Comparison and deviation from a representation formula
The author presents two formulae which he then uses to obtain simple proofs of some well known and some new comparison theorems and deviation inequalities. Let \(X_1\) and \(X_0\) be \(d\)-dimensional Gaussian vectors and \(f\) be a function in \({\mathbb R^d}\). By a comparison theorem it is understood assumptions on \(X_1, X_2\) and \(f\) under which the inequality \({\mathbf E} f(X_1)\geq {\mathbf E} f(X_0)\) holds true. The deviation results of the paper contain, in particular, the following inequality: \[ {\mathbf P}(| f(X)-{\mathbf E}f(X)|)\leq {\mathbf E}| f(X)-{\mathbf E}f(X)| \frac{e^{-\lambda^2/2}}{\lambda}. \] Here \(\lambda>0, f\) is any Lipschitz function with \(\| f\|_{\text{Lip}}\leq 1\) and \(X\sim N(0,1)\) is a standard normal vector. For the entire collection see [Zbl 0895.00024].
This page was built for publication: Comparison and deviation from a representation formula
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4218908)
