[Paper Review] Learning Thermal Response Forces: A Method for Extending the Thermodynamic Transferability of Coarse-Grained Models via Machine-Learning

Patrick G. Sahrmann, Benjamin Nebgen+2 more|arXiv (Cornell University)|Feb 26, 2026

Machine Learning in Materials Science0 citations

TL;DR

The paper introduces a data-efficient approach to embed temperature dependence into machine-learned coarse-grained force fields by training on thermal response forces of the potential of mean force, improving transferability across thermodynamic states.

ABSTRACT

Machine-learned (ML) coarse-grained (CG) models are a promising tool for significantly enhancing the efficiency of molecular simulations by systematically removing degrees of freedom while retaining fidelity to the underlying fine-grained model. The CG potential of mean force (PMF) is inherently dependent on thermodynamic conditions and, hence, a CG force-field (FF) which is trained at one thermodynamic state point is not necessarily accurate at another. We propose, in this work, a novel and data-efficient means of learning temperature dependence into ML CG force-fields via training on the thermal response forces of the PMF. We demonstrate how incorporating these terms into ML CG FFs confers significantly improved transferability for CG water models and demonstrate how this transferability enables accurate and predictive CG dynamics.

Motivation & Objective

• Motivation: improve thermodynamic transferability of ML coarse-grained force fields across different temperatures.
• Goal: incorporate temperature dependence into CG force fields via thermal response forces.
• Aim: enable accurate, predictive CG dynamics under varying thermodynamic conditions.

Proposed method

• Train ML CG force fields using thermal response forces derived from the PMF.
• Incorporate thermal response terms into the CG FF to encode temperature dependence.
• Demonstrate transferability improvements for CG water models across thermodynamic states.

Experimental results

Research questions

• RQ1 Can thermal response forces be learned from PMF data to encode temperature dependence in CG models?
• RQ2 Does including thermal response forces improve the transferability of ML CG force fields to different temperatures?
• RQ3 What is the impact of these terms on the accuracy and predictive capability of CG dynamics?

Key findings

• Incorporating thermal response forces into ML CG force fields yields significantly improved transferability across temperatures for CG water models.
• The enhanced transferability enables more accurate and predictive CG dynamics under varying thermodynamic conditions.