Statistics Seminar: "Some models of scale invariance and wavelet-based estimation in a multivariate setting"

Abstract: Scale invariant processes arise naturally in statistical contexts due to their role in various limit theorems, and in the past few decades they have been successfully applied in a wide variety of areas ranging from internet traffic modeling to climatology. However, by comparison to univariate settings, there is a relative scarcity of scale-invariance models that are characteristically multivariate, i.e. that display the scaling behavior the limit theory arguably suggests is most natural.

In this talk I will introduce a new class of scale invariance models called operator fractional Lévy motion that exhibits some characteristically multivariate scaling features, and discuss some challenges regarding wavelet-based estimation of scaling exponents in the multivariate setting.  This is joint work with Gustavo Didier (Tulane University).

Host: Jose Figueroa-Lopez