Ranking and selection in a high performance computing environment

Ranking and selection in a high performance computing environment. Ni, E. C., Hunter, S. R., & Henderson, S. G. In Pasupathy, R., Kim, S., Tolk, A., Hill, R., & Kuhl, M. E., editors, Proceedings of the 2013 Winter Simulation Conference, pages 833–845, Piscataway, NJ, 2013. Institute of Electrical and Electronics Engineers, Inc..

Paper doi abstract bibtex

We explore the adaptation of a ranking and selection procedure, originally designed for a sequential computer, to a high-performance (parallel) computing setting. We pay particular attention to screening and explaining why care is required in implementing screening in parallel settings. We develop an algorithm that allows screening at both the master and worker levels, and that apportions work to processors in such a way that excessive communication is avoided. In doing so, we reply on a random number generator with many streams and substreams.

@inproceedings{2013nihunhenWSC,
	Year = {2013},
	Author = {E. C. Ni and S. R. Hunter and S. G. Henderson},
	Title = {Ranking and selection in a high performance computing environment},
	Booktitle = {Proceedings of the 2013 Winter Simulation Conference},
	Editor = {R. Pasupathy and {S.-H.} Kim and A. Tolk and R. Hill and M. E. Kuhl},
	Publisher = {Institute of Electrical and Electronics Engineers, Inc.},
        Address = {Piscataway, NJ},
        Pages = {833--845},
        doi = {10.1109/WSC.2013.6721475},
        url_Paper = {http://web.ics.purdue.edu/~hunter63/PAPERS/2013nihunhenWSC.pdf},
        abstract = {We explore the adaptation of a ranking and selection procedure, originally designed for a sequential computer, to a high-performance (parallel) computing setting. We pay particular attention to screening and explaining why care is required in implementing screening in parallel settings. We develop an algorithm that allows screening at both the master and worker levels, and that apportions work to processors in such a way that excessive communication is avoided. In doing so, we reply on a random number generator with many streams and substreams.},
        keywords = {simulation optimization > single-objective > ranking and selection > parallel}}

Downloads: 0