非定常粒子输运蒙特卡罗自适应并行计算  

THE ADAPTIVE PARALLEL COMPUTATION OF UNSTEADY-STATE MONTE CARLO PARTICLE TRANSPORT

作　　者：邓力[1] 袁国兴[1] 黄正丰[1] 许海燕[1] 王瑞宏[1] 李树[1] 

机构地区：[1]北京应用物理与计算数学研究所计算物理实验室

出　　处：《数值计算与计算机应用》2003年第2期111-115,共5页Journal on Numerical Methods and Computer Applications

基　　金：中国工程物理研究院重点基金资助

摘　　要：The Monte Carlo method is used to simulate unsteady-state particle transport calculations. Due to the computational time existing obviously difference among the different steps and I/O percentage is rated high, it results in the parallel com-putation efficiency down if the fix processors are used in each step. So parallel I/O and the adaptive parallel algorithms are developed. The good results and high speedup are obtained.The Monte Carlo method is used to simulate unsteady-state particle transport calculations. Due to the computational time existing obviously difference among the different steps and I/O percentage is rated high, it results in the parallel computation efficiency down if the fix processors are used in each step. So parallel I/O and the adaptive parallel algorithms are developed. The good results and high speedup are obtained.

关 键 词：BOLTZMANN方程 蒙特卡罗模拟 自适应并行计算 非定常粒子输运 迭代法 MPI系统 振荡型 稳步型 并行测试 

分 类 号：O246[理学—计算数学]