8
9

A new test of multivariate normality by a double estimation in a characterizing PDE

Abstract

This paper deals with testing for nondegenerate normality of a dd-variate random vector XX based on a random sample X1,,XnX_1,\ldots,X_n of XX. The rationale of the test is that the characteristic function ψ(t)=exp(t2/2)\psi(t) = \exp(-\|t\|^2/2) of the standard normal distribution in Rd\mathbb{R}^d is the only solution of the partial differential equation Δf(t)=(t2d)f(t)\Delta f(t) = (\|t\|^2-d)f(t), tRdt \in \mathbb{R}^d, subject to the condition f(0)=1f(0) = 1. By contrast with a recent approach that bases a test for multivariate normality on the difference Δψn(t)(t2d)ψ(t)\Delta \psi_n(t)-(\|t\|^2-d)\psi(t), where ψn(t)\psi_n(t) is the empirical characteristic function of suitably scaled residuals of X1,,XnX_1,\ldots,X_n, we consider a weighted L2L^2-statistic that employs Δψn(t)(t2d)ψn(t)\Delta \psi_n(t)-(\|t\|^2-d)\psi_n(t). We derive asymptotic properties of the test under the null hypothesis and alternatives. The test is affine invariant and consistent against general alternatives, and it exhibits high power when compared with prominent competitors.

View on arXiv
Comments on this paper