1

CAAM 335 Matrix Analysis:

Solutions to HW 7

Problem 1 (5+5+5=15 points) (a)Compute the integral

R
C

f (z)dz for


C = t 2 + it : t [0, 2] .

f (z) = z

(b)-(c)Verify Cauchys theorem for the functions

f (z) = 3z2 + iz 4,
f (z) = 5 sin(2z),
if C is the square with vertices 1 i, 1 i.
(a)
Z 2

f (z)dz =

(t 2 it)(2t + i)dt =

2t 3 + t + i(t 2 2t 2 )dt

Z 2

Z 2

1 2
1 2
8
1
i t 3 |t=0
= 10 i
2t 3 + t it 2 dt = t 4 + t 2 |t=0
2
2
3
3

(b)-(c) A parametrization of the square with vertices 1 i, 1 i is given by

C = {z(t) = x(t) + iy(t) : t [0, 4)} ,
where

1 + 2t

1
x(t) =
1 2(t 2)

1 + 2(t 1)
y(t) =
1

1 2(t 3)

t [0, 1),
t [1, 2),
t [2, 3),
t [3, 4),

t [0, 1),
t [1, 2),
t [2, 3),
t [3, 4),

Let C j = {z(t) = x(t) + iy(t) : t [ j 1, j)} for j = 1, 2, 3, 4. Then

4

f (z)dz =
C

j=1 C j

f (z)dz =

(g(z( j)) g(z( j 1)).

j=1

where g(z) is an anti-derivative of f (z) so that f (z) = g0 (z).

(b) f (z) = 3z2 + iz 4 = (z3 + iz2 /2 4z)0 . Hence
Z

f (z)dz = (1 i)3 + i(1 i)2 /2 4(1 i) [(1 i)3 + i(1 i)2 /2 4(1 i)] = 2 4i

C1

f (z)dz = (1 + i)3 + i(1 + i)2 /2 4(1 + i) [(1 i)3 + i(1 i)2 /2 4(1 i)] = 14

C2

f (z)dz = (1 + i)3 + i(1 + i)2 /2 4(1 + i) [(1 + i)3 + i(1 + i)2 /2 4(1 + i)] = 2 + 4i

C3

Z
C4

f (z)dz = (1 i)3 + i(1 i)2 /2 4(1 i) [(1 + i)3 + i(1 + i)2 /2 4(1 + i)] = 10

Clearly, their sum is indeed 0.

(c) f (z) = 5 sin(2z) = ( 52 cos(2z))0 . Hence,
Z
C1

Z
C2

Z
C3

Z
C4

5
5
f (z)dz = cos(2(1 i)) + cos(2(1 i)) = 5 sin(2) sinh(2)i
2
2
5
5
f (z)dz = cos(2(1 + i)) + cos(2(1 i)) = 5 sin(2) sinh(2)i
2
2
5
5
f (z)dz = cos(2(1 + i)) + cos(2(1 + i)) = 5 sin(2) sinh(2)i
2
2
5
5
f (z)dz = cos(2(1 i)) + cos(2(1 + i)) = 5 sin(2) sinh(2)i
2
2

Again, they sums up to 0.

Problem 2 (5+5+10 =20 points) Find the residues of the following functions at 0:
(z2 + 1)/z,

ez /z2 ,

(2z + 1)/(z(z3 5)).

Let us call the three functions f j , j = 1, 2, 3. Then by the residue formula

 2

1
z +1
res( f1 , 0) = lim
z
= 1,
z0 (1 1)!
z
 z
d
e
1
z2 2 = lim ez = 1,
res( f2 , 0) = lim
z0
z0 (2 1)! dz
z
 2

1
z +1
res( f3 , 0) = lim
z 3
= 1/5.
z0 (1 1)!
z(z 5)



Problem 3 (5+10=20 points) Let C = 3eit : t [0, 2) . Compute the two integrals
sin(z2 ) + cos(z2 )
dz,
(z 1)(z 2)
C

e2z
dz.
4
C (z + 1)

You may use results from Chapter 8 to arrive quickly at the solution. Be sure to explain your rationale.

Solution 1:
Let C1 be the circle around a = 1 with radius r = 1/2 and C2 be the circle around a = 2 with radius r = 1/2.
Then,
Z
Z
Z
sin(z2 ) + cos(z2 )
sin(z2 ) + cos(z2 )
sin(z2 ) + cos(z2 )
dz =
dz +
dz
(z 1)(z 2)
(z 1)(z 2)
(z 1)(z 2)
C
C1
C2

3
sin(z2 )+cos(z2 )
(z2)

The function f (z) =

Z
C1

sin(12 ) + cos(12 )
sin(z2 ) + cos(z2 )
dz = 2i f (1) = 2i
= 2i(sin() + cos()) = 2i.
(z 1)(z 2)
(1 2)
sin(z2 )+cos(z2 )
(z1)

The function f (z) =

Z
C2

is differentiable on and inside C1 . By the Cauchy Integral Formula,

is differentiable on and inside C2 . By the Cauchy Integral Formula,

sin(z2 ) + cos(z2 )
sin(22 ) + cos(22 )
dz = 2i f (1) = 2i
= 2i(sin(4) + cos(4)) = 2i.
(z 1)(z 2)
(2 1)

Consequently,
sin(z2 ) + cos(z2 )
dz = 2i + 2i = 4i.
(z 1)(z 2)
C

We use the equation

n!
dn f
(a) =
dan
2i

Z
C

f (z)
dz
(z a)n+1

with a = 1, n = 3, f (z) = e2z . We can apply this formula for n = 4 and a = 1, since a = 1 is inside C
and f (z) = e2z is differentiable on C and inside C.
f (z) = e2z ,

f 0 (z) = 2e2z ,

f 00 (z) = 4e2z ,

f 000 (z) = 8e2z .

e2z
2i 0000
2i 2(1) 16i 2 8i 2
dz =
f (1) =
8e
=
e =
e .
4
3!
3!
6
3
C (z + 1)

Solution 2: Alternatively, we can also use the Residue Theorem to compute both integrals.
For the first function,
res(1) = lim (z 1)

sin(z2 ) + cos(z2 ) sin() + cos()

=
=1
(z 1)(z 2)
1

res(2) = lim (z 2)

sin(z2 ) + cos(z2 ) sin(4) + cos(4)

=
= 1.
(z 1)(z 2)
1

z1

z2

Hence

sin(z2 ) + cos(z2 )
dz = 2i(1 + 1) = 4i.
(z 1)(z 2)
C

For the second function,

d3
1
res(1) = lim
z1 (4 1)! dz3
Hence


(z + 1)4

e2z
(z + 1)4


= lim

z1

e2z
8i 2
dz = 2i res(1) =
e .
4
3
C (z + 1)

1 3 2z 4 2
2 e = e .
6
3

Lecture Notes Section 8.5 (P. 93)

Exercise [1] (20 points) Let us confirm the representation (8.7) in the matrix case. More precisely,
if (z) (zI B)1 is the resolvent associated with B then (8.7) states that
h

(z) =

mj

j,k

(z j )k

j=1 k=1

where
j,k =

1
2i

Z
Cj

(z)(z j )k1 dz.

Compute the j,k per (8.15) for the B in (7.13). Confirm that they agree with those appearing in (7.16).
B = [1 0 0; 1 3 0; 0 1 1];
syms z; inv(z*eye(3)-B)
ans =
[
1/(z-1),
0,
[ 1/(z-1)/(z-3),
1/(z-3),
[ 1/(z-1)2/(z-3), 1/(z-1)/(z-3),
Thus

0]
0]
1/(z-1)]

(z 1)(z 3)
0
1

(z

1)
(z

1)2
(zI B)1 =
(z 1)2 (z 3)
1
(z 1)

0
.
0
(z 1)(z 3)

and the eigenvalues of B are 1 = 1 with multiplicity m1 = 2 and 2 = 3 with multiplicity m2 = 1.

We use the Cauchys Theorem (for differentiable terms) and the Residue Theorem (for terms with singularities)
to compute the following integrals:

1
0
0
Z
Z
(z1)
1
1

1
1
0
1,1 =
(z)(z 1)0 dz =
(z1)(z3)
dz
(z3)
2i C(1,1)
2i C(1,1)
1
1
1
(z1)2 (z3)

=  (z3)
0
1
(z3)

1,2 =

1
2i

0
0
1
(z3)

0
0

z=1

21
=
14

(z)(z 1)1 dz =

C(1,1)

0
0
1
(z3)

0 0
0 0
0 0

1
2i

0
0
21

0
= 0
12
z=1

(z1)

0
0
1

1
(z3)
1
(z1)(z3)

(z1)
(z3)
(z1)
(z3)

Z
C(1,1)

(z1)(z3)

0 0
0 0
0 0

0
0
dz
1

2,1 =

1
2i

1
2i

1
2
1
4

(z)(z 3)0 dz =

C(3,1)

0
1
(z1)
1
(z1)2

0
1
1
(z1)

0
0

z=3

C(3,1)

0
1
1
2

1
(z1)
1
(z1)(z3)
1
(z1)2 (z3)

0
1
(z3)
1
(z1)(z3)

0
0
1
(z1)

dz

0
0 .
0

Exercise [2] (10 points) Use (8.14) to compute the inverse Laplace transform of 1/(s2 + 2s + 2).

Recall
(L 1 q)(t)

1
2i

q(z)ezt dz =

res( j )

j=1

where C is a simple closed curve that encloses the poles of q(z), in this case at z = 1 + i and z = 1 i.



Z
1
ezt
1
1
(t)
=
dz = res(1 + i) + res(1 i)
L
s2 + 2s + 2
2i C (z (1 + i))(z (1 i))




ezt
ezt
e(1+i)t e(1i)t


=
+
=
+


z (1 i) 
z (1 + i) 
2i
2i
z=1+i

z=1i

eit eit
= et
= et sin(t)
2i

Exercise [3] (20 points) Use the result of the previous exercise to solve, via the Laplace transform, the differential
equation
x0 (t) + x(t) = et sint, x(0) = 0.
Hint: Take the Laplace transform of each side.

Taking the Laplace transform of each side, and using the result from the previous exercise,

L (x0 (t) + x(t)) = L (et sint)

1
s2 + 2s + 2
1
(s + 1)L x = 2
(s + 2s + 2)
1
Lx =
(s + 1)(s2 + 2s + 2)

sL x x(0) + L x =

Now, we use the inverse Laplace transform to determine x(t).

Let C be a sufficiently large circle that encircles the 3 poles at 3 = 1, 2 = 1 + i and 3 = 1 i.


Z
1
1
ezt
L
(t)
=
dz =
(s + 1)(s2 + 2s + 2)
2i C (z + 1)(z (1 + i))(z (1 i))






ezt
ezt
ezt



+
+
= 2



z + 2z + 2 
(z + 1)(z (1 i)) 
(z + 1)(z (1 + i)) 
z=1
z=1+i
z=1i


t
(1+i)t
(1i)t
it
it
e
e
e
e e
=
= et (1 cos(t))
+
+
= et 1 +
1
(2i)(i)
(2i)(i)
2

Therefore, the solution to the differential equation is x(t) = et (1 cos(t)).

res( j )

j=1