摘要
本文提出了一种新的设计和抽样方法－正交表型均匀 ＬＨ设计和抽样．证明了这种抽样空间是 ＯＡＬＨ抽样空间的优良子集，这种设计和抽样空间中所有样本都与初始设计具有同阶低偏差等一些优良性质．并将它用于数值积分，证明了对有关参数估计的方差阶低于其他抽样．同时还给出了有关的模拟结果．
In this paper, we propose a new method of design and sampling, OA-based uniform Latin hypercube design and sampling. Some fine properties of the sampling, such as its sampling space being a fine subspace of OALH space, are proved and the sampling is applied to numerical integration, it is shown that its variance or mean squared error order of convergence is superior in the existing sampling methods. Also, we give some simulation results.
出处
《应用概率统计》
CSCD
北大核心
2001年第3期243-254,共12页
Chinese Journal of Applied Probability and Statistics
基金
国家自然科学基金资助项目 19771049
高校博士点基金资助项目1999005512.
