{"id":"a4bb808d-b248-41f1-bd67-b634e1b874d4","arxiv_id":"2412.10864","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A new 'quantum worldline' representation makes the classical limit of scattering amplitudes exactly match worldline field theory integrands at two loops.","lead":"This paper shows that two different calculation methods for black hole collisions, scattering amplitudes and worldline field theory, give the same mathematical integrands once a new bookkeeping trick is used. The result lets techniques from one method flow into the other, which could speed up high-precision gravitational wave predictions.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed exact integrand equality is asserted only after dropping zero-measured cuts and purely imaginary Fourier pieces by hand, and the all-order worldline identity (210) is explicitly a hypothesis, so the general equivalence is not yet established.","rationale":"The reader's weakest-assumption analysis correctly identifies zero-measured cuts as the main soft spot. My reading agrees but sharpens it in two ways. First, the paper itself flags eq. (210) as a hypothesis, so the general all-order worldline-form claim is not a theorem even within the paper's own framing; the demonstrated content is a set of two-loop examples plus automated checks at higher loop order for ladders. Second, the two-loop 'exact' comparison is not pointwise exact even in the showcased case: the text explicitly excludes cIm3 and F^Im as terms that should vanish after Fourier transformation. This means the displayed equality in eq. (267) holds modulo a class of terms that is dropped by hand. Neither issue is fatal if the excluded terms indeed integrate to zero or cancel, and the paper gives plausible physical reasons for expecting that. But the central claim 'exactly coincides' is stronger than what is proved. These are addressable concerns rather than demonstrated failures, so the appropriate verdict remains CONDITIONAL, as the reader already concluded; I would not move it to ACCEPT or REJECT without the proposed explicit verification.","tokens_in":70632,"tokens_out":5417,"duration_ms":57110,"concrete_test":"Recompute the two-loop scalar-QED ladder comparison of Section VII.B.3 without applying the Section VI.A rule: keep every term generated by GF(q)* = GF(q) + iδ(q^2), retain the cIm3-type terms of eq. (269) and F^Im of eq. (232), and Fourier-transform the full KMOC-minus-WQFT difference with e^{i(q1+q2+q3)·b}, regulating any pinched delta functions with the iε-preserving prescription of eqs. (198)–(199). If the difference has a nonvanishing real classical impulse at O(λ^2), the hand-dropped terms are load-bearing and eq. (267) is not exact; if it vanishes identically, the local equality is vindicated for the main example.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim—that the finite KMOC integrand exactly coincides with WQFT at the integrand level—rests on two truncations that are asserted rather than derived. First, Section VI.A defines zero-measured cuts and states 'we will simply set these contributions to zero by hand whenever they appear'; this underpins the replacement GF(q)* = GF(q) + iδ(q^2) with GF(q), the local cancellation of super-leading terms, and the appearance of retarded propagators. The pinched-surface example in eqs. (195)–(197) shows why these cuts are ill-defined, but it does not prove that they never contribute at classical order for the specific diagrams compared in Sections III and VII. Second, the all-order statement in eq. (210) is introduced with 'we hypothesise that...', and the two-loop comparisons in Section VII.B.3 explicitly exclude the terms cIm3 in eq. (269) and F^Im in eq. (232) because they 'give imaginary contribution upon being Fourier integrated.' Thus eq. (267), although displayed as an equality, is not a pointwise term-by-term identity unless those excluded terms are shown to vanish after Fourier integration and the zero-measured-cut contributions are shown to be harmless. These are correctness risks, not internal inconsistencies: the worked examples are substantial and the claimed equivalence may well be physical, but the advertised 'exact coincidence' is conditional on unproved vanishing statements.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a systematic diagrammatic method for taking the classical (hbar to 0) limit of observables computed in the Kosower-Maybee-O'Connell (KMOC) formalism, with the goal of making the cancellation of super-leading divergences manifest at the integrand level beyond one loop. The method rewrites matter propagators and delta functions in Schwinger proper-time form, thereby representing off-shell currents as \"quantum worldlines\"; virtual and real contributions are then combined in a common parameter space. The authors claim that the resulting finite classical integrand takes the same form as, and in explicit examples exactly coincides with, the integrand obtained from worldline quantum field theory (WQFT), with retarded propagator prescriptions emerging from the combination of diagrams. The main worked examples are the two-loop impulse in a scalar model and in scalar electrodynamics, together with one-connected-component examples to all orders and automated checks of three- and four-loop ladders. The appendices provide the WQFT Feynman rules and the detailed integrand-level comparison.","tokens_in":70924,"tokens_out":5774,"duration_ms":56362,"significance":"If the central claim is fully established, this is a strong and useful result: it connects two active approaches to classical gravitational scattering at the integrand level, goes substantially beyond the earlier one-loop comparisons, and provides a systematic handle on superclassical terms that plague amplitude-based calculations. The paper has notable strengths: the two-loop examples are worked out in great detail, the WQFT comparison uses independent Feynman rules as an external benchmark, no parameters are fitted, and the forest/arborescence formulation is elegant and likely to be reusable. The main weakness is that the advertised \"exact coincidence\" is conditional on several asserted vanishings: zero-measured cuts are set to zero by hand, the all-order identity in eq. (210) is explicitly a hypothesis, and specific terms in the two-loop comparison are discarded with the statement that they integrate to imaginary quantities. These are correctness risks rather than internal inconsistencies, and they are concentrated in identifiable places, so the work is suitable for major revision.","major_comments":[{"comment":"The procedure replaces complex-conjugate Feynman propagators by ordinary ones using G_F(q)^* = G_F(q) + i delta(q^2) and discards the resulting cut terms, citing zero-measured cuts. The illustrative pinched-surface integral in eqs. (195)-(197) demonstrates that such cuts are ill-defined, but it does not prove that they vanish after integration in the two-loop examples of Sections III and VII. Since this replacement is used to obtain the retarded-propagator form and the local cancellation of super-leading terms, the advertised exact integrand-level coincidence, eq. (267), is conditional on an unproved vanishing statement. I ask the authors to either prove that zero-measured cuts cannot contribute at the classical order for the diagrams under comparison, or state this as an explicit assumption and adjust the claims accordingly.","section":"§VI.A and §III.B"},{"comment":"Equation (210), which gives the all-order worldline form of the classical integrand, is introduced with the words \"we hypothesise that\". The all-order statements in the abstract and in Section VIII therefore go beyond what is actually proven. The two-loop comparisons are explicit and convincing, but the general equivalence between the KMOC and WQFT formalisms is not established unless eq. (210) is derived or its domain of validity is specified. Please either supply a proof or a precise set of sufficient conditions for eq. (210), or clearly separate the all-order conjecture from the proven examples.","section":"§V.B, eq. (210)"},{"comment":"The displayed equality in eq. (267) is not a pointwise term-by-term identity: the text states that F^{Im}_3 in eq. (232) \"give[s] imaginary contribution upon being Fourier integrated\" and that c^{Im}_3 in eq. (269) is excluded because it \"should vanish after Fourier transformation\". No calculation is shown establishing these vanishings. Because these terms are precisely the ones that fail to match the worldline form, the central claim of exact integrand coincidence is conditional. I request an explicit demonstration, for example using the delta-function constraints and parity under q_i -> -q_i, that these contributions vanish after Fourier integration, or a reformulation of the claim as an equivalence only after loop integration.","section":"§VII.B.1 and §VII.B.3, eqs. (232), (267), (269)"}],"minor_comments":[{"comment":"The conclusion that all unlisted self-attachment contributions to the two-loop mushroom are homogeneous and vanish is illustrated with two examples and then asserted for the remaining terms; a brief general homogeneity argument would make this step easier to verify.","section":"§VII.B.2"},{"comment":"The reduction of the integrated impulse to the q3 component via the factor 1/2 is non-obvious; an explanation of which symmetries justify eq. (230) would improve readability.","section":"§VII.B.3, eq. (230)"},{"comment":"The symbol sigma is used both for the scalar product u1.u2 in the comparison section and for the sign factors sigma^i_e in eqs. (129) and (204); these two uses should be distinguished.","section":"§VII.B.3 and §V.B"},{"comment":"The paper says the three- and four-loop ladder checks were generated automatically, but no code or auxiliary material is provided; stating the availability of the generation script, or at least describing the algorithm in more detail, would strengthen reproducibility.","section":"Appendix B"}],"recommendation":"major_revision","confidential_remarks":"I support publication after revision. The two-loop computations are impressive and the WQFT benchmark is an independent check, which gives me confidence in the main physics. The main risk is that the abstract and Section VII.B.3 claim an exact integrand identity that is actually established only up to asserted vanishings of zero-measured cuts and purely imaginary Fourier pieces. If those vanishings are proved or the claims are appropriately softened, I would be happy to accept. I see no concerns about novelty or citation practice; the related work in Refs. [57,58,59,60,84,85] is acknowledged."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a real contribution, not a stunt. The new thing is showing, at integrand level and with iε kept, that KMOC observables can be rewritten in a worldline-like form with super-leading divergences cancelling locally, and that the resulting integrand matches WQFT term by term at two loops. The quantum-worldline machinery via Schwinger parameters and the forest combinatorics is genuinely useful and clearly explained. The two-loop scalar QED comparison, including the probe and full σ coefficients, is worked out in serious detail, and the WQFT side is computed with independent Feynman rules. The higher-loop ladder tests (3 and 4 loops) for local cancellation add real support. The citation pattern is fair: they correctly position themselves against the earlier loop-integrated comparison [57] and the iε-blind off-shell current comparison [58].\n\nThe soft spots are exactly where your report puts them. The \"exact coincidence\" claim is conditional. Section VI.A defines zero-measured cuts and simply declares them zero; the pinched-surface example shows they are ill-defined but does not prove they vanish for the specific diagrams compared. Eq. (210), the all-order worldline identity, is introduced with \"we hypothesise.\" And in the two-loop comparison, terms F_3^Im (eq. 232) and c_3^Im (eq. 269) are dropped because they are said to give imaginary contributions upon Fourier integration, with the detailed proof not fully shown. So eq. (267) is more a conditional identity than the pointwise term-by-term equality the abstract suggests. To their credit, the authors are explicit about some of these limitations, but the gap between \"exactly coincides\" and \"coincides modulo asserted vanishings\" is real.\n\nI do not see an internal contradiction that invalidates the two-loop results. The derivations are painstaking and the two-loop steps are checkable from the appendices. The issue is the strength of the general statement, and the need for a rigorous treatment of the discarded pieces. That is addressable in revision.\n\nWho benefits: anyone working on the KMOC/WQFT interface, post-Minkowskian gravity, or classical limits of amplitudes. It deserves a serious referee. My recommendation: send it to peer review, and ask the authors to (a) prove or sharply state the zero-measured-cut vanishing condition, (b) justify the dropped imaginary terms in detail, and (c) clarify the status of eq. (210) as a conjecture rather than a theorem.","headline":"A substantial, honest attempt at an integrand-level KMOC/WQFT equivalence with detailed two-loop evidence, but the advertised exact coincidence is conditional on unproved vanishing statements and one explicit hypothesis.","tokens_in":71379,"tokens_out":2796,"would_cite":true,"duration_ms":27179,"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":"The paper demonstrates that, after rewriting quantum amplitudes as Schwinger-parametrized worldlines, the classical KMOC impulse integrand and the worldline quantum field theory integrand coincide term by term once super-leading quantum…","keywords":["KMOC formalism","classical limit","super-leading divergences","quantum worldline","worldline quantum field theory","post-Minkowskian expansion","scalar QED","retarded propagator"],"falsifier":"Evaluate one of the diagrams the paper sets to zero by hand, such as the degenerate vertex process shown around eq. (39), with a regulator that keeps the pinched surface finite; a non-vanishing contribution at the classical order $O(\\lambda^2)$ to the impulse would break the claimed exact correspondence.","tokens_in":70396,"feed_emoji":"⚛️","tokens_out":6473,"duration_ms":63941,"temperature":0.7,"pith_summary":"This paper tries to establish that two very different calculational frameworks for classical two-body scattering—scattering amplitudes through the KMOC formalism, and worldline quantum field theory (WQFT)—produce exactly the same integrand, not merely the same final observable. The vehicle is a rewriting of massive-particle propagators as worldline-like objects, called quantum worldlines, using Schwinger parameters, which lets virtual and real (cut) contributions be combined before the classical limit is taken. In the combined expression, the unphysical terms that diverge as $\\hbar \\to 0$ cancel point by point in loop-momentum space, and the retarded propagators that encode classical causality emerge rather than being inserted by hand. The authors demonstrate this at integrand level for the two-loop impulse in a scalar model and in scalar QED, with automated checks at three and four loops.","feed_headline":"Two-loop classical impulses match exactly in two QFT formalisms","feed_subtitle":"Rewriting quantum amplitudes as worldlines makes KMOC and WQFT integrands coincide term by term.","key_machinery":"The central object is the quantum worldline: an off-shell massive current rewritten in Schwinger-parameter space, where each matter propagator becomes a proper-time integral and each on-shell delta function a Fourier integral over a vertex time. The classical expansion of this object is organized by directed forests on the ordered vertex set, and triangulation identities collapse sums of Heaviside theta functions into single orderings. This machinery places virtual and cut contributions on exactly the same integration measure, makes the cancellation of super-leading terms algebraic, and reveals retarded time ordering as the classical causality flow.","core_discovery":"The central claim is that the finite classical KMOC integrand exactly coincides with the WQFT integrand after a systematic Schwinger-parameter manipulation, at the level of loop integrands rather than integrated observables. For the two-loop impulse in scalar QED and the scalar model, the paper obtains identities such as $\\frac{1}{2}(F_{WL}+F_{WL}|_{q_2\\leftrightarrow q_3}) = \\frac{q_1^\\mu}{2} F_1^{KMOC}$, with the retarded propagator prescriptions of WQFT reproduced rather than assumed. Super-leading divergences—terms in individual diagrams that diverge in the $\\hbar\\to 0$ limit—are cancelled locally by pairing virtual and real contributions, after a class of degenerate phase-space configurations, the zero-measured cuts, is set to zero by hand.","pith_inferences":["One implication the authors leave implicit is that WQFT diagram enumeration could serve as a template for predicting which KMOC diagrams cancel, potentially reducing amplitude-side calculations to tree-like worldline graphs at every order.","The same democratic combination of virtual and real contributions in Schwinger space may apply to other observables such as radiated momentum and waveform, since the mechanism is local in loop-momentum space and not specific to the impulse.","A natural testable extension is to push the term-by-term comparison to the first order where radiation reaction enters; at that order the imaginary pieces the paper discards, such as $c^{Im}_3$, may become physical rather than vanishing after Fourier transformation.","If the correspondence generalizes to gravity, the paper would imply that classical gravitational observables can be computed either by amplitude methods or worldline methods interchangeably, with the same integrand and the same causal propagator structure."],"forward_implications":["If the exact integrand-level equivalence holds beyond two loops, techniques developed for amplitude integrands, such as generalized unitarity and double-copy constructions, can be imported into worldline computations without rederiving the classical limit.","The retarded propagator prescription in classical worldline calculations no longer needs to be added by hand: it follows from the KMOC starting point after the local cancellation of super-leading terms.","Because the classical limit is taken before loop integration, the method avoids intermediate super-leading integrals, which should simplify higher-order post-Minkowskian computations.","The construction applies to the potential region and to conservative contributions; radiative effects and long-range Coulomb-like tails, which live on the discarded zero-measured cuts, lie outside the current setup.","The local cancellation has been verified algorithmically for three- and four-loop ladders in scalar QED, suggesting that the mechanism is not an accident of low loop order."],"supporting_citations":[{"why":"Defines the KMOC formalism this paper starts from, including the soft classical limit of quantum observables and the original one-loop local cancellation.","marker":"[14]"},{"why":"Introduces worldline quantum field theory, the formalism whose two-loop integrand is the target of the comparison.","marker":"[26]"},{"why":"Introduces the post-Minkowskian effective field theory worldline approach and the retarded-propagator causality prescription used in the comparison.","marker":"[23]"},{"why":"Establishes the retarded-propagator rules in WQFT that the KMOC-derived integrand is shown to reproduce.","marker":"[31]"},{"why":"Provides a previous diagrammatic comparison of KMOC and worldline EFT at one loop after integration, which this paper extends to integrand level.","marker":"[57]"},{"why":"Demonstrates equivalence between tree-level off-shell currents and worldline QFT; the present paper refines this by preserving the iε prescriptions.","marker":"[58]"},{"why":"Supplies the method-of-regions soft expansion used to justify the classical limit of massive propagators.","marker":"[45]"},{"why":"Provides the Schwinger-parameter derivation of the leading quantum worldline term that the forest expansion generalizes.","marker":"[81]"}],"fun_headline_variants":["KMOC and WQFT integrands coincide exactly at two loops","Worldline rewrite proves KMOC equals WQFT at loop level","Retarded causality emerges in classical QFT from amplitudes","Two-loop classical impulses: scattering amplitudes and worldlines unify","Zero-measured cuts dropped to kill super-leading divergences in KMOC"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument relies on setting to zero, by hand, all zero-measured cuts—kinematically degenerate or forbidden phase-space configurations—and if any of them contributed at the classical order, the exact KMOC-WQFT equality would fail.","fun_headline_variants_meta":{"raw":{"variants":["KMOC and WQFT integrands coincide exactly at two loops","Worldline rewrite proves KMOC equals WQFT at loop level","Retarded causality emerges in classical QFT from amplitudes","Two-loop classical impulses: scattering amplitudes and worldlines unify","Zero-measured cuts dropped to kill super-leading divergences in KMOC"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001037,"raw_usage":{"total_tokens":4345,"prompt_tokens":908,"completion_tokens":3437,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":524,"completion_tokens_details":{"reasoning_tokens":3348}},"tokens_in":524,"tokens_out":3437,"duration_ms":22041,"temperature":1.0,"reasoning_tokens":3348,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T15:32:03.996154+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate one of the diagrams the paper sets to zero by hand, such as the degenerate vertex process shown around eq. (39), with a regulator that keeps the pinched surface finite; a non-vanishing contribution at the classical order $O(\\lambda^2)$ to the impulse would break the claimed exact correspondence.","supporting_citations":[],"review_version":1}