Adaptive wavelet based estimator of long-range dependence parameter for stationary Gaussian processes
Résumé
The aim of this contribution is to provide an adaptive estimation of the long-memory parameter in the classical semi-parametric framework for Gaussian stationary processes using a wavelet method. In particular, the choice of a data-driven optimal band of scales is introduced and developed. Moreover, a central limit theorem for the estimator of the long-memory parameter reaching the minimax rate of convergence (up to a logarithm factor) is established. Simulations confirm the quality of this estimator.
Origine : Fichiers produits par l'(les) auteur(s)