{"id":"b6905c42-204e-48e9-9863-859a4c440e4a","arxiv_id":"2411.16907","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"CMS reports simultaneous constraints on six dimension-six SMEFT Wilson coefficients from VH, H to bb production at sqrt(s) = 13 TeV; all results agree with the standard model.","lead":"The CMS experiment used 138 inverse femtobarns of LHC proton-proton collision data to constrain six effective field theory coefficients in Higgs production with W or Z bosons, where the Higgs decays to bottom quarks. All measured constraints agree with the standard model, placing limits on possible new physics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Quadratic-model likelihood-ratio intervals lack guaranteed Wilks coverage; the paper discloses this, but the q<1 and q<4 thresholds in Fig. 7 can still be misread as 68% and 95% confidence intervals, and the profiled q<1 interval for cHq1 excludes the SM value zero.","rationale":"The reader identified the quadratic-model interval coverage as the weakest assumption, and the manuscript itself flags this limitation, so it is the clearest load-bearing concern. My read of the paper confirms that the central claim of SM consistency is well supported by the linear-model p-value and by the 95% intervals, so no more severe objection arises. The quadratic-model q<1 and q<4 intervals are explicitly likelihood-ratio intervals in the text, which partially addresses the concern, but the main figure presents them without a repeated warning, and the cHq1 profiled q<1 interval excludes the SM value, creating a potential misinterpretation. A coverage study would settle whether the numerical constraints need recalibration or merely clearer labeling. Since the paper already contains the caveat but the presentation could still mislead, keeping the reader's CONDITIONAL verdict is appropriate, with the condition effectively being to add prominent labeling or calibration in the summary of results.","tokens_in":58821,"tokens_out":15106,"duration_ms":141022,"concrete_test":"Run a frequentist coverage study for the quadratic model: generate thousands of toy pseudo-experiments from the fitted model at the SM point (or at the best-fit point), including all nuisance parameters, and compute the profiled profile likelihood ratio q for each of the six Wilson coefficients. Compare the empirical 68% and 95% quantiles of q with the nominal thresholds 1 and 4 used in Fig. 7. If, for cHq1, the empirical 68% quantile of q deviates substantially from 1 (e.g., exceeds 1.5 or falls below 0.7), the reported q<1 interval is not a valid 68% confidence interval and the figure and HEPData tables should carry an explicit warning that no confidence level is attached. If the quantiles are close to 1 and 4, the existing textual caveat suffices.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quadratic SMEFT parametrization makes the expected yields in analysis bins quadratic in the Wilson coefficients, so the profile likelihood ratio is not asymptotically chi-square and the Wilks regularity conditions are violated. The paper explicitly states this in Section 9: 'The likelihood ratio intervals for the quadratic model may thus undercover or overcover.' Despite this caveat, the quadratic-model results are displayed in Figure 7 with the same q<1 and q<4 thresholds as the linear model, and the figure caption labels them only as 'intervals where the test statistic is below 1 and 4' without a repeated warning. For cHq1, the profiled quadratic q<1 interval is the union [-0.068,-0.028] U [0.0085,0.074], which excludes zero; a reader who interprets q<1 as a 68% confidence interval would see a tension with the SM that is inconsistent with the reported 84% compatibility p-value. The linear-model intervals have correct coverage, so the central claim that the results are consistent with the SM is not threatened, and the linear p-value of 73% independently supports it. The weakness is therefore in the numerical constraints quoted under the quadratic model, which are a major advertised result of the paper. The paper discloses the coverage problem in the text, but the presentation in the summary figure and abstract could still mislead; calibration or prominent relabeling is needed.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a CMS measurement of standard model effective field theory (SMEFT) Wilson coefficients in VH production with H to bbbar decays, using 138 fb^-1 of proton-proton collision data at sqrt(s) = 13 TeV. The analysis combines 0-, 1-, and 2-lepton channels in resolved and boosted jet topologies and uses the boosted information tree (BIT) likelihood-free inference method to construct observables sensitive to six dimension-six operator coefficients: cHq1, cHq3, cHu, cHd, gZZ2, and gZZ4. One-dimensional profiled and frozen likelihood scans are presented for linear and quadratic SMEFT parametrizations, along with two-dimensional scans and lower limits on the cutoff scale Lambda. The observed results are reported as consistent with the standard model, with compatibility p-values of 73% for the linear model and 84% for the quadratic model.","tokens_in":59063,"tokens_out":10137,"duration_ms":97091,"significance":"This is a technically ambitious and generally well-executed analysis. It is the first CMS study in the VH(bb) channel to use likelihood-free inference with the BIT method to probe several SMEFT operators simultaneously, including CP-sensitive angular information, and it provides a six-dimensional constraint set that goes beyond earlier STXS-based interpretations. The paper is careful in its background model, with dedicated control regions, in-situ flavor-tagging scale factors, a comprehensive systematic uncertainty model, goodness-of-fit checks, and public HEPData tables. A notable strength is the explicit disclosure in Sec. 9 that the quadratic-model likelihood-ratio intervals may under- or over-cover because Wilks regularity conditions are violated. The central SM-consistency claim is robust and independently supported by the linear-model p-value of 73%, so the coverage issue does not threaten that conclusion.","major_comments":[{"comment":"The quadratic-model intervals are presented with q<1 and q<4 thresholds even though the text states that the Wilks regularity conditions are violated and that \"The likelihood ratio intervals for the quadratic model may thus undercover or overcover.\" The Fig. 7 caption does not repeat this caveat, and the abstract and summary present constraints without qualification. The problem is not merely academic: the profiled quadratic q<1 interval for cHq1 is the union [-0.068,-0.028] U [0.0085,0.074], which excludes the SM value zero, so a reader treating q<1 as a 68% confidence interval would see a tension that is inconsistent with the quoted 84% compatibility p-value. Please calibrate these intervals (e.g., with an MC-based coverage or Neyman construction) or relabel them prominently and consistently as likelihood-ratio intervals without confidence-level interpretation, including in the figure, the abstract, and the summary.","section":"Section 9, Fig. 7"},{"comment":"The lower limits on the energy scale Lambda in Fig. 8 are obtained from the q=4 thresholds on the Wilson coefficients, as stated in the text: \"The upper limits on the Wilson coefficients corresponding to q = 4 is used for translating the constraints to Lambda.\" For the quadratic parametrization these thresholds are subject to the same coverage caveat, so the Lambda limits inherit it. If the quadratic-model Lambda limits are retained as quantitative results, they should be derived from calibrated intervals or explicitly labeled as non-coverage likelihood-ratio bounds.","section":"Section 9, Fig. 8"}],"minor_comments":[{"comment":"The test statistic is written as q_theta = -log[L(D|theta)/L(D|theta0)] without the factor 2, whereas the thresholds q<1 and q<4 and the Delta(-2 ln L) axes in the figures correspond to the usual -2 log-likelihood ratio; please make the definition consistent.","section":"Section 7.1, Eq. (10)"},{"comment":"The sentence \"The modified frequentist approach [117-119] is used in this search to set intervals\" appears to describe a CLs procedure, but the reported intervals are asymptotic likelihood-ratio thresholds; either remove the sentence or explain how the CLs approach was used.","section":"Section 9"},{"comment":"The statement \"A sufficient number of nominal values are simulated which allows the interpolation to recover the full polynomial EFT dependency\" is too vague; please specify the number of simulation points and the coefficient values used to fix the quadratic polynomial in six coefficients.","section":"Section 4"},{"comment":"The sentence \"For all Wilson coefficients, the quadratic components dominate the SMEFT sensitivity, except for cHq3, where the linear and quadratic terms have comparable sensitivity and therefore result in better constraints on the Wilson coefficient observed values\" is unclear; comparable linear and quadratic contributions do not by themselves explain the better constraints, and the sentence should be rephrased.","section":"Section 9"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is technically sound and the central SM-consistency claim is well supported. My main concern is the presentation of the quadratic-model intervals, which are a headline numerical result but lack guaranteed coverage; because the collaboration has already disclosed the limitation in the text, the fix is within reach. The paper is a strong candidate for publication after the requested calibration or relabeling."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline: this is a solid, careful SMEFT measurement in VH(bb) with 138 fb^-1, and the first CMS result in this channel to use the boosted information tree (BIT) likelihood-free method and to exploit angular observables sensitive to CP. The six Wilson coefficients are constrained simultaneously, with linear and quadratic SMEFT expansions, and the results are consistent with the SM. The central claim holds up; the p-values are 73% (linear) and 84% (quadratic), and the linear-model intervals have proper coverage.\n\nWhat's genuinely new: the BIT-based extraction, the angular inputs for the 1- and 2-lepton channels, and the simultaneous fit of cHq1, cHq3, cHu, cHd, gZZ2, gZZ4 in this final state. The paper is also well documented: HEPData record, full systematic treatment, and a clear explanation of the template optimization via Bayesian optimization. The background model is detailed and validated with goodness-of-fit checks. No circularity: the Wilson coefficients are the measured targets, and the BIT training uses simulated SMEFT weights as targets, not recycled measurements.\n\nThe soft spot is exactly what the stress-test identifies: the quadratic-model likelihood-ratio intervals are reported with q<1 and q<4 thresholds even though Wilks regularity conditions are violated. The paper discloses this in Section 9, but the summary figure (Fig. 7) labels them only as 'intervals where the test statistic is below 1 and 4' without a repeated warning. A reader can easily misread the quadratic q<1 interval for cHq1 (which excludes zero) as a 68% confidence interval, which would contradict the 84% SM-compatibility p-value. This is a presentation problem, not a flaw in the analysis. The linear intervals are fine, and the central conclusion does not depend on the quadratic coverage. Still, the quadratic constraints are a headline result, so the caption should carry the caveat or the intervals should be coverage-calibrated.\n\nThe citation pattern looks appropriate; the paper builds on the STXS measurement and the BIT/MELA literature, and the LHC EFT WG note. I see no serious gaps.\n\nWho is this for: anyone doing global SMEFT fits, and phenomenologists interested in machine-learning-based EFT extraction. It deserves a serious referee; the methodology and results are important enough for JHEP-type review. My recommendation: send it to peer review, with a request that the quadratic interval labeling be tightened.","headline":"A careful and genuinely new SMEFT extraction in VH(bb) using likelihood-free inference; the quadratic-model interval labeling is the only real soft spot.","tokens_in":59645,"tokens_out":2208,"would_cite":true,"duration_ms":20595,"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":"In VH(H→bb) production at 13 TeV, six SMEFT Wilson coefficients are simultaneously constrained and all match the standard model.","keywords":["SMEFT","Higgs-strahlung","VH production","H to b bbar decay","Wilson coefficients","likelihood-free inference","boosted information trees","CP violation"],"falsifier":"Re-extract the quadratic-model intervals using a Neyman construction or Monte Carlo calibration on simulated pseudo-data and check whether the q < 1 and q < 4 thresholds actually have 68% and 95% coverage.","tokens_in":58545,"feed_emoji":"🔬","tokens_out":6485,"duration_ms":61653,"temperature":0.7,"pith_summary":"This paper asks whether LHC data on producing a Higgs boson together with a W or Z boson, with the Higgs decaying into bottom quarks, show any sign of new physics beyond the standard model. Working with 138 inverse femtobarns of 13 TeV CMS data, it fits six dimension-six SMEFT operator coefficients simultaneously and reports that all are consistent with zero. It is the first analysis in this channel to include angular observables sensitive to the CP structure of the Higgs interaction, and the most complete SMEFT interpretation of VH(H→bb) to date. If its conclusions hold, they would rule out a large class of nonresonant new-physics effects at the TeV scale in Higgs-strahlung production while tightening the global SMEFT bounds on these operators.","feed_headline":"Six new-physics couplings in Higgs-strahlung all match the SM","feed_subtitle":"CMS fits six SMEFT couplings using 138/fb of Higgs-strahlung data and sees no deviation.","key_machinery":"The central object is the boosted information tree (BIT), a likelihood-free estimator of the likelihood ratio R(x|θ, θ0) between different SMEFT hypotheses, built from per-event matrix-element weights computed with SMEFTsim. The quadratic dependence on the six Wilson coefficients is decomposed into linear, quadratic, and mixed components of the likelihood ratio, and a Bayesian optimization of the binned template shape selects the working point in coefficient space that maximizes fully profiled sensitivity to all six coefficients. The angular basis from Ref. [38] supplies the CP-sensitive angular functions, and the coefficient basis is rotated to the mass-eigenstate basis, defining gZZ2 and gZZ4 and removing unconstrained directions in the Wilson coefficient space.","core_discovery":"The paper claims that in the VH(H→bb) process at √s = 13 TeV with 138 fb⁻¹ of data, a simultaneous profiled maximum-likelihood fit to six dimension-six SMEFT Wilson coefficients — cHq1, cHq3, cHu, cHd, gZZ2, and gZZ4 — yields results consistent with the standard model. For the first time in this channel, angular observables sensitive to the CP structure of the H-V interaction are included, via a likelihood-free inference method (boosted information trees) that learns the likelihood ratio from simulated SMEFT weights. The compatibility p-values are 73% for the linear and 84% for the quadratic SMEFT expansion. The analysis claims to be the most comprehensive SMEFT interpretation in the VH(H→bb) channel to date, with constraints on vector-coupling operators generally tighter than those on gauge-coupling operators, and it reports profiled lower limits on the new-physics scale Λ for three assumptions about coefficient magnitudes.","pith_inferences":["The paper leaves implicit that the quadratic-model intervals should not be quoted as confidence intervals; only the linear-model q < 1 and q < 4 thresholds carry a coverage guarantee.","Because the data are statistically limited, the hierarchy of uncertainties suggests that a future high-luminosity run with the same method could push the new-physics scale limits to several TeV for weakly coupled coefficients.","The template-optimization routine used here is a general solution to the profiled-EFT sensitivity problem and could be reused in global SMEFT fits outside this channel.","The angular decomposition exploited in the 1- and 2-lepton channels is suppressed in the 0-lepton channel by final-state topology; extending similar observables to hadronic V decays would recover CP sensitivity there."],"forward_implications":["The linear-model constraints, which have proper coverage, provide robust bounds on the current operators, especially cHq3, in the VH(bb) channel.","Quadratic SMEFT terms dominate the sensitivity for most coefficients, so future data will tighten these bounds more than linearly.","The BIT template-optimization procedure can be extended to other multi-operator EFT analyses where profiling several coefficients degrades sensitivity.","The inclusion of CP-sensitive angular observables opens a direct path to constrain CP-violating couplings in VH production.","These constraints can be combined with electroweak-precision and top-quark SMEFT fits to tighten global limits on the same six operators."],"supporting_citations":[{"why":"Supplies the event selection, control regions, and background estimation strategy that this analysis adapts from the previous CMS VH(bb) STXS measurement.","marker":"[11]"},{"why":"Defines the Warsaw basis used for the dimension-six operators listed in Table 1.","marker":"[35]"},{"why":"Provides the angular decomposition f_i(Θ, θ, φ) into helicity amplitudes that the analysis uses to build CP-sensitive observables.","marker":"[38]"},{"why":"Introduces the tree-boosting method for learning EFT likelihood ratios that the BIT estimator is built upon.","marker":"[43]"},{"why":"Provides the cross-entropy loss function that the BIT training minimizes to estimate the likelihood ratio components.","marker":"[44]"},{"why":"Justifies the rotation to the mass-eigenstate basis that defines gZZ2 and gZZ4 and removes unconstrained directions in coefficient space.","marker":"[46]"},{"why":"Gives the LHC EFT working group prescription for simulating SMEFT effects with LO+1j matrix-element reweighting.","marker":"[68]"},{"why":"Supplies the SMEFTsim 3.0 package used to generate the per-event SMEFT weights for the six considered operators.","marker":"[72]"}],"fun_headline_variants":["Six SMEFT couplings in Higgs-strahlung all match SM, CMS fit","CMS: Higgs-strahlung data shows no SMEFT deviation across six operators","First CP-sensitive SMEFT limits in VH(bb) from CMS, all SM-consistent","Six Wilson coefficients from 138/fb VH data: no new physics","CMS SMEFT fit: six couplings in H+W/Z production stay standard"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"That the q < 1 and q < 4 thresholds define 68% and 95% confidence intervals for the quadratic SMEFT parametrization, a premise the paper itself flags as not guaranteed because Wilks-theorem regularity conditions are violated.","fun_headline_variants_meta":{"raw":{"variants":["Six SMEFT couplings in Higgs-strahlung all match SM, CMS fit","CMS: Higgs-strahlung data shows no SMEFT deviation across six operators","First CP-sensitive SMEFT limits in VH(bb) from CMS, all SM-consistent","Six Wilson coefficients from 138/fb VH data: no new physics","CMS SMEFT fit: six couplings in H+W/Z production stay standard"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000725,"raw_usage":{"total_tokens":3251,"prompt_tokens":950,"completion_tokens":2301,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":566,"completion_tokens_details":{"reasoning_tokens":2194}},"tokens_in":566,"tokens_out":2301,"duration_ms":17928,"temperature":1.0,"reasoning_tokens":2194,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:45:13.277930+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-extract the quadratic-model intervals using a Neyman construction or Monte Carlo calibration on simulated pseudo-data and check whether the q < 1 and q < 4 thresholds actually have 68% and 95% coverage.","supporting_citations":[{"cited_title":"Constraining anomalous Higgs boson couplings to virtual photons","cited_arxiv_id":"2109.13363","evidence_quote":"Justifies the rotation to the mass-eigenstate basis that defines gZZ2 and gZZ4 and removes unconstrained directions in coefficient space."},{"cited_title":"LHC EFT WG Note: SMEFT predictions, event reweighting, and simulation","cited_arxiv_id":"2406.14620","evidence_quote":"Gives the LHC EFT working group prescription for simulating SMEFT effects with LO+1j matrix-element reweighting."}],"review_version":1}