Directly patching high-level exchange-correlation potential based on fully determined optimized effective potentials.

Area of Science:

• Computational Quantum Chemistry
• Materials Science
• Condensed Matter Physics

Background:

• The exchange-correlation (XC) potential is central to Kohn-Sham (KS) density functional theory (DFT).
• Accurate XC potentials are crucial for predicting material properties but are challenging to compute for large systems.
• The recently proposed exchange-correlation potential patching (XCPP) method aims to construct high-level XC potentials in large systems by combining locally computed potentials.

Purpose of the Study:

• To investigate the performance of the XCPP method for patching exact exchange (EXX) and random phase approximation (RPA) correlation potentials.
• To address the challenge of an undetermined constant in cluster XC potentials obtained from the optimized effective potential equation.
• To develop a simple method for determining this constant and evaluate its impact on XCPP accuracy.

Main Methods:

• Developed a method to determine the unknown constant in locally computed XC potentials.
• Applied the XCPP method to patch EXX and RPA correlation potentials in one-dimensional model systems (H20, H10Li8, H19-H bond stretching).
• Investigated two definitions of EXX: adiabatic connection fluctuation dissipation theorem (ACFDT) and Hartree-Fock (HF).

Main Results:

• Effective error cancellations were observed between patched ACFDT-type EXX and patched RPA correlation potentials.
• These error cancellations were absent for HF-type EXX, linked to the integral of the HF-type EXX hole not being -1 for fractional occupations.
• KS spectra and band gaps computed using XCPP showed reasonable agreement with benchmark results as system size increased.

Conclusions:

• The developed method successfully determines the unknown constant, enabling more accurate XC potential patching.
• XCPP with ACFDT-type EXX and RPA correlation demonstrates effective error cancellation, improving accuracy for large systems.
• The XCPP method shows promise for accurate electronic structure calculations of large systems in DFT.