SpookyNet: Learning force fields with electronic degrees of freedom and nonlocal effects (2021-05-01T00:00:00.000000Z)

TL;DR

SpookyNet is a deep neural network that explicitly treats electronic degrees of freedom, closing an important remaining gap for models in quantum chemistry.

Abstract

Machine-learned force fields combine the accuracy of ab initio methods with the efficiency of conventional force fields. However, current machine-learned force fields typically ignore electronic degrees of freedom, such as the total charge or spin state, and assume chemical locality, which is problematic when molecules have inconsistent electronic states, or when nonlocal effects play a significant role. This work introduces SpookyNet, a deep neural network for constructing machine-learned force fields with explicit treatment of electronic degrees of freedom and nonlocality, modeled via self-attention in a transformer architecture. Chemically meaningful inductive biases and analytical corrections built into the network architecture allow it to properly model physical limits. SpookyNet improves upon the current state-of-the-art (or achieves similar performance) on popular quantum chemistry data sets. Notably, it is able to generalize across chemical and conformational space and can leverage the learned chemical insights, e.g. by predicting unknown spin states, thus helping to close a further important remaining gap for today’s machine learning models in quantum chemistry. Current machine-learned force fields typically ignore electronic degrees of freedom. SpookyNet is a deep neural network that explicitly treats electronic degrees of freedom, closing an important remaining gap for models in quantum chemistry.

Authors

Kristof T. Schütt

6 papers

M. Gastegger

4 papers

Oliver T. Unke

3 papers

TL;DR

Abstract

Authors

References133 items

Solving the electronic Schrödinger equation for multiple nuclear geometries with weight-sharing deep neural networks

Unsupervised Learning Methods for Molecular Simulation Data

Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications

SE(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials

Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems

Equivariant message passing for the prediction of tensorial properties and molecular spectra

QM7-X, a comprehensive dataset of quantum-mechanical properties spanning the chemical space of small organic molecules

A general-purpose machine-learning force field for bulk and nanostructured phosphorus

Machine Learning Force Fields

Rethinking Attention with Performers

A fourth-generation high-dimensional neural network potential with accurate electrostatics including non-local charge transfer

Molecular force fields with gradient-domain machine learning (GDML): Comparison and synergies with classical force fields.

Teaching a neural network to attach and detach electrons from molecules

OrbNet: Deep Learning for Quantum Chemistry Using Symmetry-Adapted Atomic-Orbital Features

QM7-X: A comprehensive dataset of quantum-mechanical properties spanning the chemical space of small organic molecules

The QMspin data set: Several thousand carbene singlet and triplet state structures and vertical spin gaps computed at MRCISD+Q-F12/cc-pVDZ-F12 level of theory

Large yet bounded: Spin gap ranges in carbenes

Directional Message Passing for Molecular Graphs

Incorporating Electronic Information into Machine Learning Potential Energy Surfaces via Approaching the Ground-State Electronic Energy as a Function of Atom-Based Electronic Populations.

Assessing conformer energies using electronic structure and machine learning methods

PyTorch: An Imperative Style, High-Performance Deep Learning Library

Exploring chemical compound space with quantum-based machine learning

Machine learning for molecular simulation

High-dimensional potential energy surfaces for molecular simulations: from empiricism to machine learning

Deep-neural-network solution of the electronic Schrödinger equation

Incorporating long-range physics in atomic-scale machine learning.

Ab-Initio Solution of the Many-Electron Schrödinger Equation with Deep Neural Networks

FCHL revisited: Faster and more accurate quantum machine learning.

Continuous and optimally complete description of chemical environments using Spherical Bessel descriptors

Unifying machine learning and quantum chemistry with a deep neural network for molecular wavefunctions

Cormorant: Covariant Molecular Neural Networks

Unmasking Clever Hans predictors and assessing what machines really learn

PhysNet: A Neural Network for Predicting Energies, Forces, Dipole Moments, and Partial Charges.

The reactive

sGDML: Constructing accurate and data efficient molecular force fields using machine learning

A generally applicable atomic-charge dependent London dispersion correction.

GFN2-xTB-An Accurate and Broadly Parametrized Self-Consistent Tight-Binding Quantum Chemical Method with Multipole Electrostatics and Density-Dependent Dispersion Contributions.

Accurate and transferable multitask prediction of chemical properties with an atoms-in-molecules neural network

SchNetPack: A Deep Learning Toolbox For Atomistic Systems.

A reactive, scalable, and transferable model for molecular energies from a neural network approach based on local information.

Tensor Field Networks: Rotation- and Translation-Equivariant Neural Networks for 3D Point Clouds

On the Convergence of Adam and Beyond

Efficient nonparametric n -body force fields from machine learning

Charge Separation in Donor-C60 Complexes with Real-Time Green Functions: The Importance of Nonlocal Correlations.

Alchemical and structural distribution based representation for universal quantum machine learning.

SchNet - A deep learning architecture for molecules and materials.

wACSF-Weighted atom-centered symmetry functions as descriptors in machine learning potentials.

The TensorMol-0.1 model chemistry: a neural network augmented with long-range physics

Non-covalent interactions across organic and biological subsets of chemical space: Physics-based potentials parametrized from machine learning.

Hierarchical modeling of molecular energies using a deep neural network.

The Expressive Power of Neural Networks: A View from the Width

Deep Potential Molecular Dynamics: a scalable model with the accuracy of quantum mechanics

Toolkit for the Construction of Reproducing Kernel-Based Representations of Data: Application to Multidimensional Potential Energy Surfaces

Interpolation of intermolecular potentials using Gaussian processes.

Attention is All you Need

Machine learning molecular dynamics for the simulation of infrared spectra

Neural Message Passing for Quantum Chemistry

First-Principles Models for van der Waals Interactions in Molecules and Materials: Concepts, Theory, and Applications.

Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learning

Machine learning of accurate energy-conserving molecular force fields

ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost

Quantum-chemical insights from deep tensor neural networks

Communication: Understanding molecular representations in machine learning: The role of uniqueness and target similarity.

Gaussian Error Linear Units (GELUs)

Identity Mappings in Deep Residual Networks

Benefits of Depth in Neural Networks

The Power of Depth for Feedforward Neural Networks

Deep Residual Learning for Image Recognition

On the Expressive Power of Deep Learning: A Tensor Analysis

Highway Networks

Interatomic potentials for ionic systems with density functional accuracy based on charge densities obtained by a neural network

Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials

A charge optimized many-body (comb) potential for titanium and titania

The stopping and range of ions in solids vol 1 : The stopping and ranges of ions in matter

Assessment and Validation of Machine Learning Methods for Predicting Molecular Atomization Energies.

Fourier series of atomic radial distribution functions: A molecular fingerprint for machine learning models of quantum chemical properties