Efficient computation of low-lying eigenmodes of non-Hermitian Wilson-Dirac type matrices

TL;DR

This paper introduces a polynomial transformation technique for non-Hermitian matrices, specifically Wilson-Dirac type, enabling efficient computation of low-lying eigenmodes and significantly speeding up the eigenmode algorithms.

Contribution

The authors present a novel polynomial transformation method that accesses spectral windows and accelerates eigenmode computations for Wilson-Dirac matrices.

Findings

Enables determination of low-lying eigenmodes efficiently

Provides substantial acceleration of eigenmode algorithms

Applicable to Wilson-Dirac type matrices

Abstract

A polynomial transformation for non-Hermitian matrices is presented, which provides access to wedge-shaped spectral windows. For Wilson-Dirac type matrices this procedure not only allows the determination of the physically interesting low-lying eigenmodes but also provides a substantial acceleration of the eigenmode algorithm employed.