摘要
考虑方差分量模型EY=Xβ,CovY=Vi,其中n×p矩阵X和非负定矩阵Vi(i=1,2,…,m)都是已知的.β∈R^P,θi>0或θi>0(i=1,2,…,m)均为参数.设Sβ是线性可估的.在本文中,我们在二次损失下,当V=>0时,给出了关于Sβ的线性估计在线性估计类中可容许的充要条件.并运用此结果,得到了共同均值模型中Sβ的线性估计在线性估计类中可容许的充要条件.
Consider the variance component model EY=Xβ,CovY =whereX: n ×p and Vi≥0 (i=1, 2,…,m) are known.β∈R^p,θi≥0 or θi>0,(i=1, 2,…,m)are parameters. Let Sβ be linearly estimable. In this paper, when V=>0, thesufficient and necessary conditions for a linear estimator of Sβ to be admissible in the classof all linear estimators are given under quadratic loss function. Applying this result to acommon mean model, we obtain the sufficient and necessary conditions for a linear estimatorof Sβto be admissible in the class of all linear estimators.
出处
《应用数学学报》
CSCD
北大核心
1998年第3期393-403,共11页
Acta Mathematicae Applicatae Sinica
基金
国家自然科学基金
关键词
方差分量模型
二次损失
线性容许估计
回归系数
Variance component model, quadratic loss function,linear admissible estimator, common mean model