线性模型中不可估参数的线性估计的可容许性

Admissible Linear Estimates of Nonestimatible Parameter Functions in a Linear Model

对一般线性模型在平方损失函数下，得到了一维不可估参数函数的线性估计为可容许估计的充要条件，以及模型中参数向量（非线性可估）的线性估计为可容许估计的两个充要条件，并得到了多维参数函数（可估或不可估）的线性估计为可容许估计的一个充分条件以及特殊情况下的一个充要条件．

Under a general linear model and the square loss function, we obtain the necessary and sufficient conditions of linear admissible estimates of one-dimensional nonestimatible parameter function and those of the nonestimatible patameter vector respectively. In addition, a sufficient condition of linear admissible estimates of arbitrary multidimensional parameter functions is given and the necessary and sufficient condition is shown in a special case.