Introduction

StationaryRandomFields.jl simulates realistic correlated noise for signal data of any given dimensions. The package follows the power-law noise procedure introduced by Timmer et al. (1995); random Gaussian noise is drawn in the Fourier frequency domain and scaled by the square root of a power law spectrum. An inverse transform back to the signal domain gives the stochastic power-law noise.

The module currently

• Provides abstract data types and methods to define noise signals, construct and modify power-law scaling functions, and generate signal noise

• Implements multiple power spectrum types:

  • Basic spectra of form $P={\nu }^{\alpha }$ (SinglePowerLaw)

  • Cut-off spectra with inner and outer scales (SaturatedPowerLaw)

• Provides methods to reverse the process and retrieve underlying power spectra from input signal noise

Installation

The package can be installed by running:

using Pkg
Pkg.add("StationaryRandomFields")