We provide an almost sure convergent expansion of fractional Brownian motion in wavelets which decorrelates the high frequencies. Our approach generalizes L~vy's midpoint displacement technique which is used to generate Brownian motion. The low-frequency terms in the expansion involve an independent fractional Brownian motion evaluated at discrete times or, alternatively, partial sums of a stationary fractional ARIMA time series. The wavelets fill in the gaps and provide the necessary high frequency corrections. We also obtain a way of constructing an arbitrary number of non-Gaussian continuous time processes whose second order properties are the same as those of fractional Brownian motion.Math Subject Classifications. Primary 60G18; secondary 41A58, 60F15.
Abstract-The aim of this communication is to propose some complementary remarks and interpretation on the waveletbased synthesis technique for fractional Brownian motion proposed by Sellan in 1995. These comments will lead us to propose a fast and efficient pyramidal filter bank-based Mallattype algorithm, which permits an easy and efficient implementation of this synthesis technique. c 1996 Academic Press, Inc. MOTIVATIONFractional Brownian motion. Fractional Brownian motion (hereafter fBm) is a continuous-time random process proposed by Mandelbrot and Van Ness [11]. Basically, it consists in a fractional integration of a white Gaussian process and is therefore a generalization of Brownian motion (as defined by P. Lévy), which consists simply in a standard integration of a white Gaussian process. Because it presents deep connections with the concepts of self-similarity, fractal, long-range dependence or 1/f-processes, fBm quickly became a major tool for the various fields where such concepts are relevant. Many efforts have therefore been devoted to the possibility of performing numerical simulation for such a process (for a review, see [14]). None of these methods, however, was able to produce a process that possesses all the properties of fBm. Very recently, Sellan proposed [14, 15] a powerful wavelet-based analysis of fBm which also provides us with a general scheme to synthesize it.Scope of the communication. Our aim here is to propose a fast and efficient implementation of this synthesis technique that relies on the use of a fast filter bank-based Mallat-type pyramidal algorithm as well as to propose further remarks that give complementary viewpoints on this technique. While the next section restates the main ideas of the construction and theorems presented in [14], Section 3 clearly details how to derive the coefficients of the filter bank involved in the fast pyramidal algorithm. Section 4 addresses both practical issues and interpretation questions. Matlab routines can be obtained at the ACHA software ftp site, and are available upon request. WAVELET-BASED SYNTHESIS FOR fBmDefinition. Let us first recall the commonly used definition for fBm [11],where where σ 2 = Γ(1 − 2H) cos πH πH .377
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.