Benchmarking the Starting Points of the GW Approximation for Molecules

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Materials Science

Background:

• The GW approximation is crucial for accurate quasiparticle energy calculations in atoms and molecules.
• Perturbative GW calculations depend heavily on the chosen self-consistent mean-field starting point.

Purpose of the Study:

• To benchmark various starting points for perturbative GW calculations.
• To evaluate the performance of different approximations for ionization energy calculations.
• To assess the reliability of the frozen-core approximation in GW calculations.

Main Methods:

• Utilized a newly developed GW code employing Gaussian basis functions.
• Benchmarked Hartree-Fock, LDA, PBE, PBE0, B3LYP, HSE06, BH&HLYP, CAM-B3LYP, and tuned CAM-B3LYP as starting points.
• Investigated the impact of the frozen-core approximation on computational speed and accuracy.

Main Results:

• Hybrid functionals significantly outperform Hartree-Fock, LDA, and semilocal approximations as starting points for ionization energy calculations.
• Among hybrid functionals, those with a higher percentage of exact exchange yield the best results.
• The frozen-core approximation offers substantial computational speed-up with reliable accuracy.

Conclusions:

• The choice of starting point critically influences the accuracy of perturbative GW calculations.
• Hybrid functionals, particularly those rich in exact exchange, are recommended for accurate ionization energy predictions using GW.
• The frozen-core approximation is a viable strategy for accelerating GW calculations without compromising reliability.