Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

We have functions $f_n\in L^2$ such that $f_n(x)$ tends to some $f(x)$ for almost all $x$, suppose that $||f_n||<M$. Does this mean that $f_n$ tend to $f$ weakly? In my previous question about $L^1$ I realized that the weak convergence fails; can one describe spaces for which this is true?

share|cite|improve this question
Every closed and bounded set is weakly relatively compact in $L^2$ hence every bounded sequence has a weakly convergent subsequence. If a sequence converges strongly it converges weakly as well (the whole sequence). – Jonas Teuwen May 18 '11 at 14:09
For the strong convergence this fails as well as for $L^1$. – peno May 18 '11 at 14:12
I want convergence for the whole sequence, not for a subsequence. – peno May 18 '11 at 14:18
your accounts have been merged. Please register your account so that you don't have login problems in the future and don't need to make new accounts. – Qiaochu Yuan May 19 '11 at 6:49
up vote 4 down vote accepted


By Fatou's lemma we have $f \in L^2$, so by replacing $f_n$ by $f_n - f$ we can assume that $f=0$.

Fix $g \in L^2$ and $\epsilon > 0$. By a standard dominated convergence argument, there is a set $A$ of finite measure with $\int_{A^c} |g|^2 < \epsilon$. There is also a $K$ so large that $\int_{|g| > K} |g|^2 < \epsilon$. Let $B = A \cap \{|g| \le K\}$; then $\lvert\int_{B^c} f_n g \rvert \le M \sqrt{2 \epsilon}$ by Cauchy--Schwarz, for every $n$.

On the other hand, for each $n$ we have $\int_B |f_n g|^2 \le K^2 M$. So $\{f_n g\}$ is an $L^2$-bounded sequence of functions on the finite measure space $B$, hence it is uniformly integrable on $B$. We also have $f_n g \to 0$ almost everywhere, and hence they converge in measure on $B$; by the Vitali convergence theorem $f_n g \to 0$ in $L^1(B)$, so $\int_B f_n g \to 0$. Letting $\epsilon \to 0$ completes the proof.

Edit: With respect to your other question (when can you say that an $L^1$-bounded, a.e. convergent sequence must converge weakly in $L^1$), the answer is: only in trivial cases. Essentially, if the measure space is anything but a finite union of atoms, you can construct a sequence of nonnegative $f_n \in L^1$ with $f_n \to 0$ a.e. but $||f_n||_{L^1} = 1$, so that taking $g=1 \in L^\infty$, you have $\int f_n g = 1$ for all $n$. (Idea: let $f_n = \mu(A_n)^{-1} 1_{A_n}$ where either $\mu(A_n) \uparrow \infty$, or $\mu(A_n) \downarrow 0$ fast (in particular try $\sum \mu(A_n) < \infty$)).

share|cite|improve this answer
Thank you so much! – peno May 18 '11 at 19:10
Nice result! Could you state what are you saying only occurs in trivial cases? – user1736 May 18 '11 at 23:58
@user1736: done. – Nate Eldredge May 19 '11 at 1:56

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.