A note on involution prefixes in Coxeter groups

Let \((W,R)\) be a Coxeter system and let \(w\in W\). The paper defines an involution prefix of \(w\) as an involution \(u\) which occurs as an initial segment of a reduced expression for \(w\) and says that \(w\) has the ancestor property if among its involution prefixes there is a unique one of maximal length. This leads to the ancestor conjecture, namely that every finite Coxeter group has the ancestor property and to the associated notion of ancestor decomposition and involution length.\N\NThe main result proves that every Coxeter element of a finitely generated Coxeter group has the ancestor property. More precisely, if \(D(w)\) is the left descent set of a Coxeter element \(w\), then the ancestor of \(w\) is the product of the mutually commuting simple reflections in \(D(w)\). The paper then studies the possible involution lengths of Coxeter elements. It shows that the minimum involution length attained by a Coxeter element is the chromatic number of the Coxeter graph, while the involution length of a given Coxeter element equals the path length determined by the relative order of simple reflections in reduced expressions. As a consequence, the maximum involution length attained by a Coxeter element is the maximum path length of the Coxeter graph.\N\NThe final part reports \textsf{Magma} computations verifying the conjectures for numerous finite Coxeter groups and then discusses the infinite case. In particular, it exhibits the infinite Coxeter group \(G_n=\langle r_1,\dots,r_n \mid r_1^2=\cdots=r_n^2=1\rangle\), where every element has a unique reduced expression (so the ancestor property holds), while the expected finite-type bound on involution length fails: for every \(k\ge 1\), the element \((r_1\cdots r_n)^k\) has involution length \(nk\).

Cites work

This page was built for publication: A note on involution prefixes in Coxeter groups

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6847799)