A fully non-perturbative charm-quark tuning using machine learning

TL;DR

This paper introduces a non-perturbative, machine learning-based method for tuning a relativistic heavy-quark action specifically for charm quarks, aiming for accurate spectrum reproduction in lattice QCD simulations.

Contribution

It develops a novel, model-independent tuning approach for charm-quark actions using machine learning, validated against experimental charmonium spectra.

Findings

Successfully tuned the charm-quark action parameters

Achieved spectrum consistent with experimental data

Established a framework for future scattering calculations

Abstract

We present a relativistic heavy-quark action tuning for the charm sector on ensembles generated by the CLS consortium. We tune a particular 5-parameter action in an entirely non-perturbative and -- up to the chosen experimental input -- model-independent way using machine learning and the continuum experimental charmonium ground-state masses with various quantum numbers. In the end we are reasonably successful; obtaining a set of simulation parameters that we then verify produces the expected spectrum. In the future, we will use this action for finite-volume calculations of hadron-hadron scattering.