Ab initio electronic structure calculations using a real-space Chebyshev-filtered subspace iteration method.

Area of Science:

• Computational physics
• Quantum chemistry
• Materials science

Background:

• Ab initio electronic structure calculations are crucial for understanding material properties.
• Traditional self-consistent field (SCF) methods scale cubically, limiting their application to large systems.
• The Chebyshev-filtered subspace iteration (CheFSI) method offers a promising alternative for accelerating SCF procedures.

Purpose of the Study:

• To implement and validate a novel computational approach for solving the Kohn-Sham equation efficiently.
• To develop and assess the ab initio Real-space Electronic Structure (ARES) software for large-scale simulations.
• To enhance the initial subspace generation scheme for Chebyshev filtering within ARES.

Main Methods:

• Employed a combination of real-space finite-difference formulation and CheFSI to solve the Kohn-Sham equation.
• Implemented the approach in the ARES software, designed for multi-processor, parallel environments.
• Developed an improved scheme for efficient initial subspace generation in Chebyshev filtering.

Main Results:

• Demonstrated the accuracy, stability, and efficiency of the ARES software.
• Successfully simulated large-scale crystalline systems containing thousands of atoms.
• Validated the effectiveness of the enhanced initial subspace generation scheme.

Conclusions:

• The ARES software, utilizing CheFSI and real-space methods, provides an efficient and accurate solution for large-scale electronic structure calculations.
• The implemented approach overcomes the limitations of traditional SCF methods for complex systems.
• The study highlights the potential of ARES for advancing materials science research through large-scale simulations.