Soluciones cercanas al optimo para el problema del ladrón viajero a través de un algoritmo genético paralelo implementado en unidades de procesamiento grafico (gpus)

Abstract

The Traveling Thief Problem (TTP) is a new and important combinatorial optimization problem that combines two outstanding problems of the NP-Hard class; which are the Traveling Salesman Problem (TSP) and the Knapsack Problem (KP). The TTP has been attempted to be solved using different algorithms and heuristics; The research proposed in this paper will seek an implementation on GPUs of a parallel genetic algorithm to find solutions close to the optimal through the exhaustive use of multiprocessing hardware and the appropriate use of memory spaces. The problems tested correspond to prominent problems in the literature (Benchmark Problems) and the results of the execution will be compared against other results documented so far to determine the validity of our algorithm.