2
$\begingroup$

Let $(X, \mathcal{A}, \mu)$ be a measurable space, $(f_n)$ a sequence of integrable functions that converges in measure to some integrable function $f$. Assume further that there is some $g : X \rightarrow [0,\infty]$ integrable such that $|f_n|\leq g$ and $|f|\leq g$. Then $(f_n) \rightarrow f$ in mean.

I am not sure how to prove this. Since $(f_n)\rightarrow f$ in measure, every subsequence of $(f_n)$ admits some subsequence that converges to $f$ almost everywhere. I can prove that such a subsequence converges in measure, since convergence a.e plus being dominated by an integrable function as above implies convergence in mean.

If $(f_n)\rightarrow f$ a.e, then $(f_n-f)\rightarrow f$ a.e. Also, $$|f_n - f|\leq |f_n| + |f| \leq 2g$$ and therefore appealing to the Dominated Convergence Theorem we obtain that $\int_X |f_n -f|d\mu \rightarrow 0$ so $(f_n)\rightarrow f$ in mean.

The problem is then in showing that if a subsequence of a sequence $(f_n)$ converges in mean, then $(f_n) \rightarrow f$ in mean, and this is the part I need some help with. Thank you!

$\endgroup$
  • 1
    $\begingroup$ @MariosGretsas Absolutely! I've edited the statement. Thank you! $\endgroup$ – user313212 Sep 5 '17 at 10:44
  • $\begingroup$ If $(f_n)$ does not converge in mean to $f$, there is a subsequence, $(f_{n_k})_k$ and an $\epsilon>0$ so that $\Vert f_{n_k}-f\Vert_k >\epsilon$ for all $k$. $\endgroup$ – David Mitra Sep 5 '17 at 11:10
1
$\begingroup$

Assume that your statement is not true.

Thus exists an $\epsilon_0>0$ and a subsequence $f_{n_k}$ of $f_n$ such that $\int_X|f_{n_k}-f| \geq \epsilon_0,\forall k\geq 1$.

Since $f_{n_k}$ converges in measure then exists a subsequence $f_{n_{k_s}}$ of $f_{n_k}$ which converges to $f$ a.e

In other words $f_{n_{k_s}}-f \to 0$ a.e

We have that $|f_{n_{k_s}}| \leq g$ a.e thus $|f| \leq g$ a.e

Now from Lebesgue's Dominated Convergence theorem we have that $$\lim_{s \to +\infty}\int_X|f_{n_{k_s}}-f|=0$$ which is a contradiction.

$\endgroup$

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.