Traveling salesman problem optimization with parallel genetic algorithm

Abstract

The TSP is a NP-hard problem in optimization studied in both operations research and computer science. Metaheuristics are efficient alternative techniques for NP-hard and greater dimensional problems and they are impossible to solve by classic mathematical techniques. The paper describes the use of MATLAB's Parallel Computing Toolbox for parallel genetic algorithm based design of a TSP. Parallel genetic algorithms (PGA) represents a stochastic optimization approach which is computed in co-operating and interconnected computation nodes in a parallel mode. Each node of the PGA can be located on the same processor, or on more processors or on more computers respectively. We also carried on the computation of a TSP example which shows a higher speedup and a better per-formance.