{"id":"311879c4-b59d-457b-a157-267048dc2d6b","arxiv_id":"2411.14985","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The paper derives the two-loop coefficient functions for the operator product expansion of two electromagnetic currents in general kinematics, the central ingredient for next-to-next-to-leading-order DDVCS predictions.","lead":"Physicists have computed the two-loop quantum corrections to the coefficient functions for double deeply virtual Compton scattering (DDVCS), a process that probes proton structure. These corrections, presented in analytic form, are essential inputs for future electron-ion collider measurements of generalized parton distributions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproven completeness of the ansatz in Eq. (3.33): the non-orthogonal moment problem (3.30) may not uniquely determine the CF, so a bulk-region term orthogonal to all P_N^{λ_N} could be missed.","rationale":"Good faith: the paper is a serious calculation with multiple nontrivial checks: one-loop reproduction, DVCS limit against Refs. [13,16], threshold expansion against Ref. [53], analyticity at q1^2+q2^2=0, and two independent functional representations of the same expressions. These checks give real support to the central claim. The numerical non-convergence of the longitudinal CFF noted by the reader is a phenomenological caveat, not a correctness objection to the CFs themselves.\n\nLoad-bearing concern: the method reduces the off-forward CF to DIS data via conformal symmetry, but the step from the moment equations (3.30) to the explicit CF uses the unproven ansatz (3.33). Because P_N^{λ_N} are not orthogonal, the moment equations do not automatically determine a unique function; one needs a completeness or injectivity statement. The ansatz provides one solution; without uniqueness, a term orthogonal to all P_N^{λ_N}, or outside the image of the convolution map, could be missed. The checks listed in the paper are not sensitive to such a term if it is localized in the bulk. This is a correctness risk, not a disagreement with consensus.\n\nWhy the verdict is unchanged: the concern is real but does not by itself warrant rejection. The numerical Gram-matrix and moment-inversion test proposed in concrete_test can settle it, and the paper's strong external checks make it plausible that the ansatz is complete even though this is not proven. CONDITIONAL is the appropriate verdict, with the condition being a demonstration of uniqueness or completeness of the moment representation, or an independent check of the bulk CF. Thus no change to the reader's verdict is needed.","tokens_in":31264,"tokens_out":7502,"duration_ms":81096,"concrete_test":"Choose a Euclidean ω (e.g., ω=0.3) and expand the r.h.s. of Eq. (3.30) to two loops. Numerically solve the truncated moment system Σ_{M even} G_{NM} a_M = R_N with G_{NM}=∫_{-1}^1 dx P_N^{λ_N(2)}(x) P_M^{λ_M(2)}(x), using the two-loop λ_N from Sect. 4, for N,M up to about 80. Reconstruct C_i(ωx,ω)=Σ a_M P_M^{λ_M}(x) and compare pointwise with the expressions in Appendix E. If the two disagree beyond integration accuracy, the convolution ansatz is not the unique solution of the moment equations and the published CFs are incomplete; if they agree stably with increasing Nmax, the ansatz captures the unique solution and the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.3 introduces Eq. (3.33) as an ansatz: C_i(ωx,ω) = ∫ dx' c_i(ω,x') K_i(x',x,ω), with c_i fixed by Eq. (3.36) and K_i an SL(2)-invariant operator whose spectrum is matched to DIS moments in Eq. (3.39). The entire two-loop result inherits this assumption. The problem is that the moment equations (3.30) involve the non-orthogonal family P_N^{λ_N}(x), with λ_N = 3/2 − ϵ* + γ_N/2 depending on N through the anomalous dimension. Matching these moments for all even N fixes the CF uniquely only if the moment problem is injective on the space of functions with the analyticity properties of a two-loop CF; no such injectivity or completeness proof is given. The ansatz is therefore one particular solution of the moment equations, not a proven general representation. The checks in Section 4 (DVCS limit ω=1, threshold expansion in Appendix F, analyticity at q1^2+q2^2=0) probe restricted kinematics or singular regions: a missing contribution that vanishes at ω=1 and at threshold, but is nonzero in the bulk, would evade all of them. If such a component exists, the published CFs are incomplete despite agreeing with all listed checks. This is the load-bearing gap in the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper computes the two-loop, flavor-nonsinglet vector coefficient functions for the OPE of two electromagnetic currents in general kinematics with two different photon virtualities, i.e., for double deeply virtual Compton scattering (DDVCS). The calculation uses conformal symmetry of large- n_f QCD at the Wilson-Fisher fixed point, following the program of Braun, Manashov, Moch, and collaborators. The authors introduce a new solution ansatz in momentum-fraction space, reconstruct SL(2)-invariant kernels from their Mellin spectra, and obtain analytic expressions for the longitudinal and transverse coefficient functions in the MS scheme. They provide the results in Appendix E and in two ancillary files, perform several internal consistency checks (one-loop limit, DVCS limit, analyticity at q1^2+q2^2=0, and agreement with the threshold resummation by Schoenleber), and estimate the numerical impact for kinematics of proposed DDVCS experiments.","tokens_in":31630,"tokens_out":5357,"duration_ms":58883,"significance":"If the results are correct, they provide the first two-loop off-forward coefficient functions for DDVCS with two different photon virtualities, an ingredient needed for NNLO phenomenology of lepton-pair electroproduction and related two-photon processes. The methodological contribution is also significant: the conformal-symmetry construction is extended to unequal photon virtualities, and the analytic results are made available in two complementary representations, including one with no spurious singularities. The paper includes several nontrivial checks, an independent threshold-resummation comparison, and reproducible ancillary files for numerical evaluation. The main open question is whether the solution ansatz in Sec. 3.3 is complete; if that gap can be closed, the paper would be a strong contribution to the field.","major_comments":[{"comment":"The central calculation rests on the ansatz C_i(ωx,ω) = ∫ dx' c_i(ω,x') K_i(x',x,ω), with K_i reconstructed from its Mellin spectrum. The moment equations (3.30) involve the N-dependent, non-orthogonal family P_N^{λ_N}(x) with λ_N = 3/2 − ϵ* + γ_N/2, and no completeness or injectivity statement is provided for this family in the space of functions with the analyticity properties expected of two-loop coefficient functions. Matching all even-N moments determines the coefficient function uniquely only if this moment problem is injective; otherwise the construction yields one particular solution. The checks reported in Sec. 4 and Appendix F (DVCS limit ω=1, analyticity for λ→0, threshold behavior, and agreement with Schoenleber) probe restricted or boundary kinematics and cannot detect a bulk term that vanishes in those limits. Please either prove that {P_N^{λ_N}} is complete in the relevant function space and that the ansatz (3.33) with the choice (3.36) spans the full space of two-loop coefficient functions, or provide an independent bulk check, such as a direct evaluation of a few lowest moments for representative ω≠1 or a numerical two-loop computation at a generic kinematic point. This is load-bearing for the abstract claim that these are the two-loop coefficient functions.","section":"Sec. 3.3, Eqs. (3.30), (3.33), (3.36), (3.39)"},{"comment":"The comparison with the independent threshold-resummation calculation by Schoenleber is stated only verbally as an agreement. Because this check is a key external validation of the singular behavior, the paper should provide at least a short quantitative comparison, e.g., the first few threshold coefficients from both calculations displayed together. This would allow the reader to assess the accuracy of the comparison and to understand exactly which terms are being tested.","section":"Sec. 4 and Appendix F"}],"minor_comments":[{"comment":"There are typographical errors: \"particulari-\" in the Introduction and \"map our results map our 1-loop and 2-loop results\" in Sec. 4. These should be corrected.","section":"Page 2 and Sec. 4"},{"comment":"The symbol \"w\" is used interchangeably with \"ω\" in several formulas, sometimes in the same expression. The notation should be unified to avoid ambiguity.","section":"Notation throughout, especially Eqs. (2.18), (E.2)-(E.15), and (F.1)"},{"comment":"The arrow notation in the limit ω→1 is nonstandard; the limit should be written explicitly with a subscript or the word \"→\" rather than \"7→\".","section":"Eq. (3.31)"},{"comment":"For the longitudinal Compton form factor, the NNLO correction is so large that the truncation does not show signs of convergence at the chosen kinematics. This is not a flaw in the coefficient functions, but it should be stated as a caveat in the numerical discussion so that the reader does not interpret Fig. 2 as a convergent prediction.","section":"Sec. 5"}],"recommendation":"major_revision","confidential_remarks":"The ansatz-completeness concern is the main obstacle: it is a genuine gap in the proof of the central claim, not merely a presentation issue. If the authors can provide a completeness argument for the moment problem or an independent bulk check, the paper should be acceptable. The self-citations are not excessive; the prior papers by the same group contain the framework on which this calculation builds. The paper fits the journal's scope and the results are of high practical interest."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the headline: this paper delivers the two-loop coefficient functions for DDVCS at general photon virtuality ratio, which is genuinely new and fills the last gap for NNLO flavor-nonsinglet DDVCS predictions. The authors extend their conformal symmetry framework from DVCS to arbitrary ω, which requires a new solution ansatz. The calculation is carefully cross-checked: they reproduce the one-loop results, match the DVCS limit, find agreement with Schoenleber's threshold resummation, and verify the expected analyticity at q1^2+q2^2=0. The expressions are provided in two ancillary files, which is good practice.\n\nThe main soft spot is the ansatz in Eq. (3.33). The CFs are written as a convolution of simple weight functions c_i with SL(2)-invariant kernels, and the kernels are reconstructed from their Mellin spectra. The spectra are fixed by the DIS moments via Eq. (3.39). The issue is that the moment problem uses the functions P_N^{λ_N}(x) with λ_N depending on N through the anomalous dimension. These are not orthogonal and no completeness or injectivity proof is given. It is conceivable that a contribution vanishes in all the limits they check (ω→1, threshold, λ→0) but is nonzero in the bulk. If that happened, the published CFs would be incomplete. I don't think this is fatal — absent evidence, the natural reading is that the ansatz is a proper parametrization and the moment problem is likely invertible on the space of functions with the expected analyticity — but the authors should say something about it. A referee should ask for a discussion or a proof that the ansatz is exhaustive, or at least an additional numerical check in the bulk region.\n\nA smaller point: the numerical section shows the longitudinal perturbative series is not converging for the chosen kinematics. The authors mention it but it is easy to miss. That is not a flaw in the calculation, but it deserves more prominence since it affects the phenomenological claims.\n\nOverall, this is a serious and careful calculation. The result is important for the DDVCS program at JLab and EIC. It deserves a normal peer review, not a desk reject, but I would send it to a referee who can pressure the authors on the completeness question.","headline":"First two-loop DDVCS coefficient functions with a solid calculation, but the completeness of the solution ansatz is assumed rather than proven.","tokens_in":32088,"tokens_out":7186,"would_cite":true,"duration_ms":71441,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["12.38.Bx","13.60.Fz"],"model":"deepseek-v4-flash","headline":"The paper derives the complete two-loop coefficient functions for double deeply virtual Compton scattering in general kinematics using conformal symmetry.","keywords":["double deeply virtual Compton scattering","coefficient functions","conformal symmetry","generalized parton distributions","NNLO QCD","Wilson-Fischer fixed point","operator product expansion","generalized polylogarithms"],"falsifier":"Evaluate the two-loop transverse coefficient function at a generic kinematic point with unequal photon virtualities, for example $z=0.3$, $\\omega=-1.27$, $Q^2=\\mu^2$, by a direct Feynman-diagram calculation and compare with the expressions in Appendix E. A mismatch away from the special limits would show that the ansatz (3.33) misses contributions.","tokens_in":31111,"feed_emoji":"⚛️","tokens_out":6693,"duration_ms":62442,"temperature":0.7,"pith_summary":"This paper calculates the two-loop coefficient functions entering the operator product expansion of two electromagnetic currents when the two photons have different virtualities, i.e. for double deeply virtual Compton scattering (DDVCS). The results supply the flavor-nonsinglet vector ingredient needed for next-to-next-to-leading-order (NNLO) predictions of this process, which can be used to extract generalized parton distributions from exclusive lepton-pair electroproduction. The authors obtain analytic expressions in momentum-fraction space in the MS scheme, expressed through generalized polylogarithms up to weight four, and verify them against known limits. Their numerical estimates show that the two-loop corrections are large for the kinematics of proposed fixed-target and collider experiments, so the new terms are relevant for interpreting upcoming data.","feed_headline":"Conformal symmetry yields two-loop DDVCS coefficient functions","feed_subtitle":"Analytic NNLO corrections for unequal photon virtualities; numerical impact is large for upcoming JLab and EIC runs.","key_machinery":"The load-bearing machinery is the conformal-symmetry reduction of off-forward coefficient functions to forward ones. At the Wilson-Fischer fixed point in $d=4-2\\epsilon$, the OPE of two conserved currents is fixed by two constants per spin, which are identified with the DIS coefficient functions; this yields the master equations (3.30). The solution uses the ansatz $C_i(\\omega x,\\omega)=\\int_{-1}^{1} dx'\\, c_i(\\omega,x') K_i(x',x,\\omega)$, where $K_i$ are SL(2)-invariant kernels determined by their Mellin spectra, and the weight functions $c_i$ are chosen so that the spectra are expressed directly through moments of the DIS coefficient functions and the eigenvalues of the rotation operator $U$ that connects the rotated scheme to the MS scheme. The SL(2)-invariant kernels are reconstructed via an intertwining operator $T$ and evaluated in momentum-fraction space through position-space convolution integrals computed with the HyperInt package. A decisive simplification is that the invariant kernels turn out not to depend on the ratio of photon virtualities $\\omega$, which is what makes the calculation feasible.","core_discovery":"The central claim is that the two-loop flavor-nonsinglet vector coefficient functions for the OPE of two electromagnetic currents in the MS scheme are now known in closed analytic form for general kinematics with two different photon virtualities. The calculation works by evaluating the conformal OPE at the Wilson-Fischer fixed point in $d=4-2\\epsilon$ dimensions, where conformal symmetry fixes the off-forward coefficient functions in terms of known deep-inelastic-scattering coefficient functions plus lower-order off-forward information, and then restoring the physical four-dimensional MS result with the QCD $\\beta$ function. Explicit results are presented in Appendix E and two ancillary files using different representations of generalized polylogarithms. The paper checks the results against the DVCS limit $\\omega=1$, against an independent threshold-resummation calculation, and against the expected analyticity at $q_1^2+q_2^2=0$.","pith_inferences":["Editorial inference: the same conformal-symmetry route should extend to NNLO coefficient functions for axial-vector, scalar, and tensor currents, and hence to pion electromagnetic and transition form factors at NNLO.","Editorial inference: the apparent lack of convergence of the longitudinal Compton form factor at the tested low lepton-pair mass suggests that NNLO alone may not be sufficient for precision phenomenology in that kinematic corner; higher scales or resummations may be needed.","Editorial inference: the analytic expressions provide a direct numerical basis for comparing future DDVCS measurements at JLab and the EIC with GPD-based predictions, making the coefficient functions part of a testable extraction framework.","Editorial inference: because the invariant kernels are independent of $\\omega$, the same kernels may be reusable for other two-photon processes whose coefficient functions differ only through the forward DIS input, saving effort in future NNLO calculations."],"forward_implications":["DDVCS can now be described at NNLO for the flavor-nonsinglet vector channel, giving the process the same perturbative accuracy as modern inclusive PDF fits.","The analytic expressions, together with the analytic-continuation prescription in Eq. (2.15), provide predictions for DVCS, TCS and DDVCS kinematics from the same Euclidean result.","The two-loop corrections are numerically large in the considered JLab and EIC kinematics, especially for the imaginary part of $F_\\perp$ and for $F_L$, so they cannot be neglected in the analysis of proposed experiments.","The method reduces the $\\ell$-loop off-forward coefficient-function problem to the known forward (DIS) coefficient function plus an $(\\ell-1)$-loop off-forward calculation in $4-2\\epsilon$ dimensions, indicating a route to higher perturbative orders.","The technique carries over to coefficient functions for correlations of other quark-antiquark currents, with light-cone sum rules for pion form factors as a stated application."],"supporting_citations":[{"why":"Supplies the conformal-OPE framework and the two-loop DVCS vector coefficient-function method that this calculation extends to unequal photon virtualities.","marker":"[13]"},{"why":"Establishes the conformal-symmetry approach to evolution equations beyond one loop that underlies the whole calculation.","marker":"[25]"},{"why":"Shows that QCD in $d=4-2\\epsilon$ at the Wilson-Fischer fixed point is conformally invariant, the premise that lets the authors use conformal OPE.","marker":"[26]"},{"why":"Provides the rotation operator and three-loop evolution kernels used to define the rotated scheme in which the calculation is performed.","marker":"[20]"},{"why":"Gives the intertwining operator $T$ needed to reconstruct SL(2)-invariant kernels from their Mellin spectra.","marker":"[21]"},{"why":"Provides the known one-loop DDVCS coefficient functions in Minkowski space used as a check of the one-loop reproduction.","marker":"[33]"},{"why":"Gives the NNLO DVCS result with which the $\\omega=1$ limit of the new coefficient functions is compared.","marker":"[16]"},{"why":"Independent threshold-resummation calculation with which the singular threshold terms of the two-loop coefficient functions are checked.","marker":"[53]"},{"why":"HyperInt package used to compute the position-space convolution integrals that produce the analytic results in momentum-fraction space.","marker":"[39]"},{"why":"Supplies the toy GPD model and the leading-region analysis used for the numerical estimates of the Compton form factors.","marker":"[27]"}],"fun_headline_variants":["Two-loop DDVCS coefficients in analytic form","Conformal symmetry yields two-loop DDVCS","Analytic NNLO DDVCS coefficient functions","Two-loop coefficient functions for DDVCS","Two-loop DDVCS coefficients at NNLO"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The solution ansatz in Eq. (3.33) assumes that every coefficient function that can occur at two loops can be represented as a convolution of simple weight functions with SL(2)-invariant kernels, and there is no proof that this representation spans all possible contributions; a missing term in the bulk kinematic region would not be caught by the checks performed.","fun_headline_variants_meta":{"raw":{"variants":["Two-loop DDVCS coefficients in analytic form","Conformal symmetry yields two-loop DDVCS","Analytic NNLO DDVCS coefficient functions","Two-loop coefficient functions for DDVCS","Two-loop DDVCS coefficients at NNLO"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000858,"raw_usage":{"total_tokens":3677,"prompt_tokens":850,"completion_tokens":2827,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":466,"completion_tokens_details":{"reasoning_tokens":2757}},"tokens_in":466,"tokens_out":2827,"duration_ms":21371,"temperature":1.0,"reasoning_tokens":2757,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:39:00.563870+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the two-loop transverse coefficient function at a generic kinematic point with unequal photon virtualities, for example $z=0.3$, $\\omega=-1.27$, $Q^2=\\mu^2$, by a direct Feynman-diagram calculation and compare with the expressions in Appendix E. A mismatch away from the special limits would show that the ansatz (3.33) misses contributions.","supporting_citations":[{"cited_title":"Evolution equations beyond one loop from conformal symmetry","cited_arxiv_id":"1306.5644","evidence_quote":"Establishes the conformal-symmetry approach to evolution equations beyond one loop that underlies the whole calculation."},{"cited_title":"Conformal symmetry of QCD in $d$-dimensions","cited_arxiv_id":"1810.04993","evidence_quote":"Shows that QCD in $d=4-2\\epsilon$ at the Wilson-Fischer fixed point is conformally invariant, the premise that lets the authors use conformal OPE."},{"cited_title":"Threshold resummation for double-deeply virtual Compton scattering","cited_arxiv_id":"2411.11686","evidence_quote":"Independent threshold-resummation calculation with which the singular threshold terms of the two-loop coefficient functions are checked."}],"review_version":1}