{"id":"b8f5e973-97ad-4fea-9104-02e7dd915e11","arxiv_id":"2412.14348","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"NNLO fits to LHC top pair production data give a top-quark pole mass of about 171.5 to 172.2 GeV depending on the PDF set, with total uncertainties near 0.2 to 0.5 GeV.","lead":"Physicists fit the top quark's mass using LHC measurements of top pair production and state-of-the-art NNLO theory predictions. They find values around 171.5 to 172.2 GeV, consistent with the world average, though different datasets show small tensions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unquantified mass dependence of the experimental unfolding could bias the extracted top-quark pole mass by an amount comparable to the quoted uncertainty.","rationale":"I read the paper in good faith and find the NNLO fitting methodology standard and the reported results plausible: the best-fit masses across four PDF sets are mutually compatible and consistent with the PDG value, and the quoted uncertainties are of the expected size for this type of analysis. The procedures described—normalized differential cross sections, seven-point scale variation, chi2 minimization with a parabolic interpolation—are all commonly used. The main vulnerability is not an internal inconsistency but an external assumption: the experimental unfolded data are imported as mass-independent, yet the unfolding is performed with Monte Carlo generators that assume a specific top-quark mass. This matters because the fit is sensitive to the shape of M(ttbar), and the unfolding corrections that determine that shape can depend on the assumed mass. The proceedings text gives no evidence that this effect was quantified, and the abstract's phrasing makes the unfolded data the starting point of the comparison. This is the same concern identified by the reader as the weakest assumption, and I agree with that assessment. My proposed closure test would resolve the issue: if re-unfolding pseudo-data at different masses with the nominal response changes the extracted m_t by less than about 0.1 GeV, the concern is not load-bearing; if it changes the result by more, the quoted total uncertainties are underestimated. Because the concern can plausibly be addressed by the companion paper and does not invalidate the existing results on its own, the reader's CONDITIONAL verdict remains the appropriate outcome. I see no reason to move to ACCEPT or REJECT based on the proceedings text alone.","tokens_in":4282,"tokens_out":6051,"duration_ms":60718,"concrete_test":"Perform an unfolding closure test using the published response matrices for the two most constraining datasets, ATLAS 1908.07305 (13 TeV semileptonic) and CMS 2108.02803 (13 TeV semileptonic). Generate parton-level pseudo-data from NNLO predictions at m_t = 170, 172.5, and 175 GeV, pass them through the detector response, unfold using the nominal response matrix built at m_t = 172.5 GeV as done in the experimental analyses, and refit m_t with the same chi2 procedure used in this paper. If the refitted mass shifts by more than about 0.1–0.2 GeV when the input mass is varied, the unfolding mass dependence is not negligible and must be included as a systematic. Alternatively, check the companion paper (Ref [5]) and supplemental material for such a closure test; if it is absent, the quoted uncertainty is incomplete.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central argument treats the unfolded parton-level double-differential cross sections (abstract: 'obtained by the ATLAS and CMS collaborations from unfolding of their experimental data to the parton level') as a fixed external input and compares them to NNLO predictions at varying m_t^pole. This is legitimate only if the unfolding corrections—acceptance, efficiency, migration matrices, and regularization—do not depend on the top-quark mass assumed in the Monte Carlo generators. That assumption is not stated, tested, or referenced in the proceedings text. In practice, the kinematic reconstruction of M(ttbar) uses the top-quark mass in the simulation, so the response matrix and the unfolding can imprint the assumed value (typically m_t = 172.5 GeV) onto the published parton-level distributions. The reported best-fit values cluster at 171.5–172.2 GeV, close to this assumed value, so even a bias of a few tenths of a GeV is comparable to the quoted data uncertainty of 0.2–0.3 GeV and to the scale and PDF uncertainties. Without a closure test showing that unfolding pseudo-data generated at m_t = 170 GeV and m_t = 175 GeV with the nominal response reproduces those input masses, the mass dependence of the input data remains an unquantified systematic that directly affects the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This proceedings paper reports NNLO fits of the top-quark pole mass using total, single-differential, and double-differential t-tbar cross-section data from ATLAS and CMS at Run 1 and Run 2. The theoretical predictions are computed with the MATRIX code at NNLO and interfaced to PineAPPL, and fits are performed with four PDF+alpha_s sets: ABMP16, CT18, MSHT20, and NNPDF40. The best-fit pole masses cluster between 171.5 and 172.2 GeV, with data uncertainties around 0.2-0.3 GeV and PDF and scale uncertainties of similar size, and are compatible with the PDG 2024 value. The paper also reports a mild tension between semileptonic and dileptonic datasets. The text is short and explicitly refers to Ref. [5] for methodological details.","tokens_in":4450,"tokens_out":3309,"duration_ms":33277,"significance":"If the results hold, this is a valuable independent NNLO extraction of the top-quark pole mass with competitive precision, and the use of multiple modern PDF sets provides a useful cross-check. The paper also highlights which Run 2 differential datasets carry the most constraining power and quantifies the current data-versus-theory uncertainty budget. A strength is that the calculation is based on established NNLO QCD predictions and fast interpolation methods, and the claimed compatibility with the PDG value gives a nontrivial consistency check. However, the proceedings format omits several pieces of information needed to verify the central numerical claims, and one unquantified systematic—the mass dependence of the experimental unfolding—could affect the extracted mass at the level of the quoted uncertainties.","major_comments":[{"comment":"The unfolded parton-level double-differential data are treated as mass-independent external inputs, but the abstract states that they are obtained 'from unfolding of their experimental data to the parton level.' If the unfolding corrections, migration matrices, or acceptance/efficiency corrections depend on the top-quark mass assumed in the Monte Carlo generators, a residual mass dependence is imprinted on the published distributions and can bias the extracted m_t^pole. The reported best-fit values (171.5-172.2 GeV) lie close to the commonly assumed generator mass of 172.5 GeV, and the quoted data uncertainty of 0.2-0.3 GeV is comparable to the size of a plausible bias. The manuscript does not mention or test this effect. The authors should either provide a closure test in which pseudo-data generated at, e.g., m_t=170 and 175 GeV are unfolded with the nominal response and shown to reproduce the input masses, or quantify the mass dependence of the unfolding corrections and propagate it into the fit.","section":"Abstract and main text"},{"comment":"The text states that 'the chi2's close to their minima show a parabolic shape' and that the best-fit mass is obtained by fitting a parabola through only three points, m_t = 170, 172.5, and 175 GeV. With only three points, parabolicity cannot be validated, and the quoted uncertainty from the parabolic interpolation is not assessed. The authors should show chi2 values at additional mass points (or a residual analysis) to justify the parabolic approximation, and should quantify the interpolation error relative to the quoted 0.2-0.3 GeV uncertainty.","section":"Main text, paragraph on chi2 fits"},{"comment":"The manuscript reports best-fit values and uncertainty decompositions (data, PDF, scale) but does not provide any goodness-of-fit information, such as chi2/ndof values, or the covariance matrices used in the global combination. Consequently, the claims that different PDF sets are 'compatible within uncertainties' and that semileptonic and dileptonic datasets are 'compatible within 2 sigma' cannot be checked from the presented material. At minimum, the authors should report the chi2 minimum per fit, the number of data points, and a quantitative definition of the tension (e.g., covariance-aware pulls between dataset groups).","section":"Main text, uncertainty summary and Fig. 2"},{"comment":"The scale uncertainties are said to be evaluated by building a separate chi2 for each of the seven-point scale variations around HT/4 and 'accounting for the spread in the fitted m_t values.' The reported asymmetric total uncertainties in the right panel of Fig. 2 (e.g., NNPDF40: +0.19/-0.49 GeV) are not explained in terms of how the spread is combined with the data and PDF uncertainties. The authors should state whether the scale uncertainty is an envelope, a quadrature sum, or a profile, and how asymmetric errors are constructed.","section":"Main text, scale uncertainty procedure"}],"minor_comments":[{"comment":"There is a typo: 'both runs simulteneously' should be 'both runs simultaneously.'","section":"Main text"},{"comment":"The phrase 'the most stringent constraints are played by the dataset' should be reworded to 'the most stringent constraints are provided by the dataset.'","section":"Main text"},{"comment":"The experimental references appear only inside the figure captions; listing the datasets with their references, kinematic ranges, and luminosities in a small table would make the input to the global fit easier to follow.","section":"Fig. 2"},{"comment":"The sentence 'In the near future, data uncertainty reduction will push for theoretical computations beyond NNLO' is vague; consider specifying which NNLL or aN3LO corrections are expected to be most relevant for the double-differential observables used here.","section":"Main text"}],"recommendation":"major_revision","confidential_remarks":"This is a very short proceedings paper that defers most methodological detail to Ref. [5], so some missing documentation (covariance matrices, chi2/ndof, dataset tables) may already exist in the full JHEP paper. However, the unfolding mass-dependence issue is not addressed anywhere in the proceedings, and it directly affects the central claim; the authors should be asked to either provide a closure test or explicitly state and justify the assumption that the unfolding is mass-independent. The three-point parabolic interpolation is also a load-bearing approximation that should be validated."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here's my honest read. The paper is a short proceedings from ICHEP 2024, reporting an updated NNLO extraction of the top-quark pole mass from LHC ttbar+X cross-section data. The genuinely new element is the combined Run 1 + Run 2 fit across four PDF sets, plus the inclusion of the full Run 2 CMS semileptonic dataset. That is a legitimate extension of the authors' earlier Ref [5], not a new method, but the numerical results are new.\n\nWhat it does well: the fitted masses are mutually compatible across ABMP16, CT18, MSHT20, and NNPDF40, and they fall close to the PDG value. The uncertainty breakdown into data, PDF, and scale components is clear, and the paper honestly notes the ~2-sigma tension between semileptonic and dileptonic datasets, saying it shrinks when PDFs are fitted simultaneously. Using normalized differential distributions to mitigate correlations of systematic uncertainties is sensible.\n\nThe soft spots are mostly of the 'proceedings' variety. The text quotes central values and uncertainties but does not give covariance matrices or goodness-of-fit numbers; that detail presumably lives in the companion paper. More substantive, the abstract says the data come from ATLAS/CMS unfolding to parton level, and the paper does not discuss whether the unfolding procedure imprints the Monte Carlo assumed top mass (typically 172.5 GeV) onto the published distributions. Since the best-fit values cluster near that value, an unquantified bias of a few tenths of a GeV is plausible and would sit alongside the quoted data uncertainty of 0.2-0.3 GeV. The stress-test note calls for a closure test with pseudo-data at different input masses; I think that is a fair request, though it may be beyond what a proceedings can contain. Also, the chi2 minimum is determined by parabolic interpolation through only three points (170, 172.5, 175 GeV). The authors state the minima are parabolic, so this is not a red flag, but it is coarse.\n\nOverall: this is a plausible, incremental result. It does not break new methodological ground, but it is a useful cross-check of the top mass with current PDFs and Run 2 data. The central claim is not undermined by anything I see here; the main worry is the unquantified unfolding mass dependence.\n\nIf this arrived as a full paper, I would send it to peer review, with the request that the authors either demonstrate the unfolding is mass-independent or quantify that systematic. As a proceedings, it is acceptable, but I would not cite it as a standalone analysis without the companion paper.","headline":"Plausible NNLO top-mass extraction from LHC differential data, but the proceedings leaves the unfolding mass-dependence unquantified and the statistical details to the companion paper.","tokens_in":5057,"tokens_out":3707,"would_cite":true,"duration_ms":32797,"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":"Combining LHC cross-section data at NNLO constrains the top-quark pole mass to about 171.5-172.2 GeV, compatible with the 2024 world average.","keywords":["top-quark pole mass","NNLO QCD","t tbar+X production","differential cross sections","parton distribution functions","LHC top physics","mass extraction","unfolding"],"falsifier":"Recompute the unfolding of the ATLAS and CMS datasets using Monte Carlo generators with the top mass set to 170 GeV and then to 175 GeV, and refit; if the extracted pole mass moves by more than the quoted ~0.3 GeV uncertainty, the mass-independent unfolding assumption is false.","tokens_in":4034,"feed_emoji":"⚛️","tokens_out":8847,"duration_ms":75214,"temperature":0.7,"pith_summary":"This paper reports next-to-next-to-leading-order (NNLO) QCD fits that extract the top-quark pole mass from LHC measurements of total, single-differential, and double-differential $t\\bar{t}+X$ cross sections. The fits use unfolded parton-level data from ATLAS and CMS, compared with NNLO theory predictions, and are repeated with four modern PDF sets. Across sets the best-fit values cluster between 171.5 and 172.2 GeV, with data uncertainties of about 0.2-0.3 GeV and PDF and scale uncertainties of similar size. The results are compatible with the 2024 world-average value of $172.4 \\pm 0.7$ GeV, while semileptonic and dileptonic datasets show a tension that stays within about two standard deviations. The paper's point is that differential top-pair data at NNLO already provide a competitive and theoretically well-defined determination of the top-quark mass.","feed_headline":"Differential LHC data narrow top-quark mass to about 172 GeV","feed_subtitle":"Combining Run 1 and Run 2 ttbar data with four PDF sets gives 171.5-172.2 GeV, matching the 2024 world average.","key_machinery":"The central machinery is the NNLO QCD computation of $t\\bar{t}+X$ production, implemented in the MATRIX code (a fully differential next-to-next-to-leading-order QCD program) and interfaced with PineAPPL (a fast interpolation tool) so that many mass values can be scanned efficiently. The fit compares normalized single- and double-differential cross-section data to these predictions using a $\\chi^2$ function whose covariance matrix includes statistical, correlated and uncorrelated systematic, PDF, and scale uncertainties. The top-quark mass enters through the pole-mass dependence of the partonic cross sections, and the extracted value is read off from a parabolic interpolation through the $\\chi^2$ values at $m_t^{\\rm pole} = 170, 172.5, 175$ GeV.","core_discovery":"The central claim is that comparing NNLO QCD predictions to normalized single- and double-differential $t\\bar{t}$ cross sections from ATLAS and CMS determines the on-shell top-quark mass with precision comparable to direct measurements. For the most global fit combining Run 1 and Run 2 differential data plus total inclusive cross sections, the paper obtains $m_t^{\\rm pole} = 171.54 \\pm 0.24$ GeV (ABMP16), $171.59 \\pm 0.22$ GeV (CT18), $171.79 \\pm 0.22$ GeV (MSHT20), and $172.15 \\pm 0.23$ GeV (NNPDF40), with additional PDF and scale uncertainties listed separately. The fit uses the $\\chi^2$ distribution around three mass points (170, 172.5, 175 GeV) and a parabolic interpolation, with seven-point renormalization and factorization scale variations. The authors find that Run 2 differential datasets, especially the CMS 13 TeV semileptonic analysis, dominate the constraint, while total cross-section data contribute little. Semileptonic data pull the mass slightly upward relative to dileptonic data, but the shift stays within about two standard deviations.","pith_inferences":["A direct test of the unfolding premise would be to recalculate the unfolded distributions with Monte Carlo generators generated at several top masses; this check is not reported, so the size of any mass-dependence bias remains unquantified.","Because normalized differential cross sections suppress unknown correlated systematics, fits to absolute double-differential distributions with full covariance information could either sharpen or dissolve the semileptonic/dileptonic tension.","The same NNLO plus fast-interpolation pipeline could be applied to other processes, such as single-top production, where the mass dependence enters differently and would provide an independent check on the extracted pole mass.","As data uncertainties continue to shrink, the fit will become dominated by NNLO scale dependence, making soft-gluon resummation for the double-differential distributions the decisive next theoretical upgrade."],"forward_implications":["If the fits are correct, current Run 1 plus Run 2 differential data already determine the top-quark pole mass to about 0.2-0.3 GeV precision, competitive with the world average.","The compatibility among four independent PDF sets indicates that, at the present precision, the choice of PDF set is not the dominant spread in the extracted mass.","The two-sigma shift between semileptonic and dileptonic channels implies that these datasets cannot yet be combined without accounting for their different systematic correlations.","Total inclusive cross-section data contribute little to the constraint, so future determinations should prioritize differential measurements.","Because data uncertainties already match the NNLO scale uncertainties, further progress in precision will require theory corrections beyond NNLO."],"supporting_citations":[{"why":"Establishes the NNLO extraction methodology and the previous version of the fit that this proceedings updates.","marker":"[5]"},{"why":"Provides the fully differential NNLO QCD predictions for top-quark pair production used as the theory input.","marker":"[2]"},{"why":"Documents the MATRIX code used to compute the NNLO cross sections.","marker":"[3]"},{"why":"Supplies the PineAPPL interpolation that makes the repeated fast evaluation needed for chi-squared fits possible.","marker":"[4]"},{"why":"Defines the ABMP16 PDF+alpha_s set, one of the four input sets against which the mass fit is checked.","marker":"[6]"},{"why":"Provides the CMS 13 TeV semileptonic double-differential dataset that gives the strongest constraint on the top-quark mass.","marker":"[7]"},{"why":"Reports simultaneous PDF and top-mass fits that reduce the dataset tension, referenced as the future direction.","marker":"[8]"},{"why":"Gives the 2024 world-average top-quark mass used as the external comparison for the extracted values.","marker":"[9]"}],"fun_headline_variants":["NNLO fits: top mass 171.5–172.2 GeV, dataset tension","Differential ttbar data pin top mass to ~172 GeV","Top quark mass from NNLO fits: ~172 GeV with slight tensions","ATLAS+CMS data constrain top mass to 0.2 GeV via NNLO fits"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The unfolded parton-level data are treated as a fixed external input whose unfolding is assumed to be independent of the top-quark mass, so a residual mass dependence in the unfolding corrections could feed the assumed mass back into the fit and bias the extracted pole mass.","fun_headline_variants_meta":{"raw":{"variants":["NNLO fits: top mass 171.5–172.2 GeV, dataset tension","Differential ttbar data pin top mass to ~172 GeV","Top quark mass from NNLO fits: ~172 GeV with slight tensions","ATLAS+CMS data constrain top mass to 0.2 GeV via NNLO fits"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000307,"raw_usage":{"total_tokens":1735,"prompt_tokens":901,"completion_tokens":834,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":517,"completion_tokens_details":{"reasoning_tokens":749}},"tokens_in":517,"tokens_out":834,"duration_ms":7507,"temperature":1.0,"reasoning_tokens":749,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T12:18:24.883780+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the unfolding of the ATLAS and CMS datasets using Monte Carlo generators with the top mass set to 170 GeV and then to 175 GeV, and refit; if the extracted pole mass moves by more than the quoted ~0.3 GeV uncertainty, the mass-independent unfolding assumption is false.","supporting_citations":[],"review_version":1}