Retrieving energetic and structural information as well as bonding features from the ground state electron density, [rho(<b>r</b>)], has been the essence of density functional theory. This has inspired the work of many, including that of our research group. The growing interest in developing machine-learning models capable of predicting the electron density of molecular systems at a low cost is, thus, of little surprise. Here, we discuss the construction of a transferable and scalable machine-learning model of rho(<b>r</b>). The atom-centered framework adapted to rotational symmetry relies only upon knowledge of the atomic coordinates and captures features of the electron density both in the covalent and intermolecular domains. Transferability across systems of different compositions, conformations, and sizes will be addressed along with implications on quantum chemical applications and high-resolution X-Ray structure refinement.

Learning (from/about) the ground-state electron density

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.