Felix A. Faber

Prediction Errors of Molecular Machine Learning Models Lower than Hybrid DFT Error

We investigate the impact of choosing regressors and molecular representations for the construction of fast machine learning (ML) models of 13 electronic ground-state properties of organic molecules. The performance of each regressor/representation/property combination is assessed using learning curves which report out-of-sample errors as a function of training set size with up to ∼118k distinct molecules. Molecular structures and properties at the hybrid density functional theory (DFT) level of theory come from the QM9 database [ Ramakrishnan et al. Sci. Data 2014 , 1 , 140022 ] and include enthalpies and free energies of atomization, HOMO/LUMO energies and gap, dipole moment, polarizability, zero point vibrational energy, heat capacity, and the highest fundamental vibrational frequency. Various molecular representations have been studied (Coulomb matrix, bag of bonds, BAML and ECFP4, molecular graphs (MG)), as well as newly developed distribution based variants including histograms of distances (HD), angles (HDA/MARAD), and dihedrals (HDAD). Regressors include linear models (Bayesian ridge regression (BR) and linear regression with elastic net regularization (EN)), random forest (RF), kernel ridge regression (KRR), and two types of neural networks, graph convolutions (GC) and gated graph networks (GG). Out-of sample errors are strongly dependent on the choice of representation and regressor and molecular property. Electronic properties are typically best accounted for by MG and GC, while energetic properties are better described by HDAD and KRR. The specific combinations with the lowest out-of-sample errors in the ∼118k training set size limit are (free) energies and enthalpies of atomization (HDAD/KRR), HOMO/LUMO eigenvalue and gap (MG/GC), dipole moment (MG/GC), static polarizability (MG/GG), zero point vibrational energy (HDAD/KRR), heat capacity at room temperature (HDAD/KRR), and highest fundamental vibrational frequency (BAML/RF). We present numerical evidence that ML model predictions deviate from DFT (B3LYP) less than DFT (B3LYP) deviates from experiment for all properties. Furthermore, out-of-sample prediction errors with respect to hybrid DFT reference are on par with, or close to, chemical accuracy. The results suggest that ML models could be more accurate than hybrid DFT if explicitly electron correlated quantum (or experimental) data were available.We investigate the impact of choosing regressors and molecular representations for the construction of fast machine learning (ML) models of thirteen electronic ground-state properties of organic molecules. The performance of each regressor/representation/property combination is assessed using learning curves which report out-of-sample errors as a function of training set size with up to

Machine Learning Energies of 2 Million Elpasolite (ABC_{2}D_{6}) Crystals.

\sim

Machine Learning Energies of 2 Million Elpasolite(ABC2D6)Crystals

117k distinct molecules. Molecular structures and properties at hybrid density functional theory (DFT) level of theory used for training and testing come from the QM9 database [Ramakrishnan et al, {\em Scientific Data} {\bf 1} 140022 (2014)] and include dipole moment, polarizability, HOMO/LUMO energies and gap, electronic spatial extent, zero point vibrational energy, enthalpies and free energies of atomization, heat capacity and the highest fundamental vibrational frequency. Various representations from the literature have been studied (Coulomb matrix, bag of bonds, BAML and ECFP4, molecular graphs (MG)), as well as newly developed distribution based variants including histograms of distances (HD), and angles (HDA/MARAD), and dihedrals (HDAD). Regressors include linear models (Bayesian ridge regression (BR) and linear regression with elastic net regularization (EN)), random forest (RF), kernel ridge regression (KRR) and two types of neural net works, graph convolutions (GC) and gated graph networks (GG). We present numerical evidence that ML model predictions deviate from DFT less than DFT deviates from experiment for all properties. Furthermore, our out-of-sample prediction errors with respect to hybrid DFT reference are on par with, or close to, chemical accuracy. Our findings suggest that ML models could be more accurate than hybrid DFT if explicitly electron correlated quantum (or experimental) data was available.

Machine Learning Energies of 2 M Elpasolite (ABC

Elpasolite is the predominant quaternary crystal structure (AlNaK_{2}F_{6} prototype) reported in the Inorganic Crystal Structure Database. We develop a machine learning model to calculate density functional theory quality formation energies of all ∼2×10^{6} pristine ABC_{2}D_{6} elpasolite crystals that can be made up from main-group elements (up to bismuth). Our models accuracy can be improved systematically, reaching a mean absolute error of 0.1  eV/atom for a training set consisting of 10×10^{3} crystals. Important bonding trends are revealed: fluoride is best suited to fit the coordination of the D site, which lowers the formation energy whereas the opposite is found for carbon. The bonding contribution of the elements A and B is very small on average. Low formation energies result from A and B being late elements from group II, C being a late (group I) element, and D being fluoride. Out of 2×10^{6} crystals, 90 unique structures are predicted to be on the convex hull-among which is NFAl_{2}Ca_{6}, with a peculiar stoichiometry and a negative atomic oxidation state for Al.

_2

Elpasolite is the predominant quaternary crystal structure (AlNaK_{2}F_{6} prototype) reported in the Inorganic Crystal Structure Database. We develop a machine learning model to calculate density functional theory quality formation energies of all ∼2×10^{6} pristine ABC_{2}D_{6} elpasolite crystals that can be made up from main-group elements (up to bismuth). Our models accuracy can be improved systematically, reaching a mean absolute error of 0.1  eV/atom for a training set consisting of 10×10^{3} crystals. Important bonding trends are revealed: fluoride is best suited to fit the coordination of the D site, which lowers the formation energy whereas the opposite is found for carbon. The bonding contribution of the elements A and B is very small on average. Low formation energies result from A and B being late elements from group II, C being a late (group I) element, and D being fluoride. Out of 2×10^{6} crystals, 90 unique structures are predicted to be on the convex hull-among which is NFAl_{2}Ca_{6}, with a peculiar stoichiometry and a negative atomic oxidation state for Al.

D

Elpasolite is the predominant quaternary crystal structure (AlNaK_{2}F_{6} prototype) reported in the Inorganic Crystal Structure Database. We develop a machine learning model to calculate density functional theory quality formation energies of all ∼2×10^{6} pristine ABC_{2}D_{6} elpasolite crystals that can be made up from main-group elements (up to bismuth). Our models accuracy can be improved systematically, reaching a mean absolute error of 0.1  eV/atom for a training set consisting of 10×10^{3} crystals. Important bonding trends are revealed: fluoride is best suited to fit the coordination of the D site, which lowers the formation energy whereas the opposite is found for carbon. The bonding contribution of the elements A and B is very small on average. Low formation energies result from A and B being late elements from group II, C being a late (group I) element, and D being fluoride. Out of 2×10^{6} crystals, 90 unique structures are predicted to be on the convex hull-among which is NFAl_{2}Ca_{6}, with a peculiar stoichiometry and a negative atomic oxidation state for Al.