Improvements in the orbitalwise scaling down of Perdew-Zunger self-interaction correction in many-electron regions
Yoh Yamamoto, Selim Romero, Tunna Baruah, and Rajendra R. Zope

TL;DR
This paper introduces an improved orbitalwise scaled down self-interaction correction (OSIC) method for density functional theory, demonstrating enhanced accuracy across various molecular properties and proposing a selective scaling approach (SOSIC) to better preserve physical potential behavior.
Contribution
The paper develops and validates an improved OSIC scheme using the Fermi-L"owdin method with the SCAN functional, and introduces SOSIC for better asymptotic potential behavior.
Findings
OSIC-SCAN outperforms previous OSIC-LSDA, PBE, and TPSS results.
SOSIC maintains proper $-1/r$ potential behavior and yields accurate orbital eigenvalues.
SOSIC provides a more balanced description of energies and barrier heights.
Abstract
The Perdew-Zunger (PZ) method provides a way to remove the self-interaction (SI) error from density functional approximations on an orbital by orbital basis. The PZ method provides significant improvements for the properties such as barrier heights or dissociation energies but results in over-correcting the properties well described by SI-uncorrected semi-local functional. One cure to rectify the over-correcting tendency is to scale down the magnitude of SI-correction of each orbital in the many electron region. We have implemented the orbitalwise scaled down SI-correction (OSIC) scheme of Vydrov et al. [J. Chem. Phys. 124, 094108 (2006)] using the Fermi-L\"owdin SI-correction method. After validating the OSIC implementation with previously reported OSIC-LSDA results, we examine its performance with the most successful non-empirical SCAN meta-GGA functional. Using different forms of…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
