Design and Experimentation of a Large Scale Distributed Stochastic Control Algorithm Applied to Energy Management Problems
Résumé
The Stochastic Dynamic Programming method often used to solve some stochastic optimization problems is only usable in low dimension, being plagued by the curse of dimensionality. In this article, we explain how to postpone this limit by using High Performance Computing: parallel and distributed algorithms design, optimized implementations and usage of large scale distributed architectures (PC clusters and Blue Gene/P).