Riemannian denoising model for molecular structure optimization with chemical accuracy
- 2 January 2026
- journal article
- research article
- Published by Springer Nature in Nature Computational Science
- Vol. 6 (2) , 134-144
Abstract
Here we introduce a framework for molecular structure optimization using a denoising model on a physics-informed Riemannian manifold (R-DM). Unlike conventional approaches operating in Euclidean space, our method leverages a Riemannian metric that better aligns with molecular energy change, enabling more robust modeling of potential energy surfaces. By incorporating internal coordinates reflective of energetic properties, R-DM achieves chemical accuracy with an energy error below 1 kcal mol−1. Comparative evaluations on QM9, QM7-X and GEOM datasets demonstrate improvements in both structural and energetic accuracy, surpassing conventional Euclidean-based denoising models. This approach highlights the potential of physics-informed coordinates for tackling complex molecular optimization problems, with implications for tasks in computational chemistry and materials science.Keywords
All Related Versions
Funding Information
- MOE | Korea Environmental Industry and Technology Institute (RS202300219144)
- MOE | Korea Environmental Industry and Technology Institute (RS202300219144)
- MOE | Korea Environmental Industry and Technology Institute (RS202300219144)
- MOE | Korea Environmental Industry and Technology Institute (RS202300219144)
- InnoCORE program of the Ministry of Science and ICT
This publication has 48 references indexed in Scilit:
- Learning Collective Variables with Synthetic Data Augmentation through Physics-Inspired Geodesic InterpolationJournal of Chemical Theory and Computation, 2024
- Dynamic Precision Approach for Accelerating Large-Scale Eigenvalue Solvers in Electronic Structure Calculations on Graphics Processing UnitsJournal of Chemical Theory and Computation, 2023
- SMICLR: Contrastive Learning on Multiple Molecular Representations for Semisupervised and Unsupervised Representation LearningJournal of Chemical Information and Modeling, 2022
- System-Specific Separable Basis Based on Tucker Decomposition: Application to Density Functional CalculationsJournal of Chemical Theory and Computation, 2022
- Dataset Construction to Explore Chemical Space with 3D Geometry and Deep LearningJournal of Chemical Information and Modeling, 2021
- Quantum Machine Learning in Chemical Compound SpaceAngewandte Chemie International Edition, 2018
- Better Informed Distance Geometry: Using What We Know To Improve Conformation GenerationJournal of Chemical Information and Modeling, 2015
- Merck molecular force field. I. Basis, form, scope, parameterization, and performance of MMFF94Journal of Computational Chemistry, 1996
- Intrinsic field theory of chemical reactionsTheoretical Chemistry Accounts, 1979
- Asymptotic probabilities and differential equationsCommunications on Pure and Applied Mathematics, 1966