Evolving one dimensional cellular automata to perform a non-trivial collective behavior task: One case study

Abstract

Here we present preliminary results in which a genetic algorithm (GA) is used to evolve one-dimensional binary-state cellular automata (CA) to perform a non-trivial task requiring collective behavior. Using a fitness function that is an average area in the iterative map, the GA discovers rules that produce a period-3 oscillation in the concentration of 1s in the lattice. We study one run in which the final state reached by the best evolved rule consists of a regular pattern plus some defects. The structural organization of the CA dynamics is uncovered using the tools of computational mechanics. PACS: 82.20Wt Computational modeling; simulation. © 2002 Springer-Verlag Berlin Heidelberg.