Synthesis of self-replication cellular automata using genetic algorithms

This paper presents an efficient searching algorithm for one-dimensional cellular automata (CAs) with self-replicating structure. In the algorithm, the CA structure is represented by a simple fitness function and a genetic algorithm is used effectively where a gene implies a rule table. Based on preliminary experimental results, we provide interesting conjectures: (1) There exists optimal mutation rate for the fitness evolution, and (2) If genes are evolved successfully, they can produce some typical patterns