Parallel Processors for Planning Under Uncertainty
STANFORD UNIV CA SYSTEMS OPTIMIZATION LAB
Pagination or Media Count:
This paper describes joint research under way whose goal is to demonstrate, for an important class of multistage stochastic models, that a variety of techniques for solving large-scale linear programs can be effectively mixed to attack this fundamental problem. The ideas involve nested primal and dual decomposition, combined with Monte Carlo simulation, high speed importance sampling, and quadrature methods for numerical integration, together with the use parallel processors. Keywords Linear programming, Mathematical programming, Large-scale optimization, Deterministic models, Times-staged systems, Staircase systems, Decomposition principle, Benders decomposition, Cutting planes, Parallel processors, Stochastic systems, Reliable systems, Hedging, Monte Carlo simulation, Importance sampling.
- Operations Research