Transfer entropy

Primary reference: Schreiber (2000) Phys. Rev. Lett. 85, 461.

Transfer entropy TE(X→Y) quantifies information flow from X to Y beyond Y’s own history. It is asymmetric: directed coupling should yield TE(X→Y) > TE(Y→X).

Assumptions

  • Discrete binning or kNN estimation; bin count affects magnitude.

  • Series should be long enough for stable density estimates.

  • Coupled logistic maps in synthetic_causal validate asymmetry (transfer_entropy_coupled_maps).

API

  • ts2net.causal.transfer_entropy.transfer_entropy()

  • ts2net.causal.transfer_entropy.transfer_entropy_network()

  • ts2net.causal.transfer_entropy.conditional_transfer_entropy()