A problem-solving environment based on cellular automata

in this paper we present a novel application to the cellular automata implementation of the diffusion-limited aggregation (DLA) model. This model, mainly used in Physics, Physical Chemistry, and Biology in the simulation of fractal growth phenomena is applied in the present work to the development of a heuristic problem-solving methodology. The method is applied in the solution of a hard combinatorial optimization problem, the Euclidean Traveling Salesman Problem. Experimental tests, carried out over a standard set of problem instances, show that the method outperforms, considering the quality of the produced solutions, another heuristic based on the real space renormalization theory. It is also found that it compares favourably (considering the computation time) with two other methods based on physical processes: The elastic net and simulated annealing methods.