Cornell University
Efficient Ranking And Selection In Parallel Computing Environments
Abstract
dc:description.abstractThe goal of ranking and selection (R&S) procedures is to identify the best stochastic system from among a finite set of competing alternatives. Such procedures require constructing estimates of each system's performance, which can be obtained simultaneously by running multiple independent replications on a parallel computing platform. However, nontrivial statistical and implementation issues arise when designing R&S procedures for a parallel computing environment. This dissertation develops efficient parallel R&S procedures. In this dissertation, several design principles are proposed for parallel R&S procedures that preserve statistical validity and maximize core utilization, especially when large numbers of alternatives or cores are involved. These principles are followed closely by the three parallel R&S procedures analyzed, each of which features a unique sampling and screening approach, and a specific statistical guarantee on the quality of the final solution. Finally, in our computational study we discuss three methods for implementing R&S procedures on parallel computers, namely the Message-Passing Interface (MPI), Hadoop MapReduce, and Apache Spark, and show that MPI performs the best while Spark provides good protection against core failures at the expense of a moderate drop in core utilization.
Degree
thesis:*- Name thesis:degree_name
- Ph. D., Operations Research
- Level thesis:degree_level
- Doctor of Philosophy
- Discipline thesis:degree_discipline
- Operations Research
- Grantor
- Cornell University
- Year dc:date.issued
- 2016
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Ni, Cao
- Chair dc:contributor.chair
-
- Henderson,Shane G.
- Committee members dc:contributor.committeemember
-
- Martinez,Jose F.
- Frazier,Peter
Subjects
dc:subject × 3Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Dc Identifier Bibid
- 9596983