6
$\begingroup$

$\newcommand\Sn{\mathit{Sn}}$A subset $A$ of a group $X$ is called algebraic if $A=\{x\in X: a_0xa_1x\dotsm xa_n=1\}$ for some elements $a_0,a_1,\dotsc,a_n\in X$.

Let $\mathcal A_X$ be the family of all algebraic sets in $X$.

Definition. The Steinhaus number $\Sn(X)$ of an infinite group $X$ is the largest cardinal $\kappa$ such that for any cover $\mathcal C\subset\mathcal A_X$ of $X$ with cardinality $\lvert\mathcal C\rvert<\kappa$ there exists a set $C\in \mathcal C$ such that $FCC^{-1}F=X$ for some finite set $F\subseteq X$.

By the famous Steinhaus–Weil Theorem, for any closed set $F$ of positive Haar measure in a compact topological group $X$ the set $FF^{-1}$ is a neighborhood of the identity in $X$.

This theorem implies that each compact topological group has $$\operatorname{cov}(\overline{\mathcal N}_X)\le \Sn(X)\le \lvert X\rvert$$ where $\operatorname{cov}(\overline{\mathcal N}_X)$ is the smallest cardinality of a cover of $X$ by closed sets of Haar measure zero.

Under Martin's Axiom, $\Sn(X)=\mathfrak c$ for each infinite compact Polish group.

Problem. Is $\Sn(X)=\mathfrak c$ for any infinite compact Polish group in ZFC?

Remark. The answer is affirmative for commutative groups. This follows from the observation that each algebraic set in a commutative group is a coset of a subgroup $\{x\in X: x^n=1\}$.

$\endgroup$
4
  • 1
    $\begingroup$ What is $F$ in the definition? $\endgroup$ Commented Aug 18, 2022 at 23:12
  • $\begingroup$ @IlyaBogdanov Thank you for the comment. $F$ is a finite set. I have added this missing part to the definition. $\endgroup$ Commented Aug 18, 2022 at 23:46
  • 3
    $\begingroup$ Could you give at least some hints for why your two assertions (the inequality and the equality) are true? I am asking because I do not see at all how to use topology and/or measure to assert something about a quantity whose definition does not involve anything but the group structure. $\endgroup$ Commented Aug 19, 2022 at 7:09
  • 2
    $\begingroup$ @მამუკაჯიბლაძე Sorry that I have not answered you comment earlier: I was in the village of my father (without computer, only with mobile phone, which does not have all functions, in particular, it does not copy you name in Georgian). The connection of the algebraic sets to topology and measure follows from the easy observation that alegbaric sets are closed in any $T_1$ semigroup topology. $\endgroup$ Commented Aug 20, 2022 at 18:44

1 Answer 1

1
$\begingroup$

The answer to this problem is negative: For the compact Polish group $X=S_3^\omega$ we have $Sn(X)\le\mathfrak r$ where $$\mathfrak r=\min\{|\mathcal R|:\mathcal R\subseteq [\omega]^\omega\;\wedge\;\forall f\in 2^\omega\;\;\exists R\in\mathcal R\;\;|f[R]|=1\}.$$ By induction it can be shown $\mathfrak r=\min\{|\mathcal R|:\mathcal R\subseteq[\omega]^\omega\;\wedge\;\forall f\in n^\omega\;\exists R\in\mathcal R\;\;|f[R]|=1\}$ for every $n\ge 2$. So, we can find a family $\mathcal R\subseteq[\omega]^\omega$ of cardinality $|\mathcal R|=\mathfrak r$ such that for every function $f:\omega\to S_3$ there exists $R\in\mathcal R$ such that $|f[R]|=1$. Then $$S_3^\omega=\bigcup_{R\in\mathcal R}\bigcup_{g\in S_3}A_{R,g},\quad\mbox{where $A_{R,g}=\{f\in S_3^\omega:\{g\}=f[R]\}$}.$$ Observe that for every $R\in\mathcal R$ and $g\in S_3$ the set $A_{R,g}A_{R,g}^{-1}=A_{R,e}$ is nowhere dense in $S_3^\omega$ and hence $FA_{R,g}A_{R,g}^{-1}F\ne S_3^\omega$ for any finite set $F\subseteq S_3^\omega$.

It remains to show that for every $R\in\mathcal R$ and $g\in S_3$, the set $A_{R,g}$ is algebraic in $S_3^\omega$.

Let $d$ and $t$ be elements of order $2$ and $3$ in the symmetric group $S_3$. It is easy to check that $S_3=\{x\in S_3:dxxdxx=e\}$ where $e$ is the identity of the group $S_3$. On the other hand, $\{x\in S_3:xtxxtx=t^2\}=\{e\}$ and hence $\{x\in S_3:g^{-1}xtg^{-1}xg^{-1}xtg^{-1}x=t^2\}=\{g\}$.

Then $A_{R,g}=\{x\in S_3^\omega:a_0xa_1xa_2xa_3x=b\}$ where $$a_0(i)=a_2(i)=\begin{cases}g^{-1}&\mbox{if $i\in R$};\\ d&\mbox{if $i\in\omega\setminus R$}, \end{cases}\quad a_1(i)=a_3(i)=\begin{cases}tg^{-1}&\mbox{if $i\in R$};\\ e&\mbox{if $i\in\omega\setminus R$}, \end{cases} $$ and $$ b(i)=\begin{cases}t^2&\mbox{if $i\in R$};\\ e&\mbox{if $i\in\omega\setminus R$}, \end{cases} $$

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .