Glueballs and quantum numbers

Prerequisites: operator overlap and glueball spectroscopy.

You will learn to identify simple operator channels, understand the cubic labels $A_1,A_2,E,T_1,T_2$, and explain why a channel energy is not automatically an isolated particle.

Scope: pure $SU(N)$ gauge theory in $3+1$ dimensions, at zero temperature and $\theta=0$. Discrete symmetries and their labels must be reconsidered when this setting changes.

Start with the operator, not the particle name

A glueball is a color-singlet excitation associated with gluonic observables. Calling an interpolating field a “glueball operator” does not prove it creates one stable particle. It can overlap with every state compatible with its quantum numbers, including multiparticle states.

For a continuum illustration write $E_i=F_{0i}$ and $B_i=\frac12\epsilon_{ijk}F_{jk}$, and take a color trace. Under spatial parity, $E_i$ is a vector and $B_i$ a pseudovector. Therefore

$$O_s=\operatorname{tr}\sum_i B_iB_i,\qquad O_p=\operatorname{tr}\sum_i E_iB_i$$

have respectively even and odd parity. Both are rotational scalars and have even charge conjugation in this setting, giving $0^{++}$ and $0^{-+}$ channels. These are formal continuum expressions until their composite-operator renormalization is specified; renormalized observables explains that qualification.

The scalar can have a vacuum expectation value. Use its connected correlator or explicitly subtract that expectation before interpreting a nonzero-energy decay.

A finite symmetry group sees less than full rotations

At nonzero lattice spacing a cubic spatial lattice has only the rotations preserving the cube. Its irreducible representations have dimensions $1,1,2,3,3$, conventionally named $A_1,A_2,E,T_1,T_2$. A lattice operator has an exact cubic label; continuum spin is inferred through a controlled approach to restored rotational symmetry.

A useful worked example is a symmetric traceless spatial tensor $Q_{ij}$. It has five components. Its two independent diagonal combinations,

$$Q_{xx}-Q_{yy},\qquad 2Q_{zz}-Q_{xx}-Q_{yy},$$

form the two-dimensional $E$ sector. Its three off-diagonal components $Q_{xy},Q_{yz},Q_{zx}$ form $T_2$. A continuum spin-two multiplet thus appears as $E\oplus T_2$. Matching energies in those sectors as the cutoff is removed supports the spin assignment; one accidental degeneracy at one spacing does not establish it.

Likewise $A_1$ can receive contributions from spins beyond zero. A label with fewer possible values cannot uniquely recover all nonnegative integer spins. The continuum interpretation needs operator construction, level patterns and scaling evidence.

How a lattice operator is assembled

Take the trace of an ordered spatial-link product around a contractible closed loop. The color trace makes it gauge invariant. Summing its translates projects to zero spatial momentum. Combining its rotated and parity-reflected copies selects a cubic representation and parity. The real and imaginary parts distinguish charge-conjugation sectors when both exist; for $SU(2)$ the fundamental trace is real, so these loop constructions have no $C=-$ channel.

The source-specific implementation is described in Athenodorou–Teper, arXiv:2106.00364v3, §§5.1 and 5.7, Table 32. This page uses its symmetry framework, without treating a numerical identification as a proof of an isolated continuum pole. Its §5.8 discusses scattering-state issues.

Thresholds and finite boxes

Two particles with opposite momenta can have the same total momentum and symmetry as a single-particle operator. In a noninteracting illustrative box, their energy is

$$E_{\mathrm{pair}}=2\sqrt{m^2+\mathbf p^2}.$$

At $\mathbf p=0$ this is $2m$, if the boundary conditions permit that momentum. Interactions shift finite-volume levels. Spatial winding flux tubes can introduce additional finite-volume states; a volume scan and an appropriate operator basis help diagnose them.

A stable level, a resonance, and the onset of a continuum are different physical objects. The existence of a positive spectral threshold does not require an isolated lowest pole. Read gap and correlation decay before interpreting a table as evidence about the entire spectrum.

Exercise: an ambiguous lattice signal

Suppose an $E^{++}$ level and a $T_2^{++}$ level approach the same mass as $a\to0$. Is spin two established solely by the two reported numbers? What extra information would make the interpretation stronger?

Solution. The five-component pattern is consistent with spin two, but higher-spin multiplets can also contain these sectors, and levels can coincide accidentally. Check the full cubic decomposition, expected partner levels, operator overlaps, volume dependence and continuum trend. Also inspect whether a multiparticle explanation is allowed. Statistical agreement is evidence within that analysis, not a uniqueness theorem.

Continue from labels to measurements

Use spectral extraction to see how an operator basis becomes a correlator matrix. Then read the glueball dossier alongside its source tables, retaining group, scale, lattice action and extrapolation uncertainty.