Etude des Théorèmes limites sous des conditions de dépendance faible.

Abstract

In this thesis, we are interested in the asymptotic behavior of a class of stochastic processes with d-Φ-subgaussian increments and weighted sums of random variables with infinite means. For the first family of processes, the Dudley metric entropy criterion is used to establish a law of iterated logarithm. For the second family, a weak law of large numbers is proved in a unified framework under weak dependence conditions. A necessary condition for the validity of this law is also derived