Challenges for Machine Learning Force Fields in Reproducing Potential Energy Surfaces of Flexible Molecules

Abstract

Dynamics of flexible molecules are often determined by an interplay between local chemical bond fluctuations and conformational changes driven by long-range electrostatics and van der Waals interactions. This interplay between interactions yields complex potential-energy surfaces (PES) with multiple minima and transition paths between them. In this work, we assess the performance of state-of-the-art Machine Learning (ML) models, namely sGDML, SchNet, GAP/SOAP, and BPNN for reproducing such PES, while using limited amounts of reference data. As a benchmark, we use the cis to trans thermal relaxation in an azobenzene molecule, where at least three different transition mechanisms should be considered. Although GAP/SOAP, SchNet, and sGDML models can globally achieve chemical accuracy of 1 kcal mol-1 with fewer than 1000 training points, predictions greatly depend on the ML method used as well as the local region of the PES being sampled. Within a given ML method, large differences can be found between predictions of close-to-equilibrium and transition regions, as well as for different transition mechanisms. We identify key challenges that the ML models face in learning long-range interactions and the intrinsic limitations of commonly used atom-based descriptors. All in all, our results suggest switching from learning the entire PES within a single model to using multiple local models with optimized descriptors, training sets, and architectures for different parts of complex PES.

References in corpus (38)

• Gaussian Approximation Potentials: the accuracy of quantum mechanics, without the electrons
• On representing chemical environments
• Fast and Accurate Modeling of Molecular Atomization Energies with Machine Learning
• Deep Potential Molecular Dynamics: a scalable model with the accuracy of quantum mechanics
• SchNet - a deep learning architecture for molecules and materials
• Machine Learning of Accurate Energy-Conserving Molecular Force Fields
• Comparing molecules and solids across structural and alchemical space
• Machine Learning Unifies the Modelling of Materials and Molecules
• Extreme Broadband Transparent Optical Phase Change Materials for High-Performance Nonvolatile Photonics
• By-passing the Kohn-Sham equations with machine learning
• Towards Exact Molecular Dynamics Simulations with Machine-Learned Force Fields
• Active learning of linearly parametrized interatomic potentials
• Machine learning prediction errors better than DFT accuracy
• Machine Learning Molecular Dynamics for the Simulation of Infrared Spectra
• VAMPnets: Deep learning of molecular kinetics
• SchNetPack: A Deep Learning Toolbox For Atomistic Systems
• i-PI 2.0: A Universal Force Engine for Advanced Molecular Simulations
• Efficient and Accurate Machine-Learning Interpolation of Atomic Energies in Compositions with Many Species
• FCHL revisited: faster and more accurate quantum machine learning
• Alchemical and structural distribution based representation for improved QML
• Symmetry-Adapted Machine-Learning for Tensorial Properties of Atomistic Systems
• sGDML: Constructing Accurate and Data Efficient Molecular Force Fields Using Machine Learning
• Accurate Interatomic Force Fields via Machine Learning with Covariant Kernels
• Machine Learning for Quantum Mechanical Properties of Atoms in Molecules
• Incorporating long-range physics in atomic-scale machine learning
• A learning scheme to predict atomic forces and accelerate materials simulations
• Non-covalent interactions across organic and biological subsets of chemical space: Physics-based potentials parametrized from machine learning
• Structure-based Sampling and Self-correcting Machine Learning for Accurate Calculations of Potential Energy Surfaces and Vibrational Levels
• Representing molecule-surface interactions with symmetry-adapted neural networks
• Efficient nonparametric $n$-body force fields from machine learning
• Operators in Machine Learning: Response Properties in Chemical Space
• Molecular Force Fields with Gradient-Domain Machine Learning: Construction and Application to Dynamics of Small Molecules with Coupled Cluster Forces
• The Many-Body Expansion Combined with Neural Networks
• Conformational Transitions of Non-Grafted Polymers Near an Adsorbing Substrate
• Solid Harmonic Wavelet Scattering for Predictions of Molecule Properties
• Dynamical Strengthening of Covalent and Non-Covalent Molecular Interactions by Nuclear Quantum Effects at Finite Temperature
• Convolutional neural networks for atomistic systems
• An Atomistic Fingerprint Algorithm for Learning Ab Initio Molecular Force Fields

Cited by in corpus (10)

• Differentiable sampling of molecular geometries with uncertainty-based adversarial attacks
• Towards Linearly Scaling and Chemically Accurate Global Machine Learning Force Fields for Large Molecules
• The MD17 Datasets from the Perspective of Datasets for Gas-Phase "Small" Molecule Potentials
• An Extendible, Graph-Neural-Network-Based Approach for Accurate Force Field Development of Large Flexible Organic Molecules
• Modeling electronic response properties with an explicit-electron machine learning potential
• Transfer-Learned Potential Energy Surfaces: Towards Microsecond-Scale Molecular Dynamics Simulations in the Gas Phase at CCSD(T) Quality
• GPU-Accelerated Approximate Kernel Method for Quantum Machine Learning
• No Headache for PIPs: A PIP Potential for Aspirin Outperforms Other Machine-Learned Potentials
• Tell machine learning potentials what they are needed for: Simulation-oriented training exemplified for glycine
• Finding Thermodynamically Favorable Pathways in Chemical Reaction Networks Using Flows in Hypergraphs and Mixed-Integer Linear Programming