Parallel implementation of ant colony optimization on MPP

An adaptive parallel ant colony algorithm (PACO) is presented. In the algorithm, we propose a strategy for information exchange between processors which make each processor choose its partner to communicate and update the pheromone adaptively. We also propose a method of adjusting the time interval of information exchange adaptively according to the diversity of the solutions so as to increase the ability of search and avoid early convergence. Experimental results show that our algorithm PACO has high convergence speed, high speedup and efficiency.