Statement of the problem:

Prove: If a sequence of complex functions $s_n$ on a set $X$ converges uniformly to a complex function $s$, then the sequence of Cesaro means $\sigma_N$ also converges uniformly to $s$.

I have already shown that a sequence of complex numbers $\{s_n\}$ which converges to $s$ has Cesaro means which also converge to $s$, but I'm not sure how to use that to prove this statement. My first instinct was to say that, for each $x$ in the domain, we just have a sequence of complex numbers, so each $\sigma_N(x)$ converges to $s(x)$, but this only establishes pointwise converges (right?). I don't know how to deal with the uniform condition.

Edit: According to my professor, this exercise he assigned is actually false. Can someone come up with an explanation why?

  • $\begingroup$ Does writing $\sigma_N - s = \frac1N\sum_{n=1}^N (s_n-s)$ help ? $\endgroup$ – Sary Mar 2 '15 at 1:56
  • $\begingroup$ @Sary I used that to prove the statement about the sequence complex numbers. Is that still a valid way to prove it for functions when there are infinitely many possibilities for $x$? I figure that each sequence of complex numbers for each point $x$ (denote it $S_x$) and any $\epsilon > 0$ has an associated $N_{x,\epsilon}$ where $n > N_x \Rightarrow |\sigma_N(x) - s(x)| < \epsilon$. While this proves pointwise convergence, does it prove uniform? If it helps, I can post my proof for the sequence of complex numbers statement. $\endgroup$ – MCT Mar 2 '15 at 2:00
  • $\begingroup$ Then write $\delta_n := \sup_x|s_n(x)-s(x)|$. What do you know about that sequence of numbers ? Can you give an upper bound for $\sup_x |\sigma_N(x) - s(x)|$ in terms of the $\delta_n$'s ? $\endgroup$ – Sary Mar 2 '15 at 2:09
  • $\begingroup$ @Sary OK, how do I know that $\delta_n$ exists and is finite? That is essentially my main confusion. If I know why it exists, the rest of the proof I can do. $\endgroup$ – MCT Mar 2 '15 at 2:11
  • 1
    $\begingroup$ @Sary see my answer. It doesn't work because $\delta_n$ might not exist for small values of $n$. $\endgroup$ – MCT Mar 2 '15 at 15:25

Turns out that the premise is incorrect, as a recent email by my professor has notified.

On $(0,1]$ let $s_1(x) = \frac{1}{x}$ and $s_n = 0$ for $n > 1$, then $s_n \to 0$ uniformly but $\sigma_N \to 0$ pointwise.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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