30 Chapter 2 Analytic Functions

Solutions to Exercises 2.1

1. We notice that the set {z : |z| 1} is the closed disc of the radius one centered at the origin.
So, the interior points are {z : |z| < 1}, or the open unit disc centered at the origin. And, the
boundary is the set {z : |z| = 1}, or the unit circle centered at the origin.
5. We see that the set {z : Re z > 0} is a right half-plane. Also, since the inequality Re z > 0 is
strict this set is open. Really, if some z
0
is in from the set {z : Re z > 0}. Then for any number
0 < r < Re z and any complex number z such that |z z
0
| < r we have Re z > Re z
0
r > 0.
Therefore z is also from the same set {z : Re z > 0}. This justies that the set {z : Re z > 0} is
open.
Also, it is clear from the picture that any two points from the set {z : Re z > 0} can be
connected even by a straight line segment, the simplest polygonal line. So, the set {z : Re z > 0}
is also connected. Since it is simultaneously open and connected the set is also a region.
9. We see that the set A = {z : z = 0, | Arg z| <

4
} {0} is an innite sector with the vertex
z = 0 included but the rays {z : z = 0, Arg z =

4
} are not included. Since any open disc with
the center at the origin z = 0 is not contained in the set A it follows this set is not open. On the
other hand since the rays {z : z = 0, Arg z =

4
} are not in this set A it follows the set A is also
not closed. It is clear from the picture that any two points from the set A can be connected by a
straight line segment. So, the set A is connected. Since it is not open the set A is not a region.
13. One of simple examples is the following. Let us take the sets A = {0} and B = {1}. Each of
these sets is a point. Any single point is a connected set but it is obviously that AB = {0} {1}
consists of only two points and they cannot be connected by a polygonal line inside A B.
17. To prove this statement we need to show that if C \ S is closed then any point z of the set S
is an interior point of S. Let us assume this is not true and there it is a point z
0
of S which is not
an interior point. This means that every neighborhood of z
0
contains at least one point not from S.
Thus, every neighborhood of z
0
contains at least one point from the complement of S, the set C\ S.
And z
0
is not from C\ S. By denition of boundary of a set it follows that z
0
is from the boundary
of the set C\S. But we know that the set C\S is closed and thus it contains all its boundary. And,
thus z
0
must be in C \ S which is impossible since z
0
is from the set S. This gives a contradiction
to our assumption. Hence the assumption was wrong and the set S is open.
21. To show that the set A B is a region we need to show that is open and connected.
Let z
0
is a point of A B. Thus, z
0
is either from A or from B. If z
0
is from A then since A is
open it follows some neighborhood of z
0
is contained in A and thus also in A B. Similar if z
0
is
from B then some neighborhood of z
0
is contained in B and thus in AB. It implies that z
0
is an
interior point of A B. Hence the set A B is open.
To show that A B is connected we need to show that any two points z
1
and z
2
of A B can
be connected by a polygonal line. Let the point z
3
is some point from the non-empty intersection
of A and B. Then z
3
and z
1
are from the same set A or B. Therefore they can be connected by a
polygonal line l
1
. Also z
3
and z
2
are from the same set A or B. And they can be connected by a
polygonal line l
2
. Now it is obvious that z
1
and z
2
are connected by the polygonal line l
1
l
2
passing
through z
3
. Hence A B is connected.
Since A B is simultaneously open and connected it is a region.
Section 2.2 Limits and Continuity 31
Solutions to Exercises 2.2
1. Using the properties from the Theorem 1 and the fact that lim
zi
z = i we have
lim
zi
3z
2
+ 2z 1 = 3 lim
zi
z
2
+ 2 lim
zi
z 1
= 3
_
lim
zi
z
_
2
+ 2 lim
zi
z 1
= 3i
2
+ 2i 1 = 3 + 2i 1 = 4 + 2i.
5. First we add to fractions under the sign of limit. We use the identity z
2
+1 = (z i)(z +i). We
have
1
z i

1
z
2
+ 1
=
z + i
(z i)(z + i)

1
(z i)(z + i)
=
z + i 1
(z i)(z + i)
.
We again use the properties from the Theorem 1 and the fact that lim
zi
z = i to get
lim
zi
1
z i

1
z
2
+ 1
= lim
zi
z + i 1
(z i)(z + i)
=
lim
zi
(z + i 1)
lim
zi
z i lim
zi
(z + i)
=
2i 1
(lim
zi
z i)2i
[since lim
zc
f(z) = 0 lim
zc
1
f(z)
= ]
=
2i 1
2i
= .
9. We evaluate using Theorem 1 and the fact that Arg z is always real to get
lim
z3
( Arg z)
2
= lim
z3
| Arg z|
2
=
_
lim
z3
| Arg z|
_
2
.
We know from Example 7 and Theorem 5 that in general Arg z is discontinuous on the ray (, 0].
But it turns out that the function f(z) = | Arg z| is continuous on the open ray (, 0). Really
if z
0
is from the open ray (, 0) and z approaches to z
0
from the second quadrant then Arg z
approaches to . Therefore | Arg z| also approaches . Now if z approaches to z
0
from the third
quadrant then Arg z approaches to . But | Arg z| approaches again. So, in either case | Arg z|
approaches . Therefore we conclude
lim
zz0
| Arg z| = .
If we take z
0
= 3 we have
lim
zz0
( Arg z)
2
=
_
lim
z3
| Arg z|
_
2
=
2
.
32 Chapter 2 Analytic Functions
13. Since z approaches (thus z = 0) we can divide by z both the numerator and the denominator.
We have
lim
z
z + 1
3i z + 2
= lim
z
1 +
1
z
3i +
2
z
=
1 + lim
z
1
z
3i + lim
z
2
z
=
1
3i
=
i
3i
2
=
i
3
.
17. We have
lim
z1
1
(z 1)
2
=
_
lim
z1
1
z 1
_
2
.
Now we notice that if f(z) approaches 0 if z approaches 1 then g(z) =
1
f(z)
approaches . Also if
lim
z1
g(z) = then lim
z1
g(z)
2
= . Using these properties we compute
_
lim
z1
1
z 1
_
2
= .
Since multiplying by a non-zero constant does not change approaching to we have
lim
z1
1
(z 1)
2
= .
21. We know that a real exponential function f(x) = e
x
has the properties
lim
x+
f(x) = 0,
and
lim
x
f(x) = +.
So, we see that even for real z the function f(z) has dierent limits in the positive and negative
directions. Hence, it also cannot have a unique limit when z approaches innity in the complex
plane C.
25. We can use the polar form of complex numbers to nd that for z = re
i
for r > 0 and real
z
|z|
=
re
i
|re
i
|
=
re
i
r
= e
i
.
And, we see that the value of
z
|z|
depends only on the argument of z. Therefore, if z approaches
to 0 along the positive x-axis then
z
|z|
is equal to and thus approaches e
i0
= 1. And, similar if z
approaches to 0 along the negative x-axis then
z
|z|
is equal to e
i
= 1. Since 1 = 1 the limit
lim
z0
z
|z|
is undened.
29. We need to show that
(1) lim
z
f(z) = L lim
z0
f
_
1
z
_
= L;
Suppose the left statement is true. This means that for any > 0 there is an R > 0 such that
|z| > R it follows |f(z) L| < . Now let > 0 be any. Let us denote =
1
R
where R is chosen as
above. Then it follows that for any |z| < we have

1
z

> R. Thus it follows |f(

1
z
) L| < . Hence
lim
z0
f
_
1
z
_
= L.
Section 2.2 Limits and Continuity 33
Now we suppose that the right statement in (??) is true. Again we pick any > 0. We can
choose > 0 such that if |z| < then |f(
1
z
) L| < . Let us denote R =
1

. If |z| > R then

|
1
z
| <
1
R
= . So,
|f(z) L| = |f(
1
1
z
L| < .
Hence lim
z
f (z) = L.
33. If z = 1 3i then by Theorem 4 the function h(z) =
zi
z+1+3i
is continuous in z. Discontinuity
at z
0
= 1 3i is not removable because z
0
i = 1 3i i = 1 4i and z
0
+ 1 + 3i = 0. And
we have
lim
z13i
z i
z + 1 + 3i
= (1 4i) lim
z13i
1
z + 1 + 3i
= (1 4i)= .
37. We use the denition of sin z using the exponential function. We have
sinz =
e
iz
e
iz
2i
.
From the Example 8 we know that e
z
is continuous everywhere in C. Since e
iz
= f(g(z)) for
g(z) = iz and f(z) = e
iz
and both functions are continuous it follows from Theorem 4 that e
iz
is
also continuous. Similar we show that the function e
iz
is continuous. Then by Theorem 4
sinz =
e
iz
e
iz
2i
=
1
2i
e
iz

1
2i
e
iz
is continuous in C.
41. (a) First, assume that f is continuous and A is open. Let z
0
be from f
1
[A]. This means
that f(z
0
) is in A. Since A is open there is > 0 such that the -neighborhood is a subset of A,
or B

(f(z
0
)) A. Since f is continuous at z
0
there is a > 0 such that if |z z
0
| < then
|f(z) f(z
0
)| < . Thus f(z) is in A. And therefore z which is in f
1
[f(z)] is in f
1
[A]. Therefore
the whole -neighborhood of z
0
is in f
1
[A]. And it follows that f
1
[A] is open.
Second, assume that f
1
[A] is open whenever A is open. Let z
0
be any complex number and
> 0 be any too. Then by assumption the -neighborhood B

(f(z
0
)) of f(z
0
) is open too. Therefore
f
1
[B

(f(z
0
))] is open. In particular, there is > 0 such that B

(z
0
) f
1
[B

(f(z
0
))]. It follows
that for any z such that |z z
0
| < we have |f(z) f(z
0
)| < . Hence f is continuous.
(b) First, assume that f
1
[A] is closed whenever A is closed. To show that f is continuous
by part(a) it is enough to show that f
1
[B] is open whenever B is open. So, let open B be given.
Since the complement of an open set is closed (problem 17 Section 2.1) we have A = C\B is closed.
Therefore, by assumption f
1
[A] is closed too. Now we use the identity
f
1
[B] = C\ f
1
[A]
which is true for any set B and A = C \ B. Really, if z is in f
1
[B]. Then f(z) is in B = C \ A.
Therefore f(z) is in the complement of f
1
[A]. On the other hand if z is in C \ f
1
[A] then f(z)
cannot be in A = C \ B. And, thus f(z) is in B. So, f
1
(z) is in f
1
[B]. Since f
1
[A] is closed
it follows from (2) and Problem 17 Section 2.1 that the complement set f
1
[B] is open. Thus f is
continuous.
Second, assume that f is continuous and B is closed. Then A = C \ B is open. Therefore by
part (a) it follows that f
1
[A] is open. By (2) it follows that f
1
[B] = C\ f
1
[A]. f
1
[A] is open.
Then by Problem 17 Section 2.1 it follows that its complement, f
1
[B] is closed.
34 Chapter 2 Analytic Functions
Solutions to Exercises 2.3
1. If we take g(z) = 3z
2
+ 2z and h(z) = z 1 then we have
f(z) = 3(z 1)
2
+ 2(z 1) = g(h(z)).
By the chain rule and the formula that (z
n
)

= nz
n1
for a positive integer n we have
f

(z) = g

(h(z))h

(z) = (6(z 1) + 2)1 = 6(z 1) + 2 = 6z 4.

5. By the quotient rule and the formula for the derivative of a polynomial it follows
_
1
z
3
+ 1
_

=
(1)

(z
3
+ 1) 1(z
3
+ 1)

(z
3
+ 1)
2
=
0 3z
2
(z
3
+ 1)
2
=
3z
2
(z
3
+ 1)
2
.
9. We have f(z) = z
2/3
= (z
1/3
)
2
. So, if we take g(z) = z
1/3
and h(z) = z
2
then f(z) = h(g(z)).
So, by the chain rule and the formulas for the derivatives of polynomial and n-th root we get
f

(z) = h

(g(z))g

(z) = 2g(z)
1
3
z
(13)/3
= 2z
1/3
1
3
z
2/3
=
2
3
z
1/3
.
13. If we take f(z) = z
100
and z
0
= 1 then the limit is the denition of the derivative of f at z
0
since f(z
0
) = 1
100
= 1. By the formula for the derivative of z
n
for a positive integer n we have
lim
z1
z
100
1
z 1
= f

(z
0
) = 100z
1001
0
= 100 1
99
= 100.
17. If z
0
is from the region {z : |z| < 1} then the function f coincides with the function g(z) = z
on some neighborhood of z
0
. Since g(z) is dierentiable everywhere then the limit used in the
denition of dierentiability of f coincides with the corresponding limit for g which we know is
equal to g

(z
0
) = 1. So, inside the region {z : |z| < 1} we have f

(z) = 1.
Now if z
0
is from the region {z : |z| > 1} then the function f coincides with the function
h(z) = z
2
on some neighborhood of z
0
. So, using the same argument as above we get that inside
this region {z : |z| > 1} we have f

(z) = (z
2
)

= 2z.
The question of dierentiability remains for z
0
from the circle {z : |z| = 1}. We have f(z
0
) = z
0
.
So, if z
0
= 1 then z
0
= z
2
0
and since f coincides with h(z) = z
2
inside the region {z : |z| > 1}. And
h(z) is continuous everywhere. So, if z approaches z
0
inside the region {z : |z| > 1} then f(z) = h(z)
approaches h(z
0
) = z
2
0
which we know is not equal to f(z
0
). So, f is not continuous at a such z
0
.
Now if z
0
= 1 then both g and h coincide at this point. Now let C
1
be any path which approaches
z
0
= 1 inside the region {z : |z| < 1} (for example C
1
= {z : z = r, 0 < r < 1}) and C
2
be any
path which approaches z
0
= 1 inside the region {z : |z| > 1} (for example C
2
= {z : z = r, r > 1}).
Since g and h are dierentiable everywhere we have
lim
zz
0
z on C1
f(z) f(z
0
)
z z
0
= lim
zz
0
z on C1
g(z) g(z
0
)
z z
0
= g

(z
0
) = 1
and
lim
zz
0
z on C2
f(z) f(z
0
)
z z
0
= lim
zz
0
z on C2
h(z) h(z
0
)
z z
0
= h

(z
0
) = 2z
0
= 2.
Section 2.3 Analytic Functions 35
Since, we have dierent limits when we approach z
0
= 1 from dierent directions it follows
that the derivative of f(z) does not exist at z
0
= 1. Finally, we see that the derivative of f exist
everywhere except the circle {z : |z| = 1 } and
f

(z) =
_
1 if |z| < 1,
2z if |z| > 1,
21. We take g(z) = z
p/q
and f(z) = z
q
in Theorem 4 and get h(z) = f(g(z)) = (z
p/q
)
q
= z
p
there.
Since f(z) = 0 for any z = 0 and z
p/q
0
= 0 for z
0
= 0 and not on the negative real axis we have
f

(g(z
0
)) = 0. We also know that
g(z) = e
p
q
( Log(z)+i2k)
for some integer k. So, g is a composition of two continuous functions on its domain. Therefore g is
also continuous. Now we can use Theorem 4 to get
d
dz
z
p/q
=
h

(z)
f

(g(z))
=
pz
p1
qg(z)
q1
=
pz
p1
qz
(p/q)(q1)
=
pz
p
z
1
qz
p
z
p/q
[if we divide the numerator and the denominator by z
p
and multiply them by z
p/q
z]
=
p
qz
z
p/q
.
25. Using the denition of derivative (??) for f(z) = z
n
and using the identity from the problem
we get
f

(z
0
) = lim
zz0
f(z) f(z
0
)
z z
0
= lim
zz0
z
n
z
n
0
z z
0
= lim
zz0
(z z
0
)(z
n1
+ z
n2
z
0
+ + z
n1
0
)
z z
0
= lim
zz0
(z
n1
+ z
n2
z
0
+ + z
n1
0
)
[using additivity and multiplicativity of limit we have]
= ( lim
zz0
z)
n1
+ ( lim
zz0
z)
n2
z
0
+ + z
n1
0
= z
n1
0
+ z
n2
0
z
0
+ + z
n1
0
= nz
n1
0
.
36 Chapter 2 Analytic Functions
Solutions to Exercises 2.4
1. For z = x + iy we get z = u(x, y) + iv(x, y) for u(x, y) = x and v(x, y) = y. Dierentiating u
with respect to x and y, we nd
u
x
= 1,
u
y
= 0.
Dierentiating v with respect to x and y, we nd
v
x
= 0,
v
y
= 1.
Comparing these derivatives we see clearly that
u
x
=
v
y
and
u
y
=
v
x
. Hence the Cauchy-
Riemann equations are satised at all points. The continuity of the partial derivatives follows from
Proposition 1, Section 2.2. Appealing to Theorem 1, we conclude that z is analytic at all points, or
entire. We compute the derivative using the formula (8). We have
d
dz
z = 1 + i0 = 1.
5. For z = x + iy, we get
e
z
= e
xiy
= e
x
(cos(y) + i sin(y))
= e
x
cos(y) i e
x
sin(y))
= u(x, y) + iv(x, y)
for u(x, y) = e
x
cos(y) and v(y) = e
x
sin(y)). Dierentiating u with respect to x and y, we nd
u
x
= e
x
cos(y),
u
y
= e
x
sin(y).
Dierentiating v with respect to x and y, we nd
v
x
= e
x
sin(y),
v
y
= e
x
cos(y).
Comparing these derivatives we notice that in general
u
x
=
v
y
and
u
y
=
v
x
. Really we can take
x = 0 and y = 0 to get a non-equality 1 = 1in the rst formula and x = 0 and y =

2
to get 1 = 1
in the second one. Hence the Cauchy-Riemann equations are not satised. So the function e
z
is not
analytic.
9. For z = x + iy we get
ze
z
= (x + iy)e
x+iy
= (x + iy)e
x
(cos y + i sin y)
= e
x
(xcos y y sin y + i(y cos y + xsiny))
= u(x, y) + iv(x, y)
for u(x, y) = e
x
(xcos y y siny) and v(x, y) = e
x
(y cos y + xsin y). Dierentiating u using the
product rule with respect to x, we nd
u
x
=

x
(e
x
xcos y e
x
y sin y)
=
_

x
(e
x
)x + e
x

x
(x)
_
cos y

x
(e
x
)y sin y
= (e
x
x + e
x
) cos y e
x
y sin y
= e
x
(xcos y + cos y y sin y).
Section 2.4 The Cauchy-Riemann Equations 37
Also dierentiating u with respect to y, we nd
u
y
=

y
(e
x
xcos y e
x
y sin y)
= e
x
x

y
(cos y) e
x
_

y
(y) sin y + y

y
(sin y)
_
= e
x
xsin y e
x
(sin y + y cos y)
= e
x
(xsin y sin y y cos y)
Similarly, we compute the derivatives v with respect to x and y
v
x
=

x
(e
x
y cos y + e
x
xsin y)
=

x
(e
x
)y cos y +
_

x
(e
x
)x + e
x

x
(x)
_
sin y
= e
x
y cos y + (e
x
x + e
x
) sin y
= e
x
(y cos y + sin y + xsiny)
and
v
y
=

y
(e
x
y cos y + e
x
xsiny)
= e
x
_

y
(y) cos y + y

y
(cos y) + x

y
(sin y)
_
= e
x
(cos y y siny + xcos y) .
Comparing these derivatives we see that
u
x
=
v
y
and
u
y
=
v
x
. Hence the Cauchy-Riemann
equations are satised at all points. Appealing to Theorem 1, we conclude that ze
z
is analytic at
all points, or entire. We compute the derivative using the formula (??). We have for z = x + i y
d
dz
ze
z
= e
x
(xcos y + cos y y sin y) + ie
x
(xsin y sin y y cos y).
13. For z = x + iy we know that |z| =
_
x
2
+ y
2
. So,
|z|
2
= x
2
+ y
2
= u(x, y) + i v(x, y).
for u(x, y) = x
2
+ y
2
and v(x, y) = 0. Dierentiating u with respect to x, we nd
u
x
=

x
_
x
2
+ y
2
_
= 2x.
Similar dierentiating u with respect to y, we nd
u
y
=

y
_
x
2
+ y
2
_
= 2y.
Obviously the derivatives of v(x, y) = 0 with respect to x and y are equal to 0,
v
x
= 0 and
v
y
= 0.
38 Chapter 2 Analytic Functions
The Cauchy-Riemann equations for the function f(z) = |z|
2
are
2x = 0 and 2y = 0.
The only complex number z = x + i y satisfying these equations is z = 0 with x = 0 and y = 0.
Since the function f(z) = |z|
2
is not dierentiable in a neighborhood of z = 0 it cannot be analytical
even at this point z = 0.
17. From the Example 2 we know that (sin z) = cos z. Now we compute the derivative of cos z.
Since
cos z =
e
iz
+ e
iz
2
,
(see (??), Section 1.6), we can appeal to (??) and conclude that cos z is entire and
d
dz
cos z =
d
dz
_
e
iz
+ e
iz
2
_
=
ie
iz
+ (i)e
iz
2
=
i(e
iz
e
iz
)
2
(i)
(i)
[since i
2
= 1]
=
e
iz
e
iz
2i
= cos z,
by (??), Section 1.6.
Now we use the product formula to compute the derivative of sin z cos z. We have
d
dz
(sin z cos z) =
_
d
dz
sinz
_
cos z + sin z
_
d
dz
cos z
_
= cos z cos z + sin z(sin z) = cos
2
z sin
2
z.
21. First, we compute the derivative of cosh z =
e
z
+e
z
2
. We have
d
dz
cosh z =
d
dz
_
e
z
+ e
z
2
_
=
e
z
e
z
2
= sinh z.
Now we use the chain rule to compute
d
dz
cosh(z
2
+ 3i) =
d cosh
dz
(z
2
+ 3i)
d
dz
(z
2
+ 3i)
[since
d
dz
(z
2
+ 3i) = 2z]
= sinh(z
2
+ 3i)(2z) = 2z sinh(z
2
+ 3i).
25. Since z
a
= e
a Logz
for a = 1/2 we can appeal to (??), (??), and the chain rule to compute
d
dz
1
(z i)
1/2
=
d
dz
_
e

1
2
Log(zi)
_
=
d
dz
_

1
2
Log (z i)
_
e

1
2
Log(zi)
=
1
2
1
z i
(z i)

1
2
Log(zi)
[since (z i)

= 1 and again e
a Logz
= z
a
]
=
1
2
1
z i
1
(z i)
1/2
=
1
2(z i)
3/2
Section 2.4 The Cauchy-Riemann Equations 39
where to get the last equality we used the property z
a
z
b
= z
a+b
which is proved as follows
z
a
z
b
= e
a Logz
e
b Logz
[since we have e
x
e
y
= e
x+y
]
= e
(a+b) Logz
= z
a+b
.
29. Since Log 1 = 0 and z = (1 + z) 1 we see that
Log (z + 1)
z
=
Log (z + 1) Log 1
(1 + z) 1
.
Therefore we can identify the limit with the derivative of the function f(z) = Log (z +1) at z
0
= 1.
Therefore we have
lim
z0
Log (z + 1)
z
=
_
d
dz
Log (z + 1)
_
(z
0
) =
1
z
0
+ 1
=
1
2
.
33. (a) These formulas are proved in the Theorem 1 or in the proof of the Cauchy-Riemann
equations.
(b) Using the rst formula from the part (a) we have
|f

(z)|
2
= |u
x
i u
y
|
2
[since |z|
2
= zz]
= (u
x
i u
y
)(u
x
i u
y
)
= (u
x
i u
y
)(u
x
+ i u
y
)
= u
2
x
+ i u
x
u
y
i u
y
u
x
=1
..
i
2
u
2
y
= u
2
x
+ u
2
y
.
By the Cauchy-Riemannn equations, u
x
= v
y
and u
y
= v
x
, we get
|f

(z)|
2
= u
2
x
+ u
2
y
= v
2
y
+ (v
x
)
2
= v
2
y
+ v
2
x
.
(c) If either u = Re f or v = Imf is constant in , then u
x
= 0 and u
y
= 0 or v
x
= 0 and v
y
= 0.
In either case by the part (b) we get |f

(z)|
2
= 0, or simply f

(z) = 0. Therefore by the Theorem 2

it follows that f is constant in .
37. Since u = v it follows that
u
x
= v
x
and u
y
= v
y
.
At the same time by the Cauchy-Riemann equations we also have
u
x
= v
y
and u
y
= v
x
.
Now we can use both sets of equations to get
u
x
= v
y
[since u
y
= v
y
]
= u
y
[since u
y
= v
x
]
= v
x
[since u
x
= v
x
]
= u
x
,
or equivalently 2u
x
= 0. So, u
x
= 0. Similar we can nd that
u
y
= v
x
= u
x
= v
y
= u
y
,
40 Chapter 2 Analytic Functions
or equivalently 2u
y
= 0 and u
y
= 0.
Since u
x
= u
y
= 0 from Theorem 1, Section 2.1, we conclude that u is constant. Since the real
part of f is constant from the result of the problem 33 it follows that f is constant itself.
41. First, we need to nd the value of f in polar coordinates. If z = re
i
then we have
f(z) = Log z = ln |z| + i Arg z
[since r = |z| and = Arg z]
= ln r + i .
Now we recall the Cauchy-Riemann equations in polar form given in the problem 39
u
r
=
1
r
v

and v
r
=
1
r
u

.
We have u(r, ) = ln r and v(r, ) = . Therefore we have
u
r
=

r
(ln r) =
1
r
, and u

(ln r) = 0.
Similar we evaluate
v
r
=

r
() = 0, and u

() = 1.
Therefore we see that
u
r
=
1
r
=
1
r
v

and v
r
= 0 =
1
r
0 =
1
r
u

.
Hence the function f is analytic. We can compute its derivative in polar form using the formula
from the part (c) of the problem 39. For z = re
i
we have
f

(z) = e
i
(u
r
+ i v
r
) = e
i
_
1
r
+ i 0
_
=
1
r
e
i
.
Section 2.5 Harmonic Functions and Laplaces Equation 41
Solutions to Exercises 2.5
1. We will directly verify Laplaces equation. For u = x
2
y
2
+2xy we evaluate partial derivatives
u
xx
= (u
x
)
x
= (2x + 2)
x
= 2
and
u
yy
= (u
y
)
y
= (2y 1)
y
= 2.
Since u
xx
+ u
yy
= 2 2 = 0 Laplaces equation is satised and u is a harmonic function.
5. Here we verify Laplaces equation again. For u =
1
x+y
we evaluate partial derivatives
u
xx
= (u
x
)
x
=
_

1
(x + y)
2
_
x
=
2
(x + y)
3
and
u
yy
= (u
y
)
y
=
_

1
(x + y)
2
_
y
=
2
(x + y)
3
.
Since u
xx
+ u
yy
=
4
(x+y)
3
= 0 Laplaces equation is not satised and u is not a harmonic function.
9. We again cab evaluate partial derivatives of u = e
x
2
y
2
cos(2xy) and verify whether u satises
Laplaces equation. But in this problem we can make a use of Theorem 1. We notice that the
following identity is true
(x + i y)
2
= x
2
+ 2i xy + (iy)
2
= x
2
y
2
+ i 2xy.
Using (1) and polar form for z = x + iy we have
e
z
2
= e
x
2
y
2
+i 2xy
= e
x
2
y
2
(cos(2xy) + i sin(2xy)).
Since u = Re (e
z
2
) we conclude from Theorem 1 that u is harmonic.
13. To nd a harmonic conjugate v for u = x+2y, we use the Cauchy-Riemann equations as follows.
We want u + i v to be analytic. Hence v must satisfy the Cauchy-Riemann equations
(1)
u
x
=
v
y
, and
u
y
=
v
x
.
Since
u
x
= 1, the rst equation implies that
1 =
v
y
.
To get v we will integrate both sides of this equation with respect to y. Doing so we x a value of
x. Therefore the result of integration with respect to y is a function of y plus some constant which
eventually may depend on x. Thus integrating with respect to y yields
v(x, y) = y + c(x),
where c(x) is a function of x alone. Plugging this into the second equation in (1), we get
2 =
_
0 +
d
dx
c(x)
_
,
42 Chapter 2 Analytic Functions
or equivalently
d
dx
c(x) = 2.
If we integrate this equation with respect to x we get that c(x) = 2x + C where C is any real
constant. Substituting the expression for c(x) into the expression (2) for v we get v(x, y) = y+c(x) =
y 2x + C.
Now we check the answer by using Cauchy-Riemann equations. We have u
x
= 1, u
y
= 2,
v
x
= 2, and v
y
= 1. Since u
x
= v
y
and u
y
= v
x
we conclude that v is a harmonic conjugate of u.
17. (a) Using the hint we evaluate f(z) = z
n
in polar coordinates. For z = re
i
we have
f(z) = z
n
= (re
i
)
n
= r
n
e
i n
= r
n
(cos(n) + i sin(n)) = r
n
cos(n) + i r
n
sin(n).
So we see that u(r, ) and v(r, ) are the real and imaginary parts of the function f. Hence, by
Theorem 1, it follows that u and v are harmonic in the domain of the function f. Since for n 0 the
function f(z) = z
n
is analytic in C it follows that u and v are harmonic in C as well. Also if n < 0
then the function f(z) = z
n
is analytic in C except z = 0. Hence u and v are harmonic everywhere
in C except z = 0.
(b) Since u(r, ) and v(r, ) are the real and imaginary part of an analytic function f(z) it
follows that v is a harmonic conjugate of u. Also from Proposition 2 it follows that correspondingly
u = r
n
cos(n) is a harmonic conjugate of v.
21. Since v is a harmonic conjugate of u it means that the function f(z) = u(z) +i v(z) is analytic.
Hence the function g(z) = (f(z))
2
= (u + iv)
2
is also analytic. We notice that
g(z) = (u(z) + i v(z))
2
= u(z)
2
+ 2i u(z)v(z) + (i v(z))
2
= u(z)
2
v(z)
2
+ 2i u(z)v(z).
Hence the real part of g is equal to u
2
v
2
and therefore it is harmonic by Theorem 1. Also the
imaginary part of g is equal to 2uv and thus it is also harmonic by Theorem 1.
25. (a) Since the function u does not depend on y it follows that all derivatives of u with respect
to y are equal to zero. In particular, u
yy
= 0. Therefore Laplaces equation u
xx
+ u
yy
= 0 can be
rewritten as u
xx
= 0. Or equivalently as (u
x
)
x
= 0. If we integrate both sides of this equation with
respect to x we generally nd that u
x
= a(y) where the constant a should depend on y. But we
know that u does not depend on y. Hence the rst derivative u
x
also does not depend on y. So
a does not depend on y and is a real constant. We can rewrite the equation like u
x
= a. If we
integrate this equation with respect to x again we nd that u = ax + b(y) where b(y) is, in general,
a function of y. But since u does not depend on y we get that b(y) = b is a constant too. Hence we
found that u = ax + b where a and b are any real constants.
(b) From the Figure we see that u = 10 if x = 3 and u = 40 if x = 5. If the harmonic function
u has the form described in the part (a), u = ax + b, it follows that the following equations are
satised
3a + b = 10 and 5a + b = 40.
If we subtract the rst equation from the second we nd 2a = 30. And thus a = 15. Now we plug-in
the value for a into the rst equation we get 45 + b = 10. Therefore we get b = 35. Hence the
function u(x, y) = 15x 35 is the solution of the boundary problem because it is harmonic by part
(a) and it satises the boundary condition as found in (2).
Section 2.5 Harmonic Functions and Laplaces Equation 43
(c) Since the function u(x, y) = 15x 35 depends only on x it follows that the isotherms are
lines parallel to the y-axis. Obviously for any such line x = c we have that for z = (x, y) lying on a
such line u(x, y) = 15c 35 does not change. Therefore a vertical line x = c is an isotherm for any
real constant c.
29. (a)
If u depends only on r then by using the Laplacian dierential equation in polar form
u =

2
u
r
2
+
1
r
u
r
+
1
r
2

2
u

2
.
we receive the equation for a harmonic u
0 =

2
u
r
2
+
1
r
u
r
= u
rr
+
1
r
u
r
.
(b) If we multiply the Euler equation received in the part (a) by r we get
0 = ru
rr
+ u
r
= (ru
r
)
r
.
We integrate this dierential equation with respect to r and we get
ru
r
= c
1
,
where c
1
is any constant. So, we get
u
r
=
c
1
r
.
And if we integrate this equation with respect to r one more time we have u(r) = c
1
lnr + c
2
for
any real constants c
1
and c
2
. And, since such functions satisfy Eulers equation we conclude they
are harmonic.
33. (a) We suppose that (z) is a harmonic conjugate of ln|z| in C \ {0}. So, the function
f(z) = ln |z| + i (z) is analytic in the same domain. We also know that the function Log z is
analytic in the region C\ (, 0] and have the real part ln|z| too. By Corollary 1, Section 2.4 we
have that f(z) = Log (z) +a for some complex number a. Therefore we have a similar equation for
imaginary parts in the region C\ (, 0]
(z) = Imf(z) = ImLog(z) + c = Arg z + c,
where c = Imy.
(b) By our assumption we have f(z) = ln|z| + i (z) should be analytic in the region C\ {0}.
By the Theorem 3 Section 2.2 it follows that f is also continuous in the same region. And, then by
the Theorem 1 from Section 2.3 it also follows that the imaginary part (z) is continuous there too.
Now if z approaches 1 from above then by part (a) we see that (z) = Arg z +c should approach c.
On the other hand if z approaches 1 from below then again by part (a) we see that (z) = Arg z +c
should approach 2 + c. Since c = 2 + c we conclude that our rst assumption was false and we
the function ln |z| does not have a harmonic conjugate in the region C\ {0}.
Argue that, since (z) is harmonic in C\ {0}, (z) is continuous on (, 0). Obtain a contra-
diction using (a) and the fact that the discontinuities of Arg z are not removable on the negative
x-axis (Example 7, Section 2.2).
37. Solve the Dirichlet problem in the upper half-plane with boundary data on the x-axis given by
u(x, 0) =
_
T
1
if a < x < b,
T
2
otherwise,
44 Chapter 2 Analytic Functions
where a < b are xed real numbers. [Hint: Use Exercise 35 and the fact that constant functions are
harmonic.]
From the Exercise 35 we know that the function
v(z) =
1

( Arg (z b) Arg (z a))

solves the Dirichlet problem
v(x, 0) =
_
1 if a < x < b,
0 otherwise,
If we pick any complex constants c
1
and c
2
then we see that the function c
1
v(z) +c
2
is also harmonic
and it satises the Laplaces equation because (c
1
v(z) + c
2
) = c
1
v(z) = 0. Now we can take
u(z) = (T
1
T
2
)v(z) + T
2
. By the remark above this function solves the Laplaces equation. And
we check that it also solves the given Dirichlet problem. For a < x < b we have
u(x0) = (T
1
T
2
)v(x0) + T
2
= (T
1
T
2
) + T
2
= T
1
.
And if x < a or x > b we have
u(x0) = (T
1
T
2
)v(x0) + T
2
= T
2
.
49. Show that if u is harmonic and independent of r then u

= 0.
Conclude that u = a + b; equivalently, u = a arg

z + b, where arg

z is a branch of the
argument.
If u is harmonic and independent of r then by the polar form of the Laplacian found in the
Exercise 47
u =

2
u
r
2
+
1
r
u
r
+
1
r
2

2
u

2
we conclude that
0 =
1
r
2

2
u

2
=
1
r
2
u

.
If we multiply this equation by r
2
we get that u

= 0. If we integrate this dierential equation with

respect to we get that u

= a for any real number a. Now we can integrate this new equation
again with respect to we get u = a + b where b is now also arbitrary.
Section 2.6 Dierentiation of Functions of Several Variables 45
Solutions to Exercises 2.6
1. If (x, y) = (0, 0) then we can compute the derivatives directly since x
2
+ y
2
= 0. We have
u
x
=
(xy)
x
(x
2
+ y
2
) xy(x
2
+ y
2
)
x
(x
2
+ y
2
)
2
=
y(x
2
+ y
2
) 2x
2
y
(x
2
+ y
2
)
2
.
Similar
u
y
=
(xy)
y
(x
2
+ y
2
) xy(x
2
+ y
2
)
y
(x
2
+ y
2
)
2
=
x(x
2
+ y
2
) 2xy
2
(x
2
+ y
2
)
2
.
Now if (x, y) = (0, 0) then we can notice that for any x
1
= 0 we have
u(x
1
, 0) =
x
1
0
x
2
1
+ 0
2
= 0.
Therefore
u
x
(0, 0) = lim
x10
u(x
1
, 0) u(0, 0)
x
1
0
= lim
x10
0
x
1
= 0.
Therefore the partial derivative of u with respect to x exists at (0, 0) and is equal to 0.
Similarly, we notice that for any y
1
= 0 we have
u(0, y
1
) =
0y
1
0
2
+ y
2
= 0.
Therefore
u
y
(0, 0) = lim
y10
u(0, y
1
) u(0, 0)
y
1
0
= lim
y10
0
y
1
= 0.
Hence u
y
exists at (0, 0) and is equal to 0.
To show that u(x, y) is not continuous at (0, 0) we pick a point (x, x) for any number x = 0. We
compute
u(x, x) =
x
2
x
2
+ x
2
=
1
2
.
Therefore if x approaches zero then (x, x) approaches (0, 0) but u(x, x) = 1/2 does not approach
u(0, 0) = 0. Therefore u is discontinuous at (0, 0).
5. Since f and g are dierentiable at x = x
0
and y = y
0
correspondingly we have
f(x) = f(x
0
) + f

(x
0
)(x x
0
) +
1
(x)|x x
0
|
and
g(y) = g(y
0
) + g

(y
0
)(y y
0
) +
2
(y)|y y
0
|
where
1
(x) 0 as x x
0
and
2
(y) 0 as y y
0
. Therefore we have
u(x, y) = f(x)g(y)
= (f(x
0
) + f

(x
0
)(x x
0
) +
1
(x)|x x
0
|)(g(y
0
) + g

(y
0
)(y y
0
)
+
2
(y)|y y
0
|)
= f(x
0
)g(y
0
) + g(y
0
)f

(x
0
)(x x
0
) + f(y
0
)g

(x
0
) + m(x, y),
46 Chapter 2 Analytic Functions
where
m(x, y) = f

(x
0
)g

(y
0
)(x x
0
)(y y
0
)
+
1
(x)|x x
0
|(g(y
0
) + g

(y
0
)(y y
0
) +
2
(y)|y y
0
|)
+
2
(y)|y y
0
|(f(x
0
) + f

(x
0
)(x x
0
)).
To show that that u is dierentiable at z
0
= (x
0
, y
0
) it is enough to check that
lim
zz0
m(z)
|z z
0
|
= 0.
First, we show that lim
zz0
(xx0)(yy0)
|zz0|
= 0. To see it we will use the squeeze theorem and the
inequality ab
a
2
+b
2
2
. To prove the inequality we start from the obvious inequality (a b)
2
0
which is equivalent to a
2
2ab + b
2
0 or if we add 2ab to both sides we get a
2
+ b
2
2ab and if
we divide by 2 we get
a
2
+b
2
2
ab which is the needed inequality. Now for a = xx
0
and b = y y
0
we get
(x x
0
)(y y
0
)
(x x
0
)
2
+ (y y
0
)
2
2
=
|z z
0
|
2
2
by Pythagorian theorem.
We also can put a = (x x
0
) and b = y y
0
to receive similar
(x x
0
)(y y
0
)
(x x
0
)
2
+ (y y
0
)
2
2
=
|z z
0
|
2
2
.
If we multiply the whole inequality by (1) then we need to reverse the inequality and we get
(x x
0
)(y y
0
)
|z z
0
|
2
2
.
If we combine the inequalities (1) and (1) we conclude that

|z z
0
|
2
2
(x x
0
)(y y
0
)
|z z
0
|
2
2
.
If we divide these inequalities by |z z
0
| we get

|z z
0
|
2

(x x
0
)(y y
0
)
|z z
0
|

|z z
0
|
2
.
We have lim
zz0

|zz0|
2
= 0 and lim
zz0
|zz0|
2
= 0. So, by squeeze theorem
lim
zz0
(x x
0
)(y y
0
)
|z z
0
|
= 0.
We notice that
lim
zz0
(g(y
0
) + g

(y
0
)(y y
0
) +
2
(y)|y y
0
|)
= lim
yy0
(g(y
0
) + g

(y
0
)(y y
0
) +
2
(y)|y y
0
|) = g(y
0
).
Now we nd that
lim
zz0

1
(x)|x x
0
|(g(y
0
) + g

(y
0
)(y y
0
) +
2
(y)|y y
0
|)
|z z
0
|
= lim
zz0

1
(x)|x x
0
|
|z z
0
|
lim
zz0
(g(y
0
) + g

(y
0
)(y y
0
) +
2
(y)|y y
0
|) = 0g(y
0
) = 0
Section 2.6 Dierentiation of Functions of Several Variables 47
since
|xx0|
|zz0|
is bounded by 1 and
1
(x) 0 if z z
0
.
Similarly, we nd that
lim
zz0

2
(y)|y y
0
|(f(x
0
) + f

(x
0
)(y y
0
))
|z z
0
|
= lim
zz0

2
(y)|y y
0
|
|z z
0
|
lim
zz0
(f(x
0
) + f

(x
0
)(x x
0
)) = 0f(x
0
) = 0.
Finally adding the limits (1)-(1) we conclude that
lim
zz0
m(z)
|z z
0
|
= 0.