ABSTRACT
Multivariate normal distribution plays a pivotal role in the theory of multivariate statistical analysis. Besides mathematical tractability, there are other reasons for this phenomenon. Often the multivariate observations are at least approximately normally distributed. Even when the original data is not multivariate normal, due to the central limit theorem, sampling distributions of certain statistics can be approximated by normal distribution. The subsequent development of the theory of random matrices was brought about by theoretical and practical considerations. Furthermore, multivariate techniques depend upon functions of random matrices such as determinants, traces and characteristic roots. Random matrices are also widely used in experimental studies in various branches such as agriculture, anthropology, biology, cybernetics, economics, education, medicine, and psychology. In these studies the observed random phenomena often can be described by random matrices which include the dependence structure of the relevant random vectors.
