Coxeter Groups and Asynchronous Cellular Automata

The dynamics group of an asynchronous cellular automaton (ACA) relates properties of its long term dynamics to the structure of Coxeter groups. The key mathematical feature connecting these diverse fields is involutions. Group-theoretic results in the latter domain may lead to insight about the dynamics in the former, and vice-versa. In this article, we highlight some central themes and common structures, and discuss novel approaches to some open and open-ended problems. We introduce the state automaton of an ACA, and show how the root automaton of a Coxeter group is essentially part of the state automaton of a related ACA.

💡 Research Summary

The paper establishes a rigorous bridge between asynchronous cellular automata (ACA) and Coxeter groups by showing that the dynamics of an ACA can be captured by a group – the dynamics group – whose generators are involutive updates of individual cells. The authors begin by formalizing ACA as a system where each cell’s local rule is applied in a non‑simultaneous, possibly arbitrary order. Because each local update is its own inverse, the set of all possible updates forms a collection of involutions. By taking the free product of these involutions and factoring by the relations that arise from the interaction of neighboring updates, one obtains a well‑defined group, the dynamics group, which acts on the state space of the ACA.

The central observation is that this dynamics group is, in many cases, a Coxeter group. Coxeter groups are generated by involutions subject to relations of the form ((s_i s_j)^{m_{ij}} = 1). The paper shows that the commutation and braid relations among cell updates in an ACA correspond precisely to the Coxeter relations with appropriate exponents (m_{ij}). Consequently, the state transition graph of the ACA is isomorphic to the Cayley graph of the associated Coxeter group with respect to its generating set.

Building on this isomorphism, the authors introduce the “state automaton” of an ACA, which records the effect of successive updates on the global configuration. They prove that the state automaton coincides with the well‑known “root automaton” of the corresponding Coxeter group. In the root automaton, each node represents a root (a signed combination of simple reflections), and edges correspond to applying a simple reflection. The mapping identifies each cell’s update with a simple reflection, so that traversing the state automaton is exactly the same as walking through the root automaton.

This identification yields several powerful consequences. First, classical results from Coxeter theory—such as the length function, weak order, and convexity properties—translate directly into dynamical properties of the ACA. For example, the length of a group element equals the minimal number of cell updates needed to reach a given configuration, providing a lower bound on convergence time. The weak order on the Coxeter group characterizes which configurations can be reached without creating cycles, thereby identifying regions of the state space that guarantee eventual fixation. Moreover, the finiteness or infiniteness of the Coxeter group determines whether the ACA exhibits only polynomial‑time convergence (finite/abelian groups) or can support arbitrarily long non‑repeating trajectories (infinite, non‑abelian groups).

The paper illustrates these ideas with two concrete families. In a one‑dimensional binary ACA where each cell updates by XOR with its right neighbor, the dynamics group is isomorphic to a finite symmetric Coxeter group (A_n). The authors show that every initial configuration converges to a fixed point in at most (O(n)) steps, a result that follows from the known diameter of the Coxeter group’s Cayley graph. In contrast, a two‑dimensional ACA based on a Life‑like rule yields a dynamics group that is an affine Coxeter group (\tilde{A}_n), which is infinite. Here the authors demonstrate the emergence of long‑period cycles and chaotic‑looking behavior, directly linked to the infinite nature of the underlying Coxeter group.

Beyond these examples, the authors outline a program for exploiting the group‑theoretic perspective on ACA. They propose algorithms for extracting the Coxeter presentation from a given ACA, for computing the length of configurations, and for using weak order to design update schedules that guarantee convergence. They also discuss how the dynamics group can be used to classify ACA up to dynamical equivalence: two ACAs with isomorphic dynamics groups share many qualitative features of their long‑term behavior.

The paper concludes with a set of open problems. Key among them is the computational complexity of finding a minimal asynchronous schedule for a given Coxeter group, the decomposition of the state automaton when the dynamics group is a non‑abelian infinite Coxeter group, and the quantitative relationship between the Coxeter length function and the actual time needed for convergence under various scheduling policies. Another intriguing direction is to explore whether the involution‑centric viewpoint can lead to new representations of Coxeter groups inspired by cellular automata dynamics.

In summary, the work reveals that the involution structure common to both ACA updates and Coxeter generators is not a superficial coincidence but a deep structural link. By translating dynamical questions about asynchronous cellular automata into algebraic questions about Coxeter groups—and vice versa—the paper opens a fertile interdisciplinary avenue that promises new insights into the complexity, convergence, and classification of discrete dynamical systems.