Solid harmonic wavelet scattering for predictions of molecule properties

Eickenberg, Michael; Exarchakis, Georgios; Hirn, Matthew; Mallat, Stéphane; Thiry, Louis

doi:10.1063/1.5023798

Cited by 86 publications

(94 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The QM9 data set represented a molecular property set of unprecedented size and consistency at the time of its inception. Subsequently, this data set has become very popular for training ML property predictors for organic molecules [69,70,71,72,73,74,75,76,77,78]. However, the principal disadvantage of the QM9 data set is that it is composed of only optimized molecular structures.…”

Section: Diversity Comparisonmentioning

confidence: 99%

The ANI-1ccx and ANI-1x Data Sets, Coupled-Cluster and Density Functional Theory Properties for Molecules

Smith¹,

Zubatyuk²,

Nebgen³

et al. 2020

Preprint

View full text Add to dashboard Cite

Maximum diversification of data is a central theme in building generalized and accurate machine learning (ML) models. In chemistry, ML has been used to develop models for predicting molecular properties, for example quantum mechanics (QM) calculated potential energy surfaces and atomic charge models. The ANI-1x and ANI-1ccx ML-based general-purpose potentials for organic molecules were developed through active learning; an automated data diversification process. Here, we describe the ANI-1x and ANI-1ccx data sets. To demonstrate data set diversity, we visualize them with a dimensionality reduction scheme, and contrast against existing data sets. The ANI-1x data set contains multiple QM properties from 5M density functional theory calculations, while the ANI-1ccx data set contains 500k data points obtained with an accurate CCSD(T)/CBS extrapolation. Approximately 14 million CPU core-hours were expended to generate this data. Multiple QM properties from density functional theory and coupled cluster are provided: energies, atomic forces, multipole moments, atomic charges, and more. We provide this data to the community to aid research and development of ML models for chemistry. ⇤ roitberg@ufl.edu † olexandr@olexandrisayev.com ‡ serg@lanl.gov ⇧ equal contribution to this work

show abstract

Section: Diversity Comparisonmentioning

confidence: 99%

The ANI-1ccx and ANI-1x Data Sets, Coupled-Cluster and Density Functional Theory Properties for Molecules

Smith¹,

Zubatyuk²,

Nebgen³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…An important issue in building machine learning models is how to handle the symmetries in the system. The handling of translational, rotational and even permutational symmetries has been discussed in depth in the literature already [25,39,40,41]. Besides these static symmetries, Boltzmann equation also possesses an important dynamic symmetry, the Galilean invariance.…”

Section: Symmetries and Galilean Invariantmentioning

confidence: 99%

Uniformly accurate machine learning-based hydrodynamic models for kinetic equations

Han

et al. 2019

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

A new framework is introduced for constructing interpretable and truly reliable reduced models for multi-scale problems in situations without scale separation. Hydrodynamic approximation to the kinetic equation is used as an example to illustrate the main steps and issues involved. To this end, a set of generalized moments are constructed first through an autoencoder to optimally represent the underlying velocity distribution. The well-known closure problem is then solved with the aim of best capturing the associated dynamics of the kinetic equation. The issue of physical constraints such as Galilean invariance is addressed and an active learning procedure is introduced to help ensure that the data set used is representative enough. The reduced system takes the form of the conventional moment systems and works regardless of the numerical discretization used. Numerical results are presented for the BGK model. We demonstrate that the reduced model achieves a uniform accuracy in a wide range of Knudsen numbers spanning from the hydrodynamic limit to free molecular flow.

show abstract

“…On the other hand, Eickenberg et al 31 introduce a ML model based on a solid harmonic wavelet scattering representation of organic molecules and demonstrate competitive performance for predicted atomization energies. Meanwhile, Hy et al 76 use a new kind of NN, called a covariant compositional network, to deduce properties from molecular graphs alone, yielding promising results on databases of small molecules.…”

Section: A Prediction Of Energies and Other Properties Throughout Chmentioning

confidence: 99%

Guest Editorial: Special Topic on Data-Enabled Theoretical Chemistry

Rupp

Lilienfeld

Burke

2018

The Journal of Chemical Physics

View full text Add to dashboard Cite

show abstract

Solid harmonic wavelet scattering for predictions of molecule properties

Cited by 86 publications

References 30 publications

The ANI-1ccx and ANI-1x Data Sets, Coupled-Cluster and Density Functional Theory Properties for Molecules

The ANI-1ccx and ANI-1x Data Sets, Coupled-Cluster and Density Functional Theory Properties for Molecules

Uniformly accurate machine learning-based hydrodynamic models for kinetic equations

Guest Editorial: Special Topic on Data-Enabled Theoretical Chemistry

Contact Info

Product

Resources

About