pith. sign in

arxiv: 2606.22500 · v1 · pith:JNOH5J2Tnew · submitted 2026-06-21 · ✦ hep-ph

Taming Symbolic IBP Reduction with Intermediate Bases

Qian Song This is my paper

Pith reviewed 2026-06-26 10:16 UTC · model grok-4.3

classification ✦ hep-ph
keywords IBP reductionintegration-by-partsFeynman integralsintermediate basesreduction coefficientsbox-trianglepentagon-trianglesymbolic reconstruction
0
0 comments X

The pith

An algorithm using sequences of intermediate bases reconstructs IBP reduction coefficients as products of a few analytic matrices with simple rational polynomial entries.

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

The paper seeks to establish that inserting a sequence of intermediate bases into the IBP reduction workflow breaks large linear systems into smaller ones that can be solved sequentially. This produces the full set of analytic reduction coefficients without ever assembling or solving the enormous general ansatz. A reader would care because the size of that ansatz quickly renders direct reconstruction impossible for multi-point massive integrals that appear in precision calculations. The method is shown to work on a three-point massive box-triangle, needing only 3289 numerical samplings against 1,407,406 unknowns, and on a four-point massive pentagon-triangle, needing 13,013 samplings against 21,638,331 unknowns.

Core claim

The central claim is that reduction coefficients for complicated Feynman integrals can be reconstructed through a sequence of intermediate bases, yielding analytic expressions that are products of a few analytic matrices whose non-zero entries are simple rational polynomials of the kinematic variables and space-time dimension. The approach is demonstrated explicitly on the massive box-triangle and pentagon-triangle families, where the required numerical evaluations remain orders of magnitude smaller than the dimension of the general ansatz.

What carries the argument

A sequence of intermediate bases that factors the full reduction into products of smaller analytic matrices with rational polynomial entries.

If this is right

  • All reduction coefficients for the massive box-triangle are obtained from 3289 numerical samplings.
  • All reduction coefficients for the massive pentagon-triangle are obtained from 13013 numerical samplings.
  • The coefficients appear as products of a small number of analytic matrices rather than a single large solution vector.
  • Non-zero entries in those matrices are simple rational polynomials of the external kinematics and dimension.

Where Pith is reading between the lines

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

  • The same intermediate-base construction might extend to other integral families that currently exceed direct ansatz size limits.
  • If the matrix entries stay simple across more families, the method could supply closed-form reduction rules for entire classes of multi-loop amplitudes.
  • The factorization into sequential matrices offers a possible route to combine IBP reduction with other algebraic techniques that also produce sparse or factored expressions.
  • Numerical stability of the final analytic matrices could be tested by comparing reconstructed coefficients against independent high-precision evaluations on specific kinematic points.

Load-bearing premise

Suitable intermediate bases exist for the target integrals and can be constructed systematically so that the coefficients factor into products of matrices whose entries remain simple rational polynomials.

What would settle it

An integral family for which no sequence of intermediate bases can be found that reduces the number of required numerical samplings below the number of unknowns in the general ansatz, or for which the resulting matrix entries are not simple rational polynomials.

Figures

Figures reproduced from arXiv: 2606.22500 by Qian Song.

Figure 1
Figure 1. Figure 1: FIG. 1: 1-loop massive bubble with non-equal mass. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2: Flowchart of the algorithm for finding [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3: left: massive box-triangle; right: massive [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
read the original abstract

Despite many years of development in integration-by-parts reduction, reconstructing all reduction coefficients, which are rational polynomials of kinematic variables and the space-time dimension, remains a non-trivial problem. The main difficulty comes from the large number of unknowns in a general ansatz, which can lead to a linear system that is too large to solve. In this paper, we present an algorithm for reconstructing reduction coefficients through a sequence of intermediate bases. The resulting analytic reduction coefficients are products of a few analytic matrices, whose non-zero entries are simple rational polynomials. We demonstrate the efficiency of this algorithm with two cutting edge examples: a three-point massive box-triangle and a four-point massive pentagon-triangle. Reconstructing all reduction coefficients for the box-triangle (pentagon-triangle) requires 3289 (13013) numerical samplings, significantly fewer than the number of unknowns in the general ansatz, 1407406 (21638331).

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.

Referee Report

1 major / 2 minor

Summary. The paper presents an algorithm for reconstructing integration-by-parts (IBP) reduction coefficients for Feynman integrals via a sequence of intermediate bases. The resulting analytic coefficients are expressed as products of a small number of analytic matrices whose nonzero entries are simple rational polynomials. Efficiency is demonstrated on two topologies: a three-point massive box-triangle (3289 numerical samplings versus 1,407,406 unknowns in the general ansatz) and a four-point massive pentagon-triangle (13,013 samplings versus 21,638,331 unknowns).

Significance. If reproducible, the intermediate-bases approach offers a concrete route to factor large IBP linear systems into smaller, sequentially solvable ones, directly addressing the scaling barrier in symbolic reduction for multi-loop integrals. The explicit sampling counts, the factored matrix representation, and the demonstration on cutting-edge massive topologies constitute verifiable strengths that could be adopted in existing reduction pipelines.

major comments (1)
  1. [Algorithm description (likely §3) and example sections (§4–5)] The central efficiency claim rests on the existence of suitable intermediate bases that can be constructed systematically. The manuscript must supply an explicit, reproducible procedure (including any selection criteria or ordering) for generating these bases; without it, independent verification that the reported sampling counts were not obtained via post-hoc adjustment is impossible.
minor comments (2)
  1. Add a short error-analysis subsection or table quantifying the numerical stability and reconstruction accuracy of the sampled coefficients for both examples.
  2. Clarify the precise definition and dimension of each intermediate basis in the two worked examples so that the matrix-product factorization can be inspected directly.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback and positive assessment of the significance of our work. We address the single major comment below.

read point-by-point responses

  1. Referee: [Algorithm description (likely §3) and example sections (§4–5)] The central efficiency claim rests on the existence of suitable intermediate bases that can be constructed systematically. The manuscript must supply an explicit, reproducible procedure (including any selection criteria or ordering) for generating these bases; without it, independent verification that the reported sampling counts were not obtained via post-hoc adjustment is impossible.

    Authors: We agree that an explicit, reproducible procedure for constructing the intermediate bases is required for independent verification. Section 3 of the manuscript presents the algorithm, including the sequential construction of intermediate bases and the resulting factored matrix representation. To strengthen reproducibility and directly address the concern about post-hoc adjustment, we will revise the manuscript to expand Section 3 with a fully detailed, step-by-step procedure. This will include the precise selection criteria, ordering rules, and decision logic used to generate the bases for both the box-triangle and pentagon-triangle examples, together with pseudocode or a flowchart if appropriate. The revised text will make clear how the reported sampling counts (3289 and 13013) follow deterministically from the algorithm. revision: yes

Circularity Check

0 steps flagged

No significant circularity; algorithmic procedure is self-contained

full rationale

The paper presents an explicit algorithmic procedure for IBP coefficient reconstruction via sequential intermediate bases and linear algebra on concrete topologies (box-triangle, pentagon-triangle). Reported sampling counts (3289, 13013) are directly compared to the size of the general ansatz (1407406, 21638331) and are verifiable without reference to fitted parameters or self-citations. No load-bearing self-citation, self-definitional loop, or ansatz smuggling appears; the central claim reduces to standard linear-system solving applied in stages, with the factored matrix form asserted as an output of the algorithm rather than an input assumption. The derivation chain is therefore independent of its own results.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper introduces an algorithmic technique relying on standard linear algebra over rational functions; no new physical entities or ad-hoc fitted parameters are introduced in the abstract.

axioms (1)
  • standard math Linear systems over the field of rational functions in kinematic variables and spacetime dimension D admit solutions that can be reconstructed from sufficiently many numerical samples.
    The method depends on solving linear systems sequentially via numerical sampling.

pith-pipeline@v0.9.1-grok · 5672 in / 1210 out tokens · 44133 ms · 2026-06-26T10:16:08.368277+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

50 extracted references · 1 canonical work pages

  1. [1]

    Integration by parts: The algorithm to calculateβ-functions in 4 loops,

    K. G. Chetyrkin and F. V. Tkachov, “Integration by parts: The algorithm to calculateβ-functions in 4 loops,” Nucl. Phys. B192, 159–204 (1981)

    1981

  2. [2]

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

    S. Laporta, “High-precision calculation of multiloop Feynman integrals by difference equations,” Int. J. Mod. Phys. A15, 5087–5159 (2000), arXiv:hep-ph/0102033

    Pith/arXiv arXiv 2000

  3. [3]

    Au- tomatic integral reduction for higher order perturba- 5 tive calculations,

    Charalampos Anastasiou and Achilleas Lazopoulos, “Au- tomatic integral reduction for higher order perturba- 5 tive calculations,” JHEP07, 046 (2004), arXiv:hep- ph/0404258

    arXiv 2004

  4. [4]

    Presenting LiteRed: a tool for the Loop In- TEgrals REDuction,

    R. N. Lee, “Presenting LiteRed: a tool for the Loop In- TEgrals REDuction,” (2012), arXiv:1212.2685 [hep-ph]

    Pith/arXiv arXiv 2012

  5. [5]

    LiteRed 1.4: a powerful tool for re- duction of multiloop integrals,

    Roman N. Lee, “LiteRed 1.4: a powerful tool for re- duction of multiloop integrals,” J. Phys. Conf. Ser.523, 012059 (2014), arXiv:1310.1145 [hep-ph]

    Pith/arXiv arXiv 2014

  6. [6]

    Algorithm FIRE – Feynman Integral REduction,

    A. V. Smirnov, “Algorithm FIRE – Feynman Integral REduction,” JHEP10, 107 (2008), arXiv:0807.3243 [hep- ph]

    Pith/arXiv arXiv 2008

  7. [7]

    FIRE4, LiteRed and accompanying tools to solve integration by parts rela- tions,

    A. V. Smirnov and V. A. Smirnov, “FIRE4, LiteRed and accompanying tools to solve integration by parts rela- tions,” Comput. Phys. Commun.184, 2820–2827 (2013), arXiv:1302.5885 [hep-ph]

    Pith/arXiv arXiv 2013

  8. [8]

    FIRE5: A C++ implementation of Feynman Integral REduction,

    Alexander V. Smirnov, “FIRE5: A C++ implementation of Feynman Integral REduction,” Comput. Phys. Com- mun.189, 182–191 (2015), arXiv:1408.2372 [hep-ph]

    Pith/arXiv arXiv 2015

  9. [9]

    FIRE6: Feynman Integral REduction with modular arithmetic,

    A. V. Smirnov and F. S. Chukharev, “FIRE6: Feynman Integral REduction with modular arithmetic,” Comput. Phys. Commun.247, 106877 (2020), arXiv:1901.07808 [hep-ph]

    arXiv 2020

  10. [10]

    Reduze – Feynman integral reduction in C++,

    C. Studerus, “Reduze – Feynman integral reduction in C++,” Comput. Phys. Commun.181, 1293–1300 (2010), arXiv:0912.2546 [physics.comp-ph]

    Pith/arXiv arXiv 2010

  11. [11]

    Reduze 2 - Distributed Feynman Integral Reduction,

    A. von Manteuffel and C. Studerus, “Reduze 2 - Distributed Feynman Integral Reduction,” (2012), arXiv:1201.4330 [hep-ph]

    Pith/arXiv arXiv 2012

  12. [12]

    Kira—A Feynman integral reduction pro- gram,

    Philipp Maierh¨ ofer, Johann Usovitsch, and Pe- ter Uwer, “Kira—A Feynman integral reduction pro- gram,” Comput. Phys. Commun.230, 99–112 (2018), arXiv:1705.05610 [hep-ph]

    Pith/arXiv arXiv 2018

  13. [13]

    Integral reduction with Kira 2.0 and finite field methods,

    Jonas Klappert, Fabian Lange, Philipp Maierh¨ ofer, and Johann Usovitsch, “Integral reduction with Kira 2.0 and finite field methods,” Comput. Phys. Commun.266, 108024 (2021), arXiv:2008.06494 [hep-ph]

    Pith/arXiv arXiv 2021

  14. [14]

    NeatIBP 1.0, a package generating small-size integration-by-parts relations for Feynman in- tegrals,

    Zihao Wu, Janko Boehm, Rourou Ma, Hefeng Xu, and Yang Zhang, “NeatIBP 1.0, a package generating small-size integration-by-parts relations for Feynman in- tegrals,” Comput. Phys. Commun.295, 108999 (2024), arXiv:2305.08783 [hep-ph]

    arXiv 2024

  15. [15]

    Feynman inte- gral reduction: balanced reconstruction of sparse rational functions and implementation on supercomputers in a co- design approach,

    Alexander V. Smirnov and Mao Zeng, “Feynman inte- gral reduction: balanced reconstruction of sparse rational functions and implementation on supercomputers in a co- design approach,” Numerical Methods and Programming 25, 30–45 (2024), arXiv:2409.19099 [hep-ph]

    arXiv 2024

  16. [16]

    A novel approach to integration by parts reduction,

    Andreas von Manteuffel and Robert M. Schabinger, “A novel approach to integration by parts reduction,” Phys. Lett. B744, 101–104 (2015), arXiv:1406.4513 [hep-ph]

    Pith/arXiv arXiv 2015

  17. [17]

    Scattering amplitudes over finite fields and multivariate functional reconstruction,

    Tiziano Peraro, “Scattering amplitudes over finite fields and multivariate functional reconstruction,” JHEP12, 030 (2016), arXiv:1608.01902 [hep-ph]

    Pith/arXiv arXiv 2016

  18. [18]

    Reconstructing ra- tional functions with FireFly,

    Jonas Klappert and Fabian Lange, “Reconstructing ra- tional functions with FireFly,” Comput. Phys. Commun. 247, 106951 (2020), arXiv:1904.00009 [cs.SC]

    arXiv 2020

  19. [19]

    FiniteFlow: multivariate functional re- construction using finite fields and dataflow graphs,

    Tiziano Peraro, “FiniteFlow: multivariate functional re- construction using finite fields and dataflow graphs,” JHEP07, 031 (2019), arXiv:1905.08019 [hep-ph]

    Pith/arXiv arXiv 2019

  20. [20]

    Scaling up to multivariate rational function reconstruction,

    Andreas Maier, “Scaling up to multivariate rational function reconstruction,” Comput. Phys. Commun.317, 109827 (2025), arXiv:2409.08757 [hep-ph]

    arXiv 2025

  21. [21]

    Balancing act: Multivariate rational reconstruc- tion for IBP,

    A. V. Belitsky, A. V. Smirnov, and R. V. Yakovlev, “Balancing act: Multivariate rational reconstruc- tion for IBP,” Nucl. Phys. B993, 116253 (2023), arXiv:2303.02511 [hep-ph]

    arXiv 2023

  22. [22]

    Maximal quotient rational recon- struction: an almost optimal algorithm for rational re- construction,

    Michael Monagan, “Maximal quotient rational recon- struction: an almost optimal algorithm for rational re- construction,” inProceedings of the 2004 International Symposium on Symbolic and Algebraic Computation, IS- SAC ’04 (Association for Computing Machinery, New York, NY, USA, 2004) p. 243–249

    2004

  23. [23]

    A deterministic algorithm for sparse multivariate polynomial interpola- tion,

    Michael Ben-Or and Prasoon Tiwari, “A deterministic algorithm for sparse multivariate polynomial interpola- tion,” inProceedings of the Twentieth Annual ACM Sym- posium on Theory of Computing, STOC ’88 (Association for Computing Machinery, New York, NY, USA, 1988) p. 301–309

    1988

  24. [24]

    Interpolating polynomials from their values,

    Richard Zippel, “Interpolating polynomials from their values,” (1989)

    1989

  25. [25]

    Computational complexity of sparse rational inter- polation,

    Dima Grigoriev, Marek Karpinski, and Michael Singer, “Computational complexity of sparse rational inter- polation,” SIAM Journal on Computing23(1997), 10.1137/S0097539791194069

    work page doi:10.1137/s0097539791194069 1997

  26. [26]

    Greatest common divisors of polynomi- als given by straight-line programs,

    Erich Kaltofen, “Greatest common divisors of polynomi- als given by straight-line programs,” J. ACM35, 231–264 (1988)

    1988

  27. [27]

    Computing with poly- nomials given by black boxes for their evaluation: Great- est common divisors, factorization, separation of numer- ators and denominators,

    Erich Kaltofen and Barry Trager, “Computing with poly- nomials given by black boxes for their evaluation: Great- est common divisors, factorization, separation of numer- ators and denominators,” (1988) pp. 296–305

    1988

  28. [28]

    On exact and ap- proximate interpolation of sparse rational functions,

    Erich Kaltofen and Zhengfeng Yang, “On exact and ap- proximate interpolation of sparse rational functions,” inProceedings of the 2007 International Symposium on Symbolic and Algebraic Computation, ISSAC ’07 (Asso- ciation for Computing Machinery, New York, NY, USA,

    2007

  29. [29]

    Algorithms for the non-monic case of the sparse modular gcd algorithm,

    Jennifer de Kleine, Michael Monagan, and Allan Wit- tkopf, “Algorithms for the non-monic case of the sparse modular gcd algorithm,” inProceedings of the 2005 In- ternational Symposium on Symbolic and Algebraic Com- putation, ISSAC ’05 (Association for Computing Machin- ery, New York, NY, USA, 2005) p. 124–131

    2005

  30. [30]

    Foxbox: a system for manipulating symbolic objects in black box representa- tion,

    Angel D´ ıaz and Erich Kaltofen, “Foxbox: a system for manipulating symbolic objects in black box representa- tion,” inProceedings of the 1998 International Sympo- sium on Symbolic and Algebraic Computation, ISSAC ’98 (Association for Computing Machinery, New York, NY, USA, 1998) p. 30–37

    1998

  31. [31]

    Early termination in ben-or/tiwari sparse interpolation and a hybrid of zippel’s algorithm,

    Erich Kaltofen, Wen-shin Lee, and Austin A. Lobo, “Early termination in ben-or/tiwari sparse interpolation and a hybrid of zippel’s algorithm,” inProceedings of the 2000 International Symposium on Symbolic and Alge- braic Computation, ISSAC ’00 (Association for Comput- ing Machinery, New York, NY, USA, 2000) p. 192–201

    2000

  32. [32]

    Sparse polyno- mial interpolation with finitely many values for the coef- ficients,

    Qiao-Long Huang and Xiao-Shan Gao, “Sparse polyno- mial interpolation with finitely many values for the coef- ficients,” (2017), arXiv:1704.04359 [cs.SC]

    Pith/arXiv arXiv 2017

  33. [33]

    Interpolation of dense and sparse rational functions and other improvements in FireFly,

    Jonas Klappert, Sven Yannick Klein, and Fabian Lange, “Interpolation of dense and sparse rational functions and other improvements in FireFly,” Comput. Phys. Com- mun.264, 107968 (2021), arXiv:2004.01463 [cs.MS]

    arXiv 2021

  34. [34]

    Barycentric ratio- nal interpolation with no poles and high rates of approx- imation,

    Michael S. Floater and Kai Hormann, “Barycentric ratio- nal interpolation with no poles and high rates of approx- imation,” Numerische Mathematik107, 315–331 (2007)

    2007

  35. [35]

    Multiloop integrals in dimensional regularization made simple,

    Johannes M. Henn, “Multiloop integrals in dimensional regularization made simple,” Phys. Rev. Lett.110, 251601 (2013), arXiv:1304.1806 [hep-th]

    Pith/arXiv arXiv 2013

  36. [36]

    Reducing differential equations for multiloop master integrals,

    Roman N. Lee, “Reducing differential equations for multiloop master integrals,” JHEP04, 108 (2015), 6 arXiv:1411.0911 [hep-ph]

    Pith/arXiv arXiv 2015

  37. [37]

    Reconstruction of rational functions made simple,

    Xiao Liu, “Reconstruction of rational functions made simple,” Phys. Lett. B850, 138491 (2024), arXiv:2306.12262 [hep-ph]

    arXiv 2024

  38. [38]

    How to choose master integrals,

    A. V. Smirnov and V. A. Smirnov, “How to choose master integrals,” Nucl. Phys. B960, 115213 (2020), arXiv:2002.08042 [hep-ph]

    arXiv 2020

  39. [39]

    Factorization of denomina- tors in integration-by-parts reductions,

    Johann Usovitsch, “Factorization of denomina- tors in integration-by-parts reductions,” (2020), arXiv:2002.08173 [hep-ph]

    arXiv 2020

  40. [40]

    Analytic Form of Planar Two-Loop Five-Gluon Scattering Amplitudes in QCD,

    S. Abreu, J. Dormans, F. Febres Cordero, H. Ita, and B. Page, “Analytic Form of Planar Two-Loop Five-Gluon Scattering Amplitudes in QCD,” Phys. Rev. Lett.122, 082002 (2019), arXiv:1812.04586 [hep-ph]

    Pith/arXiv arXiv 2019

  41. [41]

    Ans¨ atze for scat- tering amplitudes from p-adic numbers and algebraic ge- ometry,

    Giuseppe De Laurentis and Ben Page, “Ans¨ atze for scat- tering amplitudes from p-adic numbers and algebraic ge- ometry,” JHEP12, 140 (2022), arXiv:2203.04269 [hep- th]

    arXiv 2022

  42. [42]

    p-adic reconstruction of rational functions in multiloop amplitudes,

    Herschel A. Chawdhry, “p-adic reconstruction of rational functions in multiloop amplitudes,” Phys. Rev. D110, 056028 (2024), arXiv:2312.03672 [hep-ph]

    arXiv 2024

  43. [43]

    Factorization of denominators as a ‘fuel’ for Feynman integral reduction,

    Alexander V. Smirnov, Vladislav. A. Fokin, and Egor Yu. Chuvashov, “Factorization of denominators as a ‘fuel’ for Feynman integral reduction,” (2026), arXiv:2605.13256 [hep-ph]

    Pith/arXiv arXiv 2026

  44. [44]

    Critical points and syzy- gies for Feynman integrals,

    Ben Page and Qian Song, “Critical points and syzy- gies for Feynman integrals,” JHEP02, 004 (2026), arXiv:2509.17681 [hep-th]

    arXiv 2026

  45. [45]

    Feynman integral reduc- tion using syzygy-constrained symbolic reduction rules,

    Sid Smith and Mao Zeng, “Feynman integral reduc- tion using syzygy-constrained symbolic reduction rules,” JHEP01, 102 (2026), arXiv:2507.11140 [hep-th]

    arXiv 2026

  46. [46]

    Seed- less Reduction of Feynman Integrals,

    Leonardo de la Cruz and David A. Kosower, “Seed- less Reduction of Feynman Integrals,” (2026), arXiv:2602.22111 [hep-ph]

    arXiv 2026

  47. [47]

    Such an estimation is based on evaluation on 32-bit prime number with private linear solver

  48. [48]

    Kira 1.2 Re- lease Notes,

    Philipp Maierh¨ ofer and Johann Usovitsch, “Kira 1.2 Re- lease Notes,” (2018), arXiv:1812.01491 [hep-ph]

    Pith/arXiv arXiv 2018

  49. [49]

    Kira 3: integral reduction with efficient seeding and op- timized equation selection,

    Fabian Lange, Johann Usovitsch, and Zihao Wu, “Kira 3: integral reduction with efficient seeding and op- timized equation selection,” Comput. Phys. Commun. 322, 109999 (2026), arXiv:2505.20197 [hep-ph]

    arXiv 2026

  50. [50]

    For our interested part of physical amplitude,O(ϵ 2) is enough