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

TL;DR

This paper introduces a novel test for multivariate normality based on a double estimation approach involving a PDE characterization of the characteristic function, demonstrating high power and affine invariance.

Contribution

It proposes a new affine-invariant test for multivariate normality using a weighted L2-statistic derived from a PDE characterization of the characteristic function.

Findings

The test is consistent against general alternatives.

It has high power compared to existing tests.

The asymptotic properties under null and alternative hypotheses are established.

Abstract

This paper deals with testing for nondegenerate normality of a $d$-variate random vector $X$ based on a random sample $X_1,\ldots,X_n$ of $X$. The rationale of the test is that the characteristic function $\psi(t) = \exp(-\|t\|^2/2)$ of the standard normal distribution in $\mathbb{R}^d$ is the only solution of the partial differential equation $\Delta f(t) = (\|t\|^2-d)f(t)$, $t \in \mathbb{R}^d$, subject to the condition $f(0) = 1$. By contrast with a recent approach that bases a test for multivariate normality on the difference $\Delta \psi_n(t)-(\|t\|^2-d)\psi(t)$, where $\psi_n(t)$ is the empirical characteristic function of suitably scaled residuals of $X_1,\ldots,X_n$, we consider a weighted $L^2$-statistic that employs $\Delta \psi_n(t)-(\|t\|^2-d)\psi_n(t)$. 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.