Efficient and Linear-Scaling Seminumerical Method for Local Hybrid Density Functionals.

Area of Science:

• Computational Chemistry
• Quantum Mechanics
• Materials Science

Background:

• Local hybrid functionals offer improved accuracy for static correlation in electronic structures.
• Their widespread adoption is limited by high computational costs compared to conventional methods.

Purpose of the Study:

• To develop a computationally efficient, linear-scaling method for local hybrid functionals.
• To enable accurate calculations of electronic structures with static correlation.

Main Methods:

• A seminumerical integration scheme was employed for evaluating local hybrid exchange-correlation potentials and nuclear forces.
• The preLinK method was combined with explicit integral screening for enhanced computational efficiency.
• The new method is termed preLinX.

Main Results:

• The preLinX method achieves a low prefactor and linear-scaling computational cost.
• An 8-fold performance increase was observed for a DNA fragment compared to existing methods.
• The method demonstrates efficiency, accuracy, and multinode parallelization for large systems (>1000 atoms).

Conclusions:

• The preLinX method significantly reduces the computational cost of local hybrid functionals.
• This advancement makes accurate calculations of complex electronic structures more feasible.
• The method is suitable for large-scale simulations in computational chemistry and materials science.