207

Complete Asymptotic Expansions and the High-Dimensional Bingham Distributions

Test (Madrid) (TM), 2023
Abstract

Let XX denote a random vector having a Bingham distribution on Sd1{\mathcal{S}}^{d-1}, the unit sphere centered at the origin in Rd{\mathbb{R}}^d. With Σ\Sigma denoting the symmetric matrix parameter of the distribution, let Ψ(Σ)\Psi(\Sigma) be the corresponding normalizing constant of the distribution. We derive for Ψ(Σ)\Psi(\Sigma) and its first-order partial derivatives complete asymptotic expansions as dd \to \infty. These expansions are obtained under the growth condition that Σ\|\Sigma\|, the Frobenius norm of Σ\Sigma, satisfies Σ=O(dr/2)\|\Sigma\| = O(d^{r/2}), where 0r<10 \le r < 1. As a consequence, we obtain for the covariance matrix of XX an asymptotic expansion up to terms of arbitrary degree in Σ\Sigma.

View on arXiv
Comments on this paper