Acta mathematica scientia,Series A ›› 2010, Vol. 30 ›› Issue (6): 1621-1628.

• Articles • Previous Articles     Next Articles

Admissibility of Matrix Nonhomogeneous Linear Estimators in Multivariate Linear Models

 ZHOU Zai-Ying   

  1. College of Mathematics and Computer Science, |Anhui Normal University, Anhui Wuhu 241000
  • Received:2008-07-08 Revised:2009-06-06 Online:2010-12-25 Published:2010-12-25
  • Supported by:

    安徽师范大学青年基金(2008xqn51)资助

Abstract:

This paper studies the admissibility of matrix nonhomogeneous linear estimators of an estimable parameter matrix linear function in multivariate linear models with completely unknown covariance matrix under a quadratic matrix loss function. A necessary and sufficient condition is given for a matrix nonhomogeneous linear estimator to be admissible in the class of all matrix linear estimators without normality assumption. Under normality assumption, a sufficient condition that a matrix nonhomogeneous linear estimator is admissible in the class of all matrix estimators is derived.

Key words: Admissibility, Multivariate linear models, Nonhomogeneous linear estimators, Class of all linear estimators, Class of all matrix estimators, Quadratic matrix loss functions

CLC Number: 

  • 62C15
Trendmd