Static Potentials, Flux Tubes, and Effective Strings

A rectangular Wilson loop is a transfer-matrix correlator for a static source–antisource pair. Taking its Euclidean-time extent to infinity at fixed separation extracts the lowest static energy; only after that projection may one test linear growth, string breaking, or the universal long-string correction. A visible tube at finite time or distance can instead be a metastable excitation.

Required background. Line operators, screening, and symmetry diagnostics fixes the genuine probe and predicts whether dynamical matter can break its string.

Helpful background. Background-field Yang–Mills supplies the gauge-covariant expansion used in effective descriptions.

Shared comparison. The diagnostic map distinguishes a static potential from the other confinement claims, and the claim–evidence table records the relevant failure limits.

Rectangular loops project onto static energies

Work at zero temperature in $d$-dimensional Euclidean spacetime. Let $W_R(R,T_E)$ be a renormalized rectangular loop of spatial width $R$ and Euclidean-time height $T_E$ in a genuine representation $R$. Cutting the rectangle across a time slice prepares a gauge-invariant state containing two static sources. The transfer matrix gives

$$\langle W_R(R,T_E)\rangle =\sum_{n}|c_n(R)|^2e^{-E_n(R)T_E},$$

where $c_n$ is the overlap of that Wilson operator with the $n$th state in the static-source sector. Therefore

$$V_R(R)=E_0(R)-2\delta m_R =-\lim_{T_E\to\infty}\frac{1}{T_E} \log \langle W_R(R,T_E)\rangle_{\mathrm{ren}}.$$

The subtraction $2\delta m_R$ removes the scheme-dependent static self-energies, equivalently the temporal part of the loop’s perimeter divergence. Additive constants in $V_R$ are not physical; forces, level splittings, and the coefficient of asymptotic growth are.

If a large loop obeys

$$-\log\langle W(R,T_E)\rangle =\sigma RT_E+\mu(R+T_E)+\cdots,$$

then taking $T_E\to\infty$ yields $V(R)=\sigma R+\mu+\cdots$. This is the precise bridge from an area law to a linear static energy, already implicit in Wilson’s strong-coupling loop analysis Wilson 1974, §§IV–V, pp. 2450–2456. The converse requires the same loop family and compatible limits; it cannot be inferred from a finite interval that merely looks straight.

The long flux tube as an effective string

Suppose the chosen probe cannot be screened, the ground state contains a stable long tube, and

$$R\sqrt{\sigma}\gg1.$$

At wavelengths much longer than the tube thickness, the only mandatory gapless world-sheet fields are its $d-2$ transverse displacements $X^i(t,z)$. To quadratic order in static gauge,

$$S_{\mathrm{ws}} =\sigma RT_E +\frac{\sigma}{2}\sum_{i=1}^{d-2} \int\mathrm d t\,\mathrm d z\, \partial_\alpha X^i\partial_\alpha X^i+\cdots,$$

with Dirichlet endpoints at the static sources. Each transverse scalar has modes $\omega_n=\pi n/R$. Its regularized zero-point energy is

$$\frac12\sum_{n=1}^{\infty}\omega_n =\frac{\pi}{2R}\,\zeta(-1) =-\frac{\pi}{24R}.$$

Summing over transverse directions gives the universal open-string asymptotics

$$V(R)=\sigma R+\mu -\frac{\pi(d-2)}{24R} +O(R^{-3}).$$

The negative $1/R$ coefficient is the Lüscher term. It depends only on $d-2$ massless transverse modes and the open-string boundary conditions, not on the microscopic flux profile Lüscher, Symanzik, and Weisz 1980, pp. 365–396. Extra massless world-sheet fields or different boundaries change the coefficient; short strings need not follow the expansion.

The same effective theory predicts excitation gaps at leading order,

$$E_n(R)-E_0(R)=\frac{\pi n}{R}+O(R^{-3}),$$

organized by transverse oscillator quantum numbers. Agreement of both the ground-state correction and the excitation pattern tests a string description more sharply than observing approximate linearity alone.

String breaking changes the ground-state branch

With dynamical matter capable of screening the probe, compare a flux-tube state and a two-hadron state,

$$E_{S}(R)\simeq2\delta m+\sigma R, \qquad E_{M}(R)\simeq2M_{Qq}.$$

In the two-state approximation,

$$H(R)= \begin{pmatrix} E_S(R)&x(R)\\ x(R)&E_M(R) \end{pmatrix},$$

so the eigenvalues repel. The lower eigenvalue saturates rather than remaining linear. Yet a thin Wilson loop often has $|c_S|\gg|c_M|$; at practical $T_E$ it can track the upper, string-like eigenstate past the avoided crossing. Mixed correlators containing both string and two-hadron operators reveal the true ground state, as demonstrated in unquenched lattice QCD by Bali et al. 2005, §§II–IV.

Thus the order of limits is part of the claim:

$$\boxed{\text{project }T_E\to\infty\text{ at fixed }R \quad\text{before testing }R\to\infty.}$$

When the string breaks, the long-string expansion remains meaningful for a metastable level only over distances where that resonance is narrow and the world-sheet scale separation survives. It is no longer an asymptotic order parameter.

Flux profiles and discriminating measurements

A gauge-invariant flux profile can be defined by correlating the Wilson loop with a local field-strength composite, subtracting its vacuum expectation value, and taking the same large-$T_E$ projection. Useful observables include the transverse energy-density profile, its width, and excitation energies. They answer different questions:

• $V'(R)$ measures the force;
• the profile asks whether energy is spatially collimated;
• the spectrum tests the world-sheet degrees of freedom;
• mixed-operator overlaps locate string breaking.

No single one of these measurements identifies a microscopic mechanism. A dual-superconductor model, a center-vortex description, and a controlled monopole plasma can share a linear potential while predicting different field profiles or defect responses.

Exercises

1. Excited-state contamination. Suppose $W(T)=a_0e^{-E_0T}+a_1e^{-E_1T}$ with $E_1>E_0$. Show that the effective energy $E_{\mathrm{eff}}(T)=-\partial_T\log W(T)$ approaches $E_0$ from above.

Solution

Writing $r=(a_1/a_0)e^{-(E_1-E_0)T}$ gives $E_{\mathrm{eff}}=(E_0+rE_1)/(1+r)=E_0+(E_1-E_0)r/(1+r)$. For positive spectral weights, the correction is positive and vanishes exponentially. Poor ground-state overlap means $r$ can remain large for a long Euclidean time.

2. Lüscher coefficient. Evaluate the universal term in $d=3$ and $d=4$ spacetime dimensions.

Solution

There are respectively one and two transverse fields. Hence the terms are $-\pi/(24R)$ in $d=3$ and $-\pi/(12R)$ in $d=4$.

References

• Bali, Gunnar S., Hartmut Neff, Thomas Düsel, Thomas Lippert, and Klaus Schilling. “Observation of String Breaking in QCD.” Physical Review D 71 (2005): 114513. DOI. Open PDF.
• Lüscher, Martin, Kurt Symanzik, and Peter Weisz. “Anomalies of the Free Loop Wave Equation in the WKB Approximation.” Nuclear Physics B 173 (1980): 365–396. DOI.
• Wilson, Kenneth G. “Confinement of Quarks.” Physical Review D 10 (1974): 2445–2459. DOI.