Weak Convergence of sequence in a Sobolov Space.

60 Views Asked by At

Consider the question asked in here.

I understood most of the answer in the question but the part about the weak convergence I did not get. To show that $u'_{n_k}$ converges weakly to $u$ in $L^p$, we have to show that for every linear continuous functional $J:L^p\to \mathbb{R}$, we have $J(u'_{n_k})\to J(u')$ in $\mathbb{R}$. Any hints on how to solve this one?

My Try By the Riez representation theorem we can find $f\in L^{p'}$ such that for all $v\in L^p$, $$J(v)=\int_0^1 vf.$$ We can apply this to $u'_{n_k}$ to obtain $$J(u'_{n_k})=\int_0^1u'_{n_k}f,\;\;\;\text{ for all }k.$$ But then how to proceed from here?

1

There are 1 best solutions below

2
On BEST ANSWER

What you want to show is that the convergence holds for all functionals $J$ induced by an element $f\in L^{p'}$.

What Umberto P. has shown in the answer that you linked is that the convergence holds for all functionals $J$ induced by an element $\psi\in C_0^\infty(0,1)$.

There is a result in functional analysis that says that in order to show weak convergence of a bounded sequence in a space $X$ it suffices to test the sequence with functionals from a dense set of the dual space

proof: Let $u_n$ be a sequence in $X$ that is bounded by $C>0$ and let $M\subset X^*$ be dense. Suppose we already know that $f(u_n)\to f(u)$ is true for all $f\in M$. Then for arbitrary $g\in X^*$ and $\varepsilon>0$ there is an $f\in M$ such that $\|f-g\|<\varepsilon$. We have $$ |g(u_n)-g(u)|\leq |g(u_n)-f(u_n)|+|f(u_n)-f(u)|+|f(u)-g(u)| \leq \varepsilon \|u_n\|+\varepsilon \|u\| + |f(u_n)-f(u)| \leq 2C\varepsilon+ |f(u_n)-f(u)| \to 2C\varepsilon. $$ Since $\varepsilon>0$ can be arbitrarilly small, the convergence $g(u_n)\to g(u)$ follows. q.e.d.

Since $C_0^\infty(0,1)$ is dense in $L^{p'}$ (and you already use the representation $L^{p'}$ for the dual space of $L^p$) and the sequence $u_n$ is bounded, we can apply this result.