An extension of the Tusnady inequality for general distributions
Résumé
Under mild assumptions an analog of Tusnady's inequality for conditional distributions of partial sums of independent nonidentically distributed r.v.'s is obtained. These type of inequalities provide very sharp bounds for couplings of random variables via quantile transformations, which are main tools for proving various strong invariance principles for sums of independent random variables.
Domaines
Probabilités [math.PR]
Origine : Fichiers produits par l'(les) auteur(s)