1
$\begingroup$

Let's assume $u'_k \to v$ weakly where $u_k \in L^2(\Omega)$ and $\Omega \subseteq \mathbb{R}^n$. Here $'$ denotes the weak derivative. Then, for an arbitrary $\phi \in C^\infty_c(\Omega)$, $\langle u_k, \phi' \rangle = - \langle u'_k, \phi \rangle \to - \langle v, \phi \rangle$. If we can deduce that $u_k$ converges weakly, then we obtain that $u_k' \to v$ weakly implies $u_k \to u$ weakly with $u' = v$. However, I'm not sure that we can ensure the weak convergence of $u_k$. Is it actually true? I tried to construct a counter example but failed.

$\endgroup$
  • 1
    $\begingroup$ What if $u_k \equiv (-1)^k$? $\endgroup$ – PhoemueX Nov 9 '18 at 15:08
  • $\begingroup$ @PhoemueX Thanks! It was so elementary. $\endgroup$ – lionce Nov 9 '18 at 15:19
1
$\begingroup$

The example suggested by PhoemueX, namely, $u_k \equiv (-1)^k$, shows that we may not have the weak convergence of $\left(u_k\right)_{k\geqslant 1}$ (but a subsequence converges to a $u$ whose derivative is $0$).

It is also possible that no subsequence of $\left( u_k\right)_{k\geqslant 1}$ converges weakly, for example if $u_k(x)=k$ for all $x\in\Omega$.

$\endgroup$

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.