
giving
dy
dx
=

x
_

y
.
74
Similarly, the slope of the tangent to a member from the second
family at (x, y) is given by
dy
dx
=

x
_

y
.
The product of the slopes of the two tangents to the two curves at
the same point is found to be
_

x
_

y
__

x
_

y
_
= 1,
by virtue of the Cauchy-Riemann relations:

x
=

y
and

y
=

x
.
Hence, the two families of curves are mutually orthogonal to each
other.
75
Steady state temperature distribution
In a two-dimensional steady state temperature eld, the tempera-
ture function T(x, y) is harmonic:

2
T
x
2
+

2
T
y
2
= 0.
From the empirical law of heat conduction
Q = heat ux across a surface = K
T
n
, K > 0,
where K is called the thermal conductivity of the material and
T
n
denotes the normal derivative of T with respect to the surface.
Steady state temperature distribution prevails if there is no heat
source or sink inside the body and there is no net heat ux across
the bounding surface.
76
x
y
y x
y
T
K

2
,
y y
x
x
x
T
K

,
2
x
y
y y
x
x
x
T
K

,
2
x
y
y x
y
T
K

2
,
An innitesimal control volume of widths x and y is contained
inside a two-dimensional body. The heat uxes across the four sides
of the rectangular control volume are shown.
77
Within a unit time interval, the amount of heat owing across
the left vertical side into the rectangular control volume is
K
T
x
_
x
x
2
, y
_
y.
Likewise, the amount of heat owing across the right vertical
side out of the control volume is
K
T
x
_
x +
x
2
, y
_
y.
78
The net accumulation of heat per unit time per unit volume
inside the control volume is
K
_
_
T
x
_
x +
x
2
, y
_

T
x
_
x
x
2
, y
_
x
+
T
y
_
x, y +
y
2
_

T
y
_
x, y
y
2
_
y
_

_ .
Taking the limits x 0 and y 0, we then obtain

2
T
x
2
+

2
T
y
2
= 0.
Hence, the steady state temperature function is a harmonic
function.
79
Example
Supposing the isothermal curves of a steady state temperature eld
are given by the family of parabolas
y
2
=
2
+2x, is real positive,
in the complex plane, nd the general solution of the temperature
function T(x, y). Also, nd the family of ux lines of the temperature
eld.
Solution
First, we solve for the parameter in the equation of the isothermal
curves. This gives
= x +
_
x
2
+y
2
,
where the positive sign is chosen since > 0.
80
A naive guess may suggest that the temperature function T(x, y) is
given by
T(x, y) = x +
_
x
2
+y
2
.
However, since T(x, y) has to be harmonic, the above function can-
not be a feasible solution. We set
T(x, y) = f(t)
where t =
_
x
2
+y
2
x, and f is some function to be determined
such that T(x, y) is harmonic. To solve for f(t), we rst compute

2
T
x
2
= f

(t)
_
t
x
_
2
+f

(t)

2
t
x
2
= f

(t)
_
_
_
x
_
x
2
+y
2
1
_
_
_
2
+f

(t)
y
2
(x
2
+y
2
)
3/2
,

2
T
y
2
= f

(t)
y
2
x
2
+y
2
+f

(t)
x
2
(x
2
+y
2
)
3/2
.
81
Since T(x, y) satises the Laplace equation
2
_
_
_1
x
_
x
2
+y
2
_
_
_f

(t) +
1
_
x
2
+y
2
f

(t) = 0 or
f

(t)
f

(t)
=
1
2t
.
Integrating once gives
lnf

(t) =
1
2
lnt +C or f

(t) =
C

t
.
Integrating twice gives
f(t) = C
1

t +C
2
,
where C
1
and C
2
are arbitrary constants. The temperature function
is
T(x, y) = f(t) = C
1
_
_
x
2
+y
2
x +C
2
.
When expressed in polar coordinates
T(r, ) = C
1
_
r(1 cos ) +C
2
= C
1

2r sin

2
+C
2
.
82
Since T(r, ) can be expressed as

2C
1
Im z
1/2
+ C
2
, the harmonic
conjugate of T(r, ) is easily seen to be
F(r, ) =

2C
1
Re z
1/2
+C
3
= C
1

2r cos

2
+C
3
,
where C
3
is another arbitrary constant. Note that

2r cos

2
=
_
r +r cos =
_
_
x
2
+y
2
+x
so that
F(x, y) = C
1
_
_
x
2
+y
2
+x +C
3
.
The family of curves dened by
x +
_
x
2
+y
2
= or y
2
=
2
2x, > 0,
are orthogonal to the isothermal curves y
2
=
2
+2x, > 0.
83
Physically, the direction of heat ux is normal to the isothermal
lines. Therefore, the family of curves orthogonal to the isothermal
lines are called the ux lines. These ux lines indicate the ow
directions of heat in the steady state temperature eld.
The ux function F(r, ) is a harmonic conjugate of the temperature
function. The families of curves: T(r, ) = and F(r, ) = , and
being constant, are mutually orthogonal to each other.
84