Cellular Automaton Reducibility as a Measure of Complexity for Infinite Words

Infinite words, also known as streams, hold significant interest in computer science and mathematics, raising the natural question of how their complexity should be measured. We introduce cellular automaton reducibility as a measure of stream complexity: σ is at least as complex as τ when there exists a cellular automaton mapping σ to τ. This enables the categorization of streams into degrees of complexity, analogous to Turing degrees in computability theory. We investigate the algebraic properties of the hierarchy that emerges from the partial ordering of degrees, showing that it is not well-founded and not dense, that ultimately periodic streams are ordered by divisibility of their period, that sparse streams are atoms, that maximal streams have maximal subword complexity, and that suprema of sets of streams do not generally exist. We also provide a pseudo-algorithm for classifying streams up to this reducibility.

💡 Research Summary

The paper introduces a novel framework for measuring the complexity of infinite words (streams) by means of one‑dimensional cellular automata (1CA). The authors define a reduction relation σ ≥_C τ: a stream σ reduces to τ if there exists a 1CA M whose global update f_M maps σ exactly to τ. This relation is a preorder; its symmetric part (σ ⋄_C τ) is an equivalence relation, thereby partitioning all streams into “degrees” of complexity, analogous to Turing degrees but far finer.

The motivation stems from the inadequacy of stronger computational models such as Turing machines (which collapse all computable streams into a single degree) and finite‑state transducers (FSTs), which cannot distinguish many natural families of streams (e.g., all ultimately periodic streams become equivalent). By contrast, a 1CA applies a uniform local rule simultaneously to every cell, allowing it to detect fine‑grained regularities such as period divisibility.

Key technical contributions include:

1. Formal definitions and basic properties – The paper formalises 1CA as a tuple (Σ, N, δ) with finite alphabet Σ, radius N, and local rule δ. The global map f_M is defined via the neighbourhood function V_{σ,N}. It is shown that the reduction relation is reflexive (identity CA) and transitive (composition of two CAs yields a CA with summed radius). Consequently, the equivalence classes (degrees) are well‑defined.

2. Closure under finite mutations – Inserting or deleting any finite word w at the beginning of a stream does not change its degree: w·σ ⋄_C σ. This follows from simple CAs that prepend or erase a fixed symbol and the transitivity of the reduction relation.

3. Binary encoding preservation – Any finite alphabet can be encoded into binary without altering reducibility, provided the encoding respects the locality constraints of CAs. The authors discuss a naïve encoding that fails and present a corrected scheme ensuring that σ ⋄_C τ iff ψ(σ) ⋄_C ψ(τ).

4. Algebraic structure of the degree hierarchy – The hierarchy exhibits several non‑trivial features:

   • Atoms: Certain “sparse” streams (e.g., streams where 1’s appear increasingly far apart) have no non‑trivial lower degrees; they are minimal non‑zero degrees.
   • Non‑density: Because atoms exist, there are pairs of degrees with no intermediate degree.
   • Non‑well‑foundedness: Infinite descending chains are constructed (e.g., by repeatedly halving the period of a periodic stream), showing the absence of minimal elements.
   • Infinite ascending chains: By repeatedly concatenating periods, one obtains infinite strictly increasing chains.
   • Divisibility ordering of periodic streams: For ultimately periodic streams, σ ≥_C τ iff the period of σ is a multiple of the period of τ. This reflects the CA’s ability to “sample” every k‑th cell.
   • Sparse streams as atoms or constants: A stream that is eventually constant or whose non‑zero symbols become arbitrarily rare is either an atom or collapses to a constant degree.
   • Maximal streams: Streams attaining maximal subword complexity (i.e., every possible finite word of length n appears) are maximal in the degree order; every other stream reduces to them, but not vice‑versa.
   • Lack of suprema: For many sets of streams, no least upper bound exists in the degree lattice, demonstrating that the hierarchy is not a complete lattice.

5. Decision procedure – Section 6 presents a pseudo‑algorithm for testing σ ≥_C τ. The algorithm enumerates candidate radii N and local rules δ, computes the first k symbols of f_M(σ) for each candidate, and checks equality with τ’s prefix. Correctness follows from the finiteness of the search space for any fixed N and the transitivity of reductions. While exponential in the size of the neighbourhood, the procedure is decidable, contrasting with the undecidability of many FST‑based questions.

6. Alternative definitions – The authors discuss variations such as multi‑step CA reductions, asynchronous CA, and reductions via continuous functions (Wadge reducibility). They argue that most of these alternatives either coincide with the presented notion or yield strictly coarser hierarchies.

In conclusion, the paper establishes cellular‑automaton reducibility as a robust, fine‑grained measure of stream complexity. It uncovers a rich algebraic landscape—containing atoms, maximal elements, and intricate ordering relations—while providing a concrete, decidable method for classifying streams. Future work is suggested on higher‑dimensional CA, connections with Kolmogorov complexity, and potential applications to symbolic dynamics and cryptographic sequence analysis.