LibRPA

LibRPA is a comprehensive library designed for efficient and accurate first-principles calculations of random-phase approximation (RPA) correlation energies using numerical atomic orbitals (NAOs). It leverages the Localized Resolution of Identity (LRI) technique to achieve \(O(N^2)\) scaling behavior, making it highly suitable for large-scale periodic systems. Implemented in C++ and Python with MPI/OpenMP parallelism, LibRPA integrates seamlessly with various density functional theory (DFT) packages through flexible file-based and API-based interfaces.

Key Features:

• Accurate computation of physical properties using many-body perturbation theory.

• Efficient handling of large-scale periodic systems with \(O(N^2)\) scaling.

• Hybrid parallelism (MPI/OpenMP) to enhance scalability and performance.

• Seamless integration with various DFT packages through flexible interfaces.

Quick Links:

• Installation Guide

• Driver Usage

• Compile Options

• Input Parameters

How to Cite:

• Shi R, Lin P, Zhang M Y, et al. Subquadratic-scaling real-space random phase approximation correlation energy calculations for periodic systems with numerical atomic orbitals[J]. Physical Review B, 2024, 109(3): 035103.