Stochastic two-stage programming (Q1310805)

This book -- representing the author's habilitation thesis -- contributes to the methodology for two-stage stochastic programming models. It begins with recalling known definitions and characterizations. Then, two-stage stochastic programs are introduced in the first part: In a random environment, a decision \(x\) -- taken a priori -- can be corrected after the realization of the random variables by taking a second stage decision \(y\). The goal is then to minimize the total expected costs arising in the first and second stage subject to some remaining deterministic constraints. In Part II duality relations are investigated for convex optimization problems with supply/demand and prices treated as parameters. Based on these results the subdifferentiability and continuity of the optimal value function is studied. Approximations to stochastic two-stage programs by approximations of the integrands as well as by discretization of the underlying probability distribution of the random parameters are treated in Part III. Applying the stability criterion, the existence of bilinear functions is shown, which minorize (majorize) the integrand optimally with respect to the mean value. These minorants (majorants) support the integrand at generalized barycenters of the simplicial faces of the involved \(\times\)-simplices. After presenting an illustrative survey of existing approaches to two-stage stochastic programs in Part IV, the implementation of the barycentric approximation scheme is discussed in Part V. Some numerical experiences are reported in Part VI. The present interesting volume is well suited for dealing with two-stage stochastic programming problems in theory and practice.