Operator Forces For Coarse-Grained Molecular Dynamics

2025-06-24

AI Summary

This study addresses two key limitations in coarse-grained molecular dynamics (CG-MD) force field parametrization: reliance on atomistic force labels and noise-prone kernel functions causing local structural distortions. We propose an unsupervised learning framework requiring only configurational samples—no atomistic forces. Methodologically, we embed a normalized flow model into the force-matching formalism to construct an invertible, physics-constrained mapping kernel, replacing conventional noise-sensitive kernel methods. This design simultaneously ensures global conformational statistical accuracy and local physical fidelity of the CG force field without force labels. Validated on small protein systems, the learned force field accurately reproduces slow dynamical processes and significantly reduces local force errors. Our approach establishes a new high-accuracy, interpretable, label-free paradigm for large-scale biomolecular modeling.

Technology Category

• Machine Learning: Structured Learning
• Natural Language Processing: Learning & Optimization for NLP
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Coarse-grained (CG) molecular dynamics simulations extend the length and time scale of atomistic simulations by replacing groups of correlated atoms with CG beads. Machine-learned coarse-graining (MLCG) has recently emerged as a promising approach to construct highly accurate force fields for CG molecular dynamics. However, the calibration of MLCG force fields typically hinges on force matching, which demands extensive reference atomistic trajectories with corresponding force labels. In practice, atomistic forces are often not recorded, making traditional force matching infeasible on pre-existing datasets. Recently, noise-based kernels have been introduced to adapt force matching to the low-data regime, including situations in which reference atomistic forces are not present. While this approach produces force fields which recapitulate slow collective motion, it introduces significant local distortions due to the corrupting effects of the noise-based kernel. In this work, we introduce more general kernels based on normalizing flows that substantially reduce these local distortions while preserving global conformational accuracy. We demonstrate our method on small proteins, showing that flow-based kernels can generate high-quality CG forces solely from configurational samples.

Problem

• Developing accurate force fields for coarse-grained molecular dynamics without atomistic force labels
• Reducing local distortions in force fields caused by noise-based kernels
• Generating high-quality coarse-grained forces from configurational samples only

Innovation

• Machine-learned coarse-graining for accurate force fields
• Normalizing flow kernels reduce local distortions
• Flow-based kernels generate CG forces from configurations