Adapting an approximate level method to the two-stage stochastic programming problem

Abstract

We present a decomposition method for the solution of stwo-stage stochastic programming problems. This is an approximate method that can handle problems with large number scenarios. At the beginning, only rough approximation of the objective function is required. As the optimum is gradually approached, more and more accurate data are computed. The required accuracy is known at each step, hence efforts can be coordinated. The present framwork enables the application of interior-point methods because the convergence proof does not rely on basic solutions. Moreover, the classic discretization methods and stochastic approximation schemes naturally fit into the present framework.