SRB entropy of Markov transformations (Q2155945)

On a compact Riemannian manifold, each \(C^{1+\alpha}\) transitive Anosov map admits a single natural invariant measure. Given a dynamical system, the authors define the Sinai-Ruelle-Bowen (SRB) measure in different ways. The family of all piecewise \(C^{1+\alpha}\) expanding Markov transformations with a common transition matrix \(A\) is denoted by \(U^{1+\alpha}\). Depending on the family \(U^{1+\alpha}\), the authors describe two families of functions \(U^{1+\alpha}_p\) and \(U^{1+\alpha}_p(\rho)\) and give new results about the entropies for these families. In particular, the following theorem is proved: Theorem. For each piecewise Hölder continuous density function \(\rho\in C_P^{\alpha}[0, 1]\), the family \(U^{1+\alpha}_p(\rho)\) is a prime second-law family: the entropy functional \(H(\cdot)\) has a unique critical point in \(U^{1+\alpha}_p(\rho)\) where it attains the maximum value. Furthermore, the infimum of the entropy \(H( f )\) over \(U^{1+\alpha}_p(\rho)\) is zero. Furthermore, the authors show that the above claim holds true for the family of sufficiently smooth volume-preserving Anosov maps on a torus.