摘要
笔者提出基于GPU维度层面并行的局部PSO算法,换言之,基于GPU的局部粒子群优化算法求解高维优化函数,即在求解目标函数时对每一个维度进行并行处理。将粒子与线程块对应,线程块中的线程与目标函数的维度对应。实验表明,此算法在优化高维度目标函数中优势明显,概念简单,易编程实现,能有效果解决串行粒子群优化算法性能急剧下降的问题。
In other words, the GPU-based local particle swarm optimization(LPSO) algorithm is used to solve the highdimensional optimization function, i.e. to process each dimension in parallel when solving the objective function. The particles correspond to the thread block, and the threads in the thread block correspond to the dimension of the objective function. Experiments show that the algorithm has obvious advantages in optimizing high-dimensional objective function, simple concept and easy programming. It can effectively solve the problem of the performance degradation of serial particle swarm optimization algorithm.
引文
[1]KennedyJ,Eberhart R.Particle Swarm Optimization[C].//Proceedings of the IEEE International Conference on Neural Networks, 1995.
[2]张鑫源,胡晓敏,林盈.遗传算法和粒子群优化算法的性能对比分析[J].计算机科学与探索,2014,8(1):90-102.
[3]Owens J D,Houston M,Luebke D,et al.GPU Computing[J].Proceedings of the IEEE,2008,96(5):879-899.
[4]Mussi L,Nashed Y S G,Cagnoni S.GPU-based Asynchronous Particle Swarm Optimization[C].//Conference on Genetic&Evolutionary Computation,2011.
[5]张庆科,杨波,王琳,等.基于GPU的现代并行优化算法[J].计算机科学,2012,39(4):304-311.
[6]蔡勇,李光耀,王琥.基于CUDA的并行粒子群优化算法的设计与实现[J].计算机应用研究,2013,30(8):2415-2418.
[7]陈风,田雨波,杨敏.基于CUDA的并行粒子群优化算法研究及实现[J].计算机科学,2014,41(9):263-268.
[8]Kirk, D.B. and W.H. Wen-mei, Programming massively parallel processors:a hands-on approach.[M]2012:NewnesKirk D,Hwu W M.Programming Massively Parallel Processors[M].Beijing:China Machine Press,2010:146.
[9]Ivekovi,Trucco E,Petillot Y R.Human Body Pose Estimation with Particle Swarm Optimisation[J].Evolutionary Computation,2008,16(4):5 09-528.