A Parallel ACO Approach Based on One Pheromone Matrix
This paper presents and implements an approach to parallel ACO algorithms. The principal idea is to make multiple ant colonies share and utilize only one pheromone matrix. We call it SHOP (SHaring One Pheromone matrix) approach. We apply this idea to the two currently best instances of ACO sequential algorithms (MMAS and ACS), and try to hybridize these two different ACO instances. We mainly describe how to design parallel ACS and MMAS based on SHOP. We present our computing results of applying our approach to solving 10 symmetric traveling salesman problems, and give comparisons with the relevant sequential versions under the fair computing environment. The experimental results indicate that SHOP-ACO algorithms perform overall better than the sequential ACO algorithms in both the computation time and solution quality.
KeywordsMaster Thread Pheromone Matrix Chinese Word Segmentation Parallel Computing Platform Symmetric Travel Salesman Problem
Unable to display preview. Download preview PDF.
- 1.Bullnheimer, B., Kotsis, G., Strauß, C.: Parallelization strategies for the ant system. Applied Optimization 24, 87–100 (1998)Google Scholar
- 8.Delisle, P., Krajecki, M., Gravel, M., Gagné, C.: Parallel implementation of an ant colony optimization metaheuristic with openmp. In: International Conference on Parallel Architectures and Compilation Techniques (2001)Google Scholar