Let $\Omega$ be a nice open, bounded domain in $\mathbb{R}^n$. (e.g suppose $\bar \Omega$ is a smooth manifold with boundary).

Let $f_n \in W^{1,p}(\Omega),p>n$ and suppose that $f_n \rightharpoonup f$ in $W^{1,p}(\Omega)$.

Question: Is it true that $f_n$ converges uniformly to $f$?

(Since $p>n$ the $f_n$ are continuous, so this is well-defined).

I know two things:

  1. If $f_n$ converges uniformly to some function $g$, then $g=f$.

  2. There is always a subsequence of $f_n $ which converges uniformly to $f$. Indeed, any weakly convergent sequence is bounded in $W^{1,p}$, hence by the Rellich-Kondrachov theorem it has a subsequence which converges in $C(\Omega)$. By point $(1)$ the limit must be $f$.

So, the question is can we can conclude directly that the original sequence was uniformly convergent, without passing to a subsequence.

(This is the case when we have strong Sobolev convergence, by the way).


If $\{f_n\}$ does not converge to $f$ in $C(\Omega)$, then there is $\epsilon_0>0$ and a subsequence $\{f_{n_k}\}$ so that $\|f_{n_k} - f\|_{C^0} \ge \epsilon_0$.

But by taking further subsequence if necessary, $f_{n_k} \to g$ in $C^0$ for some $g$. Then $g$ has to be $f$. This is a contradiction.

  • $\begingroup$ Thanks, this is beautiful. $\endgroup$ May 6 '18 at 7:17

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.