Self-Consistent GW via Conservation of Spectral Moments

Area of Science:

• Computational Chemistry
• Quantum Mechanics
• Electronic Structure Theory

Background:

• The Green's function (GW) approximation is a powerful tool for calculating electronic properties.
• Existing GW implementations face challenges in efficiency and self-consistency.
• Accurate simulation of charged excitations is crucial for understanding material and molecular behavior.

Purpose of the Study:

• To present an enhanced framework for GW simulations of charged excitations.
• To improve the efficiency and scalability of GW calculations.
• To explore and identify optimal self-consistent GW variants for accurate predictions.

Main Methods:

• Implementation of a GW framework based on spectral moments of the self-energy.
• Development of efficiency improvements and a parallelism strategy.
• Investigation of various self-consistent GW approximations, including a novel coupled chemical potential and Fock matrix optimization.
• Comparison with established methods and validation against the GW100 molecular test set.

Main Results:

• The new GW framework demonstrates improved efficiency and scalability, comparable to Hartree-Fock methods.
• A self-consistent variant using coupled chemical potential and Fock matrix optimization shows superior accuracy.
• Tamm-Dancoff approximation-based screening yields higher accuracy than random phase approximation.
• Accurate prediction of charged excitation spectra for Chlorophyll A, matching experimental data.

Conclusions:

• The spectral moment-based GW framework offers a computationally efficient and formally robust approach.
• The identified self-consistent GW variant provides highly accurate results for molecular systems.
• The findings advance the capability of GW methods for electronic structure calculations and material discovery.