两两NQD序列下线性指数分布参数的经验Bayes检验  

Empirical Bayes Test for Parameters of the Linear Exponential Distribution with NQD Random Samples

作　　者：王亮[1] 师义民[1] 

机构地区：[1]西北工业大学应用数学系,西安710072

出　　处：《工程数学学报》2010年第4期599-604,共6页Chinese Journal of Engineering Mathematics

基　　金：国家自然科学基金(70471057);陕西省教育厅自然科学基金(03JK065)~~

摘　　要：本文讨论了线性指数分布参数的经验Bayes单边检验问题。对于两两NQD样本序列,利用核函数估计方法给出了检验函数。证明了该检验函数是渐近最优的,并在适当的条件下得到其收敛速度。In this paper, the one-sided test problem for parameters of the linear exponential families was investigated through the empirical Bayes （EB） approach. The EB test function was constructed by using the kernel-type estimation for Negative Quadrant Dependent （NQD） samples, and the convergence rates for the EB test function was obtained under suitable conditions. The results show that the EB test function is asymptotically optimal.

关 键 词：两两NQD序列 核估计 经验BAYES检验 渐近最优性 收敛速度 

分 类 号：O212.1[理学—概率论与数理统计]