Simulation Limitations of Affine Cellular Automata

TL;DR

The paper develops a general algebraic notion of intrinsic CA simulation and applies it to affine and linear cellular automata over finite fields. The central result establishes a rigidity: if a CA B is affine over $\mathbb{F}_p$ with mild bijectivity on outer components, then any CA A it can simulate must also lie in the same affine class over $\mathbb{F}_p$, strengthening negative results for additive CAs. The authors formalize simulation via algebraic operators (subalgebras, quotients, finite products, and iterative powers), prove closure properties, and derive consequences that include limitations for affine and abelian CA families. They also generalize the findings to arbitrary finite fields and to abelian CAs, and discuss connections to universal algebra and potential future work on pseudovarieties and Mal’tsev-type analyses. Overall, the work provides a principled, algebraic framework for proving negative results about CA computational power beyond constructive universality proofs.

Abstract

Cellular automata are a famous model of computation, yet it is still a challenging task to assess the computational capacity of a given automaton; especially when it comes to showing negative results. In this paper, we focus on studying this problem via the notion of CA relative simulation. We say that automaton A is simulated by B if each space-time diagram of A can be, after suitable transformations, reproduced by B. We study affine automata - i.e., automata whose local rules are affine mappings of vector spaces. This broad class contains the well-studied cases of additive automata. The main result of this paper shows that (almost) every automaton affine over a finite field F_p can only simulate affine automata over F_p. We discuss how this general result implies, and widely surpasses, limitations of additive automata previously proved in the literature. We provide a formalization of the simulation notions into algebraic language and discuss how this opens a new path to showing negative results about the computational power of cellular automata using deeper algebraic theorems.

Figures (4)

• Figure 1: Illustration of Example \ref{['ex:quotient_automaton']}. The figure shows that when ${\mathbb{A}} \in {\mathrm{H}}({\mathbb{B}})$, there exists a canonical extension ${\overline{\pi}}$ which effectively "translates" space-time diagrams of ${\mathcal{B}}$ to any given diagram of ${\mathcal{A}}$.
• Figure 2: Illustration of $\widetilde{g}$ and $\widetilde{g}^n$ for a ternary ($r=1$) function $g$ and $n=4$.
• Figure 3: Diagram of the unpacking map $o_3$.
• Figure 4: Illustration of $g^{[4]}$ for a ternary function $g$.

Theorems & Definitions (58)

• Definition 1: Cellular automaton
• Definition 2: CA canonical relations
• Example 4
• Definition 5: Unravelling a local function
• Definition 6
• Definition 7
• Definition 9: Iterative power of an automaton
• Definition 10
• Definition 11: CA simulation
• Lemma 12