On intrinsic ergodicity and weakenings of the specification property

TL;DR

This paper investigates whether weakened forms of the specification property in dynamical systems imply intrinsic ergodicity, providing counterexamples and conditions under which multiple measures of maximal entropy can or cannot coexist.

Contribution

It demonstrates that almost weak specification and almost specification do not necessarily imply intrinsic ergodicity, offering explicit counterexamples and identifying conditions for uniqueness of measures.

Findings

Counterexamples with multiple measures of maximal entropy under weakened properties

Almost weak specification with logarithmic gap functions does not imply uniqueness

Almost specification with a mistake function of 1 does not imply uniqueness

Abstract

Since seminal work of Bowen, it has been known that the specification property implies various useful properties about a topological dynamical system, among them uniqueness of the measure of maximal entropy (often referred to as intrinsic ergodicity). Weakenings of the specification property called almost weak specification and almost specification have been defined and profitably applied in various works. However, it has been an open question whether either or both of these properties imply intrinsic ergodicity. We answer this question negatively by exhibiting examples of subshifts with multiple measures of maximal entropy with disjoint support which have almost weak specification with any gap function $f(n) = O(\ln n)$ or almost specification with any mistake function $g(n) \geq 4$. We also show some results in the opposite direction, showing that subshifts with almost weak specification with gap function $f(n) = o(\ln n)$ or almost specification with mistake function $g(n) = 1$ cannot have multiple measures of maximal entropy with disjoint support.