Tuned annealing for optimization

Mir M. Atiqullah, S. S. Rao

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

4 Scopus citations

Abstract

The utility and capability of simulated annealing algorithm for generalpurpose engineering optimization is well established since introduced by Kirkpatrick et. al1. Numerous augmentations are proposed to make the algorithm effective in solving specific problems or classes of problems. Some proposed modifications were intended to enhance the performance of the algorithm in certain situations. Some specific research has been devoted to augment the convergence and related behavior of annealing algorithms by modifying its parameters, otherwise known as cooling schedule. Here we introduce an approach to tune the simulated annealing algorithm by combining algorithmic and parametric augmentations. Such tuned algorithm harnesses the benefits inherent in both types of augmentations resulting in a robust optimizer. The concept of ‘reheat’ in SA, is also used as another tune up strategy for the annealing algorithm. The beneficial effects of ‘reheat’ for escaping local optima are demonstrated by the solution of a multimodal optimization problem. Specific augmentations include handling of constraints, fast recovery from infeasible design space, immunization against premature convergence, and a simple but effective cooling schedule. Several representative optimization problems are solved to demonstrate effectiveness of tuning annealing algorithms.

Original languageEnglish (US)
Title of host publicationComputational Science – ICCS 2001 - International Conference, Proceedings
EditorsVassil N. Alexandrov, Jack J. Dongarra, Benjoe A. Juliano, Rene S. Renner, C.J. Kenneth Tan
PublisherSpringer Verlag
Pages669-679
Number of pages11
ISBN (Print)3540422331, 9783540422334
DOIs
StatePublished - Jan 1 2001
EventInternational Conference on Computational Science, ICCS 2001 - San Francisco, United States
Duration: May 28 2001May 30 2001

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2074
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

OtherInternational Conference on Computational Science, ICCS 2001
CountryUnited States
CitySan Francisco
Period5/28/015/30/01

Keywords

  • Constrained optimization
  • Cooling schedule
  • Design optimization
  • Simulated annealing
  • Tuned annealing

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'Tuned annealing for optimization'. Together they form a unique fingerprint.

  • Cite this

    Atiqullah, M. M., & Rao, S. S. (2001). Tuned annealing for optimization. In V. N. Alexandrov, J. J. Dongarra, B. A. Juliano, R. S. Renner, & C. J. Kenneth Tan (Eds.), Computational Science – ICCS 2001 - International Conference, Proceedings (pp. 669-679). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 2074). Springer Verlag. https://doi.org/10.1007/3-540-45718-6_72