Update on the computation of the quenched $SU(6)$ Yang-Mills lattice spectrum

This paper reports on the computation of the low-lying glueball and non-singlet meson spectra in quenched SU(6) Yang-Mills theory using a multilevel sampling algorithm and APE-smeared Wilson loops to reduce statistical noise and optimize operator overlap for large-$N$ extrapolation.

The Gist

Imagine the universe is built from tiny, invisible Lego bricks. In our world, the "strong force" that glues these bricks together to form protons and neutrons is described by a theory called Quantum Chromodynamics (QCD).

Usually, these bricks are made of "quarks" (the matter) and "gluons" (the glue). But sometimes, physicists want to study the glue without the matter. They imagine a world where only the glue exists. In this world, the glue can clump together to form particles called Glueballs.

Finding these glueballs in the real world is like trying to hear a whisper in a hurricane. They mix so perfectly with other particles that it's incredibly hard to tell them apart. So, instead of looking at the real world, this team of scientists built a virtual universe inside a supercomputer to study them.

Here is what they did, explained simply:

1. The Virtual Universe (The Lattice)

Think of their supercomputer simulation as a giant 3D grid, like a massive chessboard made of invisible cubes. This is their "lattice."

• The Goal: They wanted to see what happens when you have a specific type of glue (called SU(6)) in this grid. Why SU(6)? It's a mathematical step up from our real world (which is SU(3)). By studying this "bigger" version, they can use math to predict what happens in our real, smaller world.
• The Challenge: In these simulations, the "noise" (random static) is huge. It's like trying to take a photo of a ghost in a dark room; the picture comes out grainy and blurry.

2. The Noise-Canceling Headphones (Multilevel Sampling)

To fix the grainy photos, the team invented a special technique called Multilevel Sampling.

• The Analogy: Imagine you are trying to measure the average temperature of a giant swimming pool. If you just dip a thermometer in once, you might get a weird reading because of a random cold current.
• The Trick: Instead of one measurement, they divide the pool into sections. They measure the temperature of a small section many, many times to get a perfect average for that spot. Then, they combine those perfect averages to get the temperature of the whole pool.
• The Result: This "nested averaging" cancels out the random static. It allowed them to see the "ghosts" (the glueballs) clearly for the first time in this specific type of simulation.

3. The Detective Work (Finding the Particles)

Once they had clean data, they had to figure out what particles were hiding there.

• The Operators: They built different "sensors" (mathematical loops) to look for specific shapes of glue. Some sensors looked for round balls, others for squashed ones.
• The Variational Method: This is like tuning a radio. They tried different combinations of sensors until they found the "station" that played the clearest signal. That signal told them the mass (weight) of the particle.

4. The Meson Side-Quest

While the main goal was the glueballs, they also looked at Mesons.

• The Analogy: If glueballs are made of pure glue, mesons are like a "dance pair" made of two quarks holding hands.
• They simulated a world with two types of quarks dancing together and measured how heavy they were. This helps them understand how the glue and the matter interact.

5. The Big Picture: Why Do This?

Why go through all this trouble?

• The String Theory Connection: Some physicists believe that at a fundamental level, these particles aren't just blobs of matter, but tiny vibrating strings (like the strings on a guitar).
• The Prediction: If this "String Theory" idea is true, the weights of these glueballs should follow a very specific, straight-line pattern (like notes on a musical scale).
• The Verdict: By measuring these weights with extreme precision, this team is testing whether the universe really plays by the rules of String Theory.

Summary

In short, this paper is a report from a team of digital architects who built a super-clean, noise-free virtual universe. They used a clever "nested averaging" trick to filter out the static, allowing them to weigh the invisible "glue balls" and "quark dancers" with unprecedented accuracy. Their goal? To see if the universe's fundamental structure is actually made of tiny, vibrating strings.

Technical Summary

1. Problem Statement and Motivation

The paper addresses the challenge of determining the non-perturbative spectrum of Quantum Chromodynamics (QCD), specifically focusing on glueballs (bound states of gluons) and mesons.

• Experimental Difficulty: While glueballs are a fundamental prediction of QCD, their experimental identification is hindered by strong mixing with flavor-singlet meson states.
• Theoretical Context: The study operates within the Large-$N$ expansion framework (introduced by 't Hooft), where the $SU(3)$ gauge group is generalized to $SU(N)$ with $N \to \infty$. In this limit, the theory simplifies into a weakly interacting system of mesons and glueballs, exhibiting a correspondence with string perturbation theory.
• Specific Goal: The authors aim to test a Semi-Topological String Field Theory (STFT) model which predicts rigid, exactly linear Regge trajectories for glueballs. To do this, they require high-precision lattice data for $SU(6)$ Yang-Mills theory to extrapolate toward the $N \to \infty$ limit.

2. Methodology

The authors employ Lattice Field Theory simulations using the quenched approximation (ignoring dynamical fermion loops, justified in the large-$N$ limit where such contributions are suppressed by $1/N$).

A. Glueball Spectrum Computation

• Operator Basis: They construct a basis of gauge-invariant operators using spatial Wilson loops at various levels of APE smearing (0, 2, 4, 6, 8 levels with weight $\alpha=1.0$).
• Symmetry Projection: These loops are combined to form linear combinations transforming under irreducible representations ($J^{PC}$) of the cubic lattice symmetry group $O_h$. The basis includes operators for channels such as $A_1^{++}$, $E^{++}$, $T_1^{++}$, etc.
• Variational Method: To extract masses, they compute the correlation matrix $C_{ij}(t)$ and solve the Generalized Eigenvalue Problem (GEVP) to isolate the ground state and excited states, minimizing contamination from higher energy levels.
• Multilevel Sampling (Key Innovation): To overcome the exponential signal-to-noise degradation at large time separations, they implement a 2-level sampling algorithm (Lüscher-Weisz method).
  • The lattice is partitioned into two regions ($\Lambda_1, \Lambda_2$) separated by fixed boundaries.
  • The path integral factorizes, allowing for a nested Monte Carlo average: outer loops generate boundary configurations, while inner loops sample the bulk regions independently.
  • This drastically reduces statistical noise for long-time correlators.

B. Meson Spectrum Computation

• Setup: Simulations include two degenerate quark flavors ($m_{q1} = m_{q2}$) to study non-singlet mesons.
• Operators: Interpolating fields of the form $\bar{\psi}\Gamma\psi$ are used for Pseudo-scalar ($0^{-+}$, using $\gamma_5$) and Vector ($1^{--}$, using $\gamma_1$) channels.
• Inversion: Quark propagators are computed by inverting the Wilson-Dirac operator using the Generalized Conjugate Residual (GCR) algorithm, preconditioned with the Schwarz Alternating Procedure (SAP) to accelerate convergence.
• Noise Reduction: Stochastic noise vectors ($Z_2$ distribution) localized at the source time-slice are used. The calculation exploits $\gamma_5$-hermiticity and time-reversal symmetry to improve precision.

C. Lattice Parameters

• Gauge Group: $SU(6)$.
• Action: Simple Wilson plaquette action.
• Ensembles: Two lattice spacings were simulated:
  • Lattice 1: $\beta = 25.55$, $L^3 \times T = 20^3 \times 16$.
  • Lattice 2: $\beta = 26.22$, $L^3 \times T = 24^3 \times 20$.
• Statistics: Approximately 1700–2000 configurations, with multilevel parameters ($N_0$ boundary configs, $N_1$ sub-measurements) tuned to optimize noise reduction.

3. Key Contributions

1. High-Precision $SU(6)$ Data: The paper provides the first detailed computation of the low-lying glueball and meson spectra specifically for $SU(6)$, a crucial step toward the large-$N$ limit.
2. Implementation of Multilevel Sampling: The successful application of the 2-level algorithm to $SU(6)$ glueball correlators demonstrates a significant reduction in statistical noise, enabling reliable mass extraction in channels where standard sampling fails at large time separations.
3. Operator Optimization: The use of a large basis of smeared Wilson loops combined with the GEVP allows for the isolation of ground states with high fidelity, even in channels with significant excited state contamination.
4. Validation Framework: The results serve as a benchmark for testing the Semi-Topological String Field Theory (STFT) predictions regarding linear Regge trajectories.

4. Results

The authors present preliminary mass estimates (in lattice units $a \cdot m$) for both glueballs and mesons.

Glueball Spectrum (Table 5):

• $A_1^{++}$ (Scalar): $0.68(4)$ at $\beta=25.55$; $0.471(74)$ at $\beta=26.22$.
• $E^{++}$: $1.13(19)$ at $\beta=25.55$; $0.98(7)$ at $\beta=26.22$.
• $T_2^{++}$: $0.93(8)$ at $\beta=26.22$.
• $T_2^{-+}$: $0.92(7)$ at $\beta=26.22$.
• Observation: In most channels, the spectral analysis at early time separations ($t_0=2$) showed only a single exponential mode surviving the noise, necessitating the use of reduced operator bases for the GEVP.

Meson Spectrum (Table 6):

• Pseudo-scalar ($0^{-+}$): $a \cdot m = 0.1869(34)$.
• Vector ($1^{--}$): $a \cdot m = 0.36(9)$.
• Observation: Clear effective mass plateaus were identified for both channels, allowing for constant fits to determine the ground state masses.

5. Significance

• Testing String Theory Duals: The high-precision data obtained for $SU(6)$ is essential for validating or falsifying the STFT approach, which predicts specific relationships between glueball masses (linear Regge trajectories).
• Large-$N$ Extrapolation: By establishing the spectrum at $N=6$, the authors contribute to the broader goal of understanding the $N \to \infty$ limit of QCD, where the theory becomes a string theory.
• Algorithmic Advancement: The successful integration of multilevel sampling with modern hardware accelerators and domain-decomposition solvers (SAP) sets a new standard for precision in lattice gauge theory, particularly for heavy gauge groups where computational costs are high.
• Experimental Relevance: While quenched, these results provide a clean theoretical baseline for the glueball sector, helping to disentangle theoretical predictions from the complex mixing observed in real-world ($N=3$) experimental searches.