研究一般线性模型下广义最小二乘估计关于误差分布的稳健性,给出了误差分布的最大分布类,使得当误差向量的分布在此范围内变动时,广义最小二乘估计在广义均方误差准则下为一致最优估计.
Robustness of generalized least square estimator in terms of error distributions in general linear model was discussed. We explored the maximal class of distributions of error term, in which the GLSE possesses robustness, that is, the GLSE is the best linear estimator under the criterion of minimizing the generalized MSE matrix with the error distribution varying within the maximal class.