摘要
§1.模型与问题 考虑如下的生长曲线模型其中X1,X2,u≠0分别为n×k,p×l,n×s阶已知矩阵;B为k×l阶回归系数矩阵。Y=（y（1）,…,y（n））′与（?）=（ε（1）,…,ε（s））′分别为n×p阶观测资料矩阵及s×p阶随机误差矩阵。将Y、B及（?
Consider a growth curve model as follows:Y=X_1BX_2+U,where ε has a quasi-normal distribution,Eε=0,Eεε′=IΣ,and Σ≥0 is an unknown covariance matrix.This paper gives the n.s.condition for Uniformly Minimum Variance Nonne-gative Quadratic Unbiased Estimator(UMVNNQUE)of tr (CΣ)to be existent,and gives UMVNNQUE of tr (CΣ) when it exists,where C≥0(≠0) is an arbitrarily given matrix.As corollaries,this paper also gives the n.s.condition for tr(CΣ) to be UMVNNQUE of tr (CΣ),where Σ~* is Σ′s LSE in a certain sense.
出处
《应用数学学报》
CSCD
北大核心
1992年第1期83-98,共16页
Acta Mathematicae Applicatae Sinica
