Real-space density kernel method for Kohn-Sham density functional theory calculations at high temperature.

Area of Science:

• Computational Physics
• Materials Science
• Quantum Chemistry

Background:

• Kohn-Sham density functional theory (KS-DFT) calculations become computationally intensive at high temperatures.
• Conventional diagonalization methods require computing numerous partially occupied states, increasing costs.
• Efficient methods are needed for high-temperature electronic structure simulations.

Purpose of the Study:

• To develop a computationally efficient density matrix-based method for high-temperature KS-DFT.
• To eliminate the need for expensive diagonalization steps in KS-DFT calculations.
• To enable accurate simulations of materials at extreme temperatures.

Main Methods:

• A density matrix-based approach for Kohn-Sham calculations at high temperatures.
• Development of real-space expressions for electron density, free energy, forces, and stress tensor.
• Utilizing Chebyshev filtering to generate an auxiliary orbital basis.
• Employing spectral quadrature techniques for density kernel, band structure, and entropic energy calculations.
• Implementation within the SPARC electronic structure code.

Main Results:

• The new method significantly reduces computational cost by avoiding diagonalization.
• Systematic convergence of calculated quantities to exact diagonalization results was demonstrated.
• Significant speedups were achieved compared to conventional diagonalization methods.
• Accurate computation of aluminum's self-diffusion coefficient and viscosity at 116,045 K.

Conclusions:

• The presented density matrix method offers a substantial speedup for high-temperature KS-DFT.
• The method provides accurate results, comparable to traditional diagonalization techniques.
• This approach facilitates advanced quantum molecular dynamics simulations at extreme conditions.