2
$\begingroup$

Problem. Let $\lambda$ be the standard product measure on the Hilbert cube $[-\frac12,\frac12]^\omega$ and $A,B$ be two $\lambda$-positive Borel subsets of $[-\frac12,\frac12]^\omega$. Is it true that the sum-set $A+B=\{a+b:a\in A,\;b\in B\}\subset [-1,1]^\omega$ contains a homothetic copy $s+\varepsilon\cdot [-1,1]^\omega$ of the Hilbert cube for some $s\in[-1,1]^\omega$ and $\varepsilon>0$?

Remark. It can be shown that for any sequence $(\varepsilon_n)_{n\in\omega}\in\ell_1$ the sum-set $A+B$ contains a cube $s+\prod_{n\in\omega}[-\delta_n,\delta_n]$ with $\delta_n=|\varepsilon_n|$ for all sufficiently large $n$.

But this is rather far from the required $\varepsilon_n=\varepsilon$ for all $n\in\omega$.

By the way, is the restriction $(\varepsilon_n)\in\ell_1$ the best possible, or it can be improved to, say, $(\varepsilon_n^2)\in\ell_2$?

$\endgroup$
6
  • $\begingroup$ Is there an example of a subset with positive measure that does not contain such a tiny Hilbert cube ? $\endgroup$ May 19, 2018 at 13:55
  • 1
    $\begingroup$ @HenrikRüping Yes: each set of positive measure contains a zero-dimensional compact set positive measure. This zero-dimensional set cannot contain a copy of the Hilbert cube. The existence of a zero-dimensional set of positive measure follows from the observation that the space contains many spheres of zero measures (by the $\sigma$-additivity of the measure). $\endgroup$ May 19, 2018 at 14:08
  • 1
    $\begingroup$ Obviously not. Take any square summable but not summable sequence $a_k$ and consider the set $A$ (of full measure) on which $\sum_k a_k(x_k-\frac 12)$ converges. No matter how many times you add it to itself, there won't be any $\varepsilon$-cube in it. $\endgroup$
    – fedja
    May 19, 2018 at 17:09
  • $\begingroup$ @fedja Thank you for the comment. Could you explain why your set $A$ has full measure? $\endgroup$ May 19, 2018 at 20:36
  • 1
    $\begingroup$ Erm... The sum of independent symmetric random variables with convergent sum of variances converges almost surely, doesn't it? $\endgroup$
    – fedja
    May 20, 2018 at 0:05

0

Your Answer

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

Browse other questions tagged or ask your own question.