Design and implementation of parallel simulation ranking and selection procedures

Abstract

Conventional simulation ranking-and-selection(R&S) procedures are designed and implemented for serial computing environment. However, today's growth of computing power relies more on parallelism rather than faster serial execution. The conflict makes it difficult for those serial simulation R&S procedures to take advantage of the development of computing technology. In this thesis, we propose our design and implementation of simulation R&S procedures for parallel computing environment with a master-slave structure in higher level. Extensive numerical experiments show that our parallel simulation R&S procedures are capable of solving large-scale problems and are scalable when the number of available parallel computing units increases.

Date of Award	2013
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology