Title: On Causality in Dynamical Systems

Abstract: Identification of causal links is fundamental for the analysis of complex systems. In dynamical systems, however, nonlinear interactions may hamper separability of subsystems which poses a challenge for attempts to determine the directions and strengths of their mutual influences. We found that asymmetric causal influences between parts of a dynamical system lead to characteristic distortions in the mappings between the attractor manifolds reconstructed from respective local observables. These distortions can be measured in a model-free, data-driven manner. This approach extends basic intuitions about cause-effect relations to deterministic dynamical systems and suggests a mathematically well defined explanation of results obtained from previous methods based on state space reconstruction.