Neural Network Surrogates for Free Energy Computation of Complex Chemical Systems

2025/10/01 by Pornpatcharapong, Wasut · 1 citation
#Artificial Intelligence (cs.AI) #Chemical Physics (physics.chem-ph) #Computational Engineering #Computational Physics (physics.comp-ph) #FOS: Computer and information sciences #FOS: Physical sciences #Finance #Machine Learning (cs.LG) #and Science (cs.CE)

paper · doi:10.48550/arxiv.2510.01396

Abstract

Free energy reconstruction methods such as Gaussian Process Regression (GPR) require Jacobians of the collective variables (CVs), a bottleneck that restricts the use of complex or machine-learned CVs. We introduce a neural network surrogate framework that learns CVs directly from Cartesian coordinates and uses automatic differentiation to provide Jacobians, bypassing analytical forms. On an MgCl2 ion-pairing system, our method achieved high accuracy for both a simple distance CV and a complex coordination-number CV. Moreover, Jacobian errors also followed a near-Gaussian distribution, making them suitable for GPR pipelines. This framework enables gradient-based free energy methods to incorporate complex and machine-learned CVs, broadening the scope of biochemistry and materials simulations.

Citations

• Gaussian Process Regression for Materials and Molecules
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Automatic differentiation in machine learning: a survey

Cited by

• Learning Biomolecular Motion: The Physics-Informed Machine Learning Paradigm

Related

• Descriptors-free Collective Variables From Geometric Graph Neural Networks
• Neural network Gaussian processes as efficient models of potential energy surfaces for polyatomic molecules
• Extending machine learning model for implicit solvation to free energy calculations
• Solvation free energies from neural thermodynamic integration
• Sampling-free Inference for Ab-Initio Potential Energy Surface Networks