A \(\Phi \)-entropy contraction inequality for Gaussian vectors (Q1047160): Difference between revisions

Let \((X,Y)\) be a Gaussian vector. An inequality proved by \textit{O. V. Sarmanov} [Dokl. Akad. Nauk SSSR 121, 52--55 (1958; Zbl 0089.36103)] states that for any measurable function \(f\), \(\text{Var}(\mathbb E[f(Y)|X])\leq \rho^2\text{Var}(f(Y))\), where \(\rho\) is the (linear) correlation coefficient between \(X\) and \(Y\). The author gives a nonlinear generalization of this result containing an inequality for a general \(\Phi\)-entropy. His method is based on the results by \textit{D. Chafaï} [J. Math. Kyoto Univ. 44, No. 2, 326--363 (2004; Zbl 1079.26009)]. As an example, the stochastic harmonic oscillator is considered.