Granger-causality maps of diffusion processes
- 16 February 2016
- journal article
- research article
- Published by American Physical Society (APS) in Physical Review E
- Vol. 93 (2) , 022213
Abstract
Granger causality is a statistical concept devised to reconstruct and quantify predictive information flow between stochastic processes. Although the general concept can be formulated model-free it is often considered in the framework of linear stochastic processes. Here we show how local linear model descriptions can be employed to extend Granger causality into the realm of nonlinear systems. This novel treatment results in maps that resolve Granger causality in regions of state space. Through examples we provide a proof of concept and illustrate the utility of these maps. Moreover, by integration we convert the local Granger causality into a global measure that yields a consistent picture for a global Ornstein-Uhlenbeck process. Finally, we recover invariance transformations known from the theory of autoregressive processes.Keywords
Funding Information
- Grantová Agentura České Republiky (GA13-23940S)
This publication has 28 references indexed in Scilit:
- The MVGC multivariate Granger causality toolbox: A new approach to Granger-causal inferenceJournal of Neuroscience Methods, 2014
- TRENTOOL: A Matlab open source toolbox to analyse information flow in time series data with transfer entropyBMC Neuroscience, 2011
- Multivariate Granger causality and generalized variancePhysical Review E, 2010
- Assessing the strength of directed influences among neural signals using renormalized partial directed coherenceJournal of Neuroscience Methods, 2009
- Analyzing information flow in brain networks with nonparametric Granger causalityNeuroImage, 2008
- Synchronization as adjustment of information rates: Detection from bivariate time seriesPhysical Review E, 2001
- Measuring Information TransferPhysical Review Letters, 2000
- Measures of Conditional Linear Dependence and Feedback between Time SeriesJournal of the American Statistical Association, 1984
- Measurement of Linear Dependence and Feedback between Multiple Time SeriesJournal of the American Statistical Association, 1982
- Investigating Causal Relations by Econometric Models and Cross-spectral MethodsEconometrica, 1969