Balanced word

Revision as of 15:05, 1 January 2024 by Inthar (talk | contribs)

A word or necklace s is balanced if its balance satisfies the following:

[math]\displaystyle{ \mathsf{balance}(s) := \max \big\{ \big| |w|_{x_i} - |w'|_{x_i} \big| : x_i \text{ is a letter of }s\text{ and }k = \operatorname{len}(w) = \operatorname{len}(w') \big\} \leq 1, }[/math]

where |u|xi is the number of occurrences of the letter xi in the word u.

Properties

  • A balanced word or necklace on N letters has a maximum variety bound of [math]\displaystyle{ N \choose {\lceil N/2 \rceil} }[/math].
  • If w is an aperiodic infinite balanced word, then w is constructed via a finite sequence of "congruence substitutions" beginning with a Sturmian word. Over 3 or more letters, all such words have a density vector (vector of relative letter frequencies) a = (a_i) which has a pair of components that are equal. [1]
  • Some periodic balanced words are not obtainable via congruence substitutions. For alphabets of size N = 3, ..., 7, the only examples of density vectors with all components distinct are permutations of (1, 2, 4, ..., 2N-1) arising from the Fraenkel word FN, defined via [math]\displaystyle{ F_1 = \mathbf{0}, F_n = F_{n-1} \mathbf{(n-1)} F_{n-1}. }[/math] The assertion that this is true for all N ≥ 3 is Fraenkel's conjecture.

Generalizations

References

  1. Brauner, N., Crama, Y., Delaporte, E., Jost, V., & Libralesso, L. (2019). Do balanced words have a short period?. Theoretical Computer Science, 793, 169-180.