An Adaptive Markov Chain Monte Carlo Algorithm for TSP

Traveling salesman problem is widely utilized as the typical issue for algorithm performance research because of its important engineering and theoretical value. An adaptive Markov Chain Monte Carlo algorithm is employed in resolving TSP for the purpose of ameliorating the temperature management problems in the original Metropolis algorithm. In order to get the balance between the runtime and route distance accuracy, the annealing parameter is no longer fixed in advance, but optimized by an adaptive sampler with simple expression and fast convergence. The simulation results shows that the adaptive algorithm has more powerful capacity of finding global solution and stability compared with those of the original Metropolis algorithm.