Content:
An analysis of convex stochastic programs is provided if the underlying probability distribution is subjected to (relatively small) perturbations. In particular, it's shown, that the convex set of approximate solutions is Lipschitz continuous with respect to these perturbations when the distance between probability measures is measured in terms of certains metrics, (Fortet-Mourier) generated by the integrands that define the stochastic program. It also explained how these results apply to (linear) stochastic programs with recourse. Finally, we study the implications of associating such Fortet-Mourier metrics with the asymptotic behavior of empirical estimates of such stochastic programs
September 2003