Research questionHow can causal discovery recover causal structures when time series shift across latent regimes?Methods that assume one stable causal graph can produce unreliable relationships when a time series moves between regimes. The challenge is to determine where stable causal structures hold and infer the appropriate structure for each interval.