Let $1<p<\infty$ and $\{f_n\}\in L^p(\mathbb{R})$ be a uniformly bounded sequence, i.e. $\|f_n\|_p\le M, \forall n$, for some $M>0$. If $f_n \to f$ a.e. , prove that $f_n \to f$ weakly in $L^p(\mathbb{R})$ .

My attempt :

Let $A$ be any measurable subset $A \subset\mathbb{R}$, by lemma 2 we have $\lim \int_A f_n = \int_A f$ and using lemma 1 we have $f_n \to f$ weakly in $L^p(\mathbb{R})$.

lemma 1: if $1\le p<\infty$ and $\{f_n\}\in L^p(\mathbb{R})$ be a bounded sequence, then $f_n \to f$ weakly in $L^p(\mathbb{R})$ iff for every measurable subset $A \subset\mathbb{R}$, $$\lim \int_A f_n = \int_A f$$ proof:

WLOG that $m(A)<\infty$ , let $g_0 \in L^q(\mathbb{R})$, $1<q\le\infty$ be any function and denote $g= \chi_{A}$;

\begin{align} \int_{\mathbb{R}} g_0f_n - \int_{\mathbb{R}} g_0f & = \int_{\mathbb{R}} (g_0 -g)(f_n-f) + \int_{\mathbb{R}} g(f_n-f) \\ & \le \|g_0-g\|_q\|f_n-f\|_p + \frac{\epsilon}{2} \\ & \le \frac{\epsilon}{2}+\frac{\epsilon}{2} =\epsilon \implies f_n \rightharpoonup f \end{align}

where I used boundedness in the last inequality above.

lemma 2: if $A$ is measurable and $\{f_n\}\in L^p(\mathbb{R})$ be a bounded sequence, then $\{f_n\}$ is uniformly integrable over $A$. (proof in royden ch.7)


Your attempt is almost good. The only gap is that in general, you cannot make for any function $g_0$ the norm $\lVert g_0-\mathbf 1_A\rVert_q$ as small as you wish. However, we can approximate $g_0$ in $L^q$ by a finite linear combination of indicator functions.


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.