Decomposing Chemical Space: Applications to the Machine Learning of Atomic Energies

Abstract

We apply a number of atomic decomposition schemes across the standard QM7 dataset -- a small model set of organic molecules at equilibrium geometry -- to inspect the possible emergence of trends among contributions to atomization energies from distinct elements embedded within molecules. Specifically, a recent decomposition scheme of ours based on spatially localized molecular orbitals is compared to alternatives that instead partition molecular energies on account of which nuclei individual atomic orbitals are centred on. We find these partitioning schemes to expose the composition of chemical compound space in very dissimilar ways in terms of the grouping, binning, and heterogeneity of discrete atomic contributions, e.g., those associated with hydrogens bonded to different heavy atoms. Furthermore, unphysical dependencies on the one-electron basis set are found for some, but not all of these schemes. The relevance and importance of these compositional factors for training tailored neural network models based on atomic energies are next assessed. We identify both limitations and possible advantages with respect to contemporary machine learning models and discuss the design of potential counterparts based on atoms and the intrinsic energies of these as the principal decomposition units.

21+7 pages, 6 figures. SI as an ancillary file. Version 2: All PhysNet-based results are now based on NN models trained on a combination of atomic and molecular energies (as opposed to only the former in Version 1). SI also updated with a total of four figures

References in corpus (8)

• ANI-1: An extensible neural network potential with DFT accuracy at force field computational cost
• Understanding molecular representations in machine learning: The role of uniqueness and target similarity
• OrbNet Denali: A machine learning potential for biological and organic chemistry with semi-empirical cost and DFT accuracy
• Quantum chemistry-augmented neural networks for reactivity prediction: Performance, generalizability and interpretability
• BIGDML: Towards Exact Machine Learning Force Fields for Materials
• An assessment of the structural resolution of various fingerprints commonly used in machine learning
• Electronic Excitations Through the Prism of Mean-Field Decomposition Techniques
• Decomposed Mean-Field Simulations of Local Properties in Condensed Phases