|
In order to facilate implementation and understanding of the ideas presented in the above book, the following MATLAB code is available. The code is arranged in six archives and is provided entirely "as-is". Please feel free to use this code for non-profit purposes, but only with appropriate acknowledgement (i.e. citation). All algorithhims are described in the above book, and in the references provided below. If you wish to use this code for any other purpose, please contact me by email.
As this is my own personal research code-base, it may well (i.e. will) be contaminated with numerous errors. Please report these and I'll do my best to correct them.
This version, current as of 17 March 2006 (pps.c has been updated to resolve some issues with RNG).
- dimension Correlation dimension estimation code. Algorithms for estimating the correlation dimension using the grassberger-Proccacia approach (corrint.m), the Gaussian-Kernel algorithm (gka.m) and Judd's estimator (judd.m) are provided. As always, estimate the correlation dimension at your own risk.
- dynsystems A collection of the usual suspects.
- invariants Several dynamic invariants and measures that may be useful. Included are algorithms for autocorrelation (acorr.m), complexity (complexity.m), false nearest neighbours (fnn.m), local nonlinear prediction error (nlpe.m), Shannon complexity (shannon.m) and the embedding window estimator (window.m).
- surrogates Surrogate data algorithms, including the usual approaches (shuffle.m and surrogate.m), the cycle shuffle version (cycleshuffle.m) and the PPS algorithm (pps.m and findrho.m).
- ts2 A complete modelling suite. Radial basis and neural network models with various embedding strategies and basis functions. See the the separate help file (read.me).
- utilities Need for most of the above, download along with your desired specific selections.
- Install instructions Downlaod and decompress each of the archives (as need), be sure to include the utilities archive. Place all directories on the matlab search path and mex any c-files (this only applies to dimension, surrogates and invariants). That's it.
Updated 17/3/06
Homepage:
http://small.eie.polyu.edu.hk/
Research Group Homepage:
http://www.eie.polyu.edu.hk/~cktse/NSR
Departmental Homepage:
http://www.eie.polyu.edu.hk
University Homepage:
http://www.polyu.edu.hk
|