Learning QM/MM Potential using Equivariant Multiscale Model

Yuji Sugita,Yao-Kun Lei,Kiyoshi Yagi

DOI: https://doi.org/10.26434/chemrxiv-2024-80d93

2024-02-26

Abstract:The machine learning (ML) method emerges as an efficient and precise surrogate model for high-level electronic structure theory. Its application has been limited to closed chemical systems without considering external potentials from the surrounding environments. To address this limitation and incorporate the influence of external potentials, polarization effects, and long-range interactions between a chemical system and its environment, the first two terms of the Taylor expansion of an electrostatic operator have been used as extra input to the existing ML model to represent the electrostatic environments. However, high-order electrostatic interaction is often essential to account for external potentials from the environment. The existing models based only on the invariant features cannot capture significant distribution patterns of the external potentials. Here, we propose a novel ML model that includes high-order terms of the Taylor expansion of an electrostatic operator and uses an equivariant model, which can generate high-order tensors covariant with rotations as a base model. Thus, we can use the multipole-expansion equation to derive a useful representation by accounting for the polarization and intermolecular interaction. Moreover, to deal with long-range interactions, we follow the same strategy adopted to derive long-range interaction between a target system and its environment media. Our model achieves higher prediction accuracy and transferability among various environment media with these modifications.

Chemistry

What problem does this paper attempt to address?

This paper aims to address how to more accurately consider the interactions between chemical systems and the surrounding environment (such as solvents, biomolecules, etc.) in molecular simulations, especially the issues of charge distribution, polarization effects, and long-range interactions. Traditional machine learning (ML) methods have limitations in dealing with these complex interactions, often only focusing on predicting the energy and forces of closed systems while ignoring the influence of external potential fields. The paper proposes a new ML model that combines high-order Taylor expansion terms of electrostatic operators and employs equivariant models to handle rotationally invariant higher-order tensors. In this way, the model can better capture the distribution patterns of external potential fields and improve the accuracy of predicting charge distribution, polarization, and intermolecular interactions. Furthermore, to deal with long-range interactions, the model also uses a method similar to the Coulomb model to predict the total energy and forces while adjusting parameters to match the charge outputs obtained from quantum mechanical calculations. The contribution of the paper lies in the development of an ML model that can handle multi-scale problems, which can more effectively describe the environmental effects in chemical reactions, improve prediction accuracy, and enhance the generalization ability across environments, especially for large molecular systems involving long-range interactions.

Learning QM/MM Potential using Equivariant Multiscale Model

Advancing Multiscale Molecular Modeling with Machine Learning-Derived Electrostatics

Assessment of Embedding Schemes in a Hybrid Machine Learning/Classical Potentials (ML/MM) Approach

Machine Learning in QM/MM Molecular Dynamics Simulations of Condensed-Phase Systems

Assessment of embedding schemes in a hybrid machine learning/classical potentials (ML/MM) scheme

Learning the Quantum Centroid Force Correction in Molecular Systems: A Localized Approach

Electrostatic Embedding Machine Learning for Ground and Excited State Molecular Dynamics of Solvated Molecules

A quantum mechanical computational method for modeling electrostatic and solvation effects of protein

Learning Atomic Multipoles: Prediction of the Electrostatic Potential with Equivariant Graph Neural Networks

Kernel charge equilibration: efficient and accurate prediction of molecular dipole moments with a machine-learning enhanced electron density model

Learning together: Towards foundation models for machine learning interatomic potentials with meta-learning

Learning Together: Towards foundational models for machine learning interatomic potentials with meta-learning

Large-Scale Atomic Simulation via Machine Learning Potentials Constructed by Global Potential Energy Surface Exploration

Machine Learning Exchange Fields for Ab-initio Spin Dynamics

Extending the atomic decomposition and many-body representation, a chemistry-motivated monomer-centered approach for machine learning potentials

Enhancing universal machine learning potentials with polarizable long-range interactions

Machine learning potentials with Iterative Boltzmann Inversion: training to experiment

Efficient Training of Neural Network Potentials for Chemical and Enzymatic Reactions by Continual Learning

Machine learning model for non-equilibrium structures and energies of simple molecules

Exploring accurate potential energy surfaces via integrating variational quantum eigensovler with machine learning

Exploring Accurate Potential Energy Surfaces via Integrating Variational Quantum Eigensolver with Machine Learning