Distribution sensitivity in stochastic programming (Q1176576): Difference between revisions

The paper deals with stochastic programming problems depending on a random element through the corresponding probability measure only. Stochastic programming problems with penalty, two-stage stochastic programming problems and chance constrained stochastic programming problems are well-known types of optimized problems belonging to this class. The aim is to study the stability of the above-mentioned stochastic optimization problems with respect to perturbation of the underlying distribution in the sense of the topology of weak convergence. The Lipschitz metric for recourse models and the variational distance for the chance constrained case are taken to achieve the above-mentioned aim. The paper is divided into several parts. Namely, the parametric case, recourse problems and chance constrained problems are studied separately. Special attention is also paid to linear and quadratic recourse problems. Further the stability results with respect to Kolmogorov distance are mentioned in some cases. Valuable new results on the stability of stochastic programming problems are presented.