Universal Machine Learning for the Response of Atomistic Systems to External Fields

Abstract

Machine learned interatomic interaction potentials have enabled efficient and accurate molecular simulations of closed systems. However, external fields, which can greatly change the chemical structure and/or reactivity, have been seldom included in current machine learning models. This work proposes a universal field-induced recursively embedded atom neural network (FIREANN) model, which integrates a pseudo field vector-dependent feature into atomic descriptors to represent system-field interactions with rigorous rotational equivariance. This "all-in-one" approach correlates various response properties like dipole moment and polarizability with the field-dependent potential energy in a single model, very suitable for spectroscopic and dynamics simulations in molecular and periodic systems in the presence of electric fields. Especially for periodic systems, we find that FIREANN can overcome the intrinsic multiple-value issue of the polarization by training atomic forces only. These results validate the universality and capability of the FIREANN method for efficient first-principles modeling of complicated systems in strong external fields.

9 figures

References in corpus (18)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials
• Physics-inspired structural representations for molecules and materials
• SpookyNet: Learning Force Fields with Electronic Degrees of Freedom and Nonlocal Effects
• Symmetry-Adapted Machine-Learning for Tensorial Properties of Atomistic Systems
• How to remove the spurious resonances from ring polymer molecular dynamics
• Ab initio molecular dynamics study of dissociation of water under an electric field
• Raman Spectrum and Polarizability of Liquid Water from Deep Neural Networks
• Self-Consistent Determination of Long-Range Electrostatics in Neural Network Potentials
• Gaussian Moments as Physically Inspired Molecular Descriptors for Accurate and Scalable Machine Learning Potentials
• Physically Motivated Recursively Embedded Atom Neural Networks: Incorporating Local Completeness and Nonlocality
• BIGDML: Towards Exact Machine Learning Force Fields for Materials
• Universal Machine Learning for the Response of Atomistic Systems to External Fields
• Finite Field Methods for the Supercell Modelling of Charged Insulator-Electrolyte Interfaces
• Symmetry-Adapted High Dimensional Neural Network Representation of Electronic Friction Tensor of Adsorbates on Metals
• Equivariant representations for molecular Hamiltonians and N-center atomic-scale properties
• Spectroscopy from Machine Learning by Accurately Representing the Atomic Polar Tensor
• Infrared spectra at coupled cluster accuracy from neural network representations

Cited by in corpus (11)

• A practical guide to machine learning interatomic potentials -- Status and future
• Universal Machine Learning for the Response of Atomistic Systems to External Fields
• The Evolution of Machine Learning Potentials for Molecules, Reactions and Materials
• Molecular dynamics simulation with finite electric fields using Perturbed Neural Network Potentials
• Exploring the design space of machine-learning models for quantum chemistry with a fully differentiable framework
• Aqueous Solution Chemistry In Silico and the Role of Data Driven Approaches
• A simple approach to rotationally invariant machine learning of avector quantity
• Atomistic Simulations of Oxide-Water Interfaces using Machine Learning Potentials
• A Bond-Based Machine Learning Model for Molecular Polarizabilities and A Priori Raman Spectra
• MOLPIPx: an end-to-end differentiable package for permutationally invariant polynomials in Python and Rust
• Neural Polarization: Toward Electron Density for Molecules by Extending Equivariant Networks