<ul><li>CausalDynamics is a large-scale benchmark and data generation framework created to improve the structural discovery of dynamical causal models, especially in scenarios where active interventions are not possible.</li><li>The benchmark includes true causal graphs derived from numerous coupled ordinary and stochastic differential equations, along with two idealized climate models, to address limitations in existing methods tailored to deterministic, low-dimensional, and weakly nonlinear time-series data.</li><li>A comprehensive evaluation of state-of-the-art causal discovery algorithms for graph reconstruction on systems with noisy, confounded, and lagged dynamics is performed using CausalDynamics.</li><li>CausalDynamics offers a plug-and-play workflow for building a hierarchy of physical systems, aiming to support the development of robust causal discovery algorithms applicable across various domains and their unique challenges.</li></ul>

CausalDynamics: A large-scale benchmark for structural discovery of dynamical causal models

Discover more