{"id":"5f3327c7-9a76-4ec4-931f-3def0483052a","arxiv_id":"1908.02521","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In a selected kinematic region, the measured cos(psi) and cos(2psi) azimuthal asymmetries of J/psi leptoproduction can separate the color-singlet model, the 1S0[8]-dominant NRQCD picture, and the equal-color-octet NRQCD scenario.","lead":"This paper calculates the azimuthal asymmetry modulations for J/psi leptoproduction in unpolarized ep collisions and claims they can distinguish the color-singlet model from two NRQCD color-octet scenarios. A statistical analysis at 1000 inverse picobarns suggests the modulations could be measured with enough precision to tell the models apart at a future electron-ion collider.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Feed-down from ψ(2S), χ_cJ and B decays is never modeled, so the predicted azimuthal asymmetries in region D cannot be compared to an unfiltered J/ψ sample without further assumptions.","rationale":"The most load-bearing assumption is the identification of the calculated cross section with the experimentally measured J/ψ sample. The paper computes direct J/ψ production via NRQCD, but the observed sample includes feed-down from higher charmonia and b decays. This is not a minor correction: at HERA, prompt J/ψ includes a substantial feed-down component, and the kinematic region D (high z) is exactly where feed-down from higher-z parents can contribute. The global-fit LDMEs used for the NRQCD and Chao predictions may themselves have been extracted from prompt data, making the direct-only calculation inconsistent with the data set it is meant to predict. Without an estimate of feed-down asymmetries, the predicted separations are not predictions for the unfiltered sample. The reader's weakest assumption identifies this correctly. Other concerns—such as the absence of hard-scattering amplitudes and the lack of NLO uncertainty estimates—are real but concern the accuracy of the direct-production prediction rather than the validity of comparing it to the measured sample. A single concrete check, computing prompt-level asymmetries with feed-down, would settle whether the concern lands. If the prompt-level values remain separated, the paper's proposal would be robust; if not, the central claim would need revision. Thus the reader's CONDITIONAL verdict remains appropriate; no change is needed, though the condition should explicitly include a feed-down analysis.","tokens_in":15974,"tokens_out":10672,"duration_ms":109365,"concrete_test":"Compute the prompt-J/ψ azimuthal asymmetries in region D by adding, for each of the three LDME sets of Eqs. (3.11) and (3.13), the contributions from ψ(2S) and χ_cJ production (with their own LDMEs and branching ratios, smearing z and pt through two-body decay kinematics), and compare with the direct-only values in Eqs. (3.6), (3.12), and (3.14). If any of the three predictions shifts by more than the quoted statistical uncertainty (e.g., 0.013 for Acosψ), the discrimination claim fails unless direct production can be separated experimentally.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's discriminating predictions—Acosψ = 0.561 (CS) versus −0.246 (NRQCD set of Refs [55,60]) and Acos2ψ = −0.203 (Chao) versus −0.0155—are computed from Eq. (2.22), which describes direct J/ψ production through NRQCD factorization over c¯c(n) states. A real ep event sample contains J/ψ from ψ(2S) and χ_cJ feed-down and from b-hadron decays. The paper neither models these contributions nor argues they are negligible in the chosen region D (x>0.001, 0.75<z<0.9, 0.04<y<0.6, pt>1 GeV). Moreover, the LDME sets in Eqs. (3.11) and (3.13) are taken from global fits to prompt/inclusive data; if feed-down was already absorbed in those fits, the direct-production LDMEs do not by themselves predict the sample measured. Since the azimuthal modulation of the feed-down parents and the decay smearing are unknown within the paper, the observed Acosψ and Acos2ψ are not predicted by the models as presented. Statistical uncertainties of order 0.01–0.004 are irrelevant if the central values do not correspond to the unfiltered sample. Thus the central claim is conditional on an unstated direct-J/ψ selection, which is nontrivial at a collider.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper calculates the cos(ψ) and cos(2ψ) azimuthal asymmetries in J/ψ leptoproduction in unpolarized ep collisions, using a decomposition of the leptonic tensor and NRQCD factorization for the hadronic side. It presents the asymmetries as functions of x, y, z, and ξ for the four leading c-cbar intermediate states, then restricts the study to a kinematic region D (x>0.001, 0.75<z<0.9, 0.04<y<0.6, pt>1 GeV) where the cross sections are sizeable. Combining the per-channel asymmetries with three sets of NRQCD LDMEs (color-singlet, an NRQCD global fit, and a 1S0[8]-dominant set), the paper predicts that at an integrated luminosity of 1000 pb^-1 the two asymmetry modulations separate the three models: Acosψ≈0.561 (CS) versus −0.246 (NRQCD fit), and Acos2ψ≈−0.203 (Chao set) versus −0.0155 (NRQCD fit). A 12-bin statistical analysis yields uncertainties of order 0.01 and 0.004, leading the authors to conclude that a future ep collider such as the EIC could discriminate the production mechanisms.","tokens_in":16210,"tokens_out":10799,"duration_ms":99760,"significance":"If the predictions hold, the paper offers a new, falsifiable observable for discriminating charmonium production mechanisms in ep collisions; the explicit leptonic-tensor decomposition and the careful treatment of the bin-by-bin statistical analysis are strengths. The separation between models is large compared with the estimated statistical errors, making the proposed measurement potentially decisive. However, the paper does not address feed-down contributions, which are known to be important in J/ψ samples, and it does not display the hard-scattering amplitudes underlying the numerical predictions. The quantitative reach of the proposal therefore depends on additional modeling and on the theoretical stability of the LO asymmetries.","major_comments":[{"comment":"The central predictions are computed from Eq. (2.22), which describes direct J/ψ production via NRQCD factorization over c-cbar(n) states with LDMEs. However, an experimental J/ψ sample in ep collisions will contain feed-down from ψ(2S) and χ_cJ decays and from b-hadron decays. The paper never mentions or models these sources, so the predicted Acosψ and Acos2ψ in region D cannot be compared with an event sample unless a direct-J/ψ selection is imposed, which is nontrivial at a collider. Moreover, the LDME sets in Eqs. (3.11) and (3.13) are taken from global fits to prompt/inclusive hadroproduction data; if feed-down was already absorbed in those fits, using them for the direct-production cross section in Eq. (2.22) is not automatically consistent. The authors should either quantify the feed-down contribution in region D and its azimuthal modulations or explicitly restrict the prediction to a direct-J/ψ sample.","section":"Section 3, Eqs. (3.2)-(3.6)"},{"comment":"The numerical results in Section 3 depend on the hadronic tensors W_i+γ*→c cbar(n)+i, but these amplitudes are not shown. After Eq. (2.25), the paper states that they 'can be easily evaluated via Feynman diagrams,' which is not sufficient for the reader to reproduce the central predictions, e.g., the values in Eqs. (3.3)-(3.4). The authors should provide the explicit expressions (or a supplement/code) and benchmark the implementation against known cross sections from Ref. [73].","section":"Section 2.1, Eqs. (2.22)-(2.25)"},{"comment":"The statistical analysis estimates only statistical uncertainties. The cross sections are computed at LO QCD with a fixed scale choice μf = μr = sqrt(S(xy+ξ)), and no estimate of scale variation or NLO corrections is given. The claim that the three models can be told apart relies on differences in Acosψ of about 0.8 and in Acos2ψ of about 0.2, which are much larger than the quoted statistical errors (0.013 and 0.004), but a sizable NLO correction to the CS or CO channels could still shift the predicted asymmetries by more than the statistical precision. A theoretical uncertainty band should accompany the predictions.","section":"Section 3, Eq. (3.10)"}],"minor_comments":[{"comment":"The color-octet LDMEs are written with a [1] superscript (e.g., ⟨O(1S[1]0)⟩, ⟨O(3P[1]J)⟩) instead of [8]; the same typo appears in Eq. (3.3) for the 1S0[8] channel. This is misleading and should be corrected.","section":"Eqs. (3.11) and (3.13)"},{"comment":"'L = 103pb−1' should read 'L = 10^3 pb^{-1} = 1000 pb^{-1}', in agreement with the abstract and Section 4.","section":"Text after Eq. (3.7)"},{"comment":"The statistical uncertainty is quoted for the CS prediction of Acosψ and for the Chao prediction of Acos2ψ, but not for the other model scenarios; please provide the full set of uncertainties for all three models so that the separation claim is quantified uniformly.","section":"Section 3, Eqs. (3.10)-(3.15)"},{"comment":"The captions list kinematics but never define the azimuthal angle ψ; adding a definition in Fig. 1 or in the text would improve readability.","section":"Figure captions (Figs. 2-7)"},{"comment":"The analysis uses the GRV 1995 PDFs; since the publication date is 2019, a modern PDF set or a short statement on the PDF dependence of the asymmetries would be preferable.","section":"Section 3, PDF choice"}],"recommendation":"major_revision","confidential_remarks":"The paper is heavily built on the authors' own prior work (Refs. [55,60,72,73]), and the novelty relative to Ref. [73] should be made explicit. The feed-down issue is the main risk: if the authors can add a quantitative estimate of feed-down contamination in region D, the central claim would be substantially strengthened. The missing hard-scattering amplitudes are also a concern for a standalone phenomenological paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here is my read. The new thing is genuine: nobody has computed the cos(psi) and cos(2psi) azimuthal modulations for J/psi leptoproduction at LO in NRQCD, and the state-by-state differences are striking. The idea to focus on a high-z region (0.75<z<0.9) and extract the modulations from a binned fit over psi is sensible. At 1000 pb^-1 the separation between the color-singlet model, the 1S0[8]-dominance picture, and equal-color-octet NRQCD does look statistically significant. That is a useful existence proof for an EIC-era measurement.\n\nThe calculation is internally consistent as far as I can tell. The leptonic-tensor decomposition follows the authors' earlier work, the NRQCD factorization treatment is standard, and the statistical method (12 bins, Gaussian errors, linear regression) is appropriate for the event counts shown. There is no circular fitting: the LDMEs come from other data and the asymmetries are genuine predictions. I do not see a load-bearing internal error.\n\nNow the soft spots, in proportion. The biggest one is feed-down. Equation (2.22) describes direct J/psi production, but a real ep event sample also contains J/psi from psi(2S), chi_cJ, and B decays, and the paper never mentions any of those sources. If their azimuthal modulations differ from direct production, the measured Acospsi and Acos2psi will be diluted or shifted. The LDME sets are taken from global fits to prompt/inclusive hadroproduction data, so it is not even clear that the LDMEs correspond to the same production definition as the prediction. This is not a minor footnote; it is the difference between a prediction and a suggestion. Second, the hard-scattering amplitudes are not shown. The numerical coefficients in Eq. (3.4) are asserted without enough detail for a referee to verify them independently. Third, there are no scale-variation uncertainties; one scale choice is adopted, and QCD corrections could plausibly shift the asymmetries. Fourth, region D is selected after the model separation is visible. That is acceptable for a feasibility proposal, but it should be labeled as an optimization rather than presented as a natural measurement window.\n\nBottom line: this deserves a serious referee. The observable is new, the analysis is clean enough to be useful, and the field would benefit from having it on record. The revision should add a direct-vs-prompt discussion, show at least representative amplitudes, and give some scale sensitivity. I would not yet trust the central numbers as predictions, but I would cite the paper as the source of the observable proposal.\n\nFor an editor: send it to review, with clear expectations of revision. It is not a desk reject.","headline":"The paper's new observable is real and the LO NRQCD calculation is coherent, but the direct-vs-prompt/feed-down mismatch makes its central discrimination claim conditional rather than demonstrated.","tokens_in":16835,"tokens_out":3985,"would_cite":true,"duration_ms":46788,"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":"Two azimuthal modulations in J/ψ leptoproduction can separate the three leading production models at 1000 pb⁻¹.","keywords":["azimuthal asymmetry","J/psi production","leptoproduction","color-singlet model","nonrelativistic QCD","color-octet long-distance matrix elements","1S0[8] dominance","Electron-Ion Collider"],"falsifier":"At an $ep$ collider with $\\mathcal{L}=1000\\ \\mathrm{pb}^{-1}$, bin the $J/\\psi$ yield in $\\psi$ within region $D$ and fit $N_i = A v_1^{(i)} + B v_{\\cos\\psi}^{(i)} + C v_{\\cos2\\psi}^{(i)}$; if the best-fit $A_{\\cos\\psi}$ lies between $0.561$ and $-0.246$ with uncertainty comparable to the interval, or if $A_{\\cos2\\psi}$ comes out near $-0.016$ rather than $-0.203$ when the $^1S_0^{[8]}$-dominant set is assumed, the claimed discrimination is contradicted. A $\\psi(2S)$-vetoed subsample would directly test the feed-down dilution.","tokens_in":15699,"feed_emoji":"🎯","tokens_out":14441,"duration_ms":124761,"temperature":0.7,"pith_summary":"The paper aims to show that the azimuthal asymmetry of $J/\\psi$ production in unpolarized electron-proton collisions can serve as a discriminating test of competing quarkonium production mechanisms. It computes the two independent modulations, $\\cos(\\psi)$ and $\\cos(2\\psi)$, where $\\psi$ is the azimuthal angle of the lepton scattering plane relative to the hadron-production plane, over the full kinematic range. In the chosen region $D$ ($x>0.001$, $0.75<z<0.9$, $0.04<y<0.6$, $p_t>1$ GeV), the color-singlet model predicts $A_{\\cos\\psi}=0.561$, the equal-color-octet NRQCD fit predicts $-0.246$, and the $^1S_0^{[8]}$-dominance picture gives $A_{\\cos2\\psi}=-0.203$ against $-0.0155$ for the other set. A 12-bin fit to the $\\psi$ distribution at integrated luminosity $1000\\ \\mathrm{pb}^{-1}$ yields statistical uncertainties of $0.013$ on $A_{\\cos\\psi}$ and $0.004$ on $A_{\\cos2\\psi}$, small enough to tell the three models apart. If the prediction holds, a measurement at a future $ep$ collider would settle which production mechanism dominates.","feed_headline":"Two azimuthal modulations can separate three J/ψ models","feed_subtitle":"With 1000 pb⁻¹, cos(ψ) and cos(2ψ) values separate color-singlet from color-octet production.","key_machinery":"The machinery is the decomposition of the leptonic tensor into four azimuthally dependent structures, $l^{\\mu\\nu} = A_g(-g^{\\mu\\nu} - q^{\\mu}q^{\\nu}/Q^2) + A_L \\epsilon_L^{\\mu}\\epsilon_L^{\\nu} + A_{LT}(\\epsilon_L^{\\mu}\\epsilon_T^{\\nu} + \\epsilon_T^{\\mu}\\epsilon_L^{\\nu}) + A_T \\epsilon_T^{\\mu}\\epsilon_T^{\\nu}$, with the coefficients $A_g$, $A_L$, $A_{LT}$, and $A_T$ carrying the $\\cos(2\\psi)$, $\\cos(\\psi)$, and $\\cos(2\\psi)$ terms. Since the hadronic tensor for each intermediate $c\\bar c$ state is independent of $\\psi$, all azimuthal dependence enters through these coefficients, yielding the compact form $d\\sigma = \\sigma_0[1 + A_{\\cos\\psi}\\cos\\psi + A_{\\cos2\\psi}\\cos2\\psi]$ after the Fourier projections in Eq. (2.27). The paper computes the parton-level amplitudes for the four intermediate states ($^3S_1^{[1]}$, $^1S_0^{[8]}$, $^3S_1^{[8]}$, $^3P_J^{[8]}$), convolutes them with parton distribution functions, and then performs a 12-bin linear regression over $\\psi$ to estimate the statistical uncertainty of the modulation coefficients. The crucial numerical fact is that $A_{\\cos\\psi}$ cleanly separates the color-singlet channel from the color-octet channels while $A_{\\cos2\\psi}$ isolates the $^1S_0^{[8]}$ channel, so the two modulations carry orthogonal discriminative information.","core_discovery":"The central claim is that the azimuthal modulations in $J/\\psi$ leptoproduction are strongly channel-dependent and survive integration over a practical kinematic region. Because the cross section can be written as $d\\sigma = \\sigma_0[1 + A_{\\cos\\psi}\\cos\\psi + A_{\\cos2\\psi}\\cos2\\psi]\\,d\\psi$, the two modulation coefficients can be extracted by a linear regression on the $\\psi$ distribution. In region $D$ the paper finds $A_{\\cos\\psi}^{CS} = 0.561$ for the color-singlet channel, and $-0.225$, $0.531$, and $-0.346$ for the $^1S_0^{[8]}$, $^3S_1^{[8]}$, and $^3P_J^{[8]}$ channels, respectively, while $A_{\\cos2\\psi}$ is $-0.281$ for $^1S_0^{[8]}$ and near zero or small for the others. Weighting these channels with three published long-distance matrix-element sets gives the model-level predictions $A_{\\cos\\psi}=0.561$ (color-singlet), $-0.246$ (equal-color-octet NRQCD), and $A_{\\cos2\\psi}=-0.203$ ($^1S_0^{[8]}$-dominant set) versus $-0.0155$ (equal-color-octet). The statistical analysis at $1000\\ \\mathrm{pb}^{-1}$ then shows the model predictions are separated by many times the expected uncertainty.","pith_inferences":["Extension: feed-down from $\\psi(2S)$, $\\chi_{cJ}$, and $b$-hadrons is not modeled; computing the azimuthal modulations of those sources and subtracting them would show whether the discrimination survives in an inclusive $J/\\psi$ sample.","Extension: the same Fourier-projection test could be applied to $\\Upsilon$ leptoproduction, where the long-distance matrix-element hierarchy differs, to see whether the $\\cos(2\\psi)$ separation between $^1S_0^{[8]}$ dominance and equal-octet NRQCD is a general feature.","Extension: the calculation is done at leading order; repeating the region-$D$ analysis at next-to-leading order would show whether the $0.013$ and $0.004$ margins persist after radiative corrections."],"forward_implications":["If the color-singlet model is correct, region $D$ at $\\mathcal{L}=1000\\ \\mathrm{pb}^{-1}$ will show $A_{\\cos\\psi}=0.561\\pm0.013$; if the equal-color-octet NRQCD fit is correct, the same measurement gives $-0.246$, so a single run separates the two.","If the $^1S_0^{[8]}$-dominance picture is correct, $A_{\\cos2\\psi}=-0.203\\pm0.004$, versus $-0.0155$ for the equal-octet set, giving an independent test of the dominance hypothesis.","Because the modulations are ratios of integrated cross sections, most systematic uncertainties cancel, so the discrimination is statistics-limited and improves roughly as $\\sqrt{\\mathcal{L}}$.","The measurement is proposed for future $ep$ colliders, where the kinematics of region $D$ are experimentally accessible."],"supporting_citations":[{"why":"Supplies the NRQCD factorization that defines the color-singlet and color-octet long-distance matrix elements used throughout.","marker":"[16]"},{"why":"Provides the leptonic-tensor decomposition into the azimuthally dependent structures that produce the cos(ψ) and cos(2ψ) modulations.","marker":"[72]"},{"why":"Gives the leading-order NRQCD cross section for J/ψ leptoproduction at HERA on which the numerical calculation is based.","marker":"[73]"},{"why":"Supplies the GRV parton distribution functions used to evaluate the proton-side convolution.","marker":"[74]"},{"why":"Provides the charmonium wavefunction at the origin used as the color-singlet LDME value 1.16 GeV^3.","marker":"[75]"},{"why":"Together with Ref. [60], supplies the equal-order color-octet LDME set that yields Acosψ = -0.246.","marker":"[55]"},{"why":"Supplies the eta_c-data-constrained LDME set that, with Ref. [55], fixes the equal-octet NRQCD benchmark.","marker":"[60]"},{"why":"Supplies the 1S0[8]-dominant LDME set used for the Acos2ψ = -0.203 prediction.","marker":"[47]"}],"fun_headline_variants":["Azimuthal modulations in J/ψ production distinguish three models","Two cos(ψ) coefficients separate CS from CO mechanisms","J/ψ leptoproduction asymmetry: a model discriminator","Statistical analysis reveals J/ψ production mechanism via azimuthal asymmetry","cos(ψ) and cos(2ψ) in J/ψ leptoproduction tell models apart"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The predicted cross section describes direct $J/\\psi$ production, but the experimental event sample will contain feed-down from $\\psi(2S)$, $\\chi_{cJ}$, and $b$-hadron decays, which the paper does not model; if those sources have different azimuthal modulations, the measured asymmetries in region $D$ would be diluted relative to the model predictions.","fun_headline_variants_meta":{"raw":{"variants":["Azimuthal modulations in J/ψ production distinguish three models","Two cos(ψ) coefficients separate CS from CO mechanisms","J/ψ leptoproduction asymmetry: a model discriminator","Statistical analysis reveals J/ψ production mechanism via azimuthal asymmetry","cos(ψ) and cos(2ψ) in J/ψ leptoproduction tell models apart"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000748,"raw_usage":{"total_tokens":3461,"prompt_tokens":1204,"completion_tokens":2257,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":820,"completion_tokens_details":{"reasoning_tokens":2163}},"tokens_in":820,"tokens_out":2257,"duration_ms":16956,"temperature":1.0,"reasoning_tokens":2163,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:42:52.978278+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"At an $ep$ collider with $\\mathcal{L}=1000\\ \\mathrm{pb}^{-1}$, bin the $J/\\psi$ yield in $\\psi$ within region $D$ and fit $N_i = A v_1^{(i)} + B v_{\\cos\\psi}^{(i)} + C v_{\\cos2\\psi}^{(i)}$; if the best-fit $A_{\\cos\\psi}$ lies between $0.561$ and $-0.246$ with uncertainty comparable to the interval, or if $A_{\\cos2\\psi}$ comes out near $-0.016$ rather than $-0.203$ when the $^1S_0^{[8]}$-dominant set is assumed, the claimed discrimination is contradicted. A $\\psi(2S)$-vetoed subsample would directly test the feed-down dilution.","supporting_citations":[{"cited_title":"The leptonic current structure and azimuthal asymmetry in deeply inelastic scattering","cited_arxiv_id":"1701.08728","evidence_quote":"Provides the leptonic-tensor decomposition into the azimuthally dependent structures that produce the cos(ψ) and cos(2ψ) modulations."},{"cited_title":"Gluck, E","cited_arxiv_id":null,"evidence_quote":"Supplies the GRV parton distribution functions used to evaluate the proton-side convolution."},{"cited_title":"Reconciling charmonium production and polarization data within the nonrelativistic QCD framework","cited_arxiv_id":"1505.02675","evidence_quote":"Together with Ref. [60], supplies the equal-order color-octet LDME set that yields Acosψ = -0.246."}],"review_version":1}