A new dynamical evolutionary algorithm based on statistical mechanics

  • Authors:
  • YuanXiang Li;XiuFen Zou;LiShan Kang;Zbigniew Michalewicz

  • Affiliations:
  • Software Engineering Laboratory, Wuhan University, Wuhan 430072, P.R. China;Software Engineering Laboratory, Wuhan University, Wuhan 430072, P.R. China;Software Engineering Laboratory, Wuhan University, Wuhan 430072, P.R. China;Computer Science Department, University of North Carolina at Charlotte, Charlotte NC

  • Venue:
  • Journal of Computer Science and Technology
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, a new dynamical evolutionary algorithm (DEA) is presented based on the theory of statistical mechanics. The novelty of this kind of dynamical evolutionary algorithm is that all individuals in a population (called particles in a dynamical system) are running and searching with their population evolving driven by a new selecting mechanism. This mechanism simulates the principle of molecular dynamics, which is easy to design and implement. A basic theoretical analysis for the dynamical evolutionary algorithm is given and as a consequence two stopping criteria of the algorithm are derived from the principle of energy minimization and the law of entropy increasing. In order to verify the effectiveness of the scheme, DEA is applied to solving some typical numerical function minimization problems which are poorly solved by traditional evolutionary algorithms. The experimental results show that DEA is fast and reliable.