Software

Code for Simulation and Estimation in Various Stochastic Models

Fractional Brownian field (H=0.3)
Anisotropic Fractional Brownian field with self-similarity parameter H=0.3 and discrete spectrum.
Fractional Brownian field (H=0.7)
Anisotropic Fractional Brownian field with self-similarity parameter H=0.7 and discrete spectrum.
Fractional Brownian field (H=0.9)
Anisotropic Fractional Brownian field with self-similarity parameter H=0.9 and discrete spectrum.

 

Long Range Dependence

  • LASS: Local Analysis of Self-Similarity – a collection for the estimation of the Hurst long-range dependence parameter as a function of “time”. Tailored and intendend for the analysis of Internet traffic data.
    Note: This code has not been updated recently.
  • FFT-sim: FFT-based tools for the efficient simulation of LFSM and FARIMA with stable innovations.
  • ffgn.m: An routine for exact and efficient simulation of fGn (fractional Gaussian noise) time series. Hence, exact discretized paths of fBm – fractional Brownian motion – can be generated. The code uses circulant embedding and is written jointly with Yingchun Zhou (Jasmine), zhouyc@math.bu.edu.
  • Hest: Wavelet based tools for the estimation of the Hurst LRD parameter H.

Extremes

  • max spectrum: An estimator for the heavy tail exponent of a distribution based on scaling of block-maxima. Incorporates a method for the automated selection of the extreme cut-off.
    Note: Main routine maxss_3.m.
  • extremal index: An estimator for the extremal index based on the max spectrum and resampling.
    Note: Main routine max_eind_1.m.
  • max linear: an R-package for exact conditional sampling from max-linear models courtesy of Yizao Wang.