Self-consistent GW method: O(N) algorithm for polarizability and self energy

TL;DR

This paper presents an efficient self-consistent GW computational method with linear scaling for polarizability and self-energy, enabling practical calculations on larger systems in material science.

Contribution

The paper introduces a linear-scaling implementation of the self-consistent GW method within the FlapwMBPT code, improving computational efficiency for large systems.

Findings

Scaling between linear and quadratic in system size.

Successful application to complex materials like CoSbS, La2CuO4, and SmB6.

Demonstrated potential for large-scale material simulations.

Abstract

An efficient implementation of the self-consistent GW method in the FlapwMBPT code (https://www.bnl.gov/cmpmsd/flapwmbpt/) is presented. It features the evaluation of polarizability and self-energy which scales linearly with respect to the system size. Altogether the computational time scaling was measured to be between linear and quadratic in the applications to silicon supercells with up to 72 atoms. Application to such materials as paracostibite CoSbS, supercells of La$_{2}$CuO$_{4}$ (up to 56 atoms) and SmB$_{6}$ illustrate the potential of the approach in computational material science.