关于hspice的montecarlo分析的询问

huangrenjun

                               
登录/注册后可看大图

我想请问一下.ac dec 1 1 100meg sweep monte=10 lhs_sample_size=10我想使用latin hypercube sampling分析的时候，这个lhs_sample_size参数的意义是什么，所谓的采样大小不是很理解具体指的是什么，经过实验sample的次数只和monte=x有关，想请教一下大家对这个参数的理解。

JeffChen1688

lhs_sample_size might be numbers of samples using lhs
In general, lhs can have higher accuracy. It means using ihs can have fewer samples to get the same accuracy.

https://analytica.com/blog/latin ... nte-carlo-sampling/
Here, the sampling error for MC is 20 times larger than for RLHS. Increasing sample size, the sampling error of 0.005 at a sample size of N=10,000—a one-hundred-fold increase in the number of samples. Published theoretical results for the univariate show that the sampling error of Monte Carlo goes down as O( 1/sqrt(N) ), [2], whereas the sampling error for LHS is O( 1/N ), quadratically faster, for almost all distributions and statistics in common use [3], [5], [7]. In other words, if you need N samples for a desired accuracy using LHS, you’ll need N2 samples for the same accuracy using MC.