Gate lattices and the stabilized automorphism group (Q6051259)

Let \(G\) be a countable group, let \(\Sigma\) be a finite alphabet, and consider the Cantor set \(\Sigma^G\) with its natural \(G\)-action by pre-translation. A \textit{subshift} is a closed, \(G\)-invariant subset \(X\subseteq\Sigma^G\). It is of \textit{finite type} (SFT) if it is the intersection of \(G\)-translates of finitely many clopens. A classical object of study is the automorphism group \(\mathrm{Aut}(X)\) consisting of all self-homeomorphisms of \(X\) that commute with the \(G\)-action; much is known about it, but much more remains unknown or conjectural. Classically one considers \(G=\mathbb Z\); a SFT may then be defined by a graph with edge set \(\Sigma\), the elements of \(X\) being naturally in bijection with bi-infinite paths in the graph. The group \(\mathrm{Aut}(X)\) is usually not simple, for instance because of Krieger's \textit{dimension representation} on the finite-rank abelian group \(\mathbb Z^\Sigma\overset M\to\mathbb Z^\Sigma\overset M\to\cdots\) with \(M\) the adjacency matrix of the graph. The subgroup of \textit{inert} automorphisms is the kernel of this representation. The paper considers a larger, better-behaved group, the \textit{stabilized automorphism group} \(\mathrm{SAut}(X)\) consisting of all self-homeomorphisms \(\phi\) of \(X\) whose centralizer (i.e., those elements of \(G\) that commute with \(\phi\)) has finite index in \(G\). A \textit{gate} on \(X\) is a permutation of finitely many clopens; equivalently, a self-homeomorphism \(\chi\) of \(X\) with \(\chi(x)_g=x_g\) for all \(g\in G\setminus F\) for a finite subset \(F\). If \(G\) is residually finite, there will then exist a finite-index subgroup \(H\le G\) such that all \(H\)-translates of \(F\) are disjoint, so all \(\chi^h\) are disjoint and therefore commute; the associated \textit{gate lattice} is the product \(\chi^H\) of all \(\chi^h\) for \(h\in H\), and evidently belongs to \(\mathrm{SAut}(X)\). Let \(\mathfrak L(X)\) and \(\widehat{\mathfrak L}(X)\) denote respectively the subgroups of \(\mathrm{SAut}(X)\) generated by gate lattices and \textit{even} gate lattices \(\chi^H\) with \(\chi\) an even permutation of clopens. The main results of the author concern subshifts on residually finite groups which afford some amount of mixing and gluing; more precisely, a subshift has the \textit{eventual filling property} (see Definition~9) if for all finite \(F\subseteq G\) there exists a finite \(N\subseteq G\) with the following property: given \(x,y\in X\) there exists \(z\in X\) which agrees with \(x\) on \(F\) and with \(y\) on \(G\setminus N\). Theorem. Consider a residually finite countably infinite group \(G\) and an SFT \(X\subseteq\Sigma^G\) with the eventual filling property. Then \(\widehat{\mathfrak L}(X)\) is simple, coincides with the commutator subgroup of \(\mathfrak L(X)\), and is the unique minimal normal subgroup of \(\mathrm{SAut}(X)\), aka its \textit{monolith}. The next statement (Theorem 4 in the paper) gives additional assumptions (e.g., \(G=\mathbb Z\)) that guarantee \(\widehat{\mathfrak L}(X)=\mathfrak L(X)\), while Theorem 5 shows that, when \(G=\mathbb Z\), the group of stabilized inert automorphism coincides with \(\mathfrak L(X)\). In particular, if \(X=\Sigma^{\mathbb Z}\), this shows that the group of stabilized inert automorphisms is simple, recovering a result by \textit{Y. Hartman} et al. [Int. Math. Res. Not. 2022, No. 21, 17112--17186 (2022; Zbl 1510.37023)]. As is often the case, the proof of the more general statement is shorter and easier to follow. In the absence of a standard definition of inertness for subshifts on a general group \(G\), it seems therefore natural to \textit{define} the group of stabilized inert automorphisms of \(X\) as \(\mathfrak L(X)'\), and its group of inert automorphisms as \(\mathfrak L(X)'\cap\mathrm{Aut}(X)\).