OrbNet Denali: A machine learning potential for biological and organic chemistry with semi-empirical cost and DFT accuracy

We present OrbNet Denali, a machine learning model for electronic structure that is designed as a drop-in replacement for ground-state density functional theory (DFT) energy calculations. The model is a message-passing neural network that uses symmetry-adapted atomic orbital features from a low-cost quantum calculation to predict the energy of a molecule. OrbNet Denali is trained on a vast dataset of 2.3 million DFT calculations on molecules and geometries. This dataset covers the most common elements in bio- and organic chemistry (H, Li, B, C, N, O, F, Na, Mg, Si, P, S, Cl, K, Ca, Br, I) as well as charged molecules. OrbNet Denali is demonstrated on several well-established benchmark datasets, and we find that it provides accuracy that is on par with modern DFT methods while offering a speedup of up to three orders of magnitude. For the GMTKN55 benchmark set, OrbNet Denali achieves WTMAD-1 and WTMAD-2 scores of 7.19 and 9.84, on par with modern DFT functionals. For several GMTKN55 subsets, which contain chemical problems that are not present in the training set, OrbNet Denali produces a mean absolute error comparable to those of DFT methods. For the Hutchison conformers benchmark set, OrbNet Denali has a median correlation coefficient of R^2=0.90 compared to the reference DLPNO-CCSD(T) calculation, and R^2=0.97 compared to the method used to generate the training data (wB97X-D3/def2-TZVP), exceeding the performance of any other method with a similar cost. Similarly, the model reaches chemical accuracy for non-covalent interactions in the S66x10 dataset. For torsional profiles, OrbNet Denali reproduces the torsion profiles of wB97X-D3/def2-TZVP with an average MAE of 0.12 kcal/mol for the potential energy surfaces of the diverse fragments in the TorsionNet500 dataset.

Assessing impacts of discrepancies in model parameters on autoignition model performance: a case study using butanol

Side-by-side comparison of detailed kinetic models using a new tool to aid recognition of species structures reveals significant discrepancies in the published rates of many reactions and thermochemistry of many species. We present a first automated assessment of the impact of these varying parameters on observable quantities of interest---in this case, autoignition delay---using literature experimental data. A recent kinetic model for the isomers of butanol was imported into a common database. Individual reaction rate and thermodynamic parameters of species were varied using values encountered in combustion models from recent literature. The effects of over 1600 alternative parameters were considered. Separately, experimental data were collected from recent publications and converted into the standard YAML-based ChemKED format. The Cantera-based model validation tool, PyTeCK, was used to automatically simulate autoignition using the generated models and experimental data, to judge the performance of the models. Taken individually, most of the parameter substitutions have little effect on the overall model performance, although a handful have quite large effects, and are investigated more thoroughly. Additionally, models varying multiple parameters simultaneously were evolved using a genetic algorithm to give fastest and slowest autoignition delay times, showing that changes exceeding a factor of 10 in ignition delay time are possible by cherry-picking from only accepted, published parameters. All data and software used in this study are available openly.