Wilson Surfaces in Higher Gauge and Holography

• Wilson surfaces are nonlocal observables defined on two-dimensional submanifolds that extend Wilson loops to higher-dimensional gauge theories.
• They appear in various contexts including six-dimensional (2,0) theory, Yang–Mills formulations, and strict higher gauge theory through self-dual fields, surface Dirac operators, and 2-holonomy.
• They underpin holographic dualities and topological field models, linking M2/M5-brane configurations, bubbling geometries, and coadjoint-orbit constructions.

Wilson surfaces are nonlocal observables attached to two-dimensional submanifolds. In six-dimensional $\mathcal N=(2,0)$ theory they are formally built from the self-dual two-form gauge field $B^+$, in ordinary gauge theory they admit fermionic, coadjoint-orbit, equivariant-cohomological, and Poisson-$\sigma$-model realizations, and in strict higher gauge theory they are expressed through $2$-holonomy of a $2$-connection. In holographic settings they are dual to M2/M5 configurations and to half-BPS bubbling geometries locally asymptotic to $AdS_7\times S^4$ (Gentle et al., 2015, Neuberger et al., 2010, Mori et al., 2014, Zucchini, 2022, Zucchini, 2022).

1. Definitions across gauge-theoretic settings

The term “Wilson surface” does not refer to a single universal construction. Rather, it denotes a family of surface-supported observables that generalize Wilson loops from one-dimensional contours to two-dimensional supports.

In the six-dimensional $(2,0)$ theory, the basic gauge datum is a self-dual two-form $B^+$ with field strength $H=dB^+$ and $H=*H$. A natural nonlocal operator is the Wilson surface

$B^+$0

or, equivalently,

$B^+$1

with $B^+$2 or $B^+$3 a two-dimensional surface and $B^+$4 a representation label. In flat space these operators appear as planar defects, preserve $B^+$5 on the world-volume together with an $B^+$6 of transverse rotations and an $B^+$7 R-symmetry, and the preserved superalgebra is $B^+$8 (Gentle et al., 2015, Mori et al., 2014).

In ordinary $B^+$9 Yang–Mills theory, one route from loops to surfaces is furnished by a fermionic world-volume construction. For a smooth two-dimensional surface $\sigma$0 with pulled-back gauge field $\sigma$1, one introduces the surface Dirac operator

$\sigma$2

and the observable

$\sigma$3

This contains the loop case as a squeezed-surface limit in which one linear dimension of $\sigma$4 is taken to zero (Neuberger et al., 2010).

In strict higher gauge theory, the relevant datum is a crossed module together with a $\sigma$5-form and $\sigma$6-form gauge potential. Here a Wilson surface is defined from the $\sigma$7-valued surface holonomy $\sigma$8 or $\sigma$9 of the $2$0-connection, and a scalar observable is extracted through an invariant “$2$1-trace” or trace function:

$2$2

Because the $2$3-holonomy is generally only covariant under crossed-module conjugation, the trace is essential for gauge-invariant scalar observables (Zucchini, 2022, Zucchini, 2019).

Context	Basic gauge data	Representative expression
6d $2$4 theory	$2$5, $2$6, $2$7	$2$8
Ordinary Yang–Mills	Pullback $2$9 on $2$0	$2$1 from the $2$2d Dirac operator $2$3
Strict higher gauge theory	Crossed module, $2$4 or $2$5	$2$6

This multiplicity of definitions is not a contradiction. It reflects the fact that higher-dimensional gauge observables may be modeled either as direct couplings to a two-form field, as auxiliary surface theories coupled to an ordinary connection, or as genuine $2$7-holonomy in higher gauge theory.

2. Coadjoint-orbit, equivariant, Poisson, and superspace formulations

A major field-theoretic approach realizes Wilson surfaces through coadjoint orbits and equivariant cohomology. For a compact Lie group $2$8, an integral coadjoint orbit $2$9 with Kirillov–Kostant–Souriau form $AdS_7\times S^4$0 admits a closed equivariant extension

$AdS_7\times S^4$1

Using an auxiliary $AdS_7\times S^4$2-valued field $AdS_7\times S^4$3 and $AdS_7\times S^4$4, the pulled-back equivariant form on a two-manifold $AdS_7\times S^4$5 is

$AdS_7\times S^4$6

and the Wilson-surface operator is

$AdS_7\times S^4$7

This formulation allows closed surfaces as well as surfaces with boundary, and the integrality of $AdS_7\times S^4$8 is the condition that $AdS_7\times S^4$9 be an honest representation rather than a projective one (Alekseev et al., 2015).

The same observable can be recast as a Poisson $(2,0)$0-model. On a coadjoint orbit, the Kirillov Poisson structure is nondegenerate, and integrating out the auxiliary one-form of the Poisson $(2,0)$1-model reproduces the Kirillov form. A BF-type action,

$(2,0)$2

likewise produces the Diakonov–Petrov two-form and hence the Wilson-surface action. In the purely even setting this is a reformulation of the ordinary surface observable; in the super extension one replaces the target by the dual of a Lie superalgebra and the trace by the supertrace, obtaining

$(2,0)$3

The explicit “proof of concept” examples show that odd orbit parameters can induce nontrivial $(2,0)$4-graded phases in the partition function (Chekeres et al., 2024).

A different geometric formulation uses superspace and integral forms. On an $(2,0)$5-dimensional supermanifold one introduces a super-$(2,0)$6-form $(2,0)$7 and a Picture Changing Operator $(2,0)$8 localizing on a $(2,0)$9 supersurface. The super-Wilson surface is then

$B^+$0

This makes the dependence on the embedding of $B^+$1 explicit at the level of forms on the full supermanifold and is the setting in which BPS and $B^+$2-symmetry conditions for six-dimensional surface operators are derived (Cremonini et al., 2020).

These formulations share a common pattern: the surface observable is not always inserted directly as a “surface-ordered exponential” of a two-form. Often it is realized as the partition function of an auxiliary two-dimensional topological theory whose target data encode representation-theoretic information.

3. Strict higher gauge theory and higher coadjoint-orbit theory

In strict higher gauge theory, Wilson surfaces are the natural observables associated with $B^+$3-connections. A strict $B^+$4-connection on a manifold $B^+$5 is a pair

$B^+$6

subject to the fake-flatness condition

$B^+$7

with remaining curvature

$B^+$8

The $B^+$9-valued $H=dB^+$0-holonomy of a parametrized surface $H=dB^+$1 is constructed by solving a path-surface ordered system, and for fully flat $H=dB^+$2 it depends only on the $H=dB^+$3-homotopy class of $H=dB^+$4 (Zucchini, 2019).

Gauge covariance has a specifically higher form. Under a $H=dB^+$5-gauge transformation $H=dB^+$6, the transformed $H=dB^+$7-holonomy is conjugated by the $H=dB^+$8-action and by gauge parallel transport along boundary curves. For closed surface knots the boundary contributions cancel, and the holonomy transforms simply by the action of $H=dB^+$9 at the basepoint. A trace function invariant under the crossed-module conjugation rules then produces a scalar Wilson surface invariant under base change, $H=*H$0-gauge transformations, and ambient isotopy (Zucchini, 2019).

A parallel line of development formulates the representation theory behind these observables in terms of derived geometry. For a crossed module $H=*H$1 one introduces the derived Lie group

$H=*H$2

and the derived coadjoint orbit

$H=*H$3

On this orbit there is a degree-$H=*H$4 derived KKS presymplectic form $H=*H$5, a derived prequantum line bundle $H=*H$6, and a derived Bohr–Sommerfeld condition

$H=*H$7

The associated two-dimensional Topological Coadjoint Orbit model has action

$H=*H$8

and its partition function furnishes a functional-integral realization of the Wilson surface (Zucchini, 2022, Zucchini, 2022).

The vanishing fake-curvature condition is structurally central in this framework. The field equations imply, and the perturbative expansion likewise requires, that the pullback $H=*H$9-connection be fake-flat:

$B^+$00

In the fake-flat sector the partition function is homotopy invariant under motions of the surface insertion. This makes explicit a point sometimes obscured in heuristic discussions: strict higher Wilson surfaces are controlled not merely by a two-form field, but by the compatibility of one-form and two-form gauge data (Zucchini, 2022).

A persistent open issue is full quantization. The derived orbit has positive degree, there is no standard notion of derived polarization, and the bracket on Hamiltonian derived functions is only a twisted Lie bracket. The existing constructions therefore provide the geometric underpinning and partition-function realization of Wilson surfaces, but not yet a complete higher analogue of ordinary geometric quantization (Zucchini, 2022).

4. Wilson surfaces in six-dimensional $B^+$01 theory

In six dimensions, Wilson surfaces are among the canonical nonlocal observables of the $B^+$02 superconformal theory. The operator is associated with the self-dual two-form $B^+$03 and, in flat space, a planar Wilson surface filling two directions preserves the conformal subgroup $B^+$04 acting on the world-volume, an $B^+$05 rotating the four transverse directions, and an $B^+$06 R-symmetry (Gentle et al., 2015).

A standard route to calculable observables is compactification on $B^+$07. The conjecture used in the AdS$B^+$08/CFT$B^+$09 literature is that the $B^+$10 theory compactified on a circle of radius $B^+$11 is equivalent to $B^+$12d maximal super Yang–Mills, with

$B^+$13

A Wilson surface wrapping the M-theory circle and a great circle of $B^+$14 then descends to a $B^+$15-BPS Wilson loop in $B^+$16d MSYM, and one introduces

$B^+$17

for $B^+$18 radius $B^+$19. Localization yields a Chern–Simons matrix model in which large-$B^+$20 expectation values can be extracted for symmetric and antisymmetric representations (Mori et al., 2014).

The M-theory engineering of these operators is by semi-infinite M2-branes ending on M5-branes along the surface $B^+$21. In this description, $B^+$22 coincident M2-branes ending on $B^+$23 define the Wilson surface in the rank-$B^+$24 symmetric representation, whereas $B^+$25 separate M2-branes each ending once on $B^+$26 define the rank-$B^+$27 antisymmetric representation. On the tensor branch, the winding data among the $B^+$28 M5-branes fixes the Young diagram $B^+$29 (Agarwal et al., 2018).

Two independent nonperturbative computations have been compared for Wilson surfaces on the $B^+$30 inside $B^+$31-deformed $B^+$32. One uses $B^+$33d $B^+$34 instanton partition functions with Wilson lines; the other uses elliptic genera of $B^+$35d $B^+$36 gauge theories describing self-dual strings in the presence of Wilson surface defects. For minuscule representations the two expansions agree exactly. For non-minuscule representations only partial agreement is found, and the discrepancy is attributed to extra poles introducing states charged under the $B^+$37 flavor that are not part of the $B^+$38d SCFT (Agarwal et al., 2018).

The superspace construction sharpens the BPS analysis. A generalized Wilson surface may couple not only to $B^+$39 but also to the five scalars:

$B^+$40

For a purely bosonic embedding the BPS condition reduces to the projector equation

$B^+$41

which admits nontrivial constant solutions precisely when

$B^+$42

The explicit $B^+$43-BPS examples include planar surfaces

$B^+$44

and spherical wavefronts

$B^+$45

No real solutions arise for purely spacelike embeddings in Minkowski signature (Cremonini et al., 2020).

5. Holographic realizations: M5-branes, probe embeddings, and bubbling geometries

The earliest holographic description of large-representation Wilson surfaces in the $B^+$46 theory is in terms of M5-brane string solitons in $B^+$47. In this construction the M5 worldvolume is $B^+$48, a self-dual three-form flux is turned on, the BPS relation

$B^+$49

follows from the M5 equations of motion, and the solution preserves $B^+$50 of the $B^+$51 bulk supercharges. The magnetic M2 charge carried by the self-dual string is identified with the rank of the symmetric representation, and the treatment of boundary terms is subtle because the divergent “area-law” term is canceled by the $B^+$52 boundary term (0707.3978).

A later probe-brane analysis in global $B^+$53 distinguishes two half-BPS M5 embeddings. The “symmetric” M5 wraps $B^+$54 and an $B^+$55; the “antisymmetric” M5 wraps $B^+$56 and an $B^+$57 at fixed polar angle $B^+$58. For large-$B^+$59 and $B^+$60, the regularized on-shell actions give

$B^+$61

in perfect agreement with the localization computation in $B^+$62d MSYM (Mori et al., 2014).

Beyond the probe approximation, half-BPS bubbling geometries in eleven-dimensional supergravity encode heavy Wilson surfaces. Their metric is an $B^+$63 fibration over a Riemann surface $B^+$64,

$B^+$65

with $B^+$66 in Poincaré form and the planar Wilson surface located at $B^+$67, filling $B^+$68. The solution is determined by a positive harmonic function $B^+$69 and a complex function $B^+$70 obeying

$B^+$71

together with alternating boundary conditions $B^+$72 or $B^+$73 on $B^+$74 (Gentle et al., 2015).

These bubbling solutions permit direct holographic computation of one-point functions and entanglement data. The Wilson-surface contribution to the holographic stress tensor is

$B^+$75

which is traceless and respects the unbroken $B^+$76 isometry. For a spherical entangling surface, the defect contribution to the entanglement entropy is

$B^+$77

and the logarithmic divergence is associated with the two intersection points of the Wilson surface with the entangling four-sphere. The expectation value follows from the regularized on-shell supergravity action,

$B^+$78

All of these observables are expressed in terms of the real parameters $B^+$79, their moments $B^+$80, and the alternating cubic sum $B^+$81 (Gentle et al., 2015).

6. Two-dimensional topological interactions, surface sums, and recent variants

In two dimensions, Wilson surfaces admit especially explicit topological realizations. A Wilson surface labeled by a dominant weight $B^+$82 may be defined by the action

$B^+$83

Canonical quantization on a cylinder shows that the Hilbert space on each boundary circle is one-dimensional. On a closed surface, the Wilson-surface theory defines a topological invariant of the principal $B^+$84-bundle $B^+$85,

$B^+$86

where $B^+$87 corresponds to the bundle class $B^+$88. If $B^+$89 is simply connected, or if $B^+$90 descends to a weight of $B^+$91, then $B^+$92 (Chekeres, 2018).

When coupled to $B^+$93d Yang–Mills, the Wilson-surface factor modifies the partition function by weighting topological sectors:

$B^+$94

For non-simply-connected groups this produces genuinely nontrivial effects. Detailed examples include $B^+$95 and $B^+$96, while the earlier equivariant-cohomological analysis likewise finds that the surface observable is trivial for simply connected $B^+$97 and nontrivial for non-simply-connected $B^+$98, with explicit $B^+$99 and $\sigma$00 examples (Chekeres, 2018, Alekseev et al., 2015).

A distinct but related use of the phrase appears in rigorous two-dimensional Yang–Mills on the plane. There a “Wilson surface” spanning a based loop $\sigma$01 is a combinatorially defined surface $\sigma$02 obtained from a reference covering disk by choosing ramification points, cutting along non-crossing paths, and regluing edges in pairs so that $\sigma$03 covers $\sigma$04 with total degree one. The Wilson loop expectation then admits the convergent sum-over-surfaces formula

$\sigma$05

with

$\sigma$06

This construction concerns a surface expansion of loop observables rather than a surface operator inserted into the theory, but it shows that surfaces can also enter as the summation objects underlying Wilson-loop expectations (Park et al., 2023).

Recent extensions preserve the same formal architecture while changing the target geometry. In the “odd Wilson surface” construction, the ordinary Poisson-$\sigma$07-model picture is supersymmetrized by replacing the target with the dual of a Lie superalgebra and keeping track of Koszul signs, parity shifts, and supertrace. A stated consequence is that odd target-space directions can generate nontrivial graded phases, so the super version is not merely a cosmetic rewriting of the even theory (Chekeres et al., 2024).

Taken together, these developments show that Wilson surfaces sit at the intersection of extended-operator theory, higher gauge theory, topological field theory, and holography. What remains stable across these contexts is the passage from line-supported to surface-supported gauge observables; what changes is the mathematical infrastructure used to define, quantize, and evaluate them.