Testing for nonlinearity in time series: the method of surrogate data

Abstract

Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or o_herwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any ager, cy thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.By acceplance of mta article. Thepubhsher rec_gn_-"eS that the U.S Government retains a nonexcluslve, royalty-free hcenlm1opubhll_ or reproduce tr_e0ubhSr_edform Of Ih,S conlr_buhOn,or tO allow omer$ to do so. for U S Government purposes T_e Los Alamos National Laboratory requests thal the 10ubllsheridefltlfy this art,cle as work performed under the ouspfceI of the U S Depo_menl at Energy ASTEB Lo8LosAa os a, ona a0o a,or ABSTRACTWe describe a statistical approach for identifying nonlinearity in time series; in particular, we want to avoid claims of chaos when simpler models (such as linearly correlated noise) can explain the data. The method requires a careful statement of the null hypothesis which characterizes a candidate linear process, the generation of an ensemble of "surrogate" data sets which are similar to the original time series but consistent with the null hypothesis, and the computation of a discriminating statistic for the original and for each of the surrogate data sets. The idea is to test the original time series against the null hypothesis by checking whether the discriminating statistic computed for the original time series differs significantly from the statistics computed for each of the surrogate sets. We present algorithms for generating surrogate data under various null hypotheses, and we show the results of numerical experiments on artificial data using correlation dimension, Lyapunov exponent, and forecasting error as discriminating statistics.Finally, we consider a number of experimental time series m including sunspots, electroencephalogram (EEG) signals, and fluid convection --and evaluate the statistical significance of the evidence for nonlinear structure in each case.

Principal components of functional connectivity: A new approach to study dynamic brain connectivity during rest

“…We phase randomized both brain activity time courses before estimating dynamic FC (Handwerker et al, 2012), and connectivity time courses estimated from original activity (Allen et al, in press) in separate simulations. Specifically, we Fourier transformed each subject's activity or connectivity time courses, randomized the phases independently for each brain region or connectivity pair, and transformed back to the time domain using the amplitude of the real data but randomized phases (Theiler et al, 1992). Phase randomizations preserve the mean, variance and autocorrelation properties of the time courses, while randomizing their precise timing.…”

Prediction of hypertensive crisis based on average, variability and approximate entropy of 24-h ambulatory blood pressure monitoring

“…Surrogate time series are generated by phase randomizing the original time series. 16 The result is a purely random time series with properties (for example, mean, variance) similar to the original time series. In our data, surrogate time series were derived from each original time series of the ABPM, and ApEn was calculated for the surrogate time series.…”

Determining significant connectivity by 4D spatiotemporal wavelet packet resampling of functional neuroimaging data

“…Since the variance distribution is not necessarily the same as that of the original data, we propose an extension to the Theiler et al (1992) amplitude adjustment step which adjusts the data to approximate the variance distribution of the original. However, use of this extension requires the additional assumption of normality of each voxel time series.…”

“…Otherwise, we propose an extension to an amplitude adjustment step developed by Theiler et al (1992).…”

“…Thus, the variability of each intra-cranial voxel in the surrogate data is on average smaller than that of the original data as the number of voxels for which the energy is spread is much larger. We extend a 1D amplitude adjustment algorithm proposed by Theiler et al (1992) to a 4D lattice and address when it is appropriate to use such an algorithm. Finally, we apply our resampling technique to determine significant functional connectivity from resting state and motor task fMRI datasets.…”