MPS-LN 2002-12
August 2002

Long Range Dependence, Heavy Tails and Rare Events

by:

Gennady Samorodnitsky

Abstract

These notes were prepared for a Concentrated Advanced Course at the University of Copenhagen in the framework of the MaPhySto program. The notes present an attempt to propose a new approach to long range dependence, one that is not mainly based on correlations, or on (approximate) self-similarity. Rather, we advocate a point of view that regards the passage between short memory and long memory and a phase transition in the way certain rare events happen. We are especially interested in the heavy tailed case. These notes bring together ideas from large deviations, theory of heavy tailed processes, extreme value theory and relate those to the notion of memory. Main examples considered are those related to communication networks and risk theory.

Availability: [ gzipped ps-file ] [ pdf-file ]

[ Help on down-loading/viewing/printing ]