Generating combinations on the GPU and its application to the k-subset sum

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Efficiently representing and generating combinations can allow the seamless visualization, sampling, and evaluation of combinatorial architectures. In this paper, being relevant to tackle resource allocation problems ubiquitously, we address the subset sum problem by (1) using gradient-free optimization with a number-based representation of the combinatorial search space and by (2) generating combinations with minimal change order through parallel reductions in the GPU. Our computational experiments consisting of a relevant set of problem instances and gradient-free optimization algorithms show that (1) it is possible to generate combinations in the GPU efficiently, with quasi-linear complexity, (2) it is possible to tackle instances of the subset sum problem within a reasonable number of function evaluations, and (3) Particle Swarm Optimization with Fitness Euclidean Ratio converges faster. Since the search space of number-based representations is one-dimensional and amenable to parallelization schemes (e.g., GPU), we believe our work opens the door to tackle further combinatorial problems.

Original languageEnglish
Title of host publicationGECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion
PublisherAssociation for Computing Machinery, Inc
Pages1308-1316
Number of pages9
ISBN (Electronic)9781450383516
DOIs
Publication statusPublished - 2021 Jul 7
Event2021 Genetic and Evolutionary Computation Conference, GECCO 2021 - Virtual, Online, France
Duration: 2021 Jul 102021 Jul 14

Publication series

NameGECCO 2021 Companion - Proceedings of the 2021 Genetic and Evolutionary Computation Conference Companion

Conference

Conference2021 Genetic and Evolutionary Computation Conference, GECCO 2021
Country/TerritoryFrance
CityVirtual, Online
Period21/7/1021/7/14

Keywords

  • combinations
  • differential evolution
  • enumerative encoding
  • GPUs
  • gradient-free
  • knapsack problem
  • number representation
  • optimization
  • parallel reduction
  • particle swarm
  • subset sum

ASJC Scopus subject areas

  • Computer Science Applications
  • Software
  • Computational Theory and Mathematics

Fingerprint

Dive into the research topics of 'Generating combinations on the GPU and its application to the k-subset sum'. Together they form a unique fingerprint.

Cite this