Given a subset $E\subseteq\mathbb{R}$, with the length of a closed Interval $I = [a,b] $ given by $l(I)=b - a$, the Lebesgue outer measure $\lambda^*(E)$ is defined as

$\lambda^*(E) = \operatorname{inf} \{ \sum_{k=1}^\infty l(I_k) : {(I_k)_{k \in \mathbb N}} \text{ is a sequence of closed intervals with } E \subset \bigcup_{k=1}^\infty I_k \}$.

One might define a similar notion (I'll call it $\beta$) by $\beta(E) = \operatorname{sup} \{\sum_{k=1}^\infty l(I_k) : {(I_k)_{k \in \mathbb N}} \text{ is a sequence of closed intervals with} \bigcup_{k=1}^\infty I_k \subset E \}$. I would have thought this would be called "Lebesgue inner measure" but that seems to mean something different.

1) This notion $\beta$ is conspicuously missing from textbooks. Why?

My guess is that maybe $\beta(I)> l(I)$ for some closed interval $I$? I say that because the proof that that can't happen with Lebesgue outer measure (i.e. that $l^*(I)=l(I)$) uses the Heine-Borel Theorem which requires that the collection of closed intervals be a cover of $I$.

2) Are there nice sufficient conditions for $\beta(E)$ to be equal to the Lebesgue outer measure $l^*(E)$? For example, are they equal if E is Lebesgue measurable?


Because then $\beta({\bf{Q}}^{c}\cap[0,1])=-\infty$ but the inner measure of ${\bf{Q}}^{c}\cap[0,1]$ coincides with the outer measure, which the value is $1$.

We should take $\sup\emptyset=-\infty$.


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.