First-hand information for everyone
This paper presents a new way to build the raw integrands that appear in loop calculations for planar Yang–Mills theory. Instead of summing many Feynman diagrams, the authors start from off‑shell multi‑particle currents — field configurations that do not satisfy the usual on‑shell energy condition — and sew these building blocks together to form loop integrands. They solve those currents from the classical equation of motion using the perturbiner method, and then use a recursive, off‑shell sewing procedure to assemble loop integrands at any loop order.
A central technical point is that the pure gluon part of the planar integrands can be written in a matrix formalism. Each matrix encodes the contribution of a three‑point or four‑point vertex and nearby legs, so a one‑loop kernel becomes a chain of such matrices (plus a special ``contact’’ matrix for cyclic symmetry). Higher‑loop kernels are then obtained by sewing these matrix chains together, which the authors describe as attaching (or “sewing”) matrix chains to lower‑loop kernels. The paper also works out the changes needed when the ghost fields (the b and c fields required by the gauge fixing) are included; these give extra terms and signs and require some additional bookkeeping.
At a high level the method generalizes the Berends–Giele recursion, which builds tree amplitudes from off‑shell currents, to loop integrands. The recursion turns two external legs of a lower‑loop kernel into an internal line by replacing them with their comb‑component currents and the corresponding propagator. Because the matrix form isolates the local contribution of each leg and vertex, the sewing step can be handled locally: one only needs to manipulate the matrices associated with the legs being sewn, rather than redoing the whole expression.
Why this matters: constructing higher‑loop integrands in a systematic way is hard with standard Feynman rules, which produce many diagrams and obscure relations between amplitudes. The off‑shell recursive approach gives an organized path to build integrands and may help discover amplitude relations at higher loops. The matrix formalism in particular makes the off‑shell structure of planar Yang–Mills integrands clearer and could simplify practical algebraic steps in loop construction.
There are important caveats. This letter lays out the formalism and recursion steps rather than providing a complete library of evaluated high‑loop amplitudes. Some low‑point cases need special symmetry factors (for example, the three‑ and four‑leg one‑loop kernels require small adjustments). The Yang–Mills case is more involved than related scalar examples: ghost fields introduce extra cases and sign differences that must be handled carefully. The authors tested one‑loop kernels against FeynCalc and worked through a two‑loop planar example (the two‑point, two‑loop case) and give a recursion strategy for general two‑loop planar integrands, but further development will be needed to turn the method into a widely used computational tool.
In short, the paper proposes a principled, off‑shell recursion for building planar Yang–Mills loop integrands and casts the pure gluon sector in a matrix language. It supplies the recursion steps including ghost contributions and a two‑loop example. The approach is promising for organizing higher‑loop integrands, but it remains a framework that will require more work and checks before it replaces more established diagrammatic methods.