I will first state Fatou's lemma and provide a proof, then I will present the corollary I am trying to prove. I am a little lost on the proof so I to assist in the reader's help I will provide what Folland suggests to do to prove it.

If $\{f_n\}$ is any sequence in $L^{+}$, then $$\int(\lim\inf f_n)\leq \lim\inf\int f_n$$

proof:$$\int \lim_{j\rightarrow \infty} \inf f_j = \int \lim_{k\rightarrow \infty} \inf_{k\geq j}f_k$$ by Monotone Convergence Theorem, then $$= \lim_{k\rightarrow \infty}\int \inf_{k\geq j}f_k$$ observe that $\inf_{k\geq j}f_k \leq f_k$ for all $k\geq j$. So, $$\int \inf_{k\geq j}f_j \leq \int f_k \ \ \ \forall k\geq j$$ $$\leq \inf_{k\geq j}\int f_k \leq \lim_{k\rightarrow \infty}\inf_{k\geq j}\int f_k$$

Corollary - If $\{f_n\}\subset L^{+}$, $f\in L^{+}$, and $f_n\rightarrow f$ a.e., then $\int f\leq \lim \inf \int f_n$

proof (Folland) - If $f_n\rightarrow f$ everywhere, the result is immediate from Fatou's lemma, and this can be achieved by modifying $f_n$ and $f$ on a null set without affecting the integrals by proposition 2.16 which states:

If $f\in L^{+}$, then $\int f = 0$ if and only if $f = 0$ a.e.

I am lost on how to prove the corollary I just need some initial start and I think I can go from there.


2 Answers 2


Or simpler maybe: $\lim_{n}f_{n}=\liminf_{n}f_{n}=f a.e.$ By integrating, $\int \liminf_{n}f_{n}=\int f$. By Fatou's lemma, the left integral is $\leq\liminf_{n}\int f_{n}.$ The result follows immediately.


Let $C=\{x:f_n(x)\to f(x)\}$. Modify $f_n$ and $f$ s.t. $f'=f_n'=0$ on $X\setminus C$ and $f'=f, f_n'=f_n$ on $C$. Then $f_n'\to f'$ everywhere and $f-f'=0, f_n-f_n'=0$ a.e.

$$\therefore \int f=\int f'\leq \liminf \int f_n'=\liminf \int f_n$$

  • $\begingroup$ Why did you let $f' = f_n' = 0$ on $X\setminus C$? $\endgroup$
    – Wolfy
    Nov 9, 2015 at 0:48
  • $\begingroup$ @MorganWeiss In that case $f-f'\in L^{+}$ and $\int f=\int (f-f')+\int f'=\int f'$ by proposition 2.16. $\endgroup$
    – user140541
    Nov 9, 2015 at 2:29
  • $\begingroup$ I don't think I understand, why not let $f' = f'_n = 0$ on $C$? $\endgroup$
    – Wolfy
    Nov 9, 2015 at 2:33
  • $\begingroup$ @MorganWeiss Oh. $C$ is the subset on which $f_n\to f$. So you need to modify functions on its complement... (If you set $f' = f'_n = 0$ on $C$, then $\int f'=0$) $\endgroup$
    – user140541
    Nov 9, 2015 at 2:34
  • $\begingroup$ I see, so since we are modifying the functions then we have to let $f' = f'_n = 0$ be on $X\setminus C$? $\endgroup$
    – Wolfy
    Nov 9, 2015 at 2:36

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.