Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

uniformly convergent on compact subsets. prove that $\lim_{k\rightarrow \infty}a_n(k) = a_n$ for each n. This is a homework problem of mine, and I think it's probably the most difficult problem I've faced so far in my time as a math student. I have a general strategy which I think might work but every time I try to write up the solution I end up teasing out more and more complexities which need to be dealt with, it's been a pretty frustrating process. My general strategy is this:

Proceed by Contradiction. Then there is some set of n's, possibly infinite, for which the sequence $a_n(k)$ doesn't converge to $a_n$. By the well-ordering principle of the integers this set has a least element, call it $n_0$. Restrict the domain of our functions to some compact ball centered at zero. We want to show that given arbitrary $\varepsilon>0$ and $K>0$, we can always find some $k>K$ such that $|\sum_{n=0}^{\infty}(a_n(k)-a_n)z^n|>\varepsilon$. In other words, we need to show that in fact the convergence isn't uniform on this compact ball.

To do this I show that the growth of the sequence of coeffecients $(a_n(k)-a_n)$ can be no faster than a certain exponential growth determined by the size of our compact ball, since the power series difference must converge (I think!). Then by choosing a point in our compact ball sufficiently close to zero I can make all those coeffecients greater than $n_0$ in our power series difference of negligible size compared to the $n_0$ coeffecient.

As for all those coeffecients less than $n_0$ there are a finite number forming a polynomial which becomes very close to $(a_0(k)-a_0)$ for points very close to zero. Our coeffecient $(a_{n_0}(k) - a_{n_0})$ stays propped up at a certain value of some minimum magnitude. But it is multiplied by $z^{n_0}$ for some very small value of z, thus it also gets very small for points near zero, I'm hoping that by not allowing K to grow as we bring our choice of z closer and closer to zero, that $(a_0(k)-a_0)$ will stay large enough so that it doesn't cancel out $(a_{n_0}(k) - a_{n_0})$, so if I can keep them different enough in magnitude and choose a point near enough to zero, I should be able to keep the whole thing's magnitude large enough to be bigger than epsilon.

Whew, that was a mouth full, I hope this question isn't too specific to be considered appropriate for this site. Does my strategy seem reasonable? Am I missing something and making things way too complicated? (very possible). Thanks.

share|cite|improve this question
up vote 1 down vote accepted

Hint: the generalized Cauchy formula expresses the $a_n(k)$ in terms of contour integrals.

share|cite|improve this answer
Cauchy's Generalized Integral Formula? – esproff Mar 19 '12 at 0:23
Yes, it's sometimes called that. The exact name will vary from text to text. – Robert Israel Mar 19 '12 at 1:09
Gotta give mad late props on this one, very clever, completely simplified the problem. – esproff Mar 21 '12 at 1:12
Cauchy's Theorem makes this SO MUCH easier than whatever you may try to do the same thing for real power series of a real variable... The next time someone asks you why to study complex numbers, think of this. – GEdgar Jul 5 '12 at 17:49

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.