1
$\begingroup$

The task is to show $$f(x) = \sum_{n=1}^\infty \frac{\sin{nx}}{n^{5/2}}$$ is a continuous function on $\mathbb{R}$ with a continuous derivative.


Showing that the series converges at each point is easy, assume that has been shown. Next I suppose I need to show that $f(x)$ is continuous. My thought was to define $$f_N(x) = \sum_{n=1}^N \frac{\sin{nx}}{n^{5/2}},$$ and show that $f_N \to f(x)$ uniformly (then $f$ being continuous follows).

I think the following is a valid proof that $f_N \to f$ uniformly. It suffices to show $|f_N(x) - f_M(x)| < \epsilon$ for all $x$ when $N,M > $ some $Z \in \mathbb{Z},$ and assume $N > M.$ $$|f_N(x) - f_M(x)| = \sum_{n=M+1}^N \frac{\sin{nx}}{n^{5/2}} < \sum_{n=M+1}^N n^{-2} < \epsilon$$ for large enough $M,N.$


The part I am stuck on is showing $f'(x)$ exists for all $x \in \mathbb{R}.$ Rudin's book states that if

$f_n$ are differentiable
$f'_n$ uniformly converge
$f_n(x_0)$ converges for some point $x_0$
Then $f_n \to f$ uniformly and $f' = \lim f'_n(x)$

I am wondering if there is an easier way to show that $f(x)$ is continuously differentiable? And if not, would showing $f'_n$ converge uniformly be the same as above - using a Cauchy sequence for $f'_n$'s?

$\endgroup$
  • 1
    $\begingroup$ As another user190080 noted, you may use the Weierstrass M-test to prove uniform convergence. You have essentially provided a proof for this fact via Cauchy's criterion for uniform convergence with $M_n=n^{-2}$. $\endgroup$ – user293794 Jul 14 '16 at 16:40
1
$\begingroup$

Actually you could use the Weierstrass-M-criterion to show that the above sequence is converging uniformly, just observe that $$ \bigg|\frac{\sin{(nx)}}{n^{5/2}}\bigg|\leq\frac{1}{n^{5/2}} $$ and furthermore $$ \sum_{n=1}^{\infty}\frac{1}{n^{5/2}}=\xi(\frac 5 2)<\infty $$ hence while using the above criterion you can infer that the sequence $$ f_N(x) = \sum_{n=1}^N \frac{\sin{nx}}{n^{5/2}} $$ converges absolutely and uniformly to some $f$.

For the differentiability part I'd suggest the same as you already mentioned (just again, use the Weierstrass criterion, it will also work).

$\endgroup$

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.