Let $\Omega\subset\mathbb{R}^n $ be bounded smooth domain. Given a sequence $u_m$ in Sobolev space $H=\left \{v\in H^2(\Omega ):\frac{\partial v}{\partial n}=0 \text{ on } \partial \Omega \right \}$ such that $u_m$ is uniformly bounded i.e. $\|u_m\|_{H^2}\leq M$ and given the function $f(u)=u^3-u$.
If I know that $u_m\rightharpoonup u(u\in H)$ in $L^2$ sense i.e. $\int_{\Omega}u_m v\to \int_{\Omega}u v$ for every $v\in H$. Is it true that $f(u_m)\rightharpoonup f(u)$ (in $L^2$ sense)?
I post this as answer,but I am studying myself Sobolev spaces right now, so I am not sure whether it is correct (in fact it may well be completely wrong). The argument will use embedding theorems for Sobolev spaces.
First we have a sequence that converges weakly in $L^2,$ but is bounded in $H^2 = W^{2,2}.$ Since $W^{2,2}$ is reflexive, the unit ball is weakly compact. So you can extract converging subsequences. But every converging subsequence converges tu $u,$ since we know that $v_m\rightharpoonup v$ in $W^{k,p}$ implies $v_m\rightharpoonup v$ in $L^p.$ A sequence in a compact (if you want also metric) space in which every converging subsequence converges to the same limit point, is itself convergent.
Now we know that $W^{2,2}$ embeds compactly in $L^{q},$ for $q < \frac{2n}{n - 4},$ and $n \ge 4$. For smaller $n$ we have compact embedding in $C^{k, \gamma},$ for $\gamma \in (0,1)$ and $k< 2 - n/2 - \gamma.$ Compact embeddings tell us that a weakly converging sequence in the departure space is strongly converging in the arrival space. So all in all, for $n <6,$ we know that the sequence is actually strongly converging in $L^6$.
Part 2 thus tells us that $u_n^3$ is in $L^2$ and this answers some of the comments. Furthermore it is strongy convergent in $L^2.$ So for $n < 6$ we have that actually $f(u_n) \rightarrow f(u)$ strongly in $L^2.$
Last but not least we consider the border case $n = 6.$ Here we have just continuous embedding of $W^{2,2}$ in $L^6$. So we know $u_n^3$ is in $L^2$,and we know that it has some weakly converging subsequence, since it`s norm is bounded. But is $u_n^3\rightharpoonup u^3$? I suppose it is,but I am actually having some trouble proving it.