Complex Analysis Radius of Convergence.

7 replies · 2K views · last reply
kidsasd987
Messages
142
Reaction score
4
Hello, I have two questions regarding the Radius of convergence.

1. What should we do at the interval (R-eps, R)
2. It used definition to prove radius of convergence, but I am not sure if it is necessary-sufficient condition of convergence. I get that this can be a sufficient condition but not sure of the necessity
 
Attachments
  • Screenshot of a PDF page from Complex Analysis (Princeton Lectures in Analysis, Volume II), page 34, showing the proof of the radius of convergence theorem with the phrase "choose ε > 0 so small that" highlighted and the equation (L + ε)|z| = r < 1.
    스크린샷 2016-06-21 오후 4.01.05.png
    79.1 KB · Views: 716
  • Handwritten-style math proof about radius of convergence: (L+ε)|z| = r < 1 with L = 1/R and |z| < R, then |z| = R−ε < R, leading to limit expression (R+R²ε−ε²−ε)/R = 1 and a contradiction on interval (R−ε, R).
    RadiusOfConv.png
    12.8 KB · Views: 950
If I understand the question, this may help -- The ε was picked arbitrarily small so the proof shows convergence all the way to ε = 0. So there really is no specific interval (R-ε, R) to worry about.
 
kidsasd987 said:
what I am concerend about is,
hm the author used (L+eps)|Z|<1 to argue that L^n*|Z|^n < (L+eps)^n|Z|^n < r^n whereas r<1

but the thing is as I showed at the uploaded png file (if the derivation is correct), it is clear that |Z|= R-eps satisfies the condition |Z| < R and if |Z|= R-eps, r = 1.

which implies (L+eps)|Z|<1 cannot be true. for all |Z|<R
Ah, also I made a mistake. It should be R+R^2*eps-eps-R*eps^2
hmm do you agree?
 
Last edited:
FactChecker said:
If I understand the question, this may help -- The ε was picked arbitrarily small so the proof shows convergence all the way to ε = 0. So there really is no specific interval (R-ε, R) to worry about.
maybe I screwed up at somewhere.
 
Attachments
  • Handwritten-style math derivation in italic serif type with lines of algebra on the radius of convergence, including (L+ε)|z| = r < 1, L = 1/R, |z| = R-ε, a limit as ε→0 equal to 1, and the conclusion that the proof is valid if R<1.
    RadiusOfConvergence.png
    36.6 KB · Views: 796
kidsasd987 said:
maybe I screwed up at somewhere.
Maybe not. But the point is that ε >0 is arbitrarily small (or r is arbitrarily near 1) so the the interval you are worried about shrinks down to nothing.
 
FactChecker said:
Maybe not. But the point is that ε >0 is arbitrarily small (or r is arbitrarily near 1) so the the interval you are worried about shrinks down to nothing.

Although we ignore the interval, the argument that (L+eps)|Z|<1 does not hold true if we take |Z| = R - eps. (I guess) It just bugs me and I am stuck at that page ;(

and my last conclusion R<1 doesn't seem right at all too.
 
It isn't clear to me what is to be proven. We've only been shown the last part of the text that states "the theorem".
 
Stephen Tashi said:
It isn't clear to me what is to be proven. We've only been shown the last part of the text that states "the theorem".

I am sorry.
 
Attachments
  • Scanned page from Complex Analysis (Princeton Lectures in Analysis, Volume II) showing Theorem 2.5 on the radius of convergence of a power series, with Hadamard's formula 1/R = lim sup |a_n|^(1/n).
    스크린샷 2016-06-23 오후 1.03.35.png
    54.5 KB · Views: 653