Exponential Error Reduction for Glueball Calculations Using a Two-Level Algorithm in Pure Gauge Theory

TL;DR

This paper introduces a two-level algorithm that exponentially reduces errors in glueball calculations within pure gauge theory, enabling more efficient mass extraction at large distances.

Contribution

It demonstrates the effectiveness of a multi-level sampling method in reducing statistical errors exponentially for glueball spectrum calculations in SU(3) gauge theory.

Findings

Exponential error reduction achieved at large distances.

Standard methods are more efficient at short distances due to variance saturation.

Constructed a variational basis with 35 Wilson loops for spectrum analysis.

Abstract

This study explores the application of a two-level algorithm to enhance the signal-to-noise ratio of glueball calculations in four-dimensional $\mathrm{SU(3)}$ pure gauge theory. Our findings demonstrate that the statistical errors exhibit an exponential reduction, enabling reliable extraction of effective masses at distances where current standard methods would demand exponentially more samples. However, at shorter distances, standard methods prove more efficient due to a saturation of the variance reduction using the multi-level method. We discuss the physical distance at which the multi-level sampling is expected to outperform the standard algorithm, supported by numerical evidence across different lattice spacings and glueball channels. Additionally, we construct a variational basis comprising 35 Wilson loops up to length 12 and 5 smearing sizes each, presenting results for the first state in the spectrum for the scalar, pseudoscalar, and tensor channels.