IMPLEMENTING STOCHASTIC-CONTROL SOFTWARE ON SUPERCOMPUTING MACHINES

Up to now, stochastic control experiments with active learning were restricted, due to computational difficulties, to very small models. With the help of the contemporary-generation supercomputing machines, however, it is possible to employ larger models for stochastic control experiments. Stochastic control models up to 20 state equations seem to be feasible now. In this paper we report on the parallel features and implementation of an existing stochastic control program, Kendrick's DUAL program, on the latest class of supercomputing machines. © 1990.