Lattice QCD at the physical point: simulation and analysis details

S. Dürr, Z. Fodor, Christian Hoelbling, S. D. Katz, Stefan Krieg, Thorsten Kurth, Laurent Lellouch, Thomas Lippert, K. K. Szabó, G. VulvertJournal of High Energy Physics 2011 · DOI

Abstract

We present a precise determination of the light quark masses $m_{ud} = (m_u + m_d)/2$ and $m_s$ in $2+1$ flavor QCD. By simulating pion masses down to 120 MeV—reaching the physical point—at five different lattice spacings and in large volumes, we eliminate the need for traditional chiral extrapolations. Our analysis employs a tree-level $O(a^2)$-improved Symanzik gauge action and clover-improved Wilson fermions with hypercubic (HEX) smearing. We report values for $m_{ud}$, $m_s$, and individual $m_u$ and $m_d$ masses (using dispersive input), achieving sub-2% precision and definitively excluding the massless up-quark solution to the strong CP problem.

Introduction

Quark masses are fundamental parameters of the Standard Model, yet their determination is non-trivial due to color confinement. Lattice QCD provides a first-principles framework to calculate these masses by solving Quantum Chromodynamics on a discretized 4D spacetime grid. Historically, such simulations were hampered by the high computational cost of light quarks, requiring researchers to simulate at unphysically high pion masses and extrapolate results. This "chiral extrapolation" introduced significant systematic uncertainties. The Budapest-Marseille-Wuppertal (BMW) collaboration overcomes this by simulating directly at the physical point.

Datasets and Simulation Landscape

The study utilizes 47 unique gauge ensembles across five lattice spacings ($0.054$ to $0.116$ fm). This range allows for a robust continuum extrapolation ($a \to 0$). To minimize finite-size effects, simulations were conducted in large volumes with spatial extents up to $L \approx 6$ fm.

Simulation landscape showing the distribution of lattice ensembles across pion masses and spatial volumes. Shaded regions indicate the magnitude of finite-volume effects.

As shown in the figure above, the simulation points bracket the physical pion mass ($M_\pi \approx 135$ MeV), transforming the problem from one of extrapolation to one of more stable interpolation. The shaded regions represent contours of finite-volume effects, demonstrating that the chosen volumes keep these corrections well under control (mostly below 1%).

Technical Innovations: Action and Smearing

A critical component of this work is the use of two levels of HEX (hypercubic) smearing for the clover-improved Wilson fermions. Smearing is essential for simulation stability at very light quark masses, but it must be implemented carefully to maintain the locality of the fermion operator.

Comparison of the locality of the fermion operator between 2-step HEX smearing and 6-step stout smearing.

Locality tests confirm that the HEX-smeared action exhibits a rapid exponential decay of the derivative of the Dirac operator with respect to the gauge field. This ensures that the lattice action remains a valid approximation of the continuum theory while providing the numerical stability required to reach the physical point.

Renormalization and Scale Setting

The lattice spacing is set using the mass of the $\Omega$ baryon, a stable "ruler" that is relatively insensitive to light quark mass variations. For mass renormalization, the collaboration employs the non-perturbative RI/MOM scheme. To bridge the gap between lattice data and perturbative $\overline{MS}$ results (the "window problem"), the running of the scalar density is evolved to a high scale ($\mu = 4$ GeV) in the continuum before matching with perturbation theory.

Key Results

The final quark masses, reported in the $\overline{MS}$ scheme at a scale of 2 GeV, are:

  • Average light quark mass ($m_{ud}$): $3.469(47)(48)$ MeV
  • Strange quark mass ($m_s$): $95.5(1.1)(1.5)$ MeV
  • Up quark mass ($m_u$): $2.15(03)(10)$ MeV
  • Down quark mass ($m_d$): $4.79(07)(12)$ MeV
  • Mass Ratio ($m_s/m_{ud}$): $27.53(20)(08)$

By combining lattice results for $m_{ud}$ with dispersive information from $\eta \to 3\pi$ decays, the collaboration separated the up and down quark masses. The results show that $m_d$ is approximately twice as heavy as $m_u$. Crucially, the non-zero value of $m_u$ excludes the massless up-quark solution to the strong CP problem by more than 20 standard deviations.

Conclusion

This work represents a significant milestone in precision Lattice QCD. By simulating at the physical point and employing sophisticated renormalization techniques, the BMW collaboration has provided some of the most accurate determinations of light quark masses to date. Future improvements will focus on including dynamic charm quarks and incorporating electromagnetic (QED) effects directly into the lattice simulations to further reduce systematic uncertainties.

Summary updated

Abstract

We give details of our precise determination of the light quark masses m ud = (m u + m d )/2 and m s in 2 + 1 flavor QCD, with simulated pion masses down to 120 MeV, at five lattice spacings, and in large volumes. The details concern the action and algorithm employed, the HMC force with HEX smeared clover fermions, the choice of the scale setting procedure and of the input masses. After an overview of the simulation parameters, extensive checks of algorithmic stability, autocorrelation and (practical) ergodicity are reported. To corroborate the good scaling properties of our action, explicit tests of the scaling of hadron masses in N f = 3 QCD are carried out. Details of how we control finite volume effects through dedicated finite volume scaling runs are reported. To check consistency with SU(2) Chiral Perturbation Theory the behavior of M π 2 /m ud and F π as a function of m ud is investigated. Details of how we use the RI/MOM procedure with a separate continuum limit of the running of the scalar density R S (μ, μ′) are given. This procedure is shown to reproduce the known value of r 0 m s in quenched QCD. Input from dispersion theory is used to split our value of m ud into separate values of m u and m d . Finally, our procedure to quantify both systematic and statistical uncertainties is discussed.
BibTeX
@article{durr2011lattice,
  author = {S. Dürr and Z. Fodor and Christian Hoelbling and S. D. Katz and Stefan Krieg and Thorsten Kurth and Laurent Lellouch and Thomas Lippert and K. K. Szabó and G. Vulvert},
  title = {{Lattice QCD at the physical point: simulation and analysis details}},
  journal = {Journal of High Energy Physics},
  year = {2011},
  doi = {10.1007/jhep08(2011)148},
}
Figure