New Noise Subtraction Methods for Lattice QCD Calculations

Abstract

Lattice QCD is a set of numerical techniques which uses a finite space-time lattice to simulate the interactions between quarks and gluons. In lattice QCD, one of the important and challenging issues is calculation of disconnected quark loops. In order to calculate them many matrix inversions are required. To tackle this problem, instead of a direct approach, we use isolated stochastic approach along with approximation techniques Wilcox, Numerical Challenges in Lattice Quantum Chromodynamics. Lecture Notes in Computer Science and Engineering, vol. 15, (Springer Berlin, 2000), p. 127. Noise theory is used to project out the loop operator expectation values and new noise subtraction methods are applied to reduce the statistical uncertainty. Perturbative subtraction Bernardson et al. (Comput Phys Commun 78:256, 1993) is a standard noise subtraction method which we will be comparing to and attempting to improve upon. This paper will be focusing on eigenspectrum subtraction (deflation), polynomial subtraction, and combination methods.