wm-simulation — Monte Carlo simulation, counterfactual estimation, forecasting, and prediction calibration.
N21: Ports v2's wm-evolution MC suite capabilities for v4:
- Monte Carlo simulation — Bayesian MC, Quasi-MC, sensitivity analysis
- Counterfactual estimation — synthetic control projection for causal impact
- Forecasting — time series forecasting with confidence intervals
- Prediction calibration — Brier scorecard with the Murphy decomposition
- Bayesian optimization — GP surrogates + Expected Improvement search
- Information-theoretic measures — entropy, mutual information
This enables the SelfModel to forecast outcomes, the Dream cycle to simulate counterfactuals, and the Homeostatic loop to simulate action consequences before executing them.