Math 210A Homework 4

Edward Burkard

n 1
Exercise 1. Express the functions ez , cos z, sin z, (1 + z) , log(1 + z), and as a power series in z. Find
1 + z2
their radii of convergence.
Solution.
(a)
∞
X zn
ez =
n=0
n!
The radius of convergence is:
 

 an 
  1   
 (n + 1)! 
 n!   = lim n + 1 = ∞
R = lim   = lim  1  = lim  
n→∞ an+1  n→∞  n→∞ n!  n→∞
(n+1)!

(b)
∞
X z 2n
cos z =
n=0
(2n)!
2n
z
Fix a value of z ∈ C, then letting an = (2n)! :
   2n   
 an   z (2n + 2)!   (2n + 1)(2n + 2) 
lim   = lim   = lim  =∞
n→∞  an+1  n→∞ (2n)! z 2n+2  n→∞ z2 
thus the radius of convergence is infinite since this holds for all z.
(c)
∞
X z 2n+1
sin z =
n=0
(2n + 1)!
2n+1
z
Fix a value of z ∈ C, then letting an = (2n+1)! :
   2n+1   
 an 
 = lim  z (2n + 3)! 
  (2n + 2)(2n + 3) 
lim  = lim  =∞
n→∞ an+1  n→∞  (2n + 1)! z 2n+3  n→∞  z2 
thus the radius of convergence is infinite since this holds for all z.
(d)
n  
n
X n
(1 + z) = z k (by the binomial theorem)
k
k=0
The radius of convergence here must be infinite since this is a finite sum of powers of z.
(e)
∞
X (−1)(n−1)
log(1 + z) = z n (the Maclauren series)
n
k=0
The radius of convergence is:
 
 
 an   1   
n + 1
n
R = lim   = lim   = lim  =1
 
n→∞ an+1  n→∞  1  n→∞ n 
n+1

(f)
∞ ∞
1 1 X
2 n
X
(−1)n z 2n (geometric series)


= = −z =
1 + z2 1 − (−z 2 )
k=0 k=0
an z n we have that the coefficients are:
P
Embedding this into a standard power series
{an }∞
n=0 = {1, 0, −1, 0, 1, 0, −1, 0, 1, 0, −1, ...}.
Thus by using Hadamard’s formula we have:
1 p
= lim sup n |an | = 1,
R
1
2

thus the radius of convergence is 1.

Exercise 2. Give the power series of log z about z = i and find its radius of convergence.
Solution. Using the method of Taylor series we find that:
∞
πi X −in
log z = + (z − i)n .
2 n=1
n
The radius of convergence is:
 
 
 an   −in   
 = lim  n+1  = lim  n + 1  = 1.
n
 
R = lim 

n→∞ an+1  n→∞  −i  n→∞ n 

n+1

 
Exercise 3. Show that the successive derivatives of an analytic function at a point z can never satisfy f (n) (z) >
n!nn .
 
Solution. Let γ(t) = reit + z and suppose that f (n) (z) > n!nn . Then using the Cauchy integral formula we have:
 Z 
 n! f (w)  > n!nn .


n+1
dw
 2πi
γ (w − z) 
This simplifies to Z 
n!  f (w)  > n!nn

n+1
dw
2π γ (w − z)
 
and further to: Z 
1  f (w)  > nn .

n+1
dw
2π γ (w − z)
 
But now we have:
Z 
kf (γ(t))k∞
Z Z
1  f (w) ≤ 1 kf (γ(t))k∞ =M M M M
nn <

dw |dw| = |dw| = 2πr = n .
2π  γ (w − z)n+1  2π γ rn+1 2πrn+1 γ 2πrn+1 r
√
n
!n
M M
Thus the statement has reduced to nn < n = . Since M and r are positive constants, for sufficiently
r r
√
n
M
large n this inequality will not hold (since there will eventually be an n such that n > ). Thus we have a
 (n)  r
n
contradiction. Therefore it is impossible for f (z) > n!n to hold.

Exercise 4. Evaluate the following integrals:

ez
Z Z Z Z
sin z dz dz
dz, dz, 2+1
, 2−1
,
|z|=1 z |z|=1 z |z|=2 z |z|=2 z
ez ez
Z Z Z
sin z
3
dz, n
dz (n ∈ Z) , n
dz (n ∈ Z) ,
|z|=1 z |z|=1 z |z|=2 (1 − z)
Z
log z 1
n
dz where γ(t) = 1 + eit , 0 ≤ t ≤ 2π, n ≥ 0,
γ z 2
Z
dz 1

dz where γ(t) = 2eit , 0 ≤ t ≤ 2π, n ≥ 0, how about γ(t) = eit ?
1 n
γ z− 2 4
Solution. To do this problem we will make excessive use of the following lemma:
Lemma (Cauchy Integral Formula). Suppose that f is analytic everywhere inside and on a simple closed curve γ.
Let D be the region bounded by γ. Then for all z ∈ D:
2πif (n) (z)
Z
f (w)
= n+1
dw.
n! γ (w − z)

Now to begin the computations:

 3

(a)
ez ez 2πie0
Z Z
dz = dz = = 2πi
|z|=1 z |z|=1 (z − 0)0+1 0!
(b)
Z Z
sin z sin z 2πi sin(0)
dz = dz = =0
|z|=1 z |z|=1 (z − 0)0+1 0!
(c)
Z Z Z !  
dz i dz dz i 2πi 2πi
= − = − =0
|z|=2 z2 + 1 2 |z|=2 z+i |z|=2 z−i 2 0! 0!
(d)
Z Z Z !  
dz 1 dz dz 1 2πi 2πi
= − = − =0
|z|=2 z2 − 1 2 |z|=2 z−1 |z|=2 z+1 2 0! 0!
(e)
Z Z
sin z sin z 2πi(− sin 0)
dz = dz = =0
|z|=1 z3 |z|=1 (z − 0) 2+1 2!
(f)
ez
Z
dz (n ∈ Z)
|z|=1 zn
(n = 0)
Z
ez dz = 0
|z|=1
since ez is analytic.
(n > 0)
ez ez 2πie0
Z Z
2πi
dz = dz = =
|z|=1 zn |z|=1 (z − 0)(n−1)+1 (n − 1)! (n − 1)!
(n < 0) Let m = −n.
ez
Z Z
dz = z m ez dz = 0
|z|=1 zn |z|=1
since z m ez is analytic.
(g)
ez ez
Z Z
dz = dz (n ∈ Z)
|z|=2 (1 − z)n |z|=2 (−1)n (z − 1)n
(n = 0)
ez
Z Z
dz = ez dz = 0.
|z|=2 (−1)n (z − 1)n |z|=2

(n > 0)
ez ez 1
Z Z
n 2πie 2πie
dz = (−1)n dz = (−1) = (−1)n .
|z|=2 (−1)n (z − 1)n |z|=2 (z − 1)(n−1)+1 (n − 1)! (n − 1)!
(n < 0) Let m = −n. Then:
ez
Z Z
dz = ez (1 − z)m dz = 0
|z|=2 (−1) (z − 1)n
n
|z|=2

since ez (1 − z)m is analytic.

(h)
Z
log z 1
n
dz where γ(t) = 1 + eit , 0 ≤ t ≤ 2π, n ≥ 0,
γ z 2
1
This integral is zero ∀ z ∈ C since it is a circle of radius centered at 1 and hence the branch cut of log z
2
is analytic on and inside γ.
 4

(i) Z
dz 1

dz where γ! (t) = 2eit , 0 ≤ t ≤ 2π, n ≥ 0, how about γ2 (t) = eit ?
1 n
γ z− 2 4
Using the Cauchy integral formula:
Z Z
dz dz 2πi 2πi

dz =
1 n
(n−1)+1 = (n − 1)! (1) = (n − 1)!
γ1 z − 2 γ1 z − 1
2
where f (z) = 1.
1 1
As for the second integral, γ2 is a circle of radius centered at 0, so it completely misses the point .
4 2
Thus:
Z
dz

dz = 0.
1 n
γ2 z − 2

z2 + 1
Z
Exercise 5. Evaluate the integral dz, γ(t) = reit , 0 ≤ t ≤ 2π, for all possible values of r, 0 < r < 2
γ z(z 2 + 4)
and 2 < r < ∞.
Solution. We have:
z2 + 1
 
11 3 1 1
2
= + + .
z(z + 4) 4z 8 z + 2i z − 2i
Now let’s look at the two cases:
(0 < r < 2) In this case, the only pole of the integrand that lie inside the curve is 0, so:
z2 + 1
Z Z   
11 3 1 1
2
dz = + + dz
γ z(z + 4) γ 4z 8 z + 2i z − 2i
Z Z Z 
11 3 1 1
= dz + dz + dz
γ 4z 8 γ z + 2i γ z − 2i
1 2πi 3 πi
= + (0 + 0) = .
4 0! 8 2
(r > 2) In this case, all three poles of the integrand are inside the curve, so:
z2 + 1
Z Z   
11 3 1 1
2
dz = + + dz
γ z(z + 4) γ 4z 8 z + 2i z − 2i
Z Z Z 
11 3 1 1
= dz + dz + dz
γ 4z 8 γ z + 2i γ z − 2i
 
1 2πi 3 2πi 2πi
= + +
4 0! 8 0! 0!
πi 3πi
= + = 2πi.
2 2