I'm reading Real Analysis by Royden 4th Edition.

The entire problem statement is:

Let $\{f_n\}_{n=1}^\infty$ be a sequence of integrable functions on $E$ for which $f_n\to f$ pointwise a.e. on $E$ and $f$ is integrable over $E$. Show that $\int_E |f_n-f|\to0 \iff \lim\limits_{n\to\infty}\int_E|f_n|=\int_E|f|.$

My attempt at the proof is:

$(\Longrightarrow)$ Suppose $\int_E|f_n-f|\to0$ and let $\varepsilon>0$ be given. Then there exists an $N>0$ such that if $n\geq N$ then $|\int_E|f_n-f||<\varepsilon.$ Consider $$|\int_E|f_n|-\int_E|f||=|\int_E(|f_n|-|f|)|\leq|\int_E|f_n-f||<\varepsilon.$$ Thus, $\int_E|f_n|\to\int_E|f|.$

$(\Longleftarrow)$ Suppose now that $\int_E|f_n|\to\int_E|f|.$ Let $h_n=|f_n-f|$ and $g_n=|f_n|+|f|$. Then $h_n\to0$ pointwise a.e. on $E$ and $g_n\to2|f|$ pointwise a.e. on $E$. Moreover, since each $f_n$ and $f$ are integrable $\int_E g_n=\int_E|f_n|+|f|\to2\int_E|f|.$ Thus, by the General Lebesgue Dominated Convergence Theorem, $\int_E|f_n-f|\to\int_E0=0.$

I'm pretty sure I got this one down, but I was wondering if it was okay for $g_n$ to depend on $f$ or $f_n$ or does it need to be independent of them?

Thanks for any help or feedback!

  • $\begingroup$ No. That is fine. What you are using is a more general (but equivalent) version of the Dominated Convergence theorem. $\endgroup$ – Prahlad Vaidyanathan Nov 23 '13 at 7:55
  • $\begingroup$ Sounds good, thanks for checking that for me! $\endgroup$ – Shant Danielian Nov 23 '13 at 7:57

As Prahlad Vaidyanathan said, the proof is correct. I would probably not use $\varepsilon$ in the first part, instead writing

$$ \left|\int_E|f_n| - \int_E|f| \right| = \int_E \big||f_n|-|f|\big|\le \int_E |f_n-f|\to 0 $$ and invoking the squeeze lemma.


Fatou's Lemma is your friend. By Fatou,

\begin{align*} \int_{E} 2|f| &= \int_{E} \liminf_{n\to\infty} (|f| + |f_n| - |f-f_n|) \\ &\leq \liminf_{n\to\infty} \int_{E} (|f| + |f_n| - |f-f_n|) \\ &= 2\int_{E} |f| - \limsup_{n\to\infty} \int_{E} |f-f_n|. \end{align*}

So it follows that $\limsup_{n\to\infty} \int_{E} |f-f_n| = 0$ and the desired conclusion follows. You may also want to give a look on Scheffé's lemma.

  • $\begingroup$ I'm a bit of noob when it comes to using limit inferiors/superior, so it looks like you're using a duality between lim sup/lim inf in the third line? Also, doesn't lim inf satisfy a super additivity property? $\endgroup$ – user135520 Aug 1 '20 at 18:35

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.