Efficient Discovery of Actual Causality in Stochastic Systems

Arshia Rafieioskouei, Kenneth Rogale, Borzoo Bonakdarpour

Abstract: Identifying the actual cause of events in engineered systems is a fundamental challenge in systems. It provides a principled framework for capturing logical dependencies between events, offering insights into the underlying dynamics of a system. Finding such causes becomes more challenging in real-world systems. In this paper, we adopt the notion of probabilistic actual causality by Fenton-Glynn, which is a probabilistic extension of Halpern and Pearl’s actual causality, and propose a novel method to formally reason about causal effect of events in stochastic systems. We (1) formulate the discovery of probabilistic actual causes in computing systems as an SMT problem, and (2) address the scalability challenges by introducing an abstraction-refinement technique that improves efficiency by up to 95%. We demonstrate the effectiveness of our approach through three case studies, identifying probabilistic actual causes of safety violations in (1) the Mountain Car problem, (2) the Lunar Lander benchmark, and (3) MPC controller for an F-16 autopilot simulator.

External IDs:dblp:conf/vmcai/RafieioskoueiRB26