Transfer Entropy & Granger Causality

Transfer entropy T(X→Y) = I(Y_future; X_past | Y_past) quantifies directed information flow from X to Y beyond Y's own history. It generalizes Granger causality to nonlinear systems. Here, a driver signal X causally influences a target Y with adjustable coupling and lag — watch asymmetric information flow emerge.

T(X→Y) bits
T(Y→X) bits
Net flow X→Y
0.50
2
0.30