1
$\begingroup$

Statement: If $g$ is any $\Bbb{C}$-valued measurable function on a measure space $(X, \mu)$ such that $$ \int fg d\mu = 0 $$ for all $f\in L^1(\mu)$ then $g=0$ almost everywhere.

Proof. If $\mu$ is $\sigma$-finite then the statement follows from my earlier post; since the condition on $g$ implies that $$ \int_E g d\mu = \int \chi_E g d\mu = 0 $$ for all $\mu(E)<\infty$.

But I'm not sure how to (dis)prove the last "statement" for non-$\sigma$-finite $\mu$. Any help is appreciated. Thanks.

$\endgroup$
1
  • $\begingroup$ why are you asking about non-$\sigma$-finite measures, when your initial question has the standard measure on $\Bbb R$, which is $\sigma$-finite? It is unclear what your actual question is. $\endgroup$
    – supinf
    Commented Jun 18, 2021 at 9:25

2 Answers 2

3
$\begingroup$

You have missed $\sigma$ finiteness of the measure in formulating your result. There exist measures $\mu$ for which the only integrable functions are $0$ a.e. In this case your new statement is clearly false. [Take $\mu (E)=\infty$ for any non-empty $E$].

$\endgroup$
3
  • $\begingroup$ ok I understood that the $\mu$ you defined gave the counterexample for my new result (in the non sigma finite case) since: that is non sigma finite and 0 is the only integrable functions so can choose for the counterexample $g$ as any measurable function that is not $0$ a.e. ..Right? But the new result above is true if I add "sigma-finiteness" in the statement (as I have proved that above) without the assumption of "boundedness" on $g$! Isn't that? but you have mentioned "bddness of $g$ and sigma-finiteness"! $\endgroup$
    – sigma
    Commented Jun 18, 2021 at 9:44
  • $\begingroup$ You are right. Boundeness of $g$ is not necessary. @indrajit $\endgroup$ Commented Jun 18, 2021 at 9:48
  • $\begingroup$ Thank you so much $\endgroup$
    – sigma
    Commented Jun 18, 2021 at 9:49
1
$\begingroup$

If $X$ is an uncountable set, $\mathcal E$ is the $\sigma$-algebra of the countable-or-cocountable subsets of $X$ and $\mu(A)=\begin{cases}0&\text{if }\lvert A\rvert\le \aleph_0\\ \infty&\text{if }\lvert X\setminus A\rvert\le\aleph_0\end{cases}$, then every $L^1$ function is $0$ almost-everywhere, and therefore $\int fg\,d\mu=0$ for all measurable (or even non-measurable) $f$ and for all $g\in L^1$.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .