The holonomic gradient method for the distribution function of the largest root of a Wishart matrix (Q391611): Difference between revisions

The explicit evaluation of a confluent hypergeometric function of matrix argument is the topic of this paper. The case when the argument matrix is real, positive definite and diagonal is considered. The case, when the argument matrix is \(2\times 2\) and diagonal, is discussed in detail. The basic idea used is to create the corresponding partial differential equations and to solve them. The partial differential equations are simplified by using functional symmetry assumption of the hypergeometric function of matrix argument and conveniently rewritten to use the author's method called holonomic gradient method. Computer algorithms are also given for the \(2\times 2\) case. It is shown that the method enables one to compute terms beyond what the zonal polynomial approach can achieve. The results in this paper are useful when computing percentage points of densities which can be written in terms of a confluent hypergeometric function of matrix argument or to compute real matrix-variate incomplete gamma integrals etc.