Phi-entropy inequalities for diffusion semigroups (Q968870)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Phi-entropy inequalities for diffusion semigroups
scientific article

    Statements

    Phi-entropy inequalities for diffusion semigroups (English)
    0 references
    0 references
    0 references
    10 May 2010
    0 references
    Let \((P_t)_{t\geq 0}\) be a sub-Markovian semigroup with infinitesimal generator \(L\). Throughout, \(L\) is a second-order differential operator with variable coefficients. The \(\Phi\)-entropy of a function \(f\) is \[ \text{Ent}_t^\Phi(f) := \int \Phi(f)\,d\mu - \Phi\left(\int f\,d\mu\right), \] where \(\Phi:I\to\mathbb R\), \(I\subset\mathbb R\), is a convex function such that \(f, \Phi(f)\) are defined and \(\mu\)-integrable. \(\Phi\) is called admissible if it is strictly convex, of class \(C^4\) and such that \(1/\Phi''\) is again convex. The authors are interested in new \(\Phi\) entropy inequalities, i.e., bounds of \(\text{Ent}_t^\Phi(f)\) in terms of the carré-du-champ (or mean-field) operator \[ \Gamma(f,g) := \tfrac 12 \left(L(fg)-fLg - gLf\right). \] The \(\Phi\)-entropy inequalities include the Poincaré and the log-Sobolev inequalities. The authors begin with a short new proof of known equivalences of the \(CD(\rho,\infty)\) criterion (cast in terms of Bakry-Emery's \(\Gamma_2\)-operator) and (reverse) local \(\Phi\)-entropy inequalities for \((P_t)_{t\geq 0}\). They develop a general integral criterion (weaker than the \(CD(\rho,\infty)\)-criterion) which extends the Poincaré and log-Sobolev inequality cases and interpolates between them if \(\Phi(x)=x^p\), \(1<p<2\). For a large class of diffusion semigroups, the asymptotic time-behaviour is studied; in some cases, the existence of an ergodic measure is proved. For the power-law entropies \(\Phi(x)=x^p\), \(1<p<2\), the authors derive certain power-law \(\Phi\)-entropy inequalities (introduced by Arnold and Dolbeault) both for the Markov semigroup and the ergodic measure, thus strengthening the Beckner inequalities. The authors show that such results cannot be expected if the entropy function \(\Phi\) fails to be admissible. In the final section, various applications of \(\Phi\)-entropy inequalities to inhomogeneous Markov semigroups and nonlinear evolution equations (e.g., of McKean-Vlasov type) are given.
    0 references
    0 references
    0 references
    0 references
    0 references
    logarithmic Sobolev inequality
    0 references
    Poincaré inequality
    0 references
    diffusion semigroup
    0 references
    Fokker-Planck equation
    0 references
    entropy inequality
    0 references
    0 references
    0 references