A novel method for evaluating correlation functions in lattice hadron spectroscopy

TL;DR

This paper introduces a new stochastic method combining Laplacian-Heaviside smearing and noise dilution to efficiently evaluate hadronic correlation functions, especially on large lattices where exact methods are costly.

Contribution

The paper presents a novel stochastic approach that reduces variance in correlation function evaluations, enabling feasible computations on large lattice volumes.

Findings

Effective variance reduction in correlators

Successful application to disconnected diagrams and multi-hadron systems

Computational feasibility on large lattices

Abstract

We describe a new approach for evaluating hadronic correlation functions which combines Laplacian-Heaviside quark smearing with a stochastic estimator of quark propagators. This method utilizes noise dilution in a new way to reduce the variance in correlators. The efficacy of the new algorithm is demonstrated on a number of systems, including disconnected diagrams and multi-hadron correlators, on a small lattice where comparisons with the results obtained with exactly determined quark propagators are possible. On larger lattice volumes, the use of exact propagators becomes prohibitively expensive, while the stochastic method is still computationally feasible.