Spectral Analysis of Multi-dimensional Self-similar Markov Processes

TL;DR

This paper develops spectral analysis methods for discrete scale invariant Markov processes, providing spectral density functions and matrices, with an example using Brownian motion, to characterize their covariance structures.

Contribution

It introduces a spectral representation framework for discrete scale invariant Markov processes and derives explicit spectral density matrices for such processes.

Findings

Spectral density functions are derived for DT-SI and DT-SIM processes.

The spectral density matrix fully characterizes the process with specific covariance parameters.

An example with Brownian motion illustrates the theoretical results.

Abstract

In this paper we consider a discrete scale invariant (DSI) process $\{X(t), t\in {\bf R^+}\}$ with scale $l>1$. We consider to have some fix number of observations in every scale, say $T$, and to get our samples at discrete points $\alpha^k, k\in {\bf W}$ where $\alpha$ is obtained by the equality $l=\alpha^T$ and ${\bf W}=\{0, 1,...\}$. So we provide a discrete time scale invariant (DT-SI) process $X(\cdot)$ with parameter space $\{\alpha^k, k\in {\bf W}\}$. We find the spectral representation of the covariance function of such DT-SI process. By providing harmonic like representation of multi-dimensional self-similar processes, spectral density function of them are presented. We assume that the process $\{X(t), t\in {\bf R^+}\}$ is also Markov in the wide sense and provide a discrete time scale invariant Markov (DT-SIM) process with the above scheme of sampling. We present an example of DT-SIM process, simple Brownian motion, by the above sampling scheme and verify our results. Finally we find the spectral density matrix of such DT-SIM process and show that its associated $T$-dimensional self-similar Markov process is fully specified by $\{R_{j}^H(1),R_{j}^H(0),j=0, 1,..., T-1\}$ where $R_j^H(\tau)$ is the covariance function of $j$th and $(j+\tau)$th observations of the process.