0
$\begingroup$

So the series to be discussed is $\sum (nx)^{-2}$ $\ (x\ne0)$.
And I think I can show that the series is pointwise convergent in its domain by comparison test with $\frac{1}{n^2}$.
But I'm not sure how to show that it's not uniformly convergent(I think). What I currently know is that Weierstrass M-test cannot be applied. And I feel that I have to use the Cauchy Criterion? But I'm not sure how to write the proof.

$\endgroup$
  • $\begingroup$ Notice that $\sum (nx)^{-2}=ax^{-2}$, where $a=\sum n^{-2}=\pi^2/6$. $\endgroup$ – Mercy King Dec 28 '17 at 6:01
  • $\begingroup$ That's definitely a nice shortcut! $\endgroup$ – Macrophage Dec 28 '17 at 7:08
0
$\begingroup$

Assume it were uniformly convergent. Then there exists some positive integer $N$ such that for all $x\ne 0$ we have $\left|\displaystyle\sum_{k=1}^{N}\dfrac{1}{k^{2}x^{2}}-\displaystyle\sum\dfrac{1}{n^{2}x^{2}}\right|<1$, then $\displaystyle\dfrac{1}{x^{2}}\sum_{k\geq N+1}\dfrac{1}{k^{2}}=\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}x^{2}}<1$, so $0<\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}<x^{2}$ for all $x\ne 0$. Take $x\downarrow 0$ we get $\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}=0$, a contradiction.

$\endgroup$
  • $\begingroup$ Or one can choose $x=\sqrt{\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}}\ne 0$ to get a contradiction at the end. $\endgroup$ – user284331 Dec 28 '17 at 5:43
  • $\begingroup$ I suppose you could only assert the equality when the sum of 1/k^2 is greater than or equal to 0. So it seems like the final part of the proof is circular. $\endgroup$ – Macrophage Dec 28 '17 at 6:43
  • $\begingroup$ I don't quite get it. But $\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}\geq\dfrac{1}{(N+1)^{2}}>0$, doesn't it? $\endgroup$ – user284331 Dec 28 '17 at 6:45
  • $\begingroup$ The final part is due to this: If $x<\epsilon$ for all $\epsilon>0$, then $x\leq 0$. $\endgroup$ – user284331 Dec 28 '17 at 6:51
  • $\begingroup$ I think you misunderstood my argument. No matter what, we must have $\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}>0$, this is one point. Another point is that, because we have like, $\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}<\epsilon$ for all $\epsilon>0$, then we have $\displaystyle\sum_{k\geq N+1}\dfrac{1}{k^{2}}\leq 0$, so this point contradicts with the previous one. $\endgroup$ – user284331 Dec 28 '17 at 6:53

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.