Characteristic measures for language stable subshifts (Q6097047)

Given a homeomorphism \(T\) on a compact metrizable space \(X\), the automorphism group \(\operatorname{Aut}(X,T)\) is given by the set of all homeomorphisms of \(X\) which commute with \(T\). A Borel measure \(\mu\) on \(X\) is called characteristic if it \(\operatorname{Aut}(X,T)\)-invariant. A few years ago, \textit{J. Frisch} and \textit{O. Tamuz} [Ergodic Theory Dyn. Syst. 42, No. 5, 1655--1661 (2022; Zbl 1501.37014)] asked whether it is true that every \(\mathbb{Z}\)-subshift admits a characteristic measure. We know of four classes of subshifts which admit a characteristic measure for relatively simple reasons, but these classes are far from covering the space of all \(\mathbb{Z}\)-subshifts. We remark that in general the automorphism group of a subshift is non-amenable, so the existence of characteristic measures is not clear at all. The contribution of this article is to define a new class of subshifts which admit characteristic measures. These are called the ``language stable'' subshifts. Essentially, these are the subshifts which can be approximated by subshifts of finite type in such a way that a particular approximation is the best possible for arbitrarily long gaps, infinitely often (for readers familiar with ``first offender words'', a subshift is language stable if the set of lengths for which there are no first offenders has Banach upper density 1). The proof that language stable subshifts admit characteristic measures is short and clear, and most of the article is devoted to constructing examples that are language stable but do not fall into the previously known classes that admit characteristic measures. I think this article provides a very interesting class of examples and I will be looking forward to hearing more about language stable subshifts and characteristic measures in the future.