Parallel eigensolvers in plane-wave Density Functional Theory

TL;DR

This paper presents a scalable parallel eigensolver algorithm based on Chebyshev polynomials for plane-wave Density Functional Theory, outperforming traditional methods at large processor counts.

Contribution

It introduces a new parallel eigensolver method that significantly improves scalability in large-scale electronic structure calculations.

Findings

Chebyshev polynomial-based eigensolver scales into tens of thousands of processors.

Outperforms block conjugate gradient algorithms for large computations.

Demonstrates improved parallel efficiency in plane-wave DFT calculations.

Abstract

We consider the problem of parallelizing electronic structure computations in plane-wave Density Functional Theory. Because of the limited scalability of Fourier transforms, parallelism has to be found at the eigensolver level. We show how a recently proposed algorithm based on Chebyshev polynomials can scale into the tens of thousands of processors, outperforming block conjugate gradient algorithms for large computations.