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

Abstract

Deep learning is revolutionizing many areas of science and technology, especially image, text and speech recognition. In this paper, we demonstrate how a deep neural network (NN) trained on quantum mechanical (QM) DFT calculations can learn an accurate and fully transferable potential for organic molecules. We introduce ANAKIN-ME (Accurate NeurAl networK engINe for Molecular Energies) or ANI in short. ANI is a new method and procedure for training neural network potentials that utilizes a highly modified version of the Behler and Parrinello symmetry functions to build single-atom atomic environment vectors as a molecular representation. We utilize ANI to build a potential called ANI-1, which was trained on a subset of the GDB databases with up to 8 heavy atoms to predict total energies for organic molecules containing four atom types: H, C, N, and O. To obtain an accelerated but physically relevant sampling of molecular potential surfaces, we also propose a Normal Mode Sampling (NMS) method for generating molecular configurations. Through a series of case studies, we show that ANI-1 is chemically accurate compared to reference DFT calculations on much larger molecular systems (up to 54 atoms) than those included in the training data set, with root mean square errors as low as 0.56 kcal/mol.

References in corpus (4)

• Deep Learning in Neural Networks: An Overview
• How van der Waals interactions determine the unique properties of water
• Comparing the Accuracy of High-Dimensional Neural Network Potentials and the Systematic Molecular Fragmentation Method: A Benchmark Study for all-trans Alkanes
• Space-Filling Curves as a Novel Crystal Structure Representation for Machine Learning Models

Cited by in corpus (66)

• Machine learning for molecular simulation
• Machine Learning Unifies the Modelling of Materials and Molecules
• ANI-1: A data set of 20M off-equilibrium DFT calculations for organic molecules
• Resolving transition metal chemical space: feature selection for machine learning and structure-property relationships
• A deep potential model with long-range electrostatic interactions
• The prospects of quantum computing in computational molecular biology
• Unsupervised machine learning in atomistic simulations, between predictions and understanding
• Molecular Force Fields with Gradient-Domain Machine Learning: Construction and Application to Dynamics of Small Molecules with Coupled Cluster Forces
• A Universal Density Matrix Functional from Molecular Orbital-Based Machine Learning: Transferability across Organic Molecules
• Training neural nets to learn reactive potential energy surfaces using interactive quantum chemistry in virtual reality
• Classification of Local Chemical Environments from X-ray Absorption Spectra using Supervised Machine Learning
• Uncertainty estimation for molecular dynamics and sampling
• Physically Motivated Recursively Embedded Atom Neural Networks: Incorporating Local Completeness and Nonlocality
• Deep Learning of Accurate Force Field of Ferroelectric HfO${}_2$
• Gaussian approximation potentials for body-centered-cubic transition metals
• Ground state energy functional with Hartree-Fock efficiency and chemical accuracy
• Tutorial: How to Train a Neural Network Potential
• PiNN: A Python Library for Building Atomic Neural Networks of Molecules and Materials
• Towards empirical force fields that match experimental observables
• Graph Neural Networks Accelerated Molecular Dynamics
• Ensemble Learning of Coarse-Grained Molecular Dynamics Force Fields with a Kernel Approach
• BuRNN: Buffer Region Neural Network Approach for Polarizable-Embedding Neural Network / Molecular Mechanics Simulations
• An assessment of the structural resolution of various fingerprints commonly used in machine learning
• A Novel Approach to Describe Chemical Environments in High Dimensional Neural Network Potentials
• Efficient implementation of atom-density representations
• Advances in Quantum Deep Learning: An Overview
• Transfer learning for chemically accurate interatomic neural network potentials
• Combining phonon accuracy with high transferability in Gaussian approximation potential models
• Generating valid Euclidean distance matrices
• Pushing the limit of molecular dynamics with ab initio accuracy to 100 million atoms with machine learning
• Dielectric constant of supercritical water in a large pressure-temperature range
• Arrested phase separation in double-exchange models: machine-learning enabled large-scale simulation
• Potential energy surface interpolation with neural networks for instanton rate calculations
• AENET-LAMMPS and AENET-TINKER: Interfaces for Accurate and Efficient Molecular Dynamics Simulations with Machine Learning Potentials
• Learning Neural Generative Dynamics for Molecular Conformation Generation
• Deep Learning in Protein Structural Modeling and Design
• Hydrogen-induced degradation dynamics in silicon heterojunction solar cells via machine learning
• Analytical Gradients for Molecular-Orbital-Based Machine Learning
• Wavelet Scattering Networks for Atomistic Systems with Extrapolation of Material Properties
• Machine-learning of atomic-scale properties based on physical principles
• Deep Density: circumventing the Kohn-Sham equations via symmetry preserving neural networks
• A Hessian-Based Assessment of Atomic Forces for Training Machine Learning Interatomic Potentials
• Tell machine learning potentials what they are needed for: Simulation-oriented training exemplified for glycine
• Machine learning at the atomic-scale
• AI-accelerated Materials Informatics Method for the Discovery of Ductile Alloys
• A neural network potential with self-trained atomic fingerprints: a test with the mW water potential
• On the equivalence of molecular graph convolution and molecular wave function with poor basis set
• Steerable Wavelet Scattering for 3D Atomic Systems with Application to Li-Si Energy Prediction
• Machine Learning Forcefield for Silicate Glasses
• Transformative Applications of Machine Learning for Chemical Reactions
• Modeling and simulating the excited-state dynamics of a system with condensed phases: A machine learning approach
• Uncertainty quantification of molecular property prediction using Bayesian neural network models
• Machine learning approaches for analyzing and enhancing molecular dynamics simulations
• Geometric learning of the conformational dynamics of molecules using dynamic graph neural networks
• Machine learning dynamics of phase separation in correlated electron magnets
• On-the-fly Prediction of Protein Hydration Densities and Free Energies using Deep Learning
• Managing Expectations and Imbalanced Training Data in Reactive Force Field Development: an Application to Water Adsorption on Alumina
• Machine Learning for Quantum Dynamics: Deep Learning of Excitation Energy Transfer Properties
• REANN: A PyTorch-based End-to-End Multi-functional Deep Neural Network Package for Molecular, Reactive and Periodic Systems
• Detect the Interactions that Matter in Matter: Geometric Attention for Many-Body Systems
• Simulations of water and hydrophobic hydration using a neural network potential
• Deep machine learning potentials for multicomponent metallic melts: development, predictability and compositional transferability
• A2I Transformer: Permutation-equivariant attention network for pairwise and many-body interactions with minimal featurization
• HydroNet: Benchmark Tasks for Preserving Intermolecular Interactions and Structural Motifs in Predictive and Generative Models for Molecular Data
• Symmetry-driven graph neural networks
• libmolgrid: GPU Accelerated Molecular Gridding for Deep Learning Applications