4
$\begingroup$

Let $H$ be some Polish group and $X$ some standard Borel space. Assume that $H$ acts measurably on $X$, i.e. $(h,x) \mapsto hx$ is Borel. Let $H_0 \subset H$ and $A \subset X$ be some Borel sets. Is it true that the set of all $x \in X$ such that $H_0 x \subset A$ is Borel?

The following theorem of descriptive set theory is related to the question: If $A$ is some Borel subset of $P \times X$, with $P$ a Polish space and $X$ a standard Borel space, such that, for any $x \in X$, the $x$-fiber $A_x=\{p \in P ; (p,x) \in A\}$ is a $K_\sigma$ (i.e. a countable union of compact sets), then $\pi_X(A)$ is Borel, with $\pi_X : P \times X$ the projection onto $X$.

$\endgroup$

1 Answer 1

4
$\begingroup$

No, it is not true.

It suffices to show that the complement is not necessarily Borel. Let $B=X\setminus A$, which is a general Borel set. We have $$ \{x:H_0x\not\subseteq A\} = \{x:H_0x\cap B\ne\varnothing\} = \{x: (\exists h\in H_0)(hx\in B)\} = \{x: x\in H_0^{-1}B\} = H_0^{-1}B $$ Let $G=H_0^{-1}$ which is again a general Borel set since $G^{-1}=H_0$. Then we want to determine whether $GB$ must be Borel. But let the group be addition on $\mathbb R$: Erdös and Stone showed that the sum of Borel sets is not necessarily Borel.

$\endgroup$

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.