Dynamic Generation of Scenario Trees

Abstract

We present new algorithms for the dynamic generation of scenario trees for multistagestochastic optimization. The different methods described are based on random vectors, whichare drawn from conditional distributions given the past and on sample trajectories.The structure of the tree is not determined beforehand, but dynamically adapted to meeta distance criterion, which insures the quality of the approximation. The criterion is built ontransportation theory, which is extended to stochastic processes.