Uncertainties in minimax stochastic programs

Abstract

When using the minimax approach one tries to hedge against the worst possible distribution belonging to a specified class P. A suitable stability analysis of results with respect to the choice of this class is an important issue. It has to be tailored to the type of the minimax problem, to the considered class of probability distributions and to the anticipated input perturbations. We shall focus on the effect of changes in input information for classes of probability distributions with support belonging to a given set and defined by (possibly perturbed) generalized moments values.