{"id":"b03bd824-9238-41e2-8ffd-c321b9425427","arxiv_id":"1908.10154","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"A next-to-leading order fit to HERA diffractive DIS data that includes a fitted 1/Q^2 higher-twist term improves chi2/dof from 1.052 to 1.013 and changes the extracted gluon density, though the effect is not shown to be statistically robust.","lead":"This QCD analysis adds a fitted higher-twist correction to the theory of diffractive deep inelastic scattering and re-extracts diffractive parton densities from HERA data, reporting a small improvement in fit quality. The result matters because diffractive parton densities feed predictions for diffractive processes at the LHC, but the evidence for higher twists is weaker than the paper claims.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Twist-4 contribution to FL is asserted negligible after the beta<=0.80 cut without a quantitative estimate; if non-negligible, the fitted HT term in F2 and the claimed gluon suppression are biased.","rationale":"The paper does a competent NLO fit with a phenomenological 1/Q^2 term, and the Delta chi^2 improvement (390.46/371 to 376.08/371) is real in the narrow sense that adding three parameters improves the fit. This is independent support that the data at Q^2 down to 6.5 GeV^2 prefer some power correction in the adopted framework. However, the physical interpretation of that correction as 'higher twist' depends on excluding the known twist-4 piece in F_L. The manuscript flags this exclusion itself in Sec. II.D but never quantifies it; the beta<=0.80 cut is a plausible heuristic, not a demonstrated suppression. Since the extracted gluon density shifts dramatically (gamma_g from 0.741 to 2.400; alpha_g from 2.166 to 6.244), the PDF-level conclusion is sensitive to the treatment of subleading terms. The proposed check--adding the GB-L twist-4 term and refitting--would directly test whether the central claim survives. This does not move the reader's verdict: CONDITIONAL remains appropriate, with the condition being either a quantitative demonstration that twist-4 is negligible at the fitted kinematics or a full fit including it, plus uncertainties on h0, h1, h2 and a tempered novelty claim.","tokens_in":26499,"tokens_out":10003,"duration_ms":107961,"concrete_test":"Implement the twist-4 contribution to F_L from Golec-Biernat and Luszczak (Ref. [47]) and add it to Eq. (1) alongside the standard HT term, then re-fit the HT parameters and diffractive PDFs on the same datasets with the same beta<=0.80 and Q^2_min=6.5 GeV^2 cuts. If the resulting chi^2 improvement and the gluon suppression at z>0.3 relative to the no-HT fit persist within the Hessian uncertainties of h_i, the neglect is validated; if the HT parameters shift substantially or the suppression weakens, the claim is model-dependent and the 'first evidence' wording must be withdrawn.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. II.D states that a twist-4 contribution to the longitudinal diffractive structure function F_L, which is known from Refs. [47-49] to be important at large beta, 'can be strongly reduced and safely ignored' after the beta<=0.80 cut, and that the analysis is limited to the 'standard' HT term in Eq. (7). This is an assertion, not a derivation: no estimate is given of the residual twist-4 contribution at the actual fitted kinematic points with beta<=0.80 and Q^2>=6.5 GeV^2. Because the reduced cross section in Eq. (1) depends on F_L with a y-dependent weight, and because the fitted HT term multiplies F_2 only, an omitted F_L twist-4 term would be absorbed by the three parameters h0, h1, h2 (best fit h0=-25.8, h1=2.54, h2=-1.38). The central result--the improved chi^2 and the suppressed medium/large-z gluon in Fig. 4--is therefore only as secure as this untested neglect. The prior analysis of Ref. [47] explicitly found that including the twist-4 term changes the extracted diffractive gluon, so the burden is on the authors to show that the beta<=0.80 cut removes it in their kinematic region.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents an NLO QCD fit to HERA diffractive DIS data (H1 LRG-2011, H1 LRG-2012, and the H1/ZEUS combined dataset) using Regge factorization for the Pomeron and Reggeon contributions, the FONLL general-mass variable-flavor-number scheme for heavy quarks, and the APFEL/MINUIT toolchain. The claimed novelty is the inclusion of a phenomenological higher-twist (HT) correction to the diffractive structure function F2 of the form F2 = F2_LT (1 + C_HT(x)/Q^2) with C_HT(x) = h0 x^h1 (1 + h2 x), applied only to F2. The authors report that including this term improves the global chi^2 from 390.46 to 376.08 for the same 371 degrees of freedom, motivates lowering Q^2_min from 8.5 to 6.5 GeV^2 on the basis of a chi^2 scan, and leads to a gluon density that is suppressed at medium and large z while the quark density is enhanced at small z.","tokens_in":26861,"tokens_out":4415,"duration_ms":47955,"significance":"If the central claim is correct, the paper would justify extending diffractive PDF fits to lower Q^2 and would alter the extracted diffractive gluon at medium and large z, which is relevant for diffractive phenomenology at HERA and the LHC. The fit strategy is standard and the paper is transparent about datasets, parameterizations, and numerical tools, which is a strength. However, the evidence for the HT contribution is entirely in-sample: the three HT parameters are fitted to the same data used to claim the effect, the Q^2_min cut is chosen from the same chi^2 scan, the HT parameters are quoted without uncertainties, and the neglect of the twist-4 contribution to F_L is asserted rather than demonstrated. These issues are load-bearing for the abstract's claim of 'first evidence' for HT effects in diffractive DIS, so the manuscript requires substantial revision before that claim can be accepted.","major_comments":[{"comment":"The three higher-twist parameters h0, h1, and h2 are fitted to the data, yet Table II quotes them without uncertainties: h0 = -25.798, h1 = 2.536, h2 = -1.381. Since the paper's central claim is that these parameters are determined by the diffractive DIS data and improve the fit, the absence of uncertainties and of correlations with the PDF parameters makes it impossible to judge whether the HT term is statistically meaningful. The authors should provide Hessian uncertainties for h0, h1, h2, and ideally a correlation matrix with the other fitted parameters.","section":"Sec. II.D, Eqs. (7)–(8) and Table II"},{"comment":"The statement that the twist-4 contribution to F_L from longitudinally polarized photons 'can be strongly reduced and safely ignored' after the beta <= 0.80 cut is an assertion, not a quantitative estimate. The reduced cross section in Eq. (1) depends on F_L through a y-dependent coefficient, while the fitted HT correction in Eq. (7) multiplies only F2. A residual twist-4 F_L term would therefore be partially absorbed by h0, h1, h2, biasing the extracted gluon. Since Ref. [47] found that including twist-4 changes the diffractive gluon, the authors should quantify the residual twist-4 contribution at the fitted beta <= 0.80 and Q^2 >= 6.5 GeV^2 points, or perform a robustness fit that includes a twist-4 F_L term.","section":"Sec. II.D and Sec. III.B"},{"comment":"The choice Q^2_min = 6.5 GeV^2 is selected from a chi^2 scan performed on the same data that are then used to claim that HT effects improve the description. This data-driven cut selection makes the reported chi^2 improvement an in-sample statistic and can bias the apparent gain. The authors should provide an out-of-sample test, for example fitting at Q^2_min = 8.5 GeV^2 and then predicting the data below that cut, or using a hold-out subset of the combined dataset, and should separately show the stability of the HT parameters under this procedure.","section":"Sec. III.B, Fig. 3"},{"comment":"The improvement in global chi^2 from 390.46 to 376.08 for the same 371 dof is modest, especially given that it is achieved by adding three free parameters. The authors should report a significance test (e.g., an F-test or likelihood-ratio test) and should show that the improvement is not driven by a small number of data points; the largest per-dataset gain appears in H1-LRG-12, so pull distributions for that dataset would be informative. Without such a test, the abstract's wording that HT effects 'can improve the description of the data' overstates the strength of the evidence.","section":"Sec. IV, Table III"}],"minor_comments":[{"comment":"The text contains several typographical issues: 'Difrractive' in the Section III title, 'heay quark' in Section II.C, 'This analysis are enriched' in the abstract, and 'HK19-DDPF' instead of 'HK19-DPDF' in the introduction.","section":"Global"},{"comment":"The sentence 'hi should be determined along with the fit parameters and then keep fixed' is ambiguous; the authors should clarify whether the HT parameters are profiled, marginalized, or simply frozen at their best-fit values when the PDF uncertainties are computed.","section":"Sec. II.D"},{"comment":"Several theory-to-data comparisons show data with Q^2 = 2.5 and 5.09 GeV^2, which lie below the chosen Q^2_min = 6.5 GeV^2 and are excluded from the fit. The text should state explicitly whether these low-Q^2 panels are genuine predictions of the fitted model or are included only as a qualitative illustration of the HT extrapolation.","section":"Figs. 13–15"},{"comment":"The table lists 526 data points before cuts, and the text later says 499 points are included after cuts, implying 27 points excluded; this arithmetic is consistent, but the breakdown of how many points are removed by the beta, M_X, and Q^2 cuts separately would be helpful.","section":"Table I"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a useful phenomenological study that addresses a real gap in diffractive PDF fits, but the headline claim of evidence for higher-twist effects is stronger than the analysis supports. The main fixes needed are an uncertainty estimate for the HT parameters, a robustness check against the omitted twist-4 F_L term, and an out-of-sample validation of the Q^2_min choice. These are achievable within the manuscript's scope, so I would not reject the paper; with those additions it could be publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a solid, workmanlike diffractive-PDF fit. What is genuinely new is narrow: applying the standard inclusive-DIS 1/Q^2 higher-twist correction (Eqs. 7-8) inside a modern NLO fit to the H1/ZEUS combined data plus the H1 LRG sets, with heavy-quark scheme handled by APFEL/FONLL-B GM-VFNS. The fit machinery is standard and the paper is transparent about datasets and cuts. The improvement from adding three HT parameters is real but modest: chi2/dof goes from 390.46/371 = 1.052 to 376.08/371 = 1.013, with most gains in H1-LRG-12 and the combined set. The extracted gluon is suppressed at medium/large z after including HT, and that is the main phenomenological result.\n\nThe paper deserves serious referee time, but there are three soft spots, in increasing order of importance. First, the abstract says 'for the first time' HT effects are included in diffractive DIS; that is wrong. Ref [47] (Golec-Biernat and Luszczak 2007) already fitted diffractive PDFs with higher twist and found twist-4 matters for the gluon. The current paper does discuss [47], so it is an overstatement rather than ignorance, but it needs fixing.\n\nSecond, the HT parameters h0, h1, h2 are quoted as numbers with no uncertainties. For a phenomenological term whose whole claim is to describe data, you need their errors, ideally a profile or a correlation with the PDF parameters. Related: the Q^2_min = 6.5 GeV^2 cut is chosen from a chi2 scan (Fig 3) on the same data, and the improvement is in-sample. A cross-validation or an information criterion would make the 'HT improves the fit' claim much stronger.\n\nThird and most important: the paper asserts that the twist-4 contribution to F_L is 'strongly reduced and safely ignored' after the beta <= 0.80 cut, but gives no quantitative estimate. The reduced cross section depends on F_L with a y-dependent weight; an omitted twist-4 F_L term can be absorbed by the three fitted HT parameters in F2 and bias the gluon. The stress-test note is right that the burden is on the authors, especially because Ref [47] found the twist-4 term changes the extracted gluon. This does not kill the paper, but it means the strong version of the claim—'evidence for HT'—is not established. What is established is that adding three parameters improves an in-sample fit at low Q^2.\n\nBottom line: the weak claim is defensible; the strong claim is not, in current form. This paper is for people doing diffractive PDF fits and anyone who wants to use low-Q^2 HERA diffractive data; it will not change LHC phenomenology overnight. It should be refereed, with required revisions on novelty, parameter uncertainties, out-of-sample checks, and a quantitative twist-4 estimate.","headline":"A competent NLO diffractive-PDF fit with a fitted higher-twist term, but the 'first-time' novelty is overstated and the twist-4 neglect needs quantitative support.","tokens_in":27450,"tokens_out":2662,"would_cite":true,"duration_ms":27438,"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":"Including a fitted higher-twist term proportional to $1/Q^2$ in the diffractive structure function $F_2$ improves the NLO QCD description of HERA diffractive deep-inelastic scattering data and changes the extracted gluon density.","keywords":["diffractive deep inelastic scattering","higher twist effects","diffractive parton distribution functions","NLO QCD analysis","Hessian uncertainties","HERA","Pomeron flux","variable-flavor-number scheme"],"falsifier":"Use the published PDF sets to predict the diffractive reduced cross sections at $Q^2=2.5$ and $5.09$ GeV$^2$ shown in the theory-to-data comparisons; those points were excluded from the fit, so the HT model either reproduces them better than the leading-twist model or it does not.","tokens_in":2074,"feed_emoji":"⚛️","tokens_out":2922,"duration_ms":104695,"temperature":0.7,"pith_summary":"This paper sets out to establish that nonperturbative higher-twist (HT) effects are visible in diffractive deep inelastic scattering and should be included when extracting diffractive parton distribution functions (PDFs). The authors perform an NLO QCD fit to all available HERA diffractive datasets with a fitted correction of the form $F_2 = F_2^{\\rm LT}\\,(1 + C_{\\rm HT}(x)/Q^2)$, where $C_{\\rm HT}(x)=h_0 x^{h_1}(1+h_2 x)$; the fit quality improves from $\\chi^2/{\\rm dof}=1.052$ to $1.013$, and the data can be described down to $Q^2=6.5$ GeV$^2$ rather than the usual $8.5$ GeV$^2$ cut. Including the HT term changes the extracted gluon density, suppressing it at medium and large momentum fraction $z$ while slightly enhancing it at small $z$. A sympathetic reader would take the conclusion to be that leading-twist-only analyses of low-$Q^2$ diffractive data have been missing a power correction that matters for the shape of the gluon.","feed_headline":"Fitted 1/Q^2 term improves diffractive DIS fits, suppresses gluon","feed_subtitle":"Adding higher-twist corrections lets NLO QCD reproduce HERA diffraction data down to Q^2 = 6.5 GeV^2.","key_machinery":"The load-bearing object is the phenomenological higher-twist factor applied to the diffractive $F_2$: $F_2(x,Q^2) = F_2^{\\rm LT}(x,Q^2)\\,(1 + C_{\\rm HT}(x)/Q^2)$, with $C_{\\rm HT}(x)=h_0 x^{h_1}(1+h_2 x)$. This is a $1/Q^2$ power correction whose three parameters are fitted simultaneously with the diffractive PDFs; the $x^{h_1}$ factor makes the correction grow at large $x$, and the $h_2$ term allows behavior at small $x$. The argument also relies on the $\\beta\\le 0.80$ cut to justify ignoring twist-4 contributions from longitudinal virtual photons, which would otherwise dominate the large-$\\beta$ region.","core_discovery":"The central claim is that a simple power-suppressed correction to the diffractive structure function $F_2$ is required to describe HERA diffractive DIS data at NLO accuracy. The paper shows that the diffractive reduced cross section is better reproduced when the leading-twist structure function is multiplied by $(1 + C_{\\rm HT}(x)/Q^2)$ with $C_{\\rm HT}(x)=h_0 x^{h_1}(1+h_2 x)$, and the three HT parameters are fitted together with the diffractive PDFs, giving $h_0=-25.798$, $h_1=2.536$, and $h_2=-1.381$. With a $\\beta\\le 0.80$ cut and $Q^2_{\\rm min}=6.5$ GeV$^2$, the fit keeps 499 of 526 data points; the same fit without the HT term gives $\\chi^2/{\\rm dof}=1.052$, while with the HT term it gives $1.013$. The largest improvement comes from the largest rapidity-gap dataset, and the extracted gluon density at the input scale $Q_0^2=1.8$ GeV$^2$ is reduced at medium and large $z$. The authors conclude that HT effects are sizable at large $x$ and low $Q^2$ and that they should be included in diffractive PDF determinations.","pith_inferences":["A direct extension would apply the same $C_{\\rm HT}(x)/Q^2$ form to the longitudinal diffractive structure function $F_L^D$, not only to $F_2$; high-$y$ data could then discriminate between a genuine power correction and an effective rescaling of $F_2$.","If the gluon suppression at medium and large $z$ is real, diffractive dijet rates at moderate $Q^2$ should fall below current NLO predictions based on leading-twist-only fits, a cross-check that existing HERA dijet samples could perform.","The fitted $C_{\\rm HT}(x)$ has a negative overall scale and an $x$-dependent shape; testing alternative forms such as $a(1-x)^p/Q^2$ or a twist-4-motivated term would show whether the improvement is robust or an artifact of the chosen parametrization.","Releasing the $\\beta\\le 0.80$ cut and allowing an explicit twist-4 contribution would reveal whether the fitted higher-twist term is the standard $1/Q^2$ correction or a stand-in for missing large-$\\beta$ physics."],"forward_implications":["Diffractive PDF fits can safely include data with $Q^2$ down to $6.5$ GeV$^2$ instead of $8.5$ GeV$^2$, adding more data points without degrading the fit quality.","The extracted diffractive gluon density at the input scale is suppressed at medium and large $z$ when HT effects are included, so predictions for diffractive dijet and charm production built from this PDF set will differ from leading-twist-based sets.","The combined HERA data improve from $\\chi^2/{\\rm dof}=1.052$ to $1.013$ when the HT term is added, with the largest single reduction coming from the largest rapidity-gap sample.","The HT correction affects theoretical predictions most at low $Q^2$ and at medium $\\beta$ values, gradually shifting to larger $\\beta$ as $Q^2$ increases.","Accounting for HT effects changes the shape and size of the extracted diffractive quark and gluon densities while leaving the size of their uncertainty bands largely unchanged."],"supporting_citations":[{"why":"Provides the factorization theorem that lets diffractive cross sections be written as convolutions of coefficient functions and diffractive PDFs.","marker":"[10]"},{"why":"Supplies the high-precision combined HERA dataset that anchors the fit.","marker":"[15]"},{"why":"Provides the largest rapidity-gap dataset and the biggest single improvement in fit quality when HT effects are included.","marker":"[25]"},{"why":"Supplies additional HERA rapidity-gap data at three center-of-mass energies used together with the combined dataset.","marker":"[24]"},{"why":"Defines the parameterization and flux conventions adopted here and provides the comparison diffractive PDF set.","marker":"[13]"},{"why":"Sets the baseline leading-twist diffractive PDF analysis whose functional forms and kinematic cuts are extended.","marker":"[16]"},{"why":"Provides the earlier diffractive analysis from the other HERA experiment used for comparing quark and gluon densities.","marker":"[17]"},{"why":"Defines the general-mass variable-flavor-number scheme used to compute heavy-quark contributions to the structure functions.","marker":"[38]"},{"why":"Analyzes twist-4 contributions and justifies ignoring them through the $\\beta\\le 0.80$ cut.","marker":"[47]"}],"fun_headline_variants":["Higher-twist term sharpens HERA diffractive PDF fits","First NLO diffractive PDFs with higher-twist corrections","Power-suppressed term in diffractive DIS: better chi2, smaller gluon","HERA diffractive data favor higher-twist effect at low Q2","NLO diffractive PDFs with higher twist: fit quality up, gluon down"],"cache_read_input_tokens":29312,"weakest_assumption_plain":"The whole fit improvement rests on the assumption that the higher-twist correction takes exactly the factorized form $F_2 = F_2^{\\rm LT}\\,(1 + h_0 x^{h_1}(1+h_2 x)/Q^2)$, acts only on $F_2$, and that target-mass and twist-4 effects are negligible after the $\\beta\\le 0.80$ and $Q^2\\ge 6.5$ GeV$^2$ cuts.","fun_headline_variants_meta":{"raw":{"variants":["Higher-twist term sharpens HERA diffractive PDF fits","First NLO diffractive PDFs with higher-twist corrections","Power-suppressed term in diffractive DIS: better chi2, smaller gluon","HERA diffractive data favor higher-twist effect at low Q2","NLO diffractive PDFs with higher twist: fit quality up, gluon down"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000257,"raw_usage":{"total_tokens":1696,"prompt_tokens":1183,"completion_tokens":513,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":799,"completion_tokens_details":{"reasoning_tokens":416}},"tokens_in":799,"tokens_out":513,"duration_ms":5501,"temperature":1.0,"reasoning_tokens":416,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:51:26.628443+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Use the published PDF sets to predict the diffractive reduced cross sections at $Q^2=2.5$ and $5.09$ GeV$^2$ shown in the theory-to-data comparisons; those points were excluded from the fit, so the HT model either reproduces them better than the leading-twist model or it does not.","supporting_citations":[{"cited_title":"Factorization in hard diffraction","cited_arxiv_id":"hep-ph/0107252","evidence_quote":"Sets the baseline leading-twist diffractive PDF analysis whose functional forms and kinematic cuts are extended."},{"cited_title":"Diffractive parton distributions from H1 data","cited_arxiv_id":"hep-ph/0609273","evidence_quote":"Defines the general-mass variable-flavor-number scheme used to compute heavy-quark contributions to the structure functions."}],"review_version":1}