An exact penalty algorithm for recourse-constrained stochastic linear programs (Q1194451)

The paper deals with the problem of two-stage stochastic linear programming under uncertainty: \(\min f(x)=cx+E[Q(x,\tilde\omega)]\) subject to \(Ax=b\) where \(Q(x,\tilde\omega)=\min qy\) subject to \(Wy=\tilde\omega-Tx\), \(y\geq 0\). The stochastic decomposition algorithm ``is extended by developing a solution method for a model in which the recourse function appears in the constraint set rather than in the objective function''. The procedure is exemplified by several problems and computational results are described.