Local Granger causality

Granger causality (GC) is a statistical notion of causal influence based on prediction via linear vector autoregression. For Gaussian variables it equivalent to transfer entropy, an information-theoretic measure time-directed information between jointly dependent processes. We exploit such equivalence and calculate exactly the local causality, i.e., profile transferred from driver target process at each discrete time point; in this frame, GC average its version. show that variability around mean relates interplay innovation (autoregressive noise) processes, may reveal transient instances not detectable values. Our approach offers robust computationally fast method follow along history stochastic as well nonlinear complex systems studied approximation.