5
$\begingroup$

Any set of positive Lebesgue measure, have a Lebesgue measurable proper subset with the same measure?

For example: if $A$ is a measurable set, with $m(A) > 0$,

I can say that $x \in A$ (since $A$ is not empty because $m(A) > 0$), then can I say that $B = A\setminus \{x\}$ is a measurable subset of $A$ and $m(B) = m(A)$ (since $m({x}) = 0$)?

I want to use this "claim" to solve a part of a problem that I have to show that if $\int_{B} f = 0$ for any measurable subset $B$ of $A$, then $\int_{A}f = 0 = \int_{B}f$ to eventually show that $f = 0$ a.e on $A$.

Is this claim true?

By the way, I also know the Inner approximation: If $E$ is a measurable, then for any $\epsilon > 0$, exits $F \subset E$ such that $m(E\setminus F) < \epsilon$. and $F$ is closed and the similar one for open sets (Outter approximation).

Did the previous theorem says that $m(E\setminus F) = 0$?

Thank you in advance!

$\endgroup$
  • 2
    $\begingroup$ Firstly, it looks like you're specifically thinking about the Lebesgue measure - is this correct? The property that for all $E$ measurable and for all $\epsilon > 0$, there exists a closed $F \subset E$ such that $m(E \backslash F) < \epsilon$ is the inner regularity property of the Lebesgue measure. $\endgroup$ – Kenny Wong Aug 22 '17 at 13:20
  • 2
    $\begingroup$ Secondly, are you really asking whether any set of positive measure has a proper subset of the same measure? If we're using the Lebesgue measure, then this true because a set with one element is guaranteed to be null - as you pointed out. $\endgroup$ – Kenny Wong Aug 22 '17 at 13:21
  • 1
    $\begingroup$ As for your actual problem, I would personally use the fact that if $f$ is non-negative and $\int_B f = 0$, then $f = 0$ a.e. on $B$. We could then consider taking $B = \{ x : f(x) \geq 0\}$ and $B = \{ x : f(x) \leq 0\}$. $\endgroup$ – Kenny Wong Aug 22 '17 at 13:23
  • 1
    $\begingroup$ Yes, for the reason you pointed out. The only set that doesn't contain a point is the empty set. And the empty set is null. $\endgroup$ – Kenny Wong Aug 22 '17 at 13:26
  • 2
    $\begingroup$ But I'm afraid I'm missing something... How does this claim help with showing that $\int_B f = 0$ for all measurable $B \subset A$ implies that $f = 0$ a.e. on $A$? $\endgroup$ – Kenny Wong Aug 22 '17 at 13:27
3
$\begingroup$

Singletons (one-member subsets) are Lebesgue sets of measure $0.$ The complement $D^c$ of a Lebesgue set $D$ is a Lebesgue set. If $D,E$ are Lebesgue sets then so are $D\cap E$ and $D\cup E.$ If $D,E$ are disjoint Lebesgue sets then $M(D\cup E)=m(D)+m(E).$

If $a\in A$ where $A$ is $any$ non-empty Lebesgue set then $B=A$ \ $\{a\}$ is a proper subset of $A$. And $B= A\cap (\{a\})^c$ is a Lebesgue set which is disjoint from $\{a\}.$ So $$m(A)=m(B\cup \{a\})=m(B)+m(\{a\})=m(B)+0=m(B).$$

$\endgroup$
  • 1
    $\begingroup$ Countable sets are Lebesgue sets of measure $0.$ So if $A$ is a Lebesgue set of positive measure then $A$ is uncountable and we can remove any countable subset from $A$ and still have a Lebesgue set with the same measure. $\endgroup$ – DanielWainfleet Aug 23 '17 at 4:32
1
$\begingroup$

Yes, your claim is true.

However, you do not need that specific property in your application: it is not specified that $B$ is a proper subset of $A$, therefore chosing $B=A$ is allowed.

$\endgroup$
  • $\begingroup$ I was thinking about that, but this is a qualifying problem and I think they make a mistake because that one will be trivial. lol $\endgroup$ – Richard Clare Aug 22 '17 at 13:36
  • $\begingroup$ This will mean that any set with a possitive Lebesgue measure, has an uncountable number of singleton? $\endgroup$ – Richard Clare Aug 22 '17 at 13:41
  • 1
    $\begingroup$ yes, because it is an uncountable set $\endgroup$ – supinf Aug 22 '17 at 13:49
1
$\begingroup$

Let it be that $A$ is measurable and $m(B)=0$ for every measurable set $B\subseteq A$.

Then evidently $\int g1_A=0$ for every step function $g$.

Consequently $\int f1_A=0$ for every nonnegative measurable function $f$.

This because $\int f1_A$ is by definition the suprememum of the values $\int g1_A$ where $g$ is a step function with $g\leq f$.

This can be expanded to measurable functions $f$.

$\endgroup$

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.