Documente Academic
Documente Profesional
Documente Cultură
1
2
3
4
5
6
7
8
1
X
X
X
X
X
X
X
X
2
X
X
X
X
X
X
X
X
3
X
X
X
X
X
X
X
X
4
X
X
X
X
X
X
X
X
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
X X X X X X X
X X X X
X X X
X X X X X
X X X X
I.1.1
Identify and sketch the set of
(a) jz 1 ij = 1
(f)
(b) 1 < j2z 6j < 2
(g)
(c) jz 1j2 + jz + 1j2 < 8 (h)
(d) jz 1j + jz + 1j 2
(i)
(e) jz 1j < jzj
(j)
points satisfying.
0 < Im z <
< Re z <
jRe zj < jzj
Re (iz + 2) > 0
jz ij2 + jz + ij2 < 2
Solution
Let z = x + iy, where x; y 2 R.
(a) Circle, centre 1 + i, radius 1.
jz
ij = 1 , j(x
1)j = 1 , (x
1) + i (y
1)2 + (y
1)2 = 12
1 < j2z
6j < 2 , 1 < 2 jz
3j < 2 ,
jx + iy
, 1=2 < jz
p
(c) Disk, centre 0, radius 3.
, (x
3)2 + y 2 < 12
jz
q
2 , (x
1j + jz + 1j
q
,
(x
q
, (x + 1)2 + y 2
1) +
1) +
2
y2
x+1,
y2
2
2
(x + 1)2 + y 2 ,
q
2
2
2
(x + 1) + y
,
2
(x + 1) +
y2
(x + 1)2 , y = 0
x < 1;
1 x
x > 1;
then jx
1 then jx
then jx
1j + jx + 1j =
(x 1) (x + 1) =
2x 2;
1j + jx + 1j =
(x 1) + (x + 1) = 2 2
1j + jx + 1j = (x 1) + (x + 1)
= 2x 2:
jz
1j < jzj , jz
y+2>0,y <2
jz
I.1.2
Verify from the denitions each of the identities
(a) z + w = z + w
(b) zw = z w
(c) jzj = jzj
Draw sketches to illustrate (a) and (c).
(d) jzj2 = z z
Solution
Substitute z = x + iy and w = u + iv, and use the denitions.
(a)
z + w = (x + iy) + (u + iv) = (x + u) + (y + v) i =
= (x + u) (y + v) i = (x iy) + (u
iv) = z + w:
(b)
iy) (u
iv) = z w:
(c)
jzj = x + iy = jx
(d)
jzj2 = jx + iyj2 =
iyj =
p
x2 + y 2
x2 + ( y)2 =
x2 + y 2 = jx + iyj = jzj :
= x2 + y 2 =
= x2
i2 y 2 = (x + iy) (x
iy) = z z:
I.1.3
Show that the equation jzj2
centered at a with radius .
Solution
Let z = x + iy and a =
jzj2
= x2 + y 2
=x +y
represents a circle
+ i , we have
2 Re (az) + jaj2 =
2
2 Re (az) + jaj2 =
2 Re ((
i ) (x + iy)) +
2 Re (( x + y) + i ( y
2
=x +y
2 ( x + y) +
x)) +
2
2 Re (az) + jaj2 =
(x
)2 + (y
)2 + (y
)2 :
, becomes
)2 =
= (x
Thus the equation jzj2
centered at ( ; ), which in
I.1.4
Show that jzj jRe zj + jIm zj, and sketch the set of points for which
equality holds.
Solution
Apply triangle inequality to z = Re z + i Im z, to obtain jzj 6 jRe zj + jIm zj.
Now set z = x + iy, and see then equality holds
p
p
p
p
p
p
2
x2 + y 2 =
x2 + y 2
jzj = jRe zj+jIm zj , x2 + y 2 = x2 + y 2 ,
p
p p
p
2
2
x2 +
x2 +2 x2 y 2 , x2 y 2 = 0 , x = 0 or y = 0:
, x2 +y 2 =
Equality holds only when z is real or pure imaginary, which are all the points
on the real and imaginary axis.
I.1.5
Show that
jRe zj
jzj ;
jIm zj
jzj :
Show that
jz + wj2 = jzj2 + jwj2 + 2 Re (z w) :
Use this to prove the triangle inequality jz + wj
jzj + jwj.
Solution
Let z = x + iy. Then since the square root function is monotone, we have
p
p
jRe zj = jxj = x2 6 x2 + y 2 = jzj ;
p
p
jIm zj = jyj = y 2 6 x2 + y 2 = jzj :
= (z + w) (z + w) = (z + w) (z + w) = z z + z w + wz + ww =
= jzj2 + 2 Re (z w) + jwj2 ;
The desired inequality now follows by taking square root of both sides.
I.1.6
For xed a 2 C, show that jz
aj = j1
Solution
If jzj = 1, then jzj = 1 and z z = 1. Use this and get
jz
aj = jz
aj jzj = jz z
azj = j1
We have
as was to be shown.
jz
j1
aj
= 1;
azj
azj = j1
azj :
I.1.7
Fix > 0, =
6 1, and x z0 ; z1 2 C. Show that the set of z satisfying
jz z0 j = jz z1 j is a circle. Sketch it for = 12 and = 2, with
z0 = 0 and z1 = 1. What happens when = 1?
Solution
Recall that a circle in R2 centered at (a; b) with radius r is given by the
equation
a)2 + (y
(x
b)2 = r2 :
z0 j = jz
z1 j :
The solutions set of the equation above remains the same if we square both
sides,
z0 j2 =
jz
z1 j2 :
jz
(x
y0 )2 =
x1 )2 + (y
(x
y1 )2 :
x2 2 x0
x1 x+ x20
2
x2
(x0
1
x1 )
x+
(x20
1
2 2
x1
+ 1
y 2 2 y0
y1 y+ y02
), we have
2 2
x1 )
2
+ y2
(y0
1
y1 )
y+
(y02
1
2 2
y1 )
2
Now complete the squares for both the x and y terms. Recall that
x2
2ax + b = (x
So we have
a)2
a2 + b:
=0
2 2
y1
=0
x0
1
x1
y0
1
+ y
(1
) (x20
(1
y1
2 2
x1 )
2 )2
2
(1
(x0
) (y02
(1
x1 )
2 2
y1 )
2 )2
+
2
(y0
y1 )
= 0:
This becomes
x
=
x0
1
(x0
x1
+ y
y0
1
x1 ) + (y0
y1
2
2
y1 )
(1
) (x20 + y02
(x21 + y12 ))
2 )2
(1
x0 )2 + (y1
(x1
(1
2 )2
y0 )2
2
2
+y =
1
; 0 , and when = 2,
When = 21 we have a circle of radius 23 centered at
3
4
2
we have a circle of radius 3 centered at 3 ; 0 . When = 1, we have the
equation
jz
z0 j = jz
z1 j ;
which is the line bisecting the two points. When z0 = 0; z1 = 1, this is the
line Re z = 21 .
10
I.1.7
I.1.7
Im z
I.1.7
Im z
-1
-1
Re z
1
2
-1
Re z
-2
=2
11
1
-1
-2
=1
2
1
1
-1
-2
Im z
1
-1
Re z
I.1.8
Let p (z) be a polynomial of degree n
1 and let z0 2 C. Show
that there is a polynomial h (z) of degree n 1 such that p (z) =
(z z0 ) h (z)+p (z0 ). In particular, if p (z0 ) = 0, then p (z) = (z z0 ) h (z).
Solution
Set
p (z) = an z n + an 1 z n
+ an 2 z n
+ bn 3 z n
+ a2 z 2 + a1 z + a0 :
and
h (z) = bn 1 z n
+ bn 2 z n
+ : : : + b2 z 2 + b1 z + b0 :
+ an 2 z n
z0 ) h (z) +
+ a2 z 2 + a1 z + a0 =
= (z z0 ) bn 1 z n 1 + bn 2 z n 2 + bn 3 z n 3 + : : : + b2 z 2 + b1 z + b0 +p (z0 ) =
= bn 1 z n + bn 2 z n 1 + bn 3 z n 2 + : : : + b2 z 3 + b1 z 2 + b0 z + p (z0 )
bn 1 z0 z n 1 bn 2 z0 z n 2 bn 3 z0 z n 3 : : : b2 z0 z 2 b1 z0 z b0 z0 =
bn 1 z n +(bn 2 bn 1 z0 ) z n 1 +(bn 3 bn 2 z0 ) z n 2 +(bn 4 bn 3 z0 ) z n 3 +
+ (b2 b3 z0 ) z 3 + (b1 b2 z0 ) z 2 + (b0 b1 z0 ) z + p (z0 ) b0 z0 :
Equate and solve for the bj s in terms of aj s.
8
< an = b n 1
ak = bk 1 bk z0 ; 0
:
a0 = p (z0 ) b0 z0
8
b n 1 = an
>
>
>
nP
k 1
>
<
bk
=
ak+1+i z0i ; 0
1 )
i=0
>
n
>
P
>
>
ai z0i
: p (z0 ) =
i=0
: : : + a0 ;
p (z) = an (z
+ r (z) ;
z0 ) + c;
p (z) = an z n
+ q (z) (z
z0 ) + c = h (z) (z
13
1.
z0 ) + c
I.1.9
Find the polynomial h (z) in the preceding exercise for the following
choices of p (z) and z0
(a) p (z) = z 2 and z0 = i
(b) p (z) = z 3 + z 2 + z and z0 = 1
(c) p (z) = 1 + z + z 2 +
+ z m and z0 = 1
Solution
From the preceding exercise we have
p (z) = (z
z0 ) h (z) + p (z0 ) :
p (z)
z
p (z0 )
:
z0
(a)
We have that p (z) = z 2 and z0 = i, thus p (z0 ) = p (i) =
Thus
p (z)
h (z) =
z
1.
p (z0 )
z2 + 1
=
= z + i;
z0
z i
and
z 2 = (z
i) (z + i)
(b)
We have that p (z) = z 3 + z 2 + z and z0 =
Thus
h (z) =
p (z)
z
1:
1, thus p (z0 ) = p ( 1) =
z3 + z2 + z + 1
p (z0 )
(z + 1) (z 2 + 1)
=
=
= z 2 + 1;
z0
z+1
z+1
and
z 3 + z 2 + z = (z + 1) z 2 + 1
(c)
We have that p (z) = 1 + z + z 2 +
+ z m and z0 =
14
1:
1, thus
1.
0; m odd
1; m even
p (z0 ) = p ( 1) =
Thus
h (z) =
8
< (zm
=
: (zm
p (z)
z
1 +z m 3
p (z0 )
=
z0
+z 2 +1)(z+1)
= zm
z+1
+z 3 +z )(z+1)+1
z+1
1 +z m 3
= zm
+ zm
1
+ zm
+ z2 + 1
if m is odd
+ : : : + z 3 + z if m is even
and
zm + zm
+
=
z2 + z + 1 =
(z + 1) (z m
(z + 1) (z m
+ zm
1
+ zm
1
+
3
+
15
+ z 2 + 1)
if m is odd
3
+ z + z) + 1 if m is even
I.1.10
Let q (z) be a polynomial of degree m
p (z) can be expressed in the form
+ a1 z + a0
+ b1 z + b0 ;
and
q (z) = bm z m + bm 1 z m
where an 6= 0, bm 6= 0 and m
Let
p1 (z) =
n.
an n
z
bm
16
q (z)
p (z) ;
then
p1 (z) =
an n
z
bm
bm z m + bm 1 z m
+ b1 z + b0
an z n + an 1 z n
The monomials of degree n cancel, and therefore p1 (z) is a polynomial of degree at most n 1. It follows, by assumption, that p1 (z) = h1 (z) q (z)+r1 (z),
where h1 (z) and r1 (z) are the unique polynomials satisfying the conditions
above.
Let
h (z) =
an n
z
bm
h1 (z) ;
and let
r (z) =
r1 (z) :
Then
an n
z
bm
an n
z
bm
q (z)
an n
z
bm
h1 (z) q (z)
r1 (z) =
q (z)
p1 (z) =
= p (z) :
Thus given p (z) and q (z), there exist polynomials h (z) and r (z) satisfying
the above two conditions.
17
+ a1 z + a0 :
I.1.11
Find the polynomials h (z) and r (z) in the preceding exercise for
p (z) = z n and q (z) = z 2 1.
Solution
Require
z n = h (z) z 2
1 + r (z) ;
deg r (z)
1:
If n is even
zn = zn
+ zn
+ zn
+ z2 + 1
z2
1 + 1:
zn = zn
+ zn
+ zn
+ z3 + z
z2
1 + z:
If n is odd
Thus
h (z) =
zn
zn
2
2
+ zn
+ zn
4
4
+ zn
+ zn
6
6
+
+
+ z 2 + 1; if n is even,
;
+ z 3 + z; if n is odd.
and
r (z) =
1; if n is even,
:
z; if n is odd.
18
I.2.1
Express all values of the following expressions in both polar and
Cartesian
and plot them.
p coordinates,
p
4
(a)
i
(c)
1 (e) ( 8)1=3
(g) (1 + i)8
25
p
p
1+i
p
(b)
i 1 (d) 4 i
(f) (3 4i)1=8 (h)
2
Solution
(a)
p
i=
(b)
p
i( =2+2k ) 1=2
i( =4+k )
=e
1 =
2ei(3
21=4 ei3
=4+2k )
=8
o
; k = 0; 1 = ei
1=2
; 21=4 ei11
= 21=4 ei(3
=8
=8+k )
=4
i5 =4
;e
; k = 0; 1
p o
(1 + i) = 2 :
(c)
p
4
1 =
=
ei(
+2k ) 1=4
i =4
i3 i=4
;e
=4+k =2)
i5 =4
i7 =4
;e
;e
(d)
p
4
i=
ei(
o
; k = 0; 1; 2; 3 =
n
p
= (1 i)= 2; ( 1
= ei(
; k = 0; 1; 2; 3 = ei
=8
; ei5
p o
i) = 2 :
=8
; ei9
=8
; ei13
(e)
( 8)1=3 =
23 ei(
+2k ) 1=3
= 2ei
= 2ei(
=3
o
; k = 0; 1; 2 =
n
p
=3
= 1 + i 3; 2; 1
=3+2k =3)
; 2ei ; 2ei5
(f)
19
p o
i 3 :
=8
4i)1=8 =
51=8 ei(
1=8
5ei(
0 =8)
0 +2k
4=3 )
) 1=8
; 51=8 ei(
0 =8+
= 51=8 ei(
=4)
= tan
0 =8+k
=4)
; 51=8 ei(
1=8
0 =8+
4
3
o
; k = 0; 1; : : : ; 7 =
=2)
; 51=8 ei(
0 =8+3
=4)
51=8 (cos ( 0 =8 + =2) + i sin ( 0 =8 + =2)) ; 51=8 (cos ( 0 =8 + 3 =4) + i sin ( 0 =8 + 3 =4)) :
(g)
p
(1 + i)8 =
2ei(
=4+2k )
= 16ei(2
+16k )
= 16ei2 = 16:
(h)
25
1+i
p
2
= ei(
=4+2k ) 25
I.2.1a
= ei(25
=4+50k )
= ei(
=4+6 +50k )
I.2.1b
1
-1
-1
-1
1
-1
I.2.1e
I.2.1f
2
1+i
= p :
2
-1
I.2.1d
-1
=4
I.2.1c
-1
= ei
-2
-1
-2
20
-1
1
-1
I.2.1g
I.2.1h
20
1
-20
20
-1
1
-1
-20
21
I.2.2
Sketch the following sets
(a) jarg zj < =4
(c) jzj = arg z
(b) 0 < arg (z 1 i) < =3 (d) log jzj = 2 arg z
Solution
a) Sector. b) Sector. c) Is a spiral curve starting at 0, spiraling to 1. (d) Is
a spiral curve, spiraling to 0, and to 1.
I.1.2b
I.1.2a
I.2.2c
5
/3
(/2,0)
/4
/4
-5
-5
(,0)
(2,0)
(3/2,0)
-5
-5
(1,0)
22
I.2.3
For a xed complex number b, sketch the curve ei + be i : 0
2
Dierentiate between the cases jbj < 1; jbj = 1 and jbj < 1.
Hint. First consider the case b > 0, and then reduce the general
case to this case by a rotation.
Solution
For 0 < b < 1, an ellipse x2 = (1 + b)2 + y 2 = (1 b)2 = 1, traversed in positive
direction with increasing . For b = 1, an interval [ 2; 2]. For 1 < b < +1,
an ellipse traversed in negative direction. For b = ei' , express equation as
ei'=2 ei( '=2) + e i( '=2) to see that curve is rotate of ellipse or interval
by '=2.
I.2.7 (b = 0.5)
y
I.2.7 (b = 1)
2
1
-3 -2 -1
-1
I.2.7 (b = 1.5)
2
1
-3 -2 -1
-1
3
2
1
-3 -2 -1
-1
-2
-2
-2
-3
-3
-3
23
I.2.4
For which n is i an n-th root of unity?
Solution.
i is an nth root of unity for in = 1, n = 4k, k = 1; 2; 3; : : :
24
I.2.5
For n 1, show that
(a) 1 + z + z 2 +
+ z n = (1
(b) 1 + cos + cos 2 +
z n+1 ) = (1
+ cos n =
1
2
z),
1
sin n+
2
2 sin =2
z 6= 1
.
Solution
(a)
Set
+ zn;
sn = 1 + z + z 2 +
and multiply sn with z and have
+ z n+1 :
zsn = z + z 2 + z 3 +
Now subtract zsn from the sum sn , and have
sn (1
1 + ei + ei2 +
1+e
+e
i2
z n+1 :
z) = 1
z;
1
z n+1
:
1 z
;
ei(n+1)
;
1 ei
1 e i(n+1)
=
:
1 e i
+ ein =
+e
in
Add the identities, and use the denitions of sine and cosine.
25
2 (1 + cos + cos 2 +
+ cos n ) =
ei(n+1)
1 e i(n+1)
+
=
1 ei
1 e i
1 e i(n+1) ei =2
1 ei(n+1) e i =2
+
=
=
e i =2 ei =2
ei =2 e i =2
1 ei =2 1 e i(n+1)
1 e i =2 1 ei(n+1)
+
=
=
2i
sin ( =2)
2i
sin ( =2)
"
#
1
1
sin ( =2) + sin n +
1 ei =2 e i =2 + ei(n+ 2 )
e i(n+ 2 )
=
=
2i
sin ( =2)
sin ( =2)
=
1
2
sin n + 12
=1+
:
sin ( =2)
Divide both sides with 2 and get
1 + cos + cos 2 +
+ cos n =
26
1 sin n + 21
+
:
2
2 sin =2
I.2.6
Fix n 1. Show that the n th roots of unity w0 ; : : : ; wn
(a) (z w0 ) (z w1 ) : : : (z wn 1 ) = z n 1:
(b) w0 +
+ wn 1 = 0 if n 2:
(c) w0
wn 1 = ( 1)n 1 :
nP1
0; 1 k n 1;
wjk =
(d)
n; k = n:
j=0
Solution
(a)
Let w0 ; : : : ; wn 1 be the n
a roots of z n 1 since
wjn
1 = e2
ji=n n
ji
1 = e2
ji=n
satisfy
, and wj are
1 = 0:
By the fundamental theorem of algebra, and since the root are simple and
the coe cient for the z n term is 1 it follows that
zn
1 = (z
w0 ) (z
w1 )
(z
wn 1 ) :
(b)
Using exercise (a) and multiply the factors in the product we get
zn
1 = zn
wn 1 ) z n
(w0 + w1 +
+c2 z n
+( 1)n w0 w1
wn 1 :
+ wn
if
=0
2:
(c)
From (b) follows that
1 = ( 1)n w0 w1
(d)
If 1
n 1
X
j=0
wjk
n 1
X
j=0
wn
, w0 w1
wn
= ( 1)n
2 ji=n k
n 1
X
2 ki=n j
j=0
27
e2 ki=n
1 e2 ki=n
1 e2 ki
= 0:
1 e2 ki=n
If k = n
n 1
X
j=0
wjk
n 1
X
2 ji=n n
j=0
n 1
X
j=0
28
2 ji
n 1
X
j=0
1 = n:
I.2.7
Fix R > 1 and n
1, m
0. Show that
zm
zn + 1
Rm
Rn
jzj = R:
Sketch the set where equality holds. Hint. See (1.1) p.2.
Solution
I.2.7
I.2.7 (n = 1)
I.2.7 (n = 2)
a+b
b
-1
-1
a
-1
I.2.7 (n = 3)
I.2.7 (n = 4)
I.2.7 (n = 5)
-1
-1
-1
We use that jz
-1
-1
-1
wj
jzj
1
-1
jz n + 1j > jz n j
1 = Rn
1;
1
+ 1j
1
(Rn
29
1)
Rm
Rn
jzj = R:
1;
because jzjm = Rm .
We rearrarange this equality to
j 1j + jz n + 1j = jz n j
and we conclude that 1 and z n +1 must lie on the same ray. (We have used
the fact that jaj+jbj = ja + bj implies that a; b lie on the same ray, see rst gure I.2.7)
If
z = Rei
then
z n = Rn ein :
Since R > 1, we require that ein =
1 thus
z = wk Re
where wk is an n
If
i=n
i=n
th root of unity.
z = wk Re
then
zn + 1 =
Rn + 1;
and
Rm
zm
=
:
zn + 1
Rn 1
30
I.2.8
Show that cos 2 = cos2
sin2 and sin 2 = 2 cos sin using de
Moivres formulae. Find formulae for cos 4 and sin 4 in terms of
cos and sin .
Solution
Let 2 R be given. Then by de Moivres formulae (on page 8 in CA) we
have for all n 2 Z
cos n + i sin n = (cos + i sin )n
Hence for n = 2 we get
cos 2 + i sin 2 =
= (cos + i sin )2 = cos2 + i2 cos sin
= cos2
sin2 =
sin2
+ i (2 cos sin ) ;
then, by setting the real and imaginary parts equal to each other we obtain
cos 2
sin 2
= cos2
sin2
= 2 cos sin :
6 cos2 sin2
4 cos sin3
then, by setting the real and imaginary parts equal to each other we obtain
cos 4
sin 4
= cos4
6 cos2 sin2 + sin4
= 4 cos3 sin
4 cos sin3 :
as required.
31
I.3.1
Sketch the image under the spherical projection of the following
sets on the sphere
(a) the lower hemisphere Z 0
(b) the polar cap 43 Z 1
p
p
(c) lines of lattitude X = 1 Z 2 cos , Y = 1 Z 2 sin , for Z xed
and 0
2
p
p
(d) lines of longitude X = 1 Z 2 cos , Y = 1 Z 2 sin , for xed
and 1 Z 1
(e) the spherical cap A X 1, with center lying on the equator,
for xed A. Separate into cases, according to various ranges of A.
Solution
I.3.1a
I.3.1b
y
I.3.1c
-2
-1
-4
-2
-2
-1
-1
-2
-1
-2
-4
-2
I.3.1d
I.3.1e (A = -1/2)
y
I.3.1e (A = 0)
2
1
-2
-1
-6
-4
-2
-2
-1
-1
-2
-1
-2
-4
-2
32
I.3.1e (A = 1/2)
y
4
2
-2
-2
-4
(a)
Image is the unit disk.
(b)
p
2
Set Y = 0 and Z = 3=4 in X 2 + Y 2 + Zp
= 1, we have X = 7=4. We have
the formula x = X= (1 p
Z), thus x = 7. Image is the exterior of a disk,
with centre 0 and radius 7.
p
p
1
+
z=
1 z. We have that
(c) Image
is
the
disk,
with
centre
0
and
radius
p
p
2
2
Y = p 1 Z sin , we can take the radius to image
X = 1 Z cos and
p
1 Z2
.
for = 0, thus r = 1 Z = p1+Z
1 Z
(d) Image is lines of longitude issuing from 0.
(e)
Case 1: 1 < A < 0
p
Image is the exterior of the disk centered at 1=A with radius 1 A2 = jAj.
Case 2: A = 0
Image is the left halfplane.
Case 3: 0 < A < 1
p
Image is a disk centered at 1=A p
with radius p 1 A2 = jAj.
Set Y = 0 then Z goes from
1 A2 top 1 A2 . We have
p the formula
2
x = X= (1 Z) thus x goes from A= 1 + 1 A = 1
1 A2 =A to
p
p
A= 1
1 A2 = 1 + 1 A2 =A.
33
I.3.2
If the point P on the sphere corresponds to z under the stereographic projection, show that the antipodal point P on the sphere
corresponds to 1=z.
Solution
For z = x + iy the corresponding point on the sphere under transformation
given on p. 12 in CA is given by (X; Y; Z) where
2x
;
jzj2 + 1
2y
;
Y =
2
jzj + 1
jzj2 1
Z =
:
jzj2 + 1
X =
For z 6= 0, we have
y
x
i 2
2
x iy
(x
jzj
jzj
which corresponds to the point on the sphere given by (X 0 ; Y 0 ; Z 0 ) where
1
=
z
x + iy
=
iy) (x + iy)
Y0 =
Z
2x
jzj2
1 2
+
z
2y
jzj2
1 2
+
z
1 2
z
1 2
+
z
1
1
1
x + iy
=
x2 + y 2
2x
=
jzj + 1
X;
2y
=
jzj + 1
Y;
jzj2 1
=
jzj2 + 1
Z:
34
I.3.3
Show that as z travers a small circle in the complex plane in the
positive (counterclockwise) direction, the corresponding point P
on the sphere traverses a small circle in the negative (clockwise)
direction with respect to someone standing at the center of the
circle and with body outside the sphere. (Thus the stereographic
projection is orientation reversing, as a map from the sphere with
orientation determined by the unit outer normal vector to the complex plane with the usual orientation.)
Solution
Draw the picture. Or argue as follows. The orientation of the image circle
is the same for all circles on the sphere orientated so that N is outside the
circle. This can be seen by moving one circle continuously to the other, and
seeing that the image circles moves continuously. Thus we need to shrink it
only for the equator of the sphere oriented as indicated ( ), if the South Pole
is inside it. And the image circle is the unity and positive direction of the
unit circle ( ) is the converse. So the orientation of the image is clockwise
(negative).
35
I.3.4
Show that a rotation of the sphere of 180 about the X-axis corresponds under stereographic projection to the inversion z 7 ! 1=z of
C.
Solution
For z = x + iy the corresponding point on the sphere under transformation
given on p. 12 in CA is given by (X; Y; Z) where
2x
;
jzj2 + 1
2y
Y =
;
2
jzj + 1
jzj2 1
Z =
:
jzj2 + 1
X =
For z 6= 0, we have
1
1
x iy
x iy
x
=
=
= 2
=
z
x + iy
(x + iy) (x iy)
x + y2
jzj2
y
jzj2
Y0 =
Z0 =
2x
jzj2
=
1 2
+
1
z
2y
jzj2
=
1 2
+
1
z
1
1
jzj2
=
1
2 + 1
jzj
2x
= X;
jzj2 + 1
2y
=
jzj + 1
2
jzj2 1
=
jzj2 + 1
Y;
Z:
36
I.3.5
Suppose (x; y; 0) is the spherical projection of (X; Y; Z). Show that
the product of the distances from the north pole N to (X; Y; Z) and
from N to (x; y; 0) is 2. What is the situation when (X; Y; Z) lies on
the equator on the sphere?
Solution
The distance from from the north pole N = (0; 0; 1) to (X; Y; Z) is
q
X 2 + Y 2 + (Z 1)2 ;
X
; Y ;0
1 Z 1 Z
X 2 + Y 2 + (Z
1)2
1=2
X2
Y2
+
+1
(1 Z)2 (1 Z)2
1=2
X 2 + Y 2 + (Z
1 Z
1)2
When (X; Y; Z) lie on the equator, the product is simply the square of the
distance from N to a point on the equator. By the Pythagorean Law, this is
1 + 1 = 2.
37
= 2:
I.3.6
We dene the chordal distance d (z; w) between two points z; w 2 C
to be the length of the straight line segment joining to the points
P and Q on the unit sphere whose stereographic projections are z
and w, respectively.
(a) Show that the chordal distance is a metric, that is, it is symmetric, d (z; w) = d (w; z); satises the triangle inequality d (z; w)
d (z; ) + d ( ; w); and d (z; w) = 0 if and only if z = w.
(b) Show that the chordal distance from z to w is given by
2 jz wj
q
;
d (z; w) = q
2
2
1 + jzj 1 + jwj
z; w 2 C:
(c) What is d (z; 1)? Remark. The expression for d (z; w) shows
that innitesimal arc length corresponding to the chordal metric
is given by
d (z) =
2ds
;
1 + jzj2
wj2 =
= (z
= zz zw
2
= jzj + jwj
w) (z
w) = (z
w) (z
zw + ww = jzj + jwj
iy2 )
(x1
w) =
zw
= jzj + jwj2
38
zw =
2x1 x2
2y1 y2 : (1)
d (z; w)2 = (X X 0 ) + (Y Y 0 ) + (Z
= X 2 + Y 2 + Z 2 + X 02 + Y 02 + Z 02
=2
Z 0) =
2 (XX 0 + Y Y 0 + ZZ 0 ) =
!
4x1 x2 + 4y1 y2 + jzj2 1 jwj2 1
=
=2 2
jzj2 + 1 jwj2 + 1
!
jzj2 + 1 jwj2 + 1
4x1 x2 4y1 y2
jzj2 1 jwj2 1
jzj2 + 1
2jzj2 + 2 jwj2
jzj2 + 1
=2
jwj2 + 1
!
4 jz
4x1 x2 4y1 y2 (1)
=
2
2
jwj + 1
jzj + 1
wj2
:
jwj2 + 1
(c)
wj
2 jz=w 1j
2
q
q
= lim q
=q
:
w!1
2
2
2
2
2
1 + jzj 1 + jwj
1 + jzj 1= jwj + 1
1 + jzj
d (z; 1) = lim q
w!1
2 jz
z; w 2 C:
39
I.3.7
Consider the sphere of radius 21 in (X; Y; Z) space, resting on the
(X; Y; 0) plane, with south pole at the origin (0; 0; 0) and north pole
at (0; 0; 1). We dene a stereographic projection of the sphere onto
the complex plane as before, so that corresponding points (X; Y; Z)
and z (x; y; 0) lie on the same line through the north pole. Find
the equations for z = x + iy in terms of X; Y; Z, and the equations
for X; Y; Z in terms of z. What is the corresponding formula for the
chordal distance? Note. In this case, the equation of the sphere is
2
X 2 + Y 2 + Z 12 = 41 .
Solution
The line through P = (X; Y; Z) and N = (0; 0; 1) is on the form N +
t (P N ), and it meats the xy plane when
(x; y; 0) = (0; 0; 1) + t ((X; Y; Z)
1
2
, jzj2
1
t
1
= , jzj2 +
4
t
2
t + 1 = 0 , t = jzj2 + 1
thus we have
8
x
X = 1+jzj
2
>
<
y
Y = 1+jzj2
>
: Z = jzj2 :
1+jzj2
40
t2
,
4
1 2
2
1)) :
1
4
we
jz
wj2 =
= (z
= zz zw
2
= jzj + jwj
w) (z
w) = (z
w) (z
iy2 )
(x1
zw + ww = jzj + jwj
w) =
zw
= jzj2 + jwj2
41
zw =
x1 1 + jwj2
jzj2 1 + jwj2
+
y2
1 + jwj2
+ y1 1 + jwj2
2
1 + jzj2
1 + jwj2
2
1 + jzj2
1 + jzj2
1 + jzj2
1 + jzj2
jz1 j2
+
1 + jzj2
1 + jwj2
2y1 y2 1 + jwj2
1 + jwj2
+ jwj2 1 + jzj2
+ jzj4
2 (x1 x2 + y1 y2 ) 1 + jwj2
1 + jwj2
1 + jzj2
jzj2 + jwj2
=
1 + jzj2
1 + jzj2
2 (x1 x2 + y1 y2 ) 1 + jwj2
1 + jwj2
jzj2 + jwj2
=
1 + jzj2
1 + jzj2
1 + jzj2
1 + jzj2
1 + jwj2
1 + jwj2
1 + jwj2
jzj2 + jwj2
1 + jzj
1 + jwj2
2 (x1 x2 + y1 y2 )
2
1 + jwj
(1)
This gives
42
2 (x1 x2 + y1 y2 ) 1 + jwj2
2 (x1 x2 + y1 y2 )
jwj2
1 + jzj2
1 + jwj2
!2
+ jzj2
y2 1 + jzj2
+ jzj4
1 + jwj2
jz2 j2
1 + jwj2
jzj2 + 2 jzj2 jwj2 + jzj2 jwj4 + jwj2 + 2 jwj2 jzj2 + jwj2 jzj4 + jzj4
+
2x1 x2 1 + jwj2
y1
+
1 + jzj2
x2 1 + jzj2
x2
1 + jwj2
x1
1 + jzj2
1 + jzj2
1 + jzj2
=
(1)
jz wj2
:
jzj2 + 1 jwj2 + 1
d (z; w) = q
jz
wj
q
:
2
2
jzj + 1
jwj + 1
43
I.4.1
Sketch each curve and its image under w = z 2 .
(a) jz 1j = 1 (c) y = 1
(e) y 2 = x2 1; x > 0
(b) x = 1
(d) y = x + 1 (f) y = 1=x; x 6= 0
Solution
I.4.1a z-plane
y
I.4.1b z-plane
y
4
2
-4
-2
-4
-2
-4
-4
-4
-2
-4
-4
44
-4
I.4.1f z-plane
y
4
2
2
-2
-2
-2
I.4.1c w-plane
I.4.1e z-plane
y
2
-2
-4
-2
-2
-2
-4
I.4.1d z-plane
-2
-4
-2
-4
-4
-2
I.4.1b w-plane
-2
-4
I.4.1a w-plane
4
2
-2
-4
-4
4
2
-2
I.4.1c z-plane
-4
-2
2
-2
-4
I.4.1d w-plane
y
I.4.1e w-plane
y
4
2
-4
-2
4
2
2
-2
I.4.1f w-plane
-4
-2
-4
-4
45
2
-2
-4
-2
2
-2
-4
I.4.2
Sketch the image of eachpcurve in the preceding problem under the
principal branch of w = z, and also sketch, on the same grid but
in apdierent color, the image of each curve under the other branch
of z.
Solution
w = z 1=2 = jzj ei argjzj
I.4.2a z-plane
y
1=2
I.4.2b z-plane
y
4
2
-4
-2
-4
-4
-2
-4
-2
-4
-4
46
4
2
2
-2
I.4.2c w-plane
2
-2
-4
I.4.2b w-plane
y
4
2
-2
-2
I.4.2c z-plane
-4
I.4.2a w-plane
-2
p
jzjei Arg z=2
-2
-4
-4
-2
n 1=2
= jzj ei Argjzj+i2
-4
-2
-2
-4
I.4.2e z-plane
I.4.2d z-plane
y
-2
-4
-2
-4
-2
2
-2
-4
-2
-4
-4
47
4
2
2
-2
I.4.2f w-plane
-4
I.4.2e w-plane
-2
-4
I.4.2d w-plane
-2
-2
-4
-4
-2
-4
I.4.2f z-plane
-4
-2
2
-2
-4
I.4.3
(a) Give a brief description of the function z 7 ! w = z 3 , considered
as a mapping from the z
plane to the w- plane. (Describe what
happens to w as z traverses a ray emanating from the origin, and as
z traverses a ray a circle centered at the origin.) (b) Make branch
cuts and dene explicitly three branches of the inverse mapping.
(c) Describe the construction of the Riemann surface of z 1=3 .
Solution
(a)
The function w = f (z) = z 3 . For z = rei , we have z 3 = r3 ei3 . The radical
rays at angle are mapped to rays at angle 3 , that is, arg w = 3 arg z. The
magnitude of a complex number is cubed, jwj = jzj3 . Circles, centered at the
origin of radius r, are mapped to cocentric circles with radius r3 .
(b)
We make branch cuts at ( 1; 0],
z = w1=3 = jwj ei arg w
1=3
we choose
g (z) = z 1=3 = r1=3 ei =3 ,
< < .
Sheet 1 : Take f1 (z) = g (z),
Sheet 2 : Take f2 (z) = e2 i=3 g (z),
Sheet 3 : Take f3 (z) = e4 i=3 g (z).
n 1=3
= ei2
n=3
p
jwjei Arg w=3
(c) Top edge of cut on sheet 1 to bottom edge of cut on sheet 2. Top edge of
cut on sheet 2 to bottom edge of cut on sheet 3. Top edge of cut on sheet 3
to bottom edge of cut on sheet 1. The endpoints for f (z) is continuous on
surface
Spara
48
I.2.1b
I.2.1c
I.4.3a
-1
-1
Spara
49
-1
1
-1
I.4.4
Describe how to construct the
following functions
p Riemann surfaces for the 2=5
(a) w = z 1=4 ;
(b) w = z i;
(c) w = (z 1) :
Remark. To describe the Riemann surface of a multivalued function, begin
with one sheet for each branch of the function, make branch cuts so that the
branches are dened continuously on each sheet, and identify each edge of
a cut on one sheet to another appropriate edge so that the function values
match up continuously.
Solution
(a) Use four sheets, can make branch cuts along real axis from 1 to 0.
(b) Use two sheets, can make branch cuts along horizontal line from 1 + i
to i.
(c) Use ve sheets, can make branch cuts along real axis from 1 to 1.
50
I.5.1
Calculate and plot for ez for the following points z.
(a) 0
(c)
(i 1) =3
(e) i=m;
m = 1; 2; 3; : : :
(b) i + 1
(d) 37 i
(f) m (i 1)
m = 1; 2; 3; : : :
Solution
I.5.1e
I.5.1f
1
0.2
-1
-0.4
-0.2
-1
0.2
-0.2
(a)
e0 = 1:
(b)
e
i+1
= e i e1 =
e:
(c)
e
(i 1)3
=e
=3
i=3
p !
3
1
+
i = 0:351 + 0:006i:
2
2
=3
=e
(d)
e37 i = e36 i e i =
(e)
We take the limit for the sequence e
have
e
i=m
i=m
, m = 1; 2; 3; : : : as m ! 1, and
! 1;
51
1:
m mi
m mi
e j=e
! 0;
52
I.5.2
Sketch each of the following gures and its image under the exponential map w = ez . Indicate the images of horizontal and vertical
lines in your sketch.
(a) the vertical strip 0 < Re z < 1,
(b) the horizontal strip 5 =3 < Im z < 8 =3,
(c) the rectangle 0 < x < 1, 0 < y < =4,
(d) the disk jzj
=2,
(e) the disk jzj
,
(f) the disk jzj 3 =2.
Solution
I.5.2a z-plane
I.5.2b z-plane
-2
6
2
2
-4
-4
I.5.2a w-plane
-2
I.5.2b w-plane
-2
-2
-4
-4
I.5.2c z-plane
-4
-2
-2
-4
-4
53
I.5.2c w-plane
3
2
2
-2
I.5.2e z-plane
I..5.2d z-plane
-4
-2
-4
-2
-4
-2
-2
-2
-2
-4
-4
-4
I.5.2e w-plane
I..5.2d w-plane
2
-2
-4
100
10
-2
I.5.2f w-plane
20
-4
I.5.2f z-plane
-10
10
20
30
100
-10
-20
54
-100
I.5.3
Show that ez = ez .
Solution
ez = ex
iy
= ex e
iy
i sin y) =
55
I.5.4
Show that the only periods of ez are the integral multiples of 2 i,
that is, if ez+ = ez for all z, then is an integer times 2 i.
Solution
Show that the only periods of ez are the integral muliples of 2 i that is, if
ez+ = ez for all z, then is an integer times 2 i.
ez = ez+ = ez e ) e = 1 )
56
= 2 mi
I.6.1
Find and plot log z for the following complex numbers z. Specify
the principal value.
p
(a) 2
(b) i
(c) 1 + i
(d) 1 + i 3 =2
Solution
(a)
Suppose that n = 0; 1; 2; : : :
log 2 = log j2j + i Arg 2 + 2 ni = log 2 + i2 n:
(b)
Suppose that n = 0; 1; 2; : : :
log i = log jij + i Arg i + 2 ni = i =2 + i2 n:
(c)
Suppose that n = 0; 1; 2; : : :
log (1 + i) =
= log j1 + ij + i Arg (1 + i) + 2 ni = log
2 + i =4 + i2 n =
= log 2=2 + i =4 + i2 n:
(d)
Suppose that n = 0; 1; 2; : : :
log
p
1+i 3
2
=
= log
p
1+i 3
2
+ i Arg
p
1+i 3
2
+ 2 ni =
= i=3 + i2 n:
57
I.6.1a
I.6.1b
I.6.1c
20
20
20
10
10
10
-1
-1
-1
-10
-10
-10
-20
-20
-20
I.6.1d
20
10
-1
-10
-20
58
I.6.2
Sketch the image under the map w = Log z of each of the following
gures.
(a) the right halv-plane Re z > 0,
(b) the half-disk jzj < 1, Re z > 0,
(c) the unit circle jzjp= 1,
p
e),
(d) the slit annulus e < jzj < e2 , z 62 ( e2 ;
(e) the horizontal line y = e,
(f) the vertical line x = e.
Solution
(b) We have a disk with radius less then 1, this means jzj < 1, thus log jzj
< 0 and is unbounded, therefore, it goes from 0 to 1. Since Re (z) > 0,
the polar angle is between 2 and 2 .
(d) Here, we have an annulus, the polar angle is from
to , thus the argument is in this range. Since, the log function is the inverse of the exponential
map, circles in z
plane are stright lines in the w
plane (w = log z).
Therefore, the image p
of this annulus under the log function is the rectangle,
bounded by x = log j ej = 12 and x = log je2 j = 2.
I.6.2b z-plane
I.6.2a z-plane
I.6.2c z-plane
4
3
2
1
-3 -2 -1
-1
-2
-4
-2
2
-2
-2
-2
-4
-3
59
I.6.2a w-plane
I.6.2b w-plane
I.6.2c w-plane
/2
2
1
-3 -2 -1
-1
3
/2
-2
-3
I.6.2d z-plane
I.6.2e z-plane
8
6
4
2
-8 -6 -4 -2
-2
-4
-6
-8
2 4 6 8
I.6.2d w-plane
-4
-4
I.6.2f z-plane
-2
-4
-2
-2
-2
-4
-4
I.6.2e w-plane
I.6.2f w-plane
-2
-4
-2
-4
-2
-2
-2
-2
-4
-4
-4
60
I.6.3
Dene explicitly a continuous branch of log z in the complex plane
slit along the negative imaginary axis, Cn [0; i1).
Solution
I.6.3
We have log z = log rei = log r + i . To avoid the negative imaginary axis
we chose
=2 < < 3 =2. We use the branch
f rei
= log r + i ;
of log z.
61
=2 <
< 3 =2;
I.6.4
How would you make a branch cut to dene a single-valued branch
of the function log (z + i 1)? How about log (z z0 )?
Solution
Any straight line cut from z0 to 1, in any direction, will do.
62
I.7.1
Find all values and plot them.
(a) (1 + i)i
(b) ( i)1+i
(c) 2
1=2
p
1+i 3
(d)
(1 i)
Solution
(a)
Suppose that m; n = 0; 1; 2; : : :
(1 + i)i =
= ei log(1+i) = ei(logj1+ij+i Arg(1+i)+i2
m)
= ei(log
2+i =4+i2 m)
=e
2 m
= e2 n e
=4 i log
=4 i log
=2
(b)
Suppose that m; n = 0; 1; 2; : : :
( i)1+i =
= e(1+i) log(
=e
2 m
i)
= e(1+i)(logj
=2 i(
=2+2 m)
= [m =
= e(1+i)(
n] = e2 n e
i =2+i2 m)
=2 i(
=2 2 n)
ie2 n e
(c)
Suppose that m = 0; 1; 2; : : :
1=2
=e
=
log 2=2
=e
=e
=e
log 2=2
=
(d)
Suppose that n = 0; 1; 2; : : :
63
i m
=
p
1= 2:
= [m =
n] =
p
1+i 3
1 i
= e(1
= e(1
p
i) log(1+i 3)
= e(1
p
p
i)(logj1+i 3j+i Arg(1+i 3)+i2 n)
= e2 n elog 2+
=3 i(
log 2+ =3+2 n)
= e2 n elog 2+
64
=3 i(
=
log 2+ =3)
I.7.2
h
i
2i
Compute and plot log (1 + i) .
Solution
We rewrite the expression, suppose that k; m; n = 0; 1; 2; : : :
h
i
log (1 + i)2i =
i
h
log(1+i) 2i
= log e2i log(1+i) = log e2i(logj1+ij+i Arg(1+i)+i2 k) =
= log e
h
i
p
= log e2i(log 2+i =4+i2 k) = log e 4 k e =2 ei log 2 = [k = m] = log e4 m e
= log e4
=2 i log 2
+i Arg e4
=2 i log 2
I.7.2
20
10
10
-10
-20
65
-10
=2 i log 2
20
I.7.3
Sketch the image of the sector f0 < arg z < =6g under the map w =
z a for
(b) a = i
(c) a = i + 2
(a) a = 32
Use only the principal branch of z a .
Solution
I.7.3a z-plane
I.7.3b z-plane
-5
I.7.3c z-plane
-5
-5
-5
-5
-5
I.7.3b w-plane
I.7.3a w-plane
I.7.3c w-plane
-5
-2
2
1
-2
-5
Some points on the y-axis is marked, and the point (1; 0) too.
66
I.7.4
Show that (zw)a = z a wa , where on the right we take all possible
products.
Solution
Let 2 (zw)a , and k; m; n = 0; 1; 2; : : :
=
= ea log(zw) = ea(logjzwj+i Arg(zw)+i2
n)
=e
Conversely, if
m+i2 n)
a
2z w :
2 z a wa , say
=
= ea log z ea log w = ea(logjzj+i Arg z+i2
= ea(logjzj+logjwj+i Arg z+i Arg w+i2
(m+n))
= ea(logjzwj+i Arg(zw)+i2
=e
k+i2 (m+n))
=
= ea log(zw) 2 (zw)a
67
I.7.5
i
Find ii . Show that is does not coincide with ii i = i 1 .
Solution
We rewrite the expression ii , suppose that k = 0; 1; 2; : : :
ii =
= ei log i = ei(logjij+i Arg(i)+i2
k)
= ei(i
=2+i2 k)
=
=e
2 k
=2
ii =
= ii
= ei log(i ) = ei(log(e
i
i( 2 k
=e
2 k
=2+i2 m)
=2
)) = ei(logje
=e
2 k
2 m i( 2 k
= 1=i =
68
=2)
=2
j+i Arg(e
ie
2 k
2 m
=2
=
=
)+i2
ie2
m)
I.7.6
Determine the phase factors of the function z a (1 z)b at the branch
points z = 0 and z = 1. What conditions on a and b guarantee that
z a (1 z)b can be dened as a (continuous) single-valued function
on Cn [0; 1]?
Solution
Phase factors is e2 ia at 0 and e2 ib at 1. Require e2 ia e2 ib = 1, or a + b = n,
where n 2 Z , to have a continuous singlevalued determination of z a (1 z)b
on Cn [0; 1].
69
I.7.7
Let x1 < x2 <
< xn be n consecutive
points on the real axis.
p
Describe the Riemann surface of
(z x1 ) (z xn ). Show that
for n = 1 and n = 2 the surface is topologically a sphere with certain
punctures corresponding to the branch points and 1. What is it
when n = 3 or n = 4? Can you say anything for general n? (Any
compact Riemann surface is topologically a sphere with handles.
Thus a torus is topologically a sphere with one handle. For a given
n, how many handles are there, and where do they come from?)
Solution
p
p
For n = 1 and n = 2, the functions f1 p
= z x1 and f2 = (z x1 ) (z x2 ),
and for a general n we have fn (x) = (z x1 ) (z x2 ) : : : (z xn ). To constuct the Riemann Suface we use two sheets with slits [x1 ; x2 ] ; [x3 ; x4 ] ; : : :.
If n is odd, we also need a slit [xn ; +1) . Identify top edge of slit on one
sheet with bottom edge of slit on other sheet. If n = 1 and n = 2 the surface
is topologically a sphere with certain punctures corresponding to the branch
points and 1. If n = 3 or n = 4, surface is a torus. For a general even
n, fn (x) will be continuous on Cn ([x1 ; x2 ] [ [x3 ; x4 ] [ : : : [ [xn 1 ; xn ]), and
its Riemann surface will be a sphere with n2 1 holes. For a general odd n
though, fn (x) will be continuous on Cn ([x1 ; x2 ] [ [x3 ; x4 ] [ : : : [ [xn ; +1]),
its Riemann surface will be a sphere with n 2 1 holes.
70
I.7.8
p
Show that z 2 1=z can be dened as a (single-valued) continuous
function outside the unit disk, that is, for jzj > 1. Draw branch
cuts so that the function can be dened continuous o the branch
cuts. Describe the Riemann surface of the function.
Solution
p
The function is z 1 1=z 3 . If jzj > 1, can use the principal value of the
square root to dene a branch of the function. There are branch points at
z = 0 and z 3 = 1, that is at 0; 1; e2 i=3 and e 2 i=3 . Make two branch cuts by
connecting any two pairs of point by curves; for instance, connect 0 to 1 by
a straight line, and the other cube roots of unity by a straight line or arc of
unit circle. The resulting two-sheeted surface with identication of cuts and
with points at innity is a torus.
71
I.7.9
q
Consider the branch of the function z (z 3 1) (z + 1)3 that is positive at z = 2. Draw branch cuts so that this branch of the function
can be dened continuously o the branch cuts. Describe the Riemann surface of the function. To what value at z = 2 does this
branch return if it is continued continuously once counterclockwise
around the circle fjzj = 2g?
Solution
We have that the function is
p p
(z + 1) z 2 z (1
1=z 3 ) (1 + 1=z):
If jzj > 1, the second square root can be dened to be single-valued for
jzj > 1. The value of the function at z = 2 returns to thepnegative of their
initial value then we travers the circle jzj = 2 , because z does. We can
construct a Riemann Surface by making cuts at 1; 0; 1; e2 i=3 ; e 2 i=3 and
1. The function for the Riemann Surface is
q
(z + 1) z (z 1) (z + 1) (z e2 i=3 ) (z e 2 i=3 ):
72
I.7.10
q
Consider the branch of the function z (z 3 1) (z + 1)3 (z 1) that is
positive at z = 2. Draw branch cut so that this branch of the function can be dened continuously o the branch cuts. Describe the
Riemann surface of the function. To what value at z = 2 does this
branch return if it is continued continuously once counterclockwise
around the circle fjzj = 2g?
Solution
We have that the function is
q
4
z (1 1=z 3 ) (1 + 1=z)3 (1
1=z)
The branch that is positive for z = 2 is, it is dened continuously for jzj > 1.
Branch return to original value around circle jzj = 2. We can construct a
Riemann Surface by making cuts at 0; 1; e2 i=3 and e 2 i=3 .
The function for the Riemann Surface is
q
(z 1) (z + 1) z (z + 1) (z e2 i=3 ) (z e 2 i=3 ):
73
I.7.11
p
Find the branch points of 3 z 3
of the function.
Solution
I.7.11a
I.7.11b
1
I.7.11c
1
-1
-1
-1
-1
-1
-1
We rewrite as follows
p
3
z3
1=
q
3
(z
1) (z
e2
i=3 ) (z
2 i=3 ):
This equation will have 3 branch points, which are the cube root of unity.
So the phase factor is
e2 i=3 . The Riemann surface is obtained ty pasting three sheets with the
corresponding branch cuts, we end up with a one hole torus.
Make two cuts, from 1 to e 2 i=3 , on each sheet. In this case the cuts share
common endpoint.
We nee three shets where f0 (z), f1 (z) = e2 i=3 f0 (z) and f2 (z) = e 2 i=3 f0 (z)
Note: Can use Riemannformula to see that surface is a torus.
Cheek by going around each little tip what phase change to, which of for
your sheet.
74
I.8.1
Establish the following addition formulae
(a) cos (z + w) = cos z cos w sin z sin w;
(b) sin (z + w) = sin z cos w + cos z sin w;
(c) cosh (z + w) = cosh z sinh w + sinh z sinh w;
(d) sinh (z + w) = sinh z cosh w + cosh z sinh w;
Solution
(a)
Using the denitions of sine and cosine functions given on page 29 in CA we
have,
cos (z + w) =
1
ei(z+w) + e i(z+w)
=
2ei(z+w) + 2e i(z+w) =
=
2
4
1
i(z+w)
i(z w)
=
e
+e
+ e i(z w) + e i(z+w) +
4
+ ei(z+w) ei(z w) e i(z w) + e i(z+w) =
eiz + e iz eiw + e iw eiz e iz eiw e iw
=
=
2
2
2i
2i
= cos z cos w sin z sin w:
(b)
Using the addition formula (a), and using that sin z = cos (z
cos z = sin (z
=2) we have,
sin (z + w) = cos z + w
2
= cos z cos w
=2) and
sin z sin w
=
2
2
= cos z sin w + sin z cos w = sin z cos w + cos z sin w:
(c)
Using the addition formula (a) and the formulas cos (iz) = cosh z and sin (iz) =
i sinh z given on page 30 in CA we have,
75
cosh (z + w) = cos (i (z + w)) = cos (iz) cos (iw) sin (iz) sin (iw) =
= cosh z cosh w i sinh z i sinh w = cosh z cosh w + sinh z sinh w:
(d)
Using the addition formula (b) and the formulas cos (iz) = cosh z and sin (iz) =
i sinh z given on page 30 in CA we have,
sinh (z + w) = i sin (i (z + w)) = i [sin (iz) cos (iw) + cos (iz) sin (iw)] =
= i [i sinh z cosh w + cosh z i sinh w] = sinh z cosh w + cosh z sinh w:
76
I.8.2
Show that jcos zj2 = cos2 x + sinh2 y, where z = x + iy. Find all zeros
and periods of cos z.
Solution.
Use trigonometric formulas from page 29 and 30 in CA we have,
cos z = cos (x + iy) = cos x cos (iy) sin x sin (iy) = cos x cosh y i sin x sinh y:
Now take the modulus squared, and use cosh2 y = 1 + sinh2 y,
jcos zj2 = cos2 x cosh2 y + sin2 x sinh2 y =
The identity for jcos zj2 shows that the only zeros of cos z are the zeros of
cos z on the real axis, because
cos z = 0 ,
cos x = 0 , x = 2 + m ; m = 0; 1; 2; : : : ;
sinh y = 0 , y = 0:
Translation by any period of cos z sends zeros to zeros. Thus any period
is an integral multiple of , and since odd integral multiples are not periods,
the only periods of cos z are 2 n; 1 < n < 1.
77
I.8.3
Find all zeros and periods of cosh z and sinh z.
Solution
We have that cosh z = cos (iz), thus the zeros of cosh z are at z = i =2 +im ,
m = 0; 1; 2; : : :, and the periods of cosh z are 2 mi, m = 0; 1; 2; : : :.
We have that sinh z = i sin (iz), thus the zeros of sinh z are at z = m i,
m = 0; 1; 2; : : :, and the periods of cosh z are 2 mi, m = 0; 1; 2; : : :.
78
I.8.4
Show that
tan
z=
1
log
2i
1 + iz
1 iz
Solution
Set z = tan w, we have that
z = tan w =
eiw e
sin w
=
cos w
i (eiw + e
iw
iw )
e2iw 1
:
i (e2iw + 1)
1 + iz
;
1 iz
1
log
2i
z=
1 + iz
1 iz
79
1 + iz
1 iz
I.8.5
Let S denote the two slits along the imaginary axis in the complex
plane, one running from i to +i1, the other from i to i1. Show
that (1 + iz) = (1 iz) lies on the negative real axis ( 1; 0] if and
only if z 2 S. Show that the principal branch
Tan
z=
1
Log
2i
1 + iz
1 iz
maps the slit plane CnS one-to-one onto the vertical strip fjRe wj < =2g.
Solution
I.8.5 z-plane
I.8.5
I.8.5
i
-1
-i
w1 =
1+iz
1 iz
w2 = Log w1
I.8.5 w-plane
/2
/2
z=
1 w1
i
1 + w1
1 w1
i goes from i to i1 along iR, and when
As w1 goes from 1 to 1, 1+w
1
1 w1
w1 goes from 1 to 0, 1+w1 i goes from i1 to i along iR.
Since we have that
1 + iz
1 w1
= w1 ) z =
i
1 iz
1 + w1
the map is one to one.
Remark that the map makes correspondence between the interval [ i: i1]
in z plane and [ 1; 1] in w1 plane, and between interval [i; i1] in z
plane and [ 1; 0] in w1 plane.
We have that w2 = Log w1 maps Cn( 1; 0) onto fjIm wj < g, thus w = w2i2
maps jIm w2 j < onto jRe wj < 2 , se gures.
We have that the function
Tan
z=
1
Log
2i
1 + iz
1 iz
maps CnS onto jRe wj < 2 , where Tan 1 z is the principal branch for
tan 1 z. The other branches are given by fn (z) = Tan 1 z + n , 1 < n <
1.
81
I.8.6
Describe the Riemann surface for tan
z.
Solution
I.8.6 z-plane
I.8.6
I.8.5
____
++++
+++++++++++
++++++
_ _ _ _-1
__
____
++++
-i
_ _ _ _
_______
w1 =
1+iz
1 iz
w2 = Log w1
+++++++++++
___________
I.8.6 w-plane
/2
w=
1
w
2i 2
/2
1
2i
Log
1+iz
1 iz
We have that
f0 (z) = Tan
z=
1
Log
2i
1 + iz
1 iz
Make innite many copies of S and call them Sn . Dene fn (z) = Tan 1 z+n
on Sn . Identify "+" side of cut on Sn to " " side of cut on Sn+1 . Then
fn on Sn continuous to fn+1 on Sn+1 , and fn maps Sn onto vertical strip
n 21 < Re z < n + 12 . Note that composite function is not dened at
i and 1 (endpoints of slits), and its image omits the sequence 12 + n,
1 < n < 1.
83
I.8.7
Set w = cos z and
w2
i
1 ;
1. Show that
Solution
Set w = cos z and
eiz + e
2
iz
+ 1=
:
2
Solve for
=w
and set
w2
1;
i log w
w=
w2
1 :
84
w2
1 :
I.8.8
Show that the vertical strip jRe (w)j < =2 is mapped by the function
z (w) = sin w one-to-one onto the complex z plane with two slits
( 1; 1] and [+1; +1) on the real axis. Show
p that the inverse
function is the branch of sin 1 z = i Log iz + 1 z 2 obtained by
taking the principal value of the square root. Hint. First show
that the function 1 z 2 on the slit plane omits the negative real
axis, so that the principal value of the square root is dened and
continuous on the slit plane, with argument in the open interval
between
=2 and =2.
Solution
Set w = x + iy, by formulas on page 30 in CA we have that
sin w = sin (x + iy) = sin (x) cos (iy) + cos (x) sin (iy) =
= sin x cosh (y) + i cos (x) sinh y:
If jRe wj = jxj <
because
= eiw and 1= = e
iw
eiw
e iw
:
2i
in (1), we have
1=
:
2i
Mulitplying the above by 2i and doing some rearranging, we nd that
z=
85
(1)
2iz =
or
2iz
1 = 0:
z2
or
eiw = iz +
and nd
1
z2
We now take the logarithm of both sides of the last equation and divide the
result by i to obtain
w=
and, since w 2 sin
p
1
log iz + 1
i
z2
z,
sin
z=
i log iz
86
z2 :
II 1 2 3
1
X X
2 X X X
3 X X X
4
X
5 X X X
6
7
8
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
X X X X X X X X X X X X X X
X X X
X
X
X
X X
X
X X X X X
II.1.1
Establish the following:
n
(a) lim n+1
= 1 (c) lim
n!1
(b)
lim 2n
n!1 n +1
= 0 (d)
2np +5n+1
=
p
n!1 nn +3n+1
z
lim
= 0; z 2
n!1 n!
2; p > 1
C:
Solution
(a)
n
1
= lim
= 1:
n!1 n + 1
n!1 1 + 1=n
lim
(b)
1=n
n
=
lim
= 0:
n!1 1 + 1=n2
n!1 n2 + 1
lim
(c)
2np + 5n + 1
2 + 5n1
lim
= lim
n!1 np + 3n + 1
n!1 1 + 3n1
+n
p+n
p
p
= 2:
(d)
Let m be a positive integer so that m > 2 jzj and let n > m, then
0
jzjm jzjn m
zn
=
=
n!
m! (m + 1) (m + 2)
n
jzj
jzj
jzj
jzjm jzj jzj
jzj
jzjm
=
m! m + 1 m + 2
n
m! m m
m
m
m
jzj
1 1
1
jzj
1
=
!0
n
m! 2 2
2
m! 2 m
zn
n!
! 0 as n ! 1 as was to be
II.1.2
For which values of z is the sequence fz n g1
n=1 bounded? For which
values of z does the sequence converge to 0?
Solution
The sequence is bounded for jzj
II.1.3
Show that fnn z n g converges only for z = 0.
Solution
If we choose
n>
then
2
jzj
2 < n jzj
and
jnn z n j = jn zjn > 2n ! 1
when n ! 1 for z 6= 0.
II.1.4
Show that lim
N!
k
N !1 N (N k)!
= 1;
0.
Solution
We have that
Nk
N (N 1) : : : (N k + 1)
N!
=
=
(N k)!
N N ::: N
as N ! 1 (k is xed).
1
N
2
N
::: 1
1
N
!1
II.1.5
Show that the sequence
1 1
1
+ +
+
log n; n 1;
2 3
n
is decreasing, while the sequence an = bn 1=n is increasing. Show
that the sequences both converge to the same limit . Show that
1
< < 35 . Remark. The limit of the sequence is called Eulers
2
constant. It is not known whether Eulers constant is a rational
number or an irrational number.
bn = 1 +
Solution
We have that
1 1
+ +
2 3
1 1
= 1+ + +
2 3
bn = 1 +
an
1
n
1
n
1
n
log n;
log n:
1
dt
t
((1))
1
1
dt <
t
k
((2))
Zk+1
k
1
an =
n
n+1
Z
dt
> 0;
t
bn+1
1
bn =
n+1
1
log (n + 1) + log n =
n+1
n+1
Z
dt
< 0;
t
a7 =
49
20
1
log 7 > ;
2
7;
thus 21 < an , if n 7.
We have after some investigation on the decreasing sequence bn that
bn
b22 =
3
log 22 < ;
5
19 093 197
5173 168
22
II.1.6
For a complex number
choose n" by
= 1;
1)
(
n!
n + 1)
1:
! 0 if and only if Re
6= 0; 1; 2; : : :, then
6=
(d) If Re
1,
(e) If Re a >
1 and
>
1.
n+1
1, then
1.
>
n+1
1.
n
for all n
n+1
<
0.
n
for n large
Solution
(a)
We have that
n
Y
k=1
1
Y
+1
=
=
k
k=1
2 (1 + Re ) (1 + Re )2 (1 + Im )2
+
+
:
k
k2
k2
Let
(1 + Re )2 (1 + Im )2
tn =
+
k2
k2
If Re < 1, we have that tk > 0 and
diverge is n unlimited.
If Re
1, we have that tk =
(Im )2
k2
2 (1 + Re )
k
! 1 then n ! 1 because
and
+1
k
tk
! A then n ! 1, and
P
P (Im )2
> 1 and tk =
<1
k2
n
If Re > 1 we have that it exists k 2 N so that 1 < tk < 0 for all k > K.
This gives that n is a decreasing series with respect to n, and so limited.
The conclution is that
1.
b)
We have that n ! 0 if and only if n ! 0. From (a) we see that
the only possibility for this is Re > 1. For Re > 1 we can see that
0 is decreasing. Let M = lim n . Then we hav that M = lim n =
n
n
lim
= M so M =
n!1 n n+1
and only if Re > 1.
c)
If
6= 0; 1; 2; : : :, then
n
(
n+1
n
n!1
n!1
M () M = 0 so
! 0, then n ! 1 if
1):::(
(n+1)+1)
(n+1)!
(
1):::(
n+1)
n!
1
n
= n 1 !
n+1
1+ n
as n ! 1.
d)
j(n+1)j
=
j(n)j
e)
Let
nj
n+1
distance( ;n)
distance( 1;n)
> 1, if Re
1,
6=
1. Then
n)2 +(Im )2
(n+1)2
<1
for n >
j j2 1
:
2(1+Re )
Parts (a) and (b) seems to require some facts about products:
n
Q
k=1
n
Q
k=1
n
Q
k=1
1+
! 1 as n ! 1,
>0
! 1 as n ! 1,
>0
k2
! A 6= 0 as n ! 1
II.1.7
Dene x0 = 0, and dene by induction xn+1 = x2n + 14 for n 0. Show
that xn ! 12 .
Hint. Show that the sequence is bounded and monotone, and that any limit
satises x = x2 + 41 .
Solution
We rst prove by induction that for every integer n
0
xn
1
2
and
xn
1
4
1
4
0 we have
xn+1 :
and so
1
1
and
x0 = 0
= x1 :
2
4
Hence, the assertion holds for n = 0. Now suppose that the assertion holds
for n = k, that is,
0
x0
0
For 0
0 xk
x
1
2
xk
1
2
and
xk
xk+1 :
1
,
2
x2 +
and hence 0
xk+1
xk
1
,
2
xk+1 =
x2k
1
+
4
1
2
1
4
1
.
2
Since
1
1
=
4
2
and
xk+1
x2k+1 +
1
= xk+2 :
4
Hence, the inductive step holds and so the sequence fxn gn is bounded and
increasing. It follows by Bounded Monotone Sequence Theorem that it
converges, say xn ! x. Note that the real-valued function f given by
f (t) = t2 + 41 for t 2 R is continuous and therefore we have,
1
x = lim xn+1 = lim f (xn ) = f (x) = x2 + :
n!1
n!1
4
It follows that x = x2 +
1
4
and so x = 12 . Hence, xn !
10
1
2
as required.
II.1.8
Show that if sn ! s, then jsn
sn 1 j ! 0.
Solution
We use standard "=2 proof and let sn ! s.
Let " > 0, choose N such that
jsn
sj < "=2
sn j
jsn+1
for n > N .
If n > N , then we have
jsn+1
sn j = jsn+1
s+s
sj + jsn
Thus
jsn+1
sn j ! 0
as n ! 1.
11
II.1.9
Plot each sequence and determine its lim inf and lim sup.
(c) sn = sin ( n=4)
(a) sn = 1 + n1 + ( 1)n
n
(b) sn = ( n)
(d) sn = xn (x 2 R xed)
Solution
(a)
lim sup sn = 2 lim inf sn = 0
(b)
lim sup sn = 1 lim inf sn = 1
(c)
lim sup sn = 1 lim inf sn = 1
(d)
8
jxj > 1;
< +1;
1;
jxj = 1;
lim sup sn =
:
0;
jxj < 1:
and
8
1;
1<x<
>
>
>
>
x = 1;
< 1;
0;
jxj < 1;
lim inf sn =
>
>
1;
x = 1;
>
>
:
+1;
x > 1:
1;
12
II.1.10
At what points are the following function continuous? Justify your
answer.
(a) z
(c) z 2 = jzj
(b) z= jzj
(d) z 2 = jzj3
Solution
(a)
continuous everywhere
(b)
Continuous except at z = 0, where it is not dened
(c)
Continuous for jzj =
6 0. It is not dened at z = 0, but it has a limit 0 as
z ! 0. If we dene the function to be 0 at z = 0, it is contionuous there.
(d)
Contionuous for jzj =
6 0. It has no limit as z ! 0
13
II.1.11
At what points does the function Arg z have a limit? Where is Arg z
continuous? Justify your answer.
Solution
Arg z have a limit at each point of Cn ( 1; 0]. It is continuous in Cn ( 1; 0].
It is discontinuous at each point of ( 1; 0]. Values of function ! f g at
points of ( 1; 0], the function may take any value in intervall [ ; ] as
z ! 0. The value is depending in that direktion we approach z = 0.
14
II.1.12
Let h (z) be the restriction of the function Arg z to the lower halfplane fIm z < 0g. At what points does h (z) have a limit? What is
the limit?
Solution
No limit at 0, has limit at all other points of closed lower half-plane, limit
=
at points of ( 1; 0).
15
II.1.13
For which complex values of does the principal value of z have
a limit as z tends to 0? Justify your answer.
Solution
Set z = rei = eLog r ei = eLog r+i and
z = eLog r+i
Re +i Im
= eLog r Re e
Im
= eLog r Re
ei(Log r Im
+ i Im , then
= Re
+i Log r Im +i Re
+ Re )
= rRe e
Im
Im
ei(Log r Im
+ Re )
thus
jz j = rRe e
Im
16
II.1.14
Let h (t) be a continuous complex-valued function on the unit interval [0; 1], and consider
Z 1
h (t)
dt:
H (z) =
z
0 t
then th(t)
! h(t)
uniformly for 0
zn
t z
H (zn ) ! H (z).
17
II.1.15
Which of the following sets are open subsets of C? Which are close?
Sketch the sets.
(a) The punctured plane, Cn f0g .
(b) The exterior of the open unitdisk in the plane, fjzj 1g .
(c) The exterior of the closed unitdisk in the plane, fjzj > 1g .
(d) The plane with the open unitinterval removed, Cn (0; 1) .
(e) The plane with the closed unitinterval removed, Cn [0; 1] .
(f) The semidisk, fjzj < 1; Im (z) 0g .
(g) The complex plane, C.
Solution
(a) open
(c) open
(e) open
(b) closed
(d) neither
(f) neither
Note: Sketches are di cult to do reasonably.
18
II.1.16
Show that the slit plane Cn ( 1; 0] is star-shaped but not convex.
Show that the slit plane Cn [ 1; 1] is not star-shaped. Show that a
punctured disk is not star-shaped.
Solution
(a) Cn ( 1; 0] is starshaped, because we can see every point from z = 1.
To show that Cn ( 1; 0] is not convex we may choose any two points such
that the stright line segment joining them contains a point in ( 1; 0]. For
example take 1 + i and 1 i. These two points are joined by a vertical
line which contain 1. From this follows that Cn ( 1; 0] is not convex.
(b) Cn [ 1; 1] is not starshaped. Given any z we have that z can not bee
seen from a stright line between z and z that must contain 0. We also have
that if z belong to the domain then z belongs also to the domain.
(c) Same argument as in (b) shows that Cn f0g is not starshaped.
19
II.1.17
Show that a set is convex if and only if it is star-shaped with respect
to each of its points.
20
II.1.18
Show that the following are equivalent for an open subset U of the
complex plane.
(a)
Any two points of U can be joined by a path consisting of straight
linesegments parallel to the coordinate axis.
(b)
Any continuously dierentiable function h (x; y) on U such that rh =
0 is constant.
(c)
If V and W are disjoint open subsets of U such that U = V [ W ,
then either U = V or U = W . Remark. In the context of topological
spaces, this latter property is taken as denition of connectedness.
Solution
Show that the following equivalent for an open subset U of the complex plane.
(a) ) (b) Suppose rh = 0 in D. Fix z0 2 D. If z1 2 D, join by polygonal
curve. rh = 0 ) h is constant on each segment of curve ) h (z0 ) = h (z1 ).
) h is constant in D.
(b) ) (c) Suppose rh = 0, h not constant, say h (z0 ) = 0 for some z0 . Let
W = fh = 0g, V = fh 6= 0g. Evidently V is open. Since rh = 0, h is
constant in a neighborhood of each point, so fh = 0g is open and W is open.
W 6= ?, V 6= ?, W \ V = ?, W [ V = U
(c) ) (a) Suppose z0 2 U . Let D = points that can be joined to z0 by a
polygonal curves, intervals parallel to coordinate axis. Evidently D is open.
Also if can connected points near z0 2 U , then can connect to U , so U nD is
open. By (c), U nD must be empty, some D 6= ?:
21
II.1.19
Give a proof of the fundamental theorem of algebra along the following lines. Show that if p (z) is a nonconstant polynomial, then
jp (z)j attains its minimum at some point z0 2 C. Assume that
the minimum is attained at z0 = 0, and that p (z) = 1 + az m +
,
where m 1 and a 6= 0. Contradict the minimality by showing that
P "ei 0 < 1 for an appropriate choice of 0 .
Solution (K. Seip)
Set
p (z) = an z n + an 1 z n
+ a0 ;
9 R > 0 such that jp (z)j > ja0 j for all jzj > R. Thus jp (z)j attains
its minimum in jzj
R. We may assume it is attained at z0 = 0.
Suppose a0 6= 0, say a0 = 1. Then p (z) = 1 + az m + higher order
p (z) = 1
1
a
1
m
(any m
"m + O "m+1 :
Thus for " su ciently small " we have jp (z)j < 1, which is a contradiction.
22
1j R
N 1
Then p (z) 6= 0 for jzj = R, the term aN z N dominates and further jp (z)j >
jp (0)j for jzj = R.
Let z0 be a point in the disk jzj 6 R at which the continuous function for
jp (z)j attains a minimum. Then jz0 j < R, so that is a disk centred at z0 so
that jp (z)j > jp (z0 )j on the disk. Suppose p (z0 ) 6= 0. We can assume that
p (z0 ) = 1.
Write
z0 )m + O (z
p (z) = 1 + am P (z
where am 6= 0. Suppose am = r0 ei 0 .
Conside
zm = z0 + "e i
0 =m
i =m
z0 )m+1 ;
; " > 0:
We have
p (zm ) = 1 + r0 ei 0 em e
Have p (zm )
1
that p (z0 ) = 0.
+ O "m+1 = 1
r0 "m + O "m+1 :
23
II.2.1
Find the derivatives of the following function.
n
(a) z 2 1
(c) (z 2 1)
(e) 1= (z 2 + 3)
(b) z n 1
(d) 1= (1 z)
(f) z= (z 3 5)
Solution
(a) 2z
(b) nz n
(c) n (z 2
(d) 1= (1
n 1
1)
z)2
(e)
2z= (z 2 + 3)
(f) ( 2z 3 5) = (z 3
2z
24
5)
II.2.2
Show that
1 + 2z + 3z 2 +
+ nz n
zn
z)2
nz n
:
1 z
z n+1
;
1 z
z 6= 1:
1
(1
Solution
Use the geometric sum
1 + z + z2 + z3 +
+ zn =
1 + 2z + 3z 2 +
+ nz n 1 =
(n + 1) ( 1) z n (1 z) + (1 z n+1 )
=
=
(1 z)2
nz n+1 nz n + z n+1 z n + 1 z n+1
=
=
(1 z)2
1 zn
nz n+1 nz n
1 zn
=
+
=
(1 z)2
(1 z)2
(1 z)2
25
nz n
:
(1 z)
II.2.3
Show from the denition that the functions x = Re z and y = Im z
are not complex dierentiable at any point.
Solution
Dierentiation f (z) = x = Re z (from the denition)
lim
f (z +
z!0
z)
z
f (z)
= lim
z!0
Re (z +
z) Re (z)
=
z
Re z
1; if
=
= lim
0; if
z!0
z
z= x
:
z=i y
lim
z!0
g (z +
z)
z
g (z)
lim
lim
Im (z +
z!0
Im z
=
z!0
z
z)
z
Im (z)
0;
if
i; if
=
z= x
:
z=i y
26
II.2.4 (+ IV.8.2)
Suppose f (z) = az 2 + bz z + cz 2 , where a, b, and c are xed complex
numbers. By dierentiating f (z) by hand, show that f (z) is complex dierentiable at z if and only if bz + 2cz = 0. Where is f (z)
analytic?
Solution
We will nd
f (z +
a (z +
z)
z
d
f
dz
f (z)
z)2 + b (z +
a z 2 +2z z +
=
z2
z) + c(z + z)
(az 2 + bz z + cz 2 )
=
z
2
+ b z z + z z+z z + z z + c z 2 + 2z z + z
z) (z +
z
2
2az z + a z + bz z + bz z + b z z + 2cz z + c z
=
=
z
2
z
z
z
= 2az + a z + bz
+ bz + b z + 2cz
+c
=
z
z
z
z
= 2az + a z + bz + b z + c z
+ (bz + 2cz)
z
2
4z!0
z
z
z
z
d
ist. Therefore, unless bz + 2cz = 0, the derivative dz
f (z) would not exist.
Therefore f (z) is analytic on C if b = c = 0, otherwise is the function not
analytic on any open set.
27
(az 2 + bz z + cz 2 )
II.2.5
Show that if f is analytic on D, then g (z) = f (z) is analytic on the
reected domain D = fz : z 2 Dg, and g 0 (z) = f 0 (z).
Solution (A. Kumjian)
Let f be analytic on the domain D and dene g on D as above (note that
D is also a domain). For a complex-valued function ' dened near a point
z0 it is easy to show that
lim ' (z) exists i lim ' (z) exists;
z!z0
z!z0
and if either exists, then the two limits are complex conjugates of each other.
Let z 2 D be given. We rst show that g is dierentiable at z and that
g 0 (z) = f 0 (z).
1
lim (g (z + h)
h!0 h
1
f z+h
h!0 h
g (z)) = lim
1
f (z) = lim (f (z + k)
k!0 k
f (z)) = f 0 (z)
28
II.2.6
Let h (t) be a continuous complex-valued function on the unit interval [0; 1], and dene
Z 1
h (t)
dt;
z 2 Cn [0; 1] :
H (z) =
z
0 t
Show that H (z) is analytic and compute its derivative. Hint. Differentiate by hand, that is, use the dening identity (2.4) of complex derivative.
Solution
We use denition (2.4) of derivative to nd H 0 (z).
H 0 (z) =
Z 1
h (t)
h (t)
dt
dt =
= lim
z!0
z
z
z
0 t
0 t
Z 1
1
1
1
= lim
h (t)
=
z!0 z
t z
z t z
0
Z 1
Z 1
h (t)
h (t)
dt =
= lim
=
z!0
z
z) (t z)
z) (t z)
0 (t
0 (t
Z 1
h (t)
=
dt:
z)2
0 (t
H (z +
z)
z
H (z)
1
= lim
z!0 z
29
II.3.1
Find the derivatives of the following functions.
sin z
sinh z
(a) tan z = cos
(b) tanh z = cosh
(c) sec z = 1= cos z
z
z
Solution
(a)
d
d sin z
cos z cos z sin z ( sin z)
cos2 z + sin2 z
1
tan z =
=
=
=
:
2
2
dz
dz cos z
cos z
cos z
cos2 z
(b)
cosh2 z sinh2 z
1
d
d sinh z
cosh z cosh z sinh z sinh z
=
=
:
tanh z =
=
2
2
dz
dz cosh z
cosh z
cosh z
cosh2 z
(c)
d
d 1
sec z =
=
dz
dz cos z
1
sin z
(
sin
z)
=
= tan z sec z:
cos2 z
cos2 z
30
II.3.2
Show that u = sin x sinh y and v = cos x cosh y satisfy the CauchyRiemann equations. Do you recognize the analytic function f =
u + iv? (Determine its complex form)
Solution
We have
@
@
@v
@u
=
sin x sinh y = cos x sinh y =
cos x cosh y =
;
@x
@x
@y
@y
@u
@
@
@v
=
sin x sinh y = sin x cosh y =
cos x cosh y =
:
@y
@y
@x
dx
Hence, u and v satisfy Cauchy-Riemann equations, and now we calculate
f (z).
f (z) =
e ix
ey e y
eix + e ix
ey + e y
+i
=
2i
2
2
2
eix+y + eix y + e ix+y + e ix
eix+y eix y e ix+y + e ix iy
+i
4
4
ix y
ix+y
i(x+iy)
i(x+iy)
iz
e
+e
e
+e
e + e iz
=i
=i
=i
=
2
2
2
= i cos z:
= u + iv =
=
eix
31
II.3.3
1 2 3 P
L
K
LLL
Show that if f and f are both analytic on a domain D, then f is
constant.
32
II.3.4
Show that if f is analytic on a domain D, and if jf j is constant,
then f is constant.
Hint. Write f = jf j2 =f .
Solution (with use of hint)
If f (z) = 0 for some z 2 D. Then f 0 on D, since jf j is constant. Assume
that f (z) 6= 0 for every z 2 D. Since f is analytic in D we have that
1=f and f = jf j2 =f (where jf j is constant) are analytic. From this follows
that Re f and Im f are analytic and real-valued and therefore constant. So
f = Re f + i Im f is constant.
33
II.3.4
Show that if f is analytic on a domain D, and if jf j is constant,
then f is constant.
Hint. Write f = jf j2 =f .
Solution
Set f = u + iv, where u and v are realvalued functions. Now suppose that
jf j = k is constant. We have that u2 + v 2 = k 2 . Dierentiate both sides with
respect to x respectively y
0
uux + vvx = 0;
0
0
uuy + vvy = 0:
0
(1)
0
uux vuy = 0;
0
0
uuy + vux = 0:
(2)
(u2 + v 2 ) ux = 0;
0
(u2 + v 2 ) uy = 0:
(3)
0 = ux = v y ;
0
0
0 = uy = v y ;
thus by (4) f is constant.
34
(4)
II.3.5
If f = u + iv is analytic, then jruj = jrvj = jf 0 j.
Solution (K. Seip)
Because f = u + iv is analytic. Then
ru =
@u @u
;
@x @y
@u
;
@x
@v
@x
35
@v
;
@y
@v
@x
II.3.6
If f = u + iv is analytic on D, then rv is obtained by rotating ru
by 90 . In Particular, ru and rv are orthogonal.
Solution (A. Kumjian)
Let 2 R then rotation of a vector (x; y) 2 R2 by the angle
origin) is given by matrix multiplication:
x
y
x
y
7! A
cos
sin
sin
cos
x
y
(about the
x cos
y sin
x sin + y cos
@v @v
;
@x @y
@u @u
;
@y @x
=A
=2 ru:
@v
@u
=
@x
@y
and
@v @v
Since ru = @u
; @u and rv = @x
; @y
@x @y
rv rv = 0. To prove that we use (0.6).
@u
=
@y
@v
@x
@v @u @v @u
@v @v @v @v
+
=
=0
@x @x @y @y
@x @y @y @x
We conclude that ru and rv are orthogonal.
Notice that the operation that sends ru ! rv is given by the counter-clockwise rotation matrix with = 2 . Keeping in mind (0.6), one can write the
following.
ru rv =
rv =
vx
vy
0
1
1 0
vy
vx
=
36
cos 2
sin
sin 2 cos 2
vy
vx
=R
ru
Remark.
We have that
vx =
@v
;
@x
vy =
@v
;
@y
ux =
@u
;
@x
and
R ( =2) = A
37
=2 :
uy =
@u
;
@y
II.3.7
Sketch the vector elds ru and rv for the following functions f = u + iv.
(a) iz
(b) z 2
(c) 1=z
Solution
II.3.7a
II.3.7a
y
II.3.7a
y
-2
-2
-4
-2
-2
-4
4
2
2
-4
-4
-2
2
-2
-4
-4
(a)
f (z) = iz =
u= y
)
v=x
y + ix )
(b)
f (z) = z 2 = x2
ru = (0; 1)
rv = (1; 0)
u = x2 y 2
)
v = 2xy
y 2 + 2ixy )
ru = (2x; 2y)
rv = (2y; 2x)
(c)
f (z) =
cos
x
u = x2 +y
1
z
x iy
2 =
r
= 2 = 2
)
y
2
v
=
=
z
x
+
y
2
2
jzj
x +y
8
2
2
2
2
@u
2
>
= (xy2 +yx2 )2 = sin r2cos = cos
>
@x
r2
>
>
2xy
2 cos sin
< @u =
= sin2 2
2 =
2
>
>
>
>
:
@y
@v
@x
@v
@y
(x2 +y 2 )
2xy
(x2 +y 2 )2
y 2 x2
(x2 +y 2 )2
ru =
rv =
1
r2
1
r2
r
2 cos sin
r2
sin2
cos2
r2
r
sin 2
2
r
2
= cos
r2
( cos 2 ; sin 2 )
( sin 2 ; cos 2 )
38
sin
r
Solution
Set x = r cos and y = r sin , then take the partial derivative,
@x
@r
@y
@r
= cos
= sin
@x
@
@y
@
= r sin
= r cos
Use Cauchy-Riemanns equations and the derivative to get the rst equation
@u
@u @x @u @y
@u
@u
@v
@v
=
+
=
cos +
sin =
cos
sin =
@r
@x @r @y @r
@x
@y
@y
@x
1
@v
@v
1 @v @x @v @y
1 @v
=
r
sin + r cos
=
+
=
;
r
@x
@y
r @x @
@y @
r@
Use Cauchy-Riemanns equations and the derivative to get the second equation
@u @x @u @y
@u
@u
@v
@v
@u
=
+
= r
sin + r
cos = r sin
r
cos =
@
@x @
@y @
@x
@y
@y
@x
@v
@v
@v @x @v @y
@v
= r
cos +
sin
= r
+
= r :
@x
@y
@x @r @y @r
@r
we get
@u
@r
@u
@
=
=
1 @v
;
r@
r @v
:
@r
@u
@ m
=
(r cos (m )) = mrm 1 cos (m ) =
@r
@r
@u
@ m
=
(r cos (m )) = mrm sin (m ) =
@
@
40
1 m
1 @ m
1 @v
r m cos (m ) =
(r sin m ) =
;
r
r@
r@
@
@v
rmrm 1 sin (m ) = r (rm sin (m )) = r :
@r
@r
II.4.1
Sketch the gradient vector elds ru and rv for
(a) u + iv = ez
(b) u + iv = Log z
Solution
a)
We have
u + iv = ez = ex cos y + iex sin y
and get
ru = (ex cos y; ex sin y) = ex (cos y;
sin y)
ru =
41
II.4.2
Let a be a complex number a 6= 0, and let f (z) be an analytic branch
of z a on Cn ( 1; 0]. Show that f 0 (z) = af (z) =z. (Thus f 0 (z) = az a 1 ,
where we pick the branch of z a 1 that corresponds to the original
branch of z a divided by z.)
Solution
We may write the branch as follows
f (z) = z a = ea log z = ea(Log z+2
where c = e2
iam
im)
= ea Log z+2
iam
= cea Log z ;
42
af (z)
a
=
:
z
z
II.4.3
p
Consider the branch of f (z) = z (1 z) on Cn [0; 1] that has positive imaginary part at z = 2. What is f 0 (z)? Be sure to specify the
branch of the expression for f 0 (z).
Solution p
Set f (z) = z (1
f 0 (z) =
d
(z (1
z) dz
z)) =
1 1 2z
p
:
2 z (1 z)
Branch of f (z) is i increasing for x > 0 large, thus f 0 (z) is i positive for
x > 0 large.
p
Use branch of f 0 (z) = z (1 z) that it is i positive, i.e., same on f (z).
f 0 (z) =
1 1 2z
:
2 f (z)
Solution (D. p
Jakobsson)
Given f (z) = z (1 z), we can dene w = f (z) and get w2 = z (1
d
(1
dz
1
z) z =
2z = 2w
z).
d 2
w
dz
dw
dz
One gets
dp
z (1
dz
1 2z
z) = p
2 z (1 z)
and is dened on Cn [0; 1]. Take the principal branch with the positive imaginary part.
43
II.4.4
Recall that the principal branch of the inverse tangent function
was dened on the complex plane with two slits on the imaginary
axis by
Tan
1
2i
Log
1+iz
1 iz
Solution
We have that
d
Tan
dz
z=
1
1
i
2i 1 + iz
1
1
1
iz
( i) =
1 1 iz + 1 + iz
1
:
=
2 (1 + iz) (1 iz)
1 + z2
44
II.4.5
p
Recall that cos 1 (z) = i log z
z 2 1 . Suppose g (z) is an ana1
lytic branch of cos (z), dened on a domain D. Find g 0 (z). Do
dierent branches of cos 1 (z) have the same derivative?
Solution
We have that
d
cos
dz
z=
i
1 2
1=2
1
z
1
2z =
2
2
z
z
1
i
z
i
p
=
1 p
=p
=
2
2
2
z
z
1
z
1
z
1
=
=p
45
1
1
z2
II.4.6
Suppose h (z) is an analytic branch of sin 1 (z), dened on a domain
D. Find h0 (z). Do dierent branches of sin 1 (z) have the same
derivative?
h1 (0) =
1
=1
cos 0
and
1
h2 (0) =
1
cos
1:
46
II.4.7
Let f (z) be a bounded analytic function, dened on a bounded
domain D in the complex plane, and suppose that f (z) is one-oneone. Show that the area of f (D) is given by
ZZ
2
Area (f (D)) =
jf 0 (z)j dxdy:
D
by the change of variables (u (x; y) ; v (x; y)) ! (x; y), and since det Jf =
jf 0 (z)j2 , we get
ZZ
2
jf 0 (z)j dxdy:
Area (f (D)) =
D
47
II.4.8
Sketch the image of the circle fjz
Compute the area of the image.
1j
Solution
II.4.8 z-plane
y
II.4.8 w-plane
y
-4
-2
2
-2
-4
-2
-2
-4
-4
Set t = x
ZZ
t2 +y 2 61
1, get
t2 + 2t + 1 + y 2 dtdy =
2
6
= 44
ZZ
t2 + y 2 dtdy + 0 +
t2 +y 2 61
ZZ
t2 +y 2 61
7
dtdy 5 =
=4
=6 :
Area (f (D)) =
ZZ
Z
2
0
=
jf (z)j dxdy =
=
p
2Z
1
p
(x 1)
Z p1
p
4 x2 + y 2 dx dy =
1 (x 1)2
(x 1)2
1 (x 1)
2
4 jx + iyj2 dx dy =
p
8x2 2x
x2 +8=3 2x
x2
3=2
dx =
=6
Change of variables
x = 1 + sin t
dx = cos tdt
Z
=2
8 (1 + sin t)2
=2
=2
sin2 t +
8
1
3
sin2
!
3=2
cos tdt =
8
4
2
2
8 1 + 2sin
|{z}t + sin t cos + 3 cos tdtdt =
=2
odd
Z =2
2
= 16
2 cos2
cos4 tdt =
3
0
Z =2
1
=8
2 (1 + cos 2t)
1 + 2 cos 2t + cos2 2t dt =
3
0
Z =2
5 4
1
=8
+ cos 2t
(1 + cos 4t) dt =
3 3
3
0
5 1
9
=8
=4
=6 :
2 3 6
6
=
49
II.4.9
Compute
ZZ
jf 0 (z)j dxdy;
for f (z) = z and D the open unit disk fjzj < 1g. Interpret your
answer in terms of areas.
Solution
By the result above
ZZ
ZZ
2
2
0
jf (z)j dxdy =
jf 0 (z)j dx dy = ;
jzj<1
x>0
jzj<1
x<0
thus
ZZ
jf 0 (z)j dx dy = 2 :
jzj<1
Remark.
The function maps the top and bottom halves of the unit disk one-to-one
onto the unit disk, so by Exersice II.4.7 the integrals on top and bottom
halves are each .
50
II.4.10
1 2 3 P L K
For smooth functions g and h dened on a bounded domain U , we dene the
Dirichlet form DU (g; h) by
ZZ
@g @h @g @h
+
dx dy
@y @y
U @x @x
Show that if z = f ( ) is a one-to-one analytic function from the bounded
domain V onto U , then
DU (g; h) =
DU (g; h) = DV (g f; h f ) :
Remark. This shows that the Dirichlet form is a "conformal invariant".
Solution RR
Du (g; h) =
rgrhdxdy Assume h; g ????
U
RR
Du (g; h) =
r (g f ) r (h f )dudv =
U
RR
g (u (x; y) ; v (x; y)) h (u (x; y) ; v (x; y)) dudv
U
@g(u(x;y);v(x;y)) @h(u(x;y);v(x;y))
+
@x
@x
@g(u(x;y);v(x;y)) @h(u(x;y);v(x;y))
=
@y
@y
@g @u
@g @v
@g @u
@g @v
@h @u
@v
@h @u
@v
+ @v
+ @h
+ @u
+ @v
+ @h
@u @x
@x
@u @x
@v @x
@y
@y
@u @y
@v @y
@g @u @h @v
@g @v @h @u
@g @h @v 2
@g @h @u 2
+ @u
+ @v
+ @v
+
@u @u @x
@x @v @x
@x @u @x
@v @x
2
2
@g @h @u
@g @v @h @u
@g @h @v
@g @u @h @v
+ @v
+ @v
+ @u
=
@u @u @y
@y @v @y
@y @u @y
@v
@y
2
2
@g @h @v 2
@g @h @u
@g @h @v
@g @h @u 2
+
+
+
=
@u @u @x
@v @v @x
@v @v
@y
! @u @u @y
!
2
2
2
2
@u
@u
@v
@v
@g @h
@g @h
+
+ @v
+
+
@u @u
@v
@x
{z
jf 0 (z)j2
@y
@x
{z
jf 0 (z)j2
@y
@g @u @h @v @g @v @h @u @g @u @h @v @g @v @h @u
+
+
+
@u @x @v @x @v @x @u @x @u @y @v @y @v @y @u @y
|
{z
}
Use C-R equations
RR
RR
rgrh jf 0 (z)j2 dxdy =
rgrhdudv
U
51
II.5.1
Show that the following functions are harmonic, and nd its harmonic conjugates:
(a) x2 y 2
(c) sinh x sin y
(e) tan 1 (y=x) ; x > 0
2
2
(b) xy + 3x2 y y 3 (d) ex y cos (2xy) (f) x= (x2 + y 2 )
Solution
(a)
Set
u (x; y) = x2
y2 )
ux = 2x
)
0
uy = 2y
00
uxx = 2
) 4u = 0;
00
uyy = 2
u (x; y) = xy+3x2 y y 3 )
ux = y + 6xy
0
uy = x + 3x2
00
3y 2
uxx = 6y
) 4u = 0;
0
uyy = 6y
vx = uy = x 3x2 + 3y 2 ) v (x; y) =
0
0
vy = ux = y + 6xy
Thus we have that
52
x2
2
y2
g (x) = y ) g (y) =
+ C;
2
0
thus
x2
y2
+ 3xy 2
x3 + C:
2
2
In paranthesis we remark that the analytic function f (z) = u + iv is given
by
v (x; y) =
f (z) = xy +3x2 y y 3 +i
y2
+ 3xy 2
2
x2
2
x3 + C
iz 2 (z + 1=2)+iC:
(c)
Set
0
ux = cosh x sin y
)
0
uy = sinh x cos y
00
cosh x cos y + C
53
u (x; y) = ex
00
y2
cos (2xy) )
uxx = 2ex y (cos 2xy + 2x2 cos 2xy 2y 2 cos 2xy 4xy sin 2xy)
) 4u = 0;
0
2
2
uyy = 2ex y (cos 2xy + 2x2 cos 2xy 2y 2 cos 2xy 4xy sin 2xy)
y2
g 0 (y) = 0 ) g (y) = C;
thus
v (x; y) = ex
y2
sin (2xy) + C
y2
cos (2xy) + i ex
(e)
54
y2
Set
(y=x) ; x > 0 )
8
1
0
>
< ux =
1 + (y=x)2
=)
x
>
: u0y = 2
x + y2
u (x; y) = tan
y
x2
x2
y
+ y2
=)
8
00
>
>
< uxx =
0
>
>
: uyy
2xy
(x2 + y 2 )2
) 4u = 0;
2xy
=
(x2 + y 2 )2
x
) v (x; y) = a
x2 + y 2
y
0
0
>
: v y = ux = 2
x + y2
0
uy =
1
2
y
+ g 0 (y)
2
2
x +y
g 0 (y) = 0 ) g (y) = C;
thus
v (x; y) =
1
log x2 + y 2 + C:
2
(f)
Set
u (x; y) =
x2
8
0
>
>
< ux =
x
)
0
>
+ y2
>
: uy =
8
3y 2 x2
x2 y 2
00
>
>
u
=
2x
< xx
(x2 + y 2 )3
(x2 + y 2 )2 )
) 4u = 0;
2xy
3y 2 x2
0
>
>
: uyy = 2x 2
(x2 + y 2 )2
(x + y 2 )3
55
8
0
>
>
< vx =
y
2xy
0
uy =
+ g (y) ) vy =
2 ) v (x; y) =
2
2
2
2
(x + y )
(x + y )
2
2
x
y
0
0
>
>
: v y = ux =
(x2 + y 2 )2
0
x2 y 2
0
2 + g (y)
2
2
(x + y )
g 0 (y) = 0 ) g (y) = C;
thus
v (x; y) =
(x2
y
+C
+ y2)
f (z) =
x
+i
x2 + y 2
y
+C
(x2 + y 2 )
56
z
z
1
+ iC = + iC:
2 + iC =
zz
z
jzj
II.5.2
Show that if v is a harmonic conjugate for u, then
conjugate for v.
u is a harmonic
Solution
If f (z) = u (x; y) + iv (x; y) is analytic then if (z) = v (x; y)
analytic. Thus u is a harmonic conjugate of v.
57
iu (x; y) is
II.5.3
Dene u (z) = Im (1=z 2 ) for z 6= 0, and set u (0) = 0.
(a) Show that all partial derivatives of u with respect to x exist
at all points of the plane C, as do all partial derivative of u with
respect to y.
2
2
(b) Show that @@xu2 + @@yu2 = 0.
(c) Show that u is not harmonic on C.
@2u
(d) Show that @x@y
does not exist at (0; 0).
Solution
(a)
We rst dene
8
<
z2
(x iy)2
2xy
1
=
(x; y) 6= (0; 0)
Im 2 = Im 4 = Im
2
2
2
2
u (z) =
z
jzj
(x + y )
(x + y 2 )2
:
0
(x; y) = (0; 0)
(x;y)
For all (x; y) 6= (0; 0), the n
th derivative of u (z) will be (x2f+y
2 )n+2 with
f (x; y) a polynomial of degree n + 2 in (x; y). At (0; 0), the function itself is
0, thus u (x; y) 2 C1 (x). By symmetry u (x; y) 2 C1 (y).
(b)
We have
@u
@2u @2u
+ 2 =
2
@x
@y
@x
2y
3x2 y 2
(x2 + y 2 )3
=
3y 2 x2
=
(x2 + y 2 )3
x2 y 2
x2 y 2
+
24xy
=0
24xy
(x2 + y 2 )4
(x2 + y 2 )4
@u
@y
2x
(c)
Notice that the function u (x; y) is not harmonic because the function itself,
its rst and second derivative are not continuous at the origin. The limit of
u (x; y) as (x; y) ! (0; 0) does not exist, for instance u n1 ; n1 goes to 1 as
n ! 1 whereas we dened the function u to be 0 at the origin.
lim u
n!1
1 1
;
n n
= lim
n!1
58
n2
=
2
1:
(d)
@u
=
@y
2x3 + 6xy 2
2x (x2 + y 2 ) + 2xy 2 (x2 + y 2 ) 2y
=
(x2 + y 2 )4
(x2 + y 2 )3
thus
3
@2u
( 6x2 + 6y 2 ) (x2 + y 2 )
( 2x3 + 6xy 2 ) 3 (x2 + y 2 )
=
@x@y
(x2 + y 2 )6
2x
6x4
We have
@2u
=
@x@y
@2u
(x; 0) =
@x@y
(x; y) 6= (0; 0)
(x; y) = (0; 0)
6=x4 ; x 6= 0
:
0;
x=0
59
@2u
@x@y
36x2 y 2 + 6y 4
:
(x2 + y 2 )4
II.5.4
Show that if h (z) is a complex-valued harmonic function (solution
of Laplaces equation) such that zh (z) is also harmonic, then h (z)
is analytic.
Solution
First partial derivative for zh (z) is,
@z
@h
@ (zh)
=
h+z ;
@x
@x
@x
@z
@h
@ (zh)
=
h+z :
@y
@y
@y
Second derivative for zh (z) is,
@ 2 (zh)
@2z
@z @h @z @h
@2h
=
h
+
+
+
z
;
@x2
@x2
@x @x @x @x
@x2
@2z
@z @h @z @h
@2h
@ 2 (zh)
=
h
+
+
+
z
:
@y 2
@y 2
@y @y @y @y
@y 2
We have z = x+iy, thus @z=@x = 1, @z=@y = i, and @ 2 z=@x2 = @ 2 z=@y 2 = 0,
therefore
@ 2 (zh)
@h
@2h
=
2
+
z
;
@x2
@x
@x2
@ 2 (zh)
@h
@2h
=
i2
+
z
:
@y 2
@y
@y 2
Add the equations
@h
@h
@ 2 (zh) @ 2 (zh)
+
=2
+ 2i
+z
2
2
@x
@y
@x
@y
@2h @2h
+
@x2 @y 2
Because h (z) and zh (z) is harmonic, then both @ 2 (zh) =@x2 +@ 2 (zh) =@y 2 =
0 and @ 2 (h) =@x2 + @ 2 (h) =@y 2 = 0, we have
@h
=
@x
60
@h
:
@y
@u
@v
+i
@y
@y
@v
@y
@u
:
@y
@v
@y
and
@u
@y
@v
:
@x
61
II.5.5
Show that Laplaces equation in polar coordinates is
1 @2u
@ 2 u 1 @u
+
= 0:
+
@r2
r @r r2 @ 2
Solution
Let x = r cos and y = r sin , and get partial derivatives
@x
= cos ;
@r
@y
= sin ;
@r
Taking the rst derivative of u with
@x
= r sin ;
@
@y
= r cos :
@
respect to r
@u
@u @x @u @y
@u
@u
@u
@u
=
+
=
cos +
sin = cos
+ sin
:
@r
@x @r @y @r
@x
@y
@x
@y
Taking the second derivative of u with respect to r
@2u
=
@r2
@ 2 u @x
@ 2 u @y
@ 2 u @x
@ 2 u @y
cos
+
cos
+
sin
+
sin =
@x2 @r
@y@x @r
@x@y @r
@y 2 @r
@ 2 u @x
@2u
@2u
@ 2 u @y
=
cos
+
sin
cos
+
cos
sin
+
sin =
@x2 @r
@y@x
@x@y
@y 2 @r
2
@2u
@2u
2 @ u
= cos2
+
2
sin
cos
+
sin
:
@x2
@x@y
@y 2
=
62
@2u
=
@ 2
@ 2 u @y
@u
@ 2 u @x
(
r
sin
)
+
( r sin ) +
( r cos ) +
2
@x @
@y@x @
@x
@ 2 u @x
@ 2 u @y
@u
+
(r cos ) + 2
(r cos ) +
( r sin ) =
@x@y @
@y @
@y
@2u
@u
@2u
=
(r cos ) ( r sin ) +
( r cos ) +
(
r
sin
)
(
r
sin
)
+
2
@x
@y@x
@x
@2u
@2u
@u
+
( r sin ) (r cos ) + 2 (r cos ) (r cos ) +
( r sin ) =
@x@y
@y
@y
2
@2u
@2u
@u
@u
2
2
2 @ u
= r2 sin2
2r
sin
cos
+
r
cos
r cos
r sin
:
2
2
@x
@x@y
@y
@x
@y
=
@2u
@2u 1
@u
@u
+ sin2
+
cos
+ sin
+
2
@x@y
@y
r
@x
@y
@2u
@2u
@u
@u
cos
+ r2 cos2
r cos
r sin
2
@x@y
@y
@x
@y
2
2
@ u
@ u 1
@u 1
@u
+ sin2
+ cos
+ sin
+
2
@x@y
@y
r
@x r
@y
@2u
@2u 1
@u 1
@u
+ cos2
cos
sin
=
2
@x@y
@y
r
@x r
@y
@2u
@2u
@2u @2u
2
2
+
sin
+
cos
=
+
:
+ cos2
@x2
@y 2
@x2 @y 2
63
II.5.6
Show using Laplaces equation in polar coordinates that log jzj is
harmonic on the punctured plane Cn f0g.
Solution
Set z = rei , we have u (r; ) = log jzj = log r. We can compute the Laplacian
of log jzj without worrying about the origin, i.e. r 6= 0 because this case is
taken out, thus we have
log r =
1 @u2
@ 2 u 1 @u
+
+
=
@r2
r @r r2 @ 2
64
1
11
+
+ 0 = 0:
r2 r r
II.5.7
Show that log jzj has no conjugate harmonic function on the punctured plane Cn f0g, though it does have a conjugate harmonic function on the slit plane Cn ( 1; 0).
Solution (A. Kumjamin)
Set D := Cn ( 1; 0] and let u : D ! R be given by u (z) := log jzj, it is
easily veried that u (z) = Re Log z and this shows that u is harmonic on D
by the main theorem in section II.5. Moreover, it follows by Cauchy-Riemann
equations that a harmonic conjugate v is given by
v (z) := Im Log z = Arg z
for
z 2 D:
y!00+
+ c;
and
lim v (x; y) =
y!00
65
+ c:
II.5.8
Show using Laplaces equation in polar coordinates that u rei =
log r is harmonic. Use the polar form of the Cauchy-Riemann
equations. (Exercise 3.8) to nd a harmonic conjugate v for u.
What is the analytic function u + iv?
Solution
Set
u rei
= log r )
ur =
r
u = log r
0
00
urr =
u
00
=0
r2
1
1
u = urr + ur + 2 u =
+
= 0:
r
r
r2 r2
Cauchy Riemann equations in polar form (page 50) gives us
8
<
1 0
vr =
u =
: v 0 = rur0 =
r
0
log r
) v (r; ) =
r
(log r)2
0
+ g ( ) ) v = g0 ( )
2
g0 ( ) =
) g( ) =
+ C;
thus
2
(log r)2
+
+ C:
2
2
We have that the analytic function f (z) = u + iv is given by
v (r; ) =
(log r)2
+C
2
66
i
(log r + i )2 +C =
2
i
(log z)2 +iC:
2
II.6.1
Sketch the families of level curves of u and v for the following
functions f = u + iv.
(a) f (z) = 1=z
(b) f (z) = 1=z 2
(c) f (z) = z 6
Determine where f (z) is conformal and where it is not conformal.
Solution
(a)
We have that
8
< u=
x
2
z
x iy
1
x + y2
)
f (z) = = 2 = 2
y
: v=
z
x + y2
jzj
2
x + y2
67
cos(2 )
r2
sin(2 )
r2
1
1
u= ;
r2 = c cos (2 ) ;
v= ;
c
c
Conformal everywhere except at z = 0.
(c)
We have that
r2 =
c sin (2 ) :
u = r6 cos (6 )
v = r6 sin (6 )
68
II.6.2
Sketch the families of level curves of u and v for f (z) = Log z = u+iv.
Relate your sketch to a gure in Section I.6.
Solution
II.6.2
3/4
/2
/4
3/4
/2
/4
We have
u = log jzj
)
v = Arg z
u = log r
:
v=
69
II.6.3
Sketch the families of level curves of u and v for the function f =
u + iv given by (a) f (z) = ez , (b) f (z) = e z , where
is complex.
Determine where f (z) is conformal and where it is not conformal.
Solution
f (z) = ez = ex (cos y + i sin y) ) u = ex cos yv = ex sin y
u = const, x = log (c cos y) = log c log (cos y)
v = const, x = log (c sin y) = log c log (sin y)
Level curves are invariant under ????? in xdirection. Level curves are invariant under ????? in ydirection by . The function f (z) is conformal
everywhere.
f (z) = e z = e(a+ib)(x+iy) = eax by (cos (ay + bx) + i sin (ay + bx)) )
u = eax by cos (ay + bx)
v = eax by sin (ay + bx)
Figure repeat themselves
u = 0 at ay + bx = n + 2 , y = ab x + na + 2a . v = 0 at ay + bx = n ,
y = ab x + na
FIGURE II.6.3a NF FIGURE II.6.3b NF
70
II.6.4
Find a conformal map of the horizontal strip f A < Im z < Ag onto
the right half-plane fRe w > 0g. Hint. Recall the discussion of the
exponential function, or refer to the preceding problem.
e 2A
2A
for all z 2 C. Hence, h is analytic with a nowhere vanishing derivative and
so h is conformal. Note that the exponential functions maps the horizontal
strip fz 2 C :
=2 < Im z < =2g onto the right half-plane
fw 2 C : Re w > 0g = fw 2 C :
Hence, h maps the horizontal strip fz 2 C :
half-plane fw 2 C : Re w > 0g.
71
II.6.5
Find a conformal map of the wedge f B < arg z < Bg onto the right
half-plane fRe w > 0g. Assume 0 < B < .
Solution
Try w = z , multiply angels by
2B
, so eiB ! ei
=2
,w=z
72
=(2B)
II.6.6
Determine where the function f (z) = z +1=z is conformal and where
it is not conformal. Show that for each w, there are at most two
values z for which f (z) = w. Show that if r > 1, f (z) maps the circle
fjzj = rg onto an ellipse, and that f (z) maps the circle fjzj = 1=rg
onto the same ellipse. Show that f (z) is one-to-one on the exterior
domain D = fjzj > 1g. Determine the image of D under f (z). Sketch
the images under f (z) of the circles fjzj = rg for r > 1, and sketch
also the images of the parts of the rays farg z = g lying in D.
Solution
f (z) = z + 1=z, f 0 (z) = 1 1=z 2 , f 0 (z) = 0 at z 2 = 1, z =
at z = 0, not conformal at z = 1. If f (z) = , then z =
1.pNot dened
2
4 =2.
73
II.6.7
For the function f (z) = z + 1=z = u + iv, sketch the families of level
curves of u and v. Determine the images under f (z) of the top
half of the unit disk, the bottom half of the unit disk, the part
of the upper half-plane outside the unit disk, and the part of the
lower half-plane outside the unit disk. Hint. Start by locating
the images of the curves where u = 0, where v = 0, and where
v = 1. Note that the level curves are symmetric with respect to the
real and imaginary axis, and they are invariant under the inversion
z 7! 1=z in the unit circle.
Solution
74
75
II.6.9
Let f = u + iv be a continuously dierentiable complex-valued function on a domain D such that the Jacobian matrix of f does not
vanish at any point of D. Show that if f maps orthogonal curves to
orthogonal curves, then either f or f is analytic, with nonvanishing
derivative.
Solution
Suppose f (0) = 0 and ux 6= 0 at 0. The tangents in the orthogonal directions
(1; t) and ( t; 1) are mapped to the tangents in the directions (ux ; vx ) +
t (uy ; vy ) and t (ux ; vx ) + (uy ; vy ).
The orthogonality of these directions for all t gives
h(ux ; vx ) + t (uy ; vy ) ; t (ux ; vx ) + (uy ; vy )i = 0;
can be rewritten
u2y
ux u y + v x v y = 0
+ vy2 (u2x + vx2 ) = 0
most hold to get orthogonality. Put (1) into (2) and get
u2x u2y + vy2
u2x
= vx2 vy2
u4x
u2x vx2 =
vy2
u2x = 0:
76
II.7.1
Compute explicitly the fractional linear transformations determined
by the following correspondences of triples
(a) (1 + i; 2; 0) 7 ! (0; 1; i 1)
(e) (1; 2; 1) 7 ! (0; 1; 1)
(b) (0; 1; 1) 7 ! (1; 1 + i; 2)
(f) (0; 1; i) 7 ! (0; 1; 1)
(c) (1; 1 + i; 2) 7 ! (0; 1; 1)
(g) (0; 1; 1) 7 ! (0; 1; i)
(d) ( 2; i; 2) 7 ! (1 2i; 0; 1 + 2i) (h) (1; i; 1) 7 ! (1; 0; 1)
Solution
(w
(w
(z
w2 )
=
w0 )
(z
w0 ) (w1
w2 ) (w1
z0 ) (z1
z2 ) (z1
z2 )
z0 )
(a)
We have the points
z0 = 1 + i
z1 = 2
z2 = 0
w0 = 0
w1 = 1
w2 = i 1
The mapping is
(w
(w
w0 ) (1
w2 ) (1
w2 =w1 )
(z
=
w0 =w1 )
(z
z0 ) (z1
z2 ) (z1
z2 )
;
z0 )
0
(i
1)
(z
(z
(1 + i)) (2 0)
2iz + (2 2i)
)w=
:
0) (2 (1 + i))
z 2
(b)
We have the points
z0 = 0
z1 = 1
z2 = 1
w0 = 1
w1 = 1 + i
w2 = 2
The mapping is
(w
(w
w0 ) (w1
w2 ) (w1
w2 )
(z z0 ) (z1 =z2 1)
=
;
w0 )
(z=z2 1) (z1 z0 )
(w
(w
1) ((1 + i)
2) ((1 + i)
2)
z
=
1)
1
0
2z
)w=
0
z
(1 + i)
:
(1 + i)
(c)
We have the points
z0 = 1
z1 = 1 + i
z2 = 2
w0 = 0
w1 = 1
w2 = 1
The mapping is
(z=z0 1) (z1 z2 )
(w w0 ) (w1 =w2 1)
=
;
(w=w2 1) (w1 w0 )
(z z2 ) (z1 =z0 1)
take limit as z0 ; w2 ! 1, to obtain
w
1
0
(1 + i) 2
i
=
)w=
0
z 2
z
1
:
2
(d)
We have the points
z0 = 2 w0 = 1 2i
z1 = i
w1 = 0
z2 = 2
w2 = 1 + 2i
The mapping is
(w
(w
w0 ) (w1
w2 ) (w1
w2 )
(z
=
w0 )
(z
z0 ) (z1
z2 ) (z1
z2 )
;
z0 )
obtain
(w
(w
(1 2i)) (0
(1 + 2i)) (0
(1 + 2i))
(z
=
(1 2i))
(z
( 2)) (i 2)
) w = iz + 1:
2) (i ( 2))
(e)
We have the points
z0 = 1
z1 = 2
z2 = 1
w0 = 0
w1 = 1
w2 = 1
78
The mapping is
(w w0 ) (w1 =w2 1)
(z z0 ) (z1 =z2 1)
=
;
(w=w2 1) (w1 w0 )
(z=z2 1) (z1 z0 )
take limit as z2 ; w2 ! 1, to obtain
w
1
0
z
=
0
2
1
)w=z
1
1:
(f)
We have the points
z0 = 0
w0 = 0
z1 = 1 w1 = 1
z2 = i
w2 = 1
The mapping is
(w w0 ) (w1 =w2 1)
(z
=
(w=w2 1) (w1 w0 )
(z
z0 ) (1
z2 ) (1
z2 =z1 )
;
z0 =z1 )
z
0
=
0
z
0
z
)w=
i
z i
(g)
We have the points
z0 = 0
z1 = 1
z2 = 1
w0 = 0
w1 = 1
w2 = i
The mapping is
(w
(w
w0 ) (1
w2 ) (1
w2 =w1 )
(z z0 ) (z1 =z2 1)
=
;
w0 =w1 )
(z=z2 1) (z1 z0 )
0
z
=
i
1
0
iz
)w=
:
0
z 1
(h)
79
w0 ) (w1
w2 ) (w1
(z
w2 )
=
w0 )
(z
z0 ) (z1
z2 ) (z1
z2 )
;
z0 )
obtain
(w
(w
1) (0 ( 1))
(z
=
( 1)) (0 1)
(z
1) (i ( 1))
iz + 1
)w=
:
( 1)) (i 1)
z+i
80
II.7.2
Consider the fractional linear transformation in Exercise 1a above,
which maps 1 + i to 0, 2 to 1, and 0 to i 1. Without referring to
an explicit formula, determine the image of the circle fjz 1j = 1g,
the image of the disk fjz 1j < 1g, and the image of the real axis.
Solution
It is obvious that the tripple (1 + i; 2; 0) lies on the circle jz 1j = 1 in the
z
plane. From exercise II.7.1a we know that points in the w plane the
tripple is mapped on, i.e,
z0 = 1 + i
z1 = 2
z2 = 0
w0 = 0
w1 = 1
w2 = i 1
II.7.2 z-plane
II.7.2 w-plane
81
II.7.3
Consider the fractional linear transformation that maps 1 to i, 0
to 1 + i, 1 to 1. Determine the image of the unit circle fjzj = 1g,
the image of the open unit disk fjzj < 1g, and the image of the
imaginary axis. Illustrate with a sketch.
Solution
We have the points
z0 = 1
w0 = i
z1 = 0
w1 = 1 + i
z2 = 1 w2 = 1
The mapping is
(w
(w
w0 ) (w1
w2 ) (w1
(z
w2 )
=
w0 )
(z
z0 ) (z1
z2 ) (z1
z2 )
;
z0 )
obtain
(w
(w
i) ((1 + i)
1) ((1 + i)
1)
(z
=
i)
(z
1) (0 ( 1))
i 1
)w=
( 1)) (0 1)
z+i
82
II.7.3 z-plane
II.7.3 w-plane
83
II.7.4
Consider the fractional linear transformation that maps 1 to i,
1 to 2i, and i to 0. Determine the image of the unit circle fjzj = 1g,
the image of the open unit disk fjzj < 1g, and the image of the
interval [ 1; +1] on the real axis. Illustrate with a sketch.
Solution
We have the points
z0 = 1 w0 = i
z1 = 1
w1 = 2i
z2 = i
w2 = 0
The mapping is
(w
(w
w0 ) (w1
w2 ) (w1
(z
w2 )
=
w0 )
(z
z0 ) (z1
z2 ) (z1
z2 )
;
z0 )
obtain
(w
(w
( i)) (2i 0)
(z
=
0) (2i ( i))
(z
( 1)) (1 i)
( 6 2i) z 2 + 6i
)w=
i) (1 ( 1))
5z 3 + 4i
84
II.7.4 z-plane
II.7.4 w-plane
85
II.7.5
What is the image of the horizontal line through i under the fractional linear transformation that interchanges 0 and 1 and maps 1
to 1 + i? Illustrate with a sketch.
Solution
We have the points
z0 = 0
w0 = 1
z1 = 1
w1 = 0
z2 = 1 w2 = 1 + i
The mapping is
(w
(w
w0 ) (w1
w2 ) (w1
(z
w2 )
=
w0 )
(z
z0 ) (z1
z2 ) (z1
z2 )
;
z0 )
obtain
(w
(w
1) (0 (1 + i))
(z
=
(1 + i)) (0 1)
(z
0) (1 ( 1))
iz
)w=
( 1)) (1 0)
z
i
i
The tripple (0; 1; 1) on the real axis is mapped to the points (1; 0; 1 + i) in
the w plane, it must be so pthat the real axis in the z plane is mapped onto
1
+ 12 i = 22 in the w plane, since tree points determines
the circle w
2
the "circle". The real axis and the line through i have one common point at
innity. The common point at innity for both real axis and the line through
i must be mapped to the same point in the w plane, we have that the point
is w (1) = i. The two lines in the z plane is parallell, thus the images of
the lines must go through i in the w plane and be parallell in that point,
thus the mapping of the horizontal line through i must be the line through
i, which is tangent line to circle at i, i.e. Im z = Re z + 1. Remark that the
image can not be a circle because w (i) = 1, so the image of the horizontal
line must contain the point at innity.
86
II.7.5 z-plane
II.7.5 w-plane
87
II.7.6
Show that the image of a straight line under the inversion z 7! 1=z
is a straight line or circle, depending on whether the line passes
through the origin.
Solution
The image of the straight line ax+by = c is (kontrollera x iy i lsningshftet)
w=
where u = x=r2 and v =
au
bv =
1
z
1
=
= 2 = (x
z
zz
jzj
iy)
1
r2
y=r2 . Get
ax + by
c
c
= 2 = 2
= c u2 + v 2
2
r
r
x + y2
au + bv = 0.
au + bv = 0;
this is a straight line through 0.
If c 6= 0 we have
c u2 + v 2
au + bv = 0;
a
2c
it is a circle through 0.
+ v+
b
2c
0s
=@
a
2c
b
2c
12
A ;
88
II.7.7
Show that the fractional linear transformation f (z) = (az + b) = (cz + d)
is the identity mapping z if and only if b = c = 0 and a = d 6= 0.
Solution
We have that
az + b
= z , az + b = cz 2 + dz , cz 2 + (d a) z b = 0:
cz + d
Put in values on z for intance 0 and 1 the expession becomes
(1) c (d a) = 0;
(2)
b=0
(3) c + (d a) = 0;
z = 1:
z=0
z = 1:
89
II.7.8
Show that any fractional linear transformation can be represented
in the form f (z) = (az + b) (cz + d), where ad bc = 1. Is this representation unique?
Solution
If
w = ( z + )( z + )
divide each coe cient by the square root of
bc = 1:
90
1.
II.7.9
Show that the fractional linear transformations that are real on
the real axis are precisely those that can be expressed in the form
(az + b) = (cz + d), where a; b; c;and d are real.
91
II.7.9
Show that the fractional linear transformations that are real on
the real axis are precisely those that can be expressed in the form
(az + b) = (cz + d), where a; b; c;and d are real.
f (1) = lim
a=c if
1 if
c 6= 0;
c = 0:
Hence, f (R ) R .
Conversely, suppose that f (R ) R , we must show that f = fA for some 2
2 matrix with real entries. If A is a 2 2 matrix (with non-zero determinant),
then fA 1 = (fA ) 1 . Hence, f = fA for some 2 2 matrix A with real entries
i its inverse f 1 has the same property.
By the unlikeness part of the theorem on page 64, f is completely determined
by the three extended real numbers x0 = f (0), x1 = f (1) and x1 = f (1)
(recall that we have assumed f (R )
R ). So it su ces to show that
the fractional linear transformation g for which g (x0 ) = 0, g (x1 ) = 1 and
g (x1 ) = 1 has the requisite properties, because we must have g = f 1
(again by the uniqueness part of the theorem).
There are four cases to consider. The rst case is when x0 ; x1 ; x1 6= 1 and
the remaining three cases are x0 = 1; x1 = 1 and x1 = 1. The formula
for g (z) in the four cases is given as follows (in the rst case k = xx11 xx10 2 R):
92
x x0
k
if x0 ; x1 ; x1 6= 1;
x x1
x x0
if x1 = 1;
x x1
x1
x
x
x1
x1
if x0 = 1;
x1
x0
if x1 = 1:
x0
In each case g (z) has the desired form. Hence, f = fA for some 2
A with real entries as required.
93
2 matrix
II.7.10
Suppose the fractional linear transformation (az + b) = (cz + d) maps
R to R, and ad bc = 1. Show that a; b; c; and d are real or they are
all pure imaginary.
Solution
Because R is mapped on R and the orientation is perserved, f (x) will be
either increasing or decreasing.
Case 1 : Suppose f (x) is increasing, then f 0 (x) = 1= (cx + d)2 > 0 for all
x ) c; d are real. f (z) = (az + b) = (cz + d), then ax + b is real for all x 2 R,
so a; b are also real.
Case 2 : Suppose f (x) is decreasing, then f 0 (z) = 1= (cz + d)2 < 0 for all
x ) c; d are pure imaginary. f (z) = (az + b) = (cz + d), then ax + b are pure
imaginary for all x 2 R, so a; b are also pure imaginary.
94
II.7.11
Two maps f and g are conjugate if there is h such that g = h f h 1 .
Here the conjugating map h is assumed to be one-to-one, with
appropriate domain and range. We can think of f and g as the
"same" map, after the change of variable w = h (z). A point z0 is
a xed point of f if f (z0 ) = z0 . Show the following. (a) If f is
conjugate to g, then g is conjugate to f .
(b) If f1 is conjugate to f2 and f2 to f3 , then f1 is conjugate to f3 .
(c) If f is conjugate to g, then f f is conjugate to g g, and more
generally, the m fold composition f
f (m times) is conjugate
to g
g (m times).
(d) If f and g are conjugate, then the conjugating function h maps
the xed points of f to xed points of g. In particular, f and g
have the same number of xed points.
Solution
(a)
If f is conjugate to g then
g=h f
)f =h
thus g is conjugate to f
(b)
If f1 = h f2 h 1 and f2 = g f3 g
f1 = h g f3 g
g h
we get
1
= (h g) f3 (h g)
(c)
m times
g=h f
(d)
g=h f
}|
{
z
h 1 theng g : : : g =
=h f h 1 h f h
::: h f
z
=h f
h 1 , f (z0 ) = z0 , and w0 = h (z0 ), then
95
m times
}|
{
::: f h
g (w0 ) = h f h
(w0 )
= h (f (z0 )) = h (z0 ) = w0 :
We have that f and g have the same number of xed points since every point
of g is mapped to a xed point of f by the function h 1 .
96
II.7.12
Classify the conjugacy classes of fractional linear transformations
by establishing the following
(a)
A fractional linear transformation that is not the identity has either
1 or 2 xed points, that is, points satisfying f (z0 ) = z0 .
(b)
If a fractional linear transformation f (z) has two xed points, then
it is conjugate to the dilation z 7! az with a 6= 0, a 6= 1, that is, there
is a fractional linear transformation h (z) such that h (f (z)) = ah (z).
Is a unique? Hint. Consider a fractional linear transformation that
maps the xed points to 0 and 1.
(c)
If a fractional linear transformation f (z) has exactly one xed
point, then it is conjugate to the translation 7! + 1. In other
words, there is a fractional linear transformation h (z) such that
h (f (h 1 ( ))) = + 1, or equivalently, such that h (f (z)) = h (z) + 1.
Hint. Consider a fractional linear transformation that maps the
xed point to 1.
Solution
= z , az + b = cz 2 + dz , cz 2 + (d a) z b = 0. If this is = 0
a) az+b
cz+d
for every z, we get w (z) = z. It is not identically equall to zero, so it has
either 1 or 2 nite solutions. If c 6= 0, it has 2 solutions possibly one with
multiplicity and 1 is not a xed point. If c = 0, it has one nite solution
z1 = b= (d a), and z2 = 1 is a xed point, so they are two.
b) Make a change of variables h (z) = ( z + ) = ( z + ) that maps xed
points to 0 and 1 . Then h f h 1 is FLT with xed points at 0 and 1.
(h f h 1 ) (w) = aw for some a 6= 0, a 6= 1. f is conjugate to a dilation
Suppose aw is conjugate to Aw, i.e. 9h, (h (ah 1 )) (w) = Aw, h (az) =
Ah (z). h must map xed points to xed points, so either h (z) = cz or
h (z) = c=z. So either h (z) = cz ) h (az) = caz = Ah (z) = Acz ) A = a
or h (z) = c=z ) h (az) = c= (az) = Ah (z) = Ac=z ) A = 1=a. Thus a is
not unique.
c) Suppose h maps the xed points to 1, then h f h 1 has only one xed
point at 1, so (h f h 1 ) (w) = aw+b. Since this has no nite xed points,
aw +b = w has no solutions and a = 1, b 6= 0. Thus (h f h 1 ) (w) = w +b.
97
h f
()
(w) = w + 1 ()
(g h) f
(g h) 1 (w) = w + 1 ()
() ((g h) f ) (z) = (g h) (z) + 1:
98
III 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1
2
3
4
5
6
7
8
III.1.1
R 2
Evaluate
y dx + x2 dy along the following paths
from (0; 0) to
(2; 4),
(a) the arc of the parabola y = x2 ,
(b) the horizontal interval from (0; 0) to (2; 0), followed by the vertical interval from (2; 0) to (2; 4), (c) the vertical interval from (0; 0)
to (0; 4), followed by the horizontal interval from (0; 4) to (2; 4).
Solution
(a)
Z
y dx+x dy =
x=t
y = t2
dx = dt;
dy = 2t dt
dx = dt
dy = 0 dt
2 2
Z
dt+
t2 2t dt =
72
:
5
(b)
Z
y dx + x dy =
1
x=t
y=0
x=2
y=t
dx = 0 dt
dy = dt
2
2
0 dt +
t2 0 dt = 0
4
2
y dx + x dy =
=
t 0 dt +
0
2
Z
Z
Z
2
2
2
2
y dx + x dy +
y 2 dx + x2 dy = 16:
y dx + x dy =
1
22 dt = 16
(c)
Z
y dx + x dy =
1
x=0
y=t
dx = 0 dx
dy = dt
0
2
4
2
y 0 dt +
x=t
dx = dt
0 t 2
y 2 dx + x2 dy =
=
42 dt +
y
=
4
dy
=
0dt
0
2
Z
Z
Z
y 2 dx + x2 dy =
y 2 dx + x2 dy +
y 2 dx + x2 dy = 32:
1
02 dt = 0
t2 0 dt = 32
III.1.2
R
Evaluate xy dx both directly and using Greens theorem, where
R 1),
is the boundary of the square with vertices at (0; 0), (1; 0), (1;
and (0; 1).
VII.1.2
4
3
2
1
xy dx =
P dx+Q dy =
@D
ZZ
@Q
@x
@P
dy
dx dy =
x dx dy =
xy dx =
xy dx +
1
xy dx =
3
x 0 dx
x 1 dx = 0
1
=
2
1
:
2
1
:
2
III.1.3
R
Evaluate @D x2 dy both directly and using Greens theorem, where
D is the quarter-disk in the rst quadrant bounded by the unit
circle an the two coordinate axes.
SolutionR
Evaluate @D x2 dy directly, set @D =
x dy =
1
x dy =
2
x=t
y=0
x = cos t
y = sin t
dx = dt
dy = 0 dt
dx = sin t dt
dy = cos t dt
3.
+
t
t2 0 dt = 0
=2
cos2 t cos t dt =
0
1
2
3
dx = 0 dt
0 t 1
02 dt = 0
=
t
dy
=
dt
0
3
Z
Z
Z
Z
2
2
2
2
x dy =
x dy +
x dy +
x2 dy =
3
@D
1
2
3
R
Now we evaluate @D x2 dy, this time using Greens theorem. In this case,
P (x; y) = 0 and Q (x; y) = x2 .
Z
x2 dy =
ZZ
x=0
y=1
=2
x dy =
2x dxdy =
D
ZZ
x = r cos
x dxdy =
=2
y = r sin
D
@D
dx dy = r dr d
=2
0
0
1
2
r dr
1
Z
=2
=2
=2
2
cos d = :
3
=2
r cos rdrd =
III.1.4
R
Evaluate y dx both directly and using Greens theorem, where
is the semicircle in the upper half-plane from R to R.
SolutionR
Evaluate y dx directly, set
Z
y dx =
1
y dx =
2
x=t
y=0
dx = dt
dy = 0 dt
2.
x = R cos t
dx = R sin t dt
0 t
y = R sin t
dy = R cos t dt
Z
Z
Z
2
2
x2 dy =
x dy +
x dy =
@D
0 dt = 0
R
R sin t R sin t dt =
R
2
Now we evaluate
y dx, this time using Greens theorem. In this case,
P (x; y) = y and Q (x; y) = 0.
Z
ZZ
y dx =
ZZ
dx dy =
@D
dx dy =
x = r cos
y = r sin
dx dy = r dr d
0
0
r dr
R
Z
d =
0
2
R
:
2
r dr d =
R2
2
III.1.5
R
R
Show that @D x dy is the area of D, while @D y dx is minus the area
of D.
Solution
Using
R in the two cases
For @D x dy, we have P = 0 and Q = x, using Greens theorem we have
Z
Z Z
x dx =
dx dy = Area D:
For
@D
@D
III.1.6
Show that if P and Q are continuous complex-valued functions on
a curve , then
R
R Qdy
P dx
+
;
(z = x + iy)
z w
z w
is analytic for w 2 Cn . Express F 0 (w) as a line integral over .
Solution
Dierentiate by hand, use uniform convergence of
1
w z
as
1
(w +
1
w)
1
(z
w) (z
(w +
w))
R
Qdx
P dx
;
2 +
(z w)
(z w)2
z = x + iy
1
(z
w)2
III.1.7
Show that the formula in Greens theorem is invariant under coordinate changes, in the sense that if the theorem holds for a bounded
domain U with piecewise smooth boundary, and if F (x; y) is a
smooth function that maps U one-to-one onto another such domain
V and that maps the boundary of U one-to-one smoothly onto the
boundary of V , then Greens theorem holds for V . Hint. First note
the change of variable formulae for line and area integrals, given
by
ZZ
Pd
@V
Rd d
(P
Z@U
Z
@
@
dx +
dy ;
@x
@y
F)
(R F ) det JF dx dy;
where F (x; y) = ( (x; y) ; (x; y)), and where JF is the Jacobian matrix
of F . Use these formulae, with R = @P=@ . The summand
R
Q d is treated similarly.
Solution
ZZ
@P
d d =
@
ZZ
@P
@
det JF (x; y) =
@ @
@x @y
=
=
=
@
@x
@
@y
+
+
@
@x
@
@y
@ @
=
@y @x
ZZ
@P
( (x; y) ; (x; y))
@
U
@ @
@x @y
@ @
@y @x
dx dy
Pd =
@V
Z
@
@
@
@
(P F )
P ( (x; y) ; (x; y))
=
dx +
dy =
dx +
dy =
@x
@y
@x
@y
@U
@U
ZZ
@
@
@
@
=
P
+
P
dx dy =
@y
@x
@x
@y
U
ZZ
@
@
@2
@
@
@2
=
P ( (x; y) ; (x; y))
P
+
P ( (x; y) ; (x; y))
+P
dx dy =
@y
@x
@y@x @x
@y
@x@y
U
ZZ
@P @
@P @
@P @
@P @
@
@
=
+
+
dx dy =
@ @y
@ @y @x
@ @x
@ @x @y
U
ZZ
@P @ @
@P @ @
+
=
dx dy =
@ @y @x
@ @x @y
U
ZZ
@P
JF dx dy:
=
@
U
R
Similar argument as above hold for Qd
Here we use that
Z
Z
@
@
Qd =
(Q F )
dx +
dy
@x
@y
@V
@U
and replace R =
III.1.8
Prove Greens theorem for the rectangle dened by x0 < x < x1 and
y0 < y < y1 (a) directly, and (b) using the result for triangles.
Solution
VII.1.8a
VII.1.8b
III
IV
III
II
IV
II
(a)
Make gure and set the linesegment between (x0 ; y0 ) and (x1 ; y0 ) to
proceed in this way in counterclockwise direction. We have
Z
Z
Z
Z
Q dy =
P dx =
Q dy =
P dx = 0:
1
10
and
ZZ
@Q
@x
Z
y1
@P
dx dy =
@y
Z y1 Z x1
@Q
=
dx dy
x0 @x
y0
x1
Zx0x1
y1
y0
@P
dy dx =
@y
P (x; y0 )] dx =
Z x1
Z y1
Z y1
P (x; y0 ) dx =
P (x; y1 ) dx +
Q (x0 ; y) dy
Q (x1 ; y) dy
=
x0
x0
y0
y0
Z x1
Z y1
Z x1
Z y1
=
P (x; y0 ) dx+
Q (x1 ; y) dy
P (x; y1 ) dx
Q (x0 ; y) dy =
x0
y0
x0
y0
Z
Z
Z
Z
Q dy =
P dx+
Q dy +
P dx+
=
4
3
2
1
Z
=
(P dx + Q dy) :
=
[Q (x1 ; y)
Q (x0 ; y)] dy
[P (x; y1 )
Zx0x1
y0
@R
(b)
Write R = S [ T . The integrals on the diagonal are in dierent directions
and will cancel
Z
@R
(P dx + Q dy) =
Z
Z
(P dx + Q dy) =
(P dx + Q dy) +
=
@T
@S
ZZ
ZZ
@Q @P
@Q @P
=
dx dy +
dx dy =
@x
@y
@x
@y
S
T
ZZ
@Q @P
=
dx dy:
@x
@y
R
11
III.2.1
Determine whether each of the following line integrals is independent of path. If it is, nd a function h such that dh = P dx + Q dy.
If it is not, nd a closed path around which the integral is not
zero. (a) x dx + y dy, (b) x2 dx + y 5 dy, (c) y dx + x dy, (d) y dx x dy.
Solution
We compute the partial derivatve because if h exist we have that dh =
P dx + Qdy where @P=@y = @Q=@x.
(a)
P =x
Q=y
@Q
@P
=0
=0
@y
R
R@x
x2
P dx = 2 + g (y)
Qdy =
x2 +y 2
)h= 2
y2
2
+ f (x)
y6
6
+ f (x)
(b)
P = x2
Q = y5
@Q
@P
=0
=0
@x
R@y
R
3
P dx = x3 + g (y)
Qdy =
2x3 +y 6
)h= 6
(c)
P =y
Q=x
@Q
@P
=1
=1
R@y
R@x
P dx = xy + g (y)
Qdy = xy + f (x)
) h = xy
(d)
P =y Q= x
@P
= 1 @Q
= 1
@y
@x
The line integral is not independent of path, we use Greenstheorem
Z
ZZ
(y dx x dy) =
2 dx dy = 2 :
jzj=1
jzj<1
12
III.2.2
Show that the dierential
ydx + xdy
;
x2 + y 2
(x; y) 6= (0; 0) ;
Solution
P =
@P
@y
y
x2 +y 2
(x2 +y2 )+y(2y)
(x2 +y 2 )2
Q=
=
y 2 x2
(x2 +y 2 )2
@Q
@x
x
x2 +y 2
(x2 +y2 ) x(2x)
(x2 +y 2 )2
y 2 x2
(x2 +y 2 )2
P dx + Q dy =
jzj=r
jzj=r
I
x
1
y
dx+ 2
dy = 2
y dx + x dy =
x2 + y 2
x + y2
r
jzj=r
ZZ
1
1
= 2
(1 + 1) dx dy = 2 2 r2
= 2 6= 0
r
r
jzj<r
13
III.2.3
Suppose P and Q are smooth functions on the annulus fa < jzj < bg
that Hsatisfy @P=@y = @Q=dx. Show directly using Greens theorem
that jzj=r P dx + Qdy is independent of the radius r, for a < r < b.
Solution (A. Kumjian)
III.2.3
1
2 D
jzj=r2
jzj=r2
jzj=r
jzj=r1
14
III.2.4
Let P and Q be smooth functions on D satisfying @P=@y = @Q=@x.
Let 0 and 1 be two closed paths in D such that the straight line
segment Rfrom 0 (t) to 1 R(t) lies in D for every parameter value
t. Then
P dx + Q dy =
P dx + Q dy. Use this to give another
0
1
solution to the preceding exercise.
Solution
Use the theorem on page 81. Use the straight lines to deform 0 to 1 . Dene
s) 0 (t), 0 6 s 6 1, a1 6 t 6 b1 . The theorem on page
s (t) = s 1 (t) + (1
81 applies. Obtain the result for the annulus above by parameterizing the
circles jzj = r0 and jzj = r1 by 0 (t) = r0 e2 it , 1 (t) = r1 e2 it , 0 6 t 6 1. The
straight line segments joining 0 (t) to 1 (t) are radical are in the annulus,
so the rst part applies.
15
III.2.5
Let 0 (t) and 1 (t), 0 t 1, be paths in the slit annulus fa < jzj < bg n ( b; a)
from A to B. Write down explicitly a family of paths s (t) from A
to B in the slit annulus that deforms continuously to 1 .
Suggestion. Deform separately the modulus and the principal value
of the argument.
Solution
Write 0 (t) = r0 ei 0 (t) , 0 6 t 6 1 , when
continuous. (Use the fact that (t) = Arg
annulus) Also 1 (t) = r1 ei 1 (t) .
Then consider
s
s) r0 (t)] ei[s
1 (t)+(1
s)
0 (t)]
1:
sr1 (t) + (1
s) r0 (t)
and
< min (
(t) ;
(t))
(t) + (1
16
s)
(t)
max (
(t) ;
(t)) <
III.2.6
Show that any closed path (t), 0 t 1, in the annulus fa < jzj < bg
can be deformed continuously to the circular path (t) = (0) e2 imt ,
0
t
1, for some integer m. Hint. Reduce to the case where
j (t)j
j (0)j is constant. Then start by nding a subdivision
0 = t0 < t1 <
< tn = 1 such that arg (t) has a continuous determination on each interval tj 1 t tj .
Solution
We rst note that there are two cases. For the closed path (t) enclosing
the origin we can deform (t) to a curve with constant modulus. Suppose
j (t)j = r for 0 6 t 6 1 and (0) = (1) = 1 where a < r < b. Follow
hint, write (t) = rei j (t) , tj 1 6 t 6 tj . Note j (tj )
j+1 (tj ) is a integral
multiple of 2 , since (t) is continuous. Add multiples of 2 to the j s,
obtain (t) continuous for 0 6 t 6 1 such that (0) = 0 and (t) = rei (t) .
Note (1) = 2 m for some integer m. Deform by s (t) = rei[(1 s) (t)+2 sm] .
The second case is when (t) does not enclose the origin. For this case (t)
the counter of a region that lies entirely inside the annulus. Therefrom the
curve (t) may be deformed to any point of this region. In particular it
can be continuously deformed to the point (0) = (t) = (0) e2 imt when
m = 0.
17
III.2.7
Show that if 0 and 1 lie in dierent connected components of the
complement C nD of D in the extended
complex plane, then there
R
is a closed path in D such that d 6= 0. Hint. The hypothesis
means that there are > 0and a bounded subset E of CnD such
that 0 2 E; and every point of E has distance at least 5 from every
point of CnD not in E. Lay down a grid of squares in the plane
with side length ;and let F be the union of the closed squares in
the grid that meet E or that border on a square meeting
R E. Show
that @F is a nite union of a closed paths in D, and that @F d = 2 .
Solution
Proceed as in the hint.
R Assume that 0 Ris the centre of one of the squares
S0 in the grid. Then @S0 d = 2 , while @S d = 0 for any other square in
PR
d = 2 , where we sum over squares. S0 ; S1 ; : : : ; Sn
the grid. Thus
@Sj
in F . If two squares in F are ?????, the corresponding integrals along the
edge cancel. Then this is the sum over the integrals over the edges of the
Sj s that have F on oneRside and CnF on the other, that is the edges that
form @F . We have thus @F d = 2 . Note that by construction, @F
D.
Now note how edges @F can meet. Either a vertex in @F have one edge
coming in and one out, or it have two coming in and two out. By starting on
following edges, and making a left turn at vertices where 4 edges meet, we
must eventually end where we started, with a closed path. Three paths must
be disjoint,
P R except for vertices with Rfour edges. Call the paths 1 ; : : : ; m .
d = 2 , we must have d 6= 0 for one of the j s. (In fact,
Since
j R
we will have
d = 2 on the j s we get by moving from 0 to @F .)
j
18
III.3.1
For each of the following harmonic functions u, nd du, nd dv, and
nd v, the conjugate harmonic functions of u.
(a) u (x; y) = x y
(c) u (x; y) = sinh x cos y
y
(b) u (x; y) = x3 3xy 2
(d) u (x; y) = x2 +y
2
Solution
(a)
Use Cauchy-Riemanns equations page 83.
u (x; y) = x y
du = dx dy
dv = dx + dy
v (x; y) = x + y
(b)
Use Cauchy-Riemanns equations page 83.
u (x; y) = x3 3xy 2
du = (3x2 3y 2 ) dx 6xy dy
dv = 6xy dx + (3x2 3y 2 ) dy
v (x; y) = 3x2 y y 3
(c)
Use Cauchy-Riemanns equations page 83.
u (x; y) = sinh x cos y
du = cosh x cos y dx sinh x sin y dy
dv = sinh x sin y dx + cosh x cos y dy
v (x; y) = cosh x sin y
(d)
Use Cauchy-Riemanns equations page 83.
y
x2 +y 2
2xy
du = (x2 +y
2 )2 dx
2
2
dv = (xy2 +yx2 )2 dx
x
v (x; y) = x2 +y
2
u (x; y) =
x2 +y 2 y(2y)
dy
(x2 +y 2 )2
2xy
dy
(x2 +y 2 )2
19
2xy
(x2 +y 2 )2
dx +
x2 y 2
dy
(x2 +y 2 )2
III.3.2
Show that a complex-valued function h (z) on a star-shaped domain
D is harmonic if and only if h (z) = f (z) + g (z), where f (z) and g (z)
are analytic on D.
Solution
Write h = u + iw where u, w are harmonic. By the Theorem on page 83
both u and w have harmonic conjugate in D, (since D is star-shaped), so
thus there are ', analytic such that u = Re ', w = Re .
u = (' + ') =2;
w=
=2;
gives
h=
1
2
'+'+i +i
= 12 [' + i ] +
1
2
'+i
Take
g = 12 ['
f = 12 [' + i ] ;
i ];
20
+
@ 2 (Im g)
=
@y 2
@ 2 Im f
@ 2 Im g
+
@y@x
@y@x
@ 2 Re f
@ 2 Re g
=0
@y@x
@y@x
III.3.3
Let D = fa < jzj < bg n ( b; a), an annulus slit along the negative
real axis. Show that any harmonic function on D has a harmonic
conjugate on D.
Suggestion. Fix c between a and b, and dene v (z) explicitly as a line
integral along the path consisting of the straight line from c to jzj
followed by the circular arc from jzj to z. Or map the slit annulus
to a rectangle by w = Log z.
Solution
Integrate the CR equations we have
Z z
@u
1 @u
dr + r d =
v (z) =
r@
@r
z0
@u
dx +
@y
@u
dy
@x
21
Solution
u has harmonic conjugate v1 on the annulus slit along ( b; a), and also a
harmonic conjugate v2 on the annulus slit (a; b). Since v1 v2 is constant
above the slit ( b; a), and also constant below the slit, v1 jumps by a
constant across the slit. Arg z also jumps by a constant across the slit. By
appropriate choice of C, v1 C Arg z is continuous across the slit ( b; a),
and u C log jzj has a harmonic conjugate v1 C log jzj on the annulus. For
the identity, use the polar form of the Cauchy-Riemann equations to convert
the rderivative of u to a derivative of v.
22
III.3.5
The ux of a function u across a curve is dened to be
Z
Z
@u
ds = ru n ds;
@n
where n is the unit normal vector to and ds is arc length. Show
that if a harmonic function u on a domain D has a conjugate harmonic function v on D, then the integral giving the ux is independent of path in D. Further, the ux across a path in D from
A to B is v (B) v (A).
Solution
!
ds ds
t = dx
n = dy
; dy , !
; dx
,
ds
ds
R
R
R @u
@u dy
@u dx
ru!
n ds =
ds =
dx + @u
dy |{z}
=
@x ds
@y ds
@y
@x
R @v
R
@v
dx + @y
dy = dv
@x
( ) CR equations. If goes from P to Q, this is v (Q) v (P ), which shows
that the integral is independent of path in D.
23
III.4.1
Let f (z) be a continuous function on a domain D. Show that if
f (z) has the mean value property with respect to circles, as dened
above, then f (z) has the mean value property with respect to disks,
that is if z0 2 D and D0 is a disk
RR centered at z0 with area A and
1
contained in D, then f (z0 ) = A D0 f (z) dx dy.
Note: This MVP for area is not exactly what is dened as the MVP for
disks. What is then is that if h (z0 ) is the average of h (z) for any circle
fjz z0 j = rg, 0 < r < 1. Then h (z) is the average of h (z) over any disk
fjz z0 j = r0 g, 0 < r < r0 . Express dx dy in polar coordinates at z0 and
integrate rst with respect to .
Solution
Suppose that f satises the mean value property,
Z 2
1
f z0 + rei d
f (z0 ) =
2 0
ZZ
1
x = x0 + r cos dx dy = r dr d 0 r R
f (z) dx dy =
f (z) dx dy =
2
y = y0 + r sin
0
2
R
D0
D0
Z 2 Z R
Z R Z 2
Z R
1
1
1
=
f z0 + rei r dr d =
f z0 + rei d r dr =
2 f (z0 ) r d
2
2
R 0
R 0
R2 0
0
0
Z
R
1 r2
2f (z0 ) R
=
r dr = 2
f (z0 ) = f (z0 ) :
R2
R 2 0
0
1
A
ZZ
Another note: Its better to use something other than 1=A as A was already
used as the circle average.
24
III.4.2
Derive (4.2) from the polar form of the Cauchy-Riemann equations
(Exercise II.3.8).
Solution
The polar form of Cauchy-Riemanns equations (from Exercise II.3.8)
1 @v
@u
=
;
@r
r@
@u
=
@
@v
:
@r
Now we have
r
@u
z0 + rei
@r
d =
25
@v
z0 + rei
@
d =0
III.4.3
A function f (t) on an interval I = (a; b) has the mean value property
if
f
s+t
2
f (s) + f (t)
;
2
s; t 2 I:
s+t
2
=A
s+t
2
+B =
1
1
f (s) + f (t)
(As + B) + (At + B) =
:
2
2
2
t) d) = tf (c) + (1
t) f (d) :
26
tc + (1
t) d =
1
((t0 c + (1
2
t0 ) d + (t1 c + (1
t1 ) d))
hence,
f (tc + (1
t) d) =
=f
1
((t0 c + (1
2
t0 ) d + (t1 c + (1
t1 )d)
1
(f (t0 c + (1 t0 ) d) + f (t1 c + (1 t1 ) d)) = by the mean value property
2
1
= (t0 f (c) + (1 t0 ) f (d) + t1 f (c) + (1 t1 ) f (d)) = by inductive hypothesis
2
= tf (c) + (1 t) f (d) :
=
27
III.4.4
Formulate the mean value property for a function on a domain
in R3 , and show that any harmonic function has the mean value
property.
Hint. For A 2 R3 and r > 0, let Br be the ball of radius r centered
at A, with volume element d , and let @Br be its boundary sphere,
with area element d and unit outward normal vector n. Apply the
Gauss divergence theorem
ZZZ
ZZ
r Fd
F nd =
Br
@Br
to F = ru.
Solution
Prove MV theorem for harmonic functions in Rn , as follows u (R!
x)
R R @u !
!
!
u 0 = 0 @r (r x ) dr. Let d = area measure, x = unit vector, get
R @u !
R @u !
R R R @2u !
(R x ) d
(r x ) d =
(r x ) drd (!
x ). Get directional
@r
@r
S @r2
derivative of Strokes formula. Better: Apply strokes theorem to shell f 0 < <
get
R
R!
R
R !
::: V !
n d surface =
: : : r V volume. Apply to ru = !
v,
Volume
R
R
R
R
get : : : ru!
n d surface = : : : r2 !
n d volume = 0,
R
R
R @u
R @u
::: @ d
::: @ d
!
!
j x x0 j= 0
j x x0 j=R 1
R
@u
: : : ud
_ is constant.
@
!
x
x
=
j
0j
28
1 g,
III.5.1
Let D be a bounded domain, and let u be a real-valued harmonic
function on D that extends continuously to the boundary @D. Show
that if a u b, then a u b on D.
Solution
Since D [ @ is a compact set we have that u attains both maximum and
minimum values. Assume that u (z0 ) = c > b fore some z0 2 D. Then it
follows by the maximum principle that u c on D, but this is a contradiction
to the asumption that u is continuously extended to @D. Thereby u b on
D.
For u we have that b
u
a on @D. Assume that u (z0 ) = c0 > a
for seme z0 2 D. Then by the maximum principle u c0 on D, but this
contradicts that u is continuously extended to @D with u
a on @D.
So u
a on D, equivalently a u b on D.
This show that a u b on D.
29
III.5.2
Fix n
1, r > 0 and = ei' . What is the maximum modulus of
z n + over the disk fjzj rg? Where does z n + attain its maximum
modulus over the disk?
jz n j + j j = rn +
30
= (rn + ) ei' = j
+ j:
III.5.3
Use the maximum principle to prove that the fundamental theorem
of algebra, that any polynomial p (z) of degree n 1 has a zero, by
applying the maximum principle to 1=p (z) on a disk of large radius.
Solution
It is su cient to show that any p (z) has one root, for by division we can
then write p (z) = (z z0 ) g (z), with g (z) of lower degree.
Note that if
p (z) = an z n + an 1 z n
+ a0 ;
31
III.5.4
Let f (z) be an analytic function on a domain D that has no zeros
on D. (a) Show that if jf (z)j attains its minimum on D, then f (z) is
constant. (b) Show that if D is bounded, and if f (z) extends continuously to the boundary @D of D, then f (z) attains its minimum
on @D.
Solution
(a)
If jf (z)j attains its minimum on D, then j1=f (z)j attains its maximum on
D, which can only happens if f (z) is a constant, by the maximum principle.
(b)
Since D [ @D is compact and f (z) is continuous it follows that jf (z)j attains
both maximum and minimum value on D [ @D. Assume that jf (z)j does
not attain its minimum on @D. Then jf (z)j attains its minimum on D and
it follows by (a) that f is constant so jf (z)j attains its minimum on @D.
32
III.5.5
Let f (z) be a bounded analytic function on the right half-plane.
Suppose that f (z) extends continuously to the imaginary axis and
satises jf (iy)j M for all points iy on the imaginary axis. Show
that jf (z)j M for all z in the right half-plane.
Hint. For " > 0 small, consider (z + 1) " f (z) on a large semidisk.
Solution
Now we consider the function (z + 1) " f (z) in the right half-plane, because
f (z) is bounded here, we assume that jf (z)j
C in this domain. On a
semidisk of radius r in the right half-plane we have
(z + 1)
"
f (z)
jr
"
1j"
f (z)
(R
1)"
"
f (z)
"
f (z)
"
f (z) :
Then
jF" (iy)j
and
F" Rei
C R
"
M;
or
jf (z)j
M (1 + jzj)" :
34
M.
III.5.6
Let f (z) be a bounded analytic function on the right half-plane.
Suppose that lim sup jf (z)j
M as z approaches any point of the
imaginary axis. Show that jf (z)j M for all z in the half-plane.
Remark. This is a technical improvement on the preceding exercise
for students who can deal with lim sup (see Section V.1).
Solution
Set
g (z) = (z + 1)
"
f (z) :
Then
jg (z)j 6 jf (z)j
becauce jz + 1j > 1 for all z in the right half-plane.
Take R large so that jg (z)j 6 M for jzj > R. The lim sup condition implies
that there is > 0 such that
jg (z)j 6 M + ";
Apply the maximum principles to the domain fzj jzj < R; Re z > g, to obtain jg (z)j 6 M + ". Then g (z) ! f (z) as " ! 0.
35
III.5.7
Let f (z) be a bounded analytic function on the open unit disk D.
Suppose there are a nite number of points on the boundary such
that f (z) extends continuously to the arcs of @D separating the
points and satises f ei
M there. Show that jf (z)j M on D.
Hint. In the case that there is only one exceptional point z = 1,
consider the function (1 z)" f (z).
Solution
Let the points be a1 ; : : : ; an . Then (z aj )" , where 1
j
n is analytic
"
(take analytic
branch). We have that jz aj j ! 0 as z ! aj for 1 j n.
Q
Then [ (z aj )" ] f (z) = f" (z) satises f" (z) is continuous in D [ @D.
jf" (z)j 6 M on @D. jf" (z)j 6 M on D. Let " ! 0, obtain jf (z)j 6 M on
D.
Solution
Let C = supz2D jf (z)j. By denition jf (z)j C. Let z1 ; : : : ; zn 2 @D be all
the points for which f (z) does not extend continuously.
Now suppose that jf (z)j M for all z 2 @D except fz1 ; : : : ; zn g and to the
contrary M < C, then there is a z0 2 D such that jf (z0 )j = M + with
0<
C M.
Let
g (z) = (z
z1 )" : : : (z
zn )" f (z)
j(z
z1 )" : : : (z
zn )" j
and
j(z
z1 )" : : : (z
zn )" j
z1 )" : : : (z0
zn )" j f (z0 )
we have that
36
(M + )
M + =2
M+
M + =3
M
M + =2
(M + )
M+
jg (z0 )j
sup j(z
z2@D
which is a contradiction.
37
z1 )" : : : (z
zn )" j jf (z)j
M + =3
M
M
III.5.8
Let f (z) be a bounded analytic function on a horizontal strip in
the complex plane. Suppose that f (z) extends continuously to the
boundary lines of the strip and satises jf (z)j
M there. Show
that jf (z)j M for all z in the strip. Hint. Find a conformal map
of the strip onto D and apply Exercise 7.
Solution
< Im z < 2 , so it is mapped to right half-plane by
Assume strip is
2
w = i log z, z = eiw . Then g (w) = f (eiw ) hold on right halfplane, 6 M
at all point of iR except at w = 0. Modify proof idea of Exercise III.5.5 to
get jgj 6 M on the halfplane.
38
III.5.9
Let D be an unbounded domain, D 6= C, and let u (z) be a harmonic
function on D that extends continuously to the boundary @D. Suppose that u (z) is bounded below on D, and that u (z)
0 on @D.
Show that u (z)
0 on D. Hint. Suppose 0 2 @D, and consider
functions of the form u (z) + log jzj on D \ fjzj > "g.
Solution (From Hints and Solutions)
Let " > 0. Take > 0 such that u (z) > " for
> 0. Take R > 0 so large that u (z) + log jzj
maximum principle, u (z) + log jzj > " + log
hence for all z 2 D such that jzj > . Let ! 0,
u (z) > 0 on D.
";
and
so that
z 2 D \ fjzj = g
for arbitary " > 0. Since u is bounded below, there is also R0 > 0 such that
for R > R0
u (z) + log jzj
z 2 D \ fjzj = Rg ;
0;
39
III.5.10
Let D be a bounded domain, and let z0 2 @D. Let u (z) be a harmonic function on D that extends continuously to each boundary
point of D except possibly z0 . Suppose u (z) is bounded below on
D, and that u (z) 0 for all z 2 @D, z 6= z0 . Show that u (z) 0 on
D.
Solution
D hold, z0 2 @D, u > M on D, u > 0 (@D) n fz0 g. The map ' (z) =
1= (z z0 ) maps D onto an unbounded domain, to which the result of Exercise 9 applies. Or: consider w (z) = u (z) + log (1= jz z0 j). Assume
jz z0 j 6 R on D, then take " > 0 small, assume R > 1. Then w (z) >
" log (1=R) at all points of (@D) n fz0 g, w (z) ! +1 as z ! z0 . By the
maximum principle, w (z) > " log (1=R) on D. Let " ! 0, get w (z) > 0 on
D.
40
III.5.11
Let E be a bounded set of integer lattice points in the complex
plane. A point m + ni of E is an interior point of E if its four
immediate neighbors m 1 + ni, m + ni i belong to E. Otherwise
m + ni is a boundary point of E. A function E is harmonic if its
value at any interior point of E is the average of its values at the
four immediate neighbors. Show that a harmonic function on a
bounded set of lattice points attains its maximum modulus on the
boundary of the set.
Solution
Suppose u (z) attains its maximum at m + ni, call it M . Then u (m + ni) is
the average of its values at its four neighbors, so all their values must coincide
with M. Consider fm + ni : u (m + ni) = M g. This is a nite set. It have a
point with largest m. This point have a neighbor in the boundary of E. u
attains value M at that point. u = c at some point of boundary of E.
For complexvalued case, follow same approach as in book. Remark: Should
just consider real-valued u, and show it attains its maximum and minimum
at boundary points of E.
41
III.6.1
Consider the uid ow with constant velocity V = (2; 1). Find the
velocity potential (z), the stream function (z), and the complex
velocity potential f (z) of ow. Sketch the streamlines of the ow.
Determine the ux of the ow across the interval [0; 1] on the real
axis and across the interval [0; i] on the imaginary axis.
Solution
V = (2; 1), V = r , = 2x + y, @@x = @@y , @@y = @@x ) = 2y x
f (z) = + i = 2z iz = (z i) z
Streamlines are straight line with slope 1=2, ux across [0; 1] = (1)
(0) =
1, ux across [0; i] = (i)
(0) = 2.
42
III.6.2
Fix real numbers
and
polar coordinates by
V (r; ) =
ur + u ;
r
r
where ur and u are the unit vectors in r and directions, respectively.
(a)
Show that V (r; ) is the velocity vector eld of a uid ow, and nd
the velocity potential (z) of the ow.
(b)
Find the stream function (z) and the complex velocity potential
f (z) of the ow.
(c)
Determine the ux of the ow emanating from the origin. When
is 0 a source and when is 0 a sink?
(d)
Sketch the streamlines of the ow in the case = 1 and = 1.
Solution
a)
V (r; ) = r ur + r u
r = @@r ur + 1r @@ u = r ur + r u
Get @@r = r , = log r + h ( )
1@
= h0 ( ) = r , h0 ( ) = , h ( ) =
)V =r
r@
= log r +
is harmonic locally. is the velocity vector eld of a ow
b)
= Re ( log z + arg z) stream function =
= Im ( log z + arg z) =
arg z
log jzj
complex velocity of a point f (z) = (
i ) log z
c)
Flux = of around a circle centred at 2 , sin u : > 0, sin u : > 0.
d)
= 1, = 1. Note: and are only locally dened.
43
III.6.3
Consider the uid ow with velocity V = r , where (r; ) =
(cos ) =r. Show that the streamlines of the ow are circles and
sketch them. Determine the ux of the ow emanating from the
origin.
Solution
= cosr =
r cos
r2
= Re
1
z
= Re
R jzj
= Im
1
z
sin
r
44
III.6.4
Consider the uid ow with velocity V = r , where
(z) = log
z 1
:
z+1
Show that the streamlines of the ow are arcs of circles and sketch
them. Determine the ux of the ow emanating from each of the
singularities at 1.
Solution
is the composition of an analytic function w = (z 1) = (z + 1) and the harmonic function log jwj , so is harmonic, and = Re log ((z 1) = (z + 1)).
The stream function is = ImH(log (z 1) = H(z + 1)) = arg ((z 1) = (z + 1))
, then is not single-valued. jz 1j=" d = jz 1j=" d arg (z 1) = 2 = ux
H
H
emanating from +1. jz 1j=" d = jz 1j=" d arg (z + 1) = 2 = ux emanating from 1. z = 1 is a source, z = 1 is a sink.
Set = (z 1) = (z + 1), w = log . The line Im w = constant in the w
plane correspond to rays issuing from 0 in the
plane. (since = ew ), and
this correspond to arcs of circles from 1 to 1 in the zplane, since FLTs
maps circles to circles. Hence streamlines are arcs of circle from 1 to 1,
including the circle through 1.
45
III.6.5
Consider the uid ow in the horizontal strip f0 < Im z < g with
a sink at 0 and equal sources at 1. Find the stream function
(z) and the velocity vector eld V (z) of the ow. Sketch the
streamlines of the ow. Hint. Map the strip to a half-plane by
= ez and solve a Dirichlet problem with constant boundary values
on the three intervals in the boundary separating sinks and sources.
Solution
Consider uid ow in a horizontal strip. f0 < Im z < g with a sink 0, but
no source at 1 . Find the stream function (z). Sketch the ow lines.
Find the complex vector eld V (z). Hint : Map the half-plane by.
Map to half-plane = ez . A arg + B arg (
1) + C, arg = arg ez = y,
Ay+C+B arg (ez 1), c3 = 0, requires A+ B+C = 0 ) C =
(A + B),
A (y
) + B [arg (ez 1)
]. C1 = C2 require A + B
B =
z
[ A
B ], A [ (y
) + 2 [arg (e
1)
]] = A [ ( + y) + 2 arg (ez
46
1)]
III.6.6
For a uid ow with velocity potential (z), we dene the conjugate ow to be the ow whose velocity potential is the conjugate
harmonic function (z) of (z). What is the stream function of
the conjugate ow? What is the complex velocity potential of the
conjugate ow?
Solution
Stream function of conjugate ow in direction r is
potential is g (z) =
i = if (z) = i ( + i )
47
. Complex velocity
III.6.7
Find the stream function and the complex velocity potential of the
conjugate ow associated with the uid ow with velocity vector
ur =r. Sketch the streamlines of the conjugate ow. Do the particles
near the origin travel faster or slower than particles on the unit
circle?
Solution
Flow ur =r, = log jzj, = arg z. Complex velocity potential of conjugate
ow is i log z. Streamlines of the conjugate ow are
= log jzj =
constant , r = jzj = constant, which are circles centered in origin. Speed
= jV (z)j = jf 0 (z)j = 1= jzj. Particles travel faster near z = 0.
48
III.6.8
Find the stream function of the conjugate ow of
1
( ur + u ) :
r
Sketch the streamlines of both the ow and the conjugate ow on
the same axis. (See Exercise 2d.)
V (r; ) =
Solution
This is the vector eld from 2d. Hence V = r ( log r + ). The velocity
potential is = log r + . The conjugate ow has velocity potential =
log r = Re ((i 1) log z). The stream function of the conjugate ow
is then Im ((i 1) log z) =
+ log r =
. (The steam function of the
conjugate is always the negative of the velocity potential of the ow. See
Ex.6) Streamline of conjugate ow are orthogonal to the stream of the ow
(Exercise 2d) and velocity vector eld. Conjugate ow is the rotation by 90
of the velocity vector eld of the ow, so the origin is a sink of both ows in
this case.
49
III.7.1
Find the steady-state heat distribution u (x; y) in a laminar plate
corresponding to the half disk fx2 + y 2 < 1; y > 0g, where the semicircular top edge is held at temperature T1 and the lower edge
( 1; 1) is held at temperature T2 . Find and sketch the isothermal
curves for the heat distribution. Hint. Consider the steady-state
heat distribution for the full unit disk with the top held at temperature T1 and the bottom temperature T3 , where T2 = (T1 + T3 ) =2.
Solution
(z) = 2 (Arg (1 + z)
Arg
1+z
1 z
= konsant
then
(1 + z) (1
z) = 1 + 2i Im z
and
Im z
= konstant.
1 jzj2
50
jzj2
III.7.2
Find the potential function (x; y) for the electric eld for a conducting laminar plate corresponding to the quater-disk fx2 + y 2 < 1; x > 0; y > 0g,
where the two edges on the coordinate axis are grounded (that is,
= 0 on the edges), and the semicircular edge is held at constant
potential V1 . Find and sketch the equipotential lines and the lines
of force for the electric eld. Hint. Use the conformal map = z 2
and the solution to the preceding exercise.
Solution
(z) = 2 (Arg (1 + w (z))
Arg (1
w (z))),
51
= V1 2 (Arg (1 + z 2 )
Arg (1
z 2 ))
III.7.3
Find the potential function (x; y) for the electric eld for a conducting laminar plate corresponding to the unit disk where the
boundary quater-circles in each quadrant are held at a constant
voltages V1 ; V2 ; V3 and V4 . Hint. Map the disk to the upper halfplane by w = w (z) and consider potential functions of the form
Arg (w a).
Solution
Let w = i (1 z) = (1 + z), w (1) = 0, w ( 1) = 1, w (i) = 1, w ( i) = 1.
z ! w (z) maps unit disk to upper halfplane. Suers to nd a harmonic function u (w) in upper half-plane with boundary values constant
on each interval, as above. Any such function is a linear combination of
u 1 (w) = 1 Arg (w + 1), u0 (w) = 1 Arg w, u1 (w) = 1 Arg (w 1), and if
= A 1 u 1 + A0 u0 + A1 u1 + C, we get a system we can back solve.
8
8
A
+
A
+
A
+
C
=
V
C = V2
>
>
1
0
1
3
>
>
<
<
A0 + A1 + C = V3
A1 = V1 V2
)
A1 + C = V1
A0 = V4 V1
>
>
>
>
:
:
C = V2
A 1 = V3 V4
That does it-just plug in.
Example: V2 = 1, V1 = V3 = V4 = 0, this solution is
and we P
can get other elementary solutions, 1 (z),
(z) = vj j (z).
52
III.7.4
Find the steady-state heat distribution in a laminar plate corresponding to the vertical half-strip fjxj < =2; y > 0g, where the vertical sides at x =
=2 are held at temperature T0 = 0 and the
bottom edge ( =2; =2) on the real axis is held at temperature
T1 = 100. Make a rough sketch of the isothermal curves and the
lined of heat ow. Hint. Use w = sin z to map the strip to the
upper half-plane, and make use of harmonic functions of the form
Arg (w a).
Solution
Try
= a Arg (w + 1) + b Arg (w
8
< a = 50
b = 50
:
c=0
(z) = 50 Arg (sin z + 1)
8
< a+b+c=0
a b + c = 100
1) + c, want
:
a b+c=0
50 Arg (sin z
53
1).
III.7.5
Find the steady-state heat distribution in a laminar plate corresponding to the vertical half-strip fjxj < =2; y > 0g, where the side
x =
=2 is held at constant temperature T0 , the side x = =2 is
held at constant temperature T1 , and the bottom edge ( =2; =2)
on the real axis is insulated, that is, no heat passes through the
bottom edge, so the gradient ru of the solution u (x; y) is parallel
there to the x axis. Hint. Try linear functions plus constants.
Solution
=0
u (x; y) = ax + b is harmonic. @u
@y
T1 T0
a + b = T0
a=
2
)
1
a + b = T1
b = T0 +T
2
2
1
u = 1 (T1 T0 ) x + T0 +T
.
2
54
III.7.6
Find the steady-state heat distribution in a laminar plate corresponding to the upper half-plane fy > 0g, where the interval ( 1; 1)
is insulated, the interval ( 1; 1) is held at temperature T0 , and
the interval (1; 1) is held at temperature T1 . Make a rough sketch
of the isothermal curves and the lines of heat ow. Hint. Use the
solution to Exercise 5 and the conformal map from Exercise 4.
Solution
Let z = z (w) map the vertical half-strip to the upper half-plane with
z ( =2) = 1 and z ( =2) = 1, w = w (z) : upper halfplane ! vertical
semi strip. Take z = sin w, this has desired mapping property, w = sin 1 (z).
1
.
Then solution is composition of sine function with 2 (T1 T0 ) Re w + T0 +T
2
T0 +T1
1
1
2
Get u = (T1 T0 ) Re sin (z) + 2 , where we take branch sin z
with values in the vertical semi-strip. Note : This conformal map was used
in Exercise 4.
55
III.7.7
The gravitational eld near the surface of the earth is approximately constant, of the form F = ck, where k is the unit vector in
the z direction in (x; y; z) space and the surface of the earth is
represented by the plane where z = 0 (the at earth theory). Show
that F is conservative, and nd a potential function for F .
Solution
!
(x; y; z) = cz, r = c k
Since
r
F =
@ @ @
; ;
@x @y @z
56
(0; 0; c) = (0; 0; 0) ;
III.7.8
Show that the inverse square force eld F = ur =r2 on R3 is conservative. Find the potential function for F , and show that is
harmonic.
Solution
!
!
r = ur2r = r ,
(x; y; z) =
1
r
57
III.7.9
For n
3, show that the function 1=rn 2 is harmonic on Rn n f0g.
Find the vector eld F that has this function as its potential.
Solution
1
1
2
=
n 2 = u = r
rn 2
2 +:::+x2
2
x
(1
n)
@u
n
= 1 2 (x21 + : : : + x2n )
@xj
@2u
@x2j
2 n
rn
= (2
n) r
(2 n)(
+ (2
n
n
2
n) xj
n
2
n) xj r
= (x21 + : : : + x2n )
2xj = (2
n
2
(x21 + : : : + x2n )
n)x2j
n
2
2xj =
+
r n+2
(2 n)n P 2
2 n
u = n rn
xj = 0 this shows that u is harmonic on Rn n f0g
r n+2
P @u
P
x
F = u=
e = (2rnn) xj ej = r2n n1 r j ej = (2rnn)~1ur
@xj j
!
2) Area(S)
rn 1
58
IV 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1
2
3
4
5
6
7
8
IV.1.1
1 2 3 P L
K
F
KKK
FFF
Let be the boundary of the triangle f0 < y < 1 x; 0 < x < 1g, with the
usual counterclockwise
orientation.
EvaluateR the following integrals
R
R
(a)
Re z dz
(b)
Im z dz
(c)
z dz
Solution
a)
We begin do split the contur into three parts so that
And integrate along each side in the triangle
=
We parametrizize
and dz = dt
Z
(t) = t, for 0
Re z dz =
1
3.
1
t dt = :
2
1, then Re (t) = t
Re (t) dz =
1
1, then
1, then,
Im z dz =
1
Im z dz =
Z2
0 dt = 0;
t ( 1 + i) dt =
Im z dz =
(1
t) ( i) dt =
1 1
+ i;
2 2
1
i:
2
thus
Z
Im z dz =
1
:
2
Z
z dz =
2
z dz =
Z 11
0
z dz =
3
(1
1
t dt = ;
2
t + it) ( 1 + i) dt =
(i (1
1
t)) ( i) dt = :
2
thus
Z
z dz = 0:
IV.1.2
1 2 3 P
L
K
111
LLL
Absolutbeloppet i c) nagot a
r f el
Let be the unit circle fjzj = 1g, with the usual counterclockwise
orientation.
following Rintegrals, for m = 0; 1; 2; : : :.
R m EvaluateR the
(a)
z dz
(b)
z m dz
(c)
z m jdzj
Solution
Set z = ei , where 0
(a)
Z
z dz =
2
im
ie d = i
ei(m+1)
d =
m+1
i(m+1)
1
ei2
m+1
(m+1)
1 =
=
0
0; m 6=
2 i m=
1
1:
(b)
Z
z dz =
2
im
ie d = i
ei(1
m)
d =
ei2
(m 1)
ei(1 m)
=
1 m
=
0
1
1
1 =
0; m 6= 1
2 i m = 1:
1 =
0;
2
(c)
Z
z jdzj =
2
im
eim
d =
im
=
0
1
ei2
im
m 6= 0
m = 1:
IV.1.3
1 2 3 P L K
111
Let be the circle fjzj = Rg, with the usual counterclockwise orientation.
integrals,
for m = 0; 1; 2; : : :.
R mEvaluate theR following
R m
(a)
jz j dz
(b)
jz m j jdzj
(c)
z dz
Solution
Set z = Rei , where 0
(a)
jz j dz =
2
m im
R e
iRe d = iR
m+1
ei d = Rm+1 ei
2
0
= Rm+1 ei2
1 = 0:
(b)
Z
jz j jdzj =
2
m im
R e
iRe
d =R
m+1
0) = 2 Rm+1 :
(c)
Z
=
z dz =
2
m
R e
im
iRe d = iR
2
i(1 m)
Rm+1 i2
e
1 m
m+1
(1 m)
1 =
0;
m 6= 1;
=
m+1
2 iR
; m = 1;
d =R
m+1
ei(1 m)
1 m
0;
m 6= 1;
2
2 iR ; m = 1:
=
0
IV.1.4
1 2 3 P
L
K
LLL
Show that if D is a bounded domain with smooth boundary, then
Z
z dz = 2i Area (D) :
@D
Solution
Substitute z = x
IV.1.5
1 2 3 P L
K
111
KKK
Show that
I
138
jz 1j=1
ez
dz
z+1
2 e2 :
Solution
We begin with parametrizize the circle jz 1j = 1, as z = 1+eit = 1+cos t+
i sin t where 0 t 2 .
For the nominator in the integrand ez = (z + 1) we have
jez j = eRe z+i Im z = eRe z ei Im z = eRe z
e2 ;
1;
Thus
M=
jez j
ez
=
1+z
jz + 1j
e2
= e2 :
1
IV.1.6
1 2 3 P L
K
111
KKK
Show that there is a strict inequality
I
p log R
Log z
;
dz
2
2
z2
R
jzj=R
R>e :
Solution
For the integrand Log z=z 2 , with the nominator Log z = log R + i with the
principalargument
we have
Log R
=
R2
p
log2 R +
R2
log2 R +
R2
log2 R + log2 R
=
R2
2 log R
:
R2
there the last inequalitys follows by given fact R > e that gives 2 < log2 R.
Becauce the integrationcontour is a circle with radius R we have that L =
2 R.
M L estimate gives
p
p log R
H
2 log R
Log z
dz
6
2
R
=
2
2
;
R>e :
jzj=R
z2
R2
R
IV.1.7
1 2 3 P L K
Show that there is a strict inequality
I
zn
2 Rn+1
dz
;
R > 1; m
m
1
Rm 1
jzj=R z
1; n
1:
Solution
If jf (z)j 6 m0 < m on an arc
R
f (z) dz +
of the circle,
0 [ 1,
then
f (z) dz =
m) length ( 0 )+
1
,
Rm 1
L = 2 R, get
ik=m
with strict inequality unless z = R e|2 {z
} . )M=
th
zn
jzj=R z m 1
dz <
Rn
2 R=
Rm 1
2 Rn+1
,
Rm 1
Rn
,
Rm 1
root of w
IV.1.8
1 2 3 P
L
K
LLL
Suppose the continuous function f ei on the unit circle satises
R
f ei
M and jzj=1 f (z) dz = 2 M . Show that f (z) = cz for
some constant c with modulus jcj = M .
Solution
H
Multiply f by a unimodular constant, assume jf (z)j 6 M , jzj=1 f (z) dz =
R2
R2
i
f
Re f ei iei d = 2 M ,
e
d
=
2
M
.
Take
real
parts,
have
0
0
i
i
i
6 M . If have strict inequality somewhere,
Re f e ie d 6 f e
R2
then 0 < 2 M . We conclude that Re f ei iei d
M . Since f ei iei
i
i
i
i
M , we have Im f e ie d
0. ) f e ie
M , f ei
ie i M ,
f (z)
iM z. ) f (z) = cz for a constant, jcj = M .
10
IV.1.9
1 2 3 P L K
Suppose h (z) is a continuous function on a curve . Show that
Z
h (z)
dz;
w 2 Cn ;
H (w) =
z w
Solution
For the analyticity, dierentiate
h by hand. (See
i Exercise III.1.6). The DerivR
h(z)
h(z)
ative is H 0 (w) = lim 1w
dz =
z (w+ w)
z w
w!1
R
R
h(z) w
lim 1w (z (w+
dz = (zh(z)dz
w))(z w)
w)2
w!1
11
IV.2.1
1 2 3 P
L
K
K
111
LLL KKK
246
Evaluate the following integrals, for a path that travels from
i
to i in the right half-plane, and also for a path from
i to i
in theR left half-plane.R
R
R
(a)
z 4 dz
(b)
ez dz
(c)
cos z dz
(d)
sinh z dz
Solution
The integrals are all independent of path and can be evaluated by nding a
primitive.
(a)
Z
z5
5
z 4 dz =
=
i
( i)5
5
i)5
(
5
2 5i
:
5
(b)
(c)
Z
ez dz = [ez ]
cos z dz = [sin z]
= sin ( i)
i
i
i
i
=e
( 1) = 0:
sin (
i) =
ei(
i)
e
2i
i( i)
ei(
i)
2 (e
2i
e i( i)
=
2i
e )
=i e
e
(d)
Z
sinh z dz = [cosh z]
i
i
= cosh ( i)
cosh (
e i+e
=
2
12
i) =
i
+e
2
i)
= 0:
IV.2.2
1 2 3 P
L
K
K
111
LLL
246
R
Using an appropriate primitive, evaluate 1=z dz for a path that
travels from
i to i in the right half-plane, and also for a path
from
i to i in the left half-plane. For each path give a precise
denition of the primitive used to evaluate the integral.
Solution
In right halfplane use f (z) = Log z = log jzj + i , where
< <
primitive, get
Z
1
dz = [Log z] i i = log j ij + i
log j i j i
= i:
z
2
2
as a
In left halfplane use f (z) = log0 z = log jzj + i , where 0 < < 2 as a
primitive, get
Z
1
3
dz = [log0 z] i i = log j ij + i
log j i j + i
=
i:
z
2
2
13
IV.2.3
1 2 3 P
L
K
111
LLL KKK
Show that if m 6= 1, then z m has a primitive on Cn f0g.
Solution
The primitive is
z m+1
,
m+1
14
1.
IV.2.4
1 2 3 P L K
p
Let D = Cn ( 1; 1], and consider the branch of z 2 1 on D that is positive
on the interval (1; 1).
(a)
p
Show that z + z 2 1 omits the negative real axis, that is, the range of the
function on D does not include any values in the interval ( 1; 0] on the real
axis.
(b)
p
p
Show that Log z + z 2 1 is a primitive for 1= z 2 1 on D.
(c)
Evaluate
Z
dz
p
;
z2 1
where is the path from 2i to +2i in D counterclockwise around the circle
jzj = 2.
(d)
Evaluate the integral above in the case is the entire circle jzj = 2, oriented
counterclockwise. (Note that the primitive is discontinuous at z = 2.)
Solution
p
D = Cn ( 1; 1], z 2 1
a) p
p
z + z 2 1 = t (t > 0) ) z 2 1 = t z, z 2 1 = t2 2z + z 2 ,
2z = t2 + 1, z = (t2 + 1) =2.
But values of the branch are > 0 on (1; 1) = D \ R. ) It assumes no
negative values.
b)
p
p
Log z + z 2 1 is the composition of z+ z 2 1 and Log w on Cn ( 1; 0],
p
d
so its analytic. By the chain role, dz
Log z + z 2 1 = z+p1z2 1 1 + 21 (z 2 1)
p1
z+ z 2 1
p
of z 2
c)
1+
p z
z2 1
p 1
,
z2 1
15
d a
z
dz
az a
z
1=2
2z =
p
p
p
2i
= Log z + z 2 1 2i = Log 2i + i 5 Log 2i i 5 = i
p
Use fact that p
z 2 1 is positive imaginary
on positive imaginary axis, and
p
also Log 2i + i 5
Log 2i i 5 ?????. d)
p
R dz
2+i0
p
=
Log
z
+
z 2 1 2+i0
2
z 1
p
At 2 + 0t , haveLog z + z 2 1 !
p
p
log 2 + 3 + i Arg z + z 2 1 at z = 2. As z ! 2 + i0, have
p
p
p
p
3,
arg z 2 p1 ! + , jz 2 1j ! 3. ) z + z 2 1 !p 2+( 1) 3 = 2
2
2
2
1 ! . As z ! 2+i0, have arg z
1!
, jz
1j !
arg z + z
p
p
p
p
2
2
3. ) z + z
1 ! 2 + ( 1) 3 = 2
3, arg z + z
1 !
.
Values
coincide
)
p
R dz
2+i0
p
= Log z + z 2 1 2+i0 = log jj log jj + i ( i) = 2 i
z2 1
p dz
z2 1
16
IV.2.5
1 2 3 P L K
Show that
an analytic function f (z) has a primitive in D if and
R
only if
f (z) dx = 0 for every closed path in D.
Solution (A. Kumjian)
Let f be an analytic function dened on a domain D. Suppose rst that f
has a primitive in D. Then by the Fundamental
Theorem of Calculus for
R
Analytic Functions (Part I) we have f (z) dx = 0 for every closed path
2 D (since for anyRsuch path A = B).
Conversely, suppose f (z) dx = 0 for every closed path 2 D. By formula
(1.1) on p. 102, we have
Z
Z
f (z) dx + if (z) dy = f (z) dx = 0
for every close path 2 D. It follows that the integral on the left is path
independent for paths which are not necessarily closed and hence, by the
Lemma on page 77, there is a continuously dierentiable function F on D
@
@
F = f and @y
F = if . It follows that
such that dF = f dx + if dy, that is, @x
the real and imaginary parts of F satisfy the Cauchy-Riemann equations:
@F
@F
@
@
Re F = Re
= Re f = Im if = Im
=
Im F;
@x
@x
@y
@y
@
@F
@F
@
Re F = Re
= Re if = Im f = Im
=
Im F:
@y
@y
@x
@x
Moreover, F 0 (z) = f (z) for all z 2 D. Hence, f has a primitive in D, namely
F.
17
IV.3.1
1 2 3 P
L
K
LLL
2
By integrating e z =2 around a rectangle with vertices R; it R;and
sending R to 1, show that
Z 1
1
2
2
p
e x =2 e itx dx = e t =2 ;
1 < t < 1:
2
1
Use the known value of the integral for t = 0. Remark. This
2
shows that e x =2 is an eigenfunction of the Fourier transform with
eigenvalue 1. For more see the next exercise.
Solution
R
RR
R t ( R+iy)2 =2
R t (R+iy)2 =2
RR
2
z 2 =2
x2 =2
e
dz
=
e
dx
e
idy+
e
idy+
e (x+it) =2 dx =
@D
R
0
0
R
0
p
R1
2
2
e ( R+iy) =2 , 0, uniformly for 0 6 y 6 1 as R ! 1. 2 = 1 e x =2 dx =
R1
R1
2
2
(x+it)2 =2
e
dx
=
e (x +2ixt t )=2 dx =
1R
1
2
2
1
et =2 1 e x =2 e ixt dx
R1
2
2
) p12
e x =2 dx = e t =2
1
18
IV.3.2
1 2 3 P L K
We dene the Hermite polynomials Hn (x) and Hermite orthogonal functions
0 by
n (x) for n
Hn (x) = ( 1)n ex
dn
e
dxn
x2
(x) = e
x2 =2
Hn (x) :
(a)
Show that Hn (x) = 2n xn +
is a polynomial of degree n such that is even
when n is even and odd when n is odd.
(b)
By integrating the function
e(z
dn
2
e z
n
dz
R; it R and sending R to 1, show that
it)2 =2
00
dx =
(x+it)2
+ (2n + 1)
00
n n
1
n
1 dn
e
in dtn
(x+it)2
= 0.
(x) dx = 0;
n 6= m:
PartRA
2
dn
z2
dz
0 = @DR e(z it) =2 dz
ne
R 1 (x it)2 =2 dn x2
R1
2
2
dn
x2
! 1e
dx = 1 e t =2 e itx ex =2 dx
dx =
ne
ne
dx
R
n
n^
t2 =2 1
itx
t2 =2
e
e
( 1)
(x) dx = e
( 1) n (t)
1
R
R 1 x2 =2n dn (x+it)2
R 1 x2 =2 dn
2
!
e
e
dx
=
e
e (x+it) dx =
n
dx
d(it)n
(x+it)\@DR
1
1
2
n R1
n R1
2
2
2
( i) dtd n 1 ex =2 e (x+it) dx = ( i) dtd n 1 ex =2 e 2itx et dx =
Z 1
p
p
2
n
n
2
2
2
( i) dtd n e t
e x =2 e 2itx dx = 2 ( i)n dtd n e t = 2 e t =2 in n (t)
{z
}
| 1
p
2 e
(2t)2 =2
t2 =2
dn
^ (t) = in n e
n
dt
+ ( i)n
p
(t) + ( i)n 2
^ n (t) = ( i)n
n
(x) = e
x2 =2
H2 (x) = e
d
dxn
et
d
dxn
R1
et e
H0 (x) = 1, H1 (x) = e
x2
n (x)
t2
d2
dx2
x2
x2 d
dx
e
x2
=e
x2
4x2 e
x2
x2 =2
dn
dxn
x2
x2
( 2x) e
2e
x2
dx
2ixt
dx
p
= ( i)n 2
2t2
(x+it)2
d t2
e
dxn
=
x2
= 4x2
= 2x
2
R1
2
2 dn
Hn (x) = 2n xn + lower order polynomial. 1 n (x) e itx dx = 1 e x =2 ( 1)n ex dx
e
n
R
R
2
2
2
2
1
1
( 1)n 1 Dxn ex ex =2 ixt dx = 1 e x =2 Dxn ex =2 ixt dx =
R1 x2 n (x it)2 =2 t2 =2
R1
2
2
2
e Dx e
e dx = (i)n 1 e x Dtn e(x it) =2 et =2 dx =
1
R1
R1
2
2
2
2
2
2
et =2 in Dtn 1 e x e(x it) =2 dx = et =2 in Dtn 1 e x =2 e ixt e t =2 dx =
p
2
2
2
2 et =2 in Dtn e t =2 e t =2
R1
20
x2
ixt
IV.3.3
1 2 3 P L K
Let f (z) = c0 + c1 z +
+ cn z n be a polynomial.
(a)
If the ck s are real, show that
Z
f (x) dx
1
2
i
f e
d
=
2
n
X
c2k :
k=0
Hint. For the rst inequality, apply Cauchys theorem th the function f (z)2
separately on the top half an the bottom half of the unit disk.
(b)
If the ck s are complex, show that
Z
1
1
jf (x)j dx
2
i
f e
d
=
2
n
X
k=0
jck j2 :
(c)
Establish the following variant of Hilberts inequality, that
n
X
cj ck
j+k+1
j;k=0
n
X
k=0
jck j2 ;
with strict inequalityRunless the complex numbers c0 ; : : : ; cn are all zero. Hint.
1
Start by evaluating 0 f (x)2 dx.
Solution
(a)
R1
R
R
2
f (x)2 dx =
f (z)2 dz 6 f ei
d
1
0
1
R1
R
R2
2
f (x)2 dx =
f (z)2 dz 6
f ei
d
1
1
R1
R
2
2
Add, get 2 1 f (x)2 dx =6
f ei
d
n
n
n
R2
P 2
P
P
2
2
i
i
ik
f
e
d
=
2
c
,
by
from
f
e
=
c
e
cm e im
k
k
0
k=0
k=0
k=0
P
P
P
P
2
2
2
(b) f (x)2 =
ak x k + i b k x k =
ak x k +
bk xk
R1 P
R1
R1 P
P
P
2
k 2
k 2
k
f
(x)
dx
=
a
x
dx+
b
x
dx
=
a
x
+
bk x k =
k
k
k
1
1
1
P
k
ck x
21
hR
1
1
d
2
ak eik
R1 P
1
bk eik
d
2
n
R1 P
R1 P
2
cj ck xj+k dx =
(c) 0 ( cj xj ) dx = 0
R1
j;k=0
R1
"
n
P
j;k=0
xj+k+1
cj ck j+k+1
#1
0
n
P
j;k=0
cj ck
j+k+1
jj 6 0 jf (x)j dx 6 0 jf (x)j dx 6
jck j , for equality, must have
R1
R1
2
2
jf (x)j = 0 jf (x)j , so f (x) = 0, for 1 6 x 6 0, and then f (x) 0
0
Suppose P (z) = a0 + a1 z + ::: + an z n is a polynomial. By integrating
P (z)2 log z, for an appropriate branch of log z, around a keyhole contour,
n
n
R1
R
P
P
aj ak
2
show that
P
(x)
dx
6
jak j2 .
P (z)2 log z dz = 0,
=
j+k+1
R1
j;k=0
2
k=0
@D
R1
R2
2
gives 0 P (z) log x dx 0 P (z)2 (log x + 2 i) dx+ 0 P ei
iei d = 0.
R1
R
R
2
Get 2 i 0 P (x)2 dx = i 0 P ei ei d . Note that P ei d , so latter
R2
R1
R2
2
integral is i 0 P ei ei (
) d . Thus 0 P (x)2 dx = 0 P ei ei (
n
R1
R2
P
2 d
i
P
e
) 0 P (x)2 dx 6
=
jak j2
2
0
k=0
22
) 2d
IV.3.4
1 2 3 P L K
Prove that a polynomial in z without zeros is constant (the fundamental
theorem of algebra) using Cauchys theorem, along the following lines. If
P (z) is a polynomial that is not a constant, write P (z) = P (0) + zQ (z),
divide by zP (z), and integrate around a large circle. this will lead to a
contradiction if P (z) has no zeros.
Solution
P (z) is a polynomial with no zeros. Assume P (0) = 1 . Write p (z) =
1 + zQ (z).
H P (z)
H
H
H
Q(z)
1
1
1
dz
=
dz
=
+
dz
=
dz2 = i
zP (z)
zP (z)
P (z)
zP (z)
jzj=R z
jzj=R
jzj=R
z=R
1
Rn
! 0, if R ! 1, if n > 1.
23
jzj=R
IV.3.5
1 2 3 P L K
Suppose that D is a bounded domain with piecewise smooth boundary, and
that f (z) is analytic on D [ @D. Show that
sup jz
f (z)j
z2@D
Area (D)
:
Length (D)
f (z)j >
z2@D
Area(D)
2 Length(@D)
Area(D)
2 Length(@D)
=2
=1
24
f (z)j = 1
IV.3.6
1 2 3 P L K
Suppose f (z) is continuous in the closed disk
H fjzj Rg and analytic
on the open disk fjzj < Rg. Show that jzj=R f (z) dz = 0. Hint.
Approximate f (z) uniformly by fr (z) = f (rz).
Solution (A. Kumjian)
H
H
To prove jzj=R f (z) dz = 0 it su ces to show that jzj=R f (z) dz
", for
every " > 0. So let " > 0 be given. Since f is continuous in the closed disk
fjzj Rg, it must be uniformly continuous there (since the close disk is both
closed and bounded). Hence, there is a > 0 such that for all z1 ; z2 in the
closed disk, we have
jf (z1 )
f (z2 )j <
"
2 R
if
jz1
z2 j < :
For r 2 (0; 1), dene fr on the disk fz : jzj < R=rg by fr (z) = f (rz) for all
z such that jzj < R=r. Then since
H fr is analytic on a domain containing the
closed disk fjzj Rg, we have jzj=R f (z) dz = 0 by Cauchys theorem. Now
let r be chosen such that 1
=R < r < 1. Then for all z with jzj = R we
have jz zrj = R (1 r) < and so by the choice of we have
jf (z)
fr (z)j = jf (z)
f (rz)j <
"
:
2 R
j(f (z)
Hence we have
I
jzj=R
f (z) dz =
(f (z)
fr (z)) dz
jzj=R
jzj=R
25
fr (z))j jdzj
"
2 R = ":
2 R
IV.4.1
1 2 3 P
L
K
111
LLL
Evaluate
H then following integrals, Husingezthe Cauchy integral formula:
(a) jzj=2 zz 1 dz; n 0
(e) jzj=1 zm dz;
1<m<1
H
H
Log z
zn
(f) jz 1 ij=5=4 (z 1)2 dz
(b) jzj=1 z 2 dz; n 0
H
H
sin z
(c) jzj=1 z dz
(g) jzj=1 z2 (z2dz 4)ez
H
H
dz
z
dz
(h)
(d) jzj=1 cosh
3
z
jz 1j=2 z 2 (z 2 4)ez
Solution
We rearragnde the Cauchy integral formula. Let D be bounded domain with
piecewise smooth boundary @D and let f be an analytic function dened in
a domain which contains D. Then for any z0 2 D and positive integer m we
have,
Z
dm
f (z)
2 i
f (z)
:
dz
=
z0 )m
(m 1)! dz m
@D (z
z=z0
(a)
The nominator has one zero at z = 1 inside jzj = 2, thus by by Cauchys
theorem,
H
zn
jzj=2
dz =
2 i n
[z ]z=1 = 2 i:
0!
(b)
The nominator has no zeros inside jzj = 1, thus by by Cauchys theorem,
H
zn
jzj=1
dz = 0:
(c)
The nominator has one zero at z = 0 inside jzj = 1, thus by by Cauchys
theorem,
H
jzj=1
sin z
2 i
dz =
[sin z]z=0 = 0:
z
0!
(d)
The nominator has one tripel zero at z = 0 inside jzj = 1, thus by by
Cauchys theorem,
26
cosh z
2 i d2
dz =
cosh z
jzj=1
z3
2! dz 2
= i:
z=0
(e)
The nominator has one tripel zero at z = 0 inside jzj = 1, thus by by
Cauchys theorem.
If m 0, then the integrand z m ez denes an analytic function on C, hence,
by Cauchys Theorem the integral must be zero.
If m 1, the nominator has a zero of order m at z = 0 inside jzj = 1, hence
by by Cauchys theorem.
I
2 i
dm 1 z
2 i
ez
dz
=
e
:
=
m
m
1
(m 1)! dz
(m 1)!
jzj=1 z
z=0
Thus
jzj=1
ez
dz =
zm
0
2 i
(m 1)!
if m
if m
0;
1:
(f)
Let f be the principal branch of the logarithm dened on the slit plane, that
is, f (z) = Log z for all z 6= 0 with Arg z 6=
. Then f is analytic on a
domain jz 1 ij 5=4. The nominator has one double zero at z = 1 inside
jz 1 ij < 5=4, thus by by Cauchys theorem.
H
2 i d
Log z
(Log z)
2 dz =
1! dz
(z 1)
jz 1 ij=5=4
= 2 i:
z=1
(g)
The nominator has one double zero at z = 0 inside jzj = 1, thus by by
Cauchys theorem,
H
dz
jzj=1 2
z
(z 2
4) ez
2 i d
1! dz
e
z2
=
z=0
i
:
2
(h)
The nominator has one zero at z = 0, and one zero at z = 2 inside jz
thus by by Cauchys theorem,
I
jz 1j=2
dz
z
(z 2
4) ez
2 i
1
2
0! (z
4) ez
27
+
z=0
2 i
1
0! z (z + 2) ez
1j = 2,
=
z=2
i
i
+ 2:
2 4e
IV.4.2
1 2 3 P
L
K
LLL
Show that a harmonic function is C 1 , that is, a harmonic function
has partial derivatives of all orders.
Solution
@
@v
@
u = Re f 0 , @y
u = @u
= @x
= Re (f 0 ).
Write u = Re f , and note that @x
@y
@v
v = Im f = Re ( if ), @x
= Re ( if ). Now take successive derivatives, get
m
@ u
0
a multiple of Re f is Re (if 0 ).
@xp @y q
28
IV.4.3
1 2 3 P
L
K
LLL
Use the Cauchy integral formula to derive the mean value property
of harmonic functions, that
Z 2
d
;
z0 2 D;
u (z0 ) =
u z0 + ei
2
0
whenever u (z) is harmonic in a domain D and the closed disk
jz z0 j
is contained in D.
Solution
H
Write f (z0 ) = jz
f (z)
dz
z0 j= z z0
H2
0
f z0 + ei
29
d
2
IV.4.4
1 2 3 P L K
Let D be a bounded domain with smooth boundary @D, and let z0 2 D.
Using the Cauchy integral formula, show that there is a constant C such
that
jf (z0 )j
1
2 i
f (z)
dz
@D z z0
1
jz z0 j
1
dist(z0 ;@D)
n
= d1 .
L 1=n
.
2 d
30
IV.5.1
1 2 3 P
L
K
111
LLL
Denna beh
over N og f ixas till Lite
Show that if u is a harmonic function on C that is bounded above,
then u is constant.
Hint. Express u as the real part of an analytic function, and exponentiate.
Solution
Let u be the real part of the analytic function f = u + iv, where we know
that u C, for some constant C. Then set g = eu+iv where g is an entire
function. Now we have that
jgj = eu+iv = eu eiv = jeu j eiv = eu :
We can se that becauce u is bounded above, so is g bounded above. Now
we now that g is both analytic and bounded, by Liouvilles theorem g is
constant. Thus f is constant.
31
IV.5.2
1 2 3 P
L
K
111
LLL
Show that if f (z) is an entire function, and there is a nonempty
disk such that f (z) does not attain any values in the disk, then
f (z) is constant.
Solution
If f (z) does not attain values in the disk jw cj < ", then 1= (f (z) c) is
a bounded entire function, hence constant by Liouvilles theorem, and f (z)
is constant.
Let w0 be the center of the disk that f omits. Then 1= (f (z) w0 )
32
IV.5.2
1 2 3 P
L
K
LLL
Show that if f (z) is an entire function, and there is a nonempty
disk such that f (z) does not attain any values in the disk, then
f (z) is constant.
1
for all z 2 C:
f (z) z0
33
34
c) is bounded,
IV.5.3
1 2 3 P
L
K
LLL
A function f (z) on the complex plane is double periodic if there
are two periods w0 and w1 of f (z) that do not lie on the same line
thought the origin (that is, w0 and w1 are linearly independent over
the reals, and f (z + w0 ) = f (z + w1 ) = f (z) for all complex numbers
z). Prove that the only entire functions that are doubly periodic
are the constants.
Solution
The "lattice" of points mw0 +nw1 , m; n 2 Z are the corners of parallelograms
that cover C, each parallelogram being a translate F + mw0 + nw1 of "the
fundamental region" F as shown in the gure:
Figure IV.5.3
Thus each 2 C can be written = z + mw0 + nw1 with z 2 F . Since we
assume f ( ) = f (z), it follows that f i bounded, thus a constant.
35
36
IV.5.4
1 2 3 P L K
Suppose that f (z) is an entire function such that f (z) =z n is bounded
for jzj
R. Show that f (z) is a polynomial of degree at most n.
What can be said if f (z) =z n is bounded on the entire complex
plane?
Solution
Apply the Cauchy estimates for f (m+1) (z) to disk jz z0 j < R, and let
R ! 1, to obtain f (m+1) (z0 ) = 0. If f (z) is a polynomial of degree 6 m
such that f (z) =z m is bounded near 0, then f (z) = cz m .
37
IV.5.5
1 2 3 P L K
Show that if V (z) is the velocity vector eld for a uid ow in the entire
complex plane, and if the speed jV (z)j is bounded, then V (z) is a constant
ow.
Solution
Use V (z) = f 0 (z) (section III.6 p92), when f is entire. Get f 0 constant, by
Liouvilles theorem, i.e. V (z) is constant.
38
IV.6.1
1 2 3 P
L
K
LLL
Let L be a line in the complex plane. Suppose f (z) is a continuous
complex-valued function on a domain D that is analytic on DnL.
Show that f (z) is analytic on D.
Solution
Rotate and apply result in text.
39
IV.6.2
1 2 3 P L K
Let h (t) be a continuous function on the interval [a; b]. Show that the Fourier
transform
Z b
h (t) e itz dt
H (z) =
a
CeAjyj ;
z = x + iy 2 C;
40
IV.6.3
1 2 3 P L K
Let h (t) be a continuous function on a subinterval [a; b] of [0; 1). Show
that the Fourier transform H (z), dened as above, is bounded in the lower
half-plane.
Solution
41
IV.6.4
1 2 3 P L K
Let be a smooth curve in the plane R2 , let D be a domain in the complex
plane, and let P (x; y; ) and Q (x; y; ) be continuous complex-valued functions dened for (x; y) on and 2 D. Suppose that the functions depend
analytically on for each xed (x; y) on .
Show that
Z
G ( ) = P (x; y; ) dz + Q (x; y; ) dy
is analytic on D.
Solution
Reduce to Riemann integral and apply theorem in the text.
42
IV.7.1
1 2 3 P L K
Find an application for Goursats theorem in which it is not patently clear
by other means that the function in question is analytic.
Solution
Answer to this is not known (by me).
43
IV.8.1
1 2 3 P
L
K
LLL
Show from the denition that
@
@
@
z = 1;
z = 0;
z = 0;
@z
@z
@z
@
z
@z
= 1:
Solution
We use the rst-order dierential operators (page 124)
@
1 @
=
@z
2 @x
@
@y
and
@
1 @
@
=
+i
:
@z
2 @x
@y
We have that
@
z
@z
@
z
@z
@
z
@z
@
z
@z
1
2
1
=
2
1
=
2
1
=
2
=
@
@
i
@x
@y
@
@
+i
@x
@y
@
@
i
@x
@y
@
@
+i
@x
@y
1
(1 + 1) = 1
2
1
(x + iy) = (1 1) = 0
2
1
(x iy) = (1 1) = 0
2
1
(x iy) = (1 + 1) = 1
2
(x + iy) =
44
IV.8.2
1 2 3 P
L
K
LLL
Compute @@z (az 2 + bz z + cz 2 ). Use the result to determine where
az 2 + bz z + cz 2 is complex-dierentiable and where it is analytic.
(See Problem II.2.3)
Solution
By the Leibnitz rule and the results of Exercise 1, the z derivative is bz+2cz.
The function is complex dierentiable at z if and only if bz + 2cz = 0. If
b = c = 0, the function is entire. Otherwise this locus is either a point f0g
or a straight line through the origin, and there is no open set on which the
function is analytic.
45
IV.8.3
1 2 3 P
L
K
LLL
Show that the Jacobian of a smooth function f is given by
@f
@z
det Jf =
Solution
Jf = det
@f 2
@z
1
4
1
4
@u
@x
@u
@y
@f 2
=
@z
@u
@x
@v
2 @u
@x @y
@v
@y
@v
@x
@v
@y
1
4
2
@u @v
@x @y
=
@(u+iv)
@x
@v
@x
@v @u
@v
2 @x
+ 2 @u
@y
@x @y
@f
@z
@u @v
@y @x
i @(u+iv)
@y
@u
@y
1
4
@u
@x
1
4
i
@v @u
2 @x
=
@y
46
@(u+iv)
@x
@v
@y
@u @v
@x @y
+ i @(u+iv)
@y
2
@v
@x
@v @u
@x @y
@u
@y
= det Jf
IV.8.4
1 2 3 P
L
K
LLL
Show that
@2
@2
@2
:
+
=
4
@x2 @y 2
@z@ z
Deduce the following, for a smooth complex-valued function h.
(a) h is harmonic if and only if @ 2 h=@z@ z = 0.
(b) h is harmonic if and only if @h=@z is conjugate-analytic.
(c) h is harmonic if and only if @h=@ z = 0.
(d) If h is harmonic, then any mth order partial derivative of h is
a linear combination of @ m h=@z m and @ m h=@ z m .
Solution
@
@
@2
= @x
i @y
4 @z@
z
(a) clear
(b) h harmonic ,
@
@x
@2h
@ z@z
@2h
@z@ z
@
+ i @y
=
=0,
@h
@z
@
@z
@2
@x2
@
i @x@y
@
i @y@x
+
@2
@y 2
@2
@x2
@2
@y 2
is analytic
@h
=0,
= 0. Since @h
= @@hz , this ????? @h
is
(c) h harmonic ,
@z
@z
@z
@h
analytic , @ z is conjugate analytic.
@ k+l h
(d) If h is analytic, then @z
k @ z l = 0 unless l = 0. Only non-vanishing derivak+l
@mh
m
tive is @zm = h (z). If h is conjugate analytic, ????? @@zk @zhl = 0 unless l = 0.
If h is harmonic, then h = analytic + conjugate analytic, so all derivatives
l
k
of h vanish except for @@zhk = @@zhk .
47
R
d
Log zjz=1 = 2 i z1 z=1 = 2 i
(f) jz 1 ij=5=4 (zLog1)z2 dz = 21!i dz
R
( e z )(z2 4) e z (2z)
d e z
(g) jzj=1 z2 (z2dz 4)ez = 2 i dz
=2 i
z2 4
(z 2 4)2
z=0
2 i4
16
(h)
2 i
i
2
dz
=2
jz 1j=2 z(z 2 4)ez
1
1
+ 8e2 = i 12
4
i (z2
+
1
4)ez
1
4e2
z=0
1
+ 2 i (z+2)e
z
48
=
z=2
z=0
IV.8.5
1 2 3 P
L
K
LLL
With dz = dx idy, show that for a smooth function f (z) that
df =
Solution
dx +
df = @f
@x
@f
dz
@z
@f
dy
@y
= 12 @f
@x
i @f
(dx + idy) + 21
@y
h
1 @f
1 @f
1 @f
i @f
1 @f
dx
+
dy
+
dx
+
dy
+
dy
2 @x
2 @y
2 @x
2 @y
2 @x
+
@f
dz
@z
@f
@f
dz +
:
@z
@z
49
@f
@x
+ i @f
(dx
@y
@f
dx
@y
idy) =
i
@f
@f
dy
+
dx
= df
@x
@y
IV.8.6
1 2 3 P L K
Show that if D is a domain with smooth boundary, and if f (z) and g (z) are
analytic on D [ @D, then
ZZ
Z
f (z) g (z) dz = 2i
f (z) g 0 (z) dx dy:
@D
Solution R
RR @
RR
By (8.4), @D f (z) g (z) dz = 2i D @E
(f g) dz = 2i D f gdz
@g
@f
@g
0
Using the Leibnitz rule, obtain @@z (f g) =
R f @z + g @z =
RRf @z = f g
To get Ex1.4, take f = 1, g (z) = z, get @D z dz = 2i D dxdt = 2iArea
50
IV.8.7
1 2 3 P
L
K
LLL
Show that the Taylor series expansion at z0 = 0 of a smooth function
f (z), through the quadratic terms, is given by
@f
1 @2f
@f
@2f
2
(0) z+
(0) z+
(0) jzj2 +o jzj3 :
f (z) = f (0)+
(0) z + 2
2
@z
@z
2 @z
@z@ z
Solution
Since a smooth function f is a linear combination of the functions 1; z; z; z 2 ; z 2 ; jzj2
and a reminder term O jzj3 , it su ces to check the formula for these six
functions. The formula holds fro each of these. For instance, for jzj2 = z z
2
@2f
are harmonic. @f
(0) = 0, @f
(0) = 0, @@ zf2 (0) = 0, f (0) = 0, @z@
(0) = 1.
@z
@z
z
3
3
The term O jzj has Taylor series 0 + O jzj . That ????? it.
51
IV.8.8
1 2 3 P
L
K
LLL
Establish the following version of the chain rule for smooth complexvalued functions w = w (z) and h = h (w).
@h @w
@h @ w
@
(h w) =
+
;
@z
@w @ z
@w @z
@
@h @w
@h @ w
(h w) =
+
:
@z
@w @z
@ w @z
Solution
@h
Suppose w (0) = 0, h (0) = 0, write h (w) = aw + bw + O jwj2 , a = @w
(0),
2
@w
@w
@h
b = @ w (0) and w (z) = w + w + O jzj , = @z (0), b = @ z (0) then
h (w (z)) = a z + z + O jzj2 + b z + z + O jzj2 + O jwj2 =
a z + a z + b z + b z + O jzj2 = a + b z + (a + b ) z + O jzj2
This gives,
@h w
@h @w
@h @w
(0) = a + b = @w
+ @@hw @w
= @w
+ @@hw @@wz
@z
@z
@z @ w=@ z
@z
@h w
@z
(0) = a + b =
@h @w
@w @z
@h
@w
@w
=
@z
|{z}
@ w=@z
52
@h @w
@w @z
@h @ w
@ w @z
IV.8.9
1 2 3 P
L
K
LLL
Show with the aid of the preceding exercise that if both h (w) and
w (z) are analytic, then (h w) (z) and (h w)0 (z) = h0 (w (z)) w0 (z).
Solution
If the functions h and w are analytic, then @h@ zw = 0 + 0 = 0 so h
@h @w
@
h w = @w
+ 0 = h0 (w (z)) w0 (z)
analytic. (h w)0 (z) = @z
@z
53
w is
IV.8.10
1 2 3 P L K
Let g (z) be a continuously dierentiable function on the complex plane that
is zero outside of som compact set. Show that
ZZ
1
@g 1
dx dy;
w 2 C:
g (w) =
w
C @z z
54
V 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1
2
3
4
5
6
7
8
V.1.1
1 2 3 P L K
(Harmonic Series) Show that
n
X
1
k=1
P1
1
k=1 k
1
k
log n:
k+1
1
dx:
x
Solution
Let k be a nonnegative integer. Then since the function x 7! 1=x is decreasing
on the positive reals, we have for all x 2 [k; k + 1] that 1=x 1=k . Hence
locking at the oversum in the interval [k; k + 1]
Z k+1
1
1
1
((k + 1) k) =
dx:
k
k
x
k
Therefore, for n
n
X
1
k=1
n Z
X
k=1
k+1
1
dx =
x
n+1
1
dx = log (n + 1)
x
log n:
V.1.2
1 2 3 P L K
Show that if p < 1, then the series
Exercise 1 and the comparison test.
P1
k=1
Solution
If k > 1 then, k p < k for 0 < p < 1, then
1
1
> :
p
k
k
Thus we have
1
1
X
X
1
1
>
:
p
k
k
k=1
k=1
P
By the comparison test, we have that
since the sum nk=1 1=k diverge as
Pthe
n ! 1 by Exercise 1, also the sum nk=1 1=k p diverge as n ! 1, which was
to be shown.
V.1.3
1 2 3 P L K
P1
k=1
n
X
1
S
<
p
k
(p
k=1
Rk
Hint. Use the estimate k1p < k 1 dx
.
xp
1
1) np
Solution
Let k be a nonnegative integer. Then since the function x 7! 1=x is decreasing
on the positive reals, we have for all x 2 [k 1; k] that 1=k 1=x. Becauce
p > 1, it follows that 1=k p
1=xp . Hence locking att the undersum in the
interval [k 1; k]
Z k
1
1
1
(k (k 1)) = p
dx
p
p
k
k
k 1 x
Therefore
n
X
1
kp
k=1
n Z
X
k=1
k 1
Z n
dx
x p+1
dx
= 1+
= 1+
p
xp
p+1
1 x
= 1+
1
1
p
n
1 p
p+1
n
p
p+1
1+
1
p
p
p
x p+1
p+1
as N ! 1.
Thus we have
=
n
p+1
N
1
p+1
1+
1
p
thus
N
X
1
=S
kp
k=n+1
n
X
1
n
6
p
k
p
k=1
N
X
1
kp
k=n+1
(p
p+1
1
1) np
(p
1
1) np
V.1.4
1 2 3 P L K
Show that the series
1
X
( 1)k+1
=1
k
k=1
1 1
+
2 3
1
+
4
converges. Hint. Show that the partial sums of the series satisfy
S2 < S 4 < S 6 <
< S5 < S3 < S1 .
Solution
We dene the sum Sn by
Sn =
n
X
( 1)k+1
=1
k
k=1
1 1
+
2 3
1
+ ::::
4
This is the alternating series test, which is standard signs alternate, and
terms ! 0, jtermsj # , so series converge.
V.1.5
1 2 3 P L K
Show that the series
1
1
1 1 1 1 1 1
+ +
+ +
+
3 2 5 7 4 9 11 6
converges to 3S=2, where S is the sum of the series in Exercise 4.
(It turns out that S = log 2.) Hint. Organize the terms in the series
in Exercise 4 in groups of four, and relate it to the groups of three
in the above series.
1+
Solution
We have the series S from exercise V.1.5 and arrange the terms in groups of
four
1
1
1 1 1 1 1 1 1 1
+
+
+
+
+
2 3 4 5 6 7 8 9 10 11
And now we arrange the terms in groups of four
S=1
S= 1
1 1
+
2 3
1
1
+
4
5
1 1
+
6 7
1
1
+
8
9
1
1
+
10 11
1
+ :::
12
1
+ :::
12
Now we rearrange the terms in groups of two, and divide the terms by 2
S
1
=
2
2
1
1
+
4
6
1
1
+
8
10
1
+ :::
12
Now add one group of four from the sum S with a group of two from the
sum S=2, and so on
3
S =
2
=
1
3
1
1+
3
1+
1
1 1
1
1 1
1
+
+
2
+ +
2
+ :::+ =
4
5 7
8
9 11
12
1
1 1 1
1
1
1
+
+
+
+
+ :::
2
5 7 4
9 11 6
V.1.6
1 2 3 P L K
Show that
Z R
1
= [log (log k)]R
e = log (log R)
e k log k
P
1
diverges while 1
k=1 k(log k)2 converges.
log e (log e) ! 1
1
k log k
k=2n +1
there are 2n terms between k = 2n +1 and k = 2n+1 , each of the terms greater
1
1
than 2n+1 log
, so these 2n terms have sum greater than 2(n+1)
. Thus
2n+1
log 2
series diverges, by comparison with the harmonic series. The other series is
treated similarly, using an upper estimate.
For the other sum
n+1
2X
1
k (log k)2
k=2n +1
there are 2n terms between k = 2n + 1 and k = 2n+1 , each of the terms less
1
1
, so these 2n terms have sum less than 2(n+1)
. Thus series
than 2n+1 log
2n+1
log 2
converges, by comparison with the harmonic series.
Solution 2 (Integral Test)
Note that x log x and x (log x)2 are increasing for x > 1, so the terms of both
series are positive and decreasing. Vi can use Cauchys integralkriterium and
look att them corresponding integrals. Becauce of di cult with the integrand
we split the integral into two parts
Z R
1
= [log (log x)]R
log (log e) ! 1;
e = log (log R)
e x log x
8
as R ! 1.
1
=
x (log x)2
1
log x
=
e
1
log e
as R ! 1.
We have that the series
1
X
k=1
1
X
k=1
konverges.
1
k log k
1
k (log k)2
1
! 1;
log R
V.1.7
1 2 3 P L K
Show that the series
Solution
n
P
If Sn is the partial-sum
ak , then Sn
diately from
n
P
k=1
ak = S n
Sm 1 , when Sn =
n
P
n
P
1
k=m k(log k)2
converges.
k=m
k=1
k=m
the series.
d
xk
V.2.1 dx
=
k+x2k
Sm
Pk=n
(k+x2k ) xk (2k)x2k 1
= 0 at k 2 xk 1 + kx3k 1 = 2kx3k 1 ,
2
2k
(1+x
)
p
Function " for 0 6 x 6 k 1=2k , # for
k 2 xk 1 = k 3k 1 , k = x2k , k = xk . )
p
k
x
k
1
p
x > k 1=2k , ! 0 as x ! 1. k+x
= worst-case estimation. )
2k 6 k+k =
2 k
Converge to 0 uniformly.
kxk
10
V.2.1
1 2 3 P L K
Show that fk (x) = xk = k + xk converges uniformly to 0 on [0; 1).
Hint. Determine the worst-case estimator "k by calculus.
Solution
Suppose fk (x) converge pointwise to a function f (x) as k ! 1.
We have
fk (x) =
xk
;
k + x2k
thus
0
fk (x) =
kxk
k + x2k
xk (2k) x2k
(k + x2k )2
11
V.2.2
1 2 3 P L K
Show that gk = xk = 1 + xk converges pointwise on [0; 1) but not
uniformly. What is the limit function? On which subsets of [0; 1)
does the sequence converge uniformly?
Solution
Suppose gk (x) converge pointwise to a function g (x) as k ! 1, we start to
show that is is an increasing function
0
gk (x) =
kxk
xk kxk
(1 + xk )2
1
kxk 1
>0
(1 + xk )2
also is the function gk (x) increasing, now we take the limit in three intervals.
If 0 x < 1, in this interval we have that xk ! 0 as k ! 1, and we have
xk
0
=
= 0:
k
k!1 1 + x
1+0
g (x) = lim
If x = 1, then
1
1
1k
=
= :
k
k!1 1 + 1
1+1
2
k
If x > 1, in this interval we have that x ! 0 as k ! 1, and we have
g (1) = lim
xk
1
1
g (x) = lim
= lim
=
= 1:
k!1 1 + xk
k!1 1=xk + 1
0+1
We also have that gk (x) ! g (x), where
8
< 0 for 0 x < 1;
1
for x = 1;
g (x) =
: 2
1 for x > 1
Convergence is uniform on [0; 1 "] for any " > 0. Not uniform on [0; 1],
because limit function is not continuous at 1. Convergence is uniform on
[1 + "; +1] for any " > 0.
12
V.2.3
1 2 3 P L K
Show that fk (z) = z k =k converges uniformly for jzj < 1. Show that
0
fk (z) does not converge uniformly for jzj < 1. What can be said
0
about uniform convergence of fk (z)?
Solution
Suppose fk (z) converge pointwise to a function f (z) as k ! 1 in the interval
jzj 1, in this domain we have that z k =k
1=k thus
f (z) = lim
k!1
zk
k
1
= 0:
k!1 k
lim
f (z) = lim z k
k!1
If jzj = 1, then fk (z) = 1 so the series can not converge uniformly for
0
jzj < 1. But for any " > 0 the series fk (z) = z k 1 converges uniformly for
jzj 1 ".
13
V.2.4
1 2 3 P L K
Show that
1
X
1 xk
k 2 1 + x2k
k=1
Solution
We
= k12 , where the sum
P1apply the Weierstrass M test (p. 135.)k with Mk 2k
1 + x for all 1 < x <
k=0 Mk converges. First we show that x
+1.
We have
xk
xk
1 1 + x2k = 1 + x2k
x2k < 1 + x2k = 1 + x2k
for jxj 1
for jxj > 1
1
= Mk :
k2
14
V.2.5
1 2 3 P L K
For which real numbers x does
P1
xk
k 1+x2k
converge?
Solution
Suppose gk (x) converge pointwise to a function g (x) as k ! 1, and take
limit in three intervals
If x < 1, in this interval we have that xk ! 0 as k ! 1, and we have
X
X 1
X
1
xk
xk
<
<
=
k + xk
k
k 1 + x2k
1=x
1=x
k=1
k=1
k=1
1
X
1
k=1
X 1
xk
=
k 1 + x2k
2k
k=1
1
X
1
k=1
1
X
1
k=1
15
V.2.6
1 2 3 P L K
Show that for each " > 0, the series
for x 1 + ".
P1
xk
k 1+x2k
converges uniformly
Solution
We have
xk
;
1 + x2k
fk (x) =
thus
0
fk (x) =
kxk
1 + x2k
xk (2k) x2k
(1 + x2k )2
xk 1 x3k 1
(1 + x2k )2
1
X
1
k=1
xk
k 1 + x2k
1
X
1
k=1
1
X
k=1
(1 + ")k
2k
k 1 + (1 + ")
1
1
k (1 + ")k
16
1
X
1
k=1
1
k
k 1= (1 + ") + (1 + ")k
V.2.7
1 2 3 P
L
K
LLL
Let an be aPbounded sequence of complex numbers. Show that for each " > 0,
z
the series 1
converges uniformly for Re z 1 + ". Here we choose
n=1 an n
the principal branch of n z .
Solution
1
P
an n z , jan j 6 C, Re z > 1 + ". Apply Weierstrass M-test, jan n z j 6
n=1
P
P
C
Cn Re z 6 n1+"
= Mn ,
Mn < 1 )
an n z converges uniformly for
Re z > 1 + ".
17
V.2.8
1 2 3 P
L
K
LLL
P zk
Show that
converges uniformly for jzj < 1.
k2
Solution
Apply the Weierstrass M
1
1
1
X
X
zk
jzjk X 1
=
<
;
k2
k2
k2
k=1
k=1
k=1
18
1, we nd that
V.2.9
1 2 3 P
L
K
LLL
P zk
Show that
does not converges uniformly for jzj < 1.
k
Solution
Suppose that we have uniform convergence for jzj < 1, then the series must
converge for jzj 1
Apply the Weierstrass M
test with M = 1=k in jzj 1, we nd that
1
1
1
X
X
jzjk X 1
zk
=
<
;
2
k
k
k
k=1
k=1
k=1
19
V.2.10
1 2 3 P L K
Shoe that if a sequence of functions ffk (x)g converges uniformly on Ej for
1
j
n, then the sequence converges uniformly on the union E = E1 [
E2 [
[ En .
Solution
Use Cauchy criterion. Let " > 0. Choose Nj such that jfn (x) fk (x)j < k
for m; k > Nj and x 2 Ej . Let N = max fN1 ; : : : ; Nn g. Then jfm (x) fk (x)j <
" for m; k > N and x 2 E1 [ : : : [ En . ) ffm g converges uniformly.
20
V.2.11
1 2 3 P L K
Suppose that E is a bounded subset of a domain D C at a positive distance
from the boundary of D, that is > 0 such that jz wj
for all z 2 E
and w 2 CnD. Show that E can be covered by a nite number of closed
disks contained in D. Hint. Consider all closed disks with centers at points
(m + ni) =10 and radius =10 that meet E.
Solution
Follow the hint. There are only nitely many of the disks in ????? (because
E is bounded); and they cover E, because the collection of all disks centred at
(m + ni) =10 of radius =10 covers the complex plan, and they are contained
in D, because dist ("; @D) > .
21
V.2.12
1 2 3 P L K
Let f (z) be analytic on a domain D, and suppose jf (z)j M for all z 2 D.
Show that for each > 0 and m
1, f (m) (z)
m!M= m for all z 2 D
whose distance from @D is at least . Use this to show that if ffk (z)g is a
sequence of analytic functions on D that converges to f (z) on D, then for
(m)
each m the derivatives fk (z) converge uniformly to f (m) (z) on each subset
of D at a positive distance from D.
Solution
Use the Cauchy estimates for f (m) (w) on a disk centred at w of radius
d (w; @D), get f (m) (w) 6 m!m
m .
22
<
V.3.1
1 2 3 P
L
K
LLL
Find the radius of convergence of the following power series:
1
1
1
P
P
P
3k z k
kk
(a)
2k z k (d)
(g)
zk
k
k
4 +5
1+2k kk
(b)
(c)
k=0
1
P
k=0
1
P
k k
z
6k
2 k
k z
(e)
(f)
k=0
1
P
k=1
1
P
k=1
k=1
2k z 2k
k2 +k
(h)
z 2k
4k k k
(i)
k!1
k=3
1
P
k=1
Solution
(a)
We apply the ratio test, thus
R = lim
k=1
1
P
(log k)k=2 z k
k!z k
kk
2k
1
1
ak
= lim k+1 = lim = :
k!1 2
k!1 2
ak+1
2
(b)
We apply the ratio test, thus
R = lim
k!1
ak
k 6k+1
6k
6
= lim k
= lim
= lim
= 6:
k!1 6 k + 1
k!1 k + 1
k!1 1 + 1=k
ak+1
(c)
We apply the ratio test, thus
R = lim
k!1
ak
k2
k2
1
= lim
= lim
=
lim
= 1:
2
2
k!1 (k + 1)
k!1 k + 2k + 1
k!1 1 + 2=k + 1=k 2
ak+1
(d)
We apply the ratio test, thus
R = lim
k!1
ak
3k 4k+1 + 5k+1
= lim k
=
k!1 4 + 5k
ak+1
3k+1
3k 5k+1
= lim
k!1
3k+1 5k
4 k+1
5
4 k
5
+1
+1
23
= lim
k!1
4 k+1
5
4 k
5
+1
+1
5
= :
3
(e)
We apply the ratio test, thus
R = lim
k!1
ak
2k z 2k (k + 1)2 + k + 1
= lim 2
= lim
k!1 k + k
k!1
ak+1
2k+1 z 2(k+1)
= lim
k!1
1 + 2=k
1
p
The convergence radius R = 1= 2.
1
=
2 jzj2
k+2
k
0;
1;
1
=
2 jzj2
2 jzj2 < 1
:
2 jzj2 > 1
(f)
We apply the ratio test, thus
ak
z 2k 4k+1 (k + 1)k+1
4 (k + 1) (k + 1)k
R = lim
= lim k k
= lim
=
k!1 ak+1
k!1 4 k
k!1
z 2(k+1)
z2
kk
4 (k + 1)
k!1
jzj2
= lim
1 + 1=k
1
= 1:
(g)
We apply the ratio test, thus
ak
kk
1 + 2k+1 (k + 1)k+1
k k 1 + 2k+1 (k + 1)k+1
= lim
=
R = lim
= lim
k!1 1 + 2k k k
k!1 ak+1
k!1 (k + 1)k (1 + 2k k k ) (k + 1)
(k + 1)k+1
!
k
1=k
1
2k+1 (k + 1)k
= lim
+
=
k!1
1 + 1=k
(1 + 2k k k ) (k + 1)
1 + 2k k k
!
k
k
1=k
1
k+1
1
+2
= 2:
= lim
k!1
1 + 1=k
(1 + 2k k k ) (k + 1)
k
1= (2k k k ) + 1
(h)
We apply the root test, thus
24
R=
lim
k!1
1
p
k
jak j
=
lim
k!1
q
k
1
(log k)k=2
1
= 0:
lim log k
k!1
(i)
We apply the ratio test, thus
ak
k! (k + 1)k+1
R = lim
= lim k
= lim
k!1 ak+1
k!1 k
k!1
(k + 1)!
25
k+1
k
= lim
k!1
1
1+
k
= e:
V.3.2
1 2 3 P
L
K
LLL
Determine for which z the following series converge.
1
1
1
P
P
P
(a)
(z 1)k (c)
2m (z 2)m (e)
nn (z 3)n
(b)
k=1
1
P
k=0
(z i)k
k!
(d)
m=0
1
P
m=0
(z+1)m
m2
(f)
n=3
Solution
(a)
We apply the ratio test for ak = 1, thus
R = lim
k!1
n=1
1
P
2n
n2
(z
i)n
1
ak
= lim = 1:
k!1 1
ak+1
P1
1j = 1:
Since k=1 1 diverges, the power series diverges for such z also. Hence, the
given series converges for all z satisfying jz 1j < 1.
(b)
We apply the ratio test for ak = 1=k!, thus
R = lim
k!1
ak
1 (k + 1)!
= lim
= lim (k + 1) = 1:
k!1
k!1
ak+1
k!
1
m!1
2m
1
1
am
= lim m+1 = lim = :
m!1 2
m!1 2
am+1
2
j2 (z
2) j = 2
1
2
= 1:
P
Since 1
m=1 1 converges, the power series diverges for such z also. Hence,
the given series converges for all z satisfying jz 2j < 1=2.
(d)
We apply the ratio test for am = 1=m2 , thus
R = lim
m!1
1 (m + 1)2
am
m2 + 2m + 1
1 + 2=m + 1=m2
= lim 2
= lim
=
lim
= 1:
m!1 m
m!1
m!1
am+1
1
m2
1
P1
2
Since
m=1 1=m converges, the power series converges for such z also.
Hence, the given series converges for all z satisfying jz + ij 1.
(e)
We apply the ratio test for an = nn , thus
nn
an
nn
= lim
= lim
R = lim
=
lim
n
n!1 (n + 1)n+1
n!1
n!1 an+1
n!1 (n + 1) (n + 1)
Therefore, the series diverges for all z satisfying jz
such that jz 3j = 0 then
jnn (z
3)n j = nn 0n = 0:
n
n+1
P
Since 1
n=1 0 converges, the power series converges for z = 3. Hence, the
given series only converges for z = 3.
(f)
We apply the ratio test for ak = 2k =k 2 , thus
27
1
= 0:
n+1
ak
1
2k (k + 1)2
k 2 + 2k + 1
1 + 2=k + 1=k 2
R = lim
= :
= lim 2
=
lim
=
lim
k+1
2
k!1 ak+1
k!1 k
k!1
k!1
2
2k
2
2
hence the radius of convergence is R = 1=2. Therefore, the series converges for all z satisfying jz 2 ij < 1=2 and diverges for all z satisfying
jz 2 ij > 1=2. Now consider z such that jz 2 ij = 1=2 then
2n
(z
n2
i)
2n
= 2
n
1
2
1
:
n2
P
2
Since 1
k=1 1=n converges, the power series converges for such z also. Hence,
the given series converges for all z satisfying jz 2 ij 1=2.
28
V.3.3
1 2 3 P
L
K
LLL
Find the radius of convergence of the following series.
1
P
n
(a)
z 3 = z + z 3 + z 9 + z 27 + z 81 +
n=0
P p
(b)
z = z 2 + z 3 + z 5 + z 7 + z 11 +
p prime
Solutions
(a)
(b)
R=
1
lim sup
p
k
jak j
29
1
= 1:
1
V.3.4
1 2 3 P L K
P
Show that the function dened by f (z) = z n! is analytic on the open unit
disk fjzj < 1g. Show that jf (r )j ! +1 as r ! 1 whenever is a root of
unity. Remark. Thus f (z) does not extend analytically to any larger open
set than the open unit disk.
Solution
Cauchy-Hadamard formula gives R = 1 (radius of convergence) ) f (z) is
analytic for jzj < 1. Suppose is an N th root of unity, N = 1. Then
1
P
n!=N
(r )n! = rn! N
= rn! for n > N . Now
rn! " 1 as r " 1, because it is
increasing, and the nite partial sums ! 1 as r ! 1. ) jf (r )j ! +1 as
r ! 1.
30
V.3.5
1 2 3 P
L
K
LLL
What functions are represented by the following power series?
1
1
P
P
(a)
kz k
(b)
k2z k
k=1
k=1
Solution
(a)
P
k
Dierentiate the geometric series 1
k=0 z , obtain
1
X
z =
k=0
Multiply by z obtain
1
X
1
X
kz k
k=1
kz k =
k=1
z
z)2
(1
(1
1
z)2
(b)
P
k
Dierentiate the geometric series 1
k=0 z twice, obtain
1
X
z =
k=0
1
X
kz
k 1
k=1
1
(1
z)
1
X
k (k
1) z k
k=2
2
(1
Multiply by z 2 obtain
1
X
k2
k z2 =
k=2
We get
1
X
k=2
2 k
k z =
1
X
k=2
kz k +
2z 2
:
(1 z)3
2z 2
z
3 =
(1 z)
(1 z)2
31
z+
2z 2
(1 z)3
z)3
V.3.6
1 2 3 P L K
P
P
Show that the series
ak z k , the dierentiated series
kak z k 1 , and the
P ak k+1
integrated series
z
all have the same radius of convergence.
k+1
Solution
p
p
Can use Cauchy-Hadamard formula, and k k = 1. Then lim sup k jak j =
k!1
q
p
ja
j
k
lim sup k k jak j = lim sup k k+1
, so all series have the same radius of converk!1
k!1
gence. Can also use the characterization of R as the radius of the largest
disk to which the function extends and ?????. This largest disk is clearly the
same for f , f 0 , and ????.
32
V.3.7
1 2 3 P L K
Consider the series
1
X
2 + ( 1)k
zk :
k=0
Use the Cauchy-Hadamard formula to nd the radius of convergence of the series. What happens when the ratio test is applied?
Evaluate explicitly the sum of the series.
Solution
We apply the Cauchy-Hadamard formula where ak = 2 + ( 1)k
observe that
r
p
k
k
k
jak j =
2 + ( 1)k = 2 + ( 1)k
. First
Note that
3 if k is even,
1 if k is odd.
2 + ( 1)k =
Hence
p
k
jak j =
3 if k is even,
1 if k is odd.
Hence, since the lim sup of this sequence is 3 we have by the Cauchy-Hadamard
formula
R=
1
lim sup
2 + ( 1)k
2 + ( 1)k+1
1
= :
3
jak j
p
k
k
k+1
33
3k
1
3k+1
if k is even,
if k is odd.
Since the sequence does not converge, the ratio test is inconclusive, in that
limk!1 jak j = jak+1 j does not exist. However since jak j = jak+1 j
3, it does
converge for jzj < 1=3.
Then we have with jzj < 1=3
1
X
2 + ( 1)k
zk =
k=0
1
X
2n 2n 2n
2 + ( 1)
k=0
1
X
n=0
1
X
2 + ( 1)2n+1
2n+1 2n+1
1
X
1
X
k=0
2n 2n
3 z +
1
X
2n+1 2n+1
n=0
n=0
9z
2 n
+z
z2
n=0
z
1
+
:
1 9z 2 1 z 2
Note: The series is convergent because j9z 2 j ; jz 2 j < 1 since jzj < 1=3, and
that the series have singularities at z = 1=3 and z = 1.
34
V.3.7
p
Let L = lim sup k jak j. The denition
of n ln supnpimplies that given " > 0,
p
9 > 0, 9N 3 for k > N we have k jak j 6 L+", but k jak j > L " for innitely
P
many k 0 s. If k > N , then jak j < (L + ")k . If jzj < 1= (L + "), then
ak z k
converges. (L + ") jzj < 1, k z k 6 (L + ")k jzjk = ((L + ") jzj)k converges
by comparison with geometric series. ) R > 1= (L + ") ) jak j > (L ")k .
If jzj > L ", then ak z k > (L ")k (L 1")k for innitely many k 0 s. ), and
P
ak z k diverge. ) R 6 L1 " . Let " ! 0, get R 6 L1 .
35
V.3.8
1 2 3 P L K
Write out a proof of the Cauchy-Hadamard formula (3.4).
Solution
36
V.4.1
1 2 3 P
L
K
LLL
Find the radius of convergence of the power series for the following
functions, expanded about the indicated point.
(d) Log z about z = 1 + 2i;
(a) z 1 1 about z = i;
1
(b) cos z about z = 0;
(e) z 3=2
about z = 3;
1
z i
(c) cosh
about
z
=
0;
(f)
about
z = 2i:
z
z3 z
Solution
(d)
Since the values of Log z are unbounded in any neighborhood of the origin it
may not be extended to an analytic function on any open disk containing the
p
origin. But Log z is dened and analytic on the disk of radius j1 +
p2ij = 5
centered at 1 + 2i. Hence, the desired radius of convergence is 5 by the
second Corollary on p. 146.
(a) R =
37
5 (e) R = 3 (f) R = 2
V.4.2
1 2 3 P
L
K
LLL
Show that the radius
of convergence of the power series expansion of (z 2
p
about z = 2 is 7.
Solution
Rewrite f (z) as
z+1
. Singularities of f (z) are at
(z e2 i=3 )(z+e2 i=3 )
p
distance from 2 to nearest singularity is 7.
38
e2
i=3
1) = (z 3
, and
1)
V.4.3
1 2 3 P
L
K
LLL
Find the power series expansion of Log z about the point
z = i 2. Show
p
that the radius of convergence of the series is R = 5. Explain why this
does not contradict the discontinuity of Log z at z = 2.
Solution
(From Hints and Solutions)
p
Log z extends to be analytic for jz (i 2)j < 5, through the extension
does not coincide with Log z in the part of the disk in the lower halfplane.
39
V.4.4
1 2 3 P
L
K
LLL
Suppose f (z) is analytic at z = 0 and satises f (z) = z + f (z)2 . What is
the radius of convergence of the power series expansion of f (z) about z = 0?
Solution
(From Hints and Solutions)
p
1 4z =2.
Near 0 the function coincides with one of the branches of 1
The radius of convergence of the power series of either branch is 1=4, which
is the distance to the singularity at 1=4.
40
V.4.5
1 2 3 P L K
Deduce the identity eiz = cos z + i sin z from the power series expansions.
Solution
1
P
eiz =
n=0
in z n
n!
n even
n odd
1
P
m=0
( 1)m z 2m
(2m)!
+i
1
P
m=0
41
( 1)m z 2m+1
,
(2m+1)!
where n = 2m
V.4.6
1 2 3 P
L
K
LLL
Find the power series expansions of cosh z and sinh z about z = 0. What are
the radii of convergence of the series?
Solution
1
P
(a) cosh z = cos iz =
n=0
(b) sinh z =
1
P
z 2n+1
=
(2n+1)!
z 2n
(2n)!
=1+
z2
2!
z4
4!
(iz)3
3!
n=0
42
+ :::
+
(iz)5
5!
::: = z +
z3
3!
z5
5!
+ :::
V.4.7
1 2 3 P L K
Find the power series expansion of the principal branch Tan 1 (z) of
the inverse tangent function about z = 0. What is the radius of convergence of the series? Hint. Find it by integrating its derivative
(a geometric series) term by term.
Solution
We have
d
Tan
dz
thus
Tan
X
1
z=
=
1 + z2
k=0
1
z=
1
X
( 1)k
k=0
And we have R = 1.
43
z2
z 2k+1
:
2k + 1
V.4.8
1 2 3 P L K
Expand Log (1 + iz) and Log (1 iz) in power series about z = 0. By
comparing power series expansions (see the preceding exercise),
establish the identity
Tan
z=
1
Log
2i
1 + iz
1 iz
Solution
Log (1 + iz) =
1
X
k=0
( i)
k+1
z k+1
;
k+1
Log (1
iz) =
1
X
ik+1
k=0
Thus
X
1
1 + iz
2i X 2j z 2j+1
z 2j+1
Log
=
=
( 1)j
:
i
2i
1 iz
2i j=0 2j + 1
2j
+
1
j=0
1
44
z k+1
:
k+1
V.4.9
1 2 3 P L K
Let a be real, and consider the branch of z a that is real and positive on
(0; 1). Expand z a in a power series about z = 1. What is the radius of
convergence of the series? Write down the series explicitly.
Solution
f (z) = z a = ea Log z , Re z > 0. Assume a 6= 0; 1; 2; : : :. f (m) (z) = a (a 1) : : : (a
, f (m) (1) = a (a 1) : : : (a m + 1)
1
P
(m)
m+1)
a
am (z 1)m , am = f m!(1) = a(a 1):::(a
f (z) =
= m
, m > 1, and
m!
m=0
a0 = a1 =
a
0
. f (z) =
1
P
m=0
a
m
(z
1)m . R = 1 since
45
am+1
am
a m
m+1
! 1 as
m + 1) z a
= 1;
and
1)
(
n!
n + 1)
1:
zn:
Show that the binomial series represents the principal branch of the function
(1 + z) . For which does the binomial series reduce to a polynomial?
Solution
Use f (n) = (
1) : : : (
n + 1) (1 + z) n and the formula for the coe cient of z n . The series reduces to a polynomial for = 0; 1; 2; : : :. Otherwise
radius of convergence is 1, which is distance to the singularity at 1. We
can obtain the radius of convergence for the series also from the ratio test,
(n+1)!
1):::(
n+1)
an
= ( (1):::(
= n+1n ! 1 as n ! 1.
an+1
n+1)(
n) n!
46
V.4.11
1 2 3 P L K
For xed n
1
X
k=0
( 1)k z n+2k
:
k! (n + k)!2n+2k
Show that Jn (z) is an entire function. Show that w = Jn (z) satises the
dierential equation
n2
z2
1
w00 + w0 + 1
z
w = 0:
Jn00
(z) =
k=0
1
P
k=0
1
P
k=0
( 1)k z n+2k
k!(n+k)!2n+2k
zn
n!2n
+ O (z n+2 )
zn 1
(n 1)!2n
2
+ O (z n+2 )
zn 2
(n 2)!2n
+ O (z n )
n2 ) Jn is
1)k (n+2k)(n+2k 1)
k!(n+k)!2n+2k
( 1)k (n+2k)
( 1)k n2
k!(n+k)!2n+2k
k!(n+k)!2n+2k
( 1)k
=
(k 1)!(n+k 1)!2n+2k 2
( 1)k
k!(n+k)!2n+2k
(n + 2k) (n + 2k
1) + (n + 2k)
n2
4k (n + k)
=0
47
V.4.12
1 2 3 P
L
K
LLL
Suppose
that the analytic function f (z) has power series expansion
P
n
an z . Show that if f (z) is an even function, then an = 0 for n
odd. Show that if f (z) is an odd function, then an = 0 for n even.
Solution
If f (z) is analytic
f ( z) is also analytic in the same region
P1 in Dr n(0) then
n
and
P1 f ( z) = 2n+1n=0 ( 1) an z . If f is even then f (z) f ( z) = 0, and so
= 0. A power series is 0 i all of its contents are 0 (since
n=0 2a2n+1 z
(n)
an = f (0) =n! which shows an = 0 for n odd. The result for f (z) odd is
analogous.
48
49
V.4.13
1 2 3 P L K
Prove the following version of LHospitalss rule. If f (z) and g (z) are analytic, f (z0 ) = g (z0 ) = 0, and g (z) is not identically zero, then
f 0 (z)
f (z)
= lim 0
;
z!z0 g (z)
z!z0 g (z)
lim
in the sense that either both limits are nite and equal, or both limits are
innite.
Solution
1
P
Assume z0 = 0 , f (z) =
ak z k = f 0 (0) z +O (z 2 ), g (z) = g 0 (0) z +O (z 2 ) =
k=1
8
< 1 if g (0) = 0 and f (0) 6= 0 or k < N
0 (0)
aN
(z)
if g (0) = 0 , f (0) 6= 0, k = N
g 0 (0) z+aN z n +O z N +1 . fg(z)
! fg0 (0)
if g 0 (0) 6= 0 !
: bN
0 if g (0) = 0 , f (0) 6= 0, k > N
0 (0)+ka z k 1 +O z k
0
0
0
f
( ) f (z)
k
f (z)
(0)
= 0
! fg0 (0)
if g 0 (0) 6= 0
N 1
N ,
0
g 0 (z)
8 g (0)+N bN z +O(z ) g (z)
g (0) = 0 and either f (0) 6= 0 or k < N
< 1
N aN
aN
=
g (0) = 0 , f (0) 6= 0, k = N
!
bN
: N bN
0
g (0) = 0 , f (0) 6= 0, k > N
Then the limits are equal in all cases.
50
V.4.14
1 2 3 P L K
Let f be continuous function on the unit circle T = fjzj = 1g. Show that
f can be approximated uniformly on T by a sequence of polynomials in z
if and only if f has an extension F that is continuous on the closed disk
fjzj 1g and analytic on the interior fjzj < 1g. Hint. To approximate such
an F , consider dilates Fr (z) = F (rz).
Solution
If f has an analytic extension F , then Fr (z) = F (rz) has power series
1
P
P
F (z) = an z k , jzj < 1. Fr (z) =
an rn z k , jzj < 1r . Fr converge uniformly
n=j
for jzj 6 1, to F (rz) and F (rz) ! f (z) uniformly for jzj = 1 as v " 1. For
" > 0, take r < 1 with jFr (z) F (z)j < " for jzj = 1, then take N such that
N
N
P
P
an rn z n Fv (z) 6 2" for jzj = 1. Then
an rn z n F (z) 6 2" + 2" = "
n=0
n=0
51
V.5.1
1 2 3 P
L
K
LLL
Expand the following functions in power series about 1
2
2
(a) z21+1 (b) z3z 1 (c) e1=z (d) z sinh (1=z)
Solution
(b)
Denote the above function by f and dene g by
(1=w)2
w
=
g (w) = f (1=w) =
3
1 w3
(1=w)
1
for w 6= 0
and g (0) = 0. Note that g is analytic on the disk D1 (0) with power series
representation
g (w) =
1
X
w
=
w
w3
1 w3
k=0
3
1 if jwj
f (z) = g (1=z) =
1
X
(1=z)
(a)
1
P
n=0
(b)
1
z 2 +1
=1
z2 + z4
1
P
::: =
n=0
( 1)n
z 2n+2
z2
z3 1
1=z 2
(c) e
=
=
1
1
z 1 1=z 3
1
P
n=0
(d) z sinh
1
z
1 1
n! z 2n
=z
1
z
1
z
1
P
n=0
1
3!z 3
1
z 3n
1
5!z 5
1
X
w3n+1
k=0
3n+1
1
X
k=0
k=0
1
( 1)n z 2n z12 1+1=z
2 =
1
P
n=0
1
z 3n+1
+ ::: =
1
P
n=0
52
1
1
(2n+1)! z 2n
1
z 3n+1
n
1 ( 1)
z 2 z 2n
V.5.2
1 2 3 P L K
Suppose f (z) is analytic at 1, with series expansion (5.1). With the notation
f (1) = b0 and f 0 (1) = b1 , show that
f 0 (1) = lim z jf (z)
z!1
Solution 1
P
f (z) =
k=0
bk
zk
= f (1)+ f (1)
+
z
f 0 (1) as z ! 1, since
1
P
k=1
1
P
k=2
bk
.
zk
z [f (z)
f (1)j :
f (1)] = f 0 (1)+
bk+1 wk ! 0 as w ! 0.
53
1
P
k=1
bk+1
zk
V.5.3
1 2 3 P L K
Suppose f (z) is analytic at 1, with series expansion (5.1). Let
0 be the smallest number such that f (z) extends to be analytic
for jzj > . Show that the series (5.1) converges absolutely for
jzj > and diverges for jzj < .
Solution (A. Kumjian)
Recall that f is analytic at 1 i there is a function at 0 such that f (z) =
g (w) with w = 1=z wherever f (z) makes sense. We may suppose that f (z)
is dened for jzj > and that f is analytic at these points. Then setting
R = 1= if
> 0 and R = 1 if
= 0, we have that g is analytic on
DR (z0 ) and cannot be extended to an analytic function on any larger open
disk centered at z0 . Hence, g has a power series representation
g (w) =
1
X
bk w k
k=0
with radius of convergence R. Thus, the series converges absolutely for jwj <
R and diverges for jwj > R. It follows that by substituting w = 1=z, we have
1
X
bk
f (z) =
zk
k=0
54
55
jzj<jwj
RR
1>jzj>jwj
f (w) =
dxdy
w z
RR
1 dxdy
z 1 w=z
jzj6jwj
1
P
wn
n=0
56
RR
jzj6jwj
dxdy
z n+1
=0
V.6.1
1 2 3 P
L
K
LLL
Calculate the terms through order seven of the power series expansion about
z = 0 of the function 1= cos z.
Solution
1
=1+
cos z
2
z2
2!
4
z4
4!
z6
6!
1 + z2! + a4 z + a6 z 6 + O (z 8 )
2
1
5
z 4 : 4!1 + 2!1 = 14 24
= 24
, z6 :
sin z
cos z
z3
3!
= z
z:1
z 3 : 21
z 5 : 5!1
1
z 7 : 12
1
3!
1
2
z5
5!
z7
7!
+ :::
1+
3
2
+ 2!1 = 18
2!4!
5 4
1 6
z2
+ 24
z
z +
2
12
1
= 31
6
1
5
1
1
5
16
4
+ 24
= 120
+ 24
= 120
= 30
3!
12
2
5
+ 215! 7!1 = 7 6 5 2 57! 7+3 7 1 =
24 3!
420 175+21 1
= 265
= 7 653
7!
7!
432
1
2
57
2
15
1
24
:::
1
12
V.6.2
1 2 3 P
L
K
LLL
Calculate the terms through order ve of the power series expansion about
z = 0 of the function z= sin z.
Solution
z
=z
sin z
z
z 3 =3!+z 5 =5! :::
z2
3!
1
1 z 2 =3!+z 4 =5! :::
2
6
z4
+ z7! : : : + (: : :) + (: : :)3 + O (z 2 ) =
5!
1 + z 2 =6 + a4 z 4 + a6 z 6 + O (z 7 )
1
1
7
1
a4 = 5!1 + (3!)
= 103603 = 360
2 = 36
120
6(1 14)+4 5 7
2
1
62
a6 = 7!1 = 3!5!
+ (3!)
= 1403!7!78 = 3!7!
= 32
3 =
3!7!
2
z
7z 4
= 1 + z6 + 360
+ O (z 6 )
sin z
=1+
58
V.6.3
1 2 3 P
L
K
LLL
Show that
ez
1
1 3 3 4 11 5
= 1 + z2
z + z
z +
1+z
2
3
8
30
Show that the general term of the power series is given by
n
an = ( 1)
1
2!
( 1)n
+
;
n!
1
+
3!
2:
Solution
We have
X
1
( 1)k z k
=
1+z
k=0
1
and
1
X
zk
k=0
k!
for all z 2 C:
Since the given function can be expressed as product of these two we have
ez
1 z X
cn z n
=
e =
1+z
1+z
n=0
1
where
an = ( 1)n
( 1 + 1)+
( 1)n
2!
59
( 1)0
1
= ( 1)n
n!
2!
1
+
3!
( 1)n
:
n!
Solve ez = (1 + z) = a0 + a1 z + a2 z 2 + : : : by multiplying
9 by (1 + z) and coma0 = 1
>
>
Recursively
=
a
+
a
=
1
0
1
z
paring with the coe cients of e .
) a0 = 1
a1 + a2 = 1=2
>
>
an = 1=n! a n
;
an 1 + an = 1=n!
1
+ (n 1 2)! : : : + ( 1)n 0!1 (the two last terms cancel.) The
So an = n!1
(n 1)!
radius of convergence is 1 because the function has a pole at 1 . (In fact,
( 1)n an ! e 1 when n ! 1 .)
60
V.6.4
1 2 3 P L K
Dene the Bernoulli numbers Bn by
z
z2
z4
z6
cot (z=2) = 1 B1
B2
B3
:
2
2!
4!
6!
Explain why there are no odd terms in this series. What is the radius of
convergence of the series? Find the rst three Bernoulli numbers.
Solution
2
4
6
z
cot (z=2) = 1 B1 z2! B2 z4! B3 z6! : : :,
2
cot (z=2) = cos(z=2)
is odd, so z2 cot (z=2) is even, so only even terms appear in
sin(z=2)
the series. The function has singularities at z=2 =
, i.e., at 2 . Distance
from 0 to nearest singularity = 2 = radius of convergence.
w
w cos
sin w
w2
2
1
6
1 2
w
3
w4
24
1+
w2
6
1
36
1
120
1
1
4
w2
+ w4!
2!
4
w2
+ w5!
3!
w6
+:::
6!
w6
+:::
7!
w4 + O (w6 ) =
7
1
1
1
1+
w2 + 360
+ 24
w4 + O (w6 ) =
2
12
1
4
6
1
w + O (w )
45
2
4 4
cot w = 1 B1 w2! B2 2 4!w
:::
22 B 1
1
1
24
1
1
) 2! = 3 , B1 = 6 , 4! B2 = 45
, B2 = 244!45 , B2 = 30
For B3 , must keep all powers ????? to and ????? w8 . Better to just ?????
for B1 and B2 .
61
V.6.5
1 2 3 P L K
Dene the Euler numbers En by
X En
1
=
zn:
cosh z
n!
n=0
1
What is the radius of convergence of the series? Show that En = 0 for n odd.
Find the rst four nonzero Euler numbers.
Solution 1
P
1
=
cosh z
n=0
En n
z ,
n!
3
2
i;
i; : : :. Distance from 0 to
1
E2
2!
E6
6!
z4
4!
+ ::: +
=
=
1
, E2 =
2!
1
1
+ 2 2!4!
6!
E6 =
1+6 5
z2
2!
z4
4!
E4
= 4!1
4!
1
, E6 =
(2!)3
6!
= 29 6
33
1,
z6
6!
+ 2!1
61
:::
2
1
4
5 3 = 29
62
(: : :)3 , E0 = 1
1
24
90 =
5
24
, E4 = 5
61
V.6.6
1 2 3 P L K
Show that the coe cients of a power series "depend continuously" on the
function they represent, in the following sense. If ffm (z)g is a sequence of
analytic functions that converges uniformly to f (z) for jzj > , and
fm (z) =
1
X
ak;m z ;
k=0
f (z) =
m
X
k=0
0, we have ak;m ! ak as m ! 1.
63
ak z k ;
V.7.1
1 2 3 P
Find
(a)
(b)
(c)
L
K
LLL
the zeros and orders
z 2 +1
(d) cos z 1
z2 1
1
1
+ z5 (e) cos zz 1
z
z 2 sin z (f) coszz2 1
Solution
(c)
Let f (z) = z 2 sin z. Then f has zeros at all integer multiples of i.e. z0 = k
for k 2 Z. Since sin z has a zero of order 1 at 0, so sin z = zh (z) for some
analytic function h with h (0) 6= 0. Thus, f (z) = z 3 h (z) and so f has a
zero of order three at 0. Note that f 0 (z) = 2z sin z + z 2 cos z. Hence, for any
other zero z0 = k where k 2 Zn f0g, we have
f 0 (k ) = 2k sin k + (k )2 cos k = ( 1)k k 6= 0:
Thus, f has a simple zero at z0 = k when k 6= 0.
(d)
Note that g (z) = cos z 1 = 0 i z = 2k for k 2 Z. We will show
that all the zeros are double zeros. First observe that g 0 (z) = sin z and
so g 0 (2k ) = sin 2k = 0 for all k 2 Z. But g 00 (z) = cos z and so
g 0 (2k ) = cos 2k = 1 for all k 2 Z. Hence, all zeros are double zeros.
64
(a) simple zeros at i, (b) simple zeros at e i=4 , e3 i=4 , (c) triple zero at 0,
simple zeros at n , n = 1; 2; : : :, (d) double zeros at n , n = 1; 2 : : :,
(h) simple zeros at
i=4 + n i=2, n = 1; 2; : : :, (i) no zeros.
1
1
4
(b) z + z5 = (z + 1) z15 = (z z1 ) (z z2 ) (z z3 ) (z z4 ) z15 , where zj =
ei( =4+j=4 2 ) , these four points are simple zeros.
y
y
(d) cos z = cos (x + iy) = cos x cos (iy) sin x sin (iy), cos (iy) = e 2+e =
cosh y, sin (iy) =
2i
ey
=i
ey e
2
65
66
V.7.3
1 2 3 P
L
K
LLL
Show that the zeros of sin z and tan z are all simple.
Solution
Zeros of sin z are at n , 1 < n < 1, since cos (n ) =
1 6= 0, the zeros are simple.
67
d
dz
sin zjz=n =
V.7.4
1 2 3 P
L
K
LLL
Show that cos (z + w) = cos z cos w
identity for z and w real.
Solution
cos (z + w) = cos z cos w + sin z sin w = F (z; w) is entire in z for each xed
w and entire in w for each xed z. F (z; w) = 0 for z; w real. Apply theorem,
with D = C, E = R. Get F (z; w) = 0, so cos (z + w) = cos z cos w
sin z sin w.
68
V.7.5
1 2 3 P L K
Show that
Z
zt2 +2wt
dt =
ew
2 =z
z; w 2 C; Re z > 0;
where we take the principal branch of the square root. Compare the result
to Exercise IV.3.1. Hint. Show that the integral is analytic in z and w, and
evaluate it for z = x >p 0 and w real by making a change of variable and
using the known value
for z = 1 and w = 0.
Solution
R1 at2 2bt
e
dt, note the integral is improper, so a limit process may be used
1
s2
e s ds =
R1
1
R1
1
R1
R1
eat
=2)
(s
=4
ds = e
e
e
s2
e s ds =
zt2
zbt
dt =
2bt
=4
p1
x
dt =
Z1
(s
e
p
=4
R1
xt2
2bt
dt =
2 =z
eb
69
p1
x
{z
R1
s2
2b
p
s
x
=2)
ds
b2 =x
, x real
ds. F ( ) =
V.7.6
1 2 3 P
L
K
LLL
Suppose f (z) is analytic on a domain D and z0 2 D. Show that if
f (m) (z0 ) = 0 for m 1, then f (z) is constant on D.
Solution
Suppose that f (m) (z0 ) = 0 for m and dene g on D by g (z) = f (z) f (z0 ).
Then it su ces to show that g is identically zero. Suppose not, then since
g (m) (z0 ) for m 0, the power series expansion for g is trivial and therefore g
is identically zero on a disk on nonzero radius centered at z0 but this violates
the rst theorem on page 156 that asserts that an analytic function which is
not identically zero only has isolated zeros. This result may also be proved
using the Uniqueness Principle (the second theorem on page 156).
70
V.7.7
1 2 3 P
L
K
LLL
Show that if u (x; y) is a harmonic function on a domain D such that all the
partial derivatives of u (x; y) vanish at the same point of D, then u (x; y) is
constant on D.
Solution
71
V.7.8
1 2 3 P L K
With the convention that the function that is identically zero has a zero of
innite order at each point, show that if f (z) and g (z) have zeros of order n
and m respectively at z0 , then f (z) + g (z) has a zero of order k min (n; m).
Show that strict inequality can occur here, but that equality holds whenever
m 6= n.
Solution
72
V.7.9
1 2 3 P
L
K
LLL
Show that the analytic function f (z) has a zero of order N at z0 , then
f (z) = g (z)N for some function g (z) analytic near z0 and satisfying g 0 (z) 6=
0.
Solution
(From Hints and Solutions)
Write f (z) = (z z0 )N h (z), where h (z) has a convergent power series and
h (z0 ) 6= 0. Take g (z) = (z z0 ) e(log(h(z)))=N for an appropriate branch of
the logarithm.
73
V.7.10
1 2 3 P L K
Show that if f (z) is a continuous function on a domain D such that f (z)N
is analytic on D for some integer N , then f (z) is analytic on D.
Solution
f (z)N analytic. Zeros of f (z) are isolated, for this need maximum principle.
f (z) is analytic, except possibly at the isolated points where f (z) = 0. (By
the Riemanns theorem, f (z) is analytic ????? there but dont have this yet)
Write f (z)N = (z z0 )m h (z), h (z) analytic, h (z) 6= 0. Then h (z) has N th
root near z0 , by preceding exercise. f (z)N = g (z)N (z z0 )m , (f (z) =g (z))N =
(z z0 )m . (z z0 )m=N is continuous in a neighborhood of z0 . f (z) =g (z) returns to original value, doing circle around z z0 . m is an integral multiple
of N .
74
V.7.11
1 2 3 P L K
Show that if f (z) is a nonconstant analytic function on a domain D, then
the image under f (z) of any open set is open. Remark. This is the open
mapping theorem for analytic functions. The proof is easy when f 0 (z) 6= 0,
since the Jacobian of f (z) coincides with jf 0 (z)j2 . Use Exercise 9 to deal
with the points where f 0 (z) is zero.
Solution
If f 0 (z0 ) 6= 0, then f maps open disks centred at z0 onto open sets, so f (D)
contains a disk centred at f (z0 ).
If f 0 (z0 ) = 0, assume f (z0 ) = 0, write f (z) = (z z0 )N h (z), h (z0 ) 6= 0.
Then f has a N th root near z0 , f (z) = g (z)N , g 0 (z0 ) 6= 0. g covers a ?????
disk, so f cover disk centred at 0.
75
V.7.12
1 2 3 P L K
Show that the open mapping theorem for analytic functions implies the maximum principle for analytic functions.
Solutions
Clearly open mapping theorem ) strict maximum principle for analytic functions, since a non constant analytic function cant attain maximum at z0 and
cover a disk centred at z0 . Then strict maximum principle ) maximum
principle.
76
77
V.7.14
1 2 3 P L K
A set E is discrete if every point of E is isolated. Show that a closed discrete
subset of a domain D either is nite or can be arranged in a sequence fzk g
that accumulates only on f1g [ @D.
Solution
Let Kn = fz 2 D : d (z; @D) > 1=n; jzj 6 ng. En is compact, only nitely
many points of E belongs to Kn ????? points, shading with those in E \ K1 ,
the E \ (K2 nK1 ) ,?????
78
V8.1
1 2 3 P L K
p
Suppose that the principal branch of z 2 1 is continued analytically from
z = 2 around the gure eight path indicated above. What is the analytic continuation of the function at the end of the path? Answer the same question
1=3
1=3
form the functions (z 3 1) and (z 6 1) .
Solution
p
z 2 1, phase change i at +1, 1 at 1, i at +1, returns to initial value.
p
3
z3 p
1, phase change ei =3 at +1, ei =3 at +1, returns to e2 i=3 times initial
value. 3 z 6 1, phase change ei =3 at +1, e 2 i=3 at 1, ei =3 at +1, returns
to initial value.
79
V8.2
1 2 3 P L K
has an analytic continShow that f (z) = Log z = (z 1) 12 (z 1)2 +
it
uation around the unit circle (t) = e , 0
t
2 . Determine explicitly
the power series ft for each t. How is f2 related to f0 ?
Solution P
m 1
df
= z1 =
, f (m) (z) = ( 1)
(m 1)!
dz
zm
1 f (m) eit
1
P
P
( )
m
it
it
Series is
(z
e
)
=
f
(e
)
+
(
m!
m=0
ft (z) = it +
1
P
m=1
m=1
itm
1)m 1 e m
(z
it m
1)m
1 e
itm
(z
80
eit )
V.8.3
1 2 3 P L K
p
Show that each branch of z can be continued analytically along any path
in Cn f0g, and show that the radius of convergencepof the power series ft (z)
representing the continuation is j (t)j. Show that z cannot be analytically
along path containing 0.
Solution
81
V.8.4
1 2 3 P L K
Let f (z) be analytic on a domain D, x z0 2 D, and let f (z) =
be the expansion of f (z) about z0 . Let
F (z) =
z0
f ( )d =
1
X
an
(z
n
+
1
n=0
an (z
z0 )n
z0 )n+1
be the indenite integral of f (z) for z near z0 . Show that F (z) can be
continued analytically along any path in D starting at z0 . What happens in
the case D = Cn f0g, z0 = 1, and f (z) = 1=z? What happens in the case
that D is star-shaped?
Solution
82
V.8.5
1 2 3 P L K
Show that the function dened by
X n
f (z) =
z2 = z + z2 + z4 + z8 +
is analytic on the open unit disk fjzj < 1g, and that it cannot be extended
analytically to any larger open set. Hint. Observe that f (z) = z +f (z 2 ),
and that f (r) ! +1 as r ! 1.
Solution
83
V.8.6
1 2 3 P L K
P
Suppose f (z) =
an z n , where an = 0 except for n in a sequence nk that
satises nk+1 =nk 1 + for some > 0. Suppose further that the series has
radius of convergence R = 1. Show that f (z) does not extend analytically
to any point of the unit circle. Remark. Such a sequence with large gaps
between successive nonzero terms is called a lacunary sequence. This result
is the Hadamard gap theorem. There is a slick proof. If f (z) extends analytically across z = 1, consider g (w) = f (wm (1 + w) =2), where m is a large
integer.
Show that the power series for g (w) has radius of convergence r > 1, and
that this implies that the power series of f (z) converges for jz 1j < ".
Solution
84
V.8.7
1 2 3 P L K
P
Suppose f (z) = an z n , where the series has radius of convergence R < 1.
Show that there is an angle
such that f (z) does not have an analytic
continuation along the path (t) = tei , 0 t R. Determine the radius of
convergence of the power series expansion of f (z) about tei .
Solution
85
V.8.8
1 2 3 P L K
Let f (z) be analytic at z0 , and let (t), a
t
b, be a path such that
(a) = z0 . If f (z) cannot be continued analytically along , show that there
is a parameter value t1 such that there is an analytic continuation ft (z) for
a t < t1 , and the radius of convergence of the power series ft (z) tends to
0 as t ! t1 .
Solution
86
V.8.9
1 2 3 P L K
Let P (z; w) be a polynomial in z and w, of degree n in w. Suppose that
f (z) is analytic at z0 and satises P (z; f (z)) = 0. Show that if ft (z)
is any analytic continuation of f (z) along any path starting at z0 , then
P (z; ft (z)) = 0 for all t. Remark. An analytic function f (z) satises a
polynomial equation P (z;
f (z)) = 0 is called an algebraic function. For
p
n
instance, the branches of z are algebraic functions, since the satisfy z wn =
0.
Solution
87
V.8.10
1 2 3 P L K
Let D be the punctured disk f0 < jzj < "g, suppose f (z) is analytic at z0 2
D, and ew0 = z0 . Show that f (z) has an analytic continuation along any
path in D starting at z0 if and only if there is an analytic function g (w) in the
half-plane fRe w < log "g such that f (ew ) = g (w) for w near w0 . Remark.
If f (z) does not extend analytically to D but has an analytic continuation
along any path in D, we say that f (z) has a branch point at z = 0. For the
proof, use the fact that any path in D starting at z0 is the composition of
a unique path in the half-plane starting at w0 and the exponential function
ew .
Solution
88
VI 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1
2
3
4
5
6
7
8
VI.1.1
1 2 3 P
L
K
111
LLL
Find all possible Laurent expansions centered at 0 of the following
functions.
1
(a) z21 z (b) zz+11 (c) (z2 1)(z
2 4)
Solution
(a)
In the region 0 < jzj < 1, we have
1
z2
1X k
z =
z k=0
1
X
1 1
=
z1 z
k 1
= [k
1 = n] =
1
X
zn:
n= 1
k=0
1 1
= 2
z
z 1
1
z
1
1 X
= 2
z k=0
1
z
1
1 X
= 2
z
z k=0
k 2
=[ k
2 = n] =
2
X
zn:
n= 1
(b)
In the region jzj < 1 we have
2
=1
z+1
z 1
=1
z+1
X
2
( 1)n z n =
=1 2
( z)
n=0
1
1 2
1
X
( 1)n z n :
n=1
2
=1
z+1
2
z1
=1
1
z
1
X
( 1)k
= [k + 1 = n] = 1
k+1
z
k=0
(c)
From computation, we have
2
2 X ( 1)k
=
z k=0 z k
1
1
X
( 1)n
zn
n=1
=1+2
1
X
( 1)n
:
n
z
n=1
(z 2
1
1) (z 2
4)
1 1
1 1
+
2
3z
1 3 z2 4
(z 2
1
1) (z 2
4)
1 1
1 1
1 1
1
1
+
=
=
2
2
2
3z
1 3z
4
31 z
12 1 z42
1
1
1
1 X z 2n
1 X 2n
1 X
z
4 4 n z 2n :
=
=
3 n=0
12 n=0 4n
12 n=0
(z 2
1
1) (z 2
4)
=
1 1
1 1
+ 2
=
2
3z
1 3z
4
1
1 X 1
3z 2 k=0 z 2n
1
1
2
3z 1
1 X z 2n
=
12 n=0 4n
1
1
1
34 1
1
z2
1
1 X 2k
z
3 k= 1
z2
4
1 X z 2n
12 n=0 4n
1
(z 2
1
1) (z 2
4)
1 1
1 1
+
3 z2 1 3 z2
1
1
+
=
3 z 2 1 z12
1
1 X 1
=
+
3z 2 n=0 z 2n
=
4
1
1
=
3 z 2 1 z42
1
1 X 4n
=
3z 2 n=0 z 2n
1X 1
1 X 4n
1X n
+
=
(4
3 n=0 z 2(n+1) 3 n=0 z 2(n+1)
3 n=0
1
1) z
2(n+1)
= [n + 1 =
1
1 X
=
4
3 k= 1
1 k
k] =
1 z 2k :
VI.1.2
1 2 3 P
L
K
111
LLL
For each of the functions in Exercise 1, nd the Laurent expansion centered
at z = 1 that converges at z = 12 . Determine the largest open set on which
each series converges.
Solution
(a)
Partial fractions give us
1
z2
Laurent series for
1
=
z
1=z
1
z+1
1
1
1 =
z + 1 1 z+1
1
X
k=0
1
=
1
z+1
1
1
+
z z 1
1
(z + 1)
k+1
1 X
1
=
z + 1 k=0 (z + 1)k
1
= [k + 1 =
n] =
1
X
(z + 1)n
n= 1
1
1 X (z + 1)n
=
2 n=0
2n
1
1
1
=
2 1 z+1
2
1
1 X
2n+1
(z + 1)n
n=0
Thus we have
1
z2
1
X
1
1
2n+1
k= 1
The function 1= (z 2
(b)
We have
The function (z
an (z + 1)n ; where an =
for n
for n
1
0
z 1
2
=1
z+1
z+1
1) = (z + 1) converges on the set 0 < jz + 1j < 1.
4
(c)
Partial fractions give us
1 1
1 1
1 1
1 1
+
+
=
4)
6 z 1 6 z + 1 12 z 2 12 z + 2
1
1
1 1
1
1
1
1
=
+
+
=
6 z + 1 2 6 z + 1 12 z + 1 3 12 z + 1 + 1
1
1
1
1
1
1 1
1
=
z+1 +
z+1
1 =
12 1
6 z + 1 36 1
12 (z + 1) 1 + z+1
2
3
1
1
1
1 X (z + 1)n
1 X ( 1)k
1 X (z + 1)n 1 1
+
=
=
12 n=0
2n
6 z + 1 36 n=0
3n
12 k=0 (z + 1)k+1
(z 2
1
1) (z 2
1 X (z + 1)n 1 1
+
12 n=0
2n
6z +1
1
1
1 X
( 1)n (z + 1)n +
12 n=0
6
1
1) (z 2
1
12
1
n= 1
1 X (z + 1)n
1 X
+
( 1)n (z + 1)n =
n
36 n=0
3
12 1
(z + 1) 1 +
1 X
12 n=0
1
12
2n
36 3n
(z + 1)n ;
VI.1.3
1 2 3 P
L
K
LLL
Recall the powe series for the Bessel function Jn (z), n 0, given in Exercise
V.4.11, and dene J n (z) = ( 1)n Jn (z). For xed w 2 C, establish the
Laurent series expansion
exp
hw
2
(z
1
i X
1=z) =
Jn (w) z n ;
1
z 2 C:
sin z
lim (z + =2) cos
= sin
z
z!
=2
z! =2
2
+ n ; n = 0; 1; 2; : : :
=2
cos =2
sin =2
sin =2
z+ =2
cos
z
cos( =2)
=2
lim
z!
1
.
z
=2
1
sin( =2)
sin( =2)
3
2
VI.1.4
1 2 3 P
L
K
LLL
Suppose that f (z) = f0 (z) + f1 (z) is the Laurent decomposition of an analytic function f (z) on the annulus fA < jzj < Bg. Show that if f (z) is
an even function, then f0 (z) and f1 (z) are even functions, and the Laurent
series expansion of f (z) has only even powers of z. Show that if f (z) is an
odd function, then f0 (z) and f1 (z) are odd functions, ant the Laurent series
expansion of f (z) has only odd powers of z.
Solution
1
H f (z)
P
dz,
or
denition
of
f
(z)+f
(z),
f
(z)
=
an z n , f ( z) =
Use an =
0
1
n+1
z
1
jzj=r
P
( 1)n an z n . f even ) an = an for n odd ) an = 0 for n odd. )
f0 (z) + f1 (z) have only even ????? ?????.
VI.1.5
1 2 3 P
L
K
LLL
Suppose f (z) is analytic on the punctured plane D = Cn f0g. Show
that there is a constant c such that f (z) c=z has a primitive in D.
Give a formula for c in terms of an integral of f (z).
Solution
By the Laurent Expansion Theorem we have for all z 2 D and r > 0
f (z) =
1
X
ak z
k= 1
Setting c = a
1
2 i
1
ak =
2 i
where
f( )
Cr (0)
z0 )k+1
d :
f ( ) d we have
X
ak z k :
f (z) c=z =
Cr (0)
k6= 1
R
Recall that an analytic function g has a primitive on a region D i g (z) dz =
0 for every piecewise smooth
closed path in D. Let be such a path in D
P
and observe the series k6= 1 ak z k converges uniformly on any closed annulus
fz : r jzj sg with 0 < r < s and hence on . Hence, by the rst theorem
on page 126, we may integrate the series termwise we obtain:
Z X
Z
X
X Z
k
ak 0 = 0;
ak z k dz =
ak z =
f (z) c=zdz =
k6= 1
k6= 1
1. Hence, f (z)
k6= 1
VI.1.6
1 2 3 P L K
Fix an annulus D = fa < jzj < bg, and let f (z) be a continuous function on
its boundary @D. Show that f (z) can be approximated uniformly on @D by
polynomials in z and 1=z if and only if f (z) has continuous extension to the
closed annulus D [ @D that is analytic on D.
Solution
Use the conformity theorem for polynomial approximation, V5 14. If f is
analytic, continuous ????? values, unit f = f0 +f1 , its Laurent decomposing.
Then f0 + f1 have continuous boundary values. Can approximate f0 (z)
uniformly by polynomials in z for f1 (z) uniformly by polynomials in 1=z.
(?????? change of variable w = 1=z ). That does it. If f = lim gn (z)
uniformly on @D, then by the maximum principle jgk gj j ! 0 uniformly
on D, so gk ! f uniformly on D, and clearly F n@D = f .
10
VI.1.7
1 2 3 P L K
Show that a harmonic function u on an annulus fA < jzj < Bg has a unique
expansion
u rei
1
X
an rn cos (n ) +
n= 1
bn rn sin (n ) + c log r;
n6=0
which is uniformly convergent on each circle in the annulus. Show that for
each r, A < r < B, the coe cients an ; bn and c satisfy
n
an r + a
n
bn r + b
nr
nr
=
=
1
1
1
a0 + c log r =
2
Z
Z
u rei
cos (n ) d ;
n 6= 0;
u rei
sin (n ) d ;
n 6= 0;
u rei
d :
VI.1.8 (ej)
H
Multiply f by a unimodular constant, assume jf (z)j 6 M ,
f (z) dz =
R2
f ei
R2
jzj=1
Re f ei
iei
d = 2 M,
Re f ei iei d 6 f ei
6 M . If have strict inequality somewhere,
R2
M . Since f ei iei
then < 2 M . We conclude that Re f ei iei d
0
M , f ei
0. ) f ei iei
M , we have Im f ei iei d
f (z)
iM z. ) f (z) = cz for a constant, jcj = M .
ie
M,
VI.1.9 (ej)
For the analyticity, dierentiate
by hand. (See
i Exercise III.1.6). The Derivh
R
h(z)
h(z)
ative is H 0 (w) = lim 1w
dz
z (w+ w)
z w
w!1
R
R
h(z) w
dz = (zh(z)dz
= lim 1w (z (w+
w))(z w)
w)2
w!1
12
VI.2.1
1 2 3 P
L
K
LLL
Find the isolated singularities of the following functions, and determine where they are removable, essential, or poles. Determine
the order of any pole, and nd the principal part at each pole.
2
sin z
(a) z= (z 2 1) (d) tan z = cos
(g) Log 1 z1
z
z
(e) z 2 sin z1
(h) (zLog1)z3
(b) zze
2 1
2
2z
cos z
(c) e 1
(f)
(i) e1=(z +1)
z
z2
2 =4
Solution
(a) (J.R.J Groves)
with g (z) =
We have that z = 1 is a pole of order 2 since (z2 z 1)2 = (zg(z)
1)2
z
and g (z) is analytic near z = 1. Similarly we have, z = 1 is a pole
(z+1)2
h(z)
z
of order 2 since (z z1)2 = (z+1)
and h (z) is analytic near
2 with h (z) =
(z 1)2
z = 1.
(b) (J.R.J Groves)
z
g(z)
zez
We have that z = 1 is a pole of order 1 since zze
2 1 = z 1 with g (z) = z+1 ,
and g (z) is analytic near z = 1. Similarly, z = 1 is a pole of order 1 since
z
zez
= h(z)
with h (z) = zze 1 , and h (z) is analytic near z = 1.
z2 1
z+1
(c) (A. Kumjian)
Note that the given function is analytic on the punctured plane D = Cn f0g.
It has an isolated singularity at z = 0. For z 6= 0 we have
!
1
1
1
X
1
1 X 2j+1 j
e2z 1
(2z)k
1 X 2k z k
=
=
z :
1+
=
z
z
k!
z
k!
z
(j
+
1)!
j=0
k=0
k=1
Since the function has a power series expansion for all z 6= 0, it extends to an
analytic function at z = 0. Hence, the function has a removable singularity
at 0.
(d) (J.R.J Groves)
We have that cos z has simple zeros at z = =2 + k for k 2 Z. So cos1 z has
simple poles at z = =2 + k for k 2 Z. Since sin z is analytic and dierent
sin z
has simple poles at z = =2 + k for k 2 Z.
to 0 at these points, tan z = cos
z
(e) (A. Kumjian)
13
z sin
1
z
1
X
( 1)k
=z
(2k + 1)!
k=0
2
1
z
2k+1
1
X
( 1)k
1
=
2k
(2k + 1)! z
k=0
1
X
1
( 1)k
=z+
:
2k
(2k + 1)! z 1
k=1
Hence z = 0 is an essential singularity for the function since there are an innite number of nonzero terms in the Laurent series with negative exponents.
(e) (J.R.J Groves)
The function z 2 sin z1 has an isolated singularity at z = 0. Since
z 2 sin
1
z
= z2
1
1
X
X
1
( 1)n
1
( 1)n
=
;
2n+1
2n 1
(2n
+
1)!
z
(2n
+
1)!
z
n=0
n=0
z = 0 is an essential singularity.
(f) (J.R.J Groves)
The function z 2 ( =2)2 = (z
=2) (z + =2) has simple zeros at z =
=2
and z = =2. The function cos z has a simple zero at z = =2. So we suspect
a removable singularity. Set w = z
=2. Then
z2
cos (z)
=
(z
( =2)2
cos (z)
cos (w + =2)
=
=
=2) (z + =2)
w (w + )
1
w+
sin w
:
w
Since we know that sin (w) =w = 1 w2 =3! + : : :, we deduce that our function
has a removable singularity at w = 0 and so at z = =2. In a similar fashion,
it also has a removable singularity at z =
=2.
(g) (J.R.J Groves)
The function Log (z) is analytic on the region Cn ( 1; 0], it follows that
Log 1 z1 is analytic on Cn [0; 1]. There are no isolated singularities.
(h) (A. Kumjian)
First observe that for z 2 C with jz 1j < 1 we have
1
1
=
z
1 + (z
1)
1
X
k=0
14
( 1)k (z
1)k :
for jz
1)k ;
1j < 1
1
1
X
X
Log z
1
( 1)k+1
( 1)k+1
k
=
(z 1) =
(z
k
k
(z 1)3
(z 1)3 k=1
k=1
1
X
1
1
( 1)j
=
+
(z 1)j
2
2
(z
1)
k
+
3
(z 1)
j=0
1)k
1
1
1
+
2 z 1 3
i and z = i. Moreover,
i
1
1
=
+1
2 z+i
1
z
so that
1
e z2 +1 = e 2
i
1
z+i
i
1
2 z i
= e2
1
z+i
X1 ( i)n
1
:
n
n=0 2 n!
(z i)n
1
15
( 1) (z 1)
n+1
n=0
Log z
(z 1)3
n+1
1
(z 1)2
Log z = (z
1
z(z 1)
1)
(z 1)
2
(z 1)
3
:::
+ analytic
2
Double pole, principal part (z 11)2 z(z1 2) at z = 1. (i) e1=(z +1) , singularities
at z = i. Values of 1= (z 2 + 1) approach 1 from all directions as z ! i,
2
so values of e1=(z +1) do not converge, and singularities are essential.?????
2
2
check e1(1 t ) = f (it) ! +1 as t " 1 or t # 1, e1(1 t ) = f (it) ! 0 as
t # 1 or t " 1, so no limit at i.
16
VI.2.2
1 2 3 P
L
K
LLL
Find the radius of convergence of the power series for the following functions,
expanded about the indicated point.
(a) zz4 11 ;
about z = 3 + i;
(c) sinz z ;
about z = i;
cos z
z2
(b) z2 2 =4 ;
about z = 0;
(d) sin3 z ;
about z = i:
Solution
(a) Removable singularity at z = 1, poles at i; 1. R = 3 = ?????? to i
(nearest singularity).
(b) singularities at
=2 are removable.
p
2.
(c) removable at z = 0, poles at
; 2 . R=j i
j=
2
2
(d) z = sin z, simple poles at z = 0, R = .
17
VI.2.3
1 2 3 P L K
Consider the function f (z) = tan z in the annulus f3 < jzj < 4g. Let f (z)
= f0 (z) + f1 (z) be the Laurent decomposition of f (z), so that f0 (z) is
analytic for jzj < 4, and f1 (z) is analytic for jzj > 3 and vanishes at 1. (a)
Obtain an explicit expression for f1 (z).
(b)
Write down the series expansion for f1 (z), and determine the largest domain
on which it converges.
(c)
Obtain the coe cients a0 ; a1 and a2 of the power series expansion of f0 (z).
(d)
What is the radius of convergence of the power series expansion for f0 (z)?
Solution
(a) tan z has two poles in the disk jzj < 4, simple poles at
=2, principal
parts 1= (z
=2). If f1 (z) = 1= (z
=2) 1 (z + =2), then f0 (z) =
f (z) f1 (z) is analytic for jzj < 4, and f1 (z) is analytic for jzj > 3 and
! 0 as z ! 1. By uniqueness, f (z) = f1 (z) + f2 (z) is the Laurent
decomposition.
z+ =2+z
=2
=
(b) Use geometric series. Converges for jzj > =2. f1 (z) =
2 =4
z2
1
n
P 2 1
2z
2z
= , converges for jzj > =2.
2 =4 =
z2
z2
4 z2
2
1
P
n=0
k=0
z 2n+1
18
VI.2.4
1 2 3 P L K
Suppose f (z) is meromorphic on the disk fjzj < sg, with only a nite number
of poles in the disk. Show that the Laurent decomposition of f (z) with
respect to the annulus fs " < jzj < sg has the form f (z) = f0 (z) + f1 (z),
where f1 (z) is the sum of the principal parts of f (z) at its poles
Solution
Since f1 (z) is a sum of principal parts, with poles inside fjzj 6
"g, f1 (z)
is analytic for jzj >
", and f1 (z) ! 0 as z ! 1. Further f0 (z) =
f (z) f1 (z) is analytic at ????? ??????, hence for jzj < 5. By the uniqueness
of the Laurent decomposition, f is f0 (z) + f1 (z) = f (z). Use the uniqueness
of the Laurent decomposition.
19
VI.2.5
1 2 3 P L K
By estimating the coe cients of the Laurent series, prove that if z0 is an
isolated singularity of f , and if (z z0 ) f (z) ! 0 as z ! z0 , then z0 is
removable. Give a second proof based on Moreas theorem.
Solution
Suppose that z0 is an isolated singularity of f and that (z z0 )N f (z) is
bounded near z0 . Then by Riemanns Theorem on removable singularities
(see p. 172), (z z0 )N f (z) has a removable singularity at z0 , the Laurent
expansion is of the following form: there is a r > 0 so that
(z
z0 ) f (z) =
1
X
ak (z
z0 )k
k=0
1
X
ak (z
z0 )k
1
X
aj+N (z
z0 )j :
j= N
k=0
20
VI.2.6
1 2 3 P
L
K
LLL
Show that if f (z) is continuous on a domain D, and if f (z)8 is analytic on
D, then f (z) is analytic on D.
Solution
21
VI.2.7
1 2 3 P
L
K
LLL
Show that if z0 is an isolated singularity of f (z), and if (z z0 )N f (z)
is bounded near z0 , then z0 is either removable or a pole of order
at most N .
Solution
Suppose that z0 is an isolated singularity of f and that (z z0 )N f (z) is
bounded near z0 . Then by Riemanns Theorem on removable singularities
(see p. 172), (z z0 )N f (z) has a removable singularity at z0 , the Laurent
expansion is of the following form: there is r > 0 so that
(z
z0 ) f (z) =
1
X
ak (z
z0 )k
k=0
1
X
ak (z
k N
z0 )
1
X
aj+N (z
z0 )j :
j= N
k=0
22
VI.2.8
1 2 3 P L K
A meromorphic function f at z0 is said to have order N at z0 if f (z) =
(z z0 )N g (z) for some analytic function g at z0 such that g (z0 ) 6= 0. The
order of the function 0 is dened to be +1. Show that
(a) order (f g; z0 ) = order (f; z0 ) + order (g; z0 ),
(b) order (1=f; z0 ) = order (f; z0 ),
(c) order (f + g; z0 ) min forder (f; z0 ) ; order (g; z0 )g.
Show that the inequality can occur in (c), but that equality hold in (c)
whenever f and g have dierent orders at z0 .
Solution
23
VI.2.9
1 2 3 P L K
Recall that "f (z) = O (h (z)) as z ! z0 " means that there is a constant
C such that jf (z)j C jh (z)j for z near z0 . Show that if z0 is an isolated
singularity of an analytic function f (z), and if f (z) = O ((z z0 )m ) as
z ! z0 , then the Laurent coe cients of f (z) are 0 for k < m, that is the
Laurent series of f (z) has the form
f (z) = am (z
z0 ) + am+1 (z
z0 )m+1 +
Remark. This shows that the use of notation O (z m ) in section V.6 is consistent.
Solution
24
VI.2.10
1 2 3 P
L
K
LLL
Show that if f (z) and g (z) are analytic functions that both have the same
order N 0 at z0 , then
f (z)
f (N ) (z0 )
= (N )
:
z!z0 g (z)
g (z0 )
lim
Solution
25
VI.2.11
1 2 3 P L K
P
Suppose f (z) =
ak z k is analytic for jzj < R, and suppose that f (z)
extends to be meromorphic for jzj < R + ", with only one pole z0 on the
circle jzj = R. Show that ak =ak+1 ! z0 as k ! 1.
Solution P
If f (z) = an z n is analytic for jzj < R, and is meromorphic for jzj < R + ",
with poles
N on the circle jzj = R, then an = O nN 1 . If
P of ?????6
f (z) =
an z n as only one pole on the circle jzj = R, of ????? N , with
A( 1)N +1 kN
(1 + O (1=k)), so
f (z) = A= (z z0 )N + lower order, then ak = (N
1)!z N +k+1
0
N
P
P
Al
+
bk z k
that ak =ak+1 ! z0 as n ! 1. Proof: Write f (z) =
(z z0 )l
l=0
P
analytic, h (z) =
bk z k analytic. We have jbk j 6 1=sk for some s >
1 j
P
1
1
z
R, bk = O 1= jz0 jk . z 1z0 = z10 1 z=z
=
, ( 1)d (z(l z1)!)l =
z0
0
zj
1
z0
1
P
j=l
1
(z z0 )l
1
(z z0 )l
j=0
( 1)l+1
(l 1)!
( 1)l+1
(l 1)!
1
P
k=0
1
P
k=0
1
z0k+l+1
(k + l) : : : (k + 1) z k
1
z0k+l+1
(k + l) : : : (k + 1) z k
26
VI.2.12
1 2 3 P
L
K
LLL
Show that if z0 is an isolated singularity of f (z) that is not removable, then
z0 is an essential singularity for ef (z) .
Solution
Apply Arzela-Weierstrass theorem. Suppose z0 = 0, an isolated singularity
of f (z), not removable. 0 essential ) values of f (z) ???? on C as z ! 0,
! values of ef (z) ????? on C as z ! 0. 0 a pole ) f (z) ! 1, as z ! 0,
monotone, values of f (z) ?????? of 0 cover the extremum of some ??????
fjwj > Rg. In any neighbourhood of 0 there are values of ef (z) = ew that are
near 0 (e m ) and that are near to (e+m ), so ef (z) does not have a limit as
z ! 0. 0 is essential singularity of ef (z) .
27
28
VI.2.14
1 2 3 P L K
Suppose u rei is harmonic on a punctured disk f0 < r < 1g, with Laurent
series as in Exercise 7 of Section 1. Suppose > 0 is such that r u rei ! 0
as r ! 0. Show that an = 0 = bn for n
.
Solution
1
1
P
P
u rei harmonic, 0 < r < 1. u =
an rn cos n +
bn rn sin n + c log r.
n=0
n6=0
Let m > , m
> 0. Have am rm + a m r m = O (r ), am r2m + a m =
O (r +m ) as r ! 0. Let r ! 0, get a m = 0 for m > . Similarly b m = 0
for m > .
29
VI.2.15
1 2 3 P L K
Suppose u (z) is harmonic on the punctured disk f0 < jzj < g.
Show that if
u (z)
!0
log (1= jzj)
30
VI.3.1
1 2 3 P L K
Consider the functions in Exercise 1 of Section 2 above. Determine which
have isolated singularities at 1, and classify them.
Solution
31
VI.3.2
1 2 3 P L K
Suppose that f (z) is an entire function that is not a polynomial. What kind
of singularity can f (z) have at 1?
Solution
32
VI.3.3
1 2 3 P L K
Show that if f (z) is nonconstant entire function, then ef (z) has an essential
singularity at z = 1.
Solution
33
VI.3.4
1 2 3 P L K
Show that each branch of the following functions is meromorphic at 1, and
obtain the series expansion
for each branch at
p
p 1.
5=2
3
2
3
(a) (z
1)
(b)
(z
1)
(c)
z 2 1=z
Solution
34
VI.4.1
1 2 3 P
L
K
LLL
Find the partial fractions decompositions of the following functions.
z 1
(a) z21 z
(c) (z+1)(z12 +2z+2)
(e) z+1
2
z 2 +1
1
(b) z(z
(d) (z2 +1)
(f) zz2 4z+3
2
2 1)
z 6
(a)
1
z2
A
B
+
z
z 1
A = lim z f (z) =
z!0
B = lim (z
z!1
1
z2
1
z
1) f (z) =
z=0
1
z
=1
z=1
1
1
+
z z 1
(b)
z2 + 1
A
B
C
= +
+
2
z (z
1)
z
z 1 z+1
A = lim z f (z) =
z!0
z2 + 1
z2 1
z=0
2
z +1
=1
z!1
z (z + 1) z=1
z2 + 1
C = lim (z + 1) f (z) =
=1
z! 1
z (z 1) z= 1
B = lim (z
1) f (z) =
z2 + 1
=
z (z 2 1)
1
1
1
+
+
z z 1 z+1
(c)
1
(z +
1) (z 2
+ 2z + 2)
A
B
C
+
+
z+1 z+1+i z+1
35
A =
B =
C =
lim z f (z) =
z! 1
(z 2
1
+ 2z + 2)
=1
z= 1
1
(z + 1) (z + 1 i)
1
i) f (z) =
(z + 1) (z + 1 + i)
lim (z + 1 + i) f (z) =
z! 1 i
lim (z + 1
z! 1+i
1
(z +
1) (z 2
+ 2z + 2)
1
z+1
1=2
z+1+i
z= 1 i
=
z= 1+i
1=2
z+1 i
A = lim (z
1
=
(z 2 + 1)2
1=4
(z i)2
i=4
(z i)
1=4
i=4
2 +
z i
(z + i)
(e)
z 1
A
=
+B
z+1
z+1
A =
B =
lim (z + 1) f (z) = z
z! 1
lim f (z) = 1
z!1
36
1jz=
1
2
1
2
z 1
=1
z+1
2
z+1
(f)
z2
z2
A =
lim (z + 2) f (z) =
z! 2
B = lim (z
z!3
C =
A
B
4z + 3
=
+
+C
z 6
z+2 z 3
3) f (z) =
z2
4z + 3
z 3
z 2 4z + 3
z (z + 1)
lim f (z) = 1
z!1
z2
z2
4z + 3
3
=
+1
z 6
z+2
37
=
z= 2
=0
z=3
VI.4.2
1 2 3 P
L
K
LLL
Use the division algorithm to obtain the partial fractions decomposition of
the following functions
3
9
z6
(a) zz2 +1
(b) zz6 +11
(c) (z2 +1)(z
+1
1)2
(a)
z3 + 1
z+1
A
B
=
z
+
=
z
+
+
z2 + 1
z2 + 1
z+i z i
A =
1 1
z+1
=
+ i
z i z= i
2 2
1 1
z+1
=
i) f (z) =
i
z + i z=i
2 2
lim (z + i) f (z) =
z! i
B = lim (z
z!i
z+1
( 1 + i) =2
z3 + 1
=z+ 2
=z+
2
z +1
z +1
z+i
(1 + i) =2
z i
(b)
3
z9 + 1
1
3 z +1
=
z
+
= z3+ 3
= z3+
6
6
z
1
z
1
z
1
(z
1
A
B
C
= z3+
+
+
2
1) (z + z + 1)
z 1 z w z w
A = lim (z
z!1
B = lim (z
z!w
C = lim (z
z!w
1) f (z) =
z2
1
+z+1
=
z=1
i=3
, and z3 = e4
1
3
w
1
=
w) f (z) =
(z 1) (z e 2 i=3 ) z=e2 i=3
3
1
w
w)2 f (z) =
=
2
i=3
(z 1) (z e
) z=e 2 i=3
3
z9 + 1
1=3
w=3
w=3
= z3 +
+
+
;
6
z
1
z 1 z w z w
38
where w = e2
i=3
i=3
. Set
1=3
z 1
z6
(z 2 + 1) (z 2
z4
z6
2z 3 + 2z 2
w=3
z w
2z + 1)
2z + 1
w=3
z+w
z4
z6
2z 3 + 2z 2
= z 2 + 2z + 2 +
2z + 1
2z 3 z 2 + 2z 2
(z 2 + 1) (z 1)2
2z 3 z 2 + 2z 2
A
B
C
D
=
+
+
+
2
2
z + i z i (z 1)
z 1
(z 2 + 1) (z 1)
A =
2z 3 z 2 + 2z 2
=
(z i) (z 1)2 z= i
2z 3 z 2 + 2z 2
1
i) f (z) =
=
2
4
(z + i) (z 1) z=i
2z 3 z 2 + 2z 2
1
1)2 f (z) =
=
2
(z + 1)
2
z=1
3
2
d 2z
z + 2z 2
1)2 f (z) =
=
dz
(z 2 + 1)
z=1
z! i
B = lim (z
z!i
C = lim (z
z!1
D = lim (z
z!1
39
1
4
5
2
VI.4.3
1 2 3 P
L
K
LLL
Let V be the complex vector space of functions that are analytic on the
extended complex plane except possibly at the points 0 and i, where they
have poles of order at most two. What is the dimension of V ? Write down
explicitly a vector space basis for V .
Solution
40
VI.5.1
1 2 3 P L K
Show that if f (z) and g (z) have period w, then so do f (z) +g (z) and
f (z) g (z).
Solution
41
VI.5.2
1 2 3 P L K
Expand 1= cos (2 z) in a series of powers of e2 iz that converges in the upper
half-plane. Determine where the series converge absolutely and where it
converges uniformly.
Solution
42
VI.5.3
1 2 3 P L K
Expand tan z in a series of powers of exponentials eikz , 1 < k < 1, that
converges in the upper half-plane. Also nd an expansion of tan z as in an
exponential series that converges in the lower half-plane.
Solution
43
VI.5.4
1 2 3 P L K
Let f (z) be an analytic function in the upper half-plane that is periodic,
with real period 2
> 0. Suppose that there are A; C > 0 such that
jf (x + iy)j CeAy for y > 0. Show that
X
f (z) =
an einz= ;
n
44
VI.5.5
1 2 3 P L K
Suppose that 1 are periods of a nonzero doubly periodic function f (z), and
suppose that there are no periods w of f (z) satisfying 0 < jwj < 1. How
many periods of f (z) lie on the unit circle? Describe the possibilities, and
sketch the set of periods for each possibility.
Solution
45
VI.5.6
1 2 3 P L K
We say that w1 and w2 generate the periods of a double periodic function
if the periods of the function are precisely the complex numbers of the form
mw1 + nw2 where m and n are integers. Show that if w1 and w2 generate
the periods of a doubly periodic function f (z), and if 1 and 2 are complex
numbers, then 1 and 2 generate the periods of f (z) if and only if there
is a 2 2 matrix A with integer entries and with determinant 1 such that
A (w1 ; w2 ) = ( 1 ; 2 ).
Solution
46
VI.5.7
1 2 3 P L K
Let w1 and w2 be two complex numbers that do lie on the same line through
0. Let k 3. Show that the series
1
X
m;n= 1
1
(z
converges uniformly on any bounded subset of the complex plane to double periodic meromorphic function f (z), whose periods are generated by
w1 and w2 . Strategy. Show that the number of periods in any annulus
fN jzj N + 1g is bounded by CN for some constant C.
Solution
47
VI.6.1
1 2 3 P L K
. Find the
Consider the continuous function f ei = j j,
i
complex Fourier series f e and show that it can be expressed as a cosine
series. Sketch the graphs of the rst three partial sums of the cosine series.
Discuss the convergence of the series. Does it converge uniformly? Partial
answer. The cosine series is
j j=
4
2
cos +
1
1
cos 3 + 2 cos 5 +
2
3
5
Solution
48
VI.6.2
1 2 3 P L K
<
(the principal value of the argument). Find the
Let f ei ,
complex Fourier series of f ei and the sine series of f ei . Show that the
complex Fourier series diverges at =
, while the sine series converges
at
. Dierentiate the complex Fourier series term by term and determine
where the dierentiated series converges.
Solution
49
VI.6.3
1 2 3 P L K
Consider the continuous function f ei = 2 ,
. Find the
i
complex Fourier series of f e and show that it can be expressed as a cosine
series. Discuss the convergence of the series. Does it converge uniformly?
By substituting = 0, show that
2
12
=1
1
1
+
22 32
Solution
50
1
+
42
VI.6.4
1 2 3 P L K
Consider the continuous function f ei = 4 2 2 2 ,
. Find
i
the complex Fourier series of f e and show that it can be expressed as a
cosine series. Relate the Fourier series to the series of the function in Exercise
3.
Solution
51
VI.6.5
1 2 3 P L K
P
P ik
Show that if
ck converges absolutely, then
ck e converges absolutely
for each , and the series converges uniformly for
.
Solution
52
VI.6.6
1 2 3 P L K
Show that any function f ei on the unit circle with absolutely convergent
Fourier series has the form f ei = g ei + h (ei ), where g (z) and h (z)
are continuous functions on the unit circle that extend continuously to be
analytic on the open unit disk.
Solution
53
VI.6.7
1 2 3 P L K
If f ei
f e
1
X
d
=
jck j2 :
2
k= 1
54
VI.6.8
1 2 3 P L K
By applying Parsevals identity to the picewise constant function with series
(6.5), show that
2
1
1
1
+ 2+ 2+
:
2
8
3
5
7
Use this identity and som algebraic manipulation to show that
=1+
=1+
1
1
1
+ 2+ 2+
2
2
3
4
Solution
55
VI.6.9
1 2 3 P L K
By applying Parsevals identity to the function of Exercise 1, show that
4
1
1
1
+ 4+ 4+
:
4
96
3
5
7
Use this identity and som algebraic manipulation to show that
=1+
90
=1+
1
1
1
+ 4+ 4+
4
2
3
4
Solution
56
VI.6.10
1 2 3 P L K
If f (z) is analytic P
in some annulus containing the unit circle jzj = 1, with
Laurent expansion
ak z k , then
1
2
jzj=1
jf (z)j2 jdzj =
Solution
57
1
X
k= 1
jak j2 :
VI.6.11
1 2 3 P L K
Let f ei be a continuous function on the unit circle, with Fourier series
P
ck eik . Show that f ei extends to be analytic on some annulus containing the unit circle if and only if there exist r < 1 and C > 0 such that
jck j Crjkj for 1 < k < 1.
Solution
58
VI.6.12
1 2 3 P L K
Using the convergence theorem for Fourier series, prove that every contintinously function on the unit circle in the complex plane can be approximated
uniformly there by trigonometric polynomials, that is, by nite linear combinations of exponentials eik , 1 < k < 1. Strategy. First approximate
f eik by a smooth function.
Solution
59
VI.6.13
1 2 3 P L K
Let D be a domain bounded by a smooth boundary curve of length 2 . We
parametrize the boundary of D by arc length s, sot the
is given by
P boundary
iks
a smooth periodic function (s), 0
s
2 . Let
ck e be the Fourier
series of (s).
(a)
P 2
Show that
k jck j2 = 1. Hint. Apply Parsevals identity to 0 (s) and use
j 0 (s)j = 1 for a curve parameterized by arc length.
(b)
P
Show that the area of D is
k jck j2 . Hint. Use Exercise IV.1.4.
(c)
Show that the area of D is
, with equality if and only if D is a disk.
Remark. This proves the isoperimetric theorem. Among all smooth closed
curves of a given length, the curve that surrounds the largest area is a circle.
Solution
60
VI.1.14
1 2 3 P L K
Show that
Z
f e
n
X
k= m
2
ik
bk e
d
=
2
f e
n
X
k= m
2
ik
ck e
n
X
d
+
jbk
2 k= m
ck j2 ;
61
VI.1.15
1 2 3 P L K
Show that a continously dierentiable function on the unit circle has an
absolutely convergent Fourier series. Strategy. Write the Fourier coe cients
ck of f ei as ak bk , where ak = 1=ik and bk is the Fourier coe cient of
the derivative.
qPUse Bessel
qP s inequality and the Cauchy-Schwarz inequality
P
2
j kj
j k j2 .
k k
Solution
62
VI.6.16
1 2 3 P L K
Let f ei be a continuous function on the unit circle. Suppose that f ei
is piecewise continuously dierentiable, in the sense that it has a continuous
derivative except at a nite number of points, at each of which the derivative
has limits from the left and from the right. Show that the Fourier series
f ei is absolutely convergent. Strategy. Cancel the discontinuities of the
derivative using translates of the function in Exercise 3, whose Fourier series
is absolutely convergent.
Solution
63
VI.6.17
1 2 3 P L K
Let f ei be a piecewise continuously dierentiable, in the sense that it is
continuously dierentiable except at a nite number of points, at each of
which both the function and its derivative have limits from the left and from
the right. Show that the Fourier series of f ei converges at each point, to
f ei if the function is continuous at ei , and otherwise to the average of
right. Strategy. Show that
the limits of f ei from
P the ileft and from the
i
i
i
satises the hypotheses of
bj hj e , where f1 e
= f1 e +
f e
Exercise 15, and each hj ei is obtained from the function of Exercise 2 by
change of variable 7 !
j.
Solution
64
VII
1
2
3
4
5
6
7
8
1
X
X
X
X
X
X
2
X
X
X
X
X
X
X
3
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
6
X
X
X
X
X
X
X X X
X
X X X
X
10 11 12 13 14 15 16 17 18 19
X
X
VII.1.1
Evaluate the following residues.
z
(d) Res sin
(a) Res z21+4 ; 2i
;0
z2
(b) Res
(c) Res
1
;
z 2 +4
1
;1
z5 1
(g) Res
cos z
;0
z2
(e) Res
(h) Res
(f) Res [cot z; 0] (i) Res
2i
z
;1
Log z
z
e
;0
z 5n
z +1
; wk
zn 1
Solution
a)
By rule 4,
Res
z2
1
1
; 2i =
+4
2z
1
=
4i
=
z=2i
i
:
4
b)
By rule 4,
Res
z2
1
1
; 2i =
+4
2z
=
z= 2i
1
i
= :
4i
4
c)
By rule 4,
Res
1
z5
;1 =
1
5z 4
z=1
1
= :
5
d)
By rule 1,
Res
sin z
sin z
; 0 = lim
= 1:
2
z!0 z
z
e)
By rule 2,
Res
f)
By rule 3,
h cos z
z2
i
d
; 0 = lim cos z =
z!0 dz
sin zjz=0 = 0:
i
cos z
;0 =
sin z
cos z
h cos z
2
= 1:
z=0
g)
By rule 3,
Res
z
z
;1 =
Log z
1=z
= 1:
z=1
h)
By Laurent expansion
1+
ez
=
z5
z
1!
z2
2!
z3
3!
z5
z4
4!
+ o (z 5 )
1
1
1
1
1
+ o (1) ;
+
+
+
+
z 5 1!z 4 2!z 3 3!z 2 4!z
hence
Res
ez
1
1
;0 =
= :
5
z
4!
24
i)
We have poles of the form wk = e2
that wkn = 1, by rule 3,
Res
zn + 1
zn + 1
;
w
=
k
zn 1
nz n 1
ik=n
=
z=wk
, where k = 0; 1; 2; : : : n
wkn + 1
2
n 1 =
nwk
nwkn
1, remark
2e2 ik=n
2wk
=
:
n
n
VII.1.2
Calculate the residue at each singularity in the complex plane of
the following functions.
z
(d) z21+z
(a) e1=z (b) tan z (c) (z2 +1)
2
Solution
(a)
The function e1=z has a singularity at z = 0, by Laurent expansion
e1=z = 1 +
1
1
+
+ ::::
z 2!z 2
Thus by denition
Res e1=z ; 0 = 1:
(b)
sin z
has isolated singularities at z =
The function tan z = cos
z
n < 1, they are simple poles.
By rule 3,
Res
z
(z 2 +1)2
sin z
; =2 + n
cos z
sin z
sin z
z
d
z
;i =
2
dz (z + i)2
(z 2 + 1)
z
d
z
Res
2; i =
2
dz (z i)2
(z + 1)
=
z=i
=
z= i
(d)
+ n,
i)2
(z
1:
z= =2+n
(z + i)2 2z (z + i)
(z + i)4
(z
1<
2z (z
i)4
= 0;
z=i
i)
= 0:
z= i
The function
simple poles.
By rule 1,
1
z(z+1)
Res
1
1
1
; 0 = lim
=
z!0 z + 1
z (z + 1)
z+1
Res
1
1
1
; 1 = lim =
z! 1 z
z (z + 1)
z
= 1;
z=0
=
z= 1
1:
1 they are
VII.1.3
Evaluate
the followingR integrals using Rresidue theorem.
R
z
(a) jzj=1 sin
dz (c) jzj=2 cosz z dz (e) jz 1j=1 z81 1 dz
z2
R
R
R
z
z4
dz (f) jz 1=2j=3=2 tanz z dz
(b) jzj=2 z2e 1 dz (d) jzj=1 sin
z
Solution
(a)
Using the residue theorem and rule 1,
Z
jzj=1
sin z
sin z
sin z
dz = 2 i Res
; 0 = 2 i lim
2
2
z!0
z
z
z
= 2 i (1) = 2 i:
(b)
Using the residue theorem and rule 3,
Z
jzj=2
ez
z2
dz =
= 2 i Res
=2 i
ez
z2
1
z
e
ez
+
2z z=1 2z
; 1 + Res
ez
z2
=2 i
z= 1
1
e1
; 1
e
=
1
=
= 2 i sinh 1:
(c)
Using the residue theorem and rule 3,
Z
jzj=2
h z
i
h z
i
z
dz = 2 i Res
;
+ Res
;
=
cos z
cos z 2
cos z
2
!
z
z
=2 i
+
=2 i
sin z z= =2
sin z z= =2
2
=
=
(d)
Using the residue theorem and rule 3,
6
2 2 i:
jzj=1
z4
z4
dz = 2 i Res
;0 = 2 i
sin z
sin z
z4
cos z
= 0:
z=0
(e)
VII.1.3e
|z - 1| = 1
Using the residue theorem with the contour in Figure VII.1.3e and rule 4,
Z
dz = 2 i Res
8
1
jz 1j=1 z
1
1
+ 7
=2 i
7
8z z=1 8z
1
z8
z=ei
=4
; ei
=4
+ Res
z8 1
z8
i
=
1 + e i=4 + e i=4 =
4
i
=4
z=e
p
1 i 1+i
i
1+ 2 :
1+ p + p
=
4
2
2
1
8z 7
i
=
4
; 1 + Res
(f)
z=z
The function tanz z = sin
has removable singularity at z = 0, and a isolated
cos z
singularity at z = 2 inside the domain jz 1=2j = 3=2. Using the residue
theorem and rule 3,
;e
i =4
jz 1=2j=3=2
tan z
dz =
z
sin z=z
= 2 i Res
;
=2 i
cos z 2
sin z=z
lim
z! =2
sin z
=2 i
sin z=z
sin z
z= =2
= (2 i)
4i:
VII.1.4
1 2 3 P L K
Suppose P (z) and Q (z) are polynomials such that the zeros of Q (z)
are simple zeros at the points z1 ; : : : ; zm ; and deg P (z) < deg Q (z).
Show that the partial fractions decomposition P (z) =Q (z) is given
by
P (z) X P (zj ) 1
=
:
0 (z ) z
Q (z)
Q
z
j
j
j=1
m
Solution
Since the zeros of Q (z) are simple, the polynomial quotient P (z) =Q (z) has
at most simple poles. By Rule 3, the residue at zj is
P (z) =Q0 (z) :
Hence the principal part of P (z) =Q (z) at zj is
P (zj ) 1
:
Q0 (zj ) z zj
Since deg P (z) < deg Q (z), then
P (z)
!0
Q (z)
as z ! 1, and P (z) =Q (z) has no principal part at 1.
By partial fraction decomposition our polynomial quotient P (z) =Q (z)
can be written as the sum of its principal parts at its pole,
P (z) X P (zj ) 1
=
:
0 (z ) z
Q (z)
Q
z
j
j
j=1
m
or
VII.1.5
1 2 3 P L K
Let f (z) be a meromorphic function on the complex plane that is
doubly periodic, and suppose that none of the poles of f (z) lie on
the boundary of the period parallelogram P constructed in Section
VI.5. By integrating f (z) around of P , show that the sum of the
residues at the poles of f (z) in P is zero. Conclude that there is no
doubly periodic meromorphic function with only one pole, a simple
pole, in the period parallelogram.
Solution
R
Follow directions. @P = 0, the curve integrals over opposite sides cancel.
Thus the sum of the residues is zero. In particular, there can not be a single
nonzero residue.
10
VII.1.6
Consider the integral
Z
@DR
i(z 1=2)2
e
1
2 iz
dz;
as R ! 1.
(c)
Use (a) and (b) to show that
Z 1
s2
ds =
Solution
(a)
VII.1.6
2
z1
Set
11
I=
@DR
i.e, we integrate
e
1
i(z 1=2)2
2 iz
dz;
e i(z 1=2)
f (z) =
1 e 2 iz
along the parallelogram contour in Figure VII.1.6. The parallelogram has
vertices at 1=2 (1 + i) R.
Residue at a simple pole at z1 = 0, where by Rule 3,
"
#
2
2
e i(z 1=2)
e i(z 1=2)
e i=4
Res
;
0
=
=
:
1 e 2 iz
2 ie 2 iz
2 i
z=0
@DR
i.e.,
i(z 1=2)2
e
1
@DR
e
1
dz = 2 i
iz
i(z 1=2)2
2 iz
(b)
Integrate along 1 parametrized by z =
let R ! 1. This gives
Z
R
R
1+i
dz = p :
2
1
2
+ (1 + i) t with
R, and
i(z 1=2)2
z = 12 + (1 + i) t
=
dz = (1 + i) dt
e 2
R
Z R
2
2
e i((1+i)t) (1 + i)
e 2 t
dt = (1 + i)
dt !
1
2 i(1+i)t
R 1+e
1 e 2 i( 2 +(1+i)t)
Z 1
2
e 2 t
! (1 + i)
dt:
2 i(1+i)t
1 1+e
f (z) dz =
1
e i=4
;
2 i
e
1
dz =
iz
t+
1
2
+ (1 + i) R with 0
f (z) dz =
2
=
=
e
1
2
e
e
i(z 1=2)2
dz =
iz
i( t+(1+i)R)2
2 i( t+1=2+(1+i)R)
2
2
e (t 2t)i e 2 (R Rt)
sup
e (2t 1 2R)i e2 R
0 t 1 1
t + 12 + (1 + i) R
dt
z =
dz =
dt =
sup
R
R
Rt)
(R2
Rt)
(2t 1 2R)i e2 R jj
= sup
Rj
1
2
e
je
0 t 1
e2
(R2
Rt)
+ (1 + i) t with
=
! 0:
t
R,
i(z 1=2)2
1
z =
+ (1 + i) t
2
=
2
dz = (1 + i) dt
e
R
Z R 2 i(1+i)t 2 t2
2
e i( 1+(1+i)t) (1 + i) dt
e
e
dt
!
= (1 + i)
1
2
i(1+i)t
1+e
R
1 e 2 i( 2 +(1+i)t)
Z 1 2 i(1+i)t 2 t2
e
e
dt
! (1 + i)
dt:
2
i(1+i)t
1+e
1
f (z) dz =
3
2
2
e (t 2t)i e 2 (R Rt)
dt
1 e (2t 1 2R)i e2 R
0 t 1 j1
2
e 2 (R
= sup
e2
0 t 1 j1
e
1
dzdz =
iz
parameterized by z = t
13
1
2
(1 + i) R with 0
1,
f (z) dz =
4
=
Z
e i(t
1 e 2
sup
1
i(t
e (2R
dz =
iz
2Rt+t2 2t+1)i
0 t 1
i(z 1=2)2
z = t 12 (1 + i) R
=
dz = dt
e 2
Z 1 (2R 2Rt+t2 2t+1)i 2 (R2 +R
(1+i)R)2
e
e
dz =
1=2 (1+i)R)
(2t
1
2R)i
1 e
e 2 R
0
e
1
4
(R2 +R
Rt)
(2t 1 2R)i e 2 R
e
je
sup
0 t 1 j1
sup
0 t 1
(R2 +R
dz =
Rt)
(2t 1 2R)i e 2 R jj
(R2 +R
j1
Rt)
Rt)
2 Rj
! 0:
1
1
e
1+e
2 t2
2 i(1+i)t
dt + 0 + (1 + i)
1
1
2 i(1+i)t
1+e
2 t2
dt
2 i(1+i)t
dt + 0 =
= 2 i
and therefore
(1 + i)
i.e.,
2 t2
(1 + i)
1 + e 2 i(1+i)t
1+i
dt = p ;
2
i(1+i)t
1+e
2
2 t2
(c)
Comparing real parts thus yields,
Z 1
e
1+i
dt = p :
2
1
dt = p ;
2
1
and changing variables as follows gives the desired result
2 t2
14
e i=4
;
2 i
1
p =
2
i.e.,
2 t2
dt =
p
s = p2 t
ds =
2 dt
s2
dt =
15
1
=p
2
1
1
s2
ds;
VII.2.1
1 2 3 P L K
Show using residue theory that
Z 1
dx
= ;
2
2
a
1 x +a
a > 0:
VII.2.1
2
z1
-R
R
z2
Set
I=
and integrate
f (z) =
z2
1
1
dx
;
x 2 + a2
1
=
+ a2
(z
1
ia) (z + ia)
z2
1
1
; ia =
2
+a
2z
16
=
z=ia
i
:
2a
2,
! 0:
i
2a
and therefore
Z
1
1
x2
dx
= ;
+ a2
a
a > 0:
Remark.
Check of the result by evaluating the integral directly gives, because the
integrand is positive the integral can be computed taking a single limit,
namely,
Z
1
1
dx
= lim
2
x + a2 b!1
b
b
dx
1
x
= lim
arctan
2
2
b!1 a
x +a
a
17
=
b
a > 0:
VII.2.2
1 2 3 P L K
Show using residue theory that
Z 1
dx
= 3:
2
2 2
2a
1 (x + a )
Remark. Check the result by dierentiating the formula in the preceding exercise with respect to the parameter.
Solution
VII.2.2
2
z1
-R
R
z2
Set
I=
1
1
(x2
dx
;
+ a2 )2
and integrate
f (z) =
(z 2
1
=
+ a2 )2
(z
1
ia) (z + ia)2
2
1
d
1
2
; ia = lim
=
2
2
z!ia dz (z + ia)
(z 2 + a2 )
(z + ia)3
18
=
z=ia
i
:
4a3
Integrate along
Z
1,
Integrate along
2,
f (z) dz
a2 )2
(R2
R3
! 0:
I +0=2 i
and therefore
Z
1
1
(x2
dx
=
:
+ a2 )2 2a3
Remark.
From Exercise VII.2.1 we have the formula
Z 1
1
dx = ;
(1)
2
2
a
1 x +a
a > 0:
(2)
d
da
1
1
1
dx =
2
x + a2
1
1
2a
dx =
2
(x + a2 )2
2a
1
1
(x2
dx
;
+ a2 )2
and dierentiate the right hand side in the formula in the same way
(3)
d
=
da a
a2
1
1
dx
=
:
(x2 + a2 )2 2a3
20
VII.2.3
Show using residue theory that
Z 1
x2 dx
:
2 =
2
2
1 (x + 1)
Remark. Check the result by combining the preceding two exercises.
Solution
VII.2.3
2
z1
-R
z2
Set
I=
1
1
x2 dx
;
(x2 + 1)2
and integrate
f (z) =
z2
=
(z 2 + 1)2
(z
z2
i)2 (z + i)2
Res
z2
d
z2
2z (z + i)2 2z 2 (z + i)
;
i
=
lim
=
z!i dz (z 2 + i)2
(z 2 + 1)2
(z + i)4
Integrate along
1,
=
z=i
i
:
4
f (z) dz =
1
Integrate along
2,
R
R
x2 dx
!
(x2 + 1)2
1
1
x2 dx
= I:
(x2 + 1)2
f (z) dz
1)2
(R2
! 0:
i
4
and therefore
Z
1
1
x2 dx
= :
(x2 + 1)2 2
Remark.
We split the integrand using partial fraction, and put a = 1 in the formulas
from the preceding two exercises
Z 1
Z 1
Z 1
x2 dx
dx
dx
=
= :
2
2 =
2
2
2
2
2
1 (x + 1)
1 x +1
1 (x + 1)
22
VII.2.4
Using residue theory, show that
Z 1
dx
=p :
4
2
1 x +1
Solution
VII.2.4
2
-R
z2
z1
z3
z4
Set
I=
and integrate
f (z) =
z4
1
=
+1
z
1
1
dx
;
+1
x4
1
1+i
p
2
1+i
p
2
1 i
p
2
1+i
p
,
2
where by Rule 2,
23
i)
1p i
2
1
;
Res 4
z +1
Integrate along
Integrate along
1
1+i
p
= 3
4z
2
z=( 1+i)= 2
2
(1
8
i)
1,
R3
! 0:
1
1
dx
=p :
+1
2
x4
24
VII.2.5
Using residue theory, show that
Z 1
x2
dx = p :
x4 + 1
2 2
0
Solution
VII.2.5
2
-R
z2
z1
z3
z4
Set
I=
and integrate
z2
f (z) = 4
=
z +1
z
x2
dx ) 2I =
x4 + 1
1
1
x2
dx
x4 + 1
z2
1+i
p
2
1+i
p
2
1 i
p
2
1+i
p
,
2
where by Rule 3,
25
i) :
1p i
2
z2
;
Res 4
z +1
Integrate along
Z
1,
Integrate along
2,
1
1+i
p
=
4z
2
z=( 1+i)= 2
2
( 1
8
i) :
! 0:
x2
dx = p :
x4 + 1
2 2
26
VII.2.6
Show that
Z
1
1
(x2
x
dx =
+ 2x + 2) (x2 + 4)
10
Solution
VII.2.6
2
z1
z2
-R
z3
z4
Set
I=
and integrate
f (z) =
(z 2
1
1
(x2
x
dx
+ 2x + 2) (x2 + 4)
z
=
+ 2z + 2) (z 2 + 4)
(z
( 1 + i)) (z
z
2i) (z
( 1
i)) (z + 2i)
Res
(z 2
z
z
; 2i = 2
2
+ 2z + 2) (z + 4)
(z + 2z + 2) 2z
=
z=2i
1 + i, where by Rule 3,
27
1
20
1
i:
10
z
z
;
1
+
i
=
(z 2 + 2z + 2) (z 2 + 4)
(2z + 2) (z 2 + 4)
Res
Integrate along
Z
f (z) dz =
1
1,
=
z= 1+i
1
3
+ i:
20 20
R
R
(x2
x
dx !
+ 2x + 2) (x2 + 4)
Z 1
x
!
dx = I:
2
2
1 (x + 2x + 2) (x + 4)
R2
! 0:
1
20
1
1
3
i+
+ i
10
20 20
and therefore
Z
1
1
(x2
x
dx =
+ 2x + 2) (x2 + 4)
28
10
VII.2.7
1 2 3 P
Show that
Z 1
L
K
LLL
cos (ax)
dx = p e
x4 + 1
2
p
a= 2
a
a
cos p + sin p
2
2
a > 0:
Solution
VII.2.7
2
-R
z2
z1
z3
z4
Set
I=
and integrate
f (z) =
eiaz
=
z4 + 1
z
1
1
cos (ax)
dx
x4 + 1
a > 0;
eiaz
1+i
p
2
1+i
p
2
29
1 i
p
2
1p i
2
eiaz
eiaz 1 + i
p
=
;
=
z4 + 1
4z 3 z=(1+i)=p2
2
p a=p2
p a=p2
p
p
p
2e
2e
2e
cos a= 2
sin a= 2
i
8
8
Res
=
eiaz
;
z4 + 1
Integrate along
Z
f (z) dz =
1
1,
R
R
1+i
p
,
2
p
a= 2
p
p
2e
cos a= 2
i+
8
p
z=( 1+i)= 2
p
a= 2
p
p
sin a= 2
2e
i
8
p
a= 2
p
p
cos a= 2
2e
i
8
p
a= 2
I+i
p
sin a= 2
8
eax
dx !
x4 + 1
1
1
cos (ax)
dx+i
x4 + 1
1
1
sin (ax)
dx = I+i
x4 + 1
1
1
sin (ax)
dx
x4 + 1
p
sin a= 2
8
where by Rule 3,
p
a= 2
sin (ax)
dx = 2 i
x4 + 1
and therefore
Z 1
cos (ax)
dx = p e
4
2
1 x +1
2e
p
a= 2
p
a= 2
p
cos a= 2
i
4
a
a
cos p + sin p
2
2
30
2e
p
a= 2
p !
sin a= 2
i ;
4
a > 0:
VII.2.8
Show that
Z
1
1
cos x
dx = :
e
(x2 + 1)2
Solution
VII.2.8
2
z1
-R
z2
Set
I=
1
1
cos x
dx
(x2 + 1)2
and integrate
f (z) =
eiz
=
(z 2 + 1)2
(z
eiz
i)2 (z + i)2
Res
eiz
d
eiz
ieiz (z + i)2 2eiz (z + i)
;
i
=
lim
==
z!i dz (z + i)2
(z 2 + 1)2
(z + i)4
Integrate along
1,
=
z=i
e
2
f (z) dz =
1
1
1
eix
dx !
(x2 + 1)2
Integrate along
plane, this gives
2,
1
1
cos x
dx + i
(x2 + 1)2
sin x
dx =
+ 1)2
1
Z 1
sin x
=I +i
2 dx:
2
1 (x + 1)
f (z) dz
1)2
(R2
R3
(x2
! 0:
1
1
cos x
dx = :
e
(x2 + 1)2
32
VII.2.9
Show that
Z
1
1
sin2 x
dx =
1
x2 + 1
2
1
:
e2
Solution
VII.2.9
2
z1
-R
z2
Set
I=
and integrate
1
1
f (z) =
sin2 x
dx =
x2 + 1
1
1
1 cos 2x
dx
2 (x2 + 1)
1 ei2z
1 ei2z
=
2 (z 2 + 1)
2 (z i) (z + i)
z=i
33
1
= ie
4
1
i
4
f (z) dz =
1
R
R
1 ei2x
dx !
2 (x2 + 1)
Integrate along
plane, this gives
2,
1
1
1 cos 2x
dx + i
2 (x2 + 1)
sin 2x
dx =
2
1 2 (x + 1)
Z 1
sin 2x
dx
=I +i
2
1 2 (x + 1)
2
2(R2
1)
sin2 x
dx =
1
x2 + 1
2
34
! 0:
we obtain that
1
ie
4
and therefore
1
:
e2
1
i ;
4
VII.2.10
Show that
Z
1
1
cos (ax)
e jajb
;
dx
=
x 2 + b2
b
For which complex values of the parameters a and b does the integral exist? Where does the integral depend analytically on the
parameters?
Solution
The integral exists only for a real, and Re b 6= 0 or cos (iab) = 0. It depends
analytically on b for Re b 6= 0. It depends analytically on a in the intervals
1 < a < 0 and 0 < a < 1.
Case 1: a < 0
VII.2.10a
z1
-R
R
z2
Set
I=
and integrate
1
1
f (z) =
cos ax
dx;
x 2 + b2
a < 0; b > 0
eiaz
=
z 2 + b2
(z
eiaz
ib) (z + ib)
Res
Integrate along
Z
eiaz
eiaz
;
ib
=
z 2 + b2
2z
=
z= ib
eab
i
2b
f (z) dz =
1
R
R
eiax
dx !
x 2 + b2
1
1
cos (ax)
dx + i
x 2 + b2
sin (ax)
dx =
2
2
1 x +b
Z 1
sin (ax)
dx
=I +i
2
2
1 x +b
Integrate along 2 and let R ! 1. Because jeiaz j 1 in the lower half plane
if a < 0, this gives
Z
1
R
! 0:
f (z) dz
2
R
R
b2
2
Using the Residue Theorem and letting R ! 1, we obtain that
I +0=
2 i
eab
i ;
2b
and hence
(1)
I=
eab
; a < 0:
b
Case 2: a > 0
36
VII.2.10b
4
z1
-R
z2
Set
I=
and integrate
1
1
f (z) =
cos ax
dx;
x2 + b2
eiaz
=
z 2 + b2
(z
a > 0; b > 0;
eiaz
ib) (z + ib)
eiaz
eiaz
;
ib
=
z 2 + b2
2z
e ab
i
2b
=
z=ib
f (z) dz =
3
R
R
eiax
dx !
x 2 + b2
1
1
cos (ax)
dx + i
x 2 + b2
37
sin (ax)
dx =
2
2
1 x +b
Z 1
sin (ax)
=I +i
dx
2
2
1 x +b
Integrate along 4 and let R ! 1. Because jeiaz j 1 in the upper half plane
if a > 0, this gives
Z
1
f (z) dz
! 0:
R
2
R
R
b2
4
Using the Residue Theorem and letting R ! 1, we obtain that
e ab
i ;
2b
I +0=2 i
and hence
(2)
e ab
; a > 0:
b
I=
38
VII.2.11
1 2 3 P L K
Evaluate
Z
cos x
dx:
+ a2 ) (x2 + b2 )
1
Indicate the range of the parameters a and b.
(x2
Solution
The integral converges absolutely if a2 ; b2 62 ( 1; 0], a; b 62 iR. It also
converges absolutely for certain points on the imaginary axis, so that ia
are zeros of cos x, ib are zeros of cos x, but they are not the same zeros.
The integrand, is an even function of a and b, so it su ces to evaluate it for
a; b in the right half-plane. For this, we can assume that a; b > 0, and extend
the formula by analyticity. We can also assume that a 6= b, and get other
case in this limit. So we may as well assume that 0 < a < b.
VII.2.11
2
z2
z1
-R
z3
z4
R1
x
2 2
= ( 1; 0] i.e.
The integral 1 (x2 +acos
2 )(x2 +b2 ) dx converges absolutely if a ; b 2
a; b 2
= iR. It also converges absolutely for certain points on imaginary axis,
so that ia are zeros of cos x, ib are zeros of cos x, but they are not same
zeros. (Problem should may be simplied to avoid thus points.)
The integrand, is an even function of a and b, so it su ces to evaluate it for
a; b in the right half-plane. For this, we can assume that a; b > 0, and extend
the formula by analyticity. We can assume that a 6= b, and get other case in
this limit. So we may as well assume 0 < a < b.
39
1
1
(x2
cos x
dx
+ a2 ) (x2 + b2 )
eiz
=
(z 2 + a2 ) (z 2 + b2 )
(z
f (z) =
ia) (z
eiz
ib) (z + ia) (z + ib)
eiz
eiz
;
ia
=
(z 2 + a2 ) (z 2 + b2 )
2z (z 2 + b2 )
=
z=ai
2a (b2
a2 ) i
f (z) dz =
1
Z 1
!
1
Integrate along
this gives
Z
2b (b2
z=bi
a2 ) i
eix
dx !
2
2
2
2
R (x + a ) (x + b )
Z 1
cos x
sin x
dx + i
dx =
2
2
2
2
2
2
2
2
(x + a ) (x + b )
1 (x + a ) (x + b )
Z 1
sin x
=I +i
dx
2
2
2
2
1 (x + a ) (x + b )
2
f (z) dz
2
(R2
1
a2 ) (R2
b2 )
R3
! 0:
e
2a (b2
e
a2 ) i 2b (b2
40
a2 ) i
and therefore
Z
1
1
be a
cos x
dx =
(x2 + a2 ) (x2 + b2 )
ab (b2
ae b
:
a2 )
1
1
be a
cos x
dx =
(x2 + a2 ) (x2 + b2 )
ab (b2
If we let b ! a, we get
Z
1
1
ae b
(b
=
2
a)
ab (b + a)
cos x
dx = 3 (1 + a) :
2a
(x2 + a2 )2
For a = 1, we have
Z
1
1
cos x
dx = ;
e
(x2 + 1)2
41
a) e
+a e
(b a)
VII.2.12
Let Q (z) be a polynomial of degree m with no zeros
line, and let f (z) be a function that is analytic in the
plane and across the real line. Suppose there is b < m
jf (z)j jzjb for z in the upper half-plane, jzj > 1. Show
Z 1
X
f (z)
f (x)
dx = 2 i
Res
; zj ;
Q (z)
1 Q (x)
on the real
upper half1 such that
that
VII.2.12
2
zj - 1
z3
zj
z2
-R
z1
Set
I=
and integrate
1
1
f (x)
dx
Q (x)
f (z) =
f (z)
Q (z)
Res
f (z)
; zj :
Q (z)
42
Integrate along
Z
Integrate along
Z
1,
f (z) dz =
2
f Rei
d
jQ (Rei )j
Rb
Rm (C + O (1=R))
f Rei
jQ (Rei )j
R (C + O (1=R))
Q.E.D.
1
1
Res
f (z)
; zj
Q (z)
X
f (z)
f (x)
dx = 2 i
Res
; zj
Q (x)
Q (z)
43
CR
! 0:
VII.3.1
Show using residue theory that
Z 2
cos
d =2
2 + cos
0
2
p
3
Solution
VII.3.1
|z| = 1
z3
z2 z1
cos
d :
2 + cos
0
A change of variables as in book page 203-204 gives
I=
I=
cos
d =
2 + cos
jzj=1
1
2
z + z1
2 + 12 z +
1
z
dz 1
=
iz i
z2 + 1
dz:
3
2
jzj=1 z + 4z + z
Integrate
f (z) =
z2 + 1
=
z 3 + 4z 2 + z
z z
z2 + 1
p
3
z
2+
z2 + 1
z2 + 1
;
0
=
z 3 + 4z 2 + z
3z 2 + 8z + 1
44
= 1:
z=0
2+
3, where by Rule 3,
p
z2 + 1
z2 + 1
; 2+ 3 = 2
Res 3
z + 4z 2 + z
3z + 8z + 1
p
z= 2+ 3
1
I=2 i
i
and therefore
Z
cos
d =2
2 + cos
45
2
p
3
p
2 3
:
3
VII.3.2
Show using residue theory that
Z 2
d
2
=p
2
a + b sin
a
0
b2
a > b > 0:
Solution
VII.3.2 (a = 7, b = 6)
|z | = 1
z1
z2
Set
I=
1
d :
a + b sin
I=
1
d =
a + b sin
dz
=
iz
jzj=1
a+b
1
2i
1
z
jzj=1
2
bz 2 + 2azi
Integrate
f (z) =
bz 2
2
+ 2azi
=
b z
a2
b2
p
a+ a2 b2
i
b
dz
2
Res
2
bz + 2azi
a+
a2
b
b2
2
2bz + 2ai
z=
a+ a2
b
1
a2
b2
i ;
and therefore
Z
1
2
d =p
a + b sin
a2
47
b2
a > b > 0:
b2
=
i
i
a2
b2
VII.3.3
Show using residue theory that
Z
h
sin2
d =
a
0 a + cos
a2
i
1 ;
a > 1:
Solution
VII.3.3 (a = 2)
|z| = 1
z3
z2 z1
Set
I=
Z
sin2 (2
t)
=2
t
dt =
=
d = dt
t)
2 a + cos(2
Z 2
Z 2
sin2 t
sin2 (2
t)
dt =
dt:
=
a + cos(2
t)
a + cos t
sin2
d =
a + cos
We have
2I =
sin2
d :
a + cos
2I =
sin2
d =
a + cos
jzj=1
1
2i
a+
1
2
1
z
z+
48
1
z
dz
=
iz
1
2i
jzj=1
(z 2 1)
dz:
z 2 (z 2 + 2az + 1)
Integrate
2
(z 2
p
a + a2 1
(z 2 1)
=
f (z) = 2 2
z (z + 2az + 1)
z2 z
1)
z
a2
p
(z 2 1) =z
Res 2
; a + a2
z + 2az + 1
(z 2 1) =z
1 =
2z + 2a
p
= 2 a2
p
z= a+ a2 1
p
2a + 2 a2
1
2i
1 ;
and therefore
Z
sin2
d =
a + cos
49
a2
1 ;
a > 1:
1:
VII.3.4
Show using residue theory that
Z
d
1 + sin2
2:
Solution
VII.3.4
|z | = 1
z4
z3
z1
z2
Set
I=
1
d =
1 + sin2
=
d =
t
=
dt
Z 0
1
2
1+ sin (
t)
dt =
1
dt:
1+ sin2 t
We have
Z
1
d :
1 + sin2
0
A change of variables as in book page 203-204 gives
I=
I=
1
d =
1 + sin2
jzj=1
1
1+
1
2i
z
50
1
z
dz
=
iz
4
i
jzj=1
z4
z
dz:
6z 2 + 1
Integrate
f (z) =
=
z4
z
=
6z 2 + 1
p
2 1
z
z
z
2+1
2+1
z4
z
;
6z 2 + 1
z
4z
12z
z=
4
i
1
1
p
p
8 2 8 2
and therefore
Z
p
1
d
=
2:
1 + sin2
51
=
2 1)
1
p :
8 2
VII.3.5
Show using residue theory that
Z
d
1 r2
= 1;
2
1 2r cos + r 2
r < 1:
Solution
VII.3.5 (r = 1/2)
|z | = 1
z2
z1
Set
I=
1 r2
d
=
t
=
=
2
d = dt
2r cos + r 2
Z 0
1 r2
dt
=
=
2
2r cos (
t) + r 2
2 1
Z 2
dt
1 r2
=
2
1 + 2r cos t + r 2
0
52
I=
1 r2
dt
=
1 + 2r cos t + r2 2
I
1
1 r2
=
2 jzj=1 1 + 2r 21 z + z1
r2
dz
=
iz
+
I
1
1 r2
=
dz:
2 i jzj=1 (r + z) (rz + 1)
Integrate
f (z) =
1 r2
1 r2
=
(r + z) (rz + 1)
r (z + r) (z + 1=r)
= 1:
z= r
1
(1) ;
2 i
and therefore
Z
1 r2
d
= 1;
2
2r cos + r 2
53
r < 1:
VII.3.6
By expanding both sides of the identity
Z 2
1
1
d
=p
2
2 0 w + cos
w
0:
Solution
Rewrite both sides in the given identity
1
I=
2
(1)
d
=
1 + cos
w
1
w2
1=2
where the integrand in the left side in the identity can be rewritten as
k
X
1
k cos
(
1)
=
:
wk
1 + cos
w
k=0
1
Z 2 X
Z
k
1
1
1 X1 ( 1)k 2
k cos
I=
( 1)
d =
cosk d :
k=0
k=0 w k
2 0
wk
2
0
R2
Because 0 cosk d = 0 if k odd, we have
1
X
1
I =1+
2
k=1
(2)
cos2k d
1
:
w2k
Now, we use the power series expansion (MH p. 192.) for real
x < 1,
(1 + x) = 1+ x+
1)
2!
x2 +
1) (
3!
we have
54
2)
x3 +
and
xn +
1<
=1+
+
1
2
1=2
1
w2
=
1
2
1!
w2
3
2
5
2
7
2
3
2
1
2
1
w2
2!
1
1
2
+
3
2
1
2
3
2
5
2
1
w2
3!
5
2
7
2
1 2k
2
+
1
w2
k!
w2k
X1 1 3
(2k 1) 1
=1+
=
k=1
2k k!
w2k
X1 1 2 3 4 5 : : : (2k 1) 2k 1
=
=1+
k=1
2k k! 2 4 6 : : : 2k
w2k
X1 1 2 3 4 5 : : : (2k 1) 2k 1
=
1+
k=1
2k k! 2k (1 2 3 : : : k)
w2k
X1
X1
(2k)! 1
(2k)! 1
=1+
=
1
+
;
k
k
2k
k=1 2 k!2 k! w
k=1 22k (k!)2 w 2k
4!
i.e.,
(2)
1
w2
1=2
=1+
X1
k=1
(2k)! 1
:
22k (k!)2 w2k
Comparing the two power series expressions (2) and (3) of the both sides in
the identity (1) gives us together with the uniqueness theorem for the power
series
Z 2
1
(2k)!
cos2k d =
;
k 0:
2 0
22k (k!)2
55
VII.3.7
Show using residue theory that
Z 2
d
2 w
;
2 =
(w + cos )
(w2 1)3=2
0
w 2 C n [ 1; 1] :
VII.3.7 (w = 2)
|z| = 1
z2
z1
Set
I=
d
(w + cos )2
I=
d
=
(w + cos )2
4z
2
(z 2 +2wz+1)
p
2
branch of w
Integrate f (z) =
1
w+
1
2
z+
1
z
dz 4
=
2
iz i
z
2 dz:
2
jzj=1 (z + 2wz + 1)
We choose
1 on Cn [ 1; 1] that is positive for w 2 (1; 1).
It su ces to check the identity for w > p
1.
Residue at a double pole at z1 = w + w2 1, where by Rule 2,
56
p
z
; w + w2 1 =
2
(z 2 + 2wz + 1)
z
d
=
lim
p
p
2 =
z! w+ w2 1 dz z
w
w2 1
p
p
2
z
w
w2 1
2z z
w
w2 1
=
=
p
4
p
w2 1
z
w
2
z= w+ w 1
p
2
z
w
w
1
2z
=
=
p
3
p
z
w
w2 1
z= w+ w2 1
1
w
=
:
2
4 (w
1)3=2
Res
1
w
4 (w2 1)3=2
2 w
d
;
2 =
(w + cos )
(w2 1)3=2
w 2 C n [ 1; 1] :
Check.
From Exercise VII.3.6 we have the formula
Z 2
1
d
1
(1)
=p
2
2 0 w + cos
w
1
It is allowed to dierentiate both sides in the formula, because for every
compact interval with w 2 C n [ 1; 1], both the integrand and its derivative
are continuous, and that the primitive of the derivative
is uniformly bounded
R
by an w independent funtion h ( ) such that h ( ) d < 1 on every such
like interval on w.
We start by dierentiating the left hand side in the formula with respect to
the parameter w
57
(2)
d
dw
1
2
d
w + cos
1
2
d
;
(w + cos )2
and dierentiate the right hand side in the formula in the same way
d
1
p
2
dw w
(3)
w
(w2
1)3=2
d
2 w
:
2 =
2
(w + cos )
(w
1)3=2
58
VII.4.1
By integrating around the keyhole contour, show that
Z 1
x a
dx =
;
0 < a < 1:
1+x
sin ( a)
0
Solution
VII.4.1
2
z1
-R
Set
I=
and integrate
x a
dx
1+x
z a
jzj a e ia arg z
f (z) =
=
1+z
1+z
along the keyhole contour in Figure VII.4.1. We make a branchcut for z
along the positive real axis, where 0 < arg z < 2 .
Residue at a simple pole at z1 = 1, where by Rule 3,
Res
jzj
e ia arg z
; 1 = jzj
1+z
59
ia arg z
z= 1
=e
ia
Integrate along
Z
f (z) dz =
1
e ia arg z
dz =
1+z
jzj
1
Integrate along
Integrate along
3,
1,
z = xe0i
dz = dx
x a
dx !
" 1+x
Z 1
x a
dx = I:
!
1+x
0
f (z) dz =
3
jzj
3
Integrate along
e ia arg z
z = xe2 i
dz =
dz = dx
1+z
Z 0 a 2 ia
x e
!
dx = e 2
1
+
x
1
4,
=
ia
"
R
1
x a e 2 ia
dx !
1+x
x a
dx =
1+x
2 ia
I:
f (z) dz
4
2 "1
! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0
Multiplying with e
ia
2 ia
I +0=2 i
, this yields
e
ia
ia
60
I = 2 i;
ia
I=
ia
2 i
e
ia
sin ( a)
i.e.,
Z
x a
dx =
;
1+x
sin ( a)
61
0 < a < 1:
VII.4.2
By integrating around the boundary of a pie-slice domain of aperture 2 =b, show that
Z 1
dx
;
b > 1:
=
1 + xb
b sin ( =b)
0
Remark. Check the result by changing variable and comparing with Exercise
1.
Solution
VII.4.2 (b = 12)
z1
Set
I=
and integrate
f (z) =
dx
;
1 + xb
1
1
=
b
1+z
1 + jzjb eib arg z
1
;e
1 + zb
i=b
1
bz b
62
=
z=e
i=b
1
e
b
i=b
Integrate along
Z
1,
f (z) dz =
1
1
1
1+
Integrate along
Integrate along
z = xe0i
dz = dx
dz =
1
dx !
b
" 1+x
Z 1
1
!
dx = I:
1 + xb
0
f (z) dz
Rb
3,
2 R
1
b
2
Rb
! 0:
f (z) dz =
3
1
3
1+
Integrate along
dz =
z = xe2 i=b
dz = e2 i=b dx
1
e2
1 + xb
i=b
dx =
2 i=b
"
1
e2
1 + xb
1
dx =
1 + xb
i=b
dx !
e2
i=b
I:
4,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0
Multiplying with e
i=b
e2
i=b
I +0=2 i
1
e
b
, this yields
e
i=b
i=b
I=
2 i
;
b
i=b
I=
(e
i=b
2 i
e
i=b )
b sin ( =b)
i.e.,
Z
dx
;
=
1 + xb
b sin ( =b)
b > 1:
Remark.
We begin with the rst integral I, we make the substitution xb = t and use
the integral from Exercise VII.4.1 where we choose a = 1 1=b. (Because
b > 1, then 0 < 1 1=b < 1 and the restriction that 0 < a < 1 for the integral
in Exercise VII.4.1 is satised.)
I=
3
xb = t
dx
5=
dx = 4 x = t1=b
1 + xb
1 1=b 1
dx = b t
dt
Z
Z 1 1=b 1
1 1 t (1 1=b)
1t
dt =
dt =
=
b 1+t
b 0
1+t
0
1
=
=
:
b sin ( (1=b 1))
b sin ( =b)
64
VII.4.3
By integrating around the keyhole contour, show that
Z 1
2
cos ( a)
log x
dx
=
;
0 < a < 1:
a
x (x + 1)
sin2 ( a)
0
VII.4.3
2
z1
-R
Set
I=
and integrate
f (z) =
xa
xa
log x
;
(x + 1)
log x
log jzj + i arg z
= a ia arg z
(x + 1)
jzj e
(z + 1)
along the keyhole contour in Figure VII.4.3, where the cimicircle 4 have
radius ". We make a branch cut for log z along the positive real axis, so that
0 < arg z < 2 .
Residue at a simple pole at z1 = 1, where by Rule 3,
65
Res
Integrate along
Z
1,
= ie
ia
z= 1
f (z) dz =
1
Integrate along
Integrate along
3,
z = xe0i
dz = dx
R
log x
=
dx !
a
" x (x + 1)
Z 1
log x
dx = I:
!
a
x (x + 1)
0
f (z) dz =
3
Z "
log x + 2 i
log jzj + i arg z
z = xe2 i
dz =
=
!
a ia arg z
a
2 ia (x + 1)
dz = dx
jzj e
(z + 1)
R x e
Z 0
log x + 2 i
!
dx =
a
2 ia (x + 1)
1 x e
Z 1
Z 1
log x
dx
2 ia
2 ia
= e
dx 2 ie
=
a
a
x (x + 1)
x (x + 1)
0
0
= e 2 ia I 2 ie 2 ia J
and let " ! 0+ . This gives then 0 < a < 1
p
Z
log2 " + 2
f (z) dz
2 " 2 "1 a jlog "j ! 0:
a
" (1 ")
4
Integrate along
4,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
66
I +0
Multiplying with e
ia
2 ia
2 ie
2 ia
J +0=2 i
ie
ia
, this yields
ia
ia
ia
2 ie
2 2;
J=
and thus
2i sin ( a) I
2 ie
ia
2 2:
J=
2 sin ( a) I
I=
2
log x
cos ( a)
dx
=
;
a
x (x + 1)
sin2 ( a)
J=
xa
dx
=
;
(x + 1)
sin ( a)
i.e.
Z
2
cos ( a)
log x
dx
=
;
a
x (x + 1)
sin2 ( a)
0 < a < 1:
Remark.
From the Exercise VII.4.1 we have the formula
Z 1
x a
dx =
;
(1)
1+x
sin ( a)
0
0 < a < 1:
(2)
d
da
x a
dx =
1+x
1
1
log x x
1+x
67
dx =
1
1
xa
log x
dx;
(x + 1)
and dierentiate the right hand side in the formula in the same way
(3)
By (1) - (3), we have that
Z 1
0
d
=
da sin ( a)
cos ( a)
:
sin2 ( a)
2
log x
cos ( a)
dx
=
:
xa (x + 1)
sin2 ( a)
68
VII.4.4
For xed m 2, show that by integrating around the keyhole contour that
Z 1
a (a + 1)
(a + m 2)
x a
;
1 m < a < 1:
m dx =
(1 + x)
(m 1)! sin ( a)
0
Remark. The result can be obtained also by integrating the formula in Exercise 1 by parts.
Solution
VII.4.4
2
z1
-R
Set
I=
and integrate
x a
dx
(1 + x)m
z a
jzj a e ia arg z
f (z) =
=
(1 + z)m
(1 + z)m
along the keyhole contour in Figure VII.4.4. We make a branchcut for z
along the positive real axis, where 0 < arg z < 2 .
69
z!z0 (m
dm
1)! dz m
1
1
z0 )m f (z)] :
[(z
1
dm 1
z a
z a =
m ; 1 = lim
z! 1 (m
(1 + z)
1)! dz m 1
1
= lim
( a) ( a 1) ( a m + 2) z ( a (m 1)) =
z! 1 (m
1)!
1
= lim
( 1)m 1 a (a + 1)
(a + m 2) jzj( a (m 1)) ei( a (m 1)) arg z =
z! 1 (m
1)!
1
( 1)m 1 a (a + 1)
(a + m 2) j 1j( a (m 1)) ei( a (m 1)) =
=
(m 1)!
1
=
( 1)m 1 a (a + 1)
(a + m 2) ( 1)m 1 e ia =
(m 1)!
a (a + 1)
(a + m 2)
=
e ia :
(m 1)!
Res
Integrate along
Z
f (z) dz =
1
1,
jzj a e ia arg z
dz =
(1 + z)m
z = xe0i
dz = dx
!
Integrate along
m 2
Z
Integrate along
=
Z
f (z) dz
2
3,
R a
2 R
(R 1)m
2
Rm+a 1
x a
dx !
(1 + x)m
x a
dx = I:
(1 + x)m
! 0:
70
"
f (z) dz =
3
Z " a 2 ia
x e
jzj a e ia arg z
z = xe2 i
dz =
=
m
m dx !
dz = dx
(1 + z)
R (1 + x)
Z 0 a 2 ia
Z 1
x e
x a
2 ia
!
e
e 2
m dx =
m dx =
(1 + x)
1 (1 + x)
0
Integrate along
m 2
4,
"
(1
ia
I:
")m
2 "1
2 "
! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0
2 ia
Multiplying with e
e
ia
ia
I +0=2 i
a (a + 1)
(a + m
(m 1)!
2)
a (a + 1)
(a + m
(m 1)!
2)
ia
, this yields
e
ia
I=2 i
I=
(e
ia
2i
e
a (a + 1)
(a + m
ia )
(m 1)!
2)
a (a + 1)
(a + m
(m 1)! sin ( a)
i.e.
Z
x a
a (a + 1)
(a + m
m dx =
(1 + x)
(m 1)! sin ( a)
2)
m < a < 1:
Remark.
Set (this is our induction hypothesis)
Z 1
x a
a (a + 1)
(a + m
Im (a) =
m dx =
(1 + x)
(m 1)! sin ( a)
0
71
2)
2)
0 < b < 1:
sin ( b)
Z 1
1
x b
dx =
x b
dx =
1+x
1+x
0
Z 1 b+1
1
x
x b+1 1
1
=
dx =
1 b 1+x 0
1 b (1 + x)2
0
Z 1
1
x b+1
=
dx:
1 b 0 (1 + x)2
Set b = a + 1 and solve for the integral in the left side in the inequality above
Z 1
a
x a
=
;
a
2 dx =
sin ( (a + 1))
sin ( a)
(1 + x)
0
that is the same as I2 (a), thus our induction hypothesis is valid for m = 2.
Now suppose that our formula is valid for m = p, and integrate by parts,
b (b + 1)
(b + p 2)
=
(p 1)! sin ( b)
Z 1
Z 1
1
x b
x b
=
dx =
p dx =
(1 + x)
(1 + x)p
0
0
Z 1 b+1
1
1
x
p
x b+1
=
dx =
p
1 b (1 + x) 0
1 b (1 + x)p+1
0
Z 1
p
x b+1
=
dx:
1 b 0 (1 + x)p+1
Set b = a + 1 and solve for the integral in the left side in the inequality above
72
x a
dx =
(1 + x)p+1
((a + 1) + p 2)
a (a + 1) (a + 2)
=
=
p
(p 1)! sin ( (a + 1))
a (a + 1) (a + 2)
(a + (p + 1)
=
p! sin ( a)
2)
that is the same as Ip+1 (a), thus our induction hypothesis is valid for m =
p + 1.
We have showed that the formula Im (a) is valid for all m > 2, that was to
be shown.
73
VII.4.5
By integrating a branch of
show that
Z
(log z)2
(z+a)(z+b)
log x
(log a)2 (log b)2
dx =
;
(x + a) (x + b)
2 (a b)
a; b > 0; a 6= b:
Solution
VII.4.5 (a = 1, b = 2)
2
z2
z1
-R
Set
I=
and integrate
log x
dx
(x + a) (x + b)
(log z)2
(log jzj + i arg z)2
=
f (z) =
(z + a) (z + b)
(z + a) (z + b)
around the keyhole contour in Figure VII.4.5. We make a branchcut for log z
along the positive real axis, where 0 < arg z < 2 .
Residue at a simple pole at z1 = a, where by Rule 3,
74
"
#
(log jzj + i arg z)2
(log jzj + i arg z)2
Res
; a =
(z + a) (z + b)
(z + b)
Residue at a simple pole at z1 =
z= a
b, where by Rule 3,
"
#
(log jzj + i arg z)2
(log jzj + i arg z)2
Res
; b =
(z + a) (z + b)
(z + a)
Integrate along
Z
=
1,
z= b
(log b + i )2
:
=
a b
f (z) dz =
1
z = xe0i
dz = dx
!
Integrate along
Z
3,
=
Z
"
(log x)2
dx !
(x + a) (x + b)
(log x)2
dx = J:
(x + a) (x + b)
f (z) dz
2
Integrate along
Z
(log a + i )2
:
=
b a
log2 R + (2 )2
2 R
(R a) (R b)
4 log2 R
! 0:
R
f (z) dz =
3
Z "
(log jzj + i arg z)2
(log x + 2 i)2
z = xe2 i
=
dz =
=
dx !
dz = dx
(z + a) (z + b)
R (x + a) (x + b)
3
Z 0
Z 1
(log x + 2 i)2
(log x + 2 i)2
!
dx =
dx =
(x + a) (x + b)
1 (x + a) (x + b)
0
Z 1
Z 1
Z 1
(log x)2
log x
4 2
=
dx 4 i
+
dx =
(x + a) (x + b)
(x + a) (x + b)
(x + a) (x + b)
0
0
0
= J 4 iI+K
75
Integrate along
Z
4,
f (z) dz
4
log2 " + (2 )2
2 "
(a ") (b ")
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
J +0
J 4 iI+K + 0 = 2 i
log x
(log a)2 (log b)2
dx =
;
(x + a) (x + b)
2 (a b)
76
a; b > 0; a 6= b:
VII.4.6
Using residue theory, show that
Z 1 a
sin ( a) a 2 cos ( a)
x log x
dx
=
;
sin2 ( a)
(1 + x)2
0
1 < a < 1:
Solution
VII.4.6
2
z1
-R
Set
I=
xa log x
dx
(1 + x)2
and integrate
f (z) =
z a log z
jzja eia arg z (log jzj + i arg z)
=
(1 + z)2
(1 + z)2
77
Res
=
z= 1
=
z= 1
=e
Integrate along
Z
=
1,
ia
( 1
ai) :
f (z) dz =
1
z = xe0i
dz = dx
!
Integrate along
Integrate along
=
Z
xa log x
2 dx !
" (1 + x)
xa log x
dx = I:
(1 + x)2
3,
2 log R
! 0:
R1 a
78
Z
=
f (z) dz =
Z " a 2 ia
x e
(log jxj + 2 i)
jzja eia arg z (log jzj + i arg z)
z = xe2 i
=
dz =
dx !
2
dz
=
dx
(1 + z)
(1 + x)2
R
3
Z 0 a 2 ia
Z 1 a
x e
(log jxj + 2 i)
x (log jxj + 2 i)
2 ia
!
dx = e
dx =
2
(1 + x)
(1 + x)2
1
0
Z 1 a
Z 1
x log x
xa dx
2 ia
2 ia
= e
dx
2
ie
=
(1 + x)2
(1 + x)2
0
0
= e2 ia I 2 ie2 ia J
and let " ! 0+ . This gives
q
Z
a
" log2 " + (2 )2
f (z) dz
2 "
(1 ")2
4
Integrate along
4,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
ia
e2
I +0
multiplying with e
e
ia
2 ie2
ia
J +0=2 i e
ia
( 1
ai) ;
, this yields
ia
ia
2 ie
ia
J = 2 i( 1
ai) ;
and thus
2i sin ( a) I
2 ie
ia
J = 2 2a
2 i:
2 sin ( a) J = 2 2 a
2 cos ( a) J =
2 ;
I=
xa log x
=
(1 + x)2
sin ( a) a 2 cos ( a)
;
sin2 ( a)
79
J=
xa dx
a
;
2 =
sin ( a)
(1 + x)
i.e.
Z
xa log x
dx =
(1 + x)2
sin ( a) a 2 cos ( a)
;
sin2 ( a)
80
1 < a < 1:
VII.4.7
Show that
Z
xa 1
;
dx =
1 + xb
b sin ( a=b)
0 < a < b:
Determine for which complex values of the parameter a the integral exists
(in the sense that the integral of the absolute value is nite), and evaluate
it. Where does the integral depend analytically on the parameter a?
Solution
VII.4.7 (b = 12)
z1
Set
I=
and integrate
xa 1
;
1 + xb
za 1
jzja 1 ei(a 1) arg z
f (z) =
=
1 + zb
1 + jzjb eib arg z
81
za 1
Res
;e
1 + zb
Integrate along
Z
i=b
za
= b
bz
1
e
b
z=e
i=b
1,
jzja
1 i(a 1) arg z
dz =
z = xe0i
dz = dx
Integrate along
Z
Integrate along
xa 1
dx !
1 + xb
"
f (z) dz =
=
ia=b
xa 1
dx = I:
1 + xb
f (z) dz
2
3,
2
bRb
! 0:
f (z) dz =
3
jzja
3
1 i(a 1) arg z
1+
Z
!
xa 1 e2 i(a
1 + xb
Integrate along
4,
z = xe2 i=b
dz = e2 i=b dx
1)=b
2 i=b
dx =
2 ia=b
"
R
1
xa 1 e2 i(a
1 + xb
1)=b
xa 1
dx =
1 + xb
e2
e2
i=b
dx !
ia=b
I:
"a 1 2 "
1 "b
b
2 "a
! 0:
b
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0
e2
ia=b
ia=b
I +0=2 i
, this yields
82
1
e
b
ia=b
ia=b
ia=b
I=
2 i
b
b (e
2 i
e
ia=b
ia=b )
b sin ( a=b)
i.e.,
Z
xa 1
;
dx =
1 + xb
b sin ( a=b)
0 < a < b:
Now we suppose a and b complex and rewrite the nominator in the integrand
xa
= e(a
1) log x
= e(Re a
1) log x i Im a log x
= xRe a 1 ei Im a log x :
83
VII.4.8
By integrating a branch of (log z) = (z 3 + 1) around the boundary of
an indented sector of aperture 2 =3, show that
Z 1
Z 1
log x
2 2
1
2
p :
dx
=
;
dx
=
x3 + 1
27
x3 + 1
3 3
0
0
Remark. Compare the results with those of Exercise 3 (after changing variables) and Exercise 2.
Solution
VII.4.8
z1
z2
z3
Set
I=
and integrate
f (z) =
log x
dx;
x3 + 1
log z
=
z3 + 1
z
J=
x3
1
dx
+1
i=3
84
Res
Integrate along
Z
1,
i=3
z=e
i=3
i
e
9
2 i=3
f (z) dz =
1
Integrate along
z = xe0i
dz = dx
Integrate along
Z
log x
dx !
3
" x +1
Z 1
log x
!
dx = I:
x3 + 1
0
2,
3,
2 log R
! 0:
3R2
f (z) dz =
3
Z "
log x + 2 i=3 2 i=3
log jzj + i arg z
z = xe2 i=3
=
=
dz =
e
dx !
2 i=3
3
dz
=
e
dx
z +1
x3 + 1
R
3
Z 0
Z 1
log x + 2 i=3 2 i=3
log x + 2 i=3 2 i=3
e
dx
=
e
dx =
!
x3 + 1
x3 + 1
0
1
Z 1
Z
log x
2 i 2 i=3 1 1
2 i 2 i=3
2 i=3
= e
dx
e
dx = e2 i=3 I
e
J:
3
3
x +1
3
x +1
3
0
0
Z
Integrate along
4,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
85
I + 0 e2
Multiplying with e
i=3
i=3
2 i 2
e
3
i=3
i
e
9
J +0=2 i
2 i=3
, this yields
2
ie
3
2i sin ( =3) I
i=3
J=
2
9
and thus
p
i
2 2
J=
:
3
3
9
Separating real and imaginary parts, we get the simultaneous equations
3Ii +
p
and hence the solutions
Z 1
log x
dx =
I=
x3 + 1
0
3I
J = 29
J = 0;
3
2 2
;
27
J=
x3
1
2
dx = p :
+1
3 3
Remark.
We begin with the rst integral I, we make the substitution x3 = t and use
the integral from Exercise VII.4.3, where we choose a = 2=3.
I=
3
x3 = t
log x
5=
dx = 4 x = t1=3
x3 + 1
1
2=3
dx = 3 t
dt
Z 1
1
log t
1 2 cos 23
=
dt =
9 0 t2=3 (t + 1)
9 sin2 23
2
2 2
:
27
For the second integral J, we use the integral from Exercise VII.4.2, where
we choose b = 3.
Z 1
2
dx
=
J=
= p :
b
1+x
3 sin ( =3)
3 3
0
86
VII.4.9
By integrating around an appropriate contour and using the results
of Exercise 8, show that
Z
10 3
(log x)2
p :
dx
=
x3 + 1
81 3
Solution
VII.4.9
z1
z2
z3
Set
I=
Integrate
f (z) =
(log z)2
=
z3 + 1
z
(log x)2
dx
x3 + 1
i=3
(z
e i) z
e5
i=3
along the pie-slice contour with 0 < arg z < 2 =3 in Figure VII.4.9. We
make a branchcut for log z along the negative imaginary axis, where
=2 <
arg z < 3 =2.
Residue at a simple pole at z1 = e i=3 , where by Rule 3,
"
i=3
=
z=e
i=3
27
2 i=3
Integrate along
Z
1,
f (z) dz =
1
z = xe0i
dz = dx
"
!
Integrate along
Z
2,
Z
=
log2 R + 23
R3 1
3,
(log x)2
dx !
x3 + 1
1
(log x)2
dx = I:
x3 + 1
f (z) dz
Integrate along
2 log2 R
! 0:
3R2
2 R
3
f (z) dz =
3
Z "
(log x + 2 i=3)2 2 i=3
(log jzj + i arg z)2
z = xe2 i=3
dz =
e
dx !
=
dz = e2 i=3 dx
z3 + 1
x3 + 1
R
3
Z 0
Z 1
(log x + 2 i=3)2 2 i=3
(log x + 2 i=3)2 2 i=3
!
e
dx
=
e
dx =
x3 + 1
x3 + 1
1
0
Z 1
Z
Z
(log x)2
4 i 2 i=3 1 log x
4 2 2 i=3 1 1
2 i=3
e
dx
e
dx+
e
dx =
x3 + 1
3
x3 + 1
9
x3 + 1
0
0
0
Z
Z
4 i 2 i=3 1 log x
4 2 2 i=3 1 1
2 i=3
= e
I
e
dx+
e
:
3
x3 + 1
9
x3 + 1
0
0
Integrate along
Z
4,
f (z) dz
4
2 "
3
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
2 i=3
I+0 e
4 i 2
I
e
3
i=3
log x
4 2 2
dx+
e
x3 + 1
9
88
i=3
1
+0 = 2 i
3
x +1
27
2 i=3
multiplying with e
i=3
i=3
i=3
, this yields
4 i
I
3
log x
e
x3 + 1
i=3
dx+
4
9
1
e
3
x +1
i=3
dx =
and thus, substituting the values for the integrals from Exercise 9,
2i sin ( =3) I
4 i
e
3
i=3
2 2
27
4
9
i=3
3I+
2 3
4 3 12 3
+
=
;
81
81
27
and therefore
Z
10 2
(log x)2
p :
dx
=
x3 + 1
81 3
89
2
p
3 3
2 3
i:
27
2 3
i;
27
VII.4.10
By integrating a branch of (log z) = (z 3 1) around the boundary of
an indented half-disk and using the result of Exercise 8, show that
Z 1
log x
4 2
dx
=
:
x3 1
27
0
Solution
VII.4.10
4
z2
6
-R
2
1 z 1
3 R
z3
Set
I=
Integrate
f (z) =
log z
=
z3 1
(z
log x
dx
x3 1
i=3
i=3
= 0:
z=1
, where by Rule 3,
90
Res
Integrate along
Z
f (z) dz +
1
and
3,
i=3
=
z=e2
i=3
2 i
e
9
4 i=3
f (z) dz =
3
z = xe0i
=
dz = dx
Z R
log x
log x
dx +
dx !
3
3
x
1
1
1+" x
Z
!
=
=
1 "
"
Integrate along
2,
log x
dx = I:
x3 1
Integrate along
Integrate along
4,
5,
log R
! 0:
R2
91
f (z) dz =
5
xe i
e i dx
Z 1
"
log x + i
e id !
3
x
1
log x + i
dx =
x3 + 1
0
Z 1
1
dx =
3
x +1
0
Z 1
Z 1
log x
1
dx i
dx:
3
3
x +1
x +1
0
0
Integrate along
6,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0+0
log x
dx
x3 + 1
x3
1
dx + 0 = 2 i
+1
2 i
e
27
4 i=3
and thus, substituting the values for the integrals from Exercise 9,
2 2
27
2
p
I
i
=
3 3
And hence, setting imaginary parts equal,
2 2
4 2
I+
=
;
27
18
and therefore
Z
log x
4 2
dx
=
:
x3 1
27
92
4
9
4 i=3
VII.5.1
Use the keyhole contour indented on the lower edge of the axis at
x = 1 to show that
Z 1
log x
2 2
dx
=
;
0 < a < 1:
xa (x 1)
1 cos (2 a)
0
Solution
VII.5.1
2
-R
z
4
Set
I=
and integrate
f (z) =
za
1
1
xa
log x
dx
(x 1)
log z
(log jzj + i arg z)
= a(logjzj+i arg z)
(z 1)
e
(z 1)
along the contour in Figure VII.5.1. We make a branchcut for log z along
the positive imaginary axis, where 0 < arg z < 2 . Thus f (z) is analytic on
the keyhole domain. It extends analytically to (0; 1) from above, and the
93
Integrate along
Z
1,
= 2 ie
2 ia
z=1
f (z) dz =
1
z = xe0i
dz = dx
R
log x
=
dx !
a
1)
" x (x
Z 1
log x
!
dx = I:
xa (x 1)
0
Integrate along
2,
Integrate along
and
5,
f (z) dz+
f (z) dz =
5
Z
Z
(log jzj + i arg z)
(log jzj + i arg z)
=
dz+
dz =
a(logjzj+i
arg
z)
a(logjzj+i
arg z) (z
e
(z 1)
e
1)
3
5
Z 1+"
Z "
(log x + 2 i)
(log x + 2 i)
z = xe2 i
2 ia
2 ia
=e
dx+e
dx !
a
dz = dx
xa (x 1)
1)
R
1 " x (x
Z 0
Z 1
(log x + 2 i)
(log x + 2 i)
2 ia
2 ia
!e
dx = e
dx =
a
1)
xa (x 1)
1 x (x
0
Z 1
Z 1
log x
1
2 ia
2 ia
e
dx 2 ie
dx = e 2 ia I 2 ie 2 ia J:
a
a
x
(x
1)
x
(x
1)
0
0
Integrate along
4,
f (z) dz !
i(
0) 2 ie
2 ia
= 2 2e
2 ia
Integrate along
6,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I
2 ia
Multiplying with e2
ia
2 ie
2 ia
2 ia
J + 2 2e
+ 0 = 0:
, this yields
Ie2
ia
2 iJ + 2
= 0:
log x
dx =
a
x (x 1)
1
1) I + 2
= 0;
2 2
;
cos (2 a)
95
0 < a < 1:
VII.5.2
Show using residue theory that
Z 1
sin (ax)
dx =
2
1 x (x + 1)
a > 0:
Hint. Replace sin (az) by eiaz , and integrate around the boundary of a halfdisk indented at z = 0.
Solution
VII.5.2
4
-R
z1
2
z2
z3
Set
I=
and integrate
f (z) =
1
1
sin ax
dx
x (x2 + 1)
eiaz
eiaz
=
z (z 2 + 1)
z (z i) (z + i)
= 1:
z=0
Res
Integrate along
Z
f (z) dz +
1
eiaz
eiaz
;
i
=
z (z 2 + 1)
3z 2 + 1
and
3,
=
z=i
e
2
f (z) dz =
3
Z
eiaz
eiaz
z = x
dz
+
dz =
=
=
2
2
dz = dx
z (z + 1)
z (z + 1)
3
1
Z "
Z R
eiax
eiax
=
dx
+
dx !
2
2
R x (x + 1)
" x (x + 1)
Z 1
Z 1
Z 1
eiax
cos ax
sin ax
!
dx =
dx + i
dx =
2
2
2
1 x (x + 1)
1 x (x + 1)
1 x (x + 1)
Z 1
cos ax
=
dx + iI:
2
1 x (x + 1)
Integrate along
2,
e a;
I=
i.e.,
97
1
1
sin ax
dx =
x (x2 + 1)
98
a > 0:
VII.5.3
Show using residue theory that
Z 1
2
sin (ax)
dx = ;
2
2
2
ax)
1 x(
a > 0:
Solution
VII.5.3
8
2
-R
z3
4
z1
6
5
z2
7 R
Set
I=
and integrate
f (z) =
1
1
sin (ax)
dx
x ( 2 a2 x 2 )
eiaz
=
z ( 2 a2 z 2 )
a2 z (z
eiaz
=a) (z + =a)
eiaz
;0 =
z ( 2 a2 z 2 )
eiaz
3a2 z 2
=
z=0
1
2
eiaz
;
=
z ( 2 a2 z 2 ) a
99
eiaz
3a2 z 2
=
z= =a
1
2
Res
Integrate along
Z
1;
f (z) dz +
1
Z
=
3;
and
f (z) dz +
3
a
7,
f (z) dz +
5
eiaz
dz +
z ( 2 a2 z 2 )
e
z(
=
z=
=a
1
2
iaz
eiaz
3a2 z 2
a2 z 2 )
dz +
f (z) dz =
7
eiaz
dz+
z ( 2 a2 z 2 )
eiaz
dz =
z ( 2 a2 z 2 )
z = x
=
dz = dx
Z =a "
Z "
eiax
eiax
=
dx
+
dx+
2
x ( 2 a2 x 2 )
a2 x 2 )
R
=a+" x (
Z =a "
Z R
eiax
eiax
+
dx
+
dx !
2
x ( 2 a2 x 2 )
a2 x2 )
"
=a+" x (
Z 1
eiax
dx =
!
2
a2 x 2 )
1 x(
Z 1
Z 1
cos (ax)
sin (ax)
=
dx + i
dx =
2
2
2
2
ax)
a2 x 2 )
1 x(
1 x(
Z 1
cos (ax)
=
dx + iI:
2
a2 x 2 )
1 x(
=
Integrate along
2,
Integrate along
4,
100
Integrate along
6,
1
2
1
= 0;
2
i.e.,
Z
1
1
2
sin (ax)
dx = ;
2
2
2
x(
ax)
101
a > 0:
VII.5.4
Show using residue theory that
Z 1
1 cos x
dx = :
x2
2
0
Solution
VII.5.4
4
-R
z1
Set
I=
and integrate
cos x
dx ) 2I =
x2
f (z) =
1
1
cos x
dx
x2
eiz
z2
eiz
z2
and
d
1
z!0 dz
; 0 = lim
3,
eiz =
ieiz
z=0
102
f (z) dz +
1
f (z) dz =
3
"
R
Integrate along
eiz
z2
dz +
1
3
eiz
z2
dz =
z = x
=
=
dz = dx
Z R
Z 1
1 eix
1 eix
1 eix
dx
+
dx
!
dx =
x2
x2
x2
"
1
Z 1
Z 1
Z 1
1 cos x
sin x
sin x
dx i
dx = 2I i
dx:
2
2
2
x
1
1 x
1 x
2,
= 0;
cos x
dx = :
x2
2
i.e.,
Z
103
VII.5.5
By integrating (e 2iz 1) =z 2 over appropriate indented contours
and using Cauchys theorem, show that
Z 1
sin2 x
dx = :
2
1 x
Solution
VII.5.5
2
-R
4
z1
Set
I=
and integrate
1
1
sin2 x
dx =
x2
1
1
cos 2x
dx
2x2
ei2z 1
f (z) =
z2
along the contour in Figure VII.5.5.
Residue at a double pole at z1 = 0, where by Rule 2,
Res
Integrate along
ei2z 1
d i2z
;
0
=
lim
e
z!0 dz
z2
1
and
3,
1 = 2iei2z
z=0
= 2i
104
f (z) dz +
1
f (z) dz =
3
ei2z 1
dz +
z2
ei2z 1
dz =
z2
z = x
=
dz
= dx
1
3
Z " i2x
Z R i2x
Z 1 i2x
e
1
e
1
e
1
=
dx
+
dx
!
dx =
x2
x2
x2
R
"
1
Z 1
Z 1
Z 1
sin 2x
sin 2x
cos 2x 1
dx
+
i
dx
=
2I
+
i
dx:
=
x2
x2
x2
1
1
1
=
Integrate along
2,
2I + 2 = 0;
i.e.,
Z
1
1
sin2 x
dx = :
x2
105
VII.5.6
By integrating a branch of (log z) = (z 3 1) around the boundary of
an indented sector of aperture 2 =3, show that
Z 1
log x
4 2
dx
=
:
x3 1
27
0
VII.5.6
5
z2
6
8
2
1
z1
3 R
z3
Set
I=
and integrate
f (z) =
log z
=
z3 1
(z
log x
dx
x3 1
e0i )
e4
i=3
along the contour in Figure VII.5.6. We make a branchcut for log z along
the negative imaginary axis, where
=2 < arg z < 3 =2.
Residue at a simple pole at z1 = 1, where by Rule 3,
Res
i=3
= 0:
z=1
, where by Rule 3,
106
Res
Integrate along
Z
f (z) dz+
1
and
3,
i=3
=
z=e2
+
i=3
2 i 2
e
9
i=3
f (z) dz =
3
Z
log jzj + i arg z
log jzj + i arg z
z = xe0i
dz+
dz
=
=
3
dz
=
dx
z3 1
z
1
3
Z 1 "
Z R
Z 1
log x
log x
log x
=
dx+
dx !
dx = I:
3
3
x
1
1
x3 1
"
1+" x
0
Integrate along
2,
Integrate along
Integrate along
Z
f (z) dz +
5
4,
and
7,
2 log R
! 0:
R2
f (z) dz =
7
Z
log jzj + i arg z
log jzj + i arg z
z = xe2 i=3
=
=
dz+
dz
=
dz = e2 i=3 dx
z3 1
z3 1
5
7
Z 1+"
Z "
log x + 2 i=3 2 i=3
log x + 2 i=3 2 i=3
=
e
dx
+
e
dx !
x3 1
x3 1
R
1 "
Z 0
Z 1
Z
log x + 2 i=3
log x
2 i 2 i=3 1 1
2 i=3
2 i=3
!e
dx = e
dx
e
dx =
x3 1
x3 1
3
x3 1
1
0
0
Z
2 i 2 i=3 1 1
2 i=3
= e
I
e
dx:
3
x3 1
0
Integrate along
6,
f (z) dz !
i(
0)
2 i 2
e
9
i=3
Integrate along
2
9
e2
i=3
8,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
Z
2 i 2 i=3 1 1
2 2 2 i=3
2 i=3
dx
+
I +0 e
I
e
e
= 0:
3
x3 1
9
0
Multiplying with e
I
2 i=3
1
2
, this yields
p !
Z
3
2 i 1 1
2 2
i
I
dx
+
= 0:
2
3 0 x3 1
9
log x
4 2
dx
=
:
x3 1
27
108
VII.6.1
Integrate 1= (1
Show that
Z
dx
= +1;
1 x2
dx
=
1 x2
1:
Solution
Partial fractions gives
1
x2
PV
= lim
"!0
dx
= lim
1 x2 "!0
1
[log (1 + x)
2
= lim
"!0
1 "
log (1
1
2
dx
+
1 x2
x)]10
1+x
1
log
2
1 x
1
"!0 2
= lim
log
1
1
+
1+x 1 x
1 "
0
"
1+"
dx
1 x2
1
[log j1 + xj
2
1
1+x
+
log
2
1 x
"
"
log
2+"
"
=
log j1
!
xj]1
1+"
1+"
1
= lim log
"!0 2
2 "
2+"
= 0:
Inspection gives
Z
dx
= lim
1 x2 "!0+
1 "
dx
1
= lim+ [log (1 + x) log (1 x)]10 " =
2
"!0 2
1 x
0
1 "
1
1+x
1
2 "
= lim+ log
= lim+ log
= +1 :
"!0 2
"!0 2
1 x 0
"
109
dx
= lim
1 x2 "!0+
= lim+
"!0
1
dx
= lim+ [log (1 + x) log (1 x)]1
1+" =
2
"!0 2
x
1+" 1
1
1+x
1
2+"
1
log
log
! 1:
= lim+
"!0 2
2
1 x 1+"
"
110
VII.6.2
Obtain the principal value in Exercise 1 by taking imaginary parts
of the identity (5.4) p.211 in the preceding section and making a
change of variable
Solution
Identity (5.4)
Z
Z
log x
dx+
x2 1
1 "
1
Z
log jxj + i
dx+
x2 1
0
1+"
Z
log jxj + i
log z
dx+
dz = 0:
2
x2 1
1
C" z
dx + Im
C"
Note
( )
Im
C"
1 "
1
x2
dx +
0
1+"
x2
Z
dx =
1+"
x2
log z
dz = 0:
z2 1
log z
dz ! 0;
z2 1
x =
dx =
Z 1
dx +
0
x
dx
"
x2
dx ! 0 as " ! 0;
dx
= 0:
1 x2
Note. For a linear change of variables, we can take the limit either before or
after change of variable. Otherwise we can not do this.
111
VII.6.3
By integrating around the boundary of an indented half-disk in the
upper half-plane, show that
Z 1
1
a
dx =
;
1 < a < 1:
PV
2
2
a)
a +1
1 (x + 1) (x
Solution
VII.6.3 (a = 2)
4
z2
-R
2
z1
z3
Set
I = PV
and integrate
f (z) =
1
1
1
(z 2 + 1) (z
(x2
a)
1
+ 1) (x
(z
a)
dx
1
i) (z + i) (z
a)
(z 2
1
+ 1) (z
a)
;a =
z2
1
+1
=
z=a
112
a2
1
:
+1
Res
1
(z 2 + 1) (z
a)
Integrate along
Z
f (z) dz +
1
and
Za
;i =
3,
3z
1
2az + 1
z=i
f (z) dz =
3
"
1
2
(x + 1) (x
a)
dx +
ZR
Integrate along
Z
2,
(x2
a)
dx !
1
+ 1) (x
a)
dx = I:
f (z) dz !
4,
1
+ 1) (x
Z 1
(x2
a+"
Integrate along
Z
1
a
+
i:
2 (a2 + 1) 2 (a2 + 1)
i(
0)
a2
1
+1
a2 + 1
i:
f (z) dz
(R2
1
1) (R
jaj)
R2
! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I
a2 + 1
i=2 i
1
a
+
i ;
2 (a2 + 1) 2 (a2 + 1)
i.e.,
PV
1
1
(x2
1
+ 1) (x
113
a)
dx =
a2
a
:
+1
VII.6.4
Suppose m 2 and a1 < a2 <
< am . By integrating around the
boundary of an indented half-disk in the upper half-plane, show
that
Z 1
1
dx = 0:
PV
a1 ) (x a2 ) (x am )
1 (x
Solution
VII.6.4
R
-R
m - 1 m
a1
a2
a m - 1 am
Set
I = PV
and integrate
f (z) =
(z
1
1
(x
a1 ) (z
a1 ) (x
1
a2 )
(z
1
a2 )
am )
(x
am )
dx
1
k=1 (z
= Qm
ak )
Res Qm
ak )
1
k=1;k6=j (z
; aj = Qm
114
ak )
z=aj
1
k=1;k6=j (aj
= Qm
ak )
f (z) dz =
L
a1 "
R
a1 "
R
a1 "
1
k=1 (z
Qm
Integrate along
j=1
Integrate along
Z
1;
ak )
dz+
k=1
Z
m
1
X ak+1 "
ak +"
k=1
f (z) dz+
m,
2; : : :
1
k=1 (z
Qm
f (z) dz =
ak )
dz+
ak )
1
1
am +"
dx+
am +"
1
k=1 (x
Qm
i(
0)
j=1
1
Qm
k=1;k6=j (aj
ak )
ak )
1
k=1 (x
ak )
Qm
ak )
!
1
k=1 (z
Qm
and
am +"
z = x
=
dz = dx
m
X1 Z ak+1 "
1
Qm
dx+
ak )
k=1 (x
k=1 ak +"
Z
!
f (z) dz !
R,
ak +"
1
Qm
k=1 (x
m Z
X
f (z) dz+
m
X1 Z ak+1 "
L,
dx = I:
Qm
f (z) dz
R
jak j)
Rm
! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I
m
X
j=1
1
k=1;k6=j (aj
Qm
ak )
= 0:
Now, note that the the sums of the residues at a simple pole at zj = aj ;
1 j m, are real, and identifying the real part, we have
Z 1
1
PV
dx = 0:
a1 ) (x a2 ) (x am )
1 (x
115
dz =
dx !
VII.6.5
Show that
PV
1
1
xa 1
dx =
xb 1
a
;
b
cot
0 < a < b:
1 around a sector
Solution
VII.6.5 (b = 3)
5
z2
6
8
2
1
z1
3 R
z3
Set
I = PV
and integrate
1
1
xa 1
dx
xb 1
za 1
jzja 1 ei(a 1) arg z
f (z) = b
=
z
1
jzjb eib arg z 1
za 1
jzja 1 ei(a 1) arg z
;
1
=
zb 1
bz b 1
116
z=1
1
= :
b
along the
za 1 2
;e
zb 1
Integrate along
Z
f (z) dz+
1
jzja
1
and
i=b
=
3,
dz +
1 "
"
2,
Z
Integrate along
4,
Z
Integrate along
f (z) dz+
5
jzja
5
e
bz b
e2
z=e2 i=b
jzja
1 i(a 1) arg z
dz =
z = xe0i
dz = dx
=
jzjb eib arg z 1
Z R a 1
Z 1 a 1
xa 1
x
x
dx +
dx !
dx = I:
b
b
x
1
1
xb 1
1+" x
0
3
f (z) dz !
i(
0)
1
b
i
:
b
and
7,
Ra 1 2 R
Rb 1
b
2
bRb
! 0:
f (z) dz =
7
1 i(a 1) arg z
1 i(a 1) arg z
f (z) dz =
1 i(a 1) arg z
Integrate along
, where by Rule 3,
jzja
i=b
dz +
jzja
1 i(a 1) arg z
z = xe2 i=b
dz = e2 i=b dx
=
b ib arg z
jzjb eib arg z 1
1
7 jzj e
Z 1+" a 1 2 i(a 1)=b
Z " a 1 2 i(a 1)=b
x e
x e
2 i=b
=
e
dx +
e2 i=b dx!
b
b
x
1
x
1
R
1 "
Z 0 a 1 2 i(a 1)=b
Z 1 a 1
x e
x
2 i=b
2 i=b 2 i(a 1)=b
!
e
dx = e
e
dx =
b
x
1
xb 1
1
0
= e2 i=b e2 i(a 1)=b I
117
dz =
Integrate along
Z
6,
f (z) dz !
Integrate along
8,
0)
e2
e2
2 "a
! 0:
b
"a 1 2 "
1 "b
b
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
i
+ 0 e2
b
Hence solving for I,
I
i 1 + e2
I=
b 1 e2
e
i=b e2
i(a 1)=b
i 1 + e2
=
b 1 e2
e2
= 2 i 0:
ia=b
ia=b
cos
b sin
a
b
a
b
i.e.,
PV
1
1
xa 1
dx =
xb 1
cot
118
a
;
b
0 < a < b:
cot
a
;
b
VII.6.6
By integrating a branch of (log z) = z b 1 around an indented sector of aperture 2 =b, show that for b > 1,
Z
2
log x
dx
=
;
xb 1
b2 sin2 ( =b)
PV
1
xb
dx =
cot ( =b) :
Solution
VII.6.6 (b = 3)
5
z2
6
8
2
1
z1
3 R
z3
Set
I=
and integrate
log x
dx
xb 1
f (z) =
J = PV
1
xb
dx
log z
log jzj + i arg z
=
b
z
1
jzjb eib arg z 1
along the contour in Figure VII.6.5. We make a branchcut for log z along
negative imaginary axis, where
=2 < arg z < 3 =2.
Residue at a simple pole at z1 = 1, where by Rule 3,
Res
log z
log jzj + i arg z
;1 =
b
z
1
bz b 1
i=b
= 0:
z=1
, where by Rule 3,
119
Res
log z 2
;e
zb 1
Integrate along
Z
f (z) dz+
1
and
3,
=
z=e2
i=b
2 ie2
b2
i=b
f (z) dz =
3
z = xe0i
=
b ib arg z
dz = dx
jzjb eib arg z 1
1
3 jzj e
Z R
Z 1
Z 1 "
logx
logx
logx
dx +
dx !
dx = I:
=
b
b
x
1
1
xb 1
1+" x
0
"
i=b
Integrate along
2,
dz +
dz =
Integrate along
4,
Integrate along
Z
=
f (z) dz+
5
and
7,
f (z) dz =
7
dz +
dz =
z = xe2 i=b
dz = e2 i=b dx
b ib arg z
b ib arg z
1
1
5 jzj e
7 jzj e
Z 1+"
Z "
logx + 2 i=b 2 i=b
logx + 2 i=b 2
=
e
dx +
e
b
x
1
xb 1
R
1 "
Z 1
Z 0
logx + 2 i=b 2 i=b
logx
2 i 2
2 i=b
!
e
dx = e
dx
e
b
b
x
1
x
1
b
1
0
120
e2
i=b
i=b
i=b
dx !
Z 1
0
xb
2 i 2
e
b
1
i=b
dx =
J:
Integrate along
Z
6,
f (z) dz !
i(
i=b
2 ie2
b2
0)
2 2 e2
b2
i=b
Integrate along
8,
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
Multiplying with e
2 i 2
e
b
i=b
, this yields
e2
I +0
2
i=b
i=b
J+
2 2 e2
b2
i=b
= 2 i 0:
2 2
2 i
J + 2 = 0:
b
b
Separating real and imaginary parts, we get simultaneous equations
e
2 i=b
cos
1 I
2
b
1 I + 2b2 = 0
sin 2b I 2b J = 0;
i.e.,
I=
2
log x
dx
=
;
xb 1
b2 sin2 ( =b)
J = PV
121
1
xb
dx =
cot ( =b) :
VII.6.7
Suppose that P (z) and Q (z) are polynomials, deg Q (z) deg P (z)+2,
and the zeros of Q (z) on the real axis are all simple. Show that
PV
1
1
X
X
P (z)
P (z)
P (z)
dx = 2 i
Res
; zj + i
Res
; xk ;
Q (z)
Q (z)
Q (z)
summed over the poles zj of P (z) =Q (z) in the open upper halfplane and the poles xk of P (z) =Q (z) on the real axis. Remark.
In other words, the principal value of the integral is 2 i times the
sum of the residues in the upper half-plane, where we count the
poles on the real axis as being half in and half out of the upper
half-plane.
Solution
VII.6.7
R
zj - 1
z3
zj
-R
a1
a2
z2
z1
m - 1 m
a m - 1 am
Use the Residue Theorem, take limit, and the result follows
Set
I = PV
and integrate
1
1
f (z) =
P (x)
dx
Q (x)
P (z)
Q (z)
m on the real
P (z)
; xk
Q (z)
Res
k=1
P (z)
; zj
Q (z)
Res
k=1
Integrate along segments on real axis, and let R ! 1 and " ! 0+ . This
gives
Z
f (z) dz =
=
a1 "
R
a1 "
f (z) dz+
R
m
X1 Z ak+1 "
ak +"
k=1
m
X1 Z ak+1 "
P (z)
dz+
Q (z)
k=1
ak +"
a1 "
R
P (z)
dz+
Q (z)
f (z) dz =
am +"
am +"
P (z)
dz =
Q (z)
z = x
=
dz = dx
Z R
m
X1 Z ak+1 " P (x)
P (x)
P (x)
dx+
dx+
dx !
Q (x)
Q (x)
am +" Q (x)
k=1 ak +"
Z 1
P (x)
!
dx = I:
1 Q (x)
=
f (z) dz+
123
Integrate along
This gives
R,
Integrate along
m Z
X
k=1
1;
Rdeg P
Rdeg Q
f (z) dz
R
2; : : :
m,
f (z) dz !
! 0:
0)
m
X
k=1
Res
P (z)
; xk
Q (z)
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
!
j
m
X
X
P (z)
P (z)
; xk
; zk
I i
Res
=2 i
Res
;
Q (z)
Q (z)
k=1
k=1
i.e.,
PV
1
1
m
m
X
X
P (z)
P (z)
P (x)
dx = 2 i
Res
; zj + i Res
; xk :
Q (x)
Q
(z)
Q
(z)
j=1
k=1
124
VII.7.1
1 2 3 P
L
K
LLL
Show that
Z
jsin xj
dx = +1:
x
Hint: Show that the area under the mth arc of jsin xj =x is
Solution
VII.7.1
1) +
jsin xj
x
sin
, we have
1
=p ;
4
2
and
1
x
1
:
m
1 1
p
:
2m
We have
Z
=4
(m 1) + 4
jsin xj
dx
x
Hence
125
1 1
p
2m
C
:
m
1=m.
jsin xj
dx
x
m Z
X
k=1
=4
(k 1) + 4
1
1
jsin xj
dx = C 1 + + : : : +
x
2
m
! +1
as m ! 1.
We have used the fact that the harmonic series 1 + 12 + 13 + 14 +
+ m1 tends
R m+1
dx
to innity as as m ! 1, by taking the oversum for the integral 1
,
x
Z m+1
1
dx
1 1
+
= log (m + 1) ! +1
1+ + +
2 3
m
x
1
as m ! 1.
126
VII.7.2
Show that
lim
R!1
R
R
x3 sin x
dx = :
2e
(x2 + 1)2
Solution
VII.7.2
2
z1
-R
z2
Set
I=
1
1
x3 sin x
dx
(x2 + 1)2
and integrate
f (z) =
z 3 eiz
=
(z 2 + 1)2
(z
z 3 eiz
i)2 (z + i)2
z 3 eiz
d z 3 eiz
1
;
i
=
lim
= :
2
2
z!i dz (z + i)
4e
(z 2 + 1)
127
f (z) dz =
1
=
!
1
1
Z R
x3 eix
z 3 eiz
z = x
dz
=
=
2 dx !
2
dz = dx
(z 2 + 1)2
R (x + 1)
Z 1 3
Z 1 3
x3 eix
x cos x
x sin x
dx =
dx + i
2
2
2 dx =
2
2
(x2 + 1)
1 (x + 1)
1 (x + 1)
Z 1 3
x cos x
=
2 dx + iI:
2
1 (x + 1)
128
VII.7.3
Evaluate the limits
lim
R!1
R
R
x sin (ax)
dx;
x2 + 1
1 < a < 1:
VII.7.3a
z1
-R
R
z2
Case 1 a < 0:
Set
I=
and integrate
f (z) =
1
1
x sin (ax)
dx
x2 + 1
zeiaz
=
z2 + 1
(z
zeiaz
i) (z + i)
1,
zeiaz
zeiaz
;
i
=
z2 + 1
2z
=
z= i
ea
:
2
Z
zeiaz
z = xe0i
f (z) dz =
=
dz =
dz = dx
z2 + 1
1
1
Z 1
Z
Z 1
x cos (ax)
xeiax
dx =
dx + i
!
2
x2 + 1
1
1 x +1
R
R
1
xeiax
dx !
x2 + 1
x sin (ax)
dx =
2
1 x +1
Z 1
x cos (ax)
=
dx + iI:
x2 + 1
1
ea :
Case 2 a = 0:
lim
R!1
Case 3 a > 0:
R
R
x sin (ax)
dx = lim
R!1
x2 + 1
130
R
R
x2
0
=0
+1
VII.7.3b
4
z1
-R
z2
Set
J=
and integrate
f (z) =
1
1
x sin (ax)
dx
x2 + 1
zeiaz
=
z2 + 1
(z
zeiaz
i) (z + i)
3,
zeiaz
zeiaz
;
i
=
z2 + 1
2z
=
z=i
4,
Z
zeiaz
z = xe0i
dz =
=
f (z) dz =
dz = dx
z2 + 1
3
3
Z 1
Z 1
Z
xeiax
x cos (ax)
!
dx =
dx + i
2
x2 + 1
1 x +1
1
Integrate along
R
R
1
xeiax
dx !
x2 + 1
x sin (ax)
dx =
2
1 x +1
Z 1
x cos (ax)
=
dx + iJ:
x2 + 1
1
f (z) dz
4
R
R2
R
R2
! 0:
1
1
8
<
x sin (ax)
0
dx =
:
x2 + 1
ea ;
e a;
132
a < 0;
a=0
a > 0:
VII.7.4
By integrating z a 1 eiz around the boundary of a domain in the rst
quadrat bounded by the real and imaginary axes and a quatercircle, show that
lim
R!1
lim
R!1
where
xa
cos xdx =
0 < a < 1;
xa
sin xdx =
0 < a < 1;
Remark: The formula for the sine integral holds also for
this, integrate by parts.
Solution
VII.7.4
2
1
Set
I = lim
R!1
a 1
cos xdx
J = lim
R!1
xa
sin xdx
1 i(a 1) arg z iz
f (z) dz =
1
z = xe0i
jzj e
e dz =
=
dz = dx
1
Z 1
Z 1
Z
a 1 ix
a 1
!
x e dx =
x cos xdx + i
a 1 i(a 1) arg z iz
xa
xa 1 eix dx !
sin xdx = I + iJ:
3,
f (z) dz =
3
Z R
z = xe i=2
=
jzj e
e dz =
=
xa 1 e ia=2 e x dx !
i=2
dz
=
e
dx
0
3
Z 1
Z 1
!
xa 1 e ia=2 e x dx = e ia=2
xa 1 e x dx = e ia=2 (a) =
a 1 i(a 1) arg z iz
J=
(z + 1) = z (z)
Z
a 1
cos xdx = x
a 1
sin x
(a
1)
R
0
sin xdx = x
a 1
sin x
(a
1)
sin xdx = R
a 1
a 2
xa
sin xdx =
lim
R!1
xa
cos xdx
sin R
xa
xa
cos xdx
xa
cos xdx
Ra 1 sin R
sin xdx =
a 1
sin xdx
R
0
R
a 2
R
a 2
1) xa
(a
sin R
(1 a) R1
1
a
1
a
(a + 1)
cos
(a + 1) =
a
2
(a) [cos ( a=2) cos ( =2) sin ( a=2) sin ( =2)] =
= (a) sin ( a=2) :
sin xdx =
135
VII.7.5
Show that
lim
R!1
sin x
dx = lim
R!1
cos x
p
dx = p ;
2 2
VII.7.5
Set
I = lim
R!1
where
sin x
dx
J = lim
R!1
136
cos x2 dx
Z R
2
z = xe0i
eix dx !
=
f (z) dz =
e dz =
dz
=
dx
0
1
Z 1
Z 1
Z 11
2
sin x2 dx = J + iI:
cos x2 dx + i
eix dx =
!
Z
iz 2
f (z) dz =
2
=4
eiz dz =
2
eiR
2 e2it
iReit dt
=R
z = Reit
=
dz = iReit dt
Z =4
2
R
eiR (cos(2t)+i sin(2t)) dt =
0
=4
R2 sin 2t
dt
=4
4R2 t=
ds =
4R
4R
1
f (z) dz =
3
=4,
iz 2
e dz =
3
Z 1
2
!
e xe
0
z = xe i=4
dz = e i=4 dx
i=4
R2
i=4
4R2 t=
=4
=
0
4R
where 0
x2
i=4
! 0:
x
R,
dx !
Z
1 + i 1 x2
p
e dx =
dx =
2
0
Z 1
Z 1
1
1
x2
= p
e dx i p
e
2 0
2 0
x2
dx:
We will use integral (41) from Mathematics Handbook page 177 with a = 1,
r
Z 1
1
ax2
e
dx =
:
2 a
0
and hence
137
f (z) dz =
3
1
p
2
x2
dx
1
ip
2
x2
dx =
p
p
2 2
138
p
i p :
2 2
VII.8.1
1 2 3 P L K
Evaluate the residue at 1 of the following functions.
3
(c) zz2 +11
(e) q
z n e1=z , n = 0; 1; : : :
(a) z2z 1
9
1
(f) zz ab
(b) (z2 +1)
(d) zz6 +11
2
Note. There are two possibilities for (f), one for each branch of the square
root.
a)
z
z2
1 1
1
1 =
z 1 z2
z
1+
1
1
+ 4 + ::: ;
2
z
z
hence
Res
z
z2
;1 =
1:
b)
1
=O
(z 2 + 1)2
1
z4
as z ! 1;
hence
Res
(z 2
1
; 1 = 0:
+ 1)2
c)
z3 + 1
z (z 2 1) + z + 1
z
1
=
=z+ 2
+ 2
=
2
2
z
1
z
1
z
1 z
1
1
1
1
1
1
1 1
=z+
1 + 2 + 4 + ::: + 2
;
=z+
1 + 2
z
1
z
z
z
z
1
z 1 z2
hence
Res
z3 + 1
;1 =
z2 1
139
1:
d)
z9 + 1
z 3 (z 6 1) + z 3 + 1
z3 + 1
3
=
=
z
+
;
z6 1
z6 1
z6 1
hence
z9 + 1
; 1 = 0:
z6 1
Res
e)
z n e1=z = z n 1 +
1
1
1
1
1
+
+
+
+
+ ::: =
2
3
3
z 2!z
3!z
3!z
4!z 4
1
1
1
= zn + zn 1 + zn 2 + zn 3 + zn
2!
3!
4!
1
z
+:
is
1
;
(n + 1)!
hence
Res z n e1=z ; 1 =
1
;
(n + 1)!
n = 0; 1; : : : :
f)
r
z
z
analytic at w = 0:
coe cient of w is g 0 (0) =
a
=
b
1
2
1 aw
1 bw
1
1
a=z
=
b=z
1=2
b a
:
2
140
1
1
aw
;
bw
=
w=0
1 p1
2 1
(b
a) =
7.8.2
1 2 3 P L K
Show by integrating around the dogbone contour that
Z 1
x4
35
p
dx =
:
128
x (1 x)
0
Solution
VII.8.2
1
3
Set
I=
and integrate
z4
f (z) = p
z (1
z)
=p
x4
x (1
jzjei arg
141
x)
z=2
dx
z4
j1
zjei arg0 (1
z)=2
VIII.8.2
arg z
arg0 z
-2
-1
-2
arg z
arg0 (1
-1
-2
-1
-2
jzjei arg
z=2
pz
-1
j1 zjei arg0 (1
z)=2
y
3/2
/2
3/2
-2
/2
-1
3/2
/2
/2
/2
/2
/2
=
=
128 (w
Res
1
f
w2
1
w
;0 =
9=2
1)
Res 4
1
w4
1
q
w2 1 1
w
1
w
; 05 =
1
1 d4
1
1
p
p
=
lim
=
w!0 4! dz 4
w5 w 1
w 1
35
35
=
=
e i=2 =
9=2 i9 arg(w 1)=2
128
128 (jw 1j) e
Res
35
Res [f (z) ; 1] =
z)
142
35
i
128
Integrate along
Z
=
1,
f (z) dz =
1 "
"
jzjei arg
Integrate along
Integrate along
z=2
2,
Z
Z
z4
dz =
1 "
(1 + ")4
p
2 "
(1 + ") "
x4
p
dx =
jxjei 0=2 j1 xjei 0=2
x4
p
dx = I:
x (1 x)
j1
f (z) dz
3,
2 "9=2 ! 0:
f (z) dz =
"
1 "
p
jzjei arg
Integrate along
z=2
4,
z4
j1
"
1
p
dx =
jxjei 0=2 j1 xjei2 =2
1 "
Z 1
x4
p
dx !
dx = I:
x)
x (1 x)
0
dz =
f (z) dz
4
")
2 "
2 "9=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0+I +0=2 i
35
i ;
128
i.e.,
Z
x4
x (1
x)
143
dx =
35
:
128
7.8.3
1 2 3 P L K
Fix an integer n, positive or negative. Determine for which complex
values of the parameter a the integral
Z 1
xn
dx
a
x)1 a
0 x (1
converges, and evaluate it.
Solution. ( Obs fel i lsningen argumentet)
x 2 R+
1 x 2 R+
a = Re a + i Im a
xa = xRe a+i Im a = xRe a xi Im a
The integral is generalized in 0 and 1 " 2 R, x = 0
g (x) = xa1 n
f (x)
1
= (1 x)
1 a ! 1, as x ! 0
Rg(x)
R"
"
g
(x)
dx
<
1
,
a
n
<
1
,
f (x) dx < 1 , a < n + 1.
0
0
The integral is generalized in 0 and 1 " 2 R, x = 0
1
g (x) = (1 x)
1 a
n
f (x)
= xxa ! 1, as x !
g(x)
R1
g (x) dx < 1 , 1
"
1
a<1,
R"
144
VII.8.3
Set
I=
xn
xa (1
x)1
a dx
and integrate
f (z) =
zn
z a (1
z)1
145
zn
zj1 a e(1
a)i arg0 (1 z)
VIII.8.3
arg z
arg0 z
-2
-1
y
0
-2
arg z
arg0 (1
-1
-2
-1
-2
-1
zn
jzja eai arg z j1
1 a (1 a)i arg (1 z)
0
zj
e
y
2
-2
-1
2 2
1
f
Res
w2
w2 1
Res
;0 =
w(1 + w2 )
Integrate along
1,
Res [f (z) ; 1] =
z)
1
w
lim
;0 =
w!0
jw2
"
1
Res
w2
1jei arg(w
w2 + 1
1
wn
1
wa
1)=2
;0
1 1 a
w
ei
=2
=
i
Z
=
f (z) dz =
"
Integrate along
Z
2,
f (z) dz
4
Integrate along
Z
Z 1 "
zn
zn
=
dx =
zj1 a e(1 a)i arg0 (1 z)
jzja eai 0 j1 zj1 a e(1 a)i 0
"
Z 1 "
zn
dx !
=
z a (1 z)1 a
"
Z 1
zn
dx = I:
!
a
z)1 a
0 z (1
1 "
3,
2
"
2 "
! 0:
f (z) dz =
"
1 "
Integrate along
Z
Z "
zn
zn
=
dx =
a ai 0
z j1
zj1 a e(1 a)i arg0 (1 z)
j1 zj1 a e(1 a)i 2
1 " jzj e
Z 1 "
zn
2 ia
dx !
= e
z a (1 z)1 a
"
Z 1
zn
2 ia
! e
dx = e2 ia I:
a
z)1 a
0 z (1
4,
f (z) dz
4
"n
"a
(1
1 a
")
2 "
2 "n+1
! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I + 0 e2
ia
I +0=2 i
( 1)n
147
(a
1)
(a
n!
n)
1) (a n)
n!
ia
( 1)n 2 i (a 1)n!(a
=
e ia e ia
n)
= ( 1)n
(a
1)
sin ( a)
i.e.,
Z
xn
xa (1
1 a dx
x)
= ( 1)n
(a
sin ( a)
148
1)
(a
n!
n)
0 < a < 1;
n = 0; 1; 2; : : :
(a
n!
n)
VII.8.4
Show that
Z
1
1
1 x2
dx =
1 + x2
1 :
Solution
VII.8.4
1
3
Set
I=
and integrate
1
1
1 x2
dx
1 + x2
p
j1 + zjei arg
1 z2
f (z) = 2
=
z +1
along the contour in Figure VII.8.4.
p
149
p
j1
2
z +1
(1+z)=2
zjei arg0 (1
z)=2
VIII.8.4
arg z
arg0 z
-2
-1
arg (1 + z)
-2
arg0 (1
-1
-2
-1
z)
y
0
0
jzjei arg
z=2
-2
pz
-1
2
0
j1 zjei arg0 (1
z)=2
y
3/2
/2
3/2
-2
/2
-1
/2
/2
/2
/2
Res
"p
j1
i, where by Rule 3,
#
p
j1 + zjei arg(1+z)=2 j1 zjei arg(1 z)=2
; i =
z2 + 1
p
p
j1 + zjei arg(1+z)=2 j1 zjei arg(1 z)=2
=
=
2z
z= i
p
p
p
i arg(1 i)=2
i arg(1+i)=2
( i =4)=2
ije
j1 + ije
j1 ije
j1 + ije(i =4)=2
=
=
2i
2i
pp pp
p
1
2
2
2
=
=
= p i:
2i
2i
2
Res [f (z) ; 1] =
=
w2 1
;0 =
Res
w(1 + w2 )
1,
lim
;0 =
w!0
jw2
Res 4
f (z) dz =
Z 1 " p
j1 + zjei arg
1
w2
1
; 05 =
w2 1+ w12
1jei arg(w
w2 + 1
(1+z)=2
j1
2
z +1
zjei arg0 (1
z)=2
dz =
!
Integrate along
1)=2
ei
=2
1+"
1
w
1
f
w2
Integrate along
Z
Res
1 "
1+"
1 "
1+"
1 x2
dx =
1 + x2
2,
151
p
j1 + xjei 0=2 j1
1 + x2
2 "3=2 ! 0:
I:
xjei2
=2
dx !
Integrate along
Z
=
3,
f (z) dz =
Z 1+" p
j1 + zjei arg
3
(1+z)=2
j1
2
z +1
1 "
zjei arg0 (1
dz =
!
and let " ! 0+ . This gives
q
Z
1 (1 ")2
2 "
f (z) dz
1 + (1 ")2
4
Integrate along
z)=2
1 "
1 "
1+"
p
p
j1 + xjei 0=2 j1
x2 + 1
1 x2
dx =
1 + x2
1+"
I:
4,
2 "3=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0
I +0=2 i
1
1
p i+ p i
2
2
i.e.,
Z
1
1
1 x2
dx =
1 + x2
152
1 :
xjei 0=2
dx !
VII.8.5
1 2 3 P L K
Show that the sum of the residues of a rational function on the
extended complex plane is equal to zero.
Solution
I
jzj=R
Since
f (z) dz = 2 i
I
jzj=R
, we get 2 i
Res = 0.
153
VII.8.6
1 2 3 P L K
Find the residue of z= (z 2 + 1) at each pole in the extended complex
plane, and check that the sum of the residues is zero.
Solution
Residue at a simple pole at z1 = i, where by Rule 3,
Res
z2
z
z
;i =
+1
2z
z=i
1
= :
2
z2
z
z
; i =
+1
2z
z= i
1
= :
2
z2
1
= lim
= 1:
2
z!1 z + 1
z!1 1 + 1=z 2
hence
Res
z2
z
;1 =
+1
1;
and the sum of the residues is zero in the extended complex plane.
(Since
z
z 2 +1
+O
1
z2
as z ! 1).
154
VII.8.7
1 2 3 P L K
Find the sum of the residues of (3z 4 + 2z + 1) = (8z 5 + 5z 2 + 2) at its
poles in the (nite) complex plane.
Solution
For residue at 1, we have that
a
3z 5 + 2z 2 + z
3 + 2=z 3 + 1=z 4
3
=
lim
= ;
5
2
3
4
z!1 8z + 5z + z
z!1 8 + 5=z + 1=z
8
hence
3
:
8
We have that the sum of the residues in the complex nite plane is
thus the sum of residues in nite complex plane is 3=8.
Res [f; 1] =
f (z) =
+O
1
z2
as z ! 1, so again
155
VII.8.8
1 2 3 P L K
Fix n 1 and k 0. Find the residue of z k = (z n 1) at 1 by expanding 1= (z n 1) in a Laurent series. Find the residue of z k = (z n 1)
at each nite pole, and verify that the sum of all the residues are
zero.
Solution.
Let f (z) =
f (z) =
zk
zn
zk
.
zn 1
At 1, we have
zk
=
1
1
n
1
zn
= zk
1+z
+z
2n
+ : : : = zk
+z k
2n
+z k
3n
zk
nz n
=
z=wj
z k+1
nz n
wjk+1
wjk+1
=
:
nwjn
n
z=wj
+ wn
1
n
1 wn
1 w
= 0;
unless w = 1.
2 i
If w = 1, then wjk+1 = wj = 1 for every j and w = 1 , e n (k+1) = 1 ,
k+1
= m, thus k = mn 1 for some integer m 1.
n
Thus we have that the sum of the residues in the complex plane are zero.
156
+: : : :
VII.8.9
1 2 3 P L K
Show that f (z) is analytic at 1, then
Res [f (z) ; 1] =
lim z (f (z)
z!1
f (1)) :
Solution
Since f (z) is analaytic at 1, we have
a 1
+O
z
1
z2
a 1
+O
z
1
z2
a0
=a
f (1)) =
f (z) = a0 +
Then
z (f (z)
f (1)) = z a0 +
+O
as z ! 1.
From this follows that
lim z (f (z)
z!1
157
= Res [f (z) ; 1] :
1
z
!a
VII.8.10
1 2 3 P L K
Let D be an exterior domain. Suppose that f (z) is analytic on
D [ @D and at 1. Show that
Z
f( )
1
d = f ( ) f (1) ;
z 2 D:
2 i @D
z
Solution.
Cauchy integral formula for exterior domain. If D is an exterior domaina,
f (z) is analytic in D and att 1, then
Z
f( )
f( )
1
d = f (z) + Res
; 1 = f (z) f (1)
2 i @D
z
z
IIII
Res
f( )
;1 =
z
lim
!1
f( )
=
z
f (1) :
Or
f ( ) = a0 +
f( )
=
z
1
2 i
@D
f( )
1
d = f (z) +
z
2 i
a0
z
+O
2
2
+ :::
1
2
f( )
d =
z
jzj=R
Z
1
a0
= f (z)
d = f (z)
2 i jzj=R
z
158
a0 :
VII.8.11
1 2 3 P L K
If f (z) is not integrable at 1, we dene the principal value
Z 1
Z R
PV
f (x) dx = lim PV
f (x) dx:
R!1
Show that
PV
1
1
8
Im a > 0;
< i ;
1
Im a = 0;
dx = 0;
:
x a
i ; Im a < 0:
Solution
Set a = a1 + a2 i, where a1 and a2 is xed. We compute the principal value
for the following 3 cases.
Case 1, Im a < 0 , a2 < 0. Integrate and let R ! 1. This gives
Z
R
R
dx
x
= log
dx
= [log (x
a1 a2 i
a
R x
R a1 a2 i
+ i arg (R
R a1 a2 i
a1
a1
a2 i)
R a1 a2 i
R a1 a2 i
a2 i)]RR = log
i arg ( R
!0+i 0
a2 i) !
a1
i =
i :
R
R
dx
x
= log "
R
R
Z R
dx
dx
+
= [log jx a1 j]a1R " +[log jx
x
a
x
a
1
1
R
a1 +"
R a1
log jR + a1 j + log jR a1 j log " = log
! 0:
R + a1
dx
=
x a1
a1 "
a1 j]R
a1 +" =
R
R
dx
x
= log
dx
= [log (x
a1 a2 i
a
R x
R a1 a2 i
+ i arg (R
R a1 a2 i
a1
a1
a2 i)
a2 i)]RR = log
i arg ( R
!0+i 2
We have that
PV
1
1
8
Im a > 0;
< i ;
1
Im a = 0;
dx = 0;
:
x a
i ; Im a < 0:
160
a1
R a1 a2 i
R a1 a2 i
a2 i) !
i =i :
VII.8.12
1 2 3 P L K
Suppose that P (z) and Q (z) are polynomials such that the degree
of Q (z) is strictly greater than the degree of P (z). Suppose that
the zeros x1 ; : : : ; xm of Q (z) on the real axis are all simple, and set
x0 = 1. Show that
PV
1
1
X
X
P (x)
P (x)
P (x)
dx = 2 i
Res
; zj + i
Res
; zk ;
Q (x)
Q (x)
Q (x)
summed over the poles zj of P (z) =Q (z) in the open upper halfplane and summed over the xk : s including 1. Hint. Use the preceding
exercise. See also Exercise 6.7
Solution.
By VII.8.11, formula holds for
1
z a
161
VII.8.13
1 2 3 P L K
Show that the analytic dierential f (z) dz transforms the change of variable
w = 1=z to f (1=w) dw=w2 . Show that the residue of f (z) at z = 1
coincides with that of f (1=w) =w2 at w = 0.
Solution
P
We have that Res [f (z) ; 1] = a1 if f (z) = 11 aj z j , jzj > R.
We transform f (z) dz by the change of variable z = w1 , then dz =
and f (z) dz = f w1 w12 dw.
Now we take the residue for f (1=w) dw=w2 at w = 0, we have that
1
w
1
=
w2
1
1
X
1 X aj
( aj ) w
=
w2 1 wj
1
a 4 w2
a 3w
j 2
=
a 1w
a0 w
thus
Res
1
w
1
;0 =
w2
We have that
Res [f (z) ; 1] =
a1 = Res
162
1
dw,
w2
1
w
1
;0 :
w2
VII.9.1
Show using residue theory that
Z
p
3
1
x2
x3
dx =
p
3
Solution
Set
I=
p
3
and integrate
f (z) =
p
3
1
z2
z3
1
x2
dx = q
3
x3
163
dx
1
p
z=3 3 j1
zjei arg0 (1
z)=3
VIII.9.1
arg z
arg0 z
-2
-1
y
0
-2
arg z
arg0 (1
-1
-2
-1
-2
p1
z=3 3
-1
j1 zjei arg0 (1
z)=3
y
4/3
2/3
4/3
-2
2/3
-1
4/3
2/3
2/3
/3
/3
/3
/3
=
Integrate along
Res
1,
Res [f (z) ; 1] =
z)
Res
1
f
w2
1
w
;0 =
1
1
p
;0 =
3
w w 1
lim p
3
w!0
Res 4
jw
1
q
w2 3
1
1
w2
1
w
; 05 =
e
i=3
f (z) dz =
1
1 "
1 "
1
p
jxj2 e2i 0=3 3 j1
q
3
"
Integrate along
2,
Z
Integrate along
Z
"
xjei 2 =3
dz =
zjei arg0 (1
Z
2 i=3
dx = e
z)=3
1 "
p
3
"
!e
2 i=3
p
3
1
x2
1
x2
x3
x3
dx = e
dx !
2 i=3
I:
3,
1
p
z=3 3 j1
"4
p
" (1
")
2 "
2 "9=2 ! 0:
f (z) dz =
3
"
1 "
"
1 "
q
3
1
q
p
3
jxj2 e2i 0=3 3 j1
Integrate along
4,
dz =
zjei arg0 (1 z)=3
Z 1 "
1
p
dx =
dx !
3
2
3
x
x
"
i
0=3
xje
Z 1
1
p
!
dx =
3
x2 x3
0
I:
1
p
z=3 3 j1
"4
p
" (1
")
2 "
2 "9=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
165
2 i=3
I +0
I +0=2 i
i=3
i=3
2 i
= i=3
e
e
2 i=3
i=3
sin ( =3)
i.e.,
Z
p
3
1
x2
x3
dx =
166
p
3
p
3
VII.9.2
Show using residue theory that
Z 1
1
p
x x2
0
dx =
Solution
VII.9.2
2
-R
1
8
Set
I=
and integrate
1
f (z) = p
z z2
1
x x2
p
dx
1
p
= p
i
arg
(z+1)=2
1
z jz + 1je 0
jz
167
1jei arg
(z 1)=2
VIII.9.2
arg0 z
arg z
-2
-1
y
0
arg0 (z + 1)
-2
arg0 (z
-1
-2
-1
-2
-1
1p
jz+1jei arg0 (z+1)=2 jz 1jei arg
(z
1)=2
-2
/2
-1
/2
/2
/2
1
p
Res p
z jz + 1jei arg(z+1)=2 jz
Integrate along
1jei arg(z
1
p
p
jz + 1jei arg(z+1)=2 jz
1,
y
0
1)
1)=2
;0 =
1jei arg(z
1)=2
z=0
168
1
e
i=2
f (z) dz =
1
1+"
p
Z
jz +
1
p
1+"
jx +
Z R
=
1jei 0=2
1+"
1
x x2
p
1jei arg
jz
1jei 0=2
jx
1
dx !
!
Integrate along
2,
Z
Integrate along
Z
(z 1)=2
dz =
dx =
1
x x2
p
dx = I:
1
R R2
p
! 0:
3,
f (z) dz =
3
1 "
1
p
p
dz =
i
arg
(z+1)=2
z jz + 1je 0
jz 1jei arg (z 1)=2
R
Z 1 "
1
p
p
dx =
=
i
=2
x jx + 1je
jx 1jei =2
R
Z R
Z R
1
1
x= t
p
p
dx =
=
dt = I:
2
2
dx
=
dt
1
1
1 " x x
1 t t
Integrate along
4,
Z
Integrate along
5,
"4
p
" (1
")
2 "
2 "9=2 ! 0:
f (z) dz =
5
R
1 "
=
Z
Integrate along
7,
jz +
1 "
R
1 "
6,
Integrate along
1
p
jz
p
x jx + 1jei
1
p
x x2
=2
dx =
1
p
jx
1jei arg
1jei (
x=
dx =
t
dt
(z 1)=2
)=2
dz =
dx =
1
p
2
t t
dt = I:
1
R R2
p
! 0:
f (z) dz =
7
1+"
Integrate along
1
p
dz =
z jz + 1jei arg0 (z+1)=2 jz 1jei arg (z 1)=2
Z 1+"
1
p
p
=
dx =
x jx + 1jei 2 =2 jx 1jei 0=2
R
Z R
1
p
=
dx !
2
1
1+" x x
Z 1
1
p
!
dx = I:
x x2 1
1
p
8,
"4
p
" (1
")
2 "
2 "9=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
170
I + 0 + I + 0 + I + 0 + I + 0 = 2 i ( i) ;
i.e.,
Z
1
x x2
p
171
dx =
VII.9.3
Show using residue theory that
Z 1
1
p
2
1
1 (x + 1)
x2
dx = p
Solution
VII.9.3
7
5
-R
Set
I=
1
1
1
p
(x2 + 1) 1
x2
dx
and integrate
f (z) =
1
p
(z 2 + 1) 1
x2
(z
1
p
p
i) (z + i) j1 + zjei arg (1+z)=2 j1
172
zjei arg0 (1
z)=2
VIII.9.3
arg z
arg0 z
-2
-1
-2
arg (1 + z)
arg0 (1
-1
-2
-1
z)
-2
(z i)(z+i)
-1
1
j1+zjei arg
(1+z)=2
j1 zjei arg0 (1
z)=2
y
3/2
/2
3/2
-2
/2
-1
/2
/2
/2
/2
Res
"
(z
i) (z + i)
1
1
j1
zjei arg(1
z)=2
;i =
p
=
j1 + zjei arg(1+z)=2 j1 zjei arg(1 z)=2 z=i
1
1
p
p
p
= p
i
arg(1+i)=2
i
arg(1
i)=2
(i
=4)=2
(i + i) j1 + zje
j1 zje
2i j1 + zje
j1 zje(i7
1
1
1
1
p = p i
= pp pp
= p
=
2i 2( 1)
2 2i
2 2
2i
2
2e i
(z + i)
j1 + zjei arg(1+z)=2
173
=4)=2
Res
"
(z
i)
=
j1 + zjei arg(1+z)=2
(z
i)
j1 +
p
i) j1 + zjei arg(1
Integrate along
Z
zjei arg(1
j1
i)=2
zjei arg(1+z)=2
p
j1
j1
z)=2
; i =
zjei arg(1
zjei arg(1+i)=2
1
pp pp
=
2i
2
2e0i
z)=2
z= i
p
2i j1 +
j1
1
1
1
p =
p = p i
2i 2
2 2i
2 2
zje( i =4)=2
zje(i
f (z) dz =
1
=
Z
1,
( i
i, where by Rule 3,
1 "
1+"
1 "
p
dz =
j1 + zjei arg (1+z)=2 j1 zjei arg0 (1 z)=2
Z 1 "
1
1
p
p
p
dx =
2
1
(z 2 + 1) j1 + xjei0=2 j1 xjei2 =2
1+" (x + 1)
Z 1
1
p
dx =
2
1 x2
1 (x + 1)
1+"
(z 2 + 1)
Integrate along
2,
Z
Integrate along
3,
"4
p
" (1
")
2 "
174
2 "9=2 ! 0:
x2
I:
dx !
=4)=2
f (z) dz =
1
=
=
1+"
(z 2
1 "
+ 1)
1+"
(z 2
1 "
+ 1)
Integrate along
4,
j1 +
j1 +
1
xjei 0=2
j1
xjei 0=2
!
j1
dx =
zjei arg0 (1
Z 1 "
1+"
1
1
(x2
z)=2
(x2
1
p
+ 1) 1
dz =
1
p
+ 1) 1
x2
dx =
x2
dx !
I:
Integrate along
Z
f (z) dz =
5
=
Z
5,
R
1 "
R
1 "
1
p
p
(z 2 + 1) j1 + zjei arg(1+z)=2 j1
(z 2 + 1)
Integrate along
Z
Integrate along
j1 + xjei(
6,
f (z) dz
6
7,
dz =
zjei arg(1 z)=2
Z R
1
p
p
dx !
dx =
2
)=2
1 x2 ( i)
j1 xjei0=2
1 " (x + 1)
Z 1
1
p
!
dx = J:
2
(x + 1) 1 x2 ( i)
1
(R2
1
p
1) R2
2 R
2
! 0:
R3
175
f (z) dz =
7
=
=
1 "
R
1 "
(z 2
(z 2
+ 1)
Integrate along
j1 +
j1 +
xjei =2
zjei arg(1+z)=2
j1
!
p
j1
xjei2 =2
dx =
(x2
zjei arg(1
Z R
1 "
1
p
+ 1) 1
z)=2
(x2
dz =
1
p
+ 1) 1
x2 ( i)
dx =
x2 ( i)
J:
8,
+ 1)
"4
p
" (1
f (z) dz
4
")
2 "
2 "9=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
I +0
J +0+J +0=2 i
1
1
p i+ p i
2 2 2 2
i.e.,
Z
1
1
(x2
1
p
+ 1) 1
176
x2
dx = p :
2
dx !
VII.9.4
Show using residue theory that
Z 1
1
2
q
dx = p :
3
1 3 (1
x) (1 + x)2
Solution
Set
I=
and integrate
f (z) = q
3
1
(1
z) (1 + z)2
1
1
q
3
(1
= p
3
j1
dx
2
x) (1 + x)
zjei arg0 (1
177
1
q
z)=3
j1 + zj2 e2i arg
(1+z)=3
VIII.9.4
arg0 z
arg z
-2
arg0 (1
-1
-2
z)
-1
-2
arg (1 + z)
y
2
-1
-2
j1 zjei arg0 (1
-1
z)=3
1
p
(1+z)=3
y
4/3
2/3
4/3
-2
2/3
2/3
-1
2/3
0
/3
/3
/3
/3
178
Res [f (z) ; 1] =
1
1
1
; 0 = Res 4 2 q
f
2
w
w
w 3
1
3
2
1
1
; 05 =
Res 4 q
w 3
2
(w 1) (w + 1)
Res
=
2
lim 4 p
w!0
Integrate along
Z
jw
1jei arg(w
1)=3
1
q
1
1
w
5=
1+
1 2
w
i=3
; 05 =
f (z) dz =
1
=
Z
1,
1 "
1+"
1 "
1+"
p
3
j1
xjei2
Integrate along
2,
Z
Integrate along
p
3
dz =
j1 + zj2 e2i arg (1+z)=3
Z 1 "
1
1
2 i=3
q
q
dx = e
dx !
1+" 3 (1
=3
j1 + xj2 e2i 0=3
x) (1 + x)2
Z 1
1
2 i=3
q
!e
dx = e 2 i=3 I:
2
3
1
(1 x) (1 + x)
j1
3,
zjei arg0 (1
z)=3
1
q
"4
p
" (1
")
2 "
179
2 "9=2 ! 0:
f (z) dz =
3
1+"
1 "
1+"
1
q
dz =
1 "
j1 zjei arg0 (1 z)=3 j1 + zj2 e2i arg (1+z)=3
Z 1 "
1
1
q
q
dx !
dx
=
p
2 2i 0=3
2
3
3
1+"
i
0=3
j1 xje
j1 + xj e
(1 x) (1 + x)
Z 1
1
q
!
dx = I:
2
1 3 (1
x) (1 + x)
=
Z
p
3
Integrate along
4,
f (z) dz
4
")
2 "
2 "9=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
e
2 i=3
I +0
I +0=2 i
i=3
i=3
2 ie
e
2 i=3
i=3
2 i
e
i=3
2
=p ;
sin ( =3)
3
i.e.,
Z
1
1
q
3
(1
2
dx = p :
3
x) (1 + x)2
180
VII.9.5
Show using residue theory that
Z 1
1
q
0 (x + 1) 4 x (1
dx = p
:
4
2
x)
3
Solution
Set
I=
and integrate
f (z) =
1
q
(z + 1) 4 z (1
1
q
(x + 1) 4 x (1
=
z)3
dx
x)
p
(z + 1) 4 jzjei arg
181
z=4
1
q
4
j1
z)=4
VIII.9.5
arg z
arg0 z
-2
-1
y
0
arg z
-2
-1
arg0 (1
-2
-1
z)
-2
-1
1
p
z=4 4
/4
7/4
-2
/4
-1
7/4
3/2
/4
3/4
3/4
3/4
3/4
1, where by Rule 3,
182
3/4
y
7/4
z)=4
Res 4
(z + 1)
1
q
4
i
arg(z)=4
jzje
j1
p
4
1
q
jzjei arg(z)=4 4 j1
p
4
; 15 =
=
z)=4
z= 1
q
= p
4
4
i
arg(z)=4
jzje
j1
=
zj3 e3i arg(1 z)=4
1
1
1
1
1
1
p + ip
= p
= p
+i p
4
4
i=4
8e
8
2
2
2 2
242
Residue at a simple pole at z2 = 1 , where by result from Exercise VII.8.13
and Rule 1,
= p
4
Res [f (z) ; 1] =
Res
1
f
w2
1
w
Res 4
;0 =
lim 4
w!0
1
w2
(1 + w)
1
w
+1
1
q
4
(w
1
q
1
w
1)
; 05 =
; 05 = 0
f (z) dz =
1
"
1
3
1 3
w
"
1 "
1 "
(z + 1)
p
4
jzjei arg
1
q
p
(x + 1) 4 jxjei 0=4 4 j1
1
q
z=4 4 j1
dz =
zj3 e3i arg0 (1 z)=4
dx = i
xj3 e3i 2
183
"
=4
!i
1 "
1
q
(x + 1) 4 x (1
1
q
(x + 1) 4 x (1
x)3
dx = iI:
x)
dx !
Integrate along
2,
Z
Integrate along
Z
3,
f (z) dz =
Z "
=
(z + 1)
1
p
4
q
p
4
4
i
0=4
(x + 1) jxje
j1
Integrate along
4,
2 "
")
2 "9=2 ! 0:
"
1 "
"4
p
" (1
f (z) dz
1 "
1
q
4
dz =
zj3 e3i arg0 (1 z)=4
j1
dx =
"
1 "
1
q
(x + 1) 4 x (1
1
q
(x + 1) 4 x (1
dx =
x)
dx !
I:
x)
"4
p
" (1
")
2 "
2 "9=2 ! 0:
Using the Residue Theorem and letting R ! 1 and " ! 0+ , we obtain that
iI + 0
I +0=2 i
1
1
p
+i p
+0 ;
4
2 2
242
i.e.,
Z
1
q
(x + 1) 4 x (1
184
dx = p
:
4
2
x)
3
VIII 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
1
2
3
4
5
6
7
8
VIII.1.1
1 2 3 P
L
K
LLL
Show that z 4 + 2z 2 z + 1 has exactly one root in each quadrant.
Solution
VIII.1.1
Set p (z) = z 4 +2z 2 z +1, and compute 4 arg p (z) along the three segments
in the sector path in gure VIII.1.1. First of all p (z) have no real zeros.
The positive real axis 1 from 0 to R we parametizize as x : z = t; 0 t R,
2
and our function p (z) becomes p (t) = t4 + 2t2 t + 1 = (t2 + 1)
t > 0.
Since p (x) > 0 for t 0, their is no change of argument on this line segment
on the real axis, thus 4 arg p (z) = 0.
The arc 2 from R to iR we parametizize as 2 : z = Reit , 0
t
=2,
4 4it
2 2it
it
and our function p (z) becomes p (t) = R e + 2R e
Re + 1. Since
the variation in argument is determined by the dominating term R4 e4it for R
large the change of argument on this arc is 4 times the variation in argument
for the arc, and thus 4 arg p (z) 4 2 = 2 .
The positive imaginary axis 3 from iR to 0 we parametizize as 3 : z =
it; 0 t R, and our function p (z) becomes f (t) = (it)4 + 2 (it)2 it + 1
2
= (t2 1)
it . We nd that the real part have a zero at t = 1, and the
imaginary part have a zero for t = 0. We make the following table and do
the sketches,
t
1
1
0
Re z
+
0
1
Im z
arg z
0
VIII.1.1 (R = 4)
VIII.1.1 (R = 4)
4
3
1
-2 -1
-1
-2
Thus when we move from iR to 0, the argument for p (z) remain in the 4-th
quadrant except for touching the imaginary axis at i and terminating at
1, and their is no change of argument on this line segment on the imaginary
axis, thus 4 arg p (z) 0.
Now we have that the total change of argument for p (z) is 2 , so we have
exactly one zero in the rst quadrat. Because that the roots come in complex
conjugate pairs it is plain that p (z) have a root in the fourth quadrant too.
2
Becauce that p (x) = (x2 + 1) x > 0 it is clear that p (z) have no real roots
and p (z) must have on has one zero in each quadrant.
VIII.1.2
1 2 3 P L K
Find the number of zeros of the polynomial p (z) = z 4 +z 3 +4z 2 +3z+2
in each quadrant.
Solution
VIII.1.2
p
p
p
p
at t = 2
2
and
t
=
2 + 2, and the imaginary part have a zero for
p
t = 0 and t = 3. We make the following table and do the sketches,
t
1
p
p
2
+
2
p
p3 p
2
2
0
Re z
+
0
0
+
Im z
arg z
0
VIII.1.2 (R = 10)
VIII.1.2 (R = 10)
4
0
+
0
2
3
2
-4
-2
2
-2
-4
Thus when we move from iR to 0, the argument for p (z) on this line segment
on the imaginary axis changes so that, 4 arg p (z)
2 .
Now we have that the total change of argument for p (z) is 0, so we have
no zero in the 1st quadrant.
Since p (z) has real coe cients, the roots come in conjugate pairs (since there
are nor real roots), therefore there are no roots in the 4th quadras as well.
By symmetry, there are 2 roots each in the 2nd and 3rd quadrants.
VIII.1.3
1 2 3 P L K
Find the number of zeros of the polynomial p (z) = z 6 + 4z 4 + z 3 +
2z 2 + z + 5 in the rst quadrant fRe z > 0; Im z > 0g.
Solution
VIII.1.3
t
1
1; 95
1
0
Re z
Im z
arg z
VIII.1.3 (R = 3)
VIII.1.3 (R = 3)
1000
0
+
+
0
0
-2
500
0
0
-2
-4
-1000 -500
-500
500 1000
-6
Thus when we move from iR to 0, the argument for p (z) on this line segment
on the imaginary axis changes so that, 4 arg p (z)
.
Now we have that the total change of argument for p (z) is
4 , so we
have that p (z) has exactly two zeros in the rst quadrat. (None on real or
imaginary axis)
10
VIII.1.4
1 2 3 P
L
K
LLL
Find the number of zeros of the polynomial p (z) = z 9 +2z 5 2z 4 +z+3
in the right half-plane.
Solution
VIII.1.4
t
1
q
4
Re z
3
2
0
+
Im z
+
+
0
arg z
VIII.1.4 (R = 2)
VIII.1.4 (R = 2)
600
2
30
400
200
20
0
-600-400-200
-200
10
-400
-600
-20
-10
10
20
VIII.1.5
1 2 3 P L K
For a xed real number , nd the number of zeros z 4 +z 3 +4z 2 + z+3
satisfying Re z < 0. (Your answer depends on .)
Solution
VIII.1.5
a<0
t
1
p
3
1
0
Re z
+
0
0
+
Im z
a=0
t
1
p
3
1
0
Re z
+
0
0
+
Im z
arg z
0
0<a<1
t
Re z
1
+
p
3 0
1
0
p
a +
0
+
VIII.1.5 (a = 0, R = 10)
VIII.1.5 (a = 0, R = 10)
arg z
0
2
2
2
2
Im z
arg z
0
2
2
0
0
0
0
11
a=1
t
1
p
3
1
0
Re z
+
0
0
+
1<a<3
t
Re z
1
+
p
3 0
p
a
1
0
0
+
a=3
t
1
p
3
1
0
Re z
+
0
0
+
Im z
VIII.1.5 (a = 1, R = 10)
VIII.1.5 (a = 1, R = 10)
VIII.1.5 (a = 2, R = 10)
VIII.1.5 (a = 2, R = 10)
VIII.1.5 (a = 3, R = 10)
VIII.1.5 (a = 3, R = 10)
arg z
0
2
0
0
Im z
arg z
0
2
0
+
0
Im z
0
+
0
3
2
arg z
0
2
12
a>3
t
1
p
pa
3
1
0
VIII.1.5 (a = 9, R = 10)
Re z
+
+
0
0
+
Im z
0
+
+
0
VIII.1.5 (a = 9, R = 10)
arg z
0
0
2
2
Thus when we move from iR to 0, the argument for p (z) only change if
1 < a < 3 and we have that 4 arg p (z)
2 , for all other value on a we
have that 4 arg p (z) 0. Remark that we can se that for a = 1 and a = 3
we have zeros on the imaginary axis.
Now we have that the total change of argument for p (z) is 2 , so we have
exactly one zero in the rst quadrat. Because that the roots come in complex
conjugate pairs it is plain that p (z) has one zero in each quadrant.
If 1
3, the total change of argument for p (z) is 0 and no zeros on
the imaginary axis, thus we must have four zeros in the open left half-plane.
If < 1 and > 3, the total charge of argument for p (z) is
2 and no
zeros of the imaginary axis, thus we must have two zeros in the open left
half-plane.
If = 1 and = 3, the total charge of argument for p (z) is 2 and we
have two zeros on the imaginary axis, thus we must have remaining two zeros
in the open left half-plane.
13
VIII.1.6
1 2 3 P L K
For a xed real number , nd the number of solutions of z 5 +2z 3 z 2 +z =
satisfying Re z > 0.
Solution
VIII.1.6
Set p (z) = z 5 + 2z 3 z 2 + z
, and compute 4 arg p (z) along the two
segments in the contuour in gure VIII.1.6.
The arc 1 from iR to iR we parametizize as 2 : z = Reit ,
=2 t
=2,
and our function p (z) becomes p (t) = R5 e5it +2R3 e3it R2 e2it +Reit . Since
the variation in argument is determined by the dominating term R5 e5it for R
large (independent of the value of ) the change of argument on this arc is 5
times the variation in argument for the arc, and thus 4 arg p (z) 5 2 = 52 .
The positive imaginary axis 2 from iR to iR we parametizize as 2 : z =
it; R
t
R, and our function p (z) becomes f (t) = (it)5 + 2 (it)3
2
(it)2 +pit
= t2 p + it (t2 1) . We nd that the real part have a zero at
and t =
and the imaginary part have a zero for t = 1, t = 0
t=
and t = 1. We make the following table and do the sketch,
14
VIII.1.6 (a = 4, R = 4)
t
1
Re z
Im z
arg z
VIII.1.5 (a = 2, R = 10)
t
1
p
3
p
a
1
0
Re Im
+ 0
0
0
0
+
3
0
Re z
Im z
VIII.1.5 (a = 2, R = 10)
arg
0
2
3
2
VIII.1.5 (a = 0, R = 10)
t
1
VIII.1.6 (a = 4, R = 4)
arg z
15
VIII.1.5 (a = 0, R = 10)
Thus when we move from iR to 0, the argument for p (z) remain in the 4-th
quadrant except for touching the imaginary axis at i and terminating at
1, and their is no change of argument on this line segment on the imaginary
axis, thus 4 arg p (z) = 0.
Now we have that the total change of argument for p (z) is 2 , so we have
exactly one zero in the rst quadrat. Because that the roots come in complex
conjugate pairs it is plain that p (z) has one zero in each quadrant.
16
VIII.1.7
1 2 3 P L K
For a xed complex number , show that if m and n are large integers, then
the equation ez = z + has exactly m + n solutions in the horizontal strip
f 2 im < Im z < 2 ing.
Solution
VIII.1.7
Set p (z) = z 4 +2z 2 z +1, and compute 4 arg p (z) along the three segments
in the contuour in gure VIII.1.7.
The positive real axis 1 from 0 to R we parametizize as x : z = t; 0 t R,
2
and our function p (z) becomes p (t) = t4 + 2t2 t + 1 = (t2 + 1)
t > 0.
Since p (x) > 0 for t 0, their is no change of argument on this line segment
on the real axis, thus 4 arg p (z) = 0.
The arc 2 from R to iR we parametizize as 2 : z = Reit , 0
t
=2,
and our function p (z) becomes p (t) = R4 e4it + 2R2 e2it Reit + 1. Since
the variation in argument is determined by the dominating term R4 e4it for R
large the change of argument on this arc is 4 times the variation in argument
for the arc, and thus 4 arg p (z) 4 2 = 2 .
The positive imaginary axis 3 from iR to 0 we parametizize as 3 : z =
it; 0 t R, and our function p (z) becomes f (t) = (it)4 + 2 (it)2 it + 1
2
= (t2 1)
it . We nd that the real part have a zero at t = 1, and the
imaginary part have a zero for t = 0 on the positive imaginary axis. We
make the following table and do the sketch,
17
VIII.1.7
t
1
1
0
Re Im
+
0
+ 0
arg
0
20
10
-20
-10
-10
10
20
-20
Thus when we move from iR to 0, the argument for p (z) remain in the 4-th
quadrant except for touching the imaginary axis at i and terminating at
1, and their is no change of argument on this line segment on the imaginary
axis, thus 4 arg p (z) = 0.
Now we have that the total change of argument for p (z) is 2 , so we have
exactly one zero in the rst quadrat. Because that the roots come in complex
conjugate pairs it is plain that p (z) has one zero in each quadrant.
18
VIII.1.8
1 2 3 P
L
K
LLL
Show that if Re > 1, then the equation ez = z +
solution in the left half-plane.
Solution
VIII.1.8
Set p (z) = z 4 +2z 2 z +1, and compute 4 arg p (z) along the three segments
in the contuour in gure VIII.1.8.
The positive real axis 1 from 0 to R we parametizize as x : z = t; 0 t R,
2
and our function p (z) becomes p (t) = t4 + 2t2 t + 1 = (t2 + 1)
t > 0.
Since p (x) > 0 for t 0, their is no change of argument on this line segment
on the real axis, thus 4 arg p (z) = 0.
The arc 2 from R to iR we parametizize as 2 : z = Reit , 0
t
=2,
and our function p (z) becomes p (t) = R4 e4it + 2R2 e2it Reit + 1. Since
the variation in argument is determined by the dominating term R4 e4it for R
large the change of argument on this arc is 4 times the variation in argument
for the arc, and thus 4 arg p (z) 4 2 = 2 .
The positive imaginary axis 3 from iR to 0 we parametizize as 3 : z =
it; 0 t R, and our function p (z) becomes f (t) = (it)4 + 2 (it)2 it + 1
2
= (t2 1)
it . We nd that the real part have a zero at t = 1, and the
imaginary part have a zero for t = 0 on the positive imaginary axis. We
make the following table and do the sketch,
19
VIII.1.8
t
1
1
0
Re Im
+
0
+ 0
arg
0
20
10
-20
-10
-10
10
20
-20
Thus when we move from iR to 0, the argument for p (z) remain in the 4-th
quadrant except for touching the imaginary axis at i and terminating at
1, and their is no change of argument on this line segment on the imaginary
axis, thus 4 arg p (z) = 0.
Now we have that the total change of argument for p (z) is 2 , so we have
exactly one zero in the rst quadrat. Because that the roots come in complex
conjugate pairs it is plain that p (z) has one zero in each quadrant.
Set f (z) = ez z
, Re > 1, and get f (iy) = cos y Re + i (sin y y)
i Im . Thus when we move from f ( iR) to f (iR), f (iy) remains in the left
half-plane and the change in arg f is
.
When we move from f (iR) to f ( iR) along f Rei , 2 < < 32 , we cross
the real axis from the 4-th to the 1 st quadrant, thus the change in arg f is
again
. (May also use Rouches Theorem.)
t
1
q
4
Re Im
+
3
2
0
+
+
+
arg
VIII.1.4 (R = 2)
100
50
VIII.1.4 (R = 2)
30
20
0
-100
-50
50
100
10
-50
-100
20
-20
-10
10
20
t
1
q
4
Re Im
+
3
2
0
+
+
+
arg
VIII.1.4 (R = 2)
400
200
VIII.1.4 (R = 2)
30
20
0
-400 -200
200
400
10
-200
-20
-400
(it)4
(it)3 + 13 (it)2
it +36 = t4 + it3
21
13t2
it + 36
-10
10
20
VIII.1.9
1 2 3 P L K
Show that if f (z) is analytic in a domain D, and if is a closed curve in D
such that the values of f (z) on lie in the slit plane Cn ( 1; 0] then the
increase in the argument of f (z) around is zero.
Solution
22
VIII.2.1
1 2 3 P
L
K
LLL
Show that 2z 5 + 6z 1 has one root in the interval 0 < x < 1 and
four roots in the annulus f1 < jzj < 2g.
Solution
Set rst p (z) = f1 (z) + h1 (z), where f1 (z) = 2z 5 and h1 (z) = 6z 1. On
the circle jzj = 2 we have,
jf1 (z)j = j2z 5 j = 2 25 = 64 and jh1 (z)j = j6z 1j
13, then jf1 (z)j >
jh1 (z)j on jzj = 2 so by Rouches Theorem p (z) has all 5 roots in the disk
jzj = 2.
Set now p (z) = f2 (z) + h2 (z), where f2 (z) = 6z and h2 (z) = 2z 5 1. On
the circle jzj = 1 we have, jf2 (z)j = j6zj = 6 and jh2 (z)j = j2z 5 1j
3,
then jf2 (z)j > jh2 (z)j on jzj = 1 so by Rouches Theorem p (z) has 1 roots
in the disk jzj = 1.
It follows that p (z) = 2z 5 + 6z 1 has 5 1 = 4 roots in the annulus
1 < jzj < 2. As any complex roots come in conjugate pairs that have the
same magnitude, the root in the disk jzj = 1 must be real, we have that
p (0) = 1 and p (1) = 7, thus this root must be in the interval 0 < x < 1.
23
VIII.2.2
1 2 3 P
L
K
LLL
How many roots does z 9 + z 5
fjzj = 1g and fjzj = 2g?
Solution
Set rst p (z) = f1 (z)+h1 (z), where f1 (z) = z 9 and h1 (z) = z 5 8z 3 +2z+1.
On the circle jzj = 2 we have,
jf1 (z)j = jz 9 j = jzj9 = 29 = 512 and jh1 (z)j = jz 5 8z 3 + 2z + 1j
101,
then jf1 (z)j > jh1 (z)j on jzj = 2 so by Rouches Theorem p (z) has 9 roots
in the disk jzj = 2.
Set now p (z) = f2 (z) + h2 (z), where f2 (z) = z 9 + z 5 + 2z + 1 and h2 (z) =
8z 3 . On the circle jzj = 1 we have, jf2 (z)j = j 8z 3 j = 8 and jh2 (z)j =
jz 9 + z 5 + 2z + 1j
5, then jf2 (z)j > jh2 (z)j on jzj = 2 so by Rouches
Theorem p (z) has 3 roots in the disk jzj = 1.
Thus it follows that p (z) = z 9 + z 5 8z 3 + 2z + 1 has 9 3 = 7 roots in the
annulus 1 < jzj < 2.
24
VIII.2.3
1 2 3 P
L
K
LLL
Show that if m and n are positive integers, the the polynomial
z2
+
2!
has exactly n zeros in the unit disk.
p (z) = 1 + z +
zm
+ 3z n
m!
Solution
m
2
Set p (z) = f (z) + h (z), where f (z) = 3z n and h (z) = 1 + z + z2! + + zm! .
On the circle jzj = 1 we have,
2
m
jf (z)j = j3z n j = 3 jzjn = 3 1n = 3 and jh (z)j = 1 + z + z2! +
+ zm!
P1 1
k=0 k! = e, then jf (z)j > jh (z)j on jzj = 1 so by Rouches Theorem
2
m
p (z) = 1 + z + z2! + + zm! + 3z n has exactly n roots in the unit disk jzj = 1.
25
VIII.2.4
1 2 3 P
L
K
LLL
Fix a complex number such that j j < 1. For n
1, show that
n z
(z 1) e
has n zeros satisfying jz 1j < 1 and no other zeros in
the right half-plane. Determine the multiplicity of the zeros.
Solution
Set p (z) = f (z) + h (z), where f (z) = (z 1)n ez and h (z) =
. On the
i
circle jz 1j = 1 which we parametrize by z = 1 + e , 0
2 we have,
n 1+i
n z
i
1+cos
1
jf (z)j = j(z 1) e j = 1 + e
e
=e
1, then 0
2 and jh (z)j = j j = j j < 1 from the text, then jf (z)j > jh (z)j on
jz 1j = 1 so by Rouches Theorem p (z) = (z 1)n ez
has n roots in
the disk jz 1j = 1. We have that f 0 (z) = n (z 1)n 1 ez + (z 1)n ez =
(n + z 1) (z 1)n 1 ez , thus the zeros of of f 0 (z) are at z = 1 and at
z = 1 n, where the only zeros in the right half-plane are at z = 1. Since
f (1) =
, the zeros of f (z) must be simple unless = 0, in which case we
ge a zero of order n at z = 1.
***********
Hr r lite konstigt den derivarande satsen, hur funkar den med nollstllen
***********
26
VIII.2.5
1 2 3 P L K
For a xed satisfying j j < 1, show that (z 1)n ez + (z + 1)n has
n zeros in the right half-plane, which are all simple if 6= 0.
Solution
27
VIII.2.6
1 2 3 P L K
Let p (z) = z 6 + 9z 4 + z 3 + 2z + 4 be the polynomial treated in the example
in this section.
(a)
Determine which quadrants contain the four zeros of p (z) that lie inside the
unit circle.
(b)
Determine which quadrants contain the two zeros of p (z) that lie outside the
unit circle.
(c)
Show that the two zeros of p (z) that lie outside the unit circle satisfy
fjz 3ij < 1=10g.
Solution
28
VIII.2.7
1 2 3 P
L
K
LLL
Let f (z) and g (z) be analytic functions on the bounded domain D that
extend continuously to @D and satisfy jf (z) + g (z)j < jf (z)j + jg (z)j on
@D. Show that f (z) and g (z) have the same number of zeros in D, counting
multiplicity. Remark. This is a variant of Rouches theorem, in which the
hypotheses are symmetric in f (z) and g (z). Rouches theorem is obtained
by setting h (z) = f (z) g (z). For the solution of the exercise, see Exercise
9 in the preceding section.
Solution
29
VIII.2.8
1 2 3 P L K
Let D be a bounded domain, and let f (z) and h (z) be meromorphic functions
on D that extend to be analytic on @D. Suppose that jh (z)j < jf (z)j on
@D. Show by example that f (z) and f (z) + h (z) can have dierent numbers
of zeros on D. What can be said about f (z) and f (z) + g (z)? Prove your
assertion.
Solution
30
VIII.2.9
1 2 3 P L K
Let f (z) be a continuously dierentiable function on a domain D. Suppose
that for all complex constants a and b, the increase in the argument of f (z)+
az +b around any small circle in D on which f (z)+az +b 6= 0 is nonnegative.
Show that f (z) is analytic.
Solution
31
VIII.3.1
1 2 3 P
L
K
LLL
Let ffk (z)g be a sequence of analytic functions on D that converges normally
to f (z), and suppose that f (z) has a zero of order N at z0 2 D. Use Rouches
theorem to show that there exists > 0 such that for k large, fk (z) has
exactly N zeros counting multiplicity on the disk fjz z0 j < g.
Solution
32
VIII.3.2
1 2 3 P
L
K
LLL
Let S be the family of univalent functions f (z) dened on the open unit disk
fjzj < 1g that satisfy f (0) = 0 and f 0 (0) = 1. Show that S is closed under
normal convergence, that is, if a sequence in S converges normally to f (z),
then f 2 S. Remark. It is also true, but more di cult to prove, that S is
a compact family of analytic functions, that is, every sequence in S has a
normally convergent subsequence.
Solution
33
VIII.4.1
1 2 3 P
L
K
LLL
Suppose D is a bounded domain with piecewise smooth boundary. Let f (z)
be meromorphic and g (z) analytic on D. Suppose that both f (z) and g (z)
extend analytically across the boundary of D, and that f (z) 6= 0 on @D.
Show that
1
2 i
@D
X
f 0 (z)
dz =
mj g (zj ) ;
f (z)
j=1
n
g (z)
where z1 ; : : : ; zn are the zeros and poles of f (z), and mj is the order of f (z)
at zj .
Solution
34
VIII.4.2
1 2 3 P L K
Let f (z) be a meromorphic function on the complex plane that is doubly
periodic. Suppose that the zeros and the poles of f (z) are at the points
z1 ; : : : ; zn and at their translates by periods of f (z), and suppose no zj is a
translate
of f (z) at zj . Show
P by a period of another zk . Let mj be the order
that
mj zj is a period of f (z). Hint. Integrate zf 0 (z) =f (z) around the
boundary of the fundamental parallelogram P constructed in Section VI.5.
Solution
35
VIII.4.3
1 2 3 P
L
K
LLL
Let ffk (z)g be a sequence of analytic functions on a domain D that converges
normally to f (z). Suppose that fk (z) attains each value w at most m times
(counting multiplicity) in D. Show that either f (z) is constant, or f (z)
attains each value w at most m times in D.
Solution
36
VIII.4.4
1 2 3 P L K
Let f (z) be an analytic function on the open unit disk D = fjzj < 1g.
Suppose there is an annulus U = fr < jzj < 1g such that the restriction of
f (z) to U is one-to-one. Show that f (z) is one-to-one on D.
Solution
37
VIII.4.5
1 2 3 P L K
Let f (z) = p (z) =q (z) be a rational function, where p (z) and q (z) are
polynomials that are relatively prime (no common zeros). We dene the
degree of f (z) to be the larger of the degrees of p (z) and q (z). Denote the
degree of f (z) by d.
(a)
Show that each value w 2 C, w 6= f (1), is assumed d times by f (z) on C.
(b)
Show that f (z) attains each value w 2 C d times on C (as always, counting
multiplicity).
Solution
38
VIII.4.6
1 2 3 P L K
Let f (z) be a meromorphic function on the complex plane, and suppose there
is an integer m such that f 1 (w) has at most m points for all w 2 C. Show
that f (z) is a rational function.
39
VIII.4.7
1 2 3 P L K
Let F (z; w) be a continuous function of z and w that depends analytically
on z for each xed w, and let F1 (z; w) denote the derivative of F (z; w) with
respect to z. Suppose F (z0 ; w0 ) = 0, and F1 (z0 ; w0 ) 6= 0. Choose such
that F (x; w0 ) 6= 0 for 0 < jz z0 j
.
(a)
Show that there exists > 0 such that if jw w0 j < , there is a unique
z = g (w) satisfying jz z0 j < and F (z; w) = 0.
(b)
Show that
Z
F1 ( ; w)
1
d ;
jw w0 j < :
g (w) =
2 i j z0 j= F ( ; w)
(c)
Suppose further that F (z; w) is analytic in w for each xed z, and let F2 (z; w)
denote the derivative of F (z; w) with respect to w. Show that g (w) is analytic, and
g 0 (w) =
(d)
Derive the inverse function theorem given in this section, together with the
formula for the derivative of the inverse function, as a corollary of (a), (b),
and (c).
Remark. this is the implicit function theorem for analytic functions. Note
that a specic formula is given for the function g (w) dened implicitely by
F (g (w) ; w) = 0.
40
VIII.4.8
1 2 3 P L K
Let D be a bounded domain, and let f (z) be a continuous function on D[@D
that is analytic on D. Show that @ (f (D)) f (@D), that is, the boundary
of the open set f (D) is contained in the image under f (z) of the boundary
of D.
Solution
41
VIII.5.1
1 2 3 P L K
Find the critical points and critical values of f (z) = z + 1=z. Sketch the
curves where f (z) is real. Sketch the regions where Im f (z) > 0 and where
Im f (z) < 0.
Solution.
42
VIII.5.2
1 2 3 P L K
Suppose g (z) is analytic at z = 0, with power series g (z) = 2 + iz 4 + O (z 5 ).
Sketch and label the curves passing through z = 0 where Re g (z) = 2 and
Im g (z) = 0.
Solution
43
VIII.5.3
1 2 3 P L K
Find the critical points and critical values of f (z) = z 2 + 1. Sketch the set
of points z such that jf (z)j 1, and locate the critical points of f (z) on the
sketch.
Solution
44
VIII.5.4
1 2 3 P L K
Suppose that f (z) is analytic at z0 . Show that if the set of z such that
Re f (z) = Re f (z0 ) consists of just one curve passing through z0 , then
f 0 (z0 ) 6= 0. Show also that if the set of z such that jf (z)j = jf (z0 )j consists
of just one curve passing through z0 , then f 0 (z0 ) 6= 0.
Solution
45
VIII.5.5
1 2 3 P L K
How many critical points, counting multiplicity, does a polynomial of degree
m have in the complex plane? Justify your answer.
Solution
46
VIII.5.6
1 2 3 P L K
Find and plot the critical points and critical values of f (z) = z 2 + 1 and of
2
its iterates f (f (z)) = (z 2 + 1) + 1 and f (f (f (z))). Suggestion. Use the
chain rule.
Solution
47
VIII.5.7
1 2 3 P L K
Let f (z) be a polynomial of degree m. How many (nite) critical points does
the N
fold iterate f
f (N times) have? Describe them in terms of
the critical points of f (z).
Solution
48
VIII.5.8
1 2 3 P L K
We dene a pole of f (z) to be a critical point of f (z) of order k if z0 is a
critical point of 1=f (z) of order k. We dene z = 1 to be a critical point of
f (z) of order k if w = 0 is a critical point of g (w) = f (1=w) of order k. Show
that with this denition, a point z0 2 C is a critical point of order k for a
meromorphic function f (z) if and only if there are open sets U containing
z0 and V containing w0 = f (z0 ) such that each w 2 v, w 6= w0 , has exactly
k + 1 preimages in U . Remark. We say that f (z) is a (k + 1)
sheeted
covering of f 1 (V n fw0 g) \ U over V n fw0 g.
Solution
49
VIII.5.9
1 2 3 P L K
Show that a polynomial of degree m, regarded as a meromorphic function on
C , has a critical point of order m 1 at z0 = 1.
Solution
50
VIII.5.10
1 2 3 P L K
Locate the critical points ana critical values in the extended complex plane of
the polynomial f (z) = z 4 2z 2 . Determine the order of each critical point.
Sketch the set of points z such that Im z 0.
Solution
51
VIII.5.11
1 2 3 P L K
Show that if f is a rational function, and if g is a fractional linear transformation, then f and g f have the some critical points in the extended
complex plane C . What can be said about the critical values of g f ? What
can be said about the critical points and critical values f g?
Solution
52
VIII.5.12
1 2 3 P L K
Let f (z) = p (z) =q (z) be a rational function of degree d, so that p (z) and
q (z) are reactively prime, and d is the larger of the degrees of p (z) and
q (z). (See Exercise 4.5.) Show that f (z) has 2d 2 critical points, counting
multiplicity, in the extended complex plane C . Hint if deg p 6= deg q, then
the number of critical points of f (z) in the nite plane C is deg (qp0 q 0 p) =
deg p + deg q 1, while the order of the critical points at 1 is jdeg p deg qj
1.
Solution
53
VIII.5.13
1 2 3 P L K
Show that the set of solution points (w; z) of the equation z 2 2 (cos w)+1 = 0
consists of the graphs of two entire functions z1 (w) and z2 (w) of w. Specify
the entire functions, and determine where their graphs meet. Remark. The
solutions set forms a reducible one-dimensional analytic variety in C2 .
Solution
54
VIII.5.14
1 2 3 P L K
Let a0 (w) ; : : : ; am 1 (w) be analytic in a neighborhood of w = 0 and vanish
at w = 0. Consider the monic polynomial in z whose coe cients are analytic
functions in w,
P (z; w) = z m + am
(w) z m
+ a0 (w) ;
jwj < :
Suppose that for each xed w, 0 < jwj < , there are m distinct solutions of
P (z; w) = 0.
(a)
Show that the m roots of the equation P (z; w) = 0 determine analytic
functions z1 (w) ; : : : ; zm (w) in the slit disk fjwj < n ( ; 0]g. Hint. Use
the implicit function theorem (Exercise 4.7).
(b)
Glue together branch cuts to form an m sheeted (possibly disconnected)
surface over the punctured disk f0 < jwj < g on which the branches zj (w)
determine a continuous function.
(c)
Suppose that the indices are arranged so that for some xed k, 1 k n, the
continuation of zj (w) once around w = 0 is zj+1 (w) for 1 j k 1, while
the continuation of zk (w) once around w = 0 is z1 (w). Show that Q (z; w) =
(z z1 (w)) (z zk (w)) determines a polynomial in z whose coe cients
are analytic functions of w for jwj < . Show further that the polynomial
Q (z; w) is an irreducible factor of P (z; w), and that all irreducible factors of
P (z; w) arise from subsets of the zj (w)s in this way.
(d)
Show that if f (z) is an analytic function that has a zero of order m at
z = 0, and z1 (w) ; : : : ; zm (w) are solutions of the equation w = f (z), then
the polynomial P (z; w) = (z z1 (w)) (z zm (w)) is irreducible.
Solution
55
VIII.5.15
1 2 3 P L K
Consider monic polynomial in z of the form
P (z; w) = z m + am
(w) +
+ a0 (w) ;
and deg Pj+1 (z; w) < deg Pj (z; w), until eventually we reach
Pl (z; w) = 0;
Pl
56
Show that if the coe cients of P (z; w) are analytic at w = 0, then the
irreducible factors of P (z; w) can be chosen so that their coe cients are
analytic at w = 0.
(f)
Show that if P (z; w) is irreducible, then there is " > 0 such that for each
xed w, 0 < jwj < ", the roots of P (z; w) are distinct.
(g)
Show that the results of Exercise 14(a)-(c) hold without the supposition that
the solutions of P (z; w) = 0 are distinct.
Solution
57
VIII.6.1
1 2 3 P L K
Sketch the closed path (t) = eit sin (2t), 0
t
2t, and determine the
winding number W ( ; ) for each point not on the path.
Solution
58
VIII.6.2
1 2 3 P L K
Sketch the closed path (t) = e 2it cos t, 0
t
2 , and determine the
winding number W ( ; ) for each point not on the path
Solution
59
VIII.6.3
1 2 3 P L K
Let f (z) be analytic on an open set containing a closed path , and suppose
f (z) 6= 0 on . Show that the increase in arg f (z) around is 2 W (f
; 0).
Solution
60
VIII.6.4
1 2 3 P L K
Let D be a domain, and suppose z0 and z1 lie in the same connected component of CnD.
(a)
Show that the increase in the argument of f (z) = (z z0 ) (z z1 ) around
any closed curve in D is an even multiple of 2 .
(b)
Show that (z z0 ) (z z1 ) has an analytic square in D.
(c)
Show by example that (z z0 ) (z z1 ) does not unnecessarily have an analytic cube in D.
Solution
61
VIII.6.5
1 2 3 P L K
Show that if is a piecewise smooth closed curve in the complex plane, with
trace , and if z0 62 , then
Z
1
dz = 0;
n 2:
(z z0 )n
Solution
62
VIII.6.6
1 2 3 P L K
Let be a closed path in a domain D such that W ( ; ) = 0 for all 62 D.
Suppose that f (z) is analytic on D except possibly at a nite number of
isolated singularities z1 ; : : : ; zm 2 Dn . Show that
Z
X
f (z) dz = 2 i
W ( ; zk ) Res [f; zk ] :
Hint. Consider the Laurent decomposition at each zk , and use Exercise 5.
Solution
63
VIII.6.7
1 2 3 P L K
Evaluate
1
2 i
dz
z (z 2
1)
where is the closed path indicated in the gure. Hint. Either use Exercise
6, or proceed directly with partial fractions.
64
VIII.6.8
1 2 3 P L K
Let (t) and (t), a t b, be closed paths.
(a)
Show that if 2 C does not lie on the straight line segment between (t)
and (t), for a t b, then W ( ; ) = W ( ; ).
(b)
Show that if j (t)
(t)j < j
(t)j for a a
t
b, then W ( ; ) =
W ( ; ).
Solution
65
VIII.6.9
1 2 3 P L K
Let f (z) be a continuous complex-valued function on the complex plane
such that f (z) is analytic for jzj < 1, f (z) 6= 0 for jzj 1, and f (z) ! 1 as
z ! 1. Show that f (z) 6= 0 for jzj < 1.
Solution
66
VIII.6.10
1 2 3 P L K
Let K be a nonempty closed bounded subset of the complex plane, and let
f (z) be a continuous complex-valued function on the complex plane that
is analytic on CnK and at 1. Show that every value attained by f (z) on
C = C [ f1g is attained by f (z) somewhere on K, that is f (C ) = f (K).
Solution
67
VIII.7.1
1 2 3 P L K
Let f (z) be an entire function, and suppose g ( ) is analytic for in the open
upper hand lower half-planes and across the interval ( 1; 1) on the real line.
Suppose that
Z 1
f (x)
dx = g ( )
1 z
for in the upper half-plane. What is the value of the integral when
the lower half-plane? Justify your answer carefully.
Solution
68
is in
VIII.7.2
1 2 3 P L K
Show that
Z
1
1
dx
x
= Log
1
+1
2 Cn [ 1; +1] :
(Note that we use the principal branch of the logarithm here.) Reconcile this
result with your solution to Exercise 1.
Solution
69
VIII.7.3
1 2 3 P L K
Find the Cauchy integrals of the following functions around the unit circle
= fjzj =g, positively oriented.
(c) x = Re z
(d) y = Im (z)
(a) z
(b) z1
Solution
70
VIII.7.4
1 2 3 P L K
Suppose f (z) is analytic on an annulus f < jzj < g, and let f (z) = f0 (z)+
f1 (z) be the Laurent decomposition of f (z). (See Section VI.1.) Fix r
between and , and let F ( ) be the Cauchy integral of f (z) around the
circle jzj = r. Show that f0 ( ) = F ( ) for j j < r, and f1 ( ) = F ( ) for
j j > r. Show further that f0 ( ) = F ( ) and f1 ( ) = F+ ( ). Remark
The formula f (z) = f0 (z) + f1 (z) reects the jump theorem for the Cauchy
integral of f (z) around circles jzj = r.
Solution
71
VIII.7.5
1 2 3 P L K
Let be a piecewise smooth curve, and let F ( ) be the Cauchy integral (7.1)
of a continuous function f (z) on . Show that if g (z) is a smooth function
on the complex plane that is zero o some bounded set, then
Z
ZZ
@g
F (z) dx dy:
g (z) f (z) dz = 2i
C @z
Hint. Recall Pompeius formula (Section IV.8).
72
VIII.7.6
1 2 3 P L K
Determine whether the point z lie inside or outside. Explain.
Solution
73
VIII.7.7
1 2 3 P L K
A simple arc in C is the image of a continuous one-to-one function (t)
from a closed interval [a; b] to the complex plane. Show that a simple arc
in C has a connected complement, that is Cn is connected. You may
use the Tietze extension theorem, that a continuous real-valued function on
a closed subset of the complex plane can be extended to a continuous realvalued function on the entire complex plane. Hint. Suppose z0 belongs to
a bounded complement of Cn . Find at continuous determination h (z) of
log (z z0 ) on , extend h (z) to a continuous function on C , and dene
f (z) = z z0 on the component Cn containing z0 , and f (z) = eh(z) on the
remainder of C n . Consider the increase in the argument of f (z) around
circles centered at z0 .
Prove the Jordan curve theorem for a simple closed curve by lling in the
following proof outline.
(a)
Show that each component Cn has boundary . Hint. For z0 = (t0 ) 2 ,
apply the preceding exercise to the simple arc n (I), where I is a small
open parameter interval containing t0 .
(b)
Prove the Jordan curve theorem in the case where contains a straight line
segment.
(c)
Show that for any z0 = (t0 ) 2 , any small disk D0 containing z0 , and
component U of Cn , there are points z1 = (t1 ) and z2 = (t2 ) such that
the image of the parameter segment between t1 and t2 is contained in D0 and
such that z1 and z2 can be joined by a broken line segment in U \ D0 .
(d)
With notation as in (b), let be the simple curve obtained by replacing the
segment of in D0 between z0 and z1 by the broken line segment in U \ D0
between them, and let be the simple closed curve in D0 obtained by the
following the segment of in D0 from z0 to z1 and returning to z0 along the
broken line segment. Show that W ( ; ) = 0 and W ( ; ) = W ( ; ) for
2 Cn , 6= D0 .
(e)
74
Using (b) and (d), show that Cn has at least two components and that
W ( ; ) = 1 for in each bounded component of Cn .
(f)
By taking U in (c) to be a bounded component of Cn , show that W ( ; ) =
0 for in any other component of Cn .
Solution
75
VIII.8.1
1 2 3 P L K
Which of the following domains in C are simply connected? Justify your
answers. (a) D = fIm z > 0g n [0; i], the upper half-plane with a vertical
slit from 0 to i. (b) D = fIm z > 0g n [i; 2i], the upper half-plane with a
vertical slit from i to 2i. (c) D = Cn [0; +1], the complex plane slit along
the positive real axis. (d) D = Cn [ 1; 1], the complex plane with an interval
deleted.
Solution
76
VIII.8.2
1 2 3 P L K
Show that a domain D in the extended complex plane C = C [ f1g is
simply connected if and only if its complement C nD is connected. Hint. If
D 6= C , move a point in the complement of D to 1. If D = C , rst deform
a given closed path to one that does not cover the sphere, then deform it to
a point by pulling along arcs of great circles.
Solution
77
VIII.8.3
1 2 3 P L K
Which of the following domains in C are simply connected? Justify your
answers. (a) D = C n [ 1; 1], the extended plane with an interval deleted,
(b) D = C n f 1; 0; 1g, the thrice-punctured sphere.
Solution
78
VIII.8.4
1 2 3 P L K
Show that a domain D in the complex plane is simply connected if and only
if any analytic function f (z) on D does not vanish at any point of D has an
analytic logarithm on D. Hint if f (z) 6= 0 on D, consider the function
Z z 0
f (z)
G (z) =
dw:
z0 f (w)
Solution
79
VIII.8.5
1 2 3 P L K
Show that a domain D is simply connected if and only if any analytic function
f (z) on D that does not vanish at any point of D has an analytic square
root on D. Show that this occurs if and only if for any point z0 62 D the
function z z0 has an analytic square root on D.
Solution
80
VIII.8.6
1 2 3 P L K
Show that a domain D is simply connected if and only if each continuous
function f (z) on D that does not vanish at any point D has continuous
logarithm on D.
Solution
81
VIII.8.7
1 2 3 P L K
Let E be a closed connected subset of the extended complex plane C . Show
that each connected component C nE is simply connected.
Solution
82
VIII.8.8
1 2 3 P L K
Show that simple connectivity is a "topological property" that is, if U and
V are domains, and ' is a continuous map of U onto V such that ' 1 is also
continuous, then U is simply connected if and only if V is simply connected.
Solution
83
VIII.8.9
1 2 3 P L K
Suppose that f (z) is analytic on a domain D, and f 0 (z) has no zeros on D.
Suppose also that f (D) is simply connected, and that there is a branch g (w)
of f 1 that is analytic on w0 = f (z0 ) and that can be continued analytically
along any path in f (D) starting at w0 . Show that f (z) is one-to-one on D.
Solution
84
VIII.8.10
1 2 3 P L K
We
P dene an integral 1 cycle in D to be an expression of the form =
kj j , where 1 ; : : : ; m are closed paths in D and k1 ; : : : ; km are integers.
P
We dene the winding number of about to be W ( ; ) = kj W j ; ,
2 CnD. Show that if h ( ) is a continuous integer-valued function on
C nD such that h (1) = 0, then there is an integral 1-cycle on D such
that W ( ; ) = h ( ) for all 2 CnD.
Solution
85
VIII.8.11
1 2 3 P L K
An integral 1-cycle is homologous to zero i D if W ( ; ) = 0 whenever
62 D. Let U be a bounded domain whose boundary consists of a nite
number of piecewise smooth closed curves 1 ; : : : ; m , oriented positively with
respect to U , such that P
U together with its boundary is contained in D. Show
that the 1-cycle @U =
j is homologous to zero in D.
Solution
86
VIII.8.12
1 2 3 P L K
Let D be a domain in C such that C nD consists of m + 1 disjoint closed connected sets. Show that there are m piecewise smooth closed curves 1 ; : : : ; m
such thatPevery integral 1-cycle can be expressed uniquely in the form
= 0 + kj j , where the kj s are integers and 0 is homologous to zero in
D. Remark. The j s form a homology basis for D.
Solution
87