Fourier-domain transfer entropy spectrum

TL;DR

The paper introduces the Fourier-domain transfer entropy spectrum, a new model-free causality metric that captures time-varying, component-specific causal relations in complex systems, especially useful for non-stationary and high-dimensional data.

Contribution

It presents a novel Fourier-domain transfer entropy spectrum that generalizes transfer entropy for detailed causality analysis in complex, non-stationary systems.

Findings

Validates the spectrum's effectiveness through parameter dependence tests.

Demonstrates significance testing and sensitivity analysis.

Applied successfully to neuroscience and oscillator data.

Abstract

We propose the Fourier-domain transfer entropy spectrum, a novel generalization of transfer entropy, as a model-free metric of causality. For arbitrary systems, this approach systematically quantifies the causality among their different system components rather than merely analyze systems as entireties. The generated spectrum offers a rich-information representation of time-varying latent causal relations, efficiently dealing with non-stationary processes and high-dimensional conditions. We demonstrate its validity in the aspects of parameter dependence, statistic significance test, and sensibility. An open-source multi-platform implementation of this metric is developed and computationally applied on neuroscience data sets and diffusively coupled logistic oscillators.