On the Solution Quality Assessment in Multi-stage Stochastic Optimization Under Different Model Representations

Date:

In this talk we discuss the idea behind the classical hydro-thermal scheduling problem (HTSP) with different model formulations. We describe a Sampling-based Decomposition Algorithm (SBDA) and apply it to approximately solve multi-stage stochastic programs versions of the HTSP. In this case, it is important to assess the solution quality that can be obtained from the resulting policy applied to out-of-sample paths and scenario trees.