Admissibility of Elliptically Contoured Estimators of the Mean of a Multivariate Normal Distribution with Generalized Squared Error Loss.

摘要

Let X be a p-variate (p >or = 3) random vector normally distributed with mean vector theta and unknown positive definite covariance matrix sigma. Let A be a p x p Wishart matrix with a parameters (n,sigma), n >or = p,independent of X. The paper considers the problem of estimating theta with respect to a generalized squared error. (Modified author abstract)