Talk:Scheffé's lemma

A little remark

Maybe it should also be interesting to note that if $\int |f_{n}-f|\,d\mu \to 0$ then the hypothesis that $f$ is integrable can be dropped, since in this case there exists an $m\in \mathbb {N}$ such that $\int |f_{m}-f|\,d\mu <\varepsilon <+\infty$. Thus, $f=f_{m}-(f_{m}-f)\in L^{1}$ since $L^{1}$ is a vector space.

Misleading reference

I removed `<ref name=Williams>{{cite book|author=David Williams|title=Probability with Martingales|publisher=Cambridge University Press|location=New York|year=1991|ref=harv|page=55}}</ref>' from the `Applications' section because the citation makes no mention of convergence of densities. Page 55 in Williams is nothing more than Scheffé's lemma. The application is an easy exercise using Scheffé's lemma but of course it would be nice to have a reference. 130.60.188.214 (talk) 08:34, 7 August 2015 (UTC)