|
Spring 2008
STATISTICS
COLLOQUIUM
Wednesday, February 27, 2008
3:30-4:00—Refreshments, Yost 327
4:00-5:00—Talk,
Yost 101
Manfred Denker, PhD
Visiting Professor, Department of Statistics
Case Western Reserve University
Professor, Institut fur Mathematische Stochastik
Georg-August-Universitat Gottingen, Germany Maximum likelihood estimation for random matrix ensembles
Random matrix theory has been attracting attention of researchers in
different branches of science since it was introduced by Wishart in the
late 1920s. Nowadays, it has been successfully applied to an
extraordinary large variety of problems in fields as diverse as
multivariate statistics, harmonic analysis on groups, combinatorics,
nuclear physics, quantum gravity and wireless Communication.
Although H. Weyl was the first to encounter random matrices, Wishart
and later Wigner examined distributional properties of spectra of
random matrices, which are good models to describe statistical
properties in nuclear physics. These models are called random matrix
ensembles.
The talk presents results from Rada Matic' thesis on the
estimation problem of parameters in these models. This leads to new
asymptotic phenomena in maximum likelihood estimation, in particular
it will be shown that the speed of convergence of MLE is of the order
root(n) faster than ordinary MLE
|
