ABSTRACT
In the recent years some attention has been focused on the estimation of multivariate mean with constrain. This problem was originally considered by R.A.Fisher a long time ago. It is recently focused again in the works of Efron (1978), Cox and Hinkley (1977), Kariya (1989), Kariya, Giri and Perron (1988), Perron and Giri (1990), Marchand and Giri (1993), Marchand (1994), Fourdrinier and Strawderman (1996), Fourdrinier and Onassou (2000) among others. The motivation behind it is primarily based on the observed fact that in the univariate normal population with mean µ and variance s2, s becomes large proportionally to µ so that |µ|/s remains constant. This is also evident in the multivariate observations. But in the multivariate case no well accepted measure of variation between the mean vector µ and the covariance matrix S is available.
