Generating Forbidden Words and Measures of Maximal Entropy for Subshifts

Abstract

We prove that if the entropy of generating forbidden words of a subshift over a finite alphabet is strictly smaller than its topological entropy, then the subshift has only finitely many ergodic measures of maximal entropy, each of which is Bernoulli up to a finite period, and admits a characteristic measure of maximal entropy. In particular, these conclusions hold when the entropy of minimal forbidden words is strictly smaller than the topological entropy, giving a negative answer to a question of Buzzi.