Complete computation of all three-loop five-point massless planar integrals

We calculate all three-loop, five-point, massless planar Feynman integral families in the dimensional regularization scheme. This is a new milestone in Feynman integral computations. The analysis covers four distinct families of Feynman integrals for this configuration, for all of which we derive the canonical differential equations. Our results also confirm a prediction on the three-loop five-point alphabet. The boundary values are analytically determined. Using these differential equations, the integrals can be evaluated to high precision efficiently. Our work establishes the foundation for next-to-next-to-next-to-leading-order (N$^3$LO) calculation of the production of three massless final states, as well as corresponding bootstrap studies in gauge theories.

💡 Research Summary

The paper presents a complete analytic computation of all planar three‑loop five‑point massless Feynman integral families within dimensional regularisation (D = 4 − 2ε). This achievement marks a significant milestone because three‑loop five‑point integrals are the last missing ingredient for next‑to‑next‑to‑next‑to‑leading order (N³LO) predictions of processes with three massless final‑state particles, which are crucial for precision collider phenomenology and high‑precision gravitational‑wave template generation.

Four distinct planar topologies are identified (Figure 1): Pentagons‑Box‑Box (PBB), Box‑Pentagon‑Box (BPB), Box‑Hexagon‑Box (BHB) and Pentagons‑Box‑Pentagons (PBP). Each topology is described by 18 propagators D_i and a possible numerator N, with external kinematics expressed through the five Mandelstam invariants s_{12},…,s_{15} and the parity‑odd invariant ε₅. The authors adopt dual coordinates and Baikov variables to write the generic integral I_{ν₁…ν₁₈}(N).

IBP reduction, traditionally a bottleneck, is performed with the state‑of‑the‑art algebraic‑geometry‑based program NeatIBP 1.1. This tool efficiently generates IBP relations using Gröbner‑basis and module‑lift techniques, allowing the authors to find 316, 367, 431 and 734 master integrals for the PBB, BPB, BHB and PBP families respectively. The “tennis‑court” families (BHB, PBP) are considerably more complex, but the inclusion of the SpanningCut algorithm makes the reduction feasible.

A pure (ε‑uniform) integral basis is then constructed for each family. The construction combines loop‑by‑loop leading‑singularity analysis, Baikov representation, D‑log basis algorithms, and the module‑lift method to ensure each basis element has unit leading singularity and uniform transcendental weight. New sub‑topologies (Figure 2) and intricate numerators (e.g. N₄, N₁₃, N₁₇, N₁₉) are treated explicitly, demonstrating the authors’ ability to handle sectors with a large number of integrals.

With the pure basis, the differential equations take the canonical ε‑factorised form d I = ε d A I, where d A is a sum of logarithmic differentials of an alphabet of symbol letters W_i. The alphabet extends the well‑known two‑loop pentagon alphabet by 50 new letters (cW_i, fW_i, etc.), yielding a total of 56 letters. These letters are built from the Mandelstam invariants, the quartic polynomial Δ^{(1)}₄, and its cyclic images under the dihedral group D₅. The authors verify that the alphabet is closed under D₅ and that it fully captures all singularities of the three‑loop planar five‑point integrals.

Boundary conditions are fixed at the symmetric Euclidean point x₀ = {−1,…,−1} by demanding the absence of spurious poles. The integration paths are chosen to rationalise the square‑root structures: for the ladder‑type families (PBB, BPB) a single √ε₅ appears, while for the tennis‑court families (BHB, PBP) additional √Δ^{(i)}₄ terms lead to a change of variables x → t². This rationalisation reduces the effective alphabet to 17 letters along the path, allowing the authors to compute all boundary values analytically up to weight six in terms of Goncharov polylogarithms. Numerical values are obtained with the NumPolyLog library and cross‑checked against independent sector‑decomposition results from pySecDec.

The symbol analysis reveals the structure of the function space. Up to weight three the symbols involve only the two‑loop pentagon letters; from weight four onward the new three‑loop letters contribute. In total, the authors find 2 220 independent symbols at weight six, confirming the richness of the space while also noting that many possible letter pairs never appear, reflecting hidden constraints.

For practical evaluation, the canonical differential equations together with the analytically determined boundary values are solved numerically using the DiffExp package. The authors demonstrate that a full set of 734 pure integrals in the PBP family can be evaluated to high precision in less than ten minutes on a single core for a generic Euclidean kinematic point. Independent validation with pySecDec confirms the correctness of the results.

In summary, the paper delivers the first complete analytic description of all planar three‑loop five‑point massless integrals, providing the necessary ingredients for N³LO calculations of 2→3 scattering processes, high‑precision gravitational‑wave waveform modeling, and bootstrap approaches to multi‑loop amplitudes. The methodology—combining modern IBP reduction, pure‑basis construction, canonical differential equations, and systematic symbol analysis—sets a template for future extensions to non‑planar topologies, massive external legs, and even higher‑loop orders.