Update on the computation of the quenched S ​ U ​ ( 6 ) SU(6) Yang-Mills lattice spectrum

Francesco Scardino · 2026

Read Update on the computation of the quenched S ​ U ​ ( 6 ) SU(6) Yang-Mills lattice spectrum by Francesco Scardino (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-13138.txt.

Introduction

Andrea Falzetti and co-authors report progress on the quenched SU(6) Yang-Mills lattice spectrum, aimed at the large-N limit. The team measures low-lying glueball and meson states in this gauge theory. A multilevel sampling algorithm cuts statistical noise at large time separations. Their gluon operators use spatial Wilson loops at varied smearing levels, with zero momentum picked for clean overlap. Meson spectra for two light flavors are computed in parallel. The work extends a series across gauge groups.

Catalog record

CREATORFrancesco Scardino
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19097
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.13138

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Update on the computation of the quenched S ​ U ​ ( 6 ) SU(6) Yang-Mills lattice spectrum Francesco Scardino Abstract We report on our continued efforts to measure the glueball and meson spectra in SU( N N ) Yang-Mills theory and QCD with the aim of extrapolating to the large- N N limit. In particular, we document the computation of the low-lying SU( 6 6 ) spectrum. We employ a multilevel sampling algorithm to measure glueball correlators to reduce statistical noise in the large-time separation limit. The gluon operator basis is composed of spatial Wilson loop measured at different levels of (APE) smearing, with vanishing momentum selected to maximise the orthonogality of the operators and their overlap with the lowest lying states. We also report on analogous computations for the J = 0 , 1 J=0,1 non-singlet meson spectrum with two degenerate quark flavors. [a,b]Andrea Falzetti 1 Introduction Quantum Chromodynamics (QCD) is the theory of the strong interactions. It is a non-Abelian gauge theory with gauge group SU(N) coupled to fermionic matter, with N=3 in the real world. Even its pure-gauge sector, namely SU(N) Yang–Mills theory, has a highly non-trivial dynamics for N>1 [ 1 ] . Its spectrum contains the so-called glueball states, created by gauge-invariant composite operators. The existence of glueballs is one of the most important theoretical predictions of QCD [ 2 ] , and also one of the most difficult to confirm experimentally. Indeed, searches for glueball states have been ongoing for decades [ 3 , 4 ] ; nevertheless, the experimental determination of the glueball spectrum remains extremely difficult because of the strong mixing with flavour-singlet meson states predicted by the theory. As a consequence, although several glueball candidates have been proposed, no definitive direct determination of a glueball mass in real-world QCD has yet been achieved with complete confidence. At high energies, asymptotic freedom allows QCD to be successfully studied in perturbation theory, for instance in deep inelastic scattering [ 5 , 6 ] . However, because of confinement, we are presently unable to derive analytically either the physical hadronic spectrum or the full phenomenology of hadronic physics, let alone the exact \mathcal{S} -matrix of the theory. In particular, the appearance of the renormalisation-group (RG) invariant mass scale \Lambda_{\mathrm{QCD}} , which vanishes at every order in perturbation theory, implies that perturbative methods fail precisely for the physical masses of the theory. For this reason, the most successful first-principles approach to QCD is its regularisation on a spacetime lattice, followed by a numerical evaluation of the functional integral [ 7 , 8 , 9 ] . The glueball spectrum can then be extracted from the large-time behaviour of two-point correlators of purely gluonic operators. This provides a natural first-principles framework for determining the spectrum. A complementary analytical perspective is provided by the large- N expansion introduced by ’t Hooft, in which the SU(3) colour gauge group is generalized to SU(N) and observables are expanded in powers of 1/N around the N\to\infty limit [ 10 ] . In the double-line representation, the corresponding Feynman diagrams are topologically organised by the genus expansion of punctured Riemann surfaces, each topology being weighted by a factor N^{\chi} , where \chi=2-2g-n is the Euler characteristic [ 10 , 11 ] . This structure matches that of a string perturbation theory with closed-string coupling g_{s}=1/N [ 12 ] . This correspondence has long been taken as evidence for the existence of a non-perturbative string description of large- N Yang–Mills theory and large- N QCD [ 10 , 11 , 13 ] . In such a picture, glueballs and mesons arise respectively as closed- and open-string excitations, and in asymptotically free theories the string tension is expected to be set by the RG-invariant scale of the gauge theory [ 14 ] . In the N\to\infty limit, the theory simplifies considerably, becoming an effective theory of infinitely many weakly interacting mesons and glueballs. The long-term goal of the present study is to test a Semi-Topological String Field Theory (STFT) approach defined along the lines of Ref. [ 15 ] . This framework leads to very rigid predictions for the glueball spectrum in the ’t Hooft limit, in particular in terms of exactly linear Regge trajectories [ 14 ] . Our aim is to push the numerical precision of lattice determinations by combining modern hardware accelerators with multilevel sampling techniques, and to use the resulting lattice data to validate or falsify this scenario. In this contribution, we present measurements of the glueball and meson spectra for the gauge group SU( 6 ). 2 Glueball Operator Basis We employ a glueball operator basis composed of spatial loops at different levels of (APE) smearing [ 16 ] . Measurements are done at 0,2,4,6,8 smearing levels, with a smearing weight of \alpha=1.0 [ 17 ] . J^{PC} N. of Operators after Smearing A_{1}^{++} 9 45 A_{2}^{++} 2 10 E^{++} 20 100 T_{2}^{++} 12 60 T_{1}^{+-} 9 45 T_{2}^{+-} 9 45 A_{2}^{-+} 2 10 J^{PC} N. of Operators after Smearing E^{-+} 2 10 T_{1}^{-+} 27 135 T_{2}^{-+} 12 60 A_{1}^{--} 2 10 E^{--} 4 20 T_{1}^{--} 3 15 T_{2}^{--} 9 45 Table 1: Results of the subduction of the glueball operators onto the cubic group The different loop shapes included in our computation are depicted in figure 1 . These loops are combined with their rotations to produce linear combinations that transform according to irreducible representations of the symmetry group of the lattice O_{h} . At each lattice site we compute all loops and linear combinations prescribed by group theory, and then sum them over each timeslice to project onto the p=0 mode. These operators form our bases for each representation, the number of operator in each channel is listed in table 1 . Figure 1: Summary of the loop shapes used in the glueball spectrum measurements. Once we compute the full correlator matrix in each symmetry channel we extract the glueball masses with the usual variational approach. We define the matrix \displaystyle C_{ij}(\Delta t)=\left<\left<\mathcal{O}_{i}(t_{0})\mathcal{O}_{j}(t_{0}\pm\Delta t\right>\right>_{t_{0},\pm}\,, (1) where \left<\cdots\right>_{t_{0},\pm} indicates a suitable average over the sink position t_{0} illustrated in section 3 . Then, to extract the spectrum we can look for solutions v^{(k)},\lambda^{(k)} of the generalized eigenvalue problem (GEVP): \displaystyle C_{ij}(t_{0})v_{j}^{(k)}(t_{0},t_{1})=\lambda^{(k)}(t_{0},t_{1})\cdot C_{ij}(t_{1})v_{j}^{(k)}(t_{0},t_{1})\,. (2) 3 2-level Sampling The locality of the Wilson plaquette action can be exploited to improve the correlator precision in the long time separation region [ 18 , 19 ] . Indeed the contribution of a single link U_{\mu}(x) to the Wilson action is: S(U_{\mu}(x))=\frac{\beta}{N}U_{\mu}(x)V_{\mu}(x)^{\dagger}\,, (3) where V_{\mu}(x) is the sum of all the staples that one can build around the link U_{\mu}(x) . If we partition our lattice into two distinct regions, \Lambda_{1} and \Lambda_{2} , separated by fixed boundaries \partial\Lambda , in such a way that for every U_{\mu}(x) the associated V_{\mu}(x) depends only on link variables located within \Lambda_{1} or on the boundary 1 1 1 For the Wilson simple plaquette action, this can be achieved by choosing as the boundary the spatial link variables lying on two different timeslices. For improved actions that include larger Wilson loops, it may be necessary to increase the thickness of the boundary region between the two domains to guarantee their separation, at the cost

Read the complete text