The Geometry of BCFW for ABJM Loop Integrands

Livia Ferro, Ross Glew, Tomasz Łukowski, and Jonah Stalknecht · 2025

Read The Geometry of BCFW for ABJM Loop Integrands by Livia Ferro, Ross Glew, Tomasz Łukowski, and Jonah Stalknecht (2025) 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-2510-25733.txt.

Introduction

Livia Ferro, Ross Glew, Tomasz Lukowski, and Jonah Stalknecht study ABJM loop geometry. Their 2025 physics paper investigates BCFW geometry for loop integrands. ABJM theory loop-level geometry uses lightcone geometries. Dual space lightcone geometry admits a natural fibration. One loop variable lives in a fiber over lower loops. Varying lower geometry leads to half-chamber definitions.

L-loop half-chambers keep half the next fiber unchanged. A full classification and surprising bijection are provided. N-point L-loop half-chambers correspond to Feynman diagrams. Those diagrams belong to a cubic scalar theory. The scalar theory has half as many particles. Summing half-chambers computes the ABJM amplitude directly. That sum corresponds to scalar amplitude diagram sums. Loop fibers realize those Feynman diagrams geometrically. Half-chamber expansion equals loop-level BCFW recursion, they argue. Chambers emerge as intersections of two half-chambers. Concrete examples illustrate the construction.

Catalog record

CREATORLivia Ferro, Ross Glew, Tomasz Łukowski, and Jonah Stalknecht
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE81 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-25733

Browse the full library · About the Tower of Babel

Full text

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

The Geometry of BCFW for ABJM Loop Integrands Livia Ferro, Ross Glew, Tomasz Łukowski, and Jonah Stalknecht Abstract In this paper we investigate the loop-level geometry of ABJM theory from the perspective of lightcone geometries in dual space. This geometry admits a natural fibration, where one of the loop variables can be naturally interpreted as living in a fiber for each fixed point of a lower-loop geometry. When varying the latter, this leads us to the definition of L L -loop half-chambers , defined such that ‘half’ of the ( L + 1 ) (L+1) -loop fiber remains unchanged. We provide a full classification of these half-chambers, and demonstrate a surprising bijection between n n -point L L -loop half-chambers and L L -loop Feynman diagrams for a cubic scalar theory with n / 2 n/2 particles. Consequently, the sum over L L -loop half-chambers that computes the n n -point ABJM amplitude is in direct correspondence with the sum over L L -loop Feynman diagrams that computes the ( n / 2 ) (n/2) -point amplitude of Tr ⁡ ( ϕ 3 ) \Tr(\phi^3) theory. These Feynman diagrams are also realised geometrically in the structure of the loop fibers. Furthermore, we argue that the half-chamber expansion is equivalent to the loop-level BCFW recursion for ABJM, which arises naturally from our geometric construction. Finally, we will illustrate how L L -loop chambers emerge as the intersection of two L L -loop half-chambers, and we provide concrete examples of this construction. 1 Introduction Over the past decade, we have learned that scattering amplitudes can be understood as the canonical differential form of certain positive geometries Arkani-Hamed:2017tmz . This idea, first realised for planar \mathcal{N}=4 SYM through the amplituhedron Arkani-Hamed:2013jha ; Arkani-Hamed:2017vfh ; Damgaard:2019ztj ; Ferro:2022abq ; He:2021llb , has since been extended to a variety of theories, including \Tr(\phi^3) via the ABHY associahedron Arkani-Hamed:2017mur and ABJM theory via the ABJM amplituhedron He:2022cup ; He:2023rou . The positive geometry framework now encompasses a wide range of observables, such as correlation functions Eden:2017fow and cosmological wavefunctions Arkani-Hamed:2017fdk ; Benincasa:2024leu , illustrating its broader applicability. In this work, we explore the loop-level geometry of ABJM theory from the perspective of lightcone geometries in dual space. The complexity of the ABJM amplituhedron grows rapidly with both particle number and loop order, making direct computation of its canonical form intractable beyond the simplest of cases. This motivates the development of new techniques that decompose the amplituhedron into simpler pieces whose canonical forms are easier to determine. The most famous decomposition is given by the BCFW recursion relations, which provide us with an all-loop formula for the canonical form \Omega\left[\mathcal{O}_{k}^{(L)}\right] of the ABJM amplituhedron. For example, at six-points one-loop the BCFW recursion yields a sum over four on-shell diagrams given diagrammatically by \displaystyle\Omega[\mathcal{O}_{3}^{(1)}]\! \displaystyle=\ \raisebox{-10.0pt}{\includegraphics[scale]{Figures/bcfw_example_6_pts_2}}\ +\ \raisebox{-10.0pt}{\includegraphics[scale]{Figures/bcfw_example_6_pts_3}}\ +\ \raisebox{-13.0pt}{\includegraphics[scale]{Figures/bcfw_example_6_pts_1}}\ +\ \raisebox{-10.0pt}{\includegraphics[scale]{Figures/bcfw_example_6_pts}}. (1) Recently, a complementary fibration based approach has emerged, naturally formulated in dual momentum space Lukowski:2023nnf ; Ferro:2023qdp ; Ferro:2024vwn ; Glew:2024zoh . In its original formulation, the idea behind fibrations is the following. Each point in the tree-level geometry defines a null-polygon {\bf x}=(x_{1},\ldots,x_{2k}) in dual space, and the lightcones of the cusps x_{i} bound some compact region \Delta({\bf x}) . The loop integrand structure of ABJM is then captured geometrically by the space of L points (y_{1},\ldots,y_{L}) inside \Delta({\bf x}) which are additionally required to be mutually spacelike separated. As proposed in He:2023rou , by decomposing the tree-level kinematic space into chambers for which the combinatorial structure of the loop geometry remains constant, the tree-level and loop-level data effectively decouple. This leads to a factorisation of the canonical form into two components: a tree-level piece—the chambers—whose collection tiles the tree-level kinematic space, and a corresponding loop-level factor—the fibers—which encode the loop geometry. In this way, the full geometry can be viewed as a fibration of the loop geometry over the tree-level chambers. The aim of this paper is to establish a direct connection between the fibration picture and the BCFW recursion for ABJM loop integrands. In contrast to He:2023rou , in which the loop geometry is fibrated over tree-level, we consider fibrations of a single loop variable y_{L+1} over the L -loop geometry. Specifically, given a point in the L -loop amplituhedron, we define the (L+1) -loop fiber as the lightcone geometry associated to one additional loop variable y_{L+1} . In particular, the vertices of this loop-fiber are located at the intersection of three lightcones, associated to either the cusps of the null-polygon or the preceding loop momenta, and can be classified as either future or past vertices. It is then natural to triangulate the L -loop amplituhedron into regions for which the set of future or past vertices of the (L+1) -loop fiber remain constant. We refer to these regions of the L -loop amplituhedron as half-chambers 1 1 1 We refer to these as half-chambers as they fix exactly half of the vertices of the fiber. . We find that the L -loop half-chambers for 2k -particle scattering are naturally labelled by k -particle, L -loop cubic scalar Feynman diagrams, or dually, by triangulations of an L -punctured k -gon. These diagrams are realised geometrically as the skeletal structure of the (L+1) -loop fiber. This is an extension of the results of Lukowski:2023nnf , where a similar phenomenon was observed at tree-level. Since the half-chambers tile the full L -loop geometry, we arrive at the result that ABJM integrands can be computed by summing over all (tadpole-less) triangulations of an L -punctured k -gon. An example of the half-chamber expansion of the six-point one-loop integrand is given diagrammatically by \displaystyle\Omega[\mathcal{O}_{3}^{(1)}]\!=\raisebox{-17.0pt}{\includegraphics[scale]{Figures/three_point_one_loop_fac1}}\ +\ \raisebox{-17.0pt}{\includegraphics[scale]{Figures/three_point_one_loop_fac2}}\ +\ \raisebox{-17.0pt}{\includegraphics[scale]{Figures/three_point_one_loop_fl1}}\ +\ \raisebox{-17.0pt}{\includegraphics[scale]{Figures/three_point_one_loop_intro}}. (2) Comparing this to ( 1 ) we see that the terms in the half-chamber expansion are in bijection with the terms in the BCFW expansion. In fact, we will argue that the L -loop half-chamber expansion is equivalent to the L -loop BCFW recursion for ABJM integrands Huang:2014xza . It is remarkable that a simple question about the combinatorial structure of loop fibers naturally gives rise to the BCFW decomposition of the ABJM amplituhedron. Moreover, this correspondence provides a direct geometric interpretation of these recursion relations. The remainder of this paper is organised as follows. In section 2 we review the basic kinematic facts relevant for studying ABJM integrands in dual space. In section 3 we detail the construction of the ABJM momentum amplituhedron at tree-level, and the loop-level lightcone geometries. This includes an introduction to the chambers and fibers construction. In section 4 and 5 we present the main results of this paper, a detailed discussion of half-chambers and their connection to the BCFW recursion relations. Specifically, we establish that L -loop half-chambers have a natural labelling inherited from the structure of the (L+1) -loop fiber in terms of triangulations of an L -punctured polygon. Moreover, we show a diagrammatic correspondence between the BCFW recursion relations and the L -loop Berends-Giele recursion for cubic scalar Feynman diagrams. In section 6 we give a brief discussion of full L -loop chambers, which arise as the intersection of two L -loop half-chambers. Finally, we conclude with a summary and outlook to some open problems in section

Read the complete text