On the amplitude expansion of gluon correlators in AdS 4 \textrm{AdS}_{4}

Humberto Gomez a,b,c, Renann Lipinski Jusinskas b, Arthur Lipstein d, Cristhiam Lopez-Arcos e · 2026

Read On the amplitude expansion of gluon correlators in AdS 4 \textrm{AdS}_{4} by Humberto Gomez a,b,c, Renann Lipinski Jusinskas b, Arthur Lipstein d, Cristhiam Lopez-Arcos e (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-23776.txt.

Introduction

Gomez, Jusinskas, Lipstein, and Lopez-Arcos show that gluon interaction data in curved anti de Sitter space can be rebuilt entirely from flat space scattering data, in a 2026 paper on theoretical physics. Correlators in AdS space sit at the heart of the AdS CFT correspondence, which links gravity in a curved bulk to a quantum theory on its boundary. Computing them diagram by diagram is laborious.

They prove that every tree level gluon correlator breaks into a sum over energy poles. Each pole's residue gives a flat space scattering amplitude: the full one, or a lower point amplitude with merged external data. The merging rule mirrors the Berends Giele currents, carried into AdS. The formulae hold at every multiplicity and are checked against Feynman diagrams up to five points.

Catalog record

CREATORHumberto Gomez a,b,c, Renann Lipinski Jusinskas b, Arthur Lipstein d, Cristhiam Lopez-Arcos e
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.23776

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the amplitude expansion of gluon correlators in AdS 4 \textrm{AdS}_{4} Humberto Gomez a,b,c, Renann Lipinski Jusinskas b, Arthur Lipstein d, Cristhiam Lopez-Arcos e Abstract We show that tree-level gluon correlators in AdS 4 \textrm{AdS}_{4} admit a natural expansion in terms of flat-space scattering amplitudes at all multiplicities. In particular, every n n -point correlator can be decomposed into a sum over energy poles whose residues are flat-space amplitudes. The n n -point amplitude encodes the flat-space limit while curvature corrections are captured by lower-point amplitudes with merged external data. The merging of external polarizations is recursively defined via an AdS analogue of the Berends-Giele currents, giving rise to all-multiplicity formulae which we verify against Feynman diagram computations up to five points. Crucially, our approach works at the level of full correlators rather than individual diagrams, providing an elegant and transparent alternative to conventional approaches for computing correlators in anti-de Sitter space. I Introduction The study of correlators in anti-de Sitter (AdS) space occupies a central role in modern theoretical physics. It underpins the AdS/CFT correspondence and provides a controlled setting in which to explore the structure of quantum gravity and gauge theories at strong coupling. In recent years, momentum-space methods have proven particularly powerful, revealing that boundary correlators in AdS (and their de Sitter (dS) counterparts, related through a Wick rotation of the radial direction 31 ; 19 ) share deep structural similarities with flat-space scattering amplitudes. A prominent example is the flat space limit, in which flat space amplitudes can be obtained by Fourier transforming correlators along boundary directions and taking the residue around the sum of the magnitudes of the boundary momenta 35 . This observation naturally raises the question: how much flat-space data is embedded in the full correlator? In this paper we answer this question for Yang-Mills theory in \textrm{AdS}_{4} . We show that every tree-level gluon correlator admits a decomposition into a sum over energy poles, where the residue of each pole is a flat-space scattering amplitude: either the full n -point amplitude, or lower-point amplitudes with a precise merging of the external data. The lower-point amplitudes are obtained from the original one by concatenating subsets of external boundary momenta and polarizations. Their energy poles are then obtained by summing over the magnitudes of the merged boundary momenta. The first key insight underlying our construction is that the merged polarizations can be interpreted as genuine flat-space polarizations in the Lorenz gauge. They are Lorentz-covariant, associated to solutions of the free equations of motion, and enjoy a residual gauge symmetry. The second key insight is that the recursion satisfied by the multi-particle currents takes precisely the form of the Berends–Giele (BG) recursion in flat space 17 in terms of merged external data. Together, these two observations allow us to establish all-multiplicity formulae for gluon correlators in \textrm{AdS}_{4} in terms of flat-space amplitude data alone. An important feature of our approach is that it operates at the level of full correlators, rather than individual Feynman diagrams. The amplitude expansion structure only becomes manifest when all contributions to a given correlator are assembled. Our formulation makes this structure explicit from the outset, providing a more efficient and transparent route to holographic observables. Our results can also be adapted to cosmological wavefunction coefficients, complementing and extending the cosmological bootstrap program 15 . The paper is organized as follows. We begin with a detailed treatment of \mathrm{Tr}(\phi^{3}) theory in flat space with a boundary, which serves as a transparent testing ground for the key ideas. In particular, we show that the boundary correlators of this theory admit an amplitude expansion, where merged polarizations are now Lorentz scalars, and match this with explicit Feynman diagram calculations up to six points. Next, we describe our main results for Yang–Mills theory in \textrm{AdS}_{4} , including an all-multiplicity formula which we match with Feynman diagram calculations up to five points. We then present some final remarks and future directions to be investigated. In the appendix we describe our underlying framework in terms of classical multi-particle solutions. Notation: We will work in the Poincare patch of AdS 4 with metric ds^{2}=z^{-2}\eta_{mn}dx^{m}dx^{n}=z^{-2}(\eta_{\mu\nu}dx^{\mu}dx^{\nu}+dz^{2}), (1) where z\geq 0 describes the radial direction, and \eta_{\mu\nu} is the boundary metric with signature (-++) . In practice it will be convenient to perform a Weyl transformation to half of flat space and use the four-dimensional metric \eta_{mn} . Indeed, the action of Yang-Mills theory in AdS 4 (modulo gauge-fixing terms) is the same as that in flat space due to classical scale invariance. It is convenient to Fourier transform correlators to momentum space along the boundary. Each external leg (i.e. each single-particle state) will then be assigned a four-momentum k_{m}=\{k_{\mu},\mathrm{i}k\} , with k=\sqrt{\eta^{\mu\nu}k_{\mu}k_{\nu}} . More generally, we denote multi-particle labels (words) by capital letters, e.g. P=P_{1}\ldots P_{i} , where the labels P_{j} denote sub-words. Such labels appear naturally when defining amplitudes with merged external data. Single particle states will be denoted by lower case one-letter words. We denote the scalar product between two boundary vectors by a\cdot b=\eta^{\mu\nu}a_{\mu}b_{\mu} . Boundary momentum is additive in the multi-particle labels, k^{\mu}_{P}=k^{\mu}_{P_{1}}+\ldots+k_{P_{i}}^{\mu} . For the radial direction, we introduce the energy variables E_{P_{1},\ldots,P_{i}}=k_{P_{1}}+\ldots+k_{P_{i}}. (2) with k_{P}=\sqrt{k_{P}\cdot k_{P}} 1 1 1 The interpretation in terms of energy is more transparent in de Sitter space, where the bulk direction is timelike and the boundary is Euclidean . Whereas momentum is conserved along the boundary directions, this is not the case for energy. Finally, we define Lorentz-invariant Mandelstam variables in terms of (merged) momenta S_{P_{1},\ldots,P_{i}}\equiv k^{2}_{P_{1}\ldots P_{i}}-E_{P_{1},\ldots,P_{i}}^{2}, (3) which will be recurrent in our construction. In this paper, we will consider correlators of matrix-valued scalar fields and Yang-Mills fields. They can be stripped of their matrix structure and expressed in terms of cyclically symmetric color-ordered correlators, which we focus on in practice (see for example 21 for a more detailed discussion of color-ordering). II \rm{Tr}(\phi^{3}) Theory As a toy model, let us consider \textrm{Tr }(\phi^{3}) theory in half of flat space, where the field \phi is an N\times N matrix with one index in the fundamental and another index in the anti-fundamental representation of \mathrm{SU}(N) . The action is given by S=\int d^{3}xdz\,\mathrm{Tr}\left(\frac{1}{2}\partial_{\mu}\phi\partial^{\mu}\phi+\frac{1}{2}\partial_{z}\phi\partial_{z}\phi-\frac{\lambda}{3}\phi^{3}\right), (4) where z\geq 0 . For simplicity, we will set \lambda=1 in the following. Using a Weyl transformation, this can be mapped to a conformally coupled scalar theory in \textrm{AdS}_{4} with a z -dependent cubic interaction (see e.g. 8 ). Let us now consider tree-level boundary correlators with Dirichlet boundary conditions. These can be derived from Feynman diagrams of conformally coupled scalars in AdS 4 (see for example 3 for more details). As mentioned above, we will work with color-ordered correlators, in which we strip the matrix form of the external polarizations and leave behind only scalar placeholders, which are usually set to 1 . For example, at three-points we have \mathcal{C}(1,2,3)=\frac{1}{E_{1,2,3}}\phi_{1}\phi_{2}\phi_{3}, (5) where the coefficient of the energy pole is simply the flat space three-point amplitude A(1,2,3)=\phi_{1}\phi_{2}\phi_{3} . The four-point correlator is given by \mathcal{C}(1,2,3,4)=\frac{\phi_{1}\phi_{2}\phi_{3}\phi_{4}}{E_{1,2,3,4}E_{\textrm{L}}E_{\textrm{R}}}+(\{1,2,3\}\to\{2,3,1\}), (6) where E_{\textrm{L}}=E_{1,2,34} and E_{\textrm{R}}=E_{12,3,4} are known as partial energy singularities. In the flat space limit, E_{1,2,3,4}\to 0 , the product E_{\textrm{L}}E_{\textrm{R}} coincides with the Mandelstam variable S_{1,2} and we obtain \lim_{E_{1,2,3,4}\to 0}\mathcal{C}(1,2,3,4)=\frac{1}{E_{1,2,3,4}}A(1,2,3,4), (7) where the coefficient of

Read the complete text