Login / Signup

Interpretation and Automatic Generation of Fermi-Orbital Descriptors.

Sebastian SchwalbeKai TrepteLenz FiedlerAlex I JohnsonJakob KrausTorsten HahnJuan E PeraltaKoblar A JacksonJens Kortus
Published in: Journal of computational chemistry (2019)
We present an interpretation of Fermi-orbital descriptors (FODs) and argue that these descriptors carry chemical bonding information. We show that a bond order derived from these FODs agrees well with reference values, and highlight that optimized FOD positions used within the Fermi-Löwdin orbital self-interaction correction (FLO-SIC) method correspond to expectations from Linnett's double-quartet theory, which is an extension of Lewis theory. This observation is independent of the underlying exchange-correlation functional, which is shown using the local spin density approximation, the Perdew-Burke-Ernzerhof generalized gradient approximation (GGA), and the strongly constrained and appropriately normed meta-GGA. To make FOD positions generally accessible, we propose and discuss four independent methods for the generation of Fermi-orbital descriptors, their implementation as well as their advantages and drawbacks. In particular, we introduce a re-implementation of the electron force field, an approach based on the centers of mass of orbital densities, a Monte Carlo-based algorithm, and a method based on Lewis-like bonding information. All results are summarized with respect to future developments of FLO-SIC and related methods. © 2019 The Authors. Journal of Computational Chemistry published by Wiley Periodicals, Inc.
Keyphrases
  • primary care
  • healthcare
  • monte carlo
  • deep learning
  • machine learning
  • single molecule
  • health information
  • systematic review
  • current status
  • neural network