机构地区: 东北师范大学数学系
出 处: 《东北师大学报(自然科学版)》 2000年第3期13-17,共5页
摘 要: 考虑混合模型A ={ y,Xβ,Uζ,σ2 eV}和它的三个导出模型 ,其中X ,U =(U1:… :Uk)为已知设计阵 ,β为固定效应向量 ,ζ′=(ζ′1:… :ζ′k)为随机效应向量 ,且V≥ 0 ,R(X :U) R(V) .给出了可估函数Cβ在模型A和其导出模型下的最佳线性无偏估计 (BLUE)相等的充要条件 ,σ2 e 在不同模型下的最小范数二次无偏估计 (MINQUE)相等的充要条件 . Consider the mixed model A={y,Xβ,Uζ,σ 2 eV} and its three reduced models, where, X,U=(U 1:…:U k) are the design matrices. β is a vector of unknown parameter, ζ′=(ζ′ 1:…:ζ′ k) are random-effects; V is a nonnegative definite matrix and R(X:U)R(V) .In this paper, reduced necessary and sufficient conditions that BLUE of the estimable function Cβ is the same under model A and its reduced models, and obtain necessary and sufficient conditions for equivalent between minimum norm quadratic unbiased estimator (MINQUE) of σ 2 under model A and its reduced models.