Intervention-Based Time Series Causal Discovery via Simulator-Generated Interventional Distributions
For researchers in causal discovery and scientific domains with simulators, SVAR-FM provides a principled way to correct sign-reversed estimates from observational methods, with theoretical guarantees and empirical validation.
SVAR-FM uses a physics-based simulator as a do-operator to generate interventional data for time series causal discovery, proving identifiability under a coverage condition and achieving correct causal sign recovery across four scientific domains. In a laser physics case study, it achieves R-squared = 0.983 with zero bias when simulator accuracy is high.
We propose SVAR-FM (Structural VAR with Flow Matching), a framework for time series causal discovery that treats a physics-based simulator as a mechanical realization of Pearl's do operator. Clamping a variable inside the simulator physically severs confounding paths, producing interventional data by construction. Conditional Flow Matching then learns the nonlinear interventional conditionals. Theoretically, we prove that the full structural VAR becomes identifiable under a coverage condition on the simulator-clampable variables, and derive an end-to-end error bound that decomposes into Monte Carlo, simulator fidelity, and Flow Matching terms. A sign-flip corollary predicts that when simulator accuracy falls below a threshold, the estimated causal effect reverses sign. Empirically, a benchmark across four scientific domains confirms that SVAR-FM recovers the correct causal sign where observational methods produce sign-reversed estimates due to confounding. A case study in ultrafast laser physics verifies the sign-flip prediction by physically varying the accuracy level of a first-principles quantum solver: the low-accuracy setting reverses the causal sign, while the high-accuracy setting recovers the correct direction (R-squared = 0.983, zero bias).