A New Dynamical Evolutionary Algorithm Based on Statistical Mechanics

In this paper, a new dynamicalevolutionary algorithm (DEA) is presented based on the theory ofstatistical mechanics. The novelty of this kind of dynamical evolutionaryalgorithm is that all individuals in a population (called particlesin a dynamical system) are running and searching with theirpopulation evolving driven by a new selecting mechanism. This mechanismsimulates the principle of molecular dynamics, which is easy todesign and implement. A basic theoretical analysis for thedynamical evolutionary algorithm is given and as a consequence twostopping criteria of the algorithm are derived from the principle ofenergy minimization and the law of entropy increasing. In order toverify the effectiveness of the scheme, DEA is applied to solving sometypical numerical function minimization problems which are poorlysolved by traditional evolutionary algorithms. The experimental resultsshow that DEA is fast and reliable.