Finite generators for countable group actions in the Borel and Baire category settings (Q475274): Difference between revisions

In the setting of a continuous action of a countable discrete group \(G\) on a Polish space \(X\) a countable partition \(\xi=\{A_j\mid j\in J\}\) (where \(J\) is a countably infinite or a finite set) into Borel sets is said to be a generator if \(\{gA_j\mid g\in G,j\in J\}\) generates the Borel \(\sigma\)-algebra on \(X\). For many groups -- \(\mathbb{Z}\) in particular -- if there is a \(G\)-invariant probability measure then the Kolmogorov--Sinaĭ theorem shows that the entropy of the action with respect to the probability measure is no more than the entropy of a generator, so in particular infinite entropy of the action precludes the existence of a finite generator. B. Weiss raised the question of whether the non-existence of any invariant probability measure ensures the existence of a finite generator. Here this is partially answered in a striking way, showing that there is a generator with \(32\) elements in this situation for any countable group \(G\) and \(\sigma\)-compact \(X\). For aperiodic actions it is shown that there is a generator with four elements on a comeagre set, answering a question of Kechris. These ideas are developed further, showing in particular that any aperiodic Borel action of a countable group \(G\) on a Polish space has a \(G\)-equivariant Borel map from the action to the aperiodic part of the natural shift action of \(G\) on \(2^G\). A particularly interesting dichotomy is obtained for the case \(G=\mathbb{Z}\) -- conditional on a positive answer to the question of B. Weiss without a measure. That is, Weiss asked if a Borel \(\mathbb{Z}\)-space does not admit any invariant Borel probability measure, must it admit a finite generator? Assuming a positive answer to this question, it is shown here that an aperiodic Borel \(\mathbb{Z}\)-space either admits an invariant Borel probability measure with infinite entropy or admits a finite generator.