Infimum-convolution description of concentration properties of product probability measures, with applications (Q885277)

The paper is devoted to the concentration properties of product probability measures \(\mu=\mu_1\otimes \dots\otimes \mu_n\), expressed in term of dimension-free functional inequalities of the form \[ \left(\int e^{\alpha Q_\alpha f}\,d\mu\right)^{\frac{1}{\alpha}} \left(\int e^{-(1-\alpha)f}\,d\mu\right)^{\frac{1}{(1-\alpha)}}\leq 1, \] where \(\alpha\in (0,1)\) and \(Q_\alpha f\) is an appropriate infimum-convolution operator. These functional inequalities are applied to obtain optimal deviations inequalities for suprema of sums of independent random variables and the one-dimensional bin packing problem.