Surjunctivity does not characterize cosoficity of invariant random subgroups

Lewis Bowen, Michael Chapman · 2025

Read Surjunctivity does not characterize cosoficity of invariant random subgroups by Lewis Bowen, Michael Chapman (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-2511-06586.txt.

Introduction

Lewis Bowen and Michael Chapman settle a reversed question about surjunctive groups, but in a randomized generalization where the answer becomes reachable. A group is surjunctive when every injective cellular automaton on it is automatically surjective, and sofic groups are known to qualify. Whether every surjunctive group is sofic stays open for groups themselves, since no non-sofic group is known.

Working with invariant random subgroups of free groups, where non-cosofic examples were recently shown to exist, the authors construct one that is surjunctive yet not cosofic. The proof leans on complexity theory and on Rokhlin entropy methods. Their example even satisfies a stronger maximal-entropy condition than surjunctivity requires.

Catalog record

CREATORLewis Bowen, Michael Chapman
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE84 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-06586

Browse the full library · About the Tower of Babel

Full text

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

Surjunctivity does not characterize cosoficity of invariant random subgroups Lewis Bowen, Michael Chapman Abstract . A group is surjunctive if every injective cellular automaton on it is also surjective. Gottschalk famously conjectured that all groups are surjunctive. This remains a central open problem in symbolic dynamics and descriptive set theory. Gromov and Weiss termed the notion of sofic groups, and proved that all such groups are surjunctive, providing the largest class of groups which satisfy Gottschalk’s conjecture. It is still open to decide whether all groups are sofic. This became a major open problem in group theory, and is related to other well known problems such as the Aldous–Lyons conjecture in probability theory and to Connes’ embedding problem in the theory of operator algebras. A complementary natural question to ask is: Does the reverse implication to Gromov and Weiss’ result holds? Namely, are all surjunctive groups sofic? As currently there are no known non-sofic groups, answering this problem in the negative in the category of groups is still out of reach. This paper resolves this problem in the generalized setup of invariant random subgroups of free groups (IRSs), where non (co)sofic objects were recently shown to exist by Lubotzky, Vidick and the two authors. Specifically, we prove that there exists a surjunctive non (co)sofic IRS, resolving the aforementioned problem in the negative. Our proof uses a complexity theoretic approach, and in particular a recent development due to Manzoor, as well as the theory of Rokhlin entropy developed by Seward and others. As a byproduct of our proof technique, the non (co)sofic IRS we provide satisfies a condition stronger than surjunctivity; it satisfies a version of Seward’s maximal Rokhlin entropy of Bernoulli Shifts (RBS) criterion. 1. Introduction Let \Gamma be a finitely generated group, 1 1 1 Often the more generalized setup of discrete countable groups is used. But for the simplicity of various definitions, we prefer to keep the finite generation assumption throughout. \Sigma a finite set (of colors), and \Sigma^{\Gamma} the \Sigma -colorings of \Gamma . A cellular automaton on \Gamma with color palette \Sigma is a local, element independent recoloring mechanism of \Gamma ; namely, it is a recoloring map \Phi\colon\Sigma^{\Gamma}\to\Sigma^{\Gamma} , and there are elements \gamma_{1},...,\gamma_{n} and a function \phi\colon\Sigma^{n}\to\Sigma such that \Phi(c)(x)=\phi(c(x\gamma_{1}),...,c(x\gamma_{n})) for every coloring c\colon\Gamma\to\Sigma and element x\in\Gamma . A group \Gamma is said to be surjunctive if every injective cellular automaton on it is also surjective. A well known open problem due to Gottschalk [ 13 ] is: Are all finitely generated groups surjunctive? In [ 14 ] and [ 26 ] , Gromov and Weiss defined the notion of a sofic group (see Definition 1.4 ), and showed that all such groups are surjunctive. Subsequently, the following became a major open problem: Are all finitely generated groups sofic? Another natural problem that arises from their work is: Problem 1.1 ( [ 14 , 26 ] ) . Are all surjunctive groups sofic? Namely, does surjunctivity characterize soficity? Currently, no non-sofic group is known to exist, so a negative answer to Problem 1.1 is fairly out of reach. But, there is a generalized setup to this one, in which both soficity and surjunctivity are well defined, and also non sofic objects were recently shown to exist [ 4 , 5 ] — this is the setup of invariant random subgroups (IRSs) of free groups. 2 2 2 The study of invariant random subgroups was initiated independently by [ 25 ] , [ 2 ] and [ 8 ] (and appeared implicitly in [ 24 ] ). It has become a fruitful area of research, as several open problems in group theory, number theory and dynamics were resolved by studying IRSs (cf. [ 12 , 1 , 11 ] and the references therein). An invariant random subgroup of a given group is a probability measure over its subgroups which is conjugate invariant (see Section 1.1 for a detailed definition). Every pair (\Gamma,S) consisting of a group and a set of generators for it induces a short exact sequence N\hookrightarrow\mathcal{F}\twoheadrightarrow\Gamma , where \mathcal{F}=\mathcal{F}_{S} is the free group with basis S . As {\bf 1}_{N} , the Dirac measure concentrated at N , is an IRS of \mathcal{F} , the mapping (\Gamma,S)\mapsto{\bf 1}_{N} embeds the finitely generated groups into IRSs of free groups. This embedding is meaningful in our context, as IRSs can be cosofic (see Definition 1.4 ) and surjunctive (see Definition 1.6 ), and {\bf 1}_{N} is cosofic (respectively, surjunctive 3 3 3 It would be more natural to call this property cosurjunctivity , but we decided to avoid this. ) if and only if \Gamma\cong\nicefrac{{\mathcal{F}}}{{N}} is sofic (respectively, surjunctive). Moreover, similar to the result of Gromov and Weiss, one can show (see Corollary 5.4 and Theorem 3.6 ) that a cosofic IRS is always surjunctive. So, the analogoue of Problem 1.1 arises — are all surjunctive IRSs of free groups cosofic? Our main result is a negative answer to this problem: Theorem 1.2 . There exists a surjunctive non cosofic IRS of a finitely generated free group. In the rest of this introduction, we provide more detailed definitions of cosoficity and surjunctivity, as well as describe our proof technique. 1.1. Cosoficity and Surjunctivity of Invariant Random Subgroups The set \mathfrak{sub}(\mathcal{F}) consisting of all subgroups of the free group \mathcal{F} is compact when equipped with the product topology. A random subgroup is a Borel probability distribution over \mathfrak{sub}(\mathcal{F}) , and we denote the collection of all of them by {\rm Prob}(\mathfrak{sub}(\mathcal{F})) . 4 4 4 It is often the case that a random subgroup is a random variable with values in subgroups of \mathcal{F} that is distributed according to some law. We do not use this perspective at all along the paper, and a random subgroup (and later random subset) is the distribution law itself. The random subgroups can be equipped with the weak* topology, and with this topology, this space is compact. The conjugation action w.K=wKw^{-1} of \mathcal{F} on its subgroups extends naturally to random subgroups. 5 5 5 Because H will be used throughout this paper to denote entropy (of various kinds), we avoid using this notation for subgroups and use K for subgroups instead. Definition 1.3 (Invariant Random Subgroups) . An invariant random subgroup (IRS) of \mathcal{F} is a random subgroup \pi\in{\rm Prob}(\mathfrak{sub}(\mathcal{F})) such that w.\pi=\pi for every w\in\mathcal{F} . The set of all IRSs of \mathcal{F} is denoted by {\rm IRS}(\mathcal{F}) . For us there will be two main examples of IRSs. First, for every normal subgroup N\trianglelefteq\mathcal{F} , the Dirac measure {\bf 1}_{N} is an IRS. In addition, whenever \mathcal{F} acts on a finite set \Omega , it induces a finitely described IRS by choosing a uniformly random stabilizer of this action; namely, letting {\rm Stab}(x)=\{w\in\mathcal{F}\mid w.x=x\} , the IRS associated with \mathcal{F}{\curvearrowright}\Omega is (1.1) \pi=\frac{1}{|\Omega|}\sum_{x\in\Omega}{\bf 1}_{{\rm Stab}(x)}. Definition 1.4 (Soficity and Cosoficity) . An IRS \pi is cosofic if it is in the weak* closure of the finitely described IRSs, and we denote the set of these by {\rm IRS}_{\rm sof}(\mathcal{F}) . A finitely generated group \Gamma is sofic if it is isomorphic to \nicefrac{{\mathcal{F}}}{{N}} with \mathcal{F} a finitely generated free group, N\trianglelefteq\mathcal{F} , and {\bf 1}_{N} a cosofic

Read the complete text