An evolutionary cellular program on the solution of the travelling salesman problem

An evolutionary algorithm is described for solving instances of the Travelling Salesman Problem (TSP). The key point of the algorithm is the distribution of the population over a grid, being in that sense a sort of cellular automata, but having the rules of change an iteration of a genetic algorithm each time. This genetic algorithm operating in miniature just takes account of a subset of the whole population, using for that a given neighbourhood relation, among several defined. This work compares the behaviour of the algorithm against a genetic algorithm, for different program's parameters. The set of benchmark instances was taken from the worldwide known collection TSPLIB