How to measure causal influence at a single point?

Causal density functions are Radon-Nikodym derivatives that quantify how intervention changes probability distributions at each data point. Unlike existing methods that compare whole distributions, these local density ratios let you estimate and validate causal effects directly: reweight observational data by the estimated density and you should recover interventional expectations. The approach yields practical estimators for causal curves and edge strengths, validated on synthetic and real perturbation benchmarks.