Influence of template specialization degree on the search space in evolutionary synthesis of models

An algorithm for evolutionary synthesis is presented. It combines the advantages of genetic algorithms and genetic programming based on evolutionary computation, template (pattern, skeleton) of algorithm and a given set of pairs of input-output data. The influence of template specialization degree on the search space in the evolutionary synthesis is investigated. Estimates are obtained for the reduction of the search space with the introducing an additional information of template in the form of formulas represented by binary trees and refining the model. The exponential and superexponential reduction in the time of the synthesis algorithm with the increasing the template specialization degree is shown theoretically and experimentally.