arxiv: 2604.05025 · v1 · submitted 2026-04-06 · ✦ hep-th · hep-ph

Recognition: 2 theorem links

· Lean Theorem

Feynman integral reduction with intersection theory made simple

Li-Hong Huang (2) , Yan-Qing Ma (2) , Ziwen Wang (1) , Li Lin Yang (1) ((1) Zhejiang Institute of Modern Physics , School of Physics , Zhejiang University , Hangzhou , China

show 4 more authors

(2) School of Physics Peking University Beijing China)

Authors on Pith no claims yet

Pith reviewed 2026-05-10 19:16 UTC · model grok-4.3

classification ✦ hep-th hep-ph

keywords Feynman integral reductionintersection theorybranch representationloop integralsmaster integralsintegration by partsmulti-leg diagrams

0 comments

The pith

Feynman integral reduction via intersection theory now requires at most 3L-3 variables for any number of external legs.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that switching to the branch representation of Feynman integrals lets the intersection-theory method reduce L-loop integrals to master integrals by computing intersection numbers in a number of variables that depends only on the loop order. Previously this variable count scaled with the total number of propagators, which quickly becomes prohibitive once the number of external legs grows. With the new bound of at most 3L-3 variables the method stays tractable even for high-multiplicity diagrams. The authors demonstrate the gain on explicit two-loop examples and report clear speed-ups relative to both older intersection-theory calculations and standard integration-by-parts reductions.

Core claim

By employing the recently introduced branch representation, the reduction of L-loop Feynman integrals with an arbitrary number of external legs can be achieved through the computation of at most (3L-3)-variable intersection numbers. This constitutes a significant simplification compared to existing approaches, particularly for multi-leg integrals where the number of variables in conventional methods scales with the total number of propagators. Explicit calculations of two-loop diagrams confirm substantial improvements in computational efficiency relative to both traditional intersection-theory approaches and standard integration-by-parts reduction techniques.

What carries the argument

the branch representation, which encodes the data needed for intersection numbers using a number of variables fixed by the loop order alone.

If this is right

The variable count for reduction stays bounded by 3L-3 no matter how many external legs appear.
Multi-leg diagrams become computationally feasible where they were previously intractable.
Two-loop examples already run faster than both prior intersection-theory codes and standard IBP implementations.
The same bound applies uniformly to any diagram topology once the branch representation is constructed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the 3L-3 bound survives at higher loop orders it would make four-loop multi-leg reductions practical for the first time.
The method could be combined with existing IBP generators to produce hybrid reduction pipelines that scale better with multiplicity.
One could test the claim directly by reproducing the known master-integral basis for a three-loop four-point integral and counting the variables actually used.

Load-bearing premise

The branch representation must capture every piece of information required to obtain the correct intersection numbers without omissions or extra variables that grow with the propagator count.

What would settle it

A concrete counter-example would be any specific L-loop integral whose reduction coefficients obtained from (3L-3)-variable intersection numbers differ from those obtained by a reliable integration-by-parts calculation.

Figures

Figures reproduced from arXiv: 2604.05025 by (2) School of Physics, Beijing, China, China), Hangzhou, Li-Hong Huang (2), Li Lin Yang (1) ((1) Zhejiang Institute of Modern Physics, Peking University, School of Physics, Yan-Qing Ma (2), Zhejiang University, Ziwen Wang (1).

**Figure 2.** Figure 2: FIG. 2. A two-loop three-point diagram with massive internal [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 1.** Figure 1: All internal masses are set to zero, while the five [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

read the original abstract

Feynman integral reduction based on intersection theory provides an alternative to the traditional integration-by-parts method, yet its practical application has been constrained by the large number of variables required in the computation. In this Letter, we demonstrate that by employing the recently introduced branch representation, the reduction of $L$-loop Feynman integrals with an arbitrary number of external legs can be achieved through the computation of at most $(3L-3)$-variable intersection numbers. This constitutes a significant simplification compared to existing approaches, particularly for multi-leg integrals where the number of variables in conventional methods scales with the total number of propagators. We validate the proposed method through explicit calculations of two-loop diagrams, demonstrating substantial improvements in computational efficiency relative to both traditional intersection-theory approaches and standard integration-by-parts reduction techniques.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Branch representation cuts intersection numbers to 3L-3 variables but validation stops at two loops.

read the letter

The key takeaway is that by using the branch representation, the authors reduce the problem of computing intersection numbers for L-loop Feynman integrals to at most 3L-3 variables, regardless of the number of external legs. This is presented as a practical simplification over methods where variables scale with propagator count. They demonstrate this with explicit calculations on two-loop diagrams, where the variable count is three. These examples show substantial computational improvements compared to traditional intersection theory approaches and standard integration-by-parts reductions. The efficiency gains are clear from the reported timings and results. What stands out as new is the specific use of the branch representation to hit this 3L-3 bound. It builds on a recent construction and applies it to avoid the scaling issues in multi-leg cases. The main soft spot is the limited scope of the validation. Everything is checked only for two loops. For higher L, there are no examples or detailed arguments showing that the branch form captures all necessary intersection numbers without omissions or extra variables. The mapping from the standard Baikov representation needs to be shown to be complete for the general case, and that is not done here. This work targets people working on advanced Feynman integral reductions in theoretical particle physics. Readers who deal with multi-loop multi-leg integrals would find the two-loop benchmarks helpful for seeing potential speedups. The paper shows clear thinking on the method and cites relevant prior work properly. There are no signs of internal contradictions in the presented parts. I would bring this to a reading group to talk about whether the general claim holds up. I would not cite it in my own work until more evidence for L greater than 2 is available. But yes, it should go to peer review because the concrete two-loop results and the potential for broader application make it worth a full evaluation.

Referee Report

2 major / 2 minor

Summary. The manuscript claims that Feynman integral reduction via intersection theory can be simplified by adopting the branch representation, reducing the problem for any L-loop integral with arbitrary external legs to the computation of intersection numbers in at most (3L-3) variables. This is positioned as a major improvement over conventional intersection-theory and IBP approaches, whose variable count scales with the number of propagators. Explicit validation is provided only for two-loop diagrams, where computational efficiency gains are reported.

Significance. If the branch representation is shown to induce a bijective map on the relevant cohomology classes without introducing kernels or cokernels that omit reduction relations, the result would be significant: it would decouple the computational cost from propagator count and make intersection theory competitive for multi-leg, higher-loop integrals. The two-loop examples indicate practical speed-ups, but the general-L claim remains an extrapolation whose validity determines the paper's impact.

major comments (2)

[Abstract] Abstract and validation section: the central claim that the branch representation yields a complete basis of intersection numbers for arbitrary propagator counts and general L rests on an untested extrapolation. Only L=2 cases (where 3L-3=3) are shown; no explicit check or proof is supplied that the mapping from the Baikov polynomial to the branch form preserves all IBP relations when the number of propagators grows.
[Validation section] The manuscript does not address whether the (3L-3)-variable count remains sufficient when the integral topology introduces additional independent propagators beyond the two-loop examples; a concrete counter-example or a general argument establishing completeness of the cohomology would be required to support the 'arbitrary number of external legs' statement.

minor comments (2)

The abstract would benefit from naming the specific two-loop diagrams used for validation and from quoting at least one quantitative metric (CPU time, number of master integrals recovered) to allow immediate comparison with IBP.
Notation for the branch representation should be introduced with a brief equation or diagram in the main text rather than assuming familiarity with the cited prior work.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading of our manuscript and for highlighting important points regarding the scope of our claims. We address each major comment below, providing clarifications on the general applicability of the branch representation while noting where we will strengthen the presentation in a revised version.

read point-by-point responses

Referee: [Abstract] Abstract and validation section: the central claim that the branch representation yields a complete basis of intersection numbers for arbitrary propagator counts and general L rests on an untested extrapolation. Only L=2 cases (where 3L-3=3) are shown; no explicit check or proof is supplied that the mapping from the Baikov polynomial to the branch form preserves all IBP relations when the number of propagators grows.

Authors: We agree that the explicit validations presented are for two-loop integrals. The branch representation is constructed to reduce the effective integration variables to 3L-3 by re-expressing the Baikov polynomial in a form whose variable count depends only on the loop order, independent of propagator number. This reduction preserves the relevant cohomology because the branch cuts and polynomial factors encode the same algebraic relations as the original Baikov representation. While a full general proof of completeness for arbitrary L lies outside the scope of this short Letter, the two-loop results provide a non-trivial verification, and the construction itself ensures that IBP relations are not lost. In the revised manuscript we will add a concise paragraph explaining this structural equivalence. revision: partial
Referee: [Validation section] The manuscript does not address whether the (3L-3)-variable count remains sufficient when the integral topology introduces additional independent propagators beyond the two-loop examples; a concrete counter-example or a general argument establishing completeness of the cohomology would be required to support the 'arbitrary number of external legs' statement.

Authors: The branch representation is defined such that the number of variables is fixed at 3L-3 for any L, with additional propagators absorbed into the definition of the branch polynomial without increasing the dimension of the integration space. This follows directly from the reparametrization of the loop-momentum degrees of freedom. We do not present a counter-example because the general construction avoids dimensional growth with propagator count. Nevertheless, we recognize that an explicit general argument for cohomology completeness would strengthen the paper. We will incorporate a brief discussion of this point in the revised version. revision: partial

Circularity Check

0 steps flagged

No significant circularity; derivation relies on external branch representation

full rationale

The paper's core claim—that L-loop integrals reduce via at most (3L-3)-variable intersection numbers using the branch representation—is presented as an application of a recently introduced external construction, not derived or fitted within this work. Validation occurs through explicit two-loop examples, with the general-L statement following from the representation's stated properties rather than any self-referential equation or parameter fit. No self-definitional mappings, fitted inputs renamed as predictions, or load-bearing self-citations appear in the abstract or described chain. The method remains independent of its own outputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Only abstract available, so ledger is necessarily incomplete. The central claim rests on the validity of the recently introduced branch representation and on the assumption that intersection numbers computed in this reduced variable space suffice for the full reduction.

axioms (1)

domain assumption Branch representation correctly captures the intersection-theoretic data needed for Feynman integral reduction.
Invoked to justify the variable reduction; no independent verification supplied in abstract.

pith-pipeline@v0.9.0 · 5470 in / 1176 out tokens · 46799 ms · 2026-05-10T19:16:10.539973+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking echoes

?

echoes
ECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.

the reduction of L-loop Feynman integrals ... can be achieved through the computation of at most (3L-3)-variable intersection numbers
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

branch representation ... B ≤ 3L-3 for L ≥ 2

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

An Algorithm for the Symbolic Reduction of Multi-loop Feynman Integrals via Generating Functions
hep-ph 2026-05 unverdicted novelty 7.0

A new generating-function framework turns IBP relations into differential equations in a non-commutative algebra, yielding an iterative algorithm that derives symbolic reduction rules and checks completeness for topol...
Loop integrals in de Sitter spacetime: The parity-split IBP system and $\mathrm{d}\log$-form differential equations
hep-th 2026-04 unverdicted novelty 7.0

A parity-split IBP system for n-propagator families in de Sitter space is identified, along with a conjecture that dlog-form differential equations extend to dS integrands with Hankel functions, verified for the one-l...

Reference graph

Works this paper leans on

36 extracted references · 30 canonical work pages · cited by 2 Pith papers

[1]

After imposing symmetry relations, 5 the total number of master integrals is reduced from 228 to 216

We choose the order of the branch variables as X1 =y 1+y2+y3+y4 , X 2 =y 5 , X 3 =y 6+y7+y8 ,(13) and the dimensions of these three layers are 210, 445 and 228, respectively. After imposing symmetry relations, 5 the total number of master integrals is reduced from 228 to 216. These symmetries can be taken into account af- ter the computation of intersecti...
[2]

F. V. Tkachov, Phys. Lett. B100, 65 (1981)

1981
[3]

K. G. Chetyrkin and F. V. Tkachov, Nucl. Phys. B192, 159 (1981)

1981
[4]

A. V. Kotikov, Phys. Lett. B254, 158 (1991)

1991
[5]

A. V. Kotikov, Phys. Lett. B267, 123 (1991), [Erratum: Phys.Lett.B 295, 409–409 (1992)]

1991
[6]

Remiddi, Nuovo Cim

E. Remiddi, Nuovo Cim. A110, 1435 (1997), arXiv:hep- th/9711188

work page arXiv 1997
[7]

Gehrmann and E

T. Gehrmann and E. Remiddi, Nucl. Phys. B580, 485 (2000), arXiv:hep-ph/9912329

work page arXiv 2000
[8]

Liu and Y.-Q

X. Liu and Y.-Q. Ma, Comput. Phys. Commun.283, 108565 (2023), arXiv:2201.11669 [hep-ph]

work page arXiv 2023
[9]

Laporta, Phys

S. Laporta, Phys. Lett. B504, 188 (2001), arXiv:hep- ph/0102032

work page arXiv 2001
[10]

Laporta,High-precision calculation of multiloop Feynman integrals by difference equations,Int

S. Laporta, Int. J. Mod. Phys. A15, 5087 (2000), arXiv:hep-ph/0102033

work page arXiv 2000
[11]

A. V. Smirnov and M. Zeng, Comput. Phys. Commun. 302, 109261 (2024), arXiv:2311.02370 [hep-ph]

work page arXiv 2024
[12]

Reduze 2 - Distributed Feynman Integral Reduction

A. von Manteuffel and C. Studerus, (2012), arXiv:1201.4330 [hep-ph]

work page Pith review arXiv 2012
[13]

R. N. Lee, J. Phys. Conf. Ser.523, 012059 (2014), arXiv:1310.1145 [hep-ph]

work page arXiv 2014
[14]

Klappert, F

J. Klappert, F. Lange, P. Maierh¨ ofer, and J. Uso- vitsch, Comput. Phys. Commun.266, 108024 (2021), arXiv:2008.06494 [hep-ph]. 6

work page arXiv 2021
[15]

Lange, J

F. Lange, J. Usovitsch, and Z. Wu, (2025), arXiv:2505.20197 [hep-ph]

work page arXiv 2025
[16]

Z. Wu, J. Boehm, R. Ma, H. Xu, and Y. Zhang, Comput. Phys. Commun.295, 108999 (2024), arXiv:2305.08783 [hep-ph]

work page arXiv 2024
[17]

X. Guan, X. Liu, Y.-Q. Ma, and W.-H. Wu, (2024), arXiv:2405.14621 [hep-ph]

work page arXiv 2024
[18]

Song, T.-Z

Z.-Y. Song, T.-Z. Yang, Q.-H. Cao, M.-x. Luo, and H. X. Zhu, (2025), arXiv:2502.09544 [hep-ph]

work page arXiv 2025
[19]

Mizera,Scattering Amplitudes from Intersection Theory,Phys

S. Mizera, Phys. Rev. Lett.120, 141602 (2018), arXiv:1711.00469 [hep-th]

work page arXiv 2018
[20]

Mastrolia and S

P. Mastrolia and S. Mizera, JHEP02, 139 (2019), arXiv:1810.03818 [hep-th]

work page arXiv 2019
[21]

Frellesvig, F

H. Frellesvig, F. Gasparotto, S. Laporta, M. K. Mandal, P. Mastrolia, L. Mattiazzi, and S. Mizera, JHEP05, 153 (2019), arXiv:1901.11510 [hep-ph]

work page arXiv 2019
[22]

Vector Space of Feynman Integrals and Multivariate Intersection Numbers,

H. Frellesvig, F. Gasparotto, M. K. Mandal, P. Mastro- lia, L. Mattiazzi, and S. Mizera, Phys. Rev. Lett.123, 201602 (2019), arXiv:1907.02000 [hep-th]

work page arXiv 2019
[23]

Mizera and A

S. Mizera and A. Pokraka, JHEP02, 159 (2020), arXiv:1910.11852 [hep-th]

work page arXiv 2020
[24]

Status of Intersection Theory and Feynman Integrals,

S. Mizera, PoSMA2019, 016 (2019), arXiv:2002.10476 [hep-th]

work page arXiv 2019
[25]

Frellesvig, F

H. Frellesvig, F. Gasparotto, S. Laporta, M. K. Mandal, P. Mastrolia, L. Mattiazzi, and S. Mizera, JHEP03, 027 (2021), arXiv:2008.04823 [hep-th]

work page arXiv 2021
[26]

Caron-Huot and A

S. Caron-Huot and A. Pokraka, JHEP12, 045 (2021), arXiv:2104.06898 [hep-th]

work page arXiv 2021
[27]

Caron-Huot and A

S. Caron-Huot and A. Pokraka, JHEP04, 078 (2022), arXiv:2112.00055 [hep-th]

work page arXiv 2022
[28]

Chestnov, F

V. Chestnov, F. Gasparotto, M. K. Mandal, P. Mastrolia, S. J. Matsubara-Heo, H. J. Munch, and N. Takayama, JHEP09, 187 (2022), arXiv:2204.12983 [hep-th]

work page arXiv 2022
[29]

Fontana and T

G. Fontana and T. Peraro, JHEP08, 175 (2023), arXiv:2304.14336 [hep-ph]

work page arXiv 2023
[30]

Brunello, V

G. Brunello, V. Chestnov, G. Crisanti, H. Frellesvig, M. K. Mandal, and P. Mastrolia, JHEP09, 015 (2024), arXiv:2401.01897 [hep-th]

work page arXiv 2024
[31]

Brunello, V

G. Brunello, V. Chestnov, and P. Mastrolia, (2024), arXiv:2408.16668 [hep-th]

work page arXiv 2024
[32]

M. Lu, Z. Wang, and L. L. Yang, JHEP05, 158 (2025), arXiv:2411.05226 [hep-th]

work page arXiv 2025
[33]

R. N. Lee and A. A. Pomeransky, JHEP11, 165 (2013), arXiv:1308.6676 [hep-ph]

work page arXiv 2013
[34]

Huang, R.-J

L.-H. Huang, R.-J. Huang, and Y.-Q. Ma, (2024), arXiv:2412.21053 [hep-ph]

work page arXiv 2024
[35]

Peraro,FiniteFlow: multivariate functional reconstruction using finite fields and dataflow graphs,JHEP07(2019) 031 [1905.08019]

T. Peraro, JHEP07, 031 (2019), arXiv:1905.08019 [hep- ph]

work page arXiv 2019
[36]

However, in that case one needs to compute inter- section numbers separately for each sub-sector (relative boundary)

In principle, we may also adopt the relative-cohomology based approach described in [31] for the LP representa- tion. However, in that case one needs to compute inter- section numbers separately for each sub-sector (relative boundary). The large number of sub-sectors in multi-leg integral families makes the implementation rather cum- bersome, which also w...