In part of the proof of a problem I am trying to solve I need the following fact (assume that $\mu$ is the Lebesgue measure):

If $E\subseteq \mathbb{R}$ is measurable and $\delta>0$ then there exists open set $U\subseteq \mathbb{R}$, such that $E\subseteq U$ and $\,$ $\delta \mu(U)<\mu(E)$.

I know and have proven the following fact:

Suppose $E \subseteq \mathbb{R}$. Then for each $\epsilon>0$ there exists an open set $U\subseteq \mathbb{R}$ such that $E\subseteq U$ and $\mu(U)< \mu(E)+\epsilon$.

I am pretty sure I can use the second fact to prove the first fact, but I keep getting a value of $\epsilon$ that is in terms of $\mu(U)$, which isn't good because $U$ should depend on $\epsilon$, not the other way around. Some help?

  • 1
    $\begingroup$ As written the problem is trivial: you can take $U=\emptyset$ independently of $E$. Presumably you want some constraint on $U$ in terms of $E$. $\endgroup$ – Ian Mar 19 '17 at 20:33
  • $\begingroup$ @Ian You are absolutely right. I forgot to mention that $E\subseteq U$. I edited my post. Thank you. $\endgroup$ – Ana Mar 19 '17 at 20:39
  • $\begingroup$ @Ian math.stackexchange.com/questions/103306/… I was reading the first solution in this post and the comment that followed. $\endgroup$ – Ana Mar 19 '17 at 20:48
  • 4
    $\begingroup$ Sorry, I got it backwards: your inequality can only hold for $\delta<1$. Then you work backwards: you want $\delta \mu(U)< \mu(E)$, so $\mu(U) < \delta^{-1} \mu(E) = \mu(E) + (\delta^{-1}-1)\mu(E)$. Now take $(\delta^{-1}-1) \mu(E)$ to be your $\epsilon$ $\endgroup$ – Ian Mar 19 '17 at 20:56

Let me expand Ian's comment.

First, the inequality can only hold for $\delta<1$. For example, if $E=[0,1]$ and $\delta\geq1$, there never is such a $U$. But the claim is indeed true for all $\delta\in (0,1)$.

Take any such $\delta$ and work backwards: You want $\delta \mu(U)< \mu(E)$. That is, you want $$\mu(U) < \delta^{-1} \mu(E) = \mu(E) + (\delta^{-1}-1)\mu(E). $$ Now take $\epsilon=(\delta^{-1}-1) \mu(E)$ in your lemma. Then $\mu (U)<\mu (E)+\epsilon $ for some open $U \supset E$. Since this is the estimate you want, $U $ is the set you want.


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.