Edit distance in substitution systems

Abstract

Let $\ddot{I}ƒ$ be a primitive substitution on an alphabet $\mathcal{A}$, and let ${\mathcal{W}}_n$ be the set of words of length $n$ determined by $\ddot{I}ƒ$ (i.e., $w\in {\mathcal{W}}_n$ if $w$ is a subword of $\ddot{I}{ƒ}^k(a)$ for some $a\in \mathcal{A}$ and $k\ge 1$). It is known that the corresponding substitution dynamical system is loosely Kronecker (also known as zero-entropy loosely Bernoulli), so the diameter of ${\mathcal{W}}_n$ in the edit distance is $o(n)$. We improve this upper bound to $O(n/\sqrt{\log n})$. The main challenge is handling the case where $\ddot{I}ƒ$ is non-uniform; a better bound is available for the uniform case. Finally, we show that for the Thue--Morse substitution, the diameter of ${\mathcal{W}}_n$ is at least $\sqrt{n/6}-1$.

25 pages, 4 figures; updated introduction