摘要
为了提高蛋白质二级结构预测的效率,对具有完全学习策略的量子行为粒子群(CLQPSO)算法进行了研究,实现了一种融合混沌优化与完全学习策略的量子行为粒子群算法;通过在粒子群进化过程中对收缩扩张因子和局部吸引子的混沌优化,提高了敛速和精度.基于统一计算设备架构(CUDA),利用GPU的并行计算能力,将该算法并行化并应用到蛋白质二级结构预测中.实验表明:相比串行实现,该并行算法在对长度较短的残基序列进行蛋白质二级结构预测时,加速比可超过40.
In order to improve the efficiency of protein secondary structure prediction ,the Quantum behaved particle swarm with comprehensive learning strategy （CLQPSO） algorithm is studied ,this study implements the fusion of chaos op-timization and comprehensive learning strategy in Quantum behaved particle swarm algorithm .In the process of evolution of particle swarm ,the chaos optimization for contraction expansion factor and local attractor will improve the convergence speed and accuracy .Based on computer unified device architecture （CUDA ） ,the algorithm is parallelized by the GPU paral-lel computing capabilities and applied to the protein secondary structure prediction .Experiments showed that compared with serial implementation ,the speedup of the parallel algorithm can be more than 40 when we make protein secondary structure prediction on short residue sequences .
出处
《周口师范学院学报》
CAS
2014年第5期109-113,共5页
Journal of Zhoukou Normal University
基金
河南省教育厅自然科学研究计划资助项目(No.2010B520036)
关键词
混沌优化
完全学习策略
量子行为粒子群
从头预测法
CUDA
chaos optimization
comprehensive learning strategy
quantum behaved particle swarm
ab initio prediction
CUDA
