{"id":"fe4f55dd-570a-4eed-bcf4-6663829672ab","arxiv_id":"1908.08398","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Exclusive J/psi photoproduction at HERA is consistent with global PDFs at NLO when the optimal scale and a Q0 subtraction are used, enabling low-x gluon extraction.","lead":"This paper shows that exclusive J/psi photoproduction data from HERA can be described at next-to-leading order in collinear factorization using existing global PDFs, once two theoretical corrections are applied. It argues these data can constrain the gluon PDF at very low x, down to about 10^-6.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim rests on a large Q0-subtraction of the NLO coefficient function that is not specified in this paper; at Q0 ≈ m_c the subtraction is O(1) and only after it is applied do the predictions become scale-stable and match HERA.","rationale":"The reader's weakest assumption and my concern coincide: the Q0 subtraction in Sec. 4.2 is the central mechanism that makes the NLO series appear stable and makes the HERA data pass. I agree with the CONDITIONAL verdict. The paper's logic is coherent, the optimal-scale argument has independent support in [25], and the Shuvaev transform is a known construction. The issue is not internal inconsistency; it is that the decisive subtraction is imported from a self-cited paper and is large enough that it could dominate the result. Footnote 4 even acknowledges that Q0 is not identical to the parametrization scale for some of the PDF sets, which weakens the 'already in the input' claim. I would keep the reader's verdict unchanged rather than reject: the claim is plausible and testable, but it should be accepted only with access to the subtracted coefficient function and a quantitative comparison. If an independent reimplementation reproduces the curves, the concern is retired; if it does not, the central claim would become unverified.","tokens_in":11299,"tokens_out":8594,"duration_ms":96612,"concrete_test":"Derive C_NLO_rem explicitly from [29] and recompute the imaginary amplitude in Eq. (6) for NNPDF3.0 at W = 100 GeV with two implementations: (i) the subtraction exactly as described in the cited code, and (ii) an independent implementation that cuts the loop integral at |k_t|^2 = Q0^2 before applying the MS-bar subtraction. If the two implementations disagree at the level of the experimental precision, or if varying Q0 between 1.0 and 2.0 GeV shifts the HERA curve by more than the data uncertainty, the split between 'already in the PDF' and 'NLO coefficient function' is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.2 and Eq. (4) define the prediction as C_LO⊗GPD(µF) + C_NLO_rem⊗GPD(µf), where C_NLO_rem is obtained by subtracting the k_t<Q0 contribution from the full NLO coefficient function. The paper gives no explicit formula for C_NLO_rem; it is adopted from the self-cited [29]. This omission is load-bearing because the subtraction is numerically dominant: with Q0 = m_c = M_ψ/2, the ratio Q0^2/m_c^2 = 1 is not a small power correction, and the left panel of Fig. 2 shows the unsubtracted NLO amplitude is comparable in size to the LO term and changes sign with µf. The right panel of Fig. 2, the scale stability claimed in Fig. 4, and the HERA agreement all appear only after this large term is removed. The physical justification in Sec. 4.2 is that the k_t<Q0 part of Fig. 1(b) is already contained in the input PDFs/GPDs. But the input at Q0 = m_c is not a common fitted input for the PDF sets used: NNPDF3.0 starts at Q0 = 1 GeV, CT14 at 1.3 GeV, and the MMHT fit boundary is Q^2 = 2 GeV^2 (footnote 4). The input at 1.55 GeV is therefore itself partly the result of DGLAP evolution, so the double counting that is subtracted is scheme-dependent. In the absence of an explicit expression, it is possible that the subtraction is removing the large negative NLO term rather than a genuine DGLAP-generated contribution, which would make the apparent consistency with HERA an artifact of this bookkeeping.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that exclusive J/ψ photoproduction, measured at HERA and by LHCb in ultraperipheral pp collisions, can be described in NLO collinear factorization using the PDFs from existing global analyses, provided two ingredients are added: (i) the Shuvaev transform to convert conventional PDFs into the relevant GPDs, and (ii) an 'optimal scale' µF = Mψ/2 that resums double-logarithmic terms, together with a subtraction of the kt < Q0 part of the NLO coefficient function to avoid double counting with the DGLAP-evolved input PDFs. The authors show that after these ingredients the prediction becomes more stable under scale variations and that three global PDF sets (NNPDF3.0, MMHT2014, CT14) reproduce the HERA data reasonably well, while the much larger spread at LHCb energies reflects the poorly known low-x gluon. They conclude that exclusive J/ψ LHCb data can directly constrain the gluon PDF over 10^-6 < x < 10^-2 at a fixed low scale.","tokens_in":11702,"tokens_out":4517,"duration_ms":44773,"significance":"If the central claim is correct, this is a useful step toward including exclusive J/ψ data in global PDF fits: it would open a genuinely new low-x, low-scale kinematic window and would resolve the long-standing scale-instability problem of NLO exclusive vector-meson production. The use of external HERA and LHCb data makes the test non-circular, and the comparison across three independent PDF sets is a strength. However, the decisive technical ingredient, the subtracted NLO coefficient function C_NLO_rem, is not specified in this paper but is imported from a self-cited reference, and the paper's own footnote 5 concedes that the central global PDFs fail to describe the HERA data at the lowest x values. The evidence for 'consistency within uncertainties' is therefore not as complete as the abstract suggests.","major_comments":[{"comment":"The central object C_NLO_rem, defined by subtracting the kt < Q0 contribution from the NLO coefficient function, is never explicitly given in this paper. Section 4.2 states only that 'we use the NLO correction C_NLO_rem for J/ψ photoproduction excluding the contribution coming from the low virtuality domain', citing reference [29]. Because the subtraction is numerically large — the left panel of Fig. 2 shows the unsubtracted NLO term comparable in size to the LO term and even changing sign with µf — the agreement with HERA in Fig. 4 and the apparent scale stability in the right panel of Fig. 2 rest entirely on this unstated expression. The paper is not self-contained at a load-bearing point, and the reader cannot check whether the subtraction removes a genuine DGLAP-generated contribution or simply cancels a large negative NLO term. A derivation, or at least an explicit formula for C_NLO_rem, must be included.","section":"§4.2 and Eq. (4)"},{"comment":"The physical justification of the Q0 subtraction is scheme-dependent in a way the paper does not address. The argument that the |l^2| < Q0^2 part of Fig. 1(b) is 'already included in the input gluon GPD at Q0' presumes that Q0 is the parametrization scale of the input PDFs. But with Q0 = µF = mc = Mψ/2 (Eq. (9)), this is not true for the sets used: NNPDF3.0 starts at Q0 = 1 GeV, CT14 at 1.3 GeV, and MMHT2014 fits at Q^2 > 2 GeV^2 (footnote 4). At Q0 ≈ 1.55 GeV the 'input' is itself already partly a product of DGLAP evolution from a lower scale, so the double-counting subtraction is not uniquely defined. The paper should explain how C_NLO_rem is defined when Q0 differs from the fit's true input scale, and should test the sensitivity of the HERA conclusions to this choice.","section":"§4.2, Eq. (9), and footnote 4"},{"comment":"The abstract and Section 5.1 claim that the existing global PDFs are 'consistent with the data within their uncertainties', but the quantitative evidence is incomplete. Fig. 4 shows only central predictions for MMHT2014 and CT14, with no uncertainty bands, and Fig. 6 provides a 1σ band only for NNPDF3.0. Moreover footnote 5 concedes that 'when x <~ few x 10^-4 the central global partons fail to describe the HERA data.' Since the conclusion is that the data are consistent within PDF uncertainties, the paper should provide a quantitative comparison — e.g., χ² values or uncertainty bands for all three PDF sets over the HERA x range — rather than relying on visual inspection of central curves.","section":"§5.1, Fig. 4, Fig. 6, and footnote 5"}],"minor_comments":[{"comment":"The caption says 'with µF = mc before (left panel) and after (right panel) the double counting correction', but the text and Eq. (4) distinguish the fixed resummation scale µF = mc from the varying factorization scale µf. It would be clearer to state explicitly which scale is varied in each panel and which term is plotted.","section":"Figure 2 caption"},{"comment":"The symbols µF, µf, µ0 and mc are used interchangeably at several points (e.g., Section 4.1 says 'µF = µ0 = Mψ/2' while Eq. (9) sets 'Q0 = µF = mc'). A consistent notation, with a single symbol for the optimal scale, would prevent confusion.","section":"Notation in Eq. (4) and surrounding text"},{"comment":"The text says 'the above choice of Q0 and µF give a stable theoretical prediction also when the scales µf and µR are varied', but only µf variations are shown in Figs. 3 and 4; the dependence on µR is not displayed separately.","section":"Section 5.1"},{"comment":"Reference [34] appears in the text as 'Hoodhboy' but the correct name is Hoodbhoy; please correct the typo.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the reliance on reference [29] for the subtracted NLO coefficient function. If that object is not fully documented, the numerical results are not independently verifiable. I would ask the editor to require that the derivation, or at least an explicit expression and a numerical check of the subtraction, be included in the revised version. The paper's own footnote 5 weakens the 'consistency within uncertainties' claim; the authors should either strengthen the quantitative comparison or temper the abstract."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear —,\n\nThe paper you want to know about: Flett et al. claim the first NLO collinear-factorization description of exclusive J/ψ photoproduction at HERA using three global PDF sets (NNPDF3.0, MMHT2014, CT14), and argue that LHCb exclusive data can measure the gluon down to x ~ 3×10^-6 at a fixed low scale. Having read it, the claim holds up as far as it goes. The HERA description in Fig. 4 is plausible, and the scale-stability argument is honestly displayed: the left and right panels of Fig. 2 show exactly what the Q0 subtraction does. The novelty is the combined NLO treatment against the full HERA and LHCb data sets; the ingredients (Shuvaev transform, optimal scale μ_F = M_ψ/2, Q0 subtraction) come from the authors' earlier papers. That is a real demonstration, not a repackaging.\n\nCredit where due: the paper is candid that the Q0 subtraction is a large power correction, that the unsubtracted NLO term is comparable to the LO term and sign-changing, and that the subtraction is what restores scale stability. Footnote 4 even acknowledges that the PDF sets' actual input scales differ from Q0 = m_c. That is more honest than most papers in this area.\n\nThe soft spot is the weight the subtraction carries. The subtracted NLO coefficient function C_NLO_rem is not given in this paper; it is taken wholesale from ref. [29]. Because the subtraction is O(1) and flips the sign of the NLO amplitude, a reader cannot readily check whether it removes genuine DGLAP-generated physics or simply the large negative loop term. The scheme-dependence is real: NNPDF3.0 starts at 1 GeV, CT14 at 1.3 GeV, MMHT at Q^2 = 2 GeV^2, so 'already in the input PDFs' is not clean at Q0 = m_c. An explicit expression for C_NLO_rem and a comparison of the subtraction at each PDF set's actual input scale would settle it. A second, minor gap: the HERA agreement in Fig. 4 is eyeballed — no χ^2 or PDF-error band for the three sets — which is thin for a paper titled 'How to include [...] in global PDF analyses'. The LHCb data points also rest on the group's earlier gap-survival extraction, another model dependence to keep in mind when the proposal becomes a fit.\n\nWho this is for: low-x PDF people, GPD people, and vector-meson photoproduction people. It deserves a serious referee — the framework is coherent, the claim is modest, and the demonstration is in principle reproducible. I would send it to review and push for the details of the subtraction and a quantified comparison. It is not a completed global fit, but it is a legitimate step worth engaging with.","headline":"Plausible NLO collinear description of HERA exclusive J/ψ data with global PDFs, but the load-bearing Q0 subtraction is borrowed from the authors' earlier paper and left unspecified here — worth engaging, needs a closer look at the bookkeeping.","tokens_in":12270,"tokens_out":6417,"would_cite":true,"duration_ms":56143,"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 shows that exclusive J/psi photoproduction, calculated at NLO in collinear QCD with an optimal scale plus a double-counting subtraction, is consistent with HERA data and can let LHCb data directly determine the gluon PDF at very…","keywords":["exclusive J/psi photoproduction","gluon PDF at small x","NLO collinear factorization","double counting","Q0 subtraction","optimal scale","generalised parton distributions","LHCb ultraperipheral collisions"],"falsifier":"Apply the same optimum-scale plus $Q_0$-subtraction prescription to exclusive $\\Upsilon$ photoproduction, where the ratio $Q_0^2/M_\\Upsilon^2$ is much smaller than for $J/\\psi$; if the advertised scale stability and agreement with data do not survive, the success for $J/\\psi$ is tied to the charm-mass scale rather than to the general formalism.","tokens_in":11115,"feed_emoji":"⚛️","tokens_out":8106,"duration_ms":75334,"temperature":0.7,"pith_summary":"The paper tries to remove the obstacles that kept exclusive $J/\\psi$ photoproduction out of global fits of proton parton distributions. It argues that the large, sign-changing NLO corrections seen in earlier collinear-factorization calculations are largely artifacts of two mishandled pieces: unresummed double logarithmic terms and double counting between the NLO coefficient function and the input PDFs. Choosing the factorization scale $\\mu_F = M_\\psi/2$ in the leading-order term resums the double logs, and subtracting the $k_t<Q_0$ part of the NLO coefficient function removes the double counting. With those two steps, three existing global PDF sets describe the HERA exclusive $J/\\psi$ data within uncertainties, and the precise LHCb data can determine the gluon PDF over $10^{-6}<x<10^{-2}$ at a fixed low scale. This matters because the low-$x$ gluon is currently unconstrained by data and controls much of the high-energy behaviour of QCD.","feed_headline":"J/psi data can map the gluon PDF down to x = 10^-6","feed_subtitle":"A scale choice and a double-counting subtraction reconcile HERA and LHCb with NLO collinear QCD.","key_machinery":"The mechanism is a two-part prescription applied to the collinear factorization formula for $\\gamma p\\to J/\\psi p$. The first part is the optimum-scale choice $\\mu_F=M_\\psi/2$ in the leading-order term, which moves the $(\\alpha_s \\ln\\mu_F^2 \\ln(1/x))^n$ double logarithms into the incoming PDFs and resums them. The second part is the $Q_0$ subtraction: the NLO coefficient function is computed with the low-transverse-momentum region $k_t<Q_0$ removed, because that region is already present in the PDFs at the input scale $Q_0$. The Shuvaev transform, which reconstructs generalized parton distributions from integrated PDFs at small skewness, supplies the bridge from the exclusive amplitude to the usual gluon PDF.","core_discovery":"On the paper's own terms, the central discovery is that the previous failure of NLO collinear QCD to describe exclusive $J/\\psi$ photoproduction at HERA was not a sign that the process needs an intrinsically different formalism. The apparent instability came from double logarithms of $x$ and from double counting between the NLO coefficient functions and the DGLAP-evolved input PDFs. Fixing the factorization scale in the LO term to $\\mu_F=M_\\psi/2$ resums the double logarithms, while removing the $k_t<Q_0$ contribution from the NLO coefficient functions compensates for what the PDF input already contains. After both corrections, the LO plus NLO amplitude is stable under scale variation, the quark NLO term becomes negligible, and the gluon GPD extracted from ordinary PDFs describes the HERA data and gives definite predictions for LHCb. The paper states this is the first successful description of the HERA $J/\\psi$ data within NLO collinear factorization using global PDFs.","pith_inferences":["One implication the authors do not pursue explicitly: the same two-step prescription should serve as a template for other exclusive vector-meson observables, with the size of the power correction set by $Q_0^2/M_V^2$.","A natural next test is an actual global fit that includes the HERA and LHCb $J/\\psi$ data under this prescription; the paper demonstrates consistency but does not perform such a fit.","If the extracted low-$x$ gluon at $\\mu_F\\simeq 1.5$ GeV were to disagree with extrapolations from DGLAP evolution, that would be evidence for saturation or higher-twist effects that the paper treats as absorbed into the input PDFs."],"forward_implications":["HERA exclusive $J/\\psi$ data can be included in future global PDF analyses without invoking a special non-collinear treatment.","LHCb ultraperipheral $J/\\psi$ data probe the gluon PDF in the interval $10^{-6}<x<10^{-2}$ at the low scale $\\mu_F=M_\\psi/2$, a kinematic region no current global fit constrains directly.","Because the LHCb data are more precise than the current gluon uncertainty at low $x$, including them should sharply reduce the low-$x$ gluon PDF uncertainty.","After the $Q_0$ subtraction the NLO quark contribution is practically negligible, making the observable an essentially pure gluon probe."],"supporting_citations":[{"why":"One of the three global PDF sets tested; its low-x gluon uncertainty is shown to exceed the LHCb data errors.","marker":"[1]"},{"why":"One of the three global PDF sets tested; provides a second independent gluon input for the comparison.","marker":"[2]"},{"why":"One of the three global PDF sets tested with the Q0-subtracted NLO coefficient functions.","marker":"[3]"},{"why":"LHCb exclusive J/psi data, the precise low-x dataset the paper wants to use for PDF determination.","marker":"[8]"},{"why":"The Shuvaev transform relating low-x GPDs to integrated PDFs, connecting the exclusive amplitude to the PDFs.","marker":"[16]"},{"why":"Procedure for extracting the gamma-p cross section from LHCb pp exclusive data including gap survival.","marker":"[17]"},{"why":"HERA J/psi photoproduction data that the predictions must reproduce.","marker":"[18]"},{"why":"NLO collinear coefficient functions for J/psi photoproduction; their scale dependence and double counting are the problem addressed.","marker":"[19]"},{"why":"Establishes the optimal scale mu_F = M_psi/2 that resums the double logarithmic corrections.","marker":"[25]"},{"why":"Introduces the Q0 subtraction to avoid double counting between the NLO coefficient functions and the DGLAP input PDFs.","marker":"[29]"}],"fun_headline_variants":["Exclusive J/psi data pin gluon PDF down to x=1e-6","Scale and subtraction fix J/psi fit to low-x gluon","J/psi photoproduction now probes gluon at x=1e-6","HERA and LHCb agree on gluon with NLO, low-x","Small-x gluon from J/psi after NLO double-counting fix"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction stands or falls on the claim that the low-transverse-momentum part of the NLO correction is already present in the input PDFs, so subtracting it removes double counting rather than discarding real physics.","fun_headline_variants_meta":{"raw":{"variants":["Exclusive J/psi data pin gluon PDF down to x=1e-6","Scale and subtraction fix J/psi fit to low-x gluon","J/psi photoproduction now probes gluon at x=1e-6","HERA and LHCb agree on gluon with NLO, low-x","Small-x gluon from J/psi after NLO double-counting fix"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000194,"raw_usage":{"total_tokens":1332,"prompt_tokens":899,"completion_tokens":433,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":515,"completion_tokens_details":{"reasoning_tokens":328}},"tokens_in":515,"tokens_out":433,"duration_ms":4793,"temperature":1.0,"reasoning_tokens":328,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:40:10.088602+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Apply the same optimum-scale plus $Q_0$-subtraction prescription to exclusive $\\Upsilon$ photoproduction, where the ratio $Q_0^2/M_\\Upsilon^2$ is much smaller than for $J/\\psi$; if the advertised scale stability and agreement with data do not survive, the success for $J/\\psi$ is tied to the charm-mass scale rather than to the general formalism.","supporting_citations":[{"cited_title":"Exclusive $J/\\psi$ production at the LHC in the $k_T$ factorization approach","cited_arxiv_id":"1611.03711","evidence_quote":"Procedure for extracting the gamma-p cross section from LHCb pp exclusive data including gap survival."},{"cited_title":"Exclusive $J/\\psi$ and $\\Upsilon$ photoproduction and the low $x$ gluon","cited_arxiv_id":"1507.06942","evidence_quote":"Establishes the optimal scale mu_F = M_psi/2 that resums the double logarithmic corrections."},{"cited_title":"The exclusive $J/\\psi$ process at the LHC tamed to probe the low $x$ gluon","cited_arxiv_id":"1610.02272","evidence_quote":"Introduces the Q0 subtraction to avoid double counting between the NLO coefficient functions and the DGLAP input PDFs."}],"review_version":1}