[Paper Review] Freeness of automata groups vs boundary dynamics

Daniele D’Angeli, Emanuele Rodaro|arXiv (Cornell University)|Oct 22, 2014

semigroups and automata theory15 references3 citations

TL;DR

This paper establishes a dynamical characterization of freeness for automata groups by linking the algebraic property of freeness to the boundary dynamics of the group (or semigroup) generated by the enriched dual transducer. It proves that an automata group is free if and only if all Schreier graphs rooted at essentially non-trivial boundary points are infinite, with key results extending to bireversible transducers and yielding algorithmic, dynamical, and algebraic consequences.

ABSTRACT

We prove that the boundary dynamics of the (semi)group generated by the enriched dual transducer characterizes the algebraic property of being free for an automaton group. We specialize this result to the class of bireversible transducers and we show that the property of being not free is equivalent to have a finite Schreier graph in the boundary of the enriched dual pointed on some essentially non-trivial point. From these results we derive some consequences from the dynamical, algorithmic and algebraic point of view. In the last part of the paper we address the problem of finding examples of non-bireversible transducers defining free groups, we show examples of transducers with sink accessible from every state which generate free groups, and, in general, we link this problem to the nonexistence of certain words with interesting combinatorial and geometrical properties.

Motivation & Objective

• To characterize the freeness of automata groups using the dynamics of the enriched dual transducer on the boundary of the tree.
• To investigate the role of bireversible transducers in generating free groups, particularly through the structure of Schreier graphs at boundary points.
• To derive algorithmic and algebraic consequences from the dynamical characterization, including conditions for the absence of positive relations and torsion-freeness.
• To explore the existence of non-bireversible transducers generating free groups, especially those with sink states.
• To address open problems on stabilizers, amenability, and combinatorial structure of relations in automata groups.

Proposed method

• Define the enriched dual transducer to capture richer dynamics than the original automaton, enabling analysis of boundary behavior.
• Introduce the concept of 'essentially non-trivial' points in the boundary $(Q \cup Q^{-1})^\omega$, which represent elements of the Gromov boundary of the free group $F_Q$.
• Prove that Schreier graphs rooted at non-essentially non-trivial points are always finite, regardless of the group structure.
• Establish that freeness of the group generated by the dual transducer is equivalent to the infiniteness of all Schreier graphs rooted at essentially non-trivial points.
• Use the dynamical characterization to derive algebraic consequences, such as a new proof that Burnside groups with bounded exponent are finite.
• Apply the framework to analyze the existence of positive relations (i.e., $u=1$) in the semigroup of the dual, showing that their absence is necessary for freeness.

Experimental results

Research questions

• RQ1 Is the freeness of an automata group equivalent to the infiniteness of all Schreier graphs rooted at essentially non-trivial boundary points in the enriched dual transducer?
• RQ2 Can non-bireversible transducers with sink states generate free non-abelian automata groups acting transitively on the tree?
• RQ3 What is the geometric and combinatorial structure of 'strongly fragile' words, and can they be fully characterized?
• RQ4 Are the groups generated by the dual of 0-transition Cayley machines over $\mathbb{Z}_n$ free or free products of finite groups?
• RQ5 Is the problem of determining whether a Schreier graph rooted at a boundary point is finite decidable for inverse transducers?

Key findings

• An automata group is free if and only if all Schreier graphs rooted at essentially non-trivial boundary points in the enriched dual transducer are infinite.
• For bireversible transducers, the group generated by the dual is free if and only if the Schreier graphs at essentially non-trivial points are infinite, and Schreier graphs at non-essentially non-trivial points are always finite.
• The absence of positive relations (i.e., $u=1$) in the semigroup of the dual transducer is a necessary condition for the group to be free.
• If a reversible transducer generates a group with all trivial stabilizers on the boundary, then the semigroup of its dual is torsion-free and has no positive relations.
• The paper provides a new, elementary proof that Burnside automaton groups with bounded exponent are finite, without relying on Zelmanov’s theorem.
• The shortest strongly fragile word on $m$ occurrences has length at least $3(2^{m-1}) - 2$, suggesting that relations, if present, are extremely long and complex.