Critical Benchmarking of the G4(MP2) Model, the Correlation Consistent Composite Approach and Popular Density Functional Approximations on a Probabilistically Pruned Benchmark Dataset of Formation Enthalpies

Abstract

First-principles calculation of the standard formation enthalpy, $\Delta H_f^{\circ }$ (298K), in such large scale as required by chemical space explorations, is amenable only with density functional approximations (DFAs) and some composite wave function theories (cWFTs). Alas, the accuracies of popular range-separated hybrid, `rung-4' DFAs, and cWFTs that offer the best accuracy-vs.-cost trade-off have as yet been established only for datasets predominantly comprising small molecules, hence, their transferability to larger datasets remains vague. In this study, we present an extended benchmark dataset of over 1600 values of $\Delta H_f^{\circ }$ for structurally and electronically diverse molecules. We apply quartile-ranking based on boundary-corrected kernel density estimation to filter outliers and arrive at Probabilistically Pruned Enthalpies of 1694 compounds (PPE1694). For this dataset, we rank the prediction accuracies of G4, G4(MP2), ccCA, CBS-QB3 and 23 popular DFAs using conventional and probabilistic error metrics. We discuss systematic prediction errors and highlight the role an empirical higher-level correction (HLC) plays in the G4(MP2) model. Furthermore, we comment on uncertainties associated with the reference empirical data for atoms and the systematic errors stemming from these that grow with the molecular size. We believe these findings to aid in identifying meaningful application domains for quantum thermochemical methods.

arXiv commentMajor Revision. Please ignore the previous versions

References in corpus (14)

• Climbing the Density Functional Ladder: Non-Empirical Meta-Generalized Gradient Approximation Designed for Molecules and Solids
• Long-Range Corrected Hybrid Density Functionals with Improved Dispersion Corrections
• Big Data meets Quantum Chemistry Approximations: The $\Delta$-Machine Learning Approach
• W4 theory for computational thermochemistry: in pursuit of confident sub-kJ/mol predictions
• Alchemical and structural distribution based representation for improved QML
• Hierarchical modeling of molecular energies using a deep neural network
• Choosing a density functional for static molecular polarizabilities
• Error Assessment of Computational Models in Chemistry
• Probabilistic performance estimators for computational chemistry methods: the empirical cumulative distribution function of absolute errors
• Canonical and DLPNO-based G4(MP2)XK-inspired composite wavefunction methods parametrized against large and chemically diverse training sets: Are they more accurate and/or robust than double hybrid DFT?
• Impact of non-normal error distributions on the benchmarking and ranking of Quantum Machine Learning models
• The Chemical Space of B, N-substituted Polycyclic Aromatic Hydrocarbons: Combinatorial Enumeration and High-Throughput First-Principles Modeling
• Probabilistic performance estimators for computational chemistry methods: Systematic Improvement Probability and Ranking Probability Matrix. I. Theory
• Large yet bounded: Spin gap ranges in carbenes

Cited by in corpus (3)

• The long road to calibrated prediction uncertainty in computational chemistry
• Leveraging the Bias-Variance Tradeoff in Quantum Chemistry for Accurate Negative Singlet-Triplet Gap Predictions: A Case for Double-Hybrid DFT
• Stereo-Electronic Factors Influencing the Stability of Hydroperoxyalkyl Radicals: Transferability of Chemical Trends across Hydrocarbons and ab initio Methods