Entropy, volume growth and SRB measures for Banach space mappings (Q510170): Difference between revisions

Let \(B\) be a Banach space, \(f:B \longrightarrow B\) be a \(C^2\) Fréchet differentiable and injective map and let \(\mu\) be a \(f\)-invariant Borel probability measure with finitely many positive Lyapunov exponents. Denoting by \(\lambda_i(x)\) the Lyapunov exponents of \((f, \mu)\) at \(x\), with multiplicities respectively \(m_i(x)\), it is well known that the following inequality involving the entropy, \[ h_\mu(f) \leq \int \sum_i m_i(x) \lambda_i^+(x) \;d\mu,\eqno{(*)} \] holds for any Borel probability invariant measure \(\mu\) (see [\textit{P. Thieullen}, Ann. Inst. Henri Poincaré, Anal. Non Linéaire 4, 49--97 (1987; Zbl 0622.58025)]). In [\textit{F. Ledrappier} and \textit{L. S. Young}, Ann. Math. (2) 122, 509--539, 540--574 (1985; Zbl 0605.58028)] it is proved that, in case \(B\) is a finite-dimensional Banach space, a measure \(\mu\) is SRB (Sinai-Ruelle-Bowen) if and only if \((*)\) holds as equality. In the present paper the authors generalize that result in the setting of infinite-dimensional Banach spaces under the condition of no zero Lyapunov exponents. The paper is well written, self-contained as well as highly elegant. It uses local theory as John's classical theorem and multiplicative ergodic theorem. The main point is to give distortion estimates (for some volumes that the authors introduce), using Lyapunov metrics, taking advance of the uniform expansion along unstable leaves in some adapted norms, which turns out to be regular enough on unstable manifolds.