I am sitting on this multiple-choice question and I cannot answer it, nor say if it is right or wrong:
Given non-negative, Lebesgue-integrable functions $f,f_k\colon E\rightarrow \mathbb{R}^+$ with $\displaystyle\forall x \in E\setminus N: \lim_{k \to \infty}f_k(x)=f(x)$, where $\lambda(N)=0$, $E,N \subset \mathbb{R}^n$ and $\displaystyle\lim_{k \rightarrow \infty}\int_E f_k(x) d\lambda=\int_E f(x) d\lambda$.
Is it always true that $$\lim_{k \rightarrow \infty}\int_E |f-f_k(x)|d\lambda=0 ?$$
I see the striking similarity to Lebesgue's dominated convergence theorem, if one could use $\displaystyle\lim_{k \to \infty}\int_E f_k(x) d\lambda=\int_E f(x) d\lambda$ to find some majorant $g$ for our $f_k$ it would be true, especially it would be true when the $f_k$ converge against $f$ from below.