<ul><li>A new causal inference framework called assimilative causal inference (ACI) has been developed to identify instantaneous causal relationships and the dynamic evolution of causal influence range in high-dimensional systems.</li><li>ACI uses a dynamical system and a subset of state variables to trace causes backward from observed effects by solving an inverse problem via Bayesian data assimilation.</li><li>ACI captures the dynamic interplay of variables where roles as causes and effects can shift over time, provides a mathematically justified criterion for determining causal influence range, and is scalable to high-dimensional problems.</li><li>ACI is demonstrated to be effective in analyzing complex dynamical systems with intermittent and extreme events, not requiring observations of candidate causes and applicable to short time series and incomplete datasets.</li></ul>

Assimilative Causal Inference

Discover more