Planar Gauge Dynamics and String-Like Organization

At fixed ’t Hooft coupling, adjoint gauge diagrams carry the same Euler-characteristic weights as matrix ribbon graphs. Wilson loops create boundaries, handles cost $N^{-2}$, and normalized invariant correlators factorize. These properties are string-like, but they are kinematic organization. A gauge/string duality additionally requires nonperturbative dynamics, a worldsheet definition, a spectrum and observable dictionary, and consistency beyond the formal genus count.

Required background. Double-line counting and the topological expansion supplies the Euler-characteristic derivation. Static potentials, flux tubes, and effective strings supplies the observable distinction between an area law, a physical flux tube, and its long-string effective theory.

Helpful background. Large-N factorization and master-field claims supplies the state and normalization hypotheses behind planar closure.

Planar gauge-theory normalization and observables

Use $SU(N)$ generators normalized by

$$\operatorname{tr}(T^aT^b)=\frac12\delta^{ab}$$

and put the coupling outside

$$S_{\mathrm{YM}} = \frac{N}{2\lambda_{\mathrm t}(\mu)} \int\mathrm d^dx\, \operatorname{tr}F_{\mu\nu}F_{\mu\nu}.$$

The renormalized ’t Hooft coupling $\lambda_{\mathrm t}=g_{\mathrm{YM}}^2N$ is fixed at scale $\mu$ while $N\to\infty$. The regulator, state, spacetime volume, and matter scaling are fixed before counting. For adjoint fields, a connected vacuum graph of genus $h$ is $O(N^{2-2h})$.

A normalized Wilson loop in the fundamental representation is

$$W(C) = \frac1N \operatorname{tr}\, \mathcal P \exp\!\left(i\oint_C A_\mu\,\mathrm dx^\mu\right).$$

An unnormalized trace creates one boundary and changes a planar sphere factor $N^2$ to a disk factor $N$. Its explicit $1/N$ normalization then makes $\langle W(C)\rangle=O(1)$. For $r$ normalized loops,

$$\left\langle W(C_1)\cdots W(C_r) \right\rangle_{\mathrm c,h} = O\!\left(N^{2-2h-2r}\right).$$

The same formula applies to normalized local single traces, subject to operator renormalization and contact terms.

First application: planar Wilson-loop organization

Expand a Wilson loop and the gauge action in powers of the renormalized coupling. Every double-line graph ending on $C$ thickens to a surface with that loop as a boundary. At leading order in $N$, only genus-zero surfaces with one boundary contribute to $\langle W(C)\rangle$; a handle is suppressed by $N^{-2}$.

For two normalized loops,

$$\langle W(C_1)W(C_2)\rangle = \langle W(C_1)\rangle\langle W(C_2)\rangle +O(N^{-2}),$$

provided the state is clustering and the separation, volume, and infrared limits are uniform. The connected cylindrical topology is $O(N^{-2})$ after both normalizations.

This mapping between handles, boundaries, and powers of $N$ is the basis of the planar expansion ’t Hooft 1974, §§2–4, pp. 466–472. It says which color contractions dominate. It does not calculate their sum at strong coupling.

Shared calculation. The large-N counting and topology map displays the graph-to-surface count and its normalization counterexample. The large-N scaling comparison fixes the gauge and Wilson-loop conventions used here.

Confinement adds dynamical string data

In a confining phase, a sufficiently large renormalized rectangular loop of spatial width $R$ and Euclidean time $T$ can obey

$$\langle W(R,T)\rangle \sim \exp\!\left[ -\sigma RT-\mu_{\mathrm p}(2R+2T)+\cdots \right].$$

The area coefficient $\sigma$ is the string tension; perimeter and cusp terms require regularization and renormalization. Wilson’s lattice strong-coupling expansion gives a concrete setting in which an area law appears Wilson 1974, §§VI–VII, pp. 2452–2457. Large $N$ alone does not imply $\sigma>0$.

If the theory has a stable, long, thin flux tube with no additional gapless worldsheet fields, its transverse Goldstone modes give the long-distance static potential

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

This is an infrared effective-string statement, valid for $R$ large compared with the flux-tube thickness and other inverse gaps. The systematic long-string hypotheses and universality orders are described in Aharony and Komargodski 2013, §§2–4.

The two expansions answer different questions:

$$\text{large }N: \quad \sum_h N^{2-2h}(\cdots), \qquad \text{long string}: \quad \sum_k\frac{c_k}{(\sigma R^2)^k}.$$

Neither parameter guarantees control of the other. One may have planar dominance without confinement, or a useful effective string at finite $N$.

What is missing from a gauge/string duality claim

The following implications are not supplied by $N^{2-2h-b}$:

• Graph surfaces do not define a continuum measure. One needs controlled weights for arbitrarily refined worldsheets and their moduli.
• A flux-tube effective theory is infrared and sector-specific. It need not reproduce glueballs, local operators, finite-temperature physics, or short strings.
• Factorization does not identify a bulk configuration. It constrains invariant correlators in a selected state.
• Topology does not establish target-space locality. A complete dictionary must explain spectra, interactions, symmetries, and causality.
• Perturbative genus counting misses $e^{-cN}$ sectors. Brane-like saddles, eigenvalue tunneling, or level splittings can be invisible at all fixed genera.

A demonstrated holographic dual may supply these structures in a specific theory. The combinatorial expansion is necessary evidence for string-like organization in matrix-like theories, not sufficient evidence for such a dual.

Common pitfalls

Inferring confinement from planarity. Planarity organizes color factors. The area law and nonzero string tension require dynamics.

Calling the long-string EFT a complete dual. Its predictions apply to a long flux-tube sector and are organized in derivatives; they do not define every gauge-theory observable.

Mixing the $1/N$ and long-distance expansions. State both $N$ and $\sigma R^2$, and check that corrections in each remain small in the order of limits used.

Exercises

1. Determine the $N$ scaling of a connected genus-one correlator of three normalized Wilson loops.

Solution

Set $h=1$ and $r=3$ in

$$N^{2-2h-2r}.$$

The result is $N^{2-2-6}=N^{-6}$.

2. Explain why an observed Lüscher term does not prove a full gauge/string duality.

Solution

The $1/R$ term follows from the gapless transverse modes of a long flux tube under stated infrared assumptions. It tests the effective theory of that sector, not the completeness of a worldsheet formulation or a dictionary for all gauge-theory observables.

3. With fixed $N_f$, compare an adjoint planar vacuum graph and the same topology with one fundamental loop.

Solution

The adjoint sphere is $O(N^2)$. A fundamental loop creates a boundary and contributes a flavor sum, giving $O(NN_f)$. Their ratio is $N_f/N$, so it is suppressed at fixed $N_f$.

References

• Aharony, O., and Komargodski, Z. (2013). “The Effective Theory of Long Strings.” Journal of High Energy Physics 2013(05), 118. doi:10.1007/JHEP05(2013)118. Open PDF.
• ’t Hooft, G. (1974). “A Planar Diagram Theory for Strong Interactions.” Nuclear Physics B 72, 461–473. doi:10.1016/0550-3213(74)90154-0.
• Wilson, K. G. (1974). “Confinement of Quarks.” Physical Review D 10, 2445–2459. doi:10.1103/PhysRevD.10.2445.