Accurate and Efficient Parallel Implementation of an Effective Linear-Scaling Direct Random Phase Approximation...

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Materials Science

Background:

• Calculating correlation energy in large systems is computationally demanding.
• Traditional methods like canonical molecular orbital resolution-of-the-identity RPA (RI-RPA) exhibit quartic-scaling behavior.
• Memory bottlenecks hinder the application of atomic orbital (AO) based RI-RPA methods for large molecules.

Purpose of the Study:

• To develop an efficient algorithm for calculating RPA correlation energy.
• To achieve linear-scaling behavior for large molecular systems.
• To reduce memory requirements in AO-RI-RPA calculations.

Main Methods:

• Formulation in local atomic orbital space for linear scaling.
• Optimized minimax integration schemes and a local Coulomb metric for high accuracy.
• Precontraction of 3-center integral matrices with Cholesky factors to reduce memory.

Main Results:

• The new algorithm achieves accuracy comparable to canonical RI-RPA.
• Demonstrates effective linear-scaling behavior for large systems.
• Reduces memory requirements by a factor of [Formula: see text] and enables parallel computation.

Conclusions:

• The developed method offers a computationally efficient and accurate approach for RPA correlation energy calculations.
• It overcomes memory limitations of previous AO-RI-RPA implementations.
• The method extends the applicability to large and delocalized systems, facilitating studies on complex materials like covalent organic frameworks.