计算物理 ›› 2010, Vol. 27 ›› Issue (5): 733-739.

• 研究论文 • 上一篇    下一篇

计算动态系统网格通量场的驿站重要抽样方法

王瑞宏1, 姫志成1, 江松1, 黄正丰1, 裴鹿成2   

  1. 1. 北京应用物理与计算数学研究所, 北京 100094;
    2. 中国原子能科学研究院, 北京 102413
  • 收稿日期:2009-07-10 修回日期:2010-01-16 出版日期:2010-09-25 发布日期:2010-09-25
  • 作者简介:王瑞宏(1967-),男,江苏,研究员,博士,从事蒙特卡罗方法及其应用研究,北京8009(16)信箱100094.
  • 基金资助:
    国家自然科学基金(10771018);中国工程物理研究院科学技术发展基金(2008B0202018)资助项目

Time Station Importance Sampling Method for Flux Field Calculation of Time-dependent Problem

WANG Ruihong1, JI Zhicheng1, JIANG Song1, HUANG Zhengfeng1, PEI Lucheng2   

  1. 1. Institute of Applied Physics and Computational Mathematics, Beijing 100094, China;
    2. China Institute of Atomic Energy, Beijing 102413, China
  • Received:2009-07-10 Revised:2010-01-16 Online:2010-09-25 Published:2010-09-25

摘要: 对与时间有关输运问题计算条件下,全系统不同网格体通量计算统计涨落较大,甚至个别网格很难得到计数的现象,提出驿站重要抽样方法.研究可以选择的重要性函数,给出抽样方法,推导重要性偏倚抽样的源粒子纠偏权重公式.经过推导证明连续多时间步抽样计算不存在粒子纠偏权发散问题.数值模拟结果表明,采用驿站重要抽样方法,全系统精细网格通量场计算精度基本一致,整体计算结果得到明显改善.

关键词: Monte Carlo方法, 精细网格通量场, 统计涨落, 时间驿站方法, 驿站重要抽样方法

Abstract: Flux field calculation in time-dependent problems with Monte Carlo method and associated particles transportation equation are described.The random walk principle of time station sampling(TSS) method in dynamic system is introduced.A time station importance sampling(TSIS) method is developed and applied to decrease statistical fluctuation in global flux field calculation.Alternative importance function and its unbiased weight of modification are investigated.Numerical results and analyses on methods in specific model are given.The time station importance sampling(TSIS) biased method can be used to control statistical variance of global fine mesh flux adaptively in a same level.

Key words: Monte Carlo method, flux field, statistical fluctuation, time station sampling (TSS), time station importance sampling (TSIS)

中图分类号: