{"id":"22f8bad8-6626-4f72-a91b-0b7ffba4f355","arxiv_id":"2501.00101","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Any massless-internal Feynman integral is a limit of a conformal integral, letting conformal-family computations supply exact answers for many Feynman diagrams.","lead":"This paper shows that every momentum-space Feynman diagram with massless internal lines can be rewritten as a conformal integral, so one conformal integral can stand in for a whole family of Feynman diagrams. If correct, this gives a new way to compute multi-loop scattering amplitudes exactly in dimensional regularization.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The all-loop classification rests on an unproven multi-loop analogue of the imported one-loop identity (B11); a two-loop re-derivation and numerical check would settle it.","rationale":"The reader's weakest-assumption analysis points to (B11), and that is indeed the imported one-loop identity on which the framework depends. I agree that reliance on this unproven identity is a real gap. However, the more load-bearing gap for the paper's strongest claim is the multi-loop generalization: Eq. (3.3) and its L-loop counterpart are asserted by analogy with the contact case rather than derived, and the full classification claim ('computing the conformal integral in each family suffices to compute all the Feynman diagrams in the family exactly') is unsupported beyond one loop if this generalization fails. The on-shell reductions in Section V add a second layer of unexamined limit-taking. The paper has genuine independent support at one loop: the triangle formula (5.1), the on-shell box formula (5.4) matching Tarasov, and the three- and four-loop ladder values matching Baikov-Chetyrkin. These checks strengthen confidence in the one-loop machinery and in the ladder-family computations, but they do not test the multi-loop bridge that the central claim requires. Because the gap is substantial but the framework is plausible and partially checked, a conditional verdict is appropriate; this stress-test does not move the reader's verdict. The proposed test, an independent re-derivation and numerical verification of the simplest non-contact case, would directly settle whether the multi-loop generalization of (B11) holds.","tokens_in":13853,"tokens_out":19796,"duration_ms":202657,"concrete_test":"Take the simplest non-contact two-loop family (Fig. 2). Write the full position-space double-box integral with six fields, add one extra field at the non-conformal vertex so that both vertex constraints hold, perform the inversion and translation used to obtain (B11), and check that the result is exactly (3.3) with the stated prefactor and with the cross-ratio map (5.7) to (5.8). Then evaluate both sides of (3.3) numerically at one generic Euclidean kinematic point in d = 4 - 2 epsilon with epsilon = 0.1 using independent sector-decomposition code; if the prefactor or kinematics differ, the multi-loop generalization fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central mechanism is Eq. (2.4): the one-loop n-point integral equals, up to a prefactor, the conformal (n+1)-point integral, via the position-space identity (B11) imported from [10]. For L>=2, the paper needs the same statement for multi-loop diagrams: Eq. (3.3) claims the half-conformal 2-loop diagram equals the fully conformal diagram 'as we did for the contact diagram in Section B', and the L-loop generalization is then asserted without derivation. This is the load-bearing step of the classification claim: if (3.3) does not hold with the stated prefactor and cross-ratio map, computing the conformal member of a family does not determine the non-conformal members beyond one loop. The paper gives no derivation of this multi-loop identity, and the cited [10] is described as establishing contact-correlator identities, not this multi-loop statement. The on-shell reduction in Section V also exchanges limits and Mellin-Barnes contour integrals without justification; for example, the claim before (5.4) that 'only one of the five ratios is non-zero' is at best a statement about independent ratios and, as written, is not justified, and the conformal member can have compensating poles (Gamma(-2 epsilon) in (5.2)) that make limit interchange delicate. Only the kite formula (5.12) exercises the 2-loop bridge, and it has no independent check.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes that every momentum-space Feynman integral with massless internal lines and arbitrary (integer or non-integer) propagator powers can be obtained as a limit of a conformal integral in which each loop separately satisfies a conformal constraint on the sum of scaling dimensions. On the basis of this equivalence, the author organizes Feynman diagrams into 'conformal families' and claims that computing the conformal member of a family suffices to compute all members exactly. The paper gives one-loop relations (2.3)-(2.4), an L-loop generalization in Section III, and then computes several examples: ladder diagrams in four dimensions, the off-shell and on-shell triangle from the conformal box, the off-shell and on-shell box from the conformal pentagon, and the two-loop kite from the conformal double box. The on-shell box result is checked against Tarasov's result [21], and the ladder two-point values are checked against [20]. The central mechanism for multi-loop diagrams, however, is an identity (3.3) that is asserted rather than derived, and several Mellin-Barnes manipulations involve limit interchanges that are not justified.","tokens_in":14106,"tokens_out":6150,"duration_ms":65035,"significance":"If the claimed equivalence is correct, the paper provides a genuinely useful organizational principle: conformal integrals are often more symmetric than generic Feynman integrals, and known results for conformal integrals would automatically yield exact results for whole families of Feynman diagrams. The paper is honest about its checks: the one-loop on-shell box matches Tarasov [21], and the ladder two-point values at three and four loops match Baikov-Chetyrkin [20]. These are nontrivial and give real support to the one-loop framework and to the specific ladder computations. The weakness is that the load-bearing multi-loop identity (3.3) is imported from the author's prior work [10] in the contact case and then extended to internal vertices with no derivation in this paper; the L-loop classification therefore rests on an unproven step. The two-loop kite formula (5.12), which is the only new two-loop result, has no independent check. For these reasons the significance is real but conditional on closing the derivation gap for (3.3) and its L-loop generalization.","major_comments":[{"comment":"Equation (3.3) is the load-bearing step for every multi-loop claim in the paper: it asserts that a half-conformal two-loop diagram equals the fully conformal two-loop diagram up to a prefactor. The text says 'As we did for the contact diagram in §B' and then describes a position-space inversion and translation, but this is not a derivation. The position-space identity (B11) concerns n-point contact correlators, not correlators with internal vertices such as the half-conformal double box. The cited reference [10] is described as establishing contact-correlator identities, so it does not by itself justify (3.3). Because the classification claim 'Computing the conformal integral in each family suffices to compute all the Feynman diagrams in the family exactly' depends directly on (3.3), the paper should either provide a derivation of (3.3) or give an independent check, for example by evaluating both sides of (3.3) numerically for a simple two-loop topology at generic kinematics.","section":"§III, Eq. (3.3)"},{"comment":"The generalization from two loops to L loops is stated in one paragraph: 'The picture at two loops gets naturally generalized to the L-loop case.' No induction or combinatorial proof is given that repeatedly contracting lines of a conformal diagram, one loop at a time, preserves the per-loop conformal constraint, produces the stated prefactors, and generates all members of the family. The subsequent statement that 'Computing this conformal member in each family suffices to determine every other diagram in the family' is therefore not established for L≥3. A formal inductive statement of the relation between an (L−1)-loop diagram and its L-loop conformal completion, with the effect of each contraction on the scaling-dimension sums, is needed to make the classification claim rigorous.","section":"§III, L-loop classification"},{"comment":"The on-shell reduction of the kite integral collapses the q integral and four of the a_i integrations and then performs the p integral to arrive at (5.12). This procedure exchanges an on-shell kinematic limit with multi-fold Mellin-Barnes contour integrals. No justification is given for the interchange, and the presence of poles such as the Γ(−2ε) factor in (5.2) shows that the conformal member can be singular in the limit, making the operation delicate. Since (5.12) is the only new two-loop result and has no independent numerical or analytic check, the derivation is not yet convincing. I ask the author to justify the limit interchange or to provide a check of (5.12), for example by evaluating both sides numerically at generic d or ε and comparing a few terms in an ε expansion.","section":"§V.B, Eqs. (5.10)-(5.12)"},{"comment":"The sentence 'In the limit that all the external momenta are on-shell, we find that only one of the five ratios is non-zero' is not justified and, as written, is unclear. The five ratios in (5.3) are not independent, and the statement that only one is non-zero is stronger than 'only one independent ratio remains.' The derivation of the on-shell box formula (5.4) depends on this reduction. The final result matches [21], which suggests the intended claim is correct under a specific convention, but the paper should state the convention explicitly and prove which ratios vanish and which remain, or else revise the sentence to describe the number of independent ratios.","section":"§V.A, before Eq. (5.4)"}],"minor_comments":[{"comment":"The formula (4.7) is said to give the 'arbitrary loop contribution' to the massless two-point function, but for L=1 it evaluates to 2ζ(1), which is not finite. Please specify that the formula holds for L≥2, or clarify the range of validity.","section":"§IV, Eq. (4.7)"},{"comment":"The notation 'sq5' appears in (5.5) and (5.10) and is not defined; it appears to be a typo for s_5^q. Please define all Mellin variables and exponents before use.","section":"§V.B, Eqs. (5.5) and (5.10)"},{"comment":"The arguments on the left-hand sides of (2.4) and (3.3) are written as sets of ratios such as {Pij/(Pi(n+1)Pj(n+1))}^{(n)}, but the notation for evaluating a function at rescaled invariants is not formally defined. Please define this notation in Appendix A.","section":"§II, Eqs. (2.4) and (3.3)"},{"comment":"The paper uses the term 'conformal integral' in a technical sense (per-loop conformal constraint), but it is not given a formal definition in the main text. A short definition near the beginning of Section II would improve readability.","section":"§II"},{"comment":"The identity (B11) is central to the paper, but it is quoted from [10] without a derivation or a precise statement of the theorem in [10] that implies it. Since [10] is the author's own prior work, including a self-contained statement (or an appendix reproducing the key steps) would make the present paper easier to evaluate.","section":"Appendix B, Eq. (B11)"}],"recommendation":"major_revision","confidential_remarks":"The editor may wish to verify the status of [10] and whether the multi-loop identity (3.3) has been independently checked. The present paper's novelty rests on extending a contact-correlator identity from the author's own earlier preprint to internal vertices; the on-shell box and ladder checks are encouraging, but the derivation gap at two and higher loops is substantial. A revision that provides a derivation or numerical check of (3.3) and the L-loop generalization would make the paper's central claim convincing."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper offers a genuinely interesting idea: organize massless Feynman integrals into conformal families, where computing the conformal member gives every other member by taking limits. It also contains several exact results — the off-shell triangle (5.1), on-shell box (5.4), two-loop kite (5.12), and the all-loop two-point ladder (4.7) — and the on-shell box matches Tarasov [21], while the low-loop ladder values match Baikov–Chetyrkin [20]. Those external checks are real and give the paper credibility.\n\nThe soft spots are in the load-bearing structure. The one-loop identity (2.4) is imported from the author's earlier work [10] without derivation; that is acceptable if the reader trusts the prior paper. But the multi-loop generalization, equation (3.3), is asserted with \"as we did for the contact diagram,\" and Appendix B only derives the contact (one-loop) case. Without (3.3), the central claim — that computing one conformal integral per family suffices for all L-loop diagrams — does not follow beyond one loop. The paper gives no derivation and no independent numerical check for this two-loop bridge. The on-shell reduction in Section V also exchanges limits and Mellin–Barnes contours without comment; the statement before (5.4) that \"only one of the five ratios is non-zero\" is, as written, not justified. The kite formula (5.12) is the only two-loop application of the bridge and has no independent check.\n\nThese are genuine gaps, though they look like gaps in proof and exposition rather than signs the results are wrong. The external matches suggest the idea has content. The paper is most useful for practitioners who work with Feynman integrals or conformal integrals and want exact formulas or a new organizational principle; it deserves a serious referee, but the referee should insist on a derivation or independent numerical verification of the multi-loop identity before the classification claim is accepted. I would send it to peer review, with major revision required.","headline":"A plausible organizing principle and some exact formulas that pass external checks, but the multi-loop bridge is asserted, not derived, so the classification claim is conditional.","tokens_in":14657,"tokens_out":2957,"would_cite":false,"duration_ms":31247,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Every momentum-space Feynman integral with massless internal lines is an $L$-fold limit of a conformal integral, so one conformal computation per family yields all of its diagrams exactly.","keywords":["Feynman integrals","conformal integrals","conformal families","ladder diagrams","Mellin-Barnes integrals","dimensional regularization","position-space correlators","massless propagators"],"falsifier":"Take a one-loop triangle in $d = 4-2\\epsilon$ with generic off-shell momenta and propagator powers that do not sum to $d$, and evaluate both sides of (2.4) to a fixed order in $\\epsilon$: the left side by direct Schwinger-parameter integration and the right side from the conformal box Mellin-Barnes representation times the stated prefactor. Any disagreement at that order falsifies the foundational identity and with it the $L$-loop construction; a match tests the mechanism at its base.","tokens_in":13619,"feed_emoji":"⚛️","tokens_out":9873,"duration_ms":86863,"temperature":0.7,"pith_summary":"The paper claims that every momentum-space Feynman integral with massless internal lines, in any spacetime dimension and with arbitrary propagator powers, is an $L$-fold limit of a conformal integral: add one extra internal line to each loop so that the propagator powers in that loop sum to the spacetime dimension $d$, evaluate the conformal member, then collapse the extra lines. If the claim is right, Feynman integrals organize into conformal families, and computing the single conformal integral in a family fixes every other diagram in that family exactly, including off-shell and on-shell cases, with dimensionally regularized answers to all orders in the regulator. The paper demonstrates this by deriving exact answers for the triangle, box, and two-loop kite integrals from known conformal box, pentagon, and double-box results, and by converting known four-point ladder solutions into new three- and two-point ladder answers. A careful reader would care because the result reduces a large class of Feynman-integral computations to a smaller set of better-studied conformal integrals with a systematic classification.","feed_headline":"One conformal integral computes every diagram in its family","feed_subtitle":"Massless Feynman diagrams reduce to conformal integrals, giving exact answers for whole families at once.","key_machinery":"The load-bearing object is the one-loop conformal-reconstruction identity (2.4), imported from the position-space result (B11): $M_{\\Delta_1,\\dots,\\Delta_n}(\\{P_{ij}/(P_{i(n+1)}P_{j(n+1)})\\}_n) = \\frac{\\prod_{i=1}^n P_{i(n+1)}^{-\\Delta_i}}{\\Gamma(\\Delta_{n+1})} M^{\\mathrm{cft}}_{\\Delta_1,\\dots,\\Delta_{n+1}}(\\{P_{ij}\\}_{n+1})$, with the conformal constraint $\\sum_i \\Delta_i = d$ on each loop. Iterating this identity on every loop builds the $L$-fold limit (3.1), and the Mellin-Barnes representations of contact correlators supply the actual evaluations. This identity is what turns the statement 'compute the conformal member' into 'compute every member of the family,' including prefactors.","core_discovery":"The central discovery is that a position-space identity from the author's earlier work — an $n$-point massless contact correlator equals, up to a prefactor, an $(n+1)$-point conformal contact correlator for arbitrary kinematics — transports to momentum space, where each loop momentum plays the role of a vertex position. At one loop, relation (2.4) equates the general $n$-point loop integral $M_{\\Delta_1,\\dots,\\Delta_n}$ to the conformal $(n+1)$-point integral $M^{\\mathrm{cft}}_{\\Delta_1,\\dots,\\Delta_{n+1}}$ whose scaling dimensions satisfy $\\sum_i \\Delta_i = d$, up to simple factors of momentum invariants. At $L$ loops, adding one extra internal line in each loop produces an $L$-fold limit (3.1), and collapsing lines of the conformal member generates the whole conformal family; knowing the conformal member fixes every other member. The concrete payoffs include the general off-shell triangle as an Appell $F_4$ combination, the off-shell and on-shell box and the two-loop off-shell and on-shell kite as Mellin-Barnes integrals, and the new $L$-loop two-point ladder value $M^{\\mathrm{2\\text{-}point}}_{L\\text{-ladder}} = \\frac{(2L)!}{(L!)^2}\\,\\zeta(2L-1)$ obtained from the known four-point ladder family.","pith_inferences":["If the foundational identity holds, the conformal-family organization suggests a practical workflow the paper does not spell out: tabulate conformal members once, then generate all diagrams in the family by limits, rather than computing each diagram independently.","The same mechanism may extend to massive internal lines, which the paper says it will treat elsewhere; a direct test is whether the extra-line insertion can be chosen to preserve the massive propagator structure while keeping each loop's conformal constraint.","The equivalence could let techniques developed for conformal correlation functions produce Feynman-integral master integrals, reversing the usual direction in which Feynman integrals feed into conformal data.","The paper notes that the number of Mellin integrals needed for an $n$-point contact conformal diagram equals the number of cross ratios; if so, the choice of which conformal completion to use affects computational complexity, and finding the minimal completion is a testable optimization problem."],"forward_implications":["Any $L$-loop Feynman integral with massless internal lines can be obtained as an $L$-fold limit of a conformal integral, so every such diagram sits in a conformal family labelled by one conformal diagram.","Computing one conformal integral per family yields exact answers for all diagrams in the family, for general dimension, general propagator powers, and both off-shell and on-shell kinematics, including dimensionally regularized results to all orders in the regulator.","Known exact four-point ladder results in four dimensions give the corresponding three-point and two-point ladder answers; in particular, the $L$-loop massless two-point ladder is $\\frac{(2L)!}{(L!)^2}\\zeta(2L-1)$ and is finite for every $L$.","The general off-shell triangle in arbitrary dimension is a closed combination of Appell $F_4$ functions, while the off-shell and on-shell box and two-loop kite have Mellin-Barnes representations that can be expanded to any order in $\\epsilon$.","The construction applies to planar and non-planar diagrams alike, so it extends the dual-conformal picture familiar from planar supersymmetric gauge theory to general Feynman integrals."],"supporting_citations":[{"why":"It supplies the method for evaluating massless conformal contact diagrams in position space, which is used to obtain the conformal box that yields the triangle answer.","marker":"[3]"},{"why":"It shows that the same four-point conformal integral appears in conformal-block computations, providing a known evaluation used in the box derivation.","marker":"[7]"},{"why":"It provides the position-space identity (B11) from which the momentum-space conformal limits (2.3), (2.4), and (3.1) are translated, and it is the load-bearing premise.","marker":"[10]"},{"why":"It establishes the low-loop ladder equivalences between triangle and box diagrams and between two-loop ladder diagrams that the conformal-family relations generalize.","marker":"[16]"},{"why":"It gives the exact four-point ladder answers for arbitrary loop order, from which the paper derives the new three- and two-point ladder results.","marker":"[17]"},{"why":"It supplies the known three- and four-loop two-point propagator values that confirm the predicted two-point ladder formula (4.7).","marker":"[20]"},{"why":"It provides the on-shell box with arbitrary propagator powers against which the paper's expression (5.4) is checked.","marker":"[21]"}],"fun_headline_variants":["Massless Feynman diagrams are conformal integrals","One conformal integral yields every Feynman diagram in its family","Loop integrals become conformal integrals for exact results","Conformal families solve Feynman diagrams exactly","Feynman diagrams tamed by conformal integral mapping"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole construction rests on one imported premise from the author's earlier work: that the value of a certain $n$-point correlation of massless fields, up to a simple prefactor, equals an $(n+1)$-point conformal version of the same quantity, even when the points are in generic positions. If that equality fails, or breaks when translated from position space to momentum space, the claimed limits do not follow.","fun_headline_variants_meta":{"raw":{"variants":["Massless Feynman diagrams are conformal integrals","One conformal integral yields every Feynman diagram in its family","Loop integrals become conformal integrals for exact results","Conformal families solve Feynman diagrams exactly","Feynman diagrams tamed by conformal integral mapping"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000553,"raw_usage":{"total_tokens":2627,"prompt_tokens":927,"completion_tokens":1700,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":543,"completion_tokens_details":{"reasoning_tokens":1622}},"tokens_in":543,"tokens_out":1700,"duration_ms":11706,"temperature":1.0,"reasoning_tokens":1622,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:59:05.175917+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a one-loop triangle in $d = 4-2\\epsilon$ with generic off-shell momenta and propagator powers that do not sum to $d$, and evaluate both sides of (2.4) to a fixed order in $\\epsilon$: the left side by direct Schwinger-parameter integration and the right side from the conformal box Mellin-Barnes representation times the stated prefactor. Any disagreement at that order falsifies the foundational identity and with it the $L$-loop construction; a match tests the mechanism at its base.","supporting_citations":[{"cited_title":"∆2, ∆4, ˜∆01234 − ˜∆0234 + 1, − ˜∆0124 + 1; −r3 # + Γ(∆1 + 2∆0234)Γ(− ˜∆0234)Γ( ˜∆023)Γ (∆3,1) Γ ˜∆034 r ˜∆0234 3F2","cited_arxiv_id":null,"evidence_quote":"It supplies the method for evaluating massless conformal contact diagrams in position space, which is used to obtain the conformal box that yields the triangle answer."},{"cited_title":"Ferrara, R","cited_arxiv_id":null,"evidence_quote":"It shows that the same four-point conformal integral appears in conformal-block computations, providing a known evaluation used in the box derivation."},{"cited_title":"Chetyrkin and F.V","cited_arxiv_id":null,"evidence_quote":"It establishes the low-loop ladder equivalences between triangle and box diagrams and between two-loop ladder diagrams that the conformal-family relations generalize."},{"cited_title":"Usyukina and A.I","cited_arxiv_id":null,"evidence_quote":"It gives the exact four-point ladder answers for arbitrary loop order, from which the paper derives the new three- and two-point ladder results."},{"cited_title":"Bierenbaum, The massless two-loop two-point function and zeta functions in counterterms of Feynman diagrams, other thesis, 2, 2005","cited_arxiv_id":null,"evidence_quote":"It supplies the known three- and four-loop two-point propagator values that confirm the predicted two-point ladder formula (4.7)."}],"review_version":1}