Open AccessA NEW METHOD OF SURROGATE DATA TEST FOR LINEAR NON-GAUSSIAN TIME SERIES
Author(s) -
Yaozong Liu,
Xisen Wen,
Hu Niao-Qing
Publication year - 2001
Publication title -
wuli xuebao
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.199
H-Index - 47
ISSN - 1000-3290
DOI - 10.7498/aps.50.1241
Subject(s) - surrogate data , series (stratigraphy) , gaussian , gaussian process , algorithm , computer science , spectral density , cross spectrum , mathematics , gaussian noise , nonlinear system , statistics , frequency domain , physics , paleontology , quantum mechanics , biology , computer vision
Surrogate data testing is a popular method to detect nonlinearity and chaos in time series and has been widely used in many applications with erratic time series.With the explicit null hypothesis that the time series is generated from a linear,stochastic,Gaussian stationary process,the surrogate data test based on phase randomization may give false alarm for nonlinearity at a linear non-minimum phase non-Gaussian sequence.So,we propose a new method to test the hypothesis of linear non-Gaussian process in light of typical realization of surrogates.With the ARMA parameters estimated from high-order cumulants and the series itself,a method to estimate the input noise of a non-minimum phase sequence is developed based on power spectrum equivalence,which is the bottle-neck to generate surrogates for non-minimum phase time series.The results of numerical experiments confirm our approach to test non-minimum phase non-Gaussian process.