Limited Recourse in Two-Stage Stochastic Linear Programs

Abstract In several real-world applications, modelled by two-stage stochastic problems, first and second-stage decisions (or some of their components) represent identical variables of the problem that is modelled. In these cases an appropriate solution of the problem might require that the second-stage decisions do not differ substantially from the corresponding first-stage ones. In this paper we propose a parametric approach to control the variability of the first and second-stage decisions and present a suitable solution framework. The advantage of the new approach is illustrated by considering two specific applications in electric power management and financial planning.