LJKProbability & Statistics Seminar

On Thursday June 5 2003 at 14h00 in Salle 1  Tour IRMA

Seminary of Raj J. BHANSALI (University of Liverpool)

Chaotic Maps with Slowly Decaying Correlations and Intermittency

Summary

Established stochastic models for the class of discretetime longmemory stationary processes are linear and Gaussian. The chaotic intermittency maps provide alternative nonlinear, nonGaussian models for this class of processes. A formal algebraic definition of the intermittency maps, which allows for a distinction to be made between the regular and the cusp families of maps, is given. Examples of intermittency maps belonging to both these families are given, and their properties, especially the invariant density, correlation function, and the recurrence behaviour, are discussed. A discussion is also given of how intermittency maps are applied for modelling the Ethernet packet traffic.
