ML Models of Vibrating H${}_2$CO: Comparing Reproducing Kernels, FCHL and PhysNet

Abstract

Machine Learning (ML) has become a promising tool for improving the quality of atomistic simulations. Using formaldehyde as a benchmark system for intramolecular interactions, a comparative assessment of ML models based on state-of-the-art variants of deep neural networks (NN), reproducing kernel Hilbert space (RKHS+F), and kernel ridge regression (KRR) is presented. Learning curves for energies and atomic forces indicate rapid convergence towards excellent predictions for B3LYP, MP2, and CCSD(T)-F12 reference results for modestly sized (in the hundreds) training sets. Typically, learning curve off-sets decay as one goes from NN (PhysNet) to RKHS+F to KRR (FCHL). Conversely, the predictive power for extrapolation of energies towards new geometries increases in the same order with RKHS+F and FCHL performing almost equally. For harmonic vibrational frequencies, the picture is less clear, with PhysNet and FCHL yielding respectively flat learning at $\sim$ 1 and $\sim$ 0.2 cm${}^{-1}$ no matter which reference method, while RKHS+F models level off for B3LYP, and exhibit continued improvements for MP2 and CCSD(T)-F12. Finite-temperature molecular dynamics (MD) simulations with the same initial conditions yield indistinguishable infrared spectra with good performance compared with experiment except for the high-frequency modes involving hydrogen stretch motion which is a known limitation of MD for vibrational spectroscopy. For sufficiently large training set sizes all three models can detect insufficient convergence (``noise'') of the reference electronic structure calculations in that the learning curves level off. Transfer learning (TL) from B3LYP to CCSD(T)-F12 with PhysNet indicates that additional improvements in data efficiency can be achieved.

References in corpus (19)

• Neural Message Passing for Quantum Chemistry
• Gaussian Approximation Potentials: the accuracy of quantum mechanics, without the electrons
• PhysNet: A Neural Network for Predicting Energies, Forces, Dipole Moments and Partial Charges
• Big Data meets Quantum Chemistry Approximations: The $\Delta$-Machine Learning Approach
• W4 theory for computational thermochemistry: in pursuit of confident sub-kJ/mol predictions
• Machine learning prediction errors better than DFT accuracy
• ANI-1: A data set of 20M off-equilibrium DFT calculations for organic molecules
• FCHL revisited: faster and more accurate quantum machine learning
• Alchemical and structural distribution based representation for improved QML
• Comparison of permutationally invariant polynomials, neural networks, and Gaussian approximation potentials in representing water interactions through many-body expansions
• Operators in Machine Learning: Response Properties in Chemical Space
• Reactive Dynamics and Spectroscopy of Hydrogen Transfer from Neural Network-Based Reactive Potential Energy Surfaces
• Permutationally Invariant, Reproducing Kernel-Based Potential Energy Surfaces for Polyatomic Molecules: From Formaldehyde to Acetone
• The $\mathrm{N}{(}^{4}S)+{\mathrm{O}}_2(X^3{\Sigma }_g^{-})\leftrightarrow \mathrm{O}{(}^{3}P)+\mathrm{NO}(X^2\Pi )$ Reaction: Thermal and Vibrational Relaxation Rates for the ${}^2$A${}^{\prime }$, ${}^4$A${}^{\prime }$ and ${}^2$A${}^{\prime \prime }$ States
• Accurate Reproducing Kernel-Based Potential Energy Surfaces for the Triplet Ground States of N${}_2$O and Dynamics for the N+NO$\leftrightarrow$O+N${}_2$ and N${}_2$+O$\to$2N+O Reactions
• Near dissociation states for H${}_2^{+}$-He on MRCI and FCI potential energy surfaces
• Perspective: Non-conventional Force Fields for Applications in Spectroscopy and Chemical Reaction Dynamics
• Impact of non-normal error distributions on the benchmarking and ranking of Quantum Machine Learning models
• Isomerization and Decomposition Reactions of Acetaldehyde Relevant to Atmospheric Processes from Dynamics Simulations on Neural Network-Based Potential Energy Surfaces

Cited by in corpus (10)

• Perspective on integrating machine learning into computational chemistry and materials science
• Representations of molecules and materials for interpolation of quantum-mechanical simulations via machine learning
• Permutationally Invariant, Reproducing Kernel-Based Potential Energy Surfaces for Polyatomic Molecules: From Formaldehyde to Acetone
• Transfer learning for chemically accurate interatomic neural network potentials
• Impact of the characteristics of quantum chemical databases on machine learning predictions of tautomerization energies
• Transformative Applications of Machine Learning for Chemical Reactions
• Ab initio machine learning in chemical compound space
• Transfer Learned Potential Energy Surfaces: Accurate Anharmonic Vibrational Dynamics and Dissociation Energies for the Formic Acid Monomer and Dimer
• Machine-Learned Potentials for Solvation Modeling
• Quantitative Molecular Simulations