关于奇异线性模型最好线性无偏估计的计算

Computation of best linear unbiased estimation of Bizarre linear model

讨论了奇异线性模型最好线性无偏估计(BLUE)的计算问题。利用分块求逆的技术,给出BLUE的一种新的表达式,其计算量比直接利用C.R.Rao所给公式的小。对新表达式的计算,给出半正定矩阵广义逆的分块求逆算法及消去变换算法。

The computation of the best linear unbiased estimators in the singular linear model is discussed. By use of the generalized inverse of the partitioned matrices a new expression of the BLUE is given. The computation is simpler than C.R.Rao's. To the nonnegative definite matrix in the new expression the algorithm of the generalized inverses of the partitioned matrices and that of the cancellation alternation of the matrix are given.