Constructing high‐dimensional neural network potentials: A tutorial review

A lot of progress has been made in recent years in the development of atomistic potentials using machine learning (ML) techniques. In contrast to most conventional potentials, which are based on physical approximations and simplifications to derive an analytic functional relation between the atomic configuration and the potential-energy, ML potentials rely on simple but very flexible mathematical terms without a direct physical meaning. Instead, in case of ML potentials the topology of the potential-energy surface is “learned” by adjusting a number of parameters with the aim to reproduce a set of reference electronic structure data as accurately as possible. Due to this bias-free construction, they are applicable to a wide range of systems without changes in their functional form, and a very high accuracy close to the underlying first-principles data can be obtained. Neural network potentials (NNPs), which have first been proposed about two decades ago, are an important class of ML potentials. Although the first NNPs have been restricted to small molecules with only a few degrees of freedom, they are now applicable to high-dimensional systems containing thousands of atoms, which enables addressing a variety of problems in chemistry, physics, and materials science. In this tutorial review, the basic ideas of NNPs are presented with a special focus on developing NNPs for high-dimensional condensed systems. A recipe for the construction of these potentials is given and remaining limitations of the method are discussed. © 2015 Wiley Periodicals, Inc.

• Publication year

  2015

• Venue

  International Journal of Quantum Chemistry

• Publication date

  2015-08-15

• Fields of study

  Chemistry, Physics, Computer Science

• Identifiers
  DOI 10.1002/QUA.24890
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Role of physisorption states in molecular scattering: a semilocal density-functional theory study on O2/Ag(111).

  2014cited by this paper
• Next generation interatomic potentials for condensed systems

  2014cited by this paper
• Representing potential energy surfaces by high-dimensional neural network potentials

  2014cited by this paper
• A Full-Dimensional Neural Network Potential-Energy Surface for Water Clusters up to the Hexamer

  2013cited by this paper
• Machine-learning approach for one- and two-body corrections to density functional theory: Applications to molecular and condensed water

  2013cited by this paper
• Fast Crystallization of the Phase Change Compound GeTe by Large-Scale Molecular Dynamics Simulations.

  2013cited by this paper
• A density-functional theory-based neural network potential for water clusters including van der Waals corrections.

  2013cited by this paper
• Neural networks for local structure detection in polymorphic systems.

  2013cited by this paper
• Assessment and Validation of Machine Learning Methods for Predicting Molecular Atomization Energies.

  2013cited by this paper
• Six-dimensional potential energy surface of the dissociative chemisorption of HCl on Au(111) using neural networks

  2013cited by this paper
• Neural network potentials for metals and oxides – First applications to copper clusters at zinc oxide

  2013cited by this paper
• High-dimensional neural network potentials for metal surfaces: A prototype study for copper

  2012cited by this paper
• Thermal transport in phase-change materials from atomistic simulations

  2012cited by this paper
• Neural network interatomic potential for the phase change material GeTe

  2012cited by this paper
• Breakdown of Stokes–Einstein relation in the supercooled liquid state of phase change materials

  2012cited by this paper
• Construction of high-dimensional neural network potentials using environment-dependent atom pairs.

  2012cited by this paper
• Non-adiabatic effects during the dissociative adsorption of O2 at Ag(111)? A first-principles divide and conquer study

  2012cited by this paper
• Polarisable multipolar electrostatics from the machine learning method Kriging: an application to alanine

  2012cited by this paper
• A neural network potential-energy surface for the water dimer based on environment-dependent atomic energies and charges.

  2012cited by this paper
• Fast and accurate modeling of molecular atomization energies with machine learning.

  2011cited by this paper
• Neural network potential-energy surfaces in chemistry: a tool for large-scale simulations.

  2011cited by this paper
• Microscopic origins of the anomalous melting behavior of sodium under high pressure.

  2011cited by this paper
• Density functional theory with London dispersion corrections

  2011cited by this paper
• Atom-centered symmetry functions for constructing high-dimensional neural network potentials.

  2011cited by this paper
• High-dimensional neural-network potentials for multicomponent systems: Applications to zinc oxide

  2011cited by this paper
• Nucleation mechanism for the direct graphite-to-diamond phase transition.

  2011cited by this paper
• High-dimensional ab initio potential energy surfaces for reaction dynamics calculations.

  2011cited by this paper
• A model for the dissociative adsorption of N2O on Cu(1 0 0) using a continuous potential energy surface

  2010cited by this paper
• Potential energy surfaces fitted by artificial neural networks.

  2010cited by this paper
• A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu.

  2010cited by this paper
• Graphite-diamond phase coexistence study employing a neural-network mapping of the ab initio potential energy surface

  2010cited by this paper
• Neural Networks and Learning Machines

  2010cited by this paper
• Ab initio quality neural-network potential for sodium

  2010cited by this paper
• Accurate molecular van der Waals interactions from ground-state electron density and free-atom reference data.

  2009cited by this paper
• Signatures of nonadiabatic O2 dissociation at Al(111): First-principles fewest-switches study

  2009cited by this paper
• Simultaneous fitting of a potential-energy surface and its corresponding force fields using feedforward neural networks.

  2009cited by this paper
• Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons.

  2009cited by this paper
• Molecular dissociation of hydrogen peroxide (HOOH) on a neural network ab initio potential surface with a new configuration sampling method involving gradient fitting.

  2009cited by this paper
• Fitting sparse multidimensional data with low-dimensional terms

  2009cited by this paper
• Permutationally invariant potential energy surfaces in high dimensionality

  2009cited by this paper
• Ab Initio Molecular Dynamics: Basic Theory and Advanced Methods

  2009cited by this paper
• Metadynamics simulations of the high-pressure phases of silicon employing a high-dimensional neural network potential.

  2008cited by this paper
• Fingerprints for spin-selection rules in the interaction dynamics of O2 at Al(111).

  2008cited by this paper
• Pressure‐induced phase transitions in silicon studied by neural network‐based metadynamics simulations

  2008cited by this paper
• Cis-->trans, trans-->cis isomerizations and N-O bond dissociation of nitrous acid (HONO) on an ab initio potential surface obtained by novelty sampling and feed-forward neural network fitting.

  2008cited by this paper
• Silicon potentials investigated using density functional theory fitted neural networks

  2008cited by this paper
• Mapping Potential Energy Surfaces by Neural Networks : The ethanol/Au(111) interface

  2008cited by this paper
• Using neural networks, optimized coordinates, and high-dimensional model representations to obtain a vinyl bromide potential surface.

  2008cited by this paper
• Beyond Point Charges: Dynamic Polarization from Neural Net Predicted Multipole Moments.

  2008cited by this paper
• Nonadiabatic effects in the dissociation of oxygen molecules at the Al(111) surface

  2007cited by this paper
• One- and two-center physical space partitioning of the energy in the density functional theory.

  2007cited by this paper
• Ab initio molecular dynamics of hydrogen dissociation on metal surfaces using neural networks and novelty sampling.

  2007cited by this paper
• Interpolating moving least-squares methods for fitting potential energy surfaces: computing high-density potential energy surface data from low-density ab initio data points.

  2007cited by this paper
• Bond-order potentials through the ages

  2007cited by this paper
• A polarizable high-rank quantum topological electrostatic potential developed using neural networks: Molecular dynamics simulations on the hydrogen fluoride dimer

  2007cited by this paper
• Generalized neural-network representation of high-dimensional potential-energy surfaces.

  2007cited by this paper
• A new approach to potential fitting using neural networks

  2007cited by this paper
• Representing molecule-surface interactions with symmetry-adapted neural networks.

  2007cited by this paper
• Using redundant coordinates to represent potential energy surfaces with lower-dimensional functions.

  2007cited by this paper
• Descriptions of surface chemical reactions using a neural network representation of the potential-energy surface

  2006cited by this paper
• Using neural networks to represent potential surfaces as sums of products.

  2006cited by this paper
• Energy partitioning schemes.

  2006cited by this paper
• A random-sampling high dimensional model representation neural network for building potential energy surfaces.

  2006cited by this paper
• Neural Network Models of Potential Energy Surfaces:  Prototypical Examples.

  2005cited by this paper
• Dissociation of O2 at Al(111): the role of spin selection rules.

  2004cited by this paper
• Representing high-dimensional potential-energy surfaces for reactions at surfaces by neural networks

  2004cited by this paper
• Classical and quasiclassical spectral analysis of CH5+ using an ab initio potential energy surface

  2003cited by this paper
• Multi-dimensional potential energy surface determination by modified Shepard interpolation for a molecule–surface reaction: H2+Pt(111)

  2003cited by this paper
• A polarizable force field for water using an artificial neural network

  2002cited by this paper
• Escaping free-energy minima

  2002cited by this paper
• Kalman Filtering and Neural Networks

  2001cited by this paper
• ReaxFF: A Reactive Force Field for Hydrocarbons

  2001cited by this paper
• A Chemist's Guide to Density Functional Theory

  2000cited by this paper
• A local interpolation scheme using no derivatives in quantum-chemical calculations

  1999cited by this paper
• Applications of neural networks to fitting interatomic potential functions

  1999cited by this paper
• Applications of genetic algorithms and neural networks to interatomic potentials

  1999cited by this paper
• Universal Approximation Using Feedforward Neural Networks: A Survey of Some Existing Methods, and Some New Results

  1998cited by this paper
• Representation of Intermolecular Potential Functions by Neural Networks

  1998cited by this paper
• Neural Modeling of Torsional Potential Hypersurfaces in Non-rigid Molecules

  1998cited by this paper
• The fitting of potential energy surfaces using neural networks. Application to the study of the photodissociation processes

  1998cited by this paper
• The fitting of potential energy surfaces using neural networks: Application to the study of vibrational levels of H3+

  1998cited by this paper
• Approximations of Functions by a Multilayer Perceptron: a New Approach

  1997cited by this paper
• Description of the potential energy surface of the water dimer with an artificial neural network

  1997cited by this paper
• Neural network models of potential energy surfaces

  1995cited by this paper
• A second generation force field for the simulation of proteins

  1995cited by this paper
• Adaptive, global, extended Kalman filters for training feedforward neural networks

  1994cited by this paper
• Molecular Potential Energy Surfaces by Interpolation

  1994cited by this paper
• Neural networks in chemistry

  1993cited by this paper
• Neural Networks for Pattern Recognition

  1993cited by this paper
• UFF, a full periodic table force field for molecular mechanics and molecular dynamics simulations

  1992cited by this paper
• Approximation capabilities of multilayer feedforward networks

  1991cited by this paper
• DREIDING: A generic force field for molecular simulations

  1990cited by this paper
• Neural networks for self-learning control systems

  1990cited by this paper
• Molecular mechanics. The MM3 force field for hydrocarbons. 1

  1989cited by this paper
• Numerical Recipes: The Art of Scientific Computing

  1989cited by this paper
• Approximation by superpositions of a sigmoidal function

  1989cited by this paper
• Multilayer feedforward networks are universal approximators

  1989cited by this paper
• Density-functional theory of atoms and molecules

  1989cited by this paper
• An introduction to neural computing

  1988cited by this paper
• New empirical model for the structural properties of silicon.

  1986cited by this paper

• Deepcluster agent to search Global Minimum of Nanoclusters using Deep Reinforcement Learning: A Review

  2026cites this paper
• Performance of Meta's Universal Model for Atoms across the Conformational and Configurational Space of Diverse Transition-Metal Catalysts.

  2026cites this paper
• Analisis Bibliometrik Terhadap Tren dan Pertumbuhan Penelitian Quantum Machine Learning

  2026cites this paper
• Interfacial thermal transport characteristics and heat transfer enhancement mechanisms of Ga/AlN composite thermal interface materials

  2026cites this paper
• Opportunities and Challenges of Machine Learning in Atomic Layer Deposition

  2026cites this paper
• Insights into Ethylene Deep Oxidation on Pd–Au Bimetallic Catalysts: From Surface Dynamics to Pathway Selectivity

  2026cites this paper
• Machine-learning molecular dynamics of GaN: micro-displacement effects in radiation damage

  2026cites this paper
• Multiscale Simulation for Electrolyte Design: From Microstructure-Property Decoding to Macroscopic Performance Prediction

  2026cites this paper
• Nanoscale thermal transport at PbTe grain boundaries: a neural-network-potential molecular dynamics study

  2026cites this paper
• New Permutationally Invariant Polynomial Potential Energy Surfaces for H5O2 + with Fast Analytical Gradients Calculated Using Reverse Differentiation

  2026cites this paper
• Transforming MOF Modeling with Machine-Learned Potentials: Progress and Perspectives.

  2026cites this paper
• A physics-informed long-range polarizable potential based on deep learning.

  2026cites this paper
• Optimizing Prediction of Chemical Bonds in Interfacial Dynamics through Local Uncertainty Estimates with Neural Network Ensembles.

  2026cites this paper
• Investigating the effect of structural variations on the band gap of titanium dioxide using machine learning

  2026cites this paper
• Hitchhikers Guide To Training More General Machine Learning Potentials in Heterogeneous Catalysis.

  2026cites this paper
• Driving Catalytic Innovation: The Development and Application of Machine Learning Force Fields

  2026cites this paper
• Bridging Topological Over-Squashing and Physical Long-Range Interactions in ML Interatomic Potentials

  2026cites this paper
• Generating Stable and Metastable Critical Points in Uncertain Systems via Flow‐Based Models

  2026cites this paper
• Machine Learning Driven Advances in Molecular Dynamics of Bulk and Interfacial Aqueous Systems.

  2026cites this paper
• A Perspective on Employing Computational Approaches to Interrogate Catalyst Restructuring

  2026cites this paper
• Does Hessian Data Improve the Performance of Machine Learning Potentials?

  2025cites this paper
• Application of neural-network potential based on trainable descriptor to crystallographically complex lattice defects in silicon

  2025cites this paper
• A Review on Artificial Neural Networks for Structural Analysis

  2025cites this paper
• The evolution of machine learning potentials for molecules, reactions and materials.

  2025cites this paper
• Ab initio path-integral simulations of hydrogen-isotope diffusion in face-centred cubic metals

  2025cites this paper
• Study of methyl phosphate by molecular dynamics simulations based on first principles and on machine-learning force fields

  2025cites this paper
• Theoretical Approaches to Solving the Shortest Vector Problem in NP-Hard Lattice-Based Cryptography with Post-SUSY Theories of Quantum Gravity in Polynomial Time by Orch-Or

  2025cites this paper
• Free energy profiles for chemical reactions in solution from high-dimensional neural network potentials: The case of the Strecker synthesis.

  2025cites this paper
• Kernel regression methods for prediction of materials properties: Recent developments

  2025cites this paper
• On the origin of hydrophilic interactions at alumina surfaces.

  2025cites this paper
• Primary radiation damage in Si1−xGex alloys: Molecular dynamics study with machine-learning interatomic potential

  2025cites this paper
• CQPES: A GPU-Aided Software Package for Developing Full-Dimensional Accurate Potential Energy Surfaces by Permutation-Invariant-Polynomial Neural Network

  2025cites this paper
• Toward high entropy material discovery for energy applications using computational and machine learning methods

  2025cites this paper
• Electron transfer controlled by hydrogen bond donor/acceptor exchange in [H2O]6.

  2025cites this paper
• A Neural Network Interatomic Potential for the Ternary α-Fe-C-H System: Toward Million-Atom Simulations of Hydrogen Embrittlement in Steel

  2025cites this paper
• Review: machine learning approaches for diverse alloy systems

  2025cites this paper
• Transferability and expansion of the clayNN model

  2025cites this paper
• Computation of the heat capacity of water from first principles.

  2025cites this paper
• Understanding Structure-Composition-Property Relationships of Ni-P Bulk Metallic Glasses

  2025cites this paper
• ReMLP-NET: A Neural Network Interaction Potential for Molecular Energy Prediction.

  2025cites this paper
• MELTS: Fully Automated Active Learning for Fewest-Switches Surface Hopping Dynamics.

  2025cites this paper
• Applications and Advances of Machine Learning in the Development of Solid-State Electrolytes for Lithium-Ion Batteries

  2025cites this paper
• Development of a Neural Network Potential for Osmium Enables Irradiation Damage Simulations

  2025cites this paper
• Insights into the Structure and Dynamics of Water at Co3O4${\rm Co}_3{\rm O}_4$ (001) Using a High‐Dimensional Neural Network Potential

  2025cites this paper
• Efficient Grand Canonical Global Optimization with On-the-fly-trained Machine-learning Interatomic Potentials

  2025cites this paper
• LFaB: Low fidelity as Bias for Active Learning in the chemical configuration space

  2025cites this paper
• Deep reinforcement learning for identifying the global minima of platinum nanoclusters

  2025cites this paper
• Interpretability of linear regression models of glassy dynamics

  2025cites this paper
• Improved loss functions for machine-learned atomic potentials.

  2025cites this paper
• Addressing pathological deficiencies in the simulation of complex charged fluids using neural network force fields

  2025cites this paper
• Studying Noncovalent Interactions in Molecular Systems with Machine Learning

  2025cites this paper
• Alchemical Free-Energy Calculations at Quantum-Chemical Precision

  2025cites this paper
• Determination of Stable Proton Configurations by Black-Box Optimization Using an Ising Machine

  2025cites this paper
• PiNN: Equivariant Neural Network Suite for Modeling Electrochemical Systems

  2025cites this paper
• Predicting Molecular Energies of Small Organic Molecules With Multi‐Fidelity Methods

  2025cites this paper
• Effective Field Neural Network

  2025cites this paper
• Research on Classification of College Students’ Physical Fitness Test Scores Based on Neural Network

  2025cites this paper
• Machine Learning-Enhanced Calculation of Quantum-Classical Binding Free Energies

  2025cites this paper
• Computational Modeling of Solid–Liquid Interfaces: From Ab Initio Methods to Machine-Learning Potentials

  2025cites this paper
• A beginner’s approach to deep learning applied to VS and MD techniques

  2025cites this paper
• X2‐PEC: A Neural Network Model Based on Atomic Pair Energy Corrections

  2025cites this paper
• Phonon dispersion filter: A physics-inspired feature selection for machine learning potentials

  2025cites this paper
• Constant-Potential Machine Learning Molecular Dynamics Simulations Reveal Potential-Regulated Cu Cluster Formation on MoS2

  2025cites this paper
• Atomic Energy Accuracy of Neural Network Potentials: Harnessing Pretraining and Transfer Learning

  2025cites this paper
• Machine-learning potential energy surfaces implications in photodissociation process.

  2025cites this paper
• Neural Network-Based Molecular Dynamics Simulation of Water Assisted by Active Learning.

  2025cites this paper
• Δ-model correction of foundation model based on the model's own understanding.

  2025cites this paper
• Machine-learned potentials for solvation modeling

  2025cites this paper
• Charge transfer at air-water interfaces: A machine learning potential-based molecular dynamics study.

  2025cites this paper
• Machine Learning-Driven Molecular Dynamics Investigation of the Physical and Heat Transfer Properties of Magnesium

  2025cites this paper
• Tests of Artificial Neural Network-Based Diabatization Approaches on Simple 1D Models

  2025cites this paper
• Impact of the damping function in dispersion-corrected density functional theory on the properties of liquid water

  2025cites this paper
• Isomerization-assisted proton transfers in MeOH-(H2O)2H.

  2025cites this paper
• The CP2K Program Package Made Simple

  2025cites this paper
• Hot Carrier Relaxation in Halide Double Perovskites: Insights from Ab Initio Models and Structure-Informed Machine Learning

  2025cites this paper
• Multiscale Light-Matter Dynamics in Quantum Materials: From Electrons to Topological Superlattices

  2025cites this paper
• Accelerating the Structure Exploration of Diverse Bi–Pt Nanoclusters via Physics‐Informed Machine Learning Potential and Particle Swarm Optimization

  2025cites this paper
• Teachers that teach the irrelevant: Pre-training machine learned interaction potentials with classical force fields for robust molecular dynamics simulations

  2025cites this paper
• Atomistic Simulations of Thermal and Chemical Expansions of PrNixCo1‐xO3‐δ Accelerated by Machine Learning Potentials

  2025cites this paper
• Bridging quantum and computation: A bibliometric analysis of machine learning in quantum chemistry

  2025cites this paper
• A Portable Data Set for Borophene Growth Modeling with Reactive Neural Network Potentials

  2025cites this paper
• Multi-fidelity learning for atomistic models via trainable data embeddings

  2025cites this paper
• A Comprehensive Review of Machine-Learning Approaches for Crystal Structure/Property Prediction

  2025cites this paper
• Exploring Foundational Machine Learned Potentials for Treating the High Temperature Dynamics of Metal‐Organic Frameworks

  2025cites this paper
• Microscopic mechanism of thermal amorphization of ZIF-4 and melting of ZIF-zni revealed via molecular dynamics and machine learning techniques

  2024cites this paper
• Inelastic neutron scattering of hydrogen in palladium studied by semiclassical dynamics

  2024cites this paper
• Implementation and Validation of an OpenMM Plugin for the Deep Potential Representation of Potential Energy

  2024cites this paper
• Construction of a Mode-Combination Hamiltonian under the Grid-Based Representation for the Quantum Dynamics of OH + HO2 → O2 + H2O.

  2024cites this paper
• Chemistry-Informed Generative Model for Classical Dynamics Simulations.

  2024cites this paper
• No Headache for PIPs: A PIP Potential for Aspirin Runs Much Faster and with Similar Precision Than Other Machine-Learned Potentials.

  2024cites this paper
• Atomic-level mechanisms of short-circuit diffusion in materials

  2024cites this paper
• Two-Step Growth Mechanism of the Solid Electrolyte Interphase in Argyrodyte/Li-Metal Contacts.

  2024cites this paper
• SchNet_IIA: Potential Energy Surface Fitting by Interatomic Interactions Attention Based on Transfer Learning Analysis

  2024cites this paper
• Molecular dynamics studies of knotted polymers.

  2024cites this paper
• Recent Advances in Machine Learning‐Assisted Multiscale Design of Energy Materials

  2024cites this paper
• Structure optimization of UiO-66 supported metal nanoclusters based on high-dimensional neural network

  2024cites this paper
• Machine Learning Advances in High-Entropy Alloys: A Mini-Review

  2024cites this paper
• Data Generation for Machine Learning Interatomic Potentials and Beyond

  2024cites this paper
• Melting simulations of high-entropy carbonitrides by deep learning potentials

  2024cites this paper
• Generator of Neural Network Potential for Molecular Dynamics: Constructing Robust and Accurate Potentials with Active Learning for Nanosecond-Scale Simulations.

  2024cites this paper