摘要
通过对米氏累积概率分布函数(CPDF)进行自定义函数拟合,设计了一种获取显式CPDF的方法。通过对比分析事件法和权值法两种光子追踪模式的效率,证明了两种方法的计算耗时与样本容量成线性关系。事件法的计算速度要优于权值法,但权值法的收敛性要明显优于事件法。在达到相同收敛标准的前提下,权值法的样本容量可以比事件法小1-2个数量级;权值法阈值选取对计算精度和计算时间的影响主要发生在当阈值大于计算对象1/100时。总体讲权值法的计算效率要优于事件法。
A method is designed for acquiring the explicit cumulative probability distribution function (CPDF) based on curve fitting with a user defined function. By comparing and analyzing the efficiencies of two photons tracking modes named event method and weight method respectively, the relationship between sample size and calculation time in the two modes is found to be linear. The calculation speed of event method is faster than that of weight method, but the convergence of weight method is better than that of event method obviously. To achieve the same convergence standard, the sample size of weight method required can even be 1-2 orders of magnitudes less than that of event method. The influence of threshold value on the calculation accuracy and time of weight method always happens when the threshold value is larger than one spercent of the computing object, and generally speaking, the calculation efficiencies of weight method are better than that of event method.
出处
《光学学报》
EI
CAS
CSCD
北大核心
2013年第8期282-286,共5页
Acta Optica Sinica
基金
国防预研基金(1010502020202)
关键词
散射
光子传输
抽样统计
各向异性散射
追踪模式
蒙特卡罗法
scattering
photons transmission
sampling statistics
anisotropic scattering
tracking modes
MonteCarlo method