Connectivity of Coxeter group Morse boundaries

Matthew Cordes, Ivan Levcovitz · 2025

Read Connectivity of Coxeter group Morse boundaries by Matthew Cordes, Ivan Levcovitz (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-14085.txt.

Introduction

Matthew Cordes and Ivan Levcovitz ask when the Morse edge of a Coxeter group holds whole. Coxeter groups are mirror groups, coded by a graph that sets them. Their Morse edge pools the ways out where paths act like those in a bent, tree-like space. Two tests on the graph set the end state. Both hinge on how paths shun wide subgroups.

With the round form of the test, plus one end and no affine parts, the edge is joined and joined up close. Fail the plain test, and fail width too, and a one-ended group's edge splits in two. Right-angled groups get a full mark: a joined, non-void Morse edge if and only if the test holds. On the way, Morse paths are marked as those that spend a capped time in cosets of wide subgroups.

Catalog record

CREATORMatthew Cordes, Ivan Levcovitz
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE111 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-14085

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Connectivity of Coxeter group Morse boundaries Matthew Cordes, Ivan Levcovitz Abstract . We study the connectivity of Morse boundaries of Coxeter groups. We define two conditions on the defining graph of a Coxeter group: wide-avoidant and wide-spherical-avoidant. We show that wide-spherical-avoidant, one-ended, affine-free Coxeter groups have connected and locally connected Morse boundaries. On the other hand, one-ended Coxeter groups that are not wide-avoidant and not wide have disconnected Morse boundary. For the right-angled case, we get a full characterization: a one-ended right-angled Coxeter group has connected, non-empty Morse boundary if and only if it is wide-avoidant. Along the way we characterize Morse geodesic rays in affine-free Coxeter groups as those that spend uniformly bounded time in cosets of wide special subgroups. 1. Introduction Boundaries play a fundamental role in the study of (Gromov) hyperbolic spaces and groups. However, there are still many metric spaces and groups which are not hyperbolic that, nonetheless, display some hyperbolic behavior. The Morse boundary was introduced by Charney–Sultan [ CS15 ] and the first author [ Cor17 ] with the goal of identifying and encoding this behavior in a useful way. Roughly, it collects all the geodesics which have hyperbolic-like behavior into a boundary; these geodesics are called Morse geodesics. The key property of this boundary is that it is a quasi-isometry invariant, meaning the Morse boundary of a finitely generated group G is well-defined. Despite the Morse boundary being a fruitful quasi-isometry invariant, the Morse boundaries of many groups are difficult to fully describe. Even basic topological properties like connectedness or local connectedness are hard to show. Coxeter groups are a classical set of groups defined by abstracting properties of reflection groups. In this article, we study the connectivity of Morse boundaries for Coxeter groups. At odds to hyperbolic-like behavior in a group is “flat-like” behavior. Introduced by Behrstock–Drutu–Mosher in [ BDM09 ] , wide groups are a set of groups that have no Morse geodesics, i.e., they have flat-like behavior. Identifying wide subgroups in general can be difficult, but Behrstock–Hagen–Sisto in [ BHS17 ] characterize wide Coxeter groups. In order to understand the Morse directions of Coxeter groups, we study the interplay between their wide special subgroups. This interplay is captured in the notion of a wide-avoidant Coxeter group (Definition 3.3 ), which we introduce in this paper. Loosely, a Coxeter group is wide-avoidant if given any two vertices in its defining graph and a special wide subgroup, then there is a path between these two vertices that avoids the subgraph corresponding to the wide subgroup. Failure of this condition is enough to ensure that the Morse boundary of a Coxeter group is either empty or disconnected: Theorem A . A Coxeter group that is not wide and is not wide-avoidant has disconnected Morse boundary. Moreover, it admits a splitting either over a finite special subgroup or over a special subgroup that is contained in a wide special subgroup. In fact, we conjecture the full characterization: Conjecture 1.1 . A Coxeter group has connected, non-empty Morse boundary if and only if it is one-ended and wide-avoidant. We prove this conjecture for right-angled Coxeter groups and certain classes of Coxeter groups. The right-angled Coxeter characterization being: Theorem B . Let W_{\Gamma} be a right-angled Coxeter group. Then exactly one of the following holds: (1) W_{\Gamma} has empty Morse boundary and is finite or wide, (2) W_{\Gamma} has disconnected Morse boundary and has more than one end (either splits over a finite special subgroup or is virtually \mathbb{Z} ) (3) W_{\Gamma} has connected and locally connected Morse boundary and is wide-avoidant (4) W_{\Gamma} has disconnected Morse boundary, is not wide-avoidant and admits a non-trivial splitting W_{\Gamma}=W_{\Gamma_{1}}\star_{W_{\Delta}}W_{\Gamma_{2}} where W_{\Delta} is an infinite subgroup of a wide special subgroup. We were not able to prove Conjecture 1.1 for all Coxeter groups. However, we give the following sufficient conditions for the connectivity of the Morse boundary. Theorem C . An affine-free, one-ended, wide-spherical-avoidant Coxeter group has connected and locally connected Morse boundary. Recall that a Coxeter group is affine-free if it has no irreducible, special, affine subgroup of rank 3 or larger. We define a Coxeter group to be wide-spherical-avoidant if, loosely, given a certain type of subgraph corresponding to a wide subgroup and a restricted type of spherical subgroup, then there is a path between any two vertices that avoids this subgraph. See Definition 3.5 for a precise definition. Theorem C provides what we believe to be the first examples of non-relatively hyperbolic groups with locally-connected Morse boundary (e.g. Figure 1 ). In fact, the only other known class of non-hyperbolic groups with locally-connected Morse boundary are those quasi-isometric to fundamental groups of finite-volume, cusped hyperbolic n -manifolds [ CCS23 ] [ FKSZ24 ] (which are relatively hyperbolic). Note that the right-angled Coxeter defined on the 1-skeleton of a 3-cube is virtually a finite-volume, cusped hyperbolic 3 -manifold group [ HNT20 ] . \begin{overpic}[non-rel-hyp-graph.pdf] \end{overpic} Figure 1. Defining graph of a non-relatively hyperbolic RACG with locally-connected Morse boundary In proving Theorem A and Theorem C , we drew much of our inspiration from the work of Mihalik-Ruane-Tschantz. In the context of visual boundaries of the Davis–Moussang complex of right-angled Coxeter groups, Mihalik-Ruane-Tschantz give sufficient condition for their local connectivity [ MRT07 ] . In fact, their conditions on the defining graph for local connectivity of the visual boundary (no product separators and no virtual factor separators) is equivalent to wide-avoidant for right-angled Coxeter groups. Theorem C generalizes their result to Morse boundaries of Coxeter groups. Mihalik-Ruane-Tschantz define the concept of a filter , a certain type of planar van-Kampen diagram with boundary two geodesic rays in the Davis complex. We generalize their definition to general Coxeter groups and show that a filter is a stable subset connecting two Morse geodesics. We show that filters between two geodesic rays always exist given the hypothesis of Theorem C . In order to prove Theorem C , we prove a characterization of Morse geodesics in affine-free Coxeter group which is of independent interest: Theorem D . Given a geodesic ray \gamma in the Davis complex of an affine-free Coxeter group, the following are equivalent: (1) The ray \gamma is Morse, (2) there exists a constant K such that any subpath of length larger than K does not have its label contained in a wide special subgraph, and (3) the ray \gamma spends uniformly bounded time in cosets of wide special subgroups. Note that the equivalence between (2) and (3) is immediate. Although the above theorem is perhaps intuitive, its proof turned out to be quite technical. In its proof, we utilize a result of Caprace [ Cap06 ] that allowed us to recognize affine subgroups from certain configuration of walls. Our proof then required a careful analysis of CAT(0) bridges and disk diagrams in Coxeter groups. We conjecture that Theorem D holds for all Coxeter groups, not just affine-free ones. This article is organized as follows. Section 2 gives necessary background on the Morse boundary and Coxeter groups, while Section 3 gives formal definitions of wide-avoidant and wide-spherical-avoidant Coxeter groups. Section 4 proves Theorem A regarding disconnected Morse boundaries. Next, in Section 5 we prove Theorem D characterizing Morse geodesics

Read the complete text