{"id":"69344ece-2da0-4be1-92b5-9f456c093d9b","arxiv_id":"2506.13002","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Simulation-based projections show longitudinally polarised W bosons in WH production could be measured at 5-sigma significance with 300 fb^-1 at the LHC.","lead":"This paper proposes an analysis strategy for measuring longitudinally polarized W and Z bosons produced along with the Higgs boson at the LHC, using machine learning to separate them from transverse polarization. The authors' simulations predict a 5-sigma observation for the W case with 300 inverse femtobarns of data, making the measurement feasible at the LHC.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quoted >5σ at 300 fb^-1 contradicts the paper's own Table 4: the 68% interval [0.74,1.28] for μ_WLH implies σ≈0.27 and z≈3.7 under the same asymptotic approximation.","rationale":"Reader's verdict CONDITIONAL is appropriate; the direction (a feasibility study for measuring longitudinally polarised VH production) is well-motivated and the simulation chain is standard. But the strongest claim is a quantitative projection, and the paper's own outputs disagree. This is more load-bearing than the LO polarisation-fraction worry: even if higher-order corrections shift template shapes, a real experimental analysis could in principle use data-driven templates, but no amount of re-analysis can reconcile 5σ with a 68% interval of ±0.27 under the stated asymptotic procedure. The reader flagged the statistical presentation in the rationale, although the 'weakest assumption' field selected the LO polarisation fractions; I therefore mark agreement as partial. The correct handling is to retain a conditional assessment: the proposal is worth pursuing, but the quoted reach should be re-derived and either corrected or explained before being used as a planning input. If the inconsistency resolves in the direction of z≈3.7, the abstract's central claim weakens; if it resolves in the direction of z≥5, the table and curve-minimum description need repair. Either way the verdict remains conditional pending the check.","tokens_in":19481,"tokens_out":6076,"duration_ms":75428,"concrete_test":"Re-run the described pyhf maximum-likelihood fit on the binned BDT templates from Fig. 8: build the Asimov dataset as W_LH+W_TH+Wγγ (signal injected at μ=1), profile all nuisance parameters, locate the minimum of -2ΔNLL, and record both the 68% interval (2ΔNLL=1) and q(μ=0). If the minimum is at μ≈1, σ≈0.27, and z0≈3.7, then the >5σ claim in the abstract and Sec. 4 is unsupported and should be corrected. If instead z0≥5, Table 4's 68% interval is incompatible with the stated asymptotic approximation and needs to be reissued. This check uses only the paper's own templates and code stack (MadGraph/Delphes/pyhf) and settles which of the two reported numbers is wrong.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central numerical claim rests on the profile-likelihood outputs in Sec. 4 and Table 4, and those outputs are internally inconsistent. For the W_LH channel at 300 fb^-1, Table 4 gives a 68% interval for μ_WLH of [0.74,1.28]. Under the quadratic/asymptotic approximation stated in Sec. 3.1, this corresponds to a one-standard-deviation uncertainty σ≈0.27. The expected discovery significance for an injected signal at μ=1 is then z0≈1/σ≈3.7, not >5.0. The same contradiction appears with the text's claim of 5.0σ at 280 fb^-1. Furthermore, Sec. 4 says all NLL minima are at zero, while the Asimov datasets described in Sec. 3.1 are the sum of W_LH+W_TH+Wγγ, for which the true signal strength is μ=1; the confidence intervals in Table 4 are indeed centred near μ=1. One of these statements must be wrong. Because every headline precision at 300/1000/3000 fb^-1 derives from these NLL curves, the claimed sensitivity numbers are not currently reproducible. This is not a question of external NLO polarisation corrections; it is an internal consistency check that fails.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes measurement strategies for longitudinally polarised W and Z bosons in associated WH and ZH production at the 14 TeV LHC, using the l nu gamma gamma and l l gamma gamma final states. Signal and dominant triboson backgrounds are simulated with MadGraph5_aMC@NLO, Pythia8, and Delphes; polarisation-sensitive observables are combined in boosted decision trees; and binned profile-likelihood fits with pyhf are used to project sensitivities from Asimov datasets. The headline results are a >5.0 sigma significance for W_LH with 300 fb^-1 (also stated as 5.0 sigma with 280 fb^-1), inclusive cross-section precisions of approximately 30%, 15%, and 10% at 300, 1000, and 3000 fb^-1, and about 3.4 sigma significance with 35% precision for Z_LH at 3000 fb^-1.","tokens_in":19704,"tokens_out":7091,"duration_ms":83773,"significance":"If correct, the projected measurements would constitute the first observation of longitudinally polarised vector-boson production in association with a Higgs boson, providing a direct probe of electroweak symmetry breaking that complements existing diboson polarisation measurements. The paper benefits from a standard and reproducible simulation chain, large generated event samples, normalisation of signal samples to NNLO QCD plus NLO EW cross-sections, a detailed selection table, cross-validated BDTs, and use of public statistical tools. However, the headline sensitivity numbers are not currently reproducible from the paper's own statistical outputs: the quoted significances are incompatible with the confidence intervals in Table 4 under the stated asymptotic approximation, and the description of the NLL minima is inconsistent with the Asimov construction. The polarisation-fraction inputs are also unvalidated beyond leading order, which directly affects the signal yields and BDT separation.","major_comments":[{"comment":"The paper contains an internal inconsistency that affects the central claim. The Asimov datasets are defined in Sec. 3.1 as the sum of W_LH, W_TH, and W_gamma-gamma templates, i.e. with signal injected at mu=1, but Sec. 4 states that all NLL curve minima are located at zero. These two statements cannot both be true: if the Asimov data contain signal, the NLL minimum should be near mu=1, and Table 4 indeed shows 68% intervals centred near 1 (e.g. [0.74,1.28] for mu_WLH at 300 fb^-1). More importantly, the quoted significance of >5.0 sigma is not compatible with that interval. Under the asymptotic approximation stated in Sec. 3.1, a 68% interval half-width of about 0.27 implies sigma ≈ 0.27, so an injected signal at mu=1 gives an expected discovery significance of about 3.7 sigma, not 5.0 sigma. The same check for Z_LH at 3000 fb^-1 gives a 68% interval [0.65,1.35], i.e. sigma ≈ 0.35 and a significance of about 2.9 sigma, not 3.4 sigma. The authors must correct either the significance calculation or the fit outputs, and then revise all derived precision and significance numbers accordingly.","section":"Sec. 3.1, Sec. 4, Table 4"},{"comment":"The polarisation fractions and template shapes are taken from LO MadGraph+Pythia samples normalised to the NNLO+NLO total cross-section, but the relative fractions of longitudinal and transverse polarisation are not cross-checked against any higher-order polarisation calculation. These fractions enter directly into the signal yields, the BDT training, the template shapes, and therefore every quoted significance and precision. Higher-order QCD or electroweak corrections can redistribute the longitudinal and transverse contributions, and the paper provides no estimate of this effect. I ask the authors to validate the fractions against an NLO polarised VH calculation or to provide a conservative variation of the L/T fraction and show its impact on the projected sensitivities.","section":"Sec. 2.2, Table 1, Sec. 4"},{"comment":"The analysis omits backgrounds from non-prompt and mis-identified photons, considering only W_gamma-gamma and Z_gamma-gamma. The conclusion argues that statistical uncertainties will dominate and that data-driven methods can constrain backgrounds, but this is asserted rather than demonstrated. For a measurement proposal in the H->gamma-gamma channel, the reducible background contribution is a standard and potentially important component of the sensitivity. I request at least a rough estimate of the fake-photon background yield after the selection, or a demonstration that sidebands and control regions would render it negligible; without this, the absolute normalisation of the projected significance is not fully supported.","section":"Sec. 2.1 and Sec. 5"}],"minor_comments":[{"comment":"The abstract states a >5.0 sigma significance with 300 fb^-1, while Sec. 4 states 5.0 standard deviations with 280 fb^-1; these two statements need to be reconciled and the final number made consistent with the corrected statistical analysis.","section":"Abstract and Sec. 4"},{"comment":"The 95% CL entry for mu_ZLH at 300 fb^-1 is shown as a dash; the reason (e.g. upper limit outside the scan range or not converged) should be stated explicitly.","section":"Table 4"},{"comment":"The sentence 'with all curve minima located at zero, representing the Standard Model hypothesis' is unclear: if the minima were truly at zero, the confidence intervals in Table 4 would be centred near zero rather than near one. This wording should be corrected regardless of the outcome of the significance re-analysis.","section":"Sec. 4"},{"comment":"There are several typographical errors, including 'trainng' in Sec. 2.4, 'Transeverse' in Table 3, 'accross' in Sec. 3.2, and 'Transerse' in the Figure 10 caption; these should be corrected in a revision.","section":"Sec. 2.4 and Table 3"},{"comment":"In the conclusion, 'Z_L^±H' should presumably be 'Z_LH'; the superscript plus/minus is not appropriate for the neutral Z boson.","section":"Sec. 5"},{"comment":"Reference [36] appears incomplete, lacking a journal, volume, and year; a full citation should be provided.","section":"Bibliography"}],"recommendation":"major_revision","confidential_remarks":"The internal inconsistency in the significance claim is serious enough that the headline numbers should not be quoted until corrected, but it is fixable within the scope of the manuscript. I would like to see a revised version where the fits are re-run or the text is corrected so that the NLL minima, the confidence intervals, and the quoted significances are mutually consistent. The missing NLO validation of polarisation fractions and the absence of a fake-photon background estimate are additional load-bearing points that need to be addressed before the projections can be regarded as robust."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a serious, well-documented feasibility study of measuring longitudinally polarised W and Z bosons in VH production with H→γγ, and the specific projections — WLH and ZLH in the ℓνγγ and ℓℓγγ final states, with BDT separation and full likelihood fits — are new. The simulation chain is described in enough detail to re-implement: MadGraph+Pythia+Delphes, MLM merging, the polarisation-specific generation syntax, and a HistFactory/pyhf fit with MC statistical uncertainties, cross-section and branching-ratio uncertainties, and a 25% normalisation on the triboson backgrounds. The BDT discrimination (AUC ~0.78–0.80) is credible, and the paper is honest about the simplifications: no fake-photon backgrounds, no full detector systematics, LO polarisation fractions.\n\nThe soft spot is not the simplifications; it is the internal consistency of the statistical outputs. The abstract and Sec. 4 claim >5σ for WLH at 300 fb^-1, but Table 4 gives a 68% interval for μ_WLH of [0.74,1.28] at 300 fb^-1. Under the asymptotic approximation stated in Sec. 3.1, that interval implies σ≈0.27 and a discovery significance z≈3.7, not 5.0. The text also says all NLL curve minima are at zero, while the Asimov datasets are explicitly signal-injected at μ=1 and the Table 4 intervals are centered near 1. At least one of these statements is wrong, and because every quoted precision and significance in the abstract comes from these fits, the headline numbers are not currently reproducible. This is a real flaw, not a stylistic one.\n\nThe other concern, LO polarisation fractions with no NLO cross-check, is moderate: the total cross-sections are normalised to NNLO+NLO, but the L/T split and the template shapes come from LO MadGraph. A sizeable QCD or EW correction to the polarisation fractions would shift both the BDT separation and all yields. The missing fake-photon backgrounds are acknowledged and likely subdominant, but a data-driven estimate would strengthen the case.\n\nBottom line: the direction is right and the methodology is largely sound, but the central projection must be re-derived or explained before the paper can serve as a planning input. I would send it to a serious referee, with the explicit request to reproduce the fits and correct the significance and NLL-minimum statements. If that comes back clean, it becomes a useful reference; as it stands, the abstract overstates what the paper itself shows.","headline":"A genuinely new feasibility study for polarised VH with H→γγ, but the 5σ headline contradicts the paper's own Table 4 and the statistical reporting needs correction before the projections can be used.","tokens_in":20291,"tokens_out":3965,"would_cite":false,"duration_ms":38130,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper projects a >5σ measurement of longitudinally polarised W-boson production in association with a Higgs boson at the LHC using 300 fb^-1 of 14 TeV data.","keywords":["longitudinal vector boson polarisation","W H production","Z H production","Higgs-Strahlung","electroweak symmetry breaking","boosted decision trees","LHC projections","H to gamma gamma"],"falsifier":"A next-to-leading-order calculation of polarised WH and ZH production, or a measurement of the vector-boson pT spectrum in the ℓνγγ and ℓℓγγ channels with early LHC data, would show whether the leading-order polarisation mixture used here is correct; a significant mismatch would invalidate the quoted projections.","tokens_in":19230,"feed_emoji":"⚛️","tokens_out":7618,"duration_ms":89200,"temperature":0.7,"pith_summary":"The paper proposes the first dedicated measurement of longitudinally polarised W and Z bosons produced together with a Higgs boson (W_LH and Z_LH) at the LHC, using the H→γγ decay and leptonic W/Z decays. Its central projection is that W_LH production can be established with a significance above 5.0 standard deviations with 300 $fb^{-1}$ of 14 TeV proton-proton data, and that the inclusive W_LH cross section can be measured to about 30%, 15%, and 10% precision at 300, 1000, and 3000 $fb^{-1}$. For Z_LH production it projects about 3.4σ significance and 35% precision at 3000 $fb^{-1}$. Because longitudinal polarisation is acquired through electroweak symmetry breaking, such a measurement would probe the Higgs mechanism directly and could expose deviations from the Standard Model. The authors therefore recommend carrying these studies out on actual LHC data.","feed_headline":"W_LH production could cross 5σ with 300 fb^-1 of LHC data","feed_subtitle":"A proposed LHC analysis separates longitudinal from transverse W/Z polarisation using H→γγ decays and leptonic final states.","key_machinery":"The load-bearing machinery is a set of polarisation-sensitive observables defined in the rest frames of the VH system and of the vector boson: cos θ*, cos θ1, Φ1, and ΔΦ(L, Lbar), supplemented by the vector-boson and Higgs transverse momenta, rapidities, the VH invariant mass, and lepton pT balance. These variables exploit the fact that longitudinally polarised bosons are produced more centrally and decay perpendicular to their flight direction, whereas transversely polarised bosons decay more along it. For the W channel the neutrino momentum is reconstructed from missing transverse momentum with an on-shell W-mass constraint. Boosted decision trees trained on these inputs produce a score distribution that serves as the observable in a binned maximum-likelihood fit, with the polarisation templates normalised to state-of-the-art NNLO QCD and NLO electroweak cross sections.","core_discovery":"The central claim is that longitudinal and transverse polarisation states of the vector boson in WH and ZH production can be separated well enough to measure the longitudinal component. Separation is achieved with rest-frame angular observables, namely the production angle θ*, the lepton angle θ1, the azimuth Φ1, and the lepton azimuthal separation ΔΦ(L, Lbar), supplemented by kinematic variables and combined in boosted decision trees. The resulting BDT score distribution is fitted with a binned profile likelihood built from Asimov datasets, yielding expected 68% and 95% confidence intervals for the signal-strength parameters μ_WLH and μ_ZLH. These intervals are translated into the quoted significances and cross-section precisions, with W_LH production projected to exceed 5σ already at 300 $fb^{-1}$.","pith_inferences":["The quoted significances depend on the leading-order polarisation fractions used for the templates; a dedicated NNLO polarised calculation would test whether the 5σ projection survives, something the paper does not provide.","Because the longitudinal fraction grows with vector-boson pT and asymptotically tracks Goldstone-boson equivalence, a pT-binned measurement would effectively test electroweak restoration; the paper notes this possibility but leaves it to future work.","The neglected non-prompt and misidentified-photon backgrounds could alter the analysis if they populate the BDT signal region; the paper argues they are small, but data-driven background estimates would be the natural check.","The same rest-frame observables could be applied to vector-boson-fusion Higgs production or H→VV decays, though those channels are restricted to energies near the Higgs mass; the VH channel is the one that reaches the TeV scale."],"forward_implications":["With 300 fb^-1, W_LH production would be observable at more than 5σ, turning the longitudinal W-Higgs coupling into a measurable Standard Model property rather than a theoretical construct.","At 3000 fb^-1 the projected 10% inclusive precision for W_LH opens the door to pT-binned simplified template cross sections, which probe the energy dependence of the longitudinal fraction.","A Z_LH measurement at about 3.4σ and 35% precision with 3000 fb^-1 would extend the same polarisation programme to the neutral vector boson.","Deviations in the measured polarisation composition could signal a composite Higgs boson or contributions from new heavy W' and Z' gauge bosons decaying to V_LH.","The strategy transfers to other Higgs decay modes such as H→bb and H→ττ, although the extra neutrinos and larger backgrounds degrade the polarisation-sensitive observables."],"supporting_citations":[{"why":"Supplies the NNLO+NLO VH cross-sections and the H→γγ branching ratio used to normalise signal samples and their uncertainties.","marker":"[31]"},{"why":"Provides the NNLO QCD calculation of VH production used for signal normalisation.","marker":"[44]"},{"why":"Provides the NLO electroweak corrections to VH production used for signal normalisation.","marker":"[45]"},{"why":"Supplies the fast detector simulation that defines reconstruction-level objects and selection efficiencies.","marker":"[48]"},{"why":"Establishes the polarisation-sensitive angular observables in pp→ZH and pp→WH that the BDTs are trained on.","marker":"[52]"},{"why":"Gives the neutrino momentum reconstruction used in the W±H→ℓνγγ channel.","marker":"[54]"},{"why":"Provides the asymptotic profile-likelihood formulae used to convert BDT-score fits into significances and confidence intervals.","marker":"[57]"},{"why":"Implements the statistical model and likelihood minimisation used for the projections.","marker":"[59]"}],"fun_headline_variants":["W_LH: 5σ discovery potential with 300 fb^-1 at LHC","Longitudinal W in WH: 5σ reach with 300 fb^-1","Proposed analysis yields 5σ for W_LH at 300 fb^-1","Measuring W_LH polarisation: 5σ with 300 fb^-1 at LHC","ML separation enables 5σ for longitudinal W in WH"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The projections assume that the relative fractions and template shapes of longitudinal versus transverse polarisation in VH production are correctly described by leading-order simulations normalised to the NNLO+NLO total cross-section; if higher-order QCD or electroweak corrections move events between polarisation states, every quoted significance and precision changes.","fun_headline_variants_meta":{"raw":{"variants":["W_LH: 5σ discovery potential with 300 fb^-1 at LHC","Longitudinal W in WH: 5σ reach with 300 fb^-1","Proposed analysis yields 5σ for W_LH at 300 fb^-1","Measuring W_LH polarisation: 5σ with 300 fb^-1 at LHC","ML separation enables 5σ for longitudinal W in WH"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001271,"raw_usage":{"total_tokens":5274,"prompt_tokens":1092,"completion_tokens":4182,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":708,"completion_tokens_details":{"reasoning_tokens":4075}},"tokens_in":708,"tokens_out":4182,"duration_ms":30598,"temperature":1.0,"reasoning_tokens":4075,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:38:09.656312+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A next-to-leading-order calculation of polarised WH and ZH production, or a measurement of the vector-boson pT spectrum in the ℓνγγ and ℓℓγγ channels with early LHC data, would show whether the leading-order polarisation mixture used here is correct; a significant mismatch would invalidate the quoted projections.","supporting_citations":[{"cited_title":"W boson polarization as a measure of gauge-Higgs anomalous couplings at the LHC","cited_arxiv_id":"1805.06602","evidence_quote":"Gives the neutrino momentum reconstruction used in the W±H→ℓνγγ channel."}],"review_version":1}