Using parallel and distributed computing to increase the capability of selection procedures

We present a framework for deploying selection procedures in a parallel and distributed environment to identify the stochastic systems with optimal expected response when the number of alternatives is large. The goal is to speed up the process of identifying a good design with a specified probability. We present a sequential selection procedure and discuss the validity of dividing the entire set of alternative designs into several groups that can be processed in a parallel and distributed fashion. The surviving designs from each group are then processed subsequently. An experimental evaluation demonstrates the validity and efficiency of the procedure