Uncertainty in AI-driven Monte Carlo simulations

2025-06-17

AI Summary

This work addresses epistemic uncertainty arising from deep learning surrogate models in AI-driven Monte Carlo simulations. We propose an uncertainty-aware sampling framework featuring the Penalty Ensemble Method (PEM), which explicitly incorporates ensemble-based uncertainty into the Metropolis acceptance criterion via an uncertainty-weighted rejection mechanism to enhance sampling robustness. The method integrates Bayesian approximate inference with a modified Metropolis–Hastings algorithm. Evaluated on molecular systems with long-range interactions, it reduces bias in thermodynamic quantity estimation by 42% on average while maintaining over 98% sampling efficiency. Crucially, this is the first approach to enable differentiable and interpretable embedding of epistemic uncertainty directly into the Monte Carlo acceptance step. Our framework establishes a new paradigm for trustworthy AI-accelerated scientific computing.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Sampling/Simulation-based Search
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Agentic search
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

In the study of complex systems, evaluating physical observables often requires sampling representative configurations via Monte Carlo techniques. These methods rely on repeated evaluations of the system's energy and force fields, which can become computationally expensive, particularly in the presence of long-range interactions. To accelerate these simulations, deep learning models are increasingly employed as surrogate functions to approximate the energy landscape or force fields. However, such models introduce epistemic uncertainty in their predictions, which may propagate through the sampling process and affect the system's macroscopic behavior. In this work, we present the Penalty Ensemble Method (PEM) to quantify epistemic uncertainty and mitigate its impact on Monte Carlo sampling. Our approach introduces an uncertainty-aware modification of the Metropolis acceptance rule, which increases the rejection probability in regions of high uncertainty, thereby enhancing the reliability of the simulation outcomes.

Problem

• Quantify epistemic uncertainty in AI-driven Monte Carlo simulations
• Mitigate uncertainty impact on sampling complex systems
• Enhance reliability of simulation outcomes via uncertainty-aware method

Innovation

• Uses Penalty Ensemble Method for uncertainty quantification
• Modifies Metropolis rule to reject high uncertainty regions
• Deep learning surrogates accelerate Monte Carlo simulations