Parallel Global Optimization of High-Dimensional Problems

  • Authors:
  • Siegfried Höfinger;Torsten Schindler;András Aszódi

  • Affiliations:
  • -;-;-

  • Venue:
  • Proceedings of the 9th European PVM/MPI Users' Group Meeting on Recent Advances in Parallel Virtual Machine and Message Passing Interface
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

A parallel version of an optimization algorithm for arbitrary functions of arbitrary dimension N has been developed and tested on an IBM-Regatta HPC system equipped with 16 CPUs of Power4 type, each with 1.3 GHz clock frequency. The optimization algorithm follows a simplex-like stochastic search technique aimed at quasi-complete sampling of all the local minima. Parallel functionality is incorporated with the Message Passing Interface - MPI - version 1.2. The program is applied to typical problems of dimension N=60 and N=512 and the results are analyzed with respect to operability and parallel scalability.