Invariable generation and the Houghton groups

From MaRDI portal
Publication:2667906



Abstract: The Houghton groups H1,H2,ldots are a family of infinite groups. In 1975 Wiegold showed that H3 was invariably generated (IG) but H1leH3 was not. A natural question is then whether the groups H2,H3,ldots are all IG. Wiegold also ends by saying that, in the examples he had found of an IG group with a subgroup that is not IG, the subgroup was never of finite index. Another natural question is then whether there is a subgroup of finite index in H3 that is not IG. In this note we prove, for each nin2,3,ldots, that Hn and all of its finite index subgroups are IG. The independent work of Minasyan and Goffer-Lazarovich in June 2020 frames this note quite nicely: they showed that an IG group can have a finite index subgroup that is not IG.


A group \(G\) is \textit{invariably generated by a subset \(S\)} if replacing each element of \(S\) with any of its conjugates results in a generating set for \(G\). The group \(G\) is \textit{invariably generated} if such an \(S\) exists. Let us denote by \(d(G)\) the smallest cardinal of a generating set of \(G\). Let \(n\) be a positive integer and let \(X_n = \{1, \dots, n\} \times \mathbb{N}\). The \textit{Houghton group \(H_n\)} is the group of permutations of \(X_n\) that eventually act as a translation in each copy of \(\mathbb{N}\). Wiegold showed that \(H_3\) is invariably generated whereas its subgroup \(H_1\) isn't [\textit{J. Wiegold}, Arch. Math. 29, 571--573 (1977; Zbl 0382.20029)]. From now on, we assume that \(n \ge 2\). The author first proves that \(H_n\) is invariably generated by a subset with \(d(H_n) = \max(2, n - 1)\) elements (Theorem A). Then he proves that every group commensurable to \(H_n\) is invariably generated by a finite subset (Theorem B). Let us describe the invariable finite generating subsets constructed by the author. For \(k \in \{2, \dots ,n\}\), let \(g_k\) be the permutation of \(X_n\) with support \((\{1\} \times \mathbb{N}) \cup (\{k\} \times \mathbb{N})\) which is defined by \(g_k(1, m) = (1, m + 1)\) for \(m \ge 1\), \(g_k(k, 1) = (1 ,1)\) and \(g_k(k, m) = (k, m - 1)\) for \(m \ge 2\). The group $H_2$ is invariably generated by the set $\{ g_2, g_2 \sigma\}$ where $\sigma$ is the $4$-cycle $\left((1, 4)(1, 3)(1, 2)(1, 1)\right)$. If \(n \ge 3\), then \(H_n\) is invariably generated by \(\{g_2, \dots, g_{n - 1}, h\}\) where \(h = g_n g_{n - 1} \cdots g_2\). The proof of Theorem B is the most delicate part of the paper. It relies on the following elementary fact: each subgroup \(H\) of finite index in \(H_n\) contains the subgroup generated by the \(3\)-cycles of \(X_n\) and the permutations \(g_2^{2v}, \dots, g_n^{2v}\) for some \(v = v(H) > 1\). The latter subgroup, named \(U_{2v}\), is easily seen to be normal and of finite index in \(H_n\). The author's key finding is that \(U_{2v}\) is invariably generated by \(\{g_2^{2v}, \dots, g_{n - 1}^{2v}, h_{2v}, \sigma_1, g_2^{4v}\sigma_1, g_2^{4v} \sigma_2\}\) where \(h_{2v} = g_n^{2v} g_{n - 1}^{2v} \cdots g_2^{2v}\), \(\sigma_1\) is the \(3\)-cycle \(\left((1,1) (1,2) (1,3)\right)\) and \(\sigma_2\) is the product \(\left( (1, 1) (1,2)\right) \left( (1, 3) (1,4) \cdots (1, 4v)\right)\).











This page was built for publication: Invariable generation and the Houghton groups

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2667906)