Entropies, convexity, and functional inequalities: on \(\Phi\)-entropies and \(\Phi\)-Sobolev inequalities (Q2567322): Difference between revisions

For a smooth convex real-valued function \(\Phi\) on a closed interval and a probability measure \(\mu\) on a Borel space \((\Omega , \mathcal F)\), the \(\Phi\)-entropy functional \(\text{Ent}^\Phi_\mu (f)\) of a real-valued function \(f\) on \(\Omega\) is defined as \[ \text{Ent}^\Phi_\mu (f)=\int_\Omega \Phi (f)\,d\mu -\Phi \left( \int_\Omega f\,d\mu \right) . \] A number of well-known inequalities, like the Poincaré inequality and the logarithmic Sobolev inequality, can be expressed in terms of \(\text{Ent}^\Phi_\mu \) for appropriate \(\Phi\) and \(\mu\). The author gives a unified treatment and a generalization of such inequalities, covering many different situations including hyper-contracting diffusions, uniformly strictly log-concave measures, Wiener measures on Riemannian manifolds, and Poisson spaces. Relations with the Boltzmann-Shannon entropy from the kinetic theory of gases and the information theory are discussed.