Multifractal processes.

From MaRDI portal





This is a complete updating survey on multifractal processing of stochastic motions. The first part presents the toolkit for the analysis of the paths of a process, and thus also of functions. Path-free versions of the multifractal tools (using suitable averages) are also introduced. Different notions of exponents sentinelling the local singularity of paths are introduced, in particular Hölder and wavelet singularity exponents. Several fine and grain (Hausdorff) \textit{dimensional spectra} associated with a singularity exponent, together with the related \(L^q\)-\textit{spectra} obtained from the scaling of the \(q\)-th moments of a partition function, are defined. Exact results on the connection -- via Legendre transforming -- between both types of multifractal decompositions -- the so-called \textit{multifractal formalism} -- are formulated and proved. The second part discusses in detail the multifractal analysis of significant processes, e.g., binomial cascades, and statistical self-similar processes (fractional Brownian motion, Lévy motion). Wavelet-based multifractal analysis receives specific attention here. A recent class of processes with rich multifractal structure is introduced. They may be synthesized by compounding a self-similar (mono-fractal) process with a fully structured multifractal process for time change. Estimation issues and modelling features are considered throughout the manuscript. Perspectives are addressed in a closing section. For the entire collection see [Zbl 1005.00017].




Cited in
(57)








This page was built for publication: Multifractal processes.

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4407625)