Groups acting on products of locally finite trees

J.O.Button · 2026

Read Groups acting on products of locally finite trees by J.O.Button (2026) 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-2603-06139.txt.

Introduction

Button asks which finitely generated groups admit proper actions on finite products of locally finite simplicial trees, and offers evidence favoring hyperbolic surface groups. He also constructs one completely concrete embedding for the closed genus 2 hyperbolic surface group, placing it inside SL2 over F_p(x, y) for every prime p. The result supports the case for hyperbolic surface groups acting properly on such products.

Catalog record

CREATORJ.O.Button
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE89456
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.06139

Browse the full library · About the Tower of Babel

Full text

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

Groups acting on products of locally finite trees J.O.Button Abstract . We examine the question of which finitely generated groups act properly on a finite product of locally finite simplicial trees and present evidence in favour of hyperbolic surface groups having such an action. We also give a completely explicit embedding of the genus 2 closed hyperbolic surface group in S ​ L 2 ​ ( 𝔽 p ​ ( x , y ) ) SL_{2}(\mathbb{F}_{p}(x,y)) for any prime p p . 2020 Mathematics Subject Classification Primary 20F65; Secondary 20E08 1. Introduction A useful technique which can be used to understand a finitely generated group G is to find a geodesic metric space X on which it acts geometrically, that is properly and cocompactly by isometries. Of course every G has such an action on its Cayley graph but if the space X is well behaved geometrically, for instance if it is a hyperbolic space or a CAT(0) space, then this allows us to establish properties for the group G . However it can be quite restrictive asking for the action to be geometric. Indeed a group G admits a geometric action on some simply connected geodesic metric space if and only if G is finitely presented ( [ 3 ] I.8.11), so if G has a geometric action on such a space but G also has a subgroup H which is finitely generated but not finitely presented then there is no geometric action of H on any simply connected geodesic metric space. One way around this is to allow actions of a group G by isometries which are proper, but not necessarily cocompact, on a geodesic metric space X . This has the big advantage that every subgroup of G acts properly too. On the other hand, this might be too general: for instance it is an open question as to whether every countable group has a proper action on some CAT(0) space. However in this case it is possible to strengthen the hypotheses on the action without the full force of a geometric action, since the class of groups having a proper and semisimple action on a CAT(0) space is an interesting class with good geometric properties (see for instance [ 3 ] III. \Gamma .1.1). If we think of finite generation/presentability as finiteness conditions on the group and acting properly/cocompactly as finiteness conditions on the action, we might wonder about finiteness conditions on the space X . Hence we could also insist that X is a proper metric space (closed balls are compact). This has the advantage that acting properly and the stronger notion of acting metrically properly are then equivalent. We could further insist that X has some form of non-positive or negative curvature, for instance X is hyperbolic. But every countable group acts properly on a locally finite (and hence proper) hyperbolic graph. Alternatively we could specialise to the case where X is a tree, where here tree means simplicial tree throughout and group actions are always by simplicial automorphisms but they may invert edges. However if G is finitely generated then G acts properly on some tree if and only if G acts properly on some tree of bounded valence if and only if G is virtually free, thus the class of groups so obtained is very limited. Another reason to vary the type of spaces we are considering is that we would like the notion of acting properly on a space in a given class to be not just preserved by subgroups but also by finite index supergroups. With geometric actions this is far from clear, for instance if H is a CAT(0) group and it has finite index i in G , it is unknown whether G must also be CAT(0). However if we merely require a proper action then as long as our class of spaces is closed under taking direct products (say with the \ell_{2} product metric), we can induce an action of G on X^{i} from an action of H on X and here properness will be preserved. Examples of such classes \mathcal{C} where the question of which groups act properly on spaces in \mathcal{C} has been investigated recently include the class of CAT(0) cube complexes, where one might or might not restrict to finite dimensional and/or locally finite spaces, and finite products of quasi-trees. In this paper our focus is on groups that act properly (by automorphisms) on a finite product of trees, where each tree is equipped with its path metric and the space is equipped with the Euclidean product metric. This begs the question: should we consider arbitrary trees (in which case should we further consider proper or metrically proper actions), or only locally finite trees? In the first case a wide range of groups is known to have such an action (under either definition of proper) but for locally finite trees the known examples are much more restricted. Therefore our emphasis here will be on the locally finite case and seeing how it differs from the case of arbitrary trees. We provide the basic background in Section 2 and Lemma 2.7 shows that, at least for finitely generated groups, it does not matter if we consider products of locally finite or of bounded valence trees. But the question of which groups possess proper actions on finite products of locally finite trees seems quite mysterious. For instance we have virtually free groups and Burger - Mozes groups (these two examples also act cocompactly), along with subgroups and direct products of these examples. In 2.3 we look at finitely generated groups containing an infinite but locally finite subgroup, such as lamplighter groups F\wr\mathbb{Z} for F finite which also possess such an action. We also examine the Houghton groups which do not possess any proper action on a finite product of locally finite trees, although they do once the locally finite condition is removed. For any finite non-trivial group F , we show in Section 3 that, as opposed to F\wr\mathbb{Z} , the wreath product F\wr\mathbb{Z}^{2} does not have any proper action on a finite product of locally finite trees. This result sits alongside that of [ 14 ] Proposition 1.3, which shows that F\wr H cannot act properly on any finite dimensional CAT(0) cube complex (which includes any finite product of trees) if H contains a non-abelian free group. But what about any example of a word hyperbolic group with such an action but which is not virtually free? The obvious case to try would be the fundamental group S_{g} of a closed orientable hyperbolic surface \Sigma_{g} with genus \geq 2 . In fact exactly this question was raised in [ 12 ] which gives partial results and which was the motivation for much of this work. Here we present evidence in favour of the existence of a proper action for our surface groups S_{g} , by considering how these groups can act on a single locally finite tree. In Section 4, Theorem 4.4 shows that a necessary condition for such an action by S_{g} (or indeed any finitely generated, torsion free group which does not contain \mathbb{Z}\times\mathbb{Z} ) is that for any non-identity element \gamma ,

Read the complete text