{"id":"aaf8aa2f-ee5d-4a84-ba13-b03e2f88b0ba","arxiv_id":"2505.17608","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":26,"one_line_summary":"Measurement of the down-to-up quark PDF ratio from LHC forward-backward asymmetry finds deviations from PDF predictions at both high and low x, hinting at a different x dependence.","lead":"Using the asymmetry of Z boson decay products at the LHC, this paper measures the ratio of down to up quarks inside the proton. The measured ratio disagrees with global PDF predictions, being higher at high momentum and lower at low momentum, which could require revisions to proton structure models.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed x-dependence reversal at x≈0.01 rests on one statistically weak low-|Y| bin and could be an artifact of fixing Delta_u, Delta_d to CT18; a closure test should settle it.","rationale":"The reader correctly identified the fixed Delta_u and Delta_d as the weakest assumption. My stress test sharpens this: the Delta-based bias is most dangerous exactly where the new x-dependence claim lives, namely the low-x region (x < 0.02) probed by the lowest |Y| bins of the 13 TeV data. There, the extracted Pd is consistent with zero within a large experimental uncertainty, and the reversal below the PDF predictions is not quantified with a significance. No independent measurement covers this x range, so the x-dependence claim rests entirely on this one analysis. The high-x part of the claim is supported by D0 and by the re-analysis of 8 TeV data, and the fixed-Delta bias is unlikely to explain a deviation that appears in three different datasets in the same direction. Therefore the value deviation at high x is reasonably secure, but the shape deviation is not. The closure test I propose would directly determine whether the fitting procedure with fixed CT18 Delta terms can recover a genuine low-x reversal, and the alternative-PDF refit would quantify the model dependence. If both tests pass, the x-dependence claim becomes credible; if not, the abstract overstates the result. This does not change the overall conditional disposition, but it does mean the paper should not be accepted until such tests are shown. I therefore keep the verdict unchanged (conditional), consistent with the reader's recommendation, while noting that the load-bearing weakness is the low-x reversal, not the high-x excess.","tokens_in":8392,"tokens_out":11387,"duration_ms":93017,"concrete_test":"Perform a closure test: generate pseudo-experiments from the CMS 13 TeV A4 uncertainties using a modified PDF whose d/u ratio is shifted upward by ~20% for x>0.05 and downward by ~20% at x≈0.01, then apply the exact same fitting procedure with Delta_u and Delta_d fixed to CT18. If the extracted R is pulled back toward the CT18 prediction (i.e., the input shift is not recovered), the fixed-Delta assumption cannot support the claimed x-dependence. As a cross-check, refit the 13 TeV data using Delta shapes from MSHT20 and from NNPDF4.0; if the low-|Y| R values change by more than the quoted Delta uncertainties, the low-x reversal is not robust against the choice of PDF used to fix Delta.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract claims significant deviations in both value and x-dependence, but the x-dependence part is carried almost entirely by the low-|Y| bins of the 13 TeV extraction (|Y| < 0.8), which correspond to x1 < 0.02. In the lowest bin (Table I, |Y|=0-0.4), the fitted Pd is -0.0012 ± 0.0052 (exp), so R = Pd/Pu is consistent with zero and has a large relative uncertainty; the next bin (0.4-0.8) gives Pd = 0.0138 ± 0.0054, and no significance for a downward deviation from the PDF predictions is quoted. In contrast, the high-x excess (x > 0.05) is independently supported by D0 and by the 8 TeV re-analysis, which are not affected by this low-x issue. The second, more systematic problem is the fixed Delta_u and Delta_d in Eq. (4): these kinematic shapes are computed from CT18, and the fit only adjusts the constant offsets Pu and Pd within each |Y| bin. If the true M-dependence of the structure parameters differs from CT18, the fit will absorb part of that difference into Pu and Pd. The uncertainty on Delta is evaluated only by sampling the CT18 error sets, which does not cover an entirely different x-dependence, and the paper's statement that MSHT20 and NNPDF4.0 give consistent Delta values checks only the fixed shapes, not the sensitivity of the final R to those shapes. Thus the low-x reversal, which is the only evidence for a changed x-dependence, is doubly fragile: it is statistically weak and it is the region where the fixed-Delta assumption is least constrained by data. The central value claim at high x may survive, but the x-dependence claim is not yet established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper extracts the ratio R = P_d/P_u of down-type to up-type structure parameters from the forward-backward asymmetry in pp → Z/γ* → ℓ+ℓ− events, using the CMS 13 TeV 138 fb⁻¹ measurement and re-analyzing the CMS 8 TeV data with sin²θ_eff fixed. In each |Y| bin, P_u and P_d are fitted while the kinematic dependences Δ_u and Δ_d are fixed using the CT18 NNLO PDF. The authors report that R is significantly higher than current PDF predictions for x > 0.05, consistent with the D0 measurement, and lower than predictions at x ≈ 0.01, claiming a different x-dependence of the light-quark PDFs.","tokens_in":9143,"tokens_out":6027,"duration_ms":47633,"significance":"If the result holds, the paper offers a genuinely novel flavor-separated constraint on the d/u valence ratio, complementary to inclusive Drell-Yan and DIS measurements, and the high-x excess provides independent corroboration of the D0 anomaly. The use of public CMS data and the explicit separation of experimental and Δ-induced uncertainties are strengths. However, the low-x reversal, which is the truly new part of the claim, is statistically fragile and depends on the fixed CT18 shapes; the paper does not yet establish that reversal at the level claimed in the abstract. With a closure test, a significance quantification, and a goodness-of-fit report, this would be a valuable contribution.","major_comments":[{"comment":"The central claim of a changed x-dependence is not robust to the way Δ_u and Δ_d are fixed. In Eq. (4), C_u and C_d are decomposed into bin-constant P_u, P_d plus Δ_u, Δ_d, and the fit adjusts only P_u and P_d, with Δ_u, Δ_d taken from CT18 NNLO. If the true kinematic dependence inside a |Y| bin differs from CT18, that difference is absorbed into P_u and P_d, biasing R toward the CT18 prediction. The quoted Δ uncertainties (the second parentheses in Tables I and II) sample only the CT18 error sets; they do not cover an alternative x-dependence of the light-quark distributions. The paper's statement that MSHT20 and NNPDF4.0 give consistent Δ values validates the central shapes but not the sensitivity of the final R to those shapes. Please add a closure test in which pseudo-data are generated with a modified input d/u ratio and the extraction is repeated, and report the change in R when Δ_u, Δ_d are recomputed from MSHT20 or NNPDF4.0 instead of CT18.","section":"Section 3, Eq. (4) and the paragraph beginning 'The fixed Δ_u and Δ_d...'"},{"comment":"The claimed low-x reversal at x ≈ 0.01 is statistically weak. In the lowest |Y| bin (0–0.4), P_d = −0.0012 ± 0.0052, consistent with zero; in the next bin (0.4–0.8), P_d = 0.0138 ± 0.0054. No significance is quoted for the downward deviation of R from CT18, MSHT20, or NNPDF4.0, and the large relative uncertainty on P_d in these bins means the ratio R is not well determined. Please report the significance of the deviation from each PDF prediction in each |Y| bin, and the combined significance of the two low-|Y| bins for a reversal. If the combined significance is below about 3σ, the abstract and conclusions should be softened to state a 'tendency' rather than a 'significant' reversal.","section":"Table I and the paragraph beginning 'The most important results...'"},{"comment":"The paper defines a χ² but never reports its minimum value or the number of degrees of freedom for any of the fits. Without this information the reader cannot judge whether the factorized model of Eqs. (1) and (4) actually describes the CMS AFB data; a poor global fit would undermine the interpretation of the extracted P_u, P_d. Please include the minimum χ² per |Y| interval (and for the combined 13 TeV and 8 TeV fits), together with the number of mass bins used in each fit.","section":"Eq. (6) and Section 3"},{"comment":"The abstract claims 'significant deviations in both the value and x-dependence of this ratio compared to predictions from current parton distribution functions'. The value deviation at x > 0.05 is supported by the D0 measurement and by the 8 TeV re-analysis, but the x-dependence reversal rests on the low-|Y| bins discussed above, and the statistical significance is not quantified anywhere. Please state the quantitative significance for each claimed deviation, and ensure the abstract's wording matches the reported precision. Also note that the 8 TeV low-|Y| bins (|Y| < 1.25) are consistent with predictions, so the reversal is driven by the 13 TeV low-|Y| bins; this asymmetry in evidence should be acknowledged.","section":"Figures 1–2 and the abstract"}],"minor_comments":[{"comment":"There is a typo in the abstract: 'presen' should be 'present'; also, reference [32] contains 'Electronweak' instead of 'Electroweak'.","section":"Abstract and references"},{"comment":"Equation (5) writes the angular distribution as 1 + cosθ* + A4 cosθ*, which appears to omit the cos²θ* term; please check against the CMS convention and correct the formula.","section":"Equation (5)"},{"comment":"Tables I and II give correlation factors ρ, but no numerical values for R = P_d/P_u with their uncertainties; since R is the central observable, please provide a table of R with total uncertainties and correlation information, or state explicitly that the plotted R error bars include the correlated treatment.","section":"Tables I–II and Figures 1–2"},{"comment":"The phrase 'mass window not wider than 100 GeV around the Z pole' is vague; please specify the exact mass bins used in the fit and how the index i in Eq. (6) runs over them.","section":"Section 3"}],"recommendation":"major_revision","confidential_remarks":"This manuscript is a direct continuation of the authors' earlier work in Ref. [24]; the incremental novelty is the CMS 13 TeV dataset and the fixed sin²θ_eff treatment. The genuinely new physics claim is the low-x reversal, and that is exactly the part that is both statistically weakest and most sensitive to the fixed-Δ assumption. The high-x excess is already reported by D0. I would advise the editor that the paper should not be accepted until the closure test and significance quantification requested in the major comments are provided, because the abstract currently overstates the strength of the evidence."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper is a serious measurement, not a stunt. The group that introduced the AFB factorization method applies it to CMS 13 TeV data (138 fb^-1) and re-fits the 8 TeV data with sin^2 theta_eff fixed, which sharpens the extraction. The main new result is the extension to x < 0.02, where R = Pd/Pu appears to dip below PDF predictions. The high-x excess (x > 0.05) is consistent with D0's 3.5-sigma effect, so that part is likely real.\n\nWhat the paper does well: the method is clearly laid out, the observable R is cleanly factorized from the denominator cancellation, and the authors are explicit about the assumptions (u(x2) ≈ ...). They also compare against CT18, MSHT20, and NNPDF4.0, and the Delta terms are checked across PDFs. The fitting procedure is straightforward and the tables give enough numbers to reproduce the R extraction.\n\nNow the soft spots, in proportion. The low-x reversal is the novel part, but it is carried by the two lowest |Y| bins, where Pd is close to zero and the relative uncertainty is large. The lowest bin gives Pd = -0.0012 ± 0.0052, i.e., consistent with zero, and no significance is quoted for the dip relative to the PDF curves. The abstract's claim of 'significant deviations in both the value and x-dependence' overreaches; the x-dependence part is suggestive at best.\n\nMore concerning is the fixed Delta_u and Delta_d. These kinematic shapes come from CT18, and the fit only adjusts the constant offsets Pu and Pd per bin. If the true M/Y/QT dependence differs from CT18, that difference leaks into Pu and Pd and biases R toward the CT18 prediction. The paper estimates Delta uncertainties only with CT18 error sets; checking that MSHT20 and NNPDF4.0 give similar central Delta values does not test the sensitivity of the final R to those shapes. A closure test with a non-CT18 input PDF would settle this. The paper also does not report chi2 per fit, so we can't see how well the one-parameter-per-bin model actually describes the AFB mass distributions.\n\nIn short: the high-x result is credible and independently supported, the low-x reversal is a real question but not established. This deserves peer review—the method is non-trivial, the data are public, and the implications for PDF fits are interesting. I'd want the authors to add a closure test, report chi2, and downgrade the abstract before acceptance.\n\nRecommendation: send to review, with a request for those revisions.","headline":"A credible AFB-based extraction of d/u at lower x, but the new x-dependence claim rests on weak low-|Y| bins and CT18-fixed Delta terms; the high-x excess is solid and D0-consistent.","tokens_in":9793,"tokens_out":2376,"would_cite":true,"duration_ms":17851,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The down-to-up quark ratio in the proton, extracted from the Drell-Yan forward-backward asymmetry at the LHC, deviates from current PDF predictions in both its value and its x-dependence.","keywords":["forward-backward asymmetry","Drell-Yan process","parton distribution functions","up and down quarks","valence quark ratio","proton structure","flavor separation","LHC"],"falsifier":"Perform a global PDF fit that includes the CMS 13 TeV and 8 TeV AFB data with a free d/u ratio; if the best-fit d/u does not move to the high values found here for x > 0.05 and the lower value at x ~ 0.01, the claimed deviation is not reproducible. Alternatively, redo the extraction with Delta_u and Delta_d computed from MSHT20 and NNPDF4.0 instead of CT18; if R shifts by more than the quoted uncertainty, the result depends on the assumed kinematics.","tokens_in":8142,"feed_emoji":"⚛️","tokens_out":7463,"duration_ms":53339,"temperature":0.7,"pith_summary":"This paper reports a direct measurement of the ratio of down-quark to up-quark contributions inside the proton, using the forward-backward asymmetry of Drell-Yan lepton pairs at the LHC. Unlike most cross-section measurements, this asymmetry can be factorized so that the up and down quark components appear separately, and their ratio is nearly free of contamination from strange, charm, and bottom quarks. The authors extract this ratio from CMS data at 13 TeV (138 $fb^{-1}$) and re-analyze the earlier 8 TeV data. They find that the ratio is significantly larger than current PDF predictions for momentum fractions x > 0.05, consistent with a previous D0 measurement, and smaller than predictions around x ~ 0.01. These deviations point to a different x-dependence of the light-quark distributions than current global fits allow.","feed_headline":"Down-to-up quark ratio in proton deviates from PDFs","feed_subtitle":"LHC forward-backward asymmetry puts the down/up ratio above PDF predictions for x>0.05 and below it near x=0.01.","key_machinery":"The central object is the factorization of AFB into Cu and Cd (measured as bin-averaged Pu and Pd) multiplying hard-process asymmetries Au_FB and Ad_FB, with R = Cd/Cu. The hard-process asymmetries are computed with ResBos at approximate NNLO+N3LL, and the small kinematic dependences within each |Y| bin (Delta_u, Delta_d) are fixed using the CT18 NNLO PDF. This factorization cancels the contributions of s, c, and b quarks in R, so that R probes the d and u quark distributions almost without mixing.","core_discovery":"The paper claims that the ratio R = Pd/Pu, where Pu and Pd are structure parameters that factorize the forward-backward asymmetry AFB(Y,M,QT) for pp -> Z/gamma* -> l+l- into separate up-quark and down-quark contributions, is a nearly pure observable for the valence ratio dV(x1)/uV(x1) in the region x1 > 0.01. Using the latest CMS 13 TeV AFB data and a re-analysis of the CMS 8 TeV data with the weak mixing angle fixed, the measured R is found to be significantly above the predictions of CT18, MSHT20, and NNPDF4.0 for x > 0.05, while at x ~ 0.01 it lies below the predictions. The authors take this as evidence that the x-dependence of the light-quark PDFs, in particular the d/u and anti-d/anti-u ratios, is different from what current global fits assume.","pith_inferences":["One could extend the same method to lower x using forward rapidity coverage (e.g., LHCb) to see whether the downward deviation persists below x = 0.01.","If confirmed, the x-dependence mismatch would feed into electroweak precision measurements at the LHC, since the Z rapidity distribution depends on light-quark flavor separation.","The cancellation of sea-quark contributions in R relies on the assumption u(x2) ~ d(x2) at x2 ~ 10^-3; the method itself could test this by comparing extractions in different rapidity bins, where x2 varies."],"forward_implications":["If the deviation is real, PDF global fits must include Drell-Yan AFB data to constrain the d/u and anti-d/anti-u ratios, which are currently only weakly constrained.","The observed higher R at x > 0.05 implies either a larger d/u valence ratio or a suppressed anti-d/anti-u asymmetry in that region.","The sign change at x ~ 0.01 indicates that current PDF parameterizations have the wrong x-dependence for the light-quark flavor separation.","Because R is nearly insensitive to heavier quarks, the measurement provides a clean test for non-perturbative models of the proton sea."],"supporting_citations":[{"why":"Introduces the factorization of AFB into up and down quark structure parameters Pu and Pd, the method used throughout.","marker":"[23]"},{"why":"First application of the Pu/Pd extraction to LHC 8 TeV data; this paper updates it by fixing the weak mixing angle.","marker":"[24]"},{"why":"Provides the CMS 8 TeV AFB data re-analyzed here with the fixed weak mixing angle.","marker":"[25]"},{"why":"D0 measurement reporting R higher than PDF predictions at x ~ 0.1, which the new results are consistent with.","marker":"[26]"},{"why":"Supplies the CMS 13 TeV AFB data and the correlation information used for the main extraction.","marker":"[27, 28]"},{"why":"CT18 NNLO PDF used to compute the fixed kinematic-dependence terms Delta_u and Delta_d and as a comparison prediction.","marker":"[1]"},{"why":"MSHT20 and NNPDF4.0 PDFs used as comparison predictions for the extracted Pu, Pd, and R.","marker":"[2, 3]"},{"why":"Computes the hard-process-level asymmetries Au_FB and Ad_FB at approximate NNLO+N3LL, needed for the fit.","marker":"[30]"}],"fun_headline_variants":["Down/up ratio in proton flips above and below PDFs","Proton's down-up ratio deviates from all major PDFs","Flavor asymmetry in proton challenges current PDFs","Down/up ratio in proton shows unexpected x-dependence","LHC measurement of proton's down/up ratio defies PDFs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The extraction assumes that the small kinematic dependences within each rapidity bin, computed with the CT18 PDF, correctly describe how the up and down structure parameters vary; if the true light-quark distributions differ from CT18 inside a bin, the fit could absorb that difference into Pu and Pd and bias R toward the CT18 prediction.","fun_headline_variants_meta":{"raw":{"variants":["Down/up ratio in proton flips above and below PDFs","Proton's down-up ratio deviates from all major PDFs","Flavor asymmetry in proton challenges current PDFs","Down/up ratio in proton shows unexpected x-dependence","LHC measurement of proton's down/up ratio defies PDFs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000674,"raw_usage":{"total_tokens":3044,"prompt_tokens":895,"completion_tokens":2149,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":511,"completion_tokens_details":{"reasoning_tokens":2063}},"tokens_in":511,"tokens_out":2149,"duration_ms":14160,"temperature":1.0,"reasoning_tokens":2063,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:43:50.698128+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Perform a global PDF fit that includes the CMS 13 TeV and 8 TeV AFB data with a free d/u ratio; if the best-fit d/u does not move to the high values found here for x > 0.05 and the lower value at x ~ 0.01, the claimed deviation is not reproducible. Alternatively, redo the extraction with Delta_u and Delta_d computed from MSHT20 and NNPDF4.0 instead of CT18; if R shifts by more than the quoted uncertainty, the result depends on the assumed kinematics.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the factorization of AFB into up and down quark structure parameters Pu and Pd, the method used throughout."},{"cited_title":"Khachatryan et al","cited_arxiv_id":null,"evidence_quote":"First application of the Pu/Pd extraction to LHC 8 TeV data; this paper updates it by fixing the weak mixing angle."},{"cited_title":"Yuan, Phys","cited_arxiv_id":null,"evidence_quote":"Provides the CMS 8 TeV AFB data re-analyzed here with the fixed weak mixing angle."},{"cited_title":"Yuan, Phys","cited_arxiv_id":null,"evidence_quote":"D0 measurement reporting R higher than PDF predictions at x ~ 0.1, which the new results are consistent with."},{"cited_title":"( 4) via the minimum χ2 fit: χ 2 = ∑ i ( Adata FB (Mi) − Apred","cited_arxiv_id":null,"evidence_quote":"CT18 NNLO PDF used to compute the fixed kinematic-dependence terms Delta_u and Delta_d and as a comparison prediction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Computes the hard-process-level asymmetries Au_FB and Ad_FB at approximate NNLO+N3LL, needed for the fit."}],"review_version":1}