Parallel processing and dynamic programming

Abstract

One of the limitations to the practical applicability of many numerical techniques to real problems is that of the time taken to reach a solution. In the field of Dynamic Programming this is an exceptionally critical factor since the solution time in general varies exponentially with the size of the problem. The advent of parallel processing and multiprocessor computers has meant that the large Dynamic Programming problems are no longer waiting for technology to supply faster computers, but merely require a machine with a large enough capacity for parallel processing. This thesis reviews the voluminous literature available on various aspects of parallel processing, and discusses some of the computational aspects of implementing parallel processing algorithms, with particular reference to Dynamic Programming.