Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Geometric Singularities of Feynman Integrals

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper establishes that the microlocal singularity structure of a Feynman integral is encoded by the constructible function $\chi(z)=n+1-\chi(Y\cap \varphi^{-1}(z))$, from which both the Landau singularities and master-integral counts…

desk verdict A compact, computable constructible-function invariant for Feynman singularities with a single bubble check; the load-bearing rank-one D-module step is deferred to a companion paper, so the right verdict is conditional, not acceptance or rejection. read the letter →

arxiv 2506.05042 v1 pith:QY7W6PQ4 submitted 2025-06-05 hep-th math-phmath.MP

Martin Helmer , Felix Tellander This is my paper
classification hep-thmath-phmath.MP MSC 14Q2032S6035A27
keywords FeynmanintegralsLandausingularitiesconstructiblefunctionscharacteristiccyclesWhitneystratificationsEulerobstructionsmasterD-modules
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes a new way to compute all singularities of a Feynman integral at once, by encoding them in a constructible function built from the geometry of the integrand's hypersurface. The main claim is that the function $\chi(z) = n+1-\chi(Y\cap \varphi^{-1}(z))$, formed from the Euler characteristics of the fibers of the integration map, carries the full microlocal singularity data, including the Landau locus and the number of master integrals at every point of the kinematic space. The construction is presented in the Lee-Pomeransky representation and is then shown to work in the Feynman and momentum representations as well, with the one-loop bubble giving the same singular locus in all three. If correct, this gives a direct topological route from the integrand to both singularity locations and solution-space dimensions, without writing down the annihilating differential system explicitly.

What carries the argument

The load-bearing object is the constructible function $\chi^R_M(z)=n+1-\chi(Y\cap\varphi^{-1}(z))$ of Eq. (9), where $Y\subset \mathbb{P}^n_x\times\mathbb{C}^m_z$ is the hypersurface defined by the homogenized integrand and $\varphi$ is the canonical projection onto the kinematic variables. A constructible function is a finite integer-linear combination of characteristic functions of subvarieties, and by Kashiwara's index theorem it corresponds to a characteristic cycle, the conical Lagrangian cycle tracking the cotangent directions where solutions of a D-module are not analytic. The algorithm uses Whitney stratifications and Euler obstructions: stratify $Y$, compute Euler obstructions via polar varieties using formula (2), solve for the multiplicities $m_\alpha$ in (6), and then eliminate conormal directions by the ideal construction (7) to obtain the singular locus $Z$. Because the Euler characteristics $\chi(Y\cap\varphi^{-1}(z))$ are constant on each stratum, all ingredients are computable.

What would settle it

Take a one-loop triangle with generic masses, compute the annihilating D-module of the integrand explicitly, and compare its characteristic variety with the conormal variety $\mathrm{Con}(Y)$ predicted by Eq. (9); any extra characteristic component would change the projected Landau locus and the multiplicities $\mu_\alpha$, producing a visible discrepancy at a kinematic point where the predicted and actual singularity loci differ.

Watch

Extended reading notes

Core claim

The central discovery is that the characteristic cycle of the system of partial differential equations annihilating a Feynman integral can be recovered from the hypersurface $Y\subset \mathbb{P}^n_x \times \mathbb{C}^m_z$ cut out by the homogenized integrand, and that integration over the loop variables is represented by the proper pushforward of the constructible function $1_{\mathbb{P}^n\times\mathbb{C}^m}-1_Y$ along the projection $\varphi$. The resulting function $\chi^R_M(z)$ in Eq. (9) is algorithmically computable: one Whitney-stratifies $Y$, computes Euler obstructions from polar-variety multiplicities, solves the linear system for the coefficients $m_\alpha$, then eliminates conormal variables to obtain the singular locus $Z$, which is the Landau locus of the integral. The constant term $\mu_0$ of the decomposition (10) equals the number of master integrals, i.e. the dimension of the solution space, and the same function gives the master-integral count on each singularity stratum. The paper demonstrates the method on the one-loop bubble in three representations and writes out the explicit constructible function and characteristic cycle in (11).

Load-bearing premise

The paper assumes, without proof here, that the D-module annihilating the integrand is spanned by the single function defining $Y$, so that its constructible function is exactly $1_{\mathbb{P}^n\times\mathbb{C}^m}-1_Y$; this rank-one identification is deferred to the companion paper, and the only evidence given here is the one-loop bubble.

Editorial extensions

If this is right

  • The Landau locus of a Feynman integral is obtained as the projection of a characteristic variety after eliminating the conormal variables, so it can be computed without knowing the annihilating differential equations.
  • The integer $\mu_0$ in the decomposition (10) is the number of master integrals, and the same constructible function gives the master-integral count on each stratum of the singular locus.
  • The construction is representation-independent: the Lee-Pomeransky, Feynman, and momentum representations of the one-loop bubble all yield the same singular locus, and the method is set up to run in any of them.
  • Since the algorithm relies only on Whitney stratifications and Euler obstructions, it can be implemented on a computer and applied to integrals beyond one loop.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural test is to run the same construction on the one-loop triangle with arbitrary masses and verify that the predicted Landau locus and master-integral counts match the known leading Landau equation and integration-by-parts results.
  • If the rank-one identification holds generally, the method suggests a new equality: the constructible function of the integral is entirely determined by the topology of the hypersurface $Y$, so integrals with the same fiber Euler characteristics over kinematic space would share the same singularity structure.
  • The Whitney-stratification input could also be used to study how the singularity locus changes under deformations of masses and external momenta, since strata with non-zero multiplicity are the ones that contribute to the characteristic cycle.
  • The framework appears to extend naturally to integrals with arbitrary parametric powers and to dimension-shift identities, because the same $Y$ controls the singularity structure while the exponents only affect the D-module that is being integrated.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes a new method to compute the microlocal singularity structure of Feynman integrals via constructible functions and characteristic cycles. Starting from a Lee-Pomeransky representation, the integrand is homogenized and its zero locus Y = V(x0...xn(Ux0+F)) is used to define a constructible function 1_{P^n x C^m} - 1_Y. The authors claim that this function encodes the characteristic cycle of the D-module annihilating the integrand, that its pushforward under the projection to kinematic space gives a function whose discontinuities identify the Landau locus, and that the value of this function away from the singular locus equals the number of master integrals. The method is illustrated on the one-loop bubble in three representations (Lee-Pomeransky, Feynman, momentum), with the momentum computation limited to the equal-mass case m1=m2. In each case the computed Landau locus is Z = V(p^2 m1^2 m2^2 ((m1^2+m2^2-p^2)^2 - 4m1^2 m2^2)), and the paper reports three master integrals generically, dropping by one on each irreducible component of Z.

Significance. If the central rank-one D-module identification is correct, the paper provides an elegant and computationally explicit route to Landau singularities and master-integral counts that works uniformly across representations. A genuine strength is that the bubble computation reproduces the classical Landau locus in three different representations, which is a non-circular external check, and the method is implemented in a publicly available Macaulay2 package. However, the significance is currently contingent: the key structural statement about the D-module is not proved in this manuscript and is deferred entirely to the companion paper [15], while the generality claim rests on a single one-loop example. The paper is best read as a concise announcement of a method whose rigorous foundations live elsewhere, and the present text does not yet supply enough evidence for the 'any Feynman integral' conclusion.

major comments (3)
  1. [Section IV, Eq. (6) and surrounding text] The load-bearing step is the assertion that the system of PDEs annihilating the homogenized Lee-Pomeransky integrand has rank one and characteristic cycle encoded by 1_{P^n x C^m} - 1_Y. The sentence 'The rank of this system is one since the function defining Y spans the solution space' is not an argument in this text; the proof is deferred to [15]. This step is not automatic: for h = x0...xn(Ux0+F), the annihilator of h^s can in general have characteristic components not contained in Con(Y), or with multiplicities different from those encoded by 1 - 1_Y, particularly when the parametric exponents produce resonances. Since Eq. (9), the computed Landau locus Z, and the master-integral count mu0 all depend on this identification, the manuscript needs to state the precise theorem from [15], its hypotheses, and a proof or a detailed reference, rather than a one-sentence assertion.
  2. [Section V and the claims in the abstract] The paper claims that the constructible function can be 'written down for any Feynman integral' and that the method gives 'the full microlocal description of singularities of Feynman integrals', but the only evidence is the one-loop bubble, and in momentum space the equal-mass restriction m1=m2 is imposed. A single example cannot establish the general claim, especially because the rank-one D-module statement is precisely where pathologies could arise. I ask the authors to either prove the general theorem from [15], restrict the claims to a clearly stated class of integrals where the theorem applies, or add further benchmarks (e.g., triangle, box, or sunset diagrams) that exercise nontrivial multi-scale and multi-loop features.
  3. [Equation (10) and the paragraph after it] The identification of mu0 with the number of master integrals, and more generally of chi_RM(z) with the dimension of the solution space at every z, is asserted without proof. Equation (10) defines mu0 as the value of a topological Euler characteristic on the complement of Z; the relation between this number and the dimension of the solution space of a D-module is a nontrivial statement that does not follow from the characteristic-cycle computation alone. The phrase 'number of master integrals on singularities' also needs a precise definition, since master integrals are normally defined as a basis of IBP-reduced integrals and their dimension at singular kinematic points is not obviously the same as the multiplicity mu_alpha appearing in the Euler-obstruction decomposition. This point is load-bearing for the physical interpretation of the paper's main result.
minor comments (4)
  1. [Abstract] The word 'preform' should be 'perform'.
  2. [Section V, Lee-Pomeransky paragraph] The name Källén appears with corrupted encoding ('K¨ all´ en'); fix the LaTeX/Unicode rendering.
  3. [Section IV] The phrase 'all of Y is the singular locus for the system of PDEs' is ambiguous; it should say that the singular support of the system is contained in or equal to Y, or specify precisely what is meant by 'singular locus' for a system with multivalued solutions.
  4. [Section V and code availability] The text mentions an implementation in the WhitneyStatifications Macaulay2 package and gives a link, but does not provide a version, commit hash, or the exact input files used for the three bubble computations; such reproducibility details would strengthen the paper's computational claims.

Circularity Check

1 steps flagged · score 4.0 of 10

The central hinge—rank-one identification of the annihilating D-module—is imported from the authors' companion paper [15] without proof, while the bubble example is an external consistency check rather than a derivation.

  1. self citation load bearing [Section IV, 'To get all singularities...' paragraph (paragraph following Eq. (5))]
    "The main results in the present paper are based on our recent article [15] and are given here without proof. ... As every factor in the integrand is raised to a parametric power, all of Y is the singular locus for the system of PDEs annihilating the integrand. The rank of this system is one since the function defining Y spans the solution space. So the constructible function that reproduces this system's characteristic variety is 1_{P^n×C^m} − 1_Y."

    The derivation chain runs: rank-one annihilating D-module for the homogenized integrand ⇒ characteristic cycle encoded by 1 − 1_Y ⇒ pushforward (8)-(9) giving χ_RM(z) ⇒ singular locus Z and master-integral count µ0 via (10). The rank-one identification is the only nonstandard input and is not proved in this paper; the paper explicitly says its main results are based on the authors' own article [15] and are given without proof. Consequently, Eqs. (9)-(11), and the claims that Z is the Landau locus and µ0 is the number of master integrals, reduce at their hinge to a same-author citation rather than to a derivation contained in the text. The one-loop bubble is a consistency check against the classical Landau locus, not a proof of the rank-one premise for arbitrary integrals.

full rationale

No fitted parameters appear and no prediction is constructed from its own output: the bubble's singular locus is compared with the classical Landau locus and the same Z is obtained in three representations, so the computational core is externally falsifiable. The paper does not rename a known result as a new one, and the uniqueness of the constructible function is cited to an external reference [26]. The only circularity-adjacent step is the rank-one D-module identification in Section IV, which is load-bearing and is deferred to the same authors' companion paper [15]; if that theorem failed, Eq. (9), the computed Z, and the master-integral count µ0 would all be unsupported. Because this is a cited theorem from a separate work rather than a definitional identification or a fitted parameter renamed as a prediction, the overall circularity is moderate: the central claim still has independent content in the explicit constructible-function computations, but its key premise rests on a same-author citation stated without proof. Score 4 reflects a load-bearing self-citation with independent checkable content, not full definitional circularity.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No numeric parameters are fitted; D and the ν_i are physical inputs from the integral, not free parameters chosen to match data. The axioms are a mix of standard D-module and stratification facts and the paper-specific rank-one identification that fixes the characteristic cycle. The rank-one identification is the main load-bearing, paper-specific assumption and it is inherited from the authors' companion paper rather than proved here. The 'any Feynman integral' breadth is an extrapolation from the one-loop bubble. No new particles, forces, or dimensions are introduced; the constructible function and characteristic cycles are established mathematical objects, not invented entities.

assumptions (5)
  • standard math Kashiwara's index theorem gives a canonical correspondence between constructible functions and conical Lagrangian cycles.
    Section III, Eq. (3), is the bridge used to replace a system of PDEs by a constructible function and to recover the characteristic cycle from Eq. (10).
  • ad hoc to paper The D-module annihilating the homogeneous Lee-Pomeransky integrand has rank one and characteristic variety determined by Y = V(x_0...x_n(Ux_0+F)).
    Section IV, after Eq. (5): 'The rank of this system is one since the function defining Y spans the solution space.' This is the key structural assumption; it is not proved here and is deferred to [15].
  • standard math Cheng-Wu theorem allows lifting the integration variables x to projective space without changing the singularity content.
    Section IV, first paragraph: 'To get all singularities we lift the integration variables x to projective space using the Cheng-Wu theorem.'
  • domain assumption Proper pushforward of constructible functions is computed by fiber Euler characteristics: φ_*(1_V)(z) = χ(φ^{-1}(z) ∩ V).
    Section IV, Eq. (8). The formula is stated and attributed to the companion paper; it underlies the central expression χ(z) = n+1 − χ(Y∩φ^{-1}(z)).
  • standard math Whitney stratifications exist and Euler obstructions are computable from polar multiplicities.
    Section II, Eq. (2), provides the algorithmic basis for the computational steps (i)-(vii).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Geometric Singularities of Feynman Integrals." pith.science (2026). https://pith.science/paper/QY7W6PQ4

@misc{pith2026250605042,
  author       = {Pith},
  title        = {Pith review of: Geometric Singularities of Feynman Integrals},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QY7W6PQ4}},
  note         = {Machine review of arXiv:2506.05042}
}
read the original abstract

We provide a new method to calculate the full microlocal description of singularities of Feynman integrals. This is done by associating a unique constructible function to the system of partial differential equations (PDEs) annihilating the integral and from this function the singularities can directly be read-off. This function can be constructed explicitly even if the system of PDEs is unknown and describes both the location of the singularities and the number of master integrals on them. Our framework is flexible enough to preform the calculation in any of the Lee-Pomeransky, Feynman, or momentum representations.

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Positive Integrands from Feynman Integrals in the Minkowski Regime

    hep-ph 2025-06 conditional novelty 7.0 of 10

    A method for converting Minkowski-regime Feynman parameter integrals into sums of real, positive integrands with complex prefactors, eliminating contour deformation and speeding up numerical evaluation.

Reference graph

Works this paper leans on

29 extracted references · 17 canonical work pages · cited by 1 Pith paper

  1. [15]

    Correia, M

    M. Correia, M. Giroux and S. Mizera, SOFIA: Singularities of Feynman Integrals Automatized , 2503.16601

    arXiv

  2. [1]

    − k2 hm2)((k0 + khp0)2 − (k1 + khp1)2) − k2 hm2) ⊂ P2 k × C3 while in three dimensions we have Y3D ⊂ P3 k × C4 and similarly for higher dimensions. There are 18 strata in the minimal Whitney stratification of Y2D and the singular locus is Z = V(m2(p2 0 − p2 1)(p2 0 − p2 1 − 4m2)) which is the same as in the other representations with m1 = m2. ACKNOWLEDGME...

    work page

  3. [2]

    Using (7) and eliminating the conormal variables we find the Landau singularities Z = V(p2m2 1m2 2((m2 1 + m2 2 − p2)2 − 4m2 1m2 2))

    There are 24 strata in the minimal Whit- ney stratification of this surface and the seven strata of dimension zero and one appear with multiplicity zero in the characteristic cycle. Using (7) and eliminating the conormal variables we find the Landau singularities Z = V(p2m2 1m2 2((m2 1 + m2 2 − p2)2 − 4m2 1m2 2)). The mini- mal Whitney stratification of t...

    work page

  4. [3]

    Weinzierl, Feynman Integrals

    S. Weinzierl, Feynman Integrals. 1, 2022, 10.1007/978-3-030-99558-4

    doi:10.1007/978-3-030-99558-4 2022

  5. [4]

    L. D. Landau, On analytic properties of vertex parts in quantum field theory , Nucl. Phys. 13 (1959) 181–192

    work page 1959

  6. [5]

    R. J. Eden, P. V. Landshoff, D. I. Olive and J. C. Polkinghorne, The analytic S-matrix . Cambridge Univ. Press, Cambridge, 1966

    work page 1966

  7. [6]

    Nakanishi, Graph theory and Feynman integrals

    N. Nakanishi, Graph theory and Feynman integrals. Mathematics and its applications: 11. Gordon and Breach, 1971

    work page 1971

  8. [7]

    F. C. S. Brown, On the periods of some Feynman integrals, 0910.0114

    arXiv

Show all 29 references
  1. [8]

    Pham, Singularities of integrals

    F. Pham, Singularities of integrals . Universitext. Springer, London; EDP Sciences, Les Ulis, 2011, 10.1007/978-0-85729-603-0

    2011 doi

  2. [9]

    Mizera and S

    S. Mizera and S. Telen, Landau discriminants, JHEP 08 (2022) 200, [ 2109.08036]

    2022 arXiv

  3. [10]

    R. P. Klausen, Kinematic singularities of Feynman integrals and principal A-determinants , JHEP 02 (2022) 004, [ 2109.07584]

    2022 arXiv

  4. [11]

    Caron-Huot, M

    S. Caron-Huot, M. Correia and M. Giroux, Recursive Landau Analysis, 2406.05241

    arXiv

  5. [12]

    Fevola, S

    C. Fevola, S. Mizera and S. Telen, Landau Singularities Revisited: Computational Algebraic Geometry for Feynman Integrals, Phys. Rev. Lett. 132 (2024) 101601, [2311.14669]

    2024 arXiv

  6. [13]

    Helmer, G

    M. Helmer, G. Papathanasiou and F. Tellander, Landau Singularities from Whitney Stratifications , 2402.14787

    arXiv

  7. [14]

    H. S. Hannesdottir, A. J. McLeod, M. D. Schwartz and C. Vergu, Applications of the Landau bootstrap , Phys. Rev. D 111 (2025) 085003, [ 2410.02424]

    2025 arXiv

  8. [16]

    Matsubara-Heo, Hypergeometric Discriminants, 2505.13163

    S.-J. Matsubara-Heo, Hypergeometric Discriminants, 2505.13163

    arXiv

  9. [17]

    Helmer and F

    M. Helmer and F. Tellander, Spectral Decomposition of Euler-Mellin Integrals, 2505.12458

    arXiv

  10. [18]

    Macaulay2, a software system for research in algebraic geometry

    D. R. Grayson and M. E. Stillman, “Macaulay2, a software system for research in algebraic geometry.” 5 Available at http://www.math.uiuc.edu/Macaulay2

  11. [19]

    Whitney, Tangents to an analytic variety , Ann

    H. Whitney, Tangents to an analytic variety , Ann. of Math. 81 (1965) 496–549

    1965

  12. [20]

    Helmer and V

    M. Helmer and V. Nanda, Conormal spaces and Whitney stratifications, Found. Comput. Math. 23 (2023) 1745–1780

    2023

  13. [21]

    Helmer, A

    M. Helmer, A. Leykin and V. Nanda, Effective Whitney Stratification of Real Algebraic varieties , 2307.05427v3

    arXiv

  14. [22]

    Teissier, Vari´ et´ es polaires

    B. Teissier, Vari´ et´ es polaires. II. Multiplicit´ es polaires, sections planes, et conditions de Whitney , in Algebraic geometry (La R´ abida, 1981), vol. 961 of Lecture Notes in Math. , pp. 314–491. Springer, Berlin, 1982. DOI

    1981

  15. [23]

    Eisenbud, Commutative algebra: with a view toward algebraic geometry, vol

    D. Eisenbud, Commutative algebra: with a view toward algebraic geometry, vol. 150. Springer Science & Business Media, 2013

    2013

  16. [24]

    Brian¸ con, P

    J. Brian¸ con, P. Maisonobe and M. Merle,Localisation de syst` emes diff´ erentiels, stratifications de Whitney et condition de Thom , Invent. Math. 117 (1994) 531–550

    1994

  17. [25]

    Kashiwara and P

    M. Kashiwara and P. Schapira, Sheaves on Manifolds . No. 292 in Grundlehren der math. Wiss. Springer-Verlag, 1990

    1990

  18. [26]

    Kashiwara, Index theorem for a maximally overdetermined system of linear differential equations , Proc

    M. Kashiwara, Index theorem for a maximally overdetermined system of linear differential equations , Proc. Japan Acad. 49 (1973) 803–804

    1973

  19. [27]

    R. N. Lee and A. A. Pomeransky, Critical points and number of master integrals , JHEP 11 (2013) 165, [1308.6676]

    2013 arXiv

  20. [28]

    R. D. MacPherson, Chern classes for singular algebraic varieties, Ann. of Math. 100 (1974) 423–432

    1974

  21. [29]

    Helmer, Algorithms to compute the topological Euler characteristic, Chern-Schwartz-MacPherson class and Segre class of projective varieties , J

    M. Helmer, Algorithms to compute the topological Euler characteristic, Chern-Schwartz-MacPherson class and Segre class of projective varieties , J. Symbolic Comput. 73 (2016) 120–138

    2016

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.