Compactification and controlled semiclassical models

A compact direction can turn a strongly coupled infrared question into a calculation with separated scales. The calculation is useful only while the geometry, deformation, and hierarchy remain visible.

Prerequisites: dimensional transmutation and confinement and screening.
Outcomes: identify the hierarchy that permits a semiclassical treatment; distinguish a compactified mechanism from a proof for ordinary four-dimensional space.

The circle introduces a gauge-invariant variable

Take pure $SU(N)$ Yang–Mills on $\mathbb R^3\times S^1$, where the circle has circumference $L$. The holonomy is

$$\Omega(\mathbf x)=\mathcal P\exp\left(i\int_0^L A_4(\mathbf x,x_4)\,dx_4\right).$$

Its conjugacy class records parallel transport around the circle. A periodic gauge transformation conjugates $\Omega$, leaving its eigenvalues and traces invariant. A center transformation multiplies $\Omega$ by $z\in\mathbb Z_N$.

If the holonomy eigenvalues are approximately equally spaced around the unit circle, the differences of their phases shift the allowed momenta of off-diagonal gauge components. In a center-symmetric configuration the smallest corresponding scale is

$$m_W=\frac{2\pi}{NL}.$$

It is a scale of the compactified description, not a glueball mass in the uncompactified target.

Specify the controlled example

Ünsal and Yaffe’s arXiv:0803.0344v2, §2.1 and §3 add double-trace functions of $\Omega$ to stabilize center symmetry. Their deformed pure-gauge theory has analytic semiclassical control at suitable finite $N$ when $NL\Lambda\ll1$. Equation (28) defines $m_W$; §3.1 describes the nonperturbative dual-photon mass mechanism; §3.2 studies string tensions. Increasing $L$ or $N$ removes the hierarchy supporting that calculation. Their large-$N$ equivalence statements have their own observable and symmetry hypotheses, and do not establish every finite-$N$ uncompactified theory.

The source version is 22 March 2008. The cited sections and equations were inspected on 24 October 2026; this is a check of the stated model and mechanism, not an independent verification of its entire argument.

Read a scale hierarchy numerically

Let the dimensionless combination $\Lambda L$ be $0.01$. For $N=3$,

$$NL\Lambda=0.03,\qquad m_W/\Lambda=\frac{2\pi}{0.03}\approx209.$$

The charged scale is far above $\Lambda$. Holding $L\Lambda$ fixed and increasing to $N=300$ gives $NL\Lambda=3$ and $m_W/\Lambda\approx2.09$. The same “small circle” in units of $\Lambda^{-1}$ no longer gives the original large separation.

These are illustrative parameter choices, not lattice measurements or source benchmarks. They demonstrate why the order of the small-circle and large-$N$ limits matters. A phrase such as “take the volume small” leaves out the rank dependence.

Why a nonperturbative mass can be calculable

At scales below the charged modes, one can describe suitable light degrees of freedom by periodic scalar variables obtained by dualizing Abelian fields in the effective three-dimensional description. Nonperturbative events can produce a periodic potential.

To see the elementary mechanism, consider an illustrative one-field effective Lagrangian,

$$\mathcal L_{\rm eff} =\frac K2(\partial\sigma)^2+\zeta(1-\cos\sigma), \qquad K>0,\quad\zeta>0.$$

The variable $\sigma$ is dimensionless. In three Euclidean dimensions $[K]=\text{mass}$ and $[\zeta]=\text{mass}^3$. Expanding around $\sigma=0$ and setting $\varphi=\sqrt K\,\sigma$ gives

$$\mathcal L_{\rm eff} =\frac12(\partial\varphi)^2+ \frac12\frac{\zeta}{K}\varphi^2+\cdots, \qquad m_{\rm eff}^2=\frac{\zeta}{K}.$$

This small-oscillation calculation explains how a potential generates a mass in an effective model. It is deliberately schematic: the actual $SU(N)$ theory has a multi-component potential, root data, and an approximation scheme. Replacing those structures with this single cosine would not establish the source’s quantitative spectrum.

Continuity is a question to answer

To return to undeformed Yang–Mills on uncompactified space one must control increasing $L$, removing the deformation, identifying the appropriate observables, and preserving the relevant physical properties. A lack of an obvious order parameter separating regimes does not prove a controlled continuation. A finite approximation can cease to be accurate without announcing its failure through a symmetry-breaking transition.

Likewise, a spatial compactification with periodic bosonic fields is not automatically a thermal ensemble interpretation. State which direction and boundary conditions are being used before translating holonomy into a temperature statement.

Exercise: a limit that breaks control

Choose $L\Lambda=1/N$. Does taking $N\to\infty$ make the hierarchy $NL\Lambda\ll1$ progressively stronger?

Solution. No: $NL\Lambda=1$ along the entire sequence. The lightest charged scale stays $m_W=2\pi\Lambda$. A shrinking geometric circumference alone does not establish the assumed hierarchy. To make that parameter tend to zero, one would need $L\Lambda=o(1/N)$, along with control of every other approximation used.

Continue to semiclassical methods and the semiclassical dossier. For the broader comparison, return to controlled models.