Analysis of Convergence Radii for Complex Power Series, Assignments of Mathematics

Solutions to problems 1-7 from math 185, focusing on the convergence radii of complex power series. How to determine the minimum radius of convergence for a series and demonstrates the process using various examples. It also covers the effect of dividing a series by z on its radius of convergence.

Typology: Assignments

Pre 2010

Uploaded on 10/01/2009

koofers-user-n36
koofers-user-n36 🇺🇸

5

(1)

9 documents

1 / 2

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
Patrick Corn
Math 185, 7/13/05
Solution Set 5
Problem 1: If |z|<min(R1, R2), then P
n=0 anznand P
n=0 bnznboth converge, so then
P
n=0(an+bn)znconverges as well. So the radius of convergence of this series is at least min(R1, R2).
(You can surely work out what this means when one or both of the radii is .)
And for the second part, if Ris the radius of convergence of P
n=0 anbnzn, we have (by property
(ii) on p. 52):
1
R= lim sup
n→∞ anbn1
R1
1
R2
so RR1R2(provided this product is well-defined).
Problem 2: It is easier to show that P
n=1 nanznhas the same radius of convergence; it should
be clear that dividing a series (with no constant term) by zdoes not affect its radius of convergence,
so this is good enough. Suppose the radius of convergence of P
n=0 anznis R.
This follows directly from the following computation:
lim sup
n→∞ |nan|1/n = lim
n→∞ n1/n ·lim sup
n→∞ |an|1/n = 1 ·1
R=1
R
by property (ii) on p. 52. (I’m omitting the proof that n1/n 1 here, but you should probably
have put it on your homework, to remind yourself and the reader that you know why it’s true–good
old L’Hopital.)
Problem 3: (a) 1, by the ratio test.
(b) If we replaced z3nby zn, the answer would be 27, by the ratio test:
lim
n→∞
cn
cn+1
= lim
n→∞
(n!)3(3n+ 3)!
(3n)!((n+ 1)!)3= lim
n→∞
(3n+ 3)(3n+ 2)(3n+ 1)
(n+ 1)3= 27.
Now, if P
n=0 anznhas radius of convergence R, then P
n=0 anz3nhas radius of convergence R1/3
(compare exercise V.13.3), which you can check by noting that it converges for |z3|< R and diverges
for |z3|> R. So the final answer is 3.
(c) It is not hard to see that the radius is the reciprocal of
lim sup
n→∞ 1
n1/n!
= lim sup
n→∞ 1
n1/n!1/(n1)!
,
but the thing in parentheses goes to 1 as n , and the exponent goes to 0, so this sequence has
a limit of 1, which must equal its lim sup. Then the answer is 1/1 = 1.
(d) It should be clear that the sequence (n!)1/n!is just a subsequence of n1/n, which converges
to 1, so it must also converge to 1. So R= 1.
(e) Well, we want the reciprocal of
lim sup
n→∞ (nn)1/n2= li m sup
n→∞ n1/n = 1,
so R= 1.
Problem 4: This is a tricky one. Let z=e. Assume that θ(0,2π). Then the real part
of the series is P
n=0
cos()
n, and we’d like to show that this converges if θis not a multiple of 2π.
The proof when we replace cos by sin will be similar.
Anyway, we will use a criterion for convergence due to Dirichlet: Suppose the partial sums of
a series Pbnare all bounded, and suppose that cnis a decreasing sequence of positive numbers
1
pf2

Partial preview of the text

Download Analysis of Convergence Radii for Complex Power Series and more Assignments Mathematics in PDF only on Docsity!

Patrick Corn Math 185, 7/13/ Solution Set 5 Problem 1: If |z| < min(R 1 , R 2 ), then

n=0 anz

n (^) and ∑∞ n=0 bnz

n (^) both converge, so then ∑∞ n=0(an+bn)z

n (^) converges as well. So the radius of convergence of this series is at least min(R 1 , R 2 ).

(You can surely work out what this means when one or both of the radii is ∞.) And for the second part, if R is the radius of convergence of

n=0 anbnz

n, we have (by property

(ii) on p. 52): 1 R

= lim sup n→∞

anbn ≤

R 1

R 2

so R ≥ R 1 R 2 (provided this product is well-defined).

Problem 2: It is easier to show that

n=1 nanz

n (^) has the same radius of convergence; it should

be clear that dividing a series (with no constant term) by z does not affect its radius of convergence, so this is good enough. Suppose the radius of convergence of

n=0 anz

n (^) is R. This follows directly from the following computation:

lim sup n→∞

|nan|^1 /n^ = lim n→∞ n^1 /n^ · lim sup n→∞

|an|^1 /n^ = 1 ·

R

R

by property (ii) on p. 52. (I’m omitting the proof that n^1 /n^ → 1 here, but you should probably have put it on your homework, to remind yourself and the reader that you know why it’s true–good old L’Hopital.)

Problem 3: (a) 1, by the ratio test. (b) If we replaced z^3 n^ by zn, the answer would be 27, by the ratio test:

nlim→∞

cn cn+

= (^) nlim→∞

(n!)^3 (3n + 3)! (3n)!((n + 1)!)^3

= lim n→∞

(3n + 3)(3n + 2)(3n + 1) (n + 1)^3

Now, if

n=0 anz

n (^) has radius of convergence R, then ∑∞ n=0 anz

3 n (^) has radius of convergence R 1 / 3

(compare exercise V.13.3), which you can check by noting that it converges for |z^3 | < R and diverges for |z^3 | > R. So the final answer is 3.

(c) It is not hard to see that the radius is the reciprocal of

lim sup n→∞

n

) 1 /n! = lim sup n→∞

n

) 1 /n) 1 /(n−1)! ,

but the thing in parentheses goes to 1 as n → ∞, and the exponent goes to 0, so this sequence has a limit of 1, which must equal its lim sup. Then the answer is 1/1 = 1.

(d) It should be clear that the sequence

(n!)^1 /n!

is just a subsequence of

n^1 /n

, which converges to 1, so it must also converge to 1. So R = 1.

(e) Well, we want the reciprocal of

lim sup n→∞

(nn)^1 /n

2 = lim sup n→∞

n^1 /n^ = 1,

so R = 1.

Problem 4: This is a tricky one. Let z = eiθ^. Assume that θ ∈ (0, 2 π). Then the real part

of the series is

n=

cos(nθ) n , and we’d like to show that this converges if^ θ^ is not a multiple of 2π. The proof when we replace cos by sin will be similar. Anyway, we will use a criterion for convergence due to Dirichlet: Suppose the partial sums of a series

bn are all bounded, and suppose that cn is a decreasing sequence of positive numbers 1

tending to 0. Then

bncn converges. (I leave the proof to you; it’s probably in any good real analysis book.) Apply Dirichlet’s criterion with bn = cos(nθ) and cn = 1/n. By exercise I.10.3 (which you can easily get from de Moivre’s theorem), the partial sums of

bn are bounded in absolute value by 1 | sin(θ/2)| , and the same would be true if we had taken^ bn^ = sin(nθ) instead. (Think about it.) Note that we are using that θ is not a multiple of 2π.

Problem 5: Let us consider the coefficients of zj 1 zk 2 on both sides of the purported equality. If they are the same for all j, k, then we are done. Well, the coefficient of zj 1 zk 2 on the left is simply (^) j!^1 k! , because the only way to get this monomial is

by multiplying z

j 1 j! by^

zk 2 k!. To get the coefficient of^ z

j 1 z

k 2 on the right, we must look only at^

(z 1 +z 2 )j+k (j+k)! ; expanding the binomial, we get that the coefficient of z 1 jz 2 k is (j+k j

(j + k)!

j!k!

as desired.

Problem 6: By exercise V.6.1, it is enough to show that the partial sums sk (z) of

gn(z) are uniformly Cauchy (the absolute convergence follows immediately from conditions (1) and (2)). Well,

|sm+p(z) − sm(z)| =

m∑+p

m+

gn(z)

m∑+p

m+

|gn(z)| ≤

m∑+p

m+

Mn,

and since

Mn converges, its partial sums satisfy the Cauchy condition; for all ǫ > 0 there exists a positive integer N such that that last sum is always < ǫ for m > N. This immediately shows that the sequence (sk(z)) is uniformly Cauchy, so we are done.

Problem 7: For all ǫ > 0 there exists a positive integer N such that |gn(z) − g(z)| < ǫ/ 3 for n ≥ N and for all z ∈ G. Now pick a point z ∈ G. Then there exists δ > 0 such that |w − z| < δ ⇒ |gN (w) − gN (z)| < ǫ/3. So then if |w − z| < δ,

|g(w) − g(z)| ≤ |g(w) − gN (w)| + |gN (w) − gN (z)| + |gN (z) − g(z)| <

ǫ 3

ǫ 3

ǫ 3

= ǫ,

and we are done.

2