2
$\begingroup$

Let $p\in [1,\infty [$ and $\Omega \subset \mathbb R^n$ an open. Let $u_n,\in L^p$. We say that $u_n\rightharpoonup u$ (weak convergence) if $$\lim_{n\to \infty }\int_\Omega (u_n-u)\varphi=0$$ for all $\varphi\in L^{p'}$ (the dual of $L^p$).

Q1) First, what is a the intuition behind this definition ?

Q2) How can I bu sure that $(u_n-u)\varphi\in L^1$ ?In other word that $\int (u_n-u)\varphi$ exist for all $n$.

$\endgroup$
2
$\begingroup$

To complete Marko Karbevski answer : for Q2), as you can remark, it's a convergence in $\mathbb R$. Here we consider $L^p$ for $p\in ]1,\infty [$. Let denote $$T(\varphi)=\int u\varphi\quad \text{and}\quad T_n(\varphi)=\int u_n\varphi.$$

So $T$ and $T_n$ are in $(L^{p'})^*$ (the dual of $L^{p'}$) You say that $(u_n)$ converge weakly if $$\lim_{n\to \infty }T_n(\varphi)=T(\varphi).$$ In other word, if the sequence $(T_n)$ converge pointwise to $T(\varphi)$.

Now, suppose $u_n\to u$ strongly. Then, $$|T_n(\varphi)-T(\varphi)|\leq \int |(u_n-u)\varphi|\leq \|u_n-u\|_{L^p}\|\varphi\|_{L^{p'}}\implies \|T_n-T\|=\sup_{\|\varphi\|_{L^{p'}\leq 1}}|T_n(\varphi)-T(\varphi)|\leq \|u_n-u_n\|_{L^p}\underset{n\to \infty }{\longrightarrow }0,$$ and thus the convergence of $(T_n)$ is uniform.

$\endgroup$
2
$\begingroup$

For $Q2$ use Hölder's inequality.

For $Q1$, I personally keep the intuition for the weak vs convergence in norm from the one of the pointwise vs uniform convergence on (say) bounded functions (which is usually taught earlier than $L^p$ spaces). I do not claim that this is the best or even the right way to think of it though.


If you are in for a more advanced explanaition, look for the definition of the weak topology on a normed space. Since $L^{p'}$ is the dual of $L^{p}$ for $1 \le p < \infty$, the definition you wrote gives exactly the sequential characterisation of the weak topology.

$\endgroup$
  • 1
    $\begingroup$ I think it's a good way to see weak and strong convergence. You can see my answer for more detail. $\endgroup$ – Surb Jun 12 '17 at 11:36

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.