Word Complexity And Repetitions In Words

With ideas from data compression and combinatorics on words, we introduce a complexity measure for words, called repetition complexity, which quantifies the amount of repetition in a word. The repetition complexity of w, r(w), is defined as the smallest amount of space needed to store w when reduced by repeatedly applying the following procedure: n consecutive occurrences uu . . . u of the same subword u of w are stored as (u, n). The repetition complexity has interesting relations with well-known complexity measures, such as subword complexity, sub, and Lempel-Ziv complexity, lz. We have always r(w) ≥ lz(w) and could even be that the former is linear while the latter is only logarithmic; e.g., this happens for prefixes of certain infinite words obtained by iterated morphisms. An infinite word α being ultimately periodic is equivalent to: (i) sub(pref n (α)) = O(n), (ii) lz(pref n (α)) = O(1), and (iii) r(pref n (α)) = lgn + O(1). De Bruijn words, well known for their high subword complexity, are shown to have almost highest repetition complexity; the precise complexity remains open. r(w) can be computed in time O(n(logn)) and it is open, and probably very difficult, to find fast algorithms.