3
$\begingroup$

Let $\mathcal F$ be a field of subsets of a set $\Omega$. Equip the space $[0,1]^\mathcal F$ of functions from $\mathcal F$ into $[0,1]$ with the product topology. Then, the set $\Delta$ of finitely additive probability measures on $\mathcal F$ is a convex and compact subset of $[0,1]^\mathcal F$.

If $\mu \in \Delta$, define the inner measure $\mu^i: 2^\Omega \to [0,1]$ for $\mu$ by $$\mu^i(A) = \sup \big\{\mu(F): F \subset A, F \in \mathcal F \big\}, \ A \subset \Omega.$$ We can view the inner measure as a mapping $\mu \mapsto \mu^i$ from $\Delta$ into $[0,1]^{2^\Omega}$.

Question. Is the inner measure mapping continuous? That is, if $\mu_{\alpha}$ is a net in $\Delta$ that converges to $\mu$ (i.e. $\mu_\alpha(F) \to \mu(F)$ for all $F \in \mathcal F$), then is it true that $\mu^i_\alpha \to \mu^i$ (i.e. $\mu_\alpha^i(A) \to \mu^i(A)$ for every subset $A$ of $\Omega$)?

A similar question arises for outer measure, though I assume the answers are the same.

$\endgroup$

1 Answer 1

2
$\begingroup$

Let $\Omega = \mathbb{N}$ and let $\mathcal{F}$ be the field consisting of all finite and cofinite subsets of $\mathbb{N}$. Let $\mu_n = \delta_{2n}$ be a point mass at the integer $2n$, and let $\mu$ be the finitely additive measure that assigns measure $0$ to every finite set and $1$ to every cofinite set. Then $\mu_n(F) \to \mu(F)$ for every $F \in \mathcal{F}$.

Let $A \subset \mathbb{N}$ be the set of all even integers. Then we have $\mu_n^i(A) = 1$ for every $n$ but $\mu^i(A) = 0$.

$\endgroup$
1
  • $\begingroup$ Thanks. On the other hand, if $\mathcal F$ is closed under arbitrary unions, then I think the result holds. I wonder if something like that is necessary, though. $\endgroup$
    – aduh
    Mar 29, 2020 at 23:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.