Convergence of quadratic forms for random fields and its application to the convergence of empirical covariances
Résumé
Limit theorems are proved for quadratic forms of Gaussian random fields in presence of long memory. Similarly to the one dimensional case, we prove that the quadratic forms, appropriately normalized, may have Gaussian or non-Gaussian limits. However the dichotomy observed in $d=1$ cannot be stated so easily, due to the possible occurrence of anisotropic strong dependence in $d>1$. We apply our theorems to obtain the asymptotic behavior of the empirical covariances, which is a particular example of quadratic forms.
Origine : Fichiers produits par l'(les) auteur(s)