摘要
蚁群算法在处理大规模TSP问题耗时较长,为解决这一不足,给出了一种基于MapReduce编程模式的并行蚁群算法。采用MapReduce的并行优化技术对蚁群算法中最耗时的循环迭代和循环赋值部分进行改进,同时运用PC集群环境的优势将具有一定规模的小蚁群分配到对应的PC机上,使其并行执行,减少运行时间。实验证明改进后的并行蚁群算法在大数据集上运行时间明显缩短,执行效率显著提高。
As ant colony algorithm is time consuming in dealing with large-scale TSP problems, a parallel opti- mization algorithm based on MapReduce programming mode is proposed, which improves the loop and loop assign- ment part with the most time-consuming by MapReduce parallel optimization technique. Simultaneously, it takes ad- vantage of PC integration environment to assign small ant colony with certain scale to corresponding PC machine and to make it execute in parallel as well as reduce its running time. Experiments show that the operation time of the im- proved parallel ant colony algorithm dealing with large data sets is significantly reduced and execution efficiency is significantly improved.
出处
《电子科技》
2013年第2期146-149,共4页
Electronic Science and Technology