2
$\begingroup$

If a real measurable function $f : E \rightarrow \mathbb{R}\ \cup \{\infty\}$ is finite almost everywhere (i.e. $f^{-1}(\{\infty\})$ has measure zero) and $E \subset \mathbb{R}$ is of finite Lebesgue measure, then $\forall \varepsilon > 0\ \exists\ F \subset E$ with $|E \setminus F| \leq \varepsilon$ and $f$ bounded in $F$.

I supposed that no such $F$ exists, and I tried to prove that $f = \infty$ in a set with positive measure, but without success. Any help? :)

$\endgroup$
  • 1
    $\begingroup$ What does "finite almost everywhere" mean here? Are you assuming that $f:\mathbb R\to\mathbb R\cup\{\infty\}$ and simply saying that $f^{-1}(\{\infty\})$ is null? $\endgroup$ – Henning Makholm May 14 '16 at 18:44
  • $\begingroup$ not null, but of measure zero. I'll edit the question though, I noticed it's not very clear. $\endgroup$ – Guillermo Mosse May 14 '16 at 20:26
  • $\begingroup$ Using the result of this question which I asked for the purpose, we can see that the assumption that $f$ is measurable is crucial. Its answer shows that, assuming the Axiom of Choice, $\mathbb R$ can be partitioned into continuum many sets $(X_t)_{t\in\mathbb R}$ such that every $X_t$ intersects every set of positive measure. Then let $f(x)=$the $t$ such that $x\in X_t$, and then $f$ is not bounded in any $F$. $\endgroup$ – Henning Makholm May 15 '16 at 14:56
2
$\begingroup$

For each $n\in\mathbb{N}$, define $F_n=\{x\in E:|f(x)|> n\}$. Then $F_1\supset F_2\supset\cdots$ and $$\bigcap_{n=1}^{\infty}F_n=\{x\in E:|f(x)|=\infty\}$$

Since $E$ has finite measure, we can apply continuity from above to conclude that $$ 0=|\{x\in E:|f(x)|=\infty\}|=\lim_{n\to\infty}|F_n|$$ Therefore given $\varepsilon>0$ there exists $n$ such that $|F_n|<\varepsilon$, so the set $F=E\setminus F_n$ has the desired properties.

$\endgroup$
  • $\begingroup$ How do you know $|F_n|$ exists? It doesn't seem to be assumed that $f$ is measurable. $\endgroup$ – Henning Makholm May 14 '16 at 19:29
  • 1
    $\begingroup$ I suppose I assumed $f$ is measurable, because it's not clear how to interpret the statement that $f$ is finite a.e. otherwise. $\endgroup$ – carmichael561 May 14 '16 at 19:35
  • 1
    $\begingroup$ Even if $f$ is not measurable, it still seems to make sense to ask whether $f^{-1}(\{\infty\})$ happens to be a measurable set with measure $0$. $\endgroup$ – Henning Makholm May 14 '16 at 19:37
  • 1
    $\begingroup$ I guess. While OP should clarify, I'm reasonably confident that $f$ is measurable. $\endgroup$ – carmichael561 May 14 '16 at 19:38
  • 1
    $\begingroup$ @SeñorBilly: Non-measurable functions are elusive things, difficult to construct. One needs to use the Axiom of Choice to show that they even exist. $\endgroup$ – Henning Makholm May 14 '16 at 20:53

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.