Direct inversion of the iterative subspace with contracted planewave basis functions

Area of Science:

• Computational Chemistry
• Materials Science
• Quantum Mechanics

Background:

• Periodic electronic structure calculations are computationally intensive.
• Contracted planewave (CPW) basis functions offer advantages of both planewave (PW) and localized basis functions.
• Efficient methods are needed to reduce computational cost in these calculations.

Purpose of the Study:

• To develop and assess a direct inversion of the iterative subspace (DIIS) method optimized for CPW basis functions.
• To leverage unique CPW properties for efficient electronic wavefunction convergence.
• To reduce the computational expense of periodic electronic structure calculations.

Main Methods:

• Development of a DIIS method utilizing CPW basis functions.
• Exploitation of PW-based electronic structure representation for efficient matrix-vector products.
• Transformation of matrix-vector products for direct diagonalization and DIIS application.
• Periodic Hartree-Fock calculations for assessment.

Main Results:

• The developed DIIS method achieves approximately five times speedup compared to full Fock matrix calculations with CPW basis sets.
• The method shows weak dependence on basis set size, allowing for larger basis sets with minimal performance impact.
• Efficient convergence of electronic wavefunctions was demonstrated.

Conclusions:

• The novel DIIS method significantly enhances the efficiency of periodic electronic structure calculations using CPW basis functions.
• This approach provides a computationally tractable route to increased variational flexibility.
• The findings pave the way for more extensive applications of CPW basis functions in computational chemistry and materials science.