Investigation of the genetic algorithm with an alternative representation of solutions

This paper describes the reaserch of the various options of representation of solutions in the genetic algorithm. Besides traditional binary coding and Gray code used to represent the real variables, Elias Gamma-codes and Deltacodes, Levenstein Gamma-codes, Golomb codes, Rice codes and others are examined. To test the data representations, a modified genetic algorithm with variable-length strings is used. This paper deals with the statistical significance of these parameters for the algorithm. The results of numerical studies are presented. The expediency of the use of some alternative coding for the individual tasks is shown.