{"id":"f24fb466-61f3-49f6-8106-2d691acb4671","arxiv_id":"2506.16489","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A new multi-parton model for DIS computes NLO structure functions as iterated discontinuities of Feynman integrals and is equivalent to the parton model up to a scheme change.","lead":"This paper builds a perturbative model of deep inelastic scattering in which several partons from the proton can enter the hard collision, with initial-state partons clustered by momentum conservation alone. The authors show that at next-to-leading order the model reproduces the standard parton model up to a scheme change, and they express the cross-section as a double discontinuity of a Feynman integral.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The doubled optical theorem rests on the uniform-density collapse of eq. (22), which is asserted from infrared safety rather than derived; a correlated initial-state density would break the double-discontinuity representation of eq. (136).","rationale":"The reader's weakest assumption correctly identifies the simplest-clustering uniform density as the load-bearing input: it is the step that turns the partonic tensor into iterated discontinuities. My reading agrees that this is an assumption, not a derived QCD result, and that the higher-winding truncation is a second admitted limitation. However, the paper is explicitly a model and repeatedly acknowledges these restrictions, including the need for an NLO Drell-Yan comparison to test the physical relevance of the truncation. The internal checks (Ward identity, IR cancellation per embedding, reproduction of parton-model results when the extra diagrams are removed) and the reproducible ancillary data give real support to the technical computation. The concern therefore does not invalidate the paper's conditional claims; it confirms that the present verdict CONDITIONAL is the right level of endorsement, so no adjustment is needed.","tokens_in":60906,"tokens_out":7126,"duration_ms":86080,"concrete_test":"Recompute the NLO leading-virtuality quark structure functions using the general density of eq. (20) with a minimal non-uniform, still collinear-safe ansatz, e.g. f_2(p1,p2) = C (p1·n_+)^a (p2·n_+)^a delta(p1+p2-p), and compare the resulting W_q against disc_p2 disc_1-x I_Gamma for the relevant class-one and class-two embeddings. If the equality (136) fails, or if F_L no longer matches the parton-model value, the doubled optical theorem and the claimed scheme change are artifacts of the uniform-density choice rather than generic consequences of infrared safety.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central identity eq. (136) is established for the specific density matrix of eq. (12), whose general form eq. (20) is collapsed to the p-only form eq. (22) by the statement that an infrared-safe density must, 'by the usual degeneracy arguments', depend only on p. No derivation is supplied, and this is a modeling assumption rather than a consequence of QCD. If the true proton state has correlations among initial-state partons, e.g. f(p1,...,pn) depending on individual momentum fractions, then the interference diagrams of a given embedding are weighted by that f and the partonic tensor is no longer a sum of unweighted double discontinuities of a single embedding integral. The KLN theorem constrains the S-matrix, not the model density; choosing f to be uniform is what makes the cross-section a sum of iterated discontinuities. The paper itself flags the related exclusion of higher-winding embeddings with |r_e|>1 as the least physically motivated constraint (sect. IIIB3) and notes in sect. VIh that the full KLN sum over windings is infinite and requires an additional truncation or resummation principle. Thus the theorem, as stated, is demonstrated only for the truncated winding-one embedding set under a specific ansatz for the proton density, and its claimed universality is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a perturbative model for multi-partonic initial states in deep inelastic scattering. Initial-state partons are clustered into a proton-like object using momentum conservation alone, with a uniform Fock-space density and a cutoff Lambda^2 on the cluster virtuality. The authors organize the resulting diagrammatic contributions by graph embeddings, i.e. vacuum graphs endowed with a puncture and winding numbers, and show that at leading virtuality the contribution of each embedding is an iterated double discontinuity of the corresponding embedding integral (eq. (136)). The framework is applied to NLO DIS structure functions; the authors find the same Altarelli-Parisi scale evolution as the parton model, identical longitudinal structure functions, and finite pieces that differ from the parton model. They also present a two-loop example and a first extension to Drell-Yan. The paper contains extensive cross-checks, including the Ward identity, reproduction of parton-model results when MP diagrams are removed, embedding-by-embedding infrared cancellations, integrability tests, and supplementary code and embedding data.","tokens_in":61125,"tokens_out":9253,"duration_ms":106675,"significance":"If the doubled optical theorem in eq. (136) holds at the claimed level of generality, it is a substantial technical result: infrared-finite multi-parton building blocks would be computable by standard cut-integral and reverse-unitarity methods, and the model provides explicit, falsifiable NLO predictions. The paper is also valuable for introducing a systematic graph-theoretic classification of KLN-related embeddings and for releasing code and data. The central caveat is that the theorem and the parton-model equivalence rest on modeling assumptions, in particular the uniform initial-state density and the truncation of higher-winding embeddings, whose scope is not fully established. The strengths of the paper are real, but the strongest claims need to be delimited more carefully.","major_comments":[{"comment":"The collapse of the general diagonal density to the uniform density is asserted, not derived. The text moves from f(p1,...,pn) to a function of p alone via the statement that infrared safety 'by the usual degeneracy arguments' forces this collapse. The KLN theorem constrains cancellations among degenerate states; it does not fix the relative weights of different initial-state configurations. If f retains dependence on the individual momentum fractions, the sum over interference diagrams in eq. (90) is weighted by f, and the identification with the unweighted double discontinuity of a single embedding integral in eq. (136) is not guaranteed. This collapse is load-bearing for the doubled optical theorem and for the NLO comparison with the parton model. Please either provide an explicit derivation of the collapse or state it as a defining axiom of the model, and test its necessity, for example by computing an NLO embedding with a non-uniform f and showing what breaks in eq. (136).","section":"§II.B, eqs. (20)–(22)"},{"comment":"The exclusion of higher-winding embeddings is a truncation, not a derived property. The paper itself describes the |r_e| <= 1 condition as the least physically motivated constraint, and §VI.B.h states that the full KLN sum over windings is infinite and requires an additional truncation or resummation principle. In addition, the leading-order matching to the parton model requires the extra condition f_q(1/q) = q^alpha f_q(1), which is not fixed by the model. Consequently the NLO structure functions and the claimed scheme-change equivalence to the parton model are established only for the particular finite subset Embdis. The abstract and Section VI state the equivalence more strongly than what is proven. Please either give a physical principle that selects Embdis, bound the contribution of higher-winding embeddings at NLO, or rephrase the equivalence claim as applying only to the truncated set.","section":"§III.B.3, §VI.A.b, §VI.B.h"},{"comment":"Equation (136) is presented as a general doubled optical theorem, but the evidence is an explicit match for NLO one-loop embeddings (triangle, box, pentagon) and one two-loop example. The derivation relies on the leading-virtuality analytic forms of eqs. (119)–(120) and the reverse-unitarity reduction of eqs. (132)–(135); these are checked, not proven, at arbitrary order. Furthermore, the text itself notes that the discontinuity and the virtuality expansion do not commute, so eq. (136) holds only order-by-order in the virtuality expansion. Please state the theorem with its actual domain (verified at NLO, conjectural beyond), or supply a proof, and make the order-by-order qualification part of the statement of eq. (136).","section":"§V.D, eq. (136)"}],"minor_comments":[{"comment":"In eq. (22) the density is written as f(p), while the surrounding text says f does not depend on p^2 but can depend on the momentum fraction; please clarify the notation by writing f(ξ) or specifying the p-dependence after integration over the phase space.","section":"§II.B, eq. (22)"},{"comment":"The sentence 'By using Z_n we bound the possible winding numbers to be less than n' uses an integer n that is not defined; it presumably depends on the edge or cut set, but this should be stated explicitly.","section":"§III.B.3"},{"comment":"The validation tests a)–d) are listed but not documented in detail; a short table or a description of how the embedding-by-embedding cancellation and the integrability checks were performed would make these claims reproducible.","section":"§VI.B.a"},{"comment":"The claim that equality of the longitudinal structure functions is necessary and sufficient for a scheme change is stated without proof; please provide a derivation or a reference, since the sufficiency direction is not immediate for two structure functions that both depend on the PDF transformation.","section":"§II.D.c, eq. (50)"},{"comment":"There are several typos in this section, including 're-routing okk' and 'compeletely' in §II.B; a careful proofread is needed.","section":"§VII.B"},{"comment":"The code is described as hosted on GitHub 'at this link', but no URL is given in the text; please include the actual repository address.","section":"§III.B.3.d"},{"comment":"The acknowledgments thank 'the work of M.H.' but the author list contains M. Ruf; this should be corrected to M.R.","section":"Acknowledgments"},{"comment":"The embedding tables are very hard to read in the text version; since the .m data files are provided as supplementary material, it would help to keep a few illustrative rows in the paper and move the full tables to the supplementary material.","section":"Appendix C"}],"recommendation":"major_revision","confidential_remarks":"This is an ambitious and technically rich paper. My main recommendation for the editor is that the publication version must clearly separate what is proven (NLO calculations for the truncated embedding set, explicit cross-checks) from what is conjectured (the general doubled optical theorem and the full equivalence with the parton model). The authors are clearly aware of the main limitations, because the body of the paper states them, but the abstract and conclusions overstate the results. With a revision that removes that mismatch, the paper would be a valuable contribution; in its current form the claims are too strong relative to the demonstrated domain of validity."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this paper delivers a genuinely new perturbative framework: the simplest clustering criterion (momentum conservation only), the graph-embedding classification with automorphism-invariant canonical forms, and the identification of the MP cross-section with the double discontinuity of a Feynman-type embedding integral, eq. (136). Second, the NLO DIS structure functions in this model are computed for the first time, and they reproduce the parton model's scale evolution and longitudinal structure functions, with finite pieces that differ by a scheme change, plus a genuinely new F2qq. That is real, checkable work, with code and ancillary data shipped.\n\nWhat I find strongest is the honesty and density of internal cross-checks: Ward identity, embedding-by-embedding IR cancellations, reproduction of parton-model results when the new diagrams are removed, and integrability over x in (0,1) all check out. The embedding generation pipeline itself is a useful contribution.\n\nNow the soft spots, in proportion. Eq. (136) is demonstrated for the NLO one-loop embeddings and a two-loop example, not proven for the general case; the paper itself notes that the discontinuity and virtuality expansions do not commute. More importantly, the whole construction rests on the uniform density ansatz of eq. (22), justified by infrared-safety 'degeneracy arguments' rather than derived from QCD. The stress-test note is fair on this point, but it does not sink the paper, because the authors explicitly frame this as a modeling assumption, not a theorem about QCD. They also flag the higher-winding exclusion (|r_e| <= 1) as the least physically motivated constraint, and they note the full KLN sum over windings is infinite and needs a truncation or resummation principle. The LO matching to the parton model for higher windings uses the admittedly ad hoc f_q(1/q) = q^alpha f_q(1).\n\nSo the central theorem is conditional: it holds for the truncated embedding set under a specific density ansatz. That conditionality is stated in the paper, not hidden. Within its self-defined framework, the math and the computations appear solid; the NLO results are derived, not fitted. My only real complaints are that class three and four contributions lack an independent validation beyond integrability and the Ward identity, and that the paper's abstract slightly oversells the generality of eq. (136).\n\nWho this is for: anyone working on KLN, multi-parton interactions, optical theorems, or alternative models of hadronic initial states. It deserves a serious referee: the novelty is genuine, the calculations are nontrivial, and the limitations are explicit. I would engage with it, and I would expect the report to focus on the generality of eq. (136) and the physical status of the uniform-density collapse.","headline":"A serious, internally consistent KLN-based model of multi-partonic DIS whose central doubled optical theorem is shown in examples, not proven in general, and whose physical relevance rests on an explicitly modeled uniform-density ansatz.","tokens_in":61701,"tokens_out":2626,"would_cite":true,"duration_ms":26855,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"At leading virtuality, multi-partonic DIS cross-section building blocks equal iterated discontinuities of Feynman-type embedding integrals; the NLO structure functions match the parton model up to a scheme change.","keywords":["multi-partonic interactions","deep inelastic scattering","KLN theorem","virtuality expansion","graph embeddings","iterated discontinuities","reverse unitarity","parton model"],"falsifier":"Perform the same model's next-to-leading-order Drell-Yan calculation: the paper's equivalence claim requires that the same scheme redefinition that converts multi-parton DIS into parton-model DIS also converts multi-parton Drell-Yan into parton-model Drell-Yan; a difference in the two conversions would make the model experimentally distinguishable from the parton model.","tokens_in":60625,"feed_emoji":"⚛️","tokens_out":11151,"duration_ms":112705,"temperature":0.7,"pith_summary":"This paper introduces a perturbative model of multi-partonic interactions in deep inelastic scattering. The proton is described as a uniform mixture of massless partons, and the leading term in an expansion in the invariant mass $\\Lambda^2$ of the initial-state cluster is studied. The paper's central claim is that each infrared-finite class of interference diagrams, organized by \"embeddings\" of vacuum graphs, can be computed as a double discontinuity of a Feynman-type integral, a result the authors call a doubled optical theorem. If correct, single- and multi-parton initial states are treated on the same footing with building blocks that are separately infrared finite and computable by standard cut-integral and Reverse Unitarity techniques. At next-to-leading order the resulting structure functions agree with the parton model after a scheme change, with the same Altarelli-Parisi scale evolution in $\\Lambda^2/Q^2$.","feed_headline":"Multi-parton DIS becomes iterated discontinuities of Feynman graphs","feed_subtitle":"The result makes every leading-virtuality multi-parton building block computable with standard cut-integral technology.","key_machinery":"The load-bearing object is an \"embedding\": a vacuum Feynman graph together with a puncture whose position is encoded by winding numbers of the graph's cycles. An embedding groups all cuts that can be moved around the puncture into one equivalence class, and the sum of the corresponding interference diagrams is separately infrared finite by the KLN theorem. The second essential element is the simplest clustering criterion: the initial-state partons are taken to be a uniform density matrix constrained only by total momentum, so that in the $\\Lambda^2\\to0$ limit the density collapses to a function of the proton momentum $p$ alone. The identity carrying the argument is the doubled optical theorem, eq. (136), together with the expansion-by-regions form $I_\\Gamma=(p^2)^{-1}H+(-p^2)^{-1-\\epsilon}C+O((p^2)^0)$; applying $\\operatorname{disc}_{p^2}\\operatorname{disc}_{1-x}$ to that expression reproduces the sum of cut diagrams, with the hard and collinear coefficients $H$ and $C$ carrying the KLN cancellations.","core_discovery":"The central discovery is that at leading virtuality the partonic tensor of an embedding $\\Gamma$ satisfies $$W_\\Gamma(p,q)=\\frac{1}{\\operatorname{av}(q)}\\,\\operatorname{disc}_{$p^{2}$}\\,\\operatorname{disc}_{1-x}\\,I_\\Gamma($p^{2}$,$q^{2}$,1-x)+O(($p^{2}$)^0).$$ Interference diagrams that cancel infrared singularities through the KLN theorem are therefore obtained not one by one but as the iterated discontinuity of the embedding's loop integral, generalising the optical theorem and Cutkosky's cutting rules. The identity is checked in detail on triangle, box, and pentagon embeddings by direct Reverse Unitarity evaluation of the cut integrals. Using it, the paper computes next-to-leading order DIS structure functions: the longitudinal structure functions coincide with the parton model, the $F_2$ scale dependence is driven by the usual splitting kernels, and the finite pieces differ only by a scheme change, with genuinely new contributions such as $F_{2qq}$ appearing.","pith_inferences":["Editorial inference: because the scheme-change equivalence is established for only one process, the paper's own Drell-Yan roadmap makes it falsifiable; if the same redefinition does not convert the multi-parton Drell-Yan result into the parton-model one, the two models are experimentally distinguishable.","Editorial inference: adopting simplest clustering as an effective definition rather than a derived fact, the doubled optical theorem gives a direct route to $\\Lambda^2/Q^2$ power corrections in terms of the $p^2$ derivative of the initial-state density; precision Drell-Yan data could constrain those corrections.","Editorial inference: the distinctive $\\log|2x-1|$ term in $F_{2qq}$, though suppressed by $1/(2N_c)$, is a clean kinematic signature that could be searched for in low-virtuality DIS measurements to test the model against the parton model.","Editorial inference: the mechanism points toward a general rule that one discontinuity corresponds to each clustered external state; if that pattern holds, any momentum-conservation-only clustering prescription would define an optical theorem for multi-jet observables, but the paper does not prove this generality."],"forward_implications":["Every infrared-finite multi-parton contribution at leading virtuality can be computed by taking discontinuities of Feynman-type integrals, so standard integration-by-parts, differential-equation, and Reverse Unitarity technology applies.","NLO DIS structure functions match the parton model up to a scheme change: the longitudinal structure functions are identical, and the $Q^2/\\Lambda^2$ evolution of $F_{2q}$ and $F_{2g}$ is governed by the same Altarelli-Parisi kernels.","Class-three and class-four embeddings introduce branch cuts at $x=1/2$ and new objects such as the double-quark structure function $F_{2qq}$, which are integrable, suppressed by $1/(2N_c)$, and carry no $\\Lambda^2$ logarithms.","The simplest clustering criterion generalises to Drell-Yan, where the cross-section becomes a sum of iterated discontinuities (one per incoming proton) and is finite after virtuality integration, with leading behaviour proportional to $\\log(\\Lambda_1^2/m_\\star^2)\\log(\\Lambda_2^2/m_\\star^2)$.","At NNLO, an explicit two-loop example shows that double, single, and on-shell discontinuities cancel their infrared poles after virtuality integration, leaving a finite expression in terms of powers of $\\log(Q^2/\\Lambda^2)$."],"supporting_citations":[{"why":"Kinoshita's argument that cross-section contributions can be seen as double cuts of vacuum diagrams and that mass singularities cancel with summed multiplicities; this is the basis for the embedding classes.","marker":"[16]"},{"why":"Provides the degenerate-state KLN theorem requiring sums over initial and final multiplicities; the model's density-matrix sum is built to realise it.","marker":"[17]"},{"why":"Proposes modelling initial-state hadrons as jets, the conceptual origin of the simplest clustering criterion.","marker":"[52]"},{"why":"Reverse Unitarity cut-integral technology used throughout to compute interference diagrams and compare them with discontinuities.","marker":"[65–69]"},{"why":"Cutkosky's cutting rules, which the doubled optical theorem extends by treating forward-scattering diagrams as discontinuities of embeddings.","marker":"[82]"},{"why":"Standard NLO DIS parton-model result that the pipeline must reproduce when multi-partonic contributions are excluded, and against which the scheme change is identified.","marker":"[93]"}],"fun_headline_variants":["Multi-parton DIS via iterated discontinuities","Iterated cuts compute multi-parton DIS","DIS at NLO through iterated discontinuities","Multi-parton building blocks via double cuts"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction rests on modeling the proton as a uniform mixture of massless partons constrained only by total momentum, with no correlations among initial-state partons, together with the omission of multi-winding diagrams; if a real proton state has non-trivial initial-state correlations, or if those diagrams contribute, the claimed equivalence with the parton model is not guaranteed.","fun_headline_variants_meta":{"raw":{"variants":["Multi-parton DIS via iterated discontinuities","Iterated cuts compute multi-parton DIS","DIS at NLO through iterated discontinuities","Multi-parton building blocks via double cuts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000393,"raw_usage":{"total_tokens":2041,"prompt_tokens":897,"completion_tokens":1144,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":513,"completion_tokens_details":{"reasoning_tokens":1087}},"tokens_in":513,"tokens_out":1144,"duration_ms":7981,"temperature":1.0,"reasoning_tokens":1087,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:25:31.121300+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Perform the same model's next-to-leading-order Drell-Yan calculation: the paper's equivalence claim requires that the same scheme redefinition that converts multi-parton DIS into parton-model DIS also converts multi-parton Drell-Yan into parton-model Drell-Yan; a difference in the two conversions would make the model experimentally distinguishable from the parton model.","supporting_citations":[],"review_version":1}