Directional entropy of rotation sets

Let \((X,T)\) be a compact topological dynamical system. Any \(n\)-tuple \(f= (f_1,\dots, f_n)\) of continuous real-valued maps on \(X\) induces a map \(\mu\mapsto f_*(\mu)= (\int f_1 d\mu,\dots,\int f_n d\mu)\) from the set \({\mathcal M}\) of \(T\)-invariant probability measures on \(X\) to \(\mathbb{R}^n\), and the set \(f_*({\mathcal M})\) is called the rotation set of \(f\). Following Geller and Misiurewicz, a measure \(\mu\in{\mathcal M}\) is called directional if \(f_*(\nu)= f_*(\mu)\) for every \(\nu\in{\mathcal M}\) whose support is contained in that of \(\mu\), and lost otherwise.NEWLINENEWLINENEWLINEThe author investigates certain geometric properties of \(f_*({\mathcal M})\) under additional hypotheses on \((X,T)\). The main result (Theorem 2) states that, if \((X,T)\) is a mixing shift of finite type, and if the components of \(f\) are cohomologically independent functions of summable variation, then any \(\rho\) which is either an interior or an exposed extremal point of \(f_*({\mathcal M})\) has the property that NEWLINE\[NEWLINE\sup_{\substack{ \nu\in f^{-1}_*(\rho)\\ \nu\text{ is directional}}} h_\nu(T)= \sup_{\nu\in f^{-1}_*(\rho)} h_\nu(T),NEWLINE\]NEWLINE where \(h_\nu(T)\) denotes metric entropy. If \(\rho\) is interior, then this supremum is attained by a unique lost measure in \(f^{-1}_*(\rho)\), and if \(\rho\) is exposed, the supremum is attained by at least one ergodic and directional measure, but not by any lost measure.NEWLINENEWLINENEWLINEThe author gives no information about non-exposed extremal points in \(f_*({\mathcal M})\).

Recommendations

Cited in

(9)

This page was built for publication: Directional entropy of rotation sets

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2732248)