"ADMISSIBLE METHOD FOR IMPROVED GENETIC SEARCH IN CELLULAR AUTOMATA MODEL (AMMISCA)": A STRATEGY IN GENETIC CALIBRATION - PRELIMINARY RESULTS

Genetic Algorithms (GAs) are widely used to incrementally reach admissible solutions for hard problems such as parameter tuning in Cellular Automata (CA) models. This paper presents a genetic strategy, specifically developed for CA model calibration, exploiting the circumstance that the considered CA parameters have a physical meaning. The proposed approach has proved to be comparable and, in some cases outperforming, if compared with the standard GA proposed by Holland. As a further result, the goodness of the proposed genetic strategy opens the door to genetic tuning algorithms lacking of a standard crossover operator. Results, though preliminary, can be considered encouraging and routes to a wider analysis of the proposed approach.