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

Is the set of functions $f$ in $C^1([0,1])$ equipped with $\lVert f \lVert = \lVert f\lVert_\infty+\lVert f'\lVert_\infty$ (Case 1) or $\lVert f\lVert=\lVert f \lVert_\infty$ (Case 2) satisfying $\lvert f(0.5) \lvert \leq 1$ and $\int_0^1 \lvert f'(x)\lvert^2 dx \leq1$ relatively compact in $C^1([0,1])$?

I think this might be an application of the Arzela-Ascoli theorem. So I'd like to show that the set of functions is uniformly bounded and equicontinous. Boundedness is easy since

\begin{align}\lvert f(x)\lvert \leq 1+\int_{0.5}^x \lvert f'(t)\lvert dt\leq1+0.5+\int_0^1\lvert f'(t)\lvert^2 dt \quad \forall x. \end{align}

But equicontinuity seems harder since $\int_0^1 \lvert f'(x)\lvert^2 dx \leq1$ doesn't imply anything about the boundedness of $f'(x)$.

share|cite|improve this question
@DavideGiraudo This is a nice counterexample – Norbert Jun 24 '12 at 22:31
up vote 3 down vote accepted

Let $$K:=\{f\in C^1[0,1]\mid |f(0.5)|\leq 1\mbox{ and }\int_0^1|f'(t)|^2\leq 1\}.$$ This set have a compact closure, since for $f\in K$, $x,y\in [0,1]$: $$|f(x)-f(y)|\leq \int_x^y1\cdot |f'(t)|dt\leq \sqrt{x-y},$$ by Cauchy-Scharz inequality. Uniform boundedness have already be shown by the OP, but can also be deduced from equi-continuity and boundedness at one point.

But $K$ hasn't a compact closure for the norm $\lVert f\rVert:=\sup_{0\leq x\leq 1}|f(x)|+\sup_{0\leq x\leq 1}|f'(x)|$. Indeed, consider $f_n(x):=\frac{\sin(nx)}n$. Then $f_n\in K$ for all $k$, and assume that $\{f_{n_k}\}$ is a converging subsequence for $\lVert\cdot\rVert$, to $f$ for example. In particular, $f_{n_k}'$ converges uniformly to $f'$. But $f'_{n_k}(x)=\cos(n_kx)$ and we should have, by uniform convergence, $$f'(0)=\lim_{k\to +\infty}f'_{n_k}\left(\frac{2\pi}{n_k}\right)=\lim_{k\to +\infty}f'_{n_k}\left(\frac{\pi}{2n_k}\right).$$ But the first sequence is always $1$ whereas the other one is always $0$.

share|cite|improve this answer
very nice answer! didn't think of applying Cauchy-Schwarz... – Abhillascha Jun 24 '12 at 22:45

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.