Off-constraint structure of Wilson-loop minimal surfaces in AdS/CFT: determining $G(z)$ and the prefactor $\Delta$ to test the loop equation for cusped and self-intersecting loops

The AdS/CFT correspondence predicts that the Maldacena-Wilson loop of planar $\mathcal{N}=4$ super Yang-Mills theory at strong 't Hooft coupling $\lambda$ is computed by a string worldsheet in $\mathrm{AdS}_5\times S^5$ ending on the boundary curve, $\langle W\rangle=\Delta\exp(-\sqrt{\lambda}\,\tilde A)$, where $\tilde A$ is the Legendre transform of the regularized minimal-surface area and $\Delta$ is a multiplicative prefactor set by string fluctuations. Drukker, Gross and Ooguri (1999) showed that smooth loops satisfying the BPS constraint $\dot x^2=\dot y^2$ solve the Makeenko-Migdal loop equation, but the quantitative test for loops off the constraint — cusped and self-intersecting loops — could not be completed: the coefficient of the residual linear divergence, a function $G(z)$ of $z=\dot y^2/\dot x^2$, is unknown away from $z=1$, and the prefactor $\Delta$, possibly depending on loop topology, is unknown. The open problem is to determine this off-constraint structure of the string functional and thereby verify or refute at strong coupling the loop-equation contact terms for cusps, which require $\mathcal{G}(f(\Omega/2))=(\pi-\Omega)/(8\sin\Omega)$ (with $f(\Omega/2)$ a known function of the cusp geometry fixed by the minimal-surface solution and $\mathcal{G}$ the cusp function related to but distinct from the linear-divergence coefficient $G(z)$), and for self-intersections, which require the ratio $\Delta_1\Delta_2/\Delta_{1+2}$. Later literature has solved adjacent slices exactly (the near-BPS linear-divergence coefficient, the generalized cusp anomalous dimension, and one-loop determinants for supersymmetric loops), but no determination of $G(z)$ for general $z$, and no strong-coupling verification of the cusp or self-intersection contact terms, is known; the problem remains open.

Background

The Wilson loop is the trace of the path-ordered exponential of a gauge connection around a closed curve $C$; in non-Abelian gauge theory it encodes confinement and the heavy-quark potential. In four-dimensional $\mathcal{N}=4$ supersymmetric Yang-Mills (SYM) theory the natural loop operator couples not only to the gauge field but also to the six adjoint scalar fields through couplings $y^i(s)$ with $(y,y)\le 1$, so the loop is a curve $(x^\mu(s),y^i(s))$ in $\mathbb{R}^4\times S^5$; this is the Maldacena-Wilson loop. The theory is controlled by the 't Hooft coupling $\lambda=g_{\mathrm{YM}}^2N$, with $N$ the number of colors.

The AdS/CFT correspondence identifies large-$N$ $\mathcal{N}=4$ SYM at strong coupling with type IIB string theory on $\mathrm{AdS}_5\times S^5$ (five-dimensional anti-de Sitter space times a five-sphere), the conformal boundary of $\mathrm{AdS}_5$ carrying the gauge theory. Under this dictionary the Wilson loop becomes an open string whose worldsheet ends on the curve drawn at the boundary; as $\lambda\to\infty$ the string path integral is dominated by the classical minimal surface (surface of least area) spanning the curve, giving $\langle W\rangle=\Delta\exp(-\sqrt{\lambda}\,\tilde A)$, where $\tilde A$ is a regularized area functional and $\Delta$ a prefactor generated by quantum fluctuations of the string.

Several constructions of the source paper are needed to state the problem. First, boundary conditions: the $S^5$ coordinates of the string endpoint are fixed by the scalar couplings $y^i(s)$ while their conjugate momenta obey Neumann boundary conditions, and consistency of the mixed Dirichlet-Neumann problem selects loops satisfying the constraint $\dot x^2=\dot y^2$ (a point of the curve uses half of its allowed $S^5$ coupling); such loops preserve half of the supersymmetry and are called BPS loops. For them the naive area must be replaced by its Legendre transform with respect to the $S^5$ coordinates — the area plus a boundary term that fixes the conjugate momenta instead of the coordinates — and this transform exactly removes the linear $1/\epsilon$ ultraviolet divergence of the regularized area, so BPS loops are UV finite as supersymmetry requires. Second, off the constraint: for loops violating $\dot x^2=\dot y^2$, the Legendre-transformed functional $\tilde A(x,y)$ retains a linear divergence of the form $\tilde A\approx(1/\pi\epsilon)\int ds\,|\dot x|\,G(z)$ with $z=\dot y^2/\dot x^2$ and $G$ an unknown function, defined by eq. (5.8) of the source and fixed near $z=1$ by its near-BPS expansion $\tilde A\approx(1/\pi\epsilon)\int ds\,(|\dot x|-|\dot y|)$ (eq. (5.2) of the source). Third, the loop equation: the Makeenko-Migdal loop equation is a functional differential equation in the space of loops obtained by inserting the gauge-field propagator along the contour; its right-hand side is a contact term concentrated where the loop nearly self-intersects. Drukker, Gross and Ooguri formulated it for the supersymmetric Maldena-Wilson loop and evaluated the contact terms: a cusp — a corner where the tangent jumps by an opening angle $\Omega$ and the $S^5$ couplings jump by an angle $\Theta$ — contributes at order $\lambda/\epsilon^2$ a term proportional to $(\pi-\Omega)(\cos\Omega+\cos\Theta)/\sin\Omega$ (their eq. (4.14)), and a self-intersection contributes the same structure without the $(\pi-\Omega)$ factor (their eq. (4.13)). At one loop in weakly coupled gauge theory the same angular factor multiplies the logarithmic divergence of the cusped loop (their eq. (2.12)). Fourth, zig-zag symmetry — invariance of the loop under retraced (backtracking) segments, expected when the loop does not couple to the scalars — turned out to depend on whether the string boundary condition is imposed at the AdS boundary $Y=0$ or on a cutoff hypersurface $Y=\epsilon$, a regularization subtlety the source left open. Fifth, quantum corrections: the prefactor $\Delta$ is set by the determinant of quadratic string fluctuations around the minimal surface with the Neumann boundary conditions and boundary terms of the Legendre-transformed action; the same determinants control subleading $1/\sqrt{\lambda}$ terms and Lüscher-type terms — the universal correction to a static potential from massless worldsheet fluctuations — in confining generalizations. Finally, in other realizations of the duality the analogous observables are Wilson surfaces, two-dimensional surface operators in the six-dimensional superconformal theory on M5-branes (the five-dimensional membranes of M-theory), computed by the area of an M2-brane (a two-dimensional membrane) ending on the boundary surface in $\mathrm{AdS}_7\times S^4$; these surfaces have both a quadratic and a logarithmic ultraviolet divergence, the latter identifiable with the conformal anomaly of submanifold observables.

By the time of the source paper, the minimal surfaces for the line, the circle and antiparallel lines were known, the Legendre transform had been shown to render BPS loops finite, and the zig-zag property had been shown to emerge from the minimal surface for loops that do not couple the scalars. What was missing, and what the loop-equation test exposed, is everything about the functional away from the constraint: the linear-divergence coefficient $G(z)$, and the normalization $\Delta$.

Problem Statement

Determine the off-constraint structure of the string-theory functional for the Maldacena-Wilson loop in $\mathrm{AdS}_5\times S^5$, and use it to complete the strong-coupling test of the Makeenko-Migdal loop equation that Drukker, Gross and Ooguri left unfinished. The common objective is to make the string functional $\langle W\rangle=\Delta\exp(-\sqrt{\lambda}\,\tilde A(x,y))$ a controlled, normalized object for loops that violate the BPS constraint $\dot x^2=\dot y^2$, and thereby to test the loop equation away from the protected sector. The required results, which jointly serve this objective, are:

(1) Classical, off-constraint: determine the coefficient of the linear divergence of the regularized, Legendre-transformed minimal-surface functional for general boundary data $(x(s),y(s))$ — in the parametrization of the source paper, the function $G(z)$ with $z=\dot y^2/\dot x^2$ (eq. (5.8)), or a demonstration that this coefficient is not a local function of $z$, together with its actual dependence on the boundary curve. The result must reproduce the BPS cancellation at $z=1$, be consistent with the known near-BPS (Bremsstrahlung) limit at strong coupling, and settle the behavior as $z\to0$ (vanishing scalar velocity, the regime of self-intersections), including whether $G$ is singular there.

(2) Quantum, normalization: determine the multiplicative prefactor $\Delta$ — the fluctuation determinant of the string around the minimal surface, with the Neumann boundary conditions and boundary terms of the Legendre-transformed action — including its dependence on the topology of the loop, in particular the ratio $\Delta_1\Delta_2/\Delta_{1+2}$ for a self-intersecting loop $W_{1+2}$ decomposed into two loops $W_1$ and $W_2$.

(3) The test: with (1) and (2), verify or refute at leading order in $1/\sqrt{\lambda}$ the loop-equation contact terms for non-BPS loops: the cusp term, which requires $\mathcal{G}(f(\Omega/2))=(\pi-\Omega)/(8\sin\Omega)$ (eq. (5.10)), where $\mathcal{G}$ is the cusp function related to but distinct from the linear-divergence coefficient $G(z)$ of eq. (5.8); here $f(\phi)$ is the known function in the scale-invariant ansatz $Y=r/f(\phi)$ for the minimal surface spanning a cusp of opening angle $\Omega$ ($r,\phi$ radial coordinates near the cusp), whose minimum $f(\Omega/2)$ sets the boundary data at the cusp via $\dot x^2=(1+f(\Omega/2)^2)\dot y^2$ — so that the AdS computation reproduces the full angular prefactor $(\pi-\Omega)(\cos\Omega+\cos\Theta)/\sin\Omega$ of the contact term at order $\lambda/\epsilon^2$; and the self-intersection term, which requires the $\Delta$ ratio above and the factor $(\cos\Omega+\cos\Theta)/\sin\Omega$.

(4) Boundary prescription, supporting: fix the canonical regularization of the string boundary data (imposing the boundary condition at $Y=0$ versus at the cutoff $Y=\epsilon$) that gives a consistent physical interpretation of zig-zag invariance, resolving the regularization dependence the source paper identified.

(5) Secondary, same boundary-functional question in other dualities: state the general criterion for which ultraviolet divergences of the classical brane action are removed by the Legendre transform; in particular, whether the logarithmic divergence of the M5-brane surface observables in $\mathrm{AdS}_7\times S^4$ is removed by it or is the physical conformal anomaly.

The completion boundary is the quantitative strong-coupling evaluation in (3): the problem is complete when the cusp and self-intersection contact terms of the loop equation for the Maldacena-Wilson loop are reproduced — or shown to fail — by the $\mathrm{AdS}_5\times S^5$ string computation, with $G$ and $\Delta$ under control; items (4) and (5) are boundary-prescription questions that the same analysis is expected to settle along the way.

References

1. N. Drukker, D. J. Gross, H. Ooguri, "Wilson loops and minimal surfaces", Phys. Rev. D 60 (1999) 125006, arXiv:hep-th/9904191, doi:10.1103/PhysRevD.60.125006, https://arxiv.org/abs/hep-th/9904191
2. N. Drukker, D. J. Gross, A. A. Tseytlin, "Green-Schwarz string in AdS5 x S5: Semiclassical partition function", JHEP 0004 (2000) 021, arXiv:hep-th/0001204, doi:10.1088/1126-6708/2000/04/021, https://arxiv.org/abs/hep-th/0001204
3. S. Förste, D. Ghoshal, S. Theisen, "Stringy corrections to the Wilson loop in N=4 super Yang-Mills theory", JHEP 9908 (1999) 013, arXiv:hep-th/9903042, doi:10.1088/1126-6708/1999/08/013, https://arxiv.org/abs/hep-th/9903042
4. Y. Kinar, E. Schreiber, J. Sonnenschein, N. Weiss, "Quantum fluctuations of Wilson loops from string models", Nucl. Phys. B 573 (2000) 764, arXiv:hep-th/9911123, doi:10.1016/S0550-3213(00)00238-8, https://arxiv.org/abs/hep-th/9911123
5. N. Drukker, D. J. Gross, "An exact prediction of N=4 supersymmetric Yang-Mills theory for string theory", J. Math. Phys. 42 (2001) 2896, arXiv:hep-th/0010274, doi:10.1063/1.1372177, https://arxiv.org/abs/hep-th/0010274
6. G. W. Semenoff, D. Young, "Wavy Wilson line and AdS/CFT", Int. J. Mod. Phys. A 20 (2005) 2833, arXiv:hep-th/0405288, doi:10.1142/S0217751X0502077X, https://arxiv.org/abs/hep-th/0405288
7. B. Fiol, B. Garolera, G. Torrents, "Exact momentum fluctuations of an accelerated quark in N=4 super Yang-Mills", JHEP 1306 (2013) 011, arXiv:1302.6991, doi:10.1007/JHEP06(2013)011, https://arxiv.org/abs/1302.6991
8. Y. Makeenko, P. Olesen, G. W. Semenoff, "Cusped SYM Wilson loop at two loops and beyond", Nucl. Phys. B 738 (2006) 243, arXiv:hep-th/0602100, doi:10.1016/j.nuclphysb.2006.05.002, https://arxiv.org/abs/hep-th/0602100
9. H. Hata, A. Miwa, "Loop equation in D=4, N=4 super Yang-Mills theory and string field equation on AdS5 x S5", Phys. Rev. D 73 (2006) 046001, arXiv:hep-th/0510150, doi:10.1103/PhysRevD.73.046001, https://arxiv.org/abs/hep-th/0510150
10. D. Correa, J. Henn, J. Maldacena, A. Sever, "The cusp anomalous dimension at three loops and beyond", JHEP 1205 (2012) 098, arXiv:1203.1019, doi:10.1007/JHEP05(2012)098, https://arxiv.org/abs/1203.1019
11. D. Correa, J. Maldacena, A. Sever, "The quark anti-quark potential and the cusp anomalous dimension from a TBA equation", JHEP 1208 (2012) 134, arXiv:1203.1913, doi:10.1007/JHEP08(2012)134, https://arxiv.org/abs/1203.1913
12. V. Forini, V. Giangreco M. Puletti, O. Ohlsson Sax, "The generalized cusp in AdS(4) x CP(3) and more one-loop results from semiclassical strings", J. Phys. A 46 (2013) 115402, arXiv:1204.3302, doi:10.1088/1751-8113/46/11/115402, https://arxiv.org/abs/1204.3302
13. R. Espíndola, J. A. García, "Cusp anomalous dimension and rotating open strings in AdS/CFT", JHEP 1803 (2018) 116, arXiv:1607.05305, doi:10.1007/JHEP03(2018)116, https://arxiv.org/abs/1607.05305
14. A. I. Karanikas, C. N. Ktorides, "Zig zag symmetry in AdS/CFT duality", Eur. Phys. J. C 52 (2007) 67, arXiv:0704.2103, doi:10.1140/epjc/s10052-007-0504-3, https://arxiv.org/abs/0704.2103
15. C. R. Graham, E. Witten, "Conformal anomaly of submanifold observables in AdS/CFT correspondence", Nucl. Phys. B 546 (1999) 52, arXiv:hep-th/9901021, doi:10.1016/S0550-3213(99)00055-3, https://arxiv.org/abs/hep-th/9901021
16. A. Gustavsson, "Conformal anomaly of Wilson surface observables: A field theoretical computation", JHEP 0407 (2004) 074, arXiv:hep-th/0404150, doi:10.1088/1126-6708/2004/07/074, https://arxiv.org/abs/hep-th/0404150
17. M. Kruczenski, "Wilson loops and minimal area surfaces in hyperbolic space", JHEP 1411 (2014) 065, arXiv:1406.4945, doi:10.1007/JHEP11(2014)065, https://arxiv.org/abs/1406.4945
18. N. Drukker, "Integrable Wilson loops", JHEP 1310 (2013) 135, arXiv:1203.1617, doi:10.1007/JHEP10(2013)135, https://arxiv.org/abs/1203.1617
19. H. Münkler, "The cross anomalous dimension in maximally supersymmetric Yang-Mills theory", JHEP 1810 (2018) 162, arXiv:1805.06448, doi:10.1007/JHEP10(2018)162, https://arxiv.org/abs/1805.06448
20. V. Forini, A. A. Tseytlin, E. Vescovi, "Perturbative computation of string one-loop corrections to Wilson loop minimal surfaces in AdS5 x S5", JHEP 1703 (2017) 003, arXiv:1702.02164, doi:10.1007/JHEP03(2017)003, https://arxiv.org/abs/1702.02164
21. A. Cagnazzo, D. Medina-Rincon, K. Zarembo, "String corrections to circular Wilson loop and anomalies", JHEP 1802 (2018) 120, arXiv:1712.07730, doi:10.1007/JHEP02(2018)120, https://arxiv.org/abs/1712.07730
22. D. Medina-Rincon, A. A. Tseytlin, K. Zarembo, "Precision matching of circular Wilson loops and strings in AdS5 x S5", JHEP 1805 (2018) 199, arXiv:1804.08925, doi:10.1007/JHEP05(2018)199, https://arxiv.org/abs/1804.08925
23. S. Giombi, S. Komatsu, "Loop equation and exact soft anomalous dimension in N=4 super Yang-Mills", JHEP 2006 (2020) 075, arXiv:2003.04460, doi:10.1007/JHEP06(2020)075, https://arxiv.org/abs/2003.04460
24. N. Drukker, M. Trépanier, "Ironing out the crease", JHEP 2208 (2022) 193, arXiv:2204.12627, doi:10.1007/JHEP08(2022)193, https://arxiv.org/abs/2204.12627