Real-Space Stochastic GW Calculations Benchmark on GW20

Area of Science:

• Computational Quantum Chemistry
• Materials Science

Background:

• The G0W0 method is crucial for calculating electronic properties.
• Stochastic approaches offer scalability for large systems.
• Previous methods faced challenges with accuracy and computational cost.

Purpose of the Study:

• To implement and benchmark the stochastic G0W0 method in the PARSEC DFT code.
• To improve the accuracy and efficiency of quasiparticle energy calculations.
• To identify optimal embedded states for stochastic G0W0 calculations.

Main Methods:

• Stochastic resolution of identity for Green's function approximation.
• Hybrid approach combining stochastic and deterministic calculations.
• Implementation within the real-space finite-difference DFT code PARSEC.
• Benchmarking on GW100 and GW20 datasets.

Main Results:

• Excellent agreement with established G0W0 implementations.
• Demonstrated a reduction in stochastic error by up to a factor of 5.
• Identified key Kohn-Sham orbitals for embedding to enhance accuracy.
• Preserved computational efficiency despite error reduction.

Conclusions:

• The stochastic G0W0 method with embedded orbitals is accurate and efficient for large systems.
• Embedding a small set of Kohn-Sham orbitals significantly improves accuracy.
• Provides practical guidance for optimizing stochastic G0W0 calculations.