Parallel processing and OR/MS

Parallel processing has moved in recent years from a largely theoretical issue to the commercial arena. Much of the interest in parallel architecture is related to its potential for complementing or replacing existing hardware in engineering and scientific computations. However, there are a number of OR/MS problems that can greatly benefit from the application of this technology. These include tree-search problems with either optimizing or heuristic rules, virtually all dynamic programming problems, large-scale mathematical programming problems and simulation of managerial and economic systems. This paper describes some of the current parallel processing architectures, and identifies characteristics of Operational Research OR problems which are amenable to solution using coarse-grained multiple-instruction, multiple-data parallel computers.