{"id":"db21c719-06e5-4378-861d-1c9f4f3ec4ef","arxiv_id":"2412.14452","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"By recasting ATLAS's Z to three dark photon analysis, the authors set the best limits to date on the exotic Higgs decay chain H to aa to VVVV to eight fermions, bounding the branching fraction to roughly 4e-5 to 1e-4.","lead":"Researchers reinterpreted an existing ATLAS search, designed for Z boson decays to three dark photons, to constrain a new exotic Higgs decay into eight leptons. This narrows the space for hidden sector models and shows how published LHC results can be repurposed to test new theories.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unvalidated transfer of rlep/rtrig from Z→6f to H→8f makes the low-ma BR limits potentially over-optimistic; the quoted 4e-5–1e-4 range needs a detector-level cross-check.","rationale":"The paper's reproduction of the ATLAS Z→6f MC study is credible: the cut-flow efficiencies match within a few percent after applying constant recalibration factors, and the authors assign a large log-normal systematic (sigma_epsilon=0.59) that makes the limits statistically conservative. The recast idea is sound and the order-of-magnitude conclusion that BR(H→aa→VVVV) is constrained near 1e-4 is likely robust. The single most load-bearing step is the transfer of rlep and rtrig from Z→6f at m_hD=50 GeV to H→aa→VVVV over the full (ma,mV) grid, with no validation against any H→8f Monte Carlo or detector-level simulation. This is especially concerning in the low-ma, low-mV region, where the paper itself identifies collimation-induced isolation losses as a dominant effect. A constant per-lepton efficiency cannot capture a geometry-dependent isolation inefficiency, and the trigger turn-on may also differ because the a bosons are far more boosted than the hD in Z→6f. This does not invalidate the paper's main qualitative conclusion, but it means the specific numeric limits, and the claimed ma dependence of those limits, should be treated as conditional pending a detector-level cross-check. The reader's weakest_assumption identifies the same issue, so I agree with the reader's CONDITIONAL verdict and see no reason to change it.","tokens_in":14695,"tokens_out":13877,"duration_ms":122118,"concrete_test":"Run a fast detector simulation, e.g., Delphes 3 with the ATLAS Run-2 card and tuned electron/muon reconstruction, isolation, and trigger turn-on efficiencies, on H→aa→VVVV→8f samples at (ma,mV) = (20,7), (30,10), and (50,15), as well as on the Z→6f control points (mhD=50, mA'=8,15,20) used in Section 3.1.1. Apply the exact ATLAS cut flow and compare the resulting overall signal efficiencies to those obtained with the paper's rlep=0.78, rtrig=0.81 model. If the control points match within ~10% but the H→8f efficiencies deviate by more than ~30% (or more than the assigned sigma_epsilon) in the low-ma bins, the quoted BR range must be rescaled and the low-ma limits are not reliable.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central numerical claim (Section 4: BR(H→aa→VVVV) between 4e-5 and 1e-4) depends on transferring the constant recalibration factors rlep=0.78 per lepton and rtrig=0.81 from the ATLAS Z→6f MC study (Section 3.1.1, fitted at m_hD=50 GeV) to the H→aa→VVVV→8f topology over the whole (ma,mV) grid. The paper provides no validation of this transfer: Section 4 only notes that the lepton pT spectra are 'roughly as expected' because mZ/6≈mH/8, and that average |η| is slightly smaller. The transfer is least secure at low ma and mV. The paper itself states in Section 4 that at low ma the a is much more boosted than the hD in Z→6f, making its decay products more collimated and more likely to fail isolation; Section 5 reiterates that the ATLAS isolation variables include energy from other leptons. However, a constant per-lepton recalibration fitted to the Z→6f topology cannot model this geometry dependence of the detector-level isolation (nor the trigger turn-on for a different lepton pT spectrum). If the true low-ma efficiency is lower, the corresponding limits are too strong by a topology-dependent factor. The large sigma_epsilon=0.59 covers the width of the efficiency uncertainty, but not a bias in its central value, and the low-ma, low-mV points lie inside the quoted range. Thus the order-of-magnitude conclusion is probably robust, but the specific numeric range, and especially the ma dependence, is not established without revalidation.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper recasts the ATLAS search for Z -> hD A' -> A'A'A' with four-lepton final states (ATLAS PRL 131, 251801) to constrain the exotic Higgs decay H -> aa -> VVVV -> 8f. The authors first reproduce ATLAS's Monte Carlo study for Z -> 6f using truth-level simulation with two constant recalibration factors, rlep = 0.78 per lepton and rtrig = 0.81 for triggering, and then benchmark their limit-setting procedure against ATLAS's published limits, finding agreement within a factor of 1.5 to 1.9. They then apply the same methodology to H -> aa -> VVVV, assuming dark-photon-like branching fractions for V and BR(a -> VV) = 1, and obtain 95% CL limits on BR(H -> aa -> VVVV) of roughly 4e-5 to 1e-4 across the (ma, mV) grid. The central numerical claim, and especially its dependence on ma and mV, rests on transferring detector-level efficiency corrections fitted at one mass point in a different topology to the H -> 8f final state.","tokens_in":14947,"tokens_out":5061,"duration_ms":40893,"significance":"If the central claim holds, this is the strongest limit on H -> aa -> VVVV -> 8f, improving on earlier multilepton recasts by roughly an order of magnitude, and it demonstrates that a published four-lepton analysis can be repurposed for novel multi-resonance signatures. The paper is transparent about its crude detector modeling, makes use of public ATLAS background bins and MC benchmarks rather than fitting to the Higgs-signal hypothesis, and is therefore not circular. The authors also give credit to, and compare with, competing recasts and newer ATLAS/CMS searches. The main significance risk is that the quoted numerical range, especially at low ma and mV, may be systematically too strong because the efficiency corrections are transferred from a different topology.","major_comments":[{"comment":"The central claim in Section 4, that BR(H -> aa -> VVVV) is bounded between 4e-5 and 1e-4, depends directly on applying rlep = 0.78 and rtrig = 0.81, fitted to the ATLAS Z -> 6f MC study at m_hD = 50 GeV, to the H -> 8f topology over the full (ma, mV) grid. This transfer is not validated. The paper itself notes in Sections 4 and 5 that at low ma the a is more boosted than the hD in Z -> 6f, so its decay products are more collimated and more likely to fail ATLAS isolation, and that the isolation variables include energy from other leptons. A constant per-lepton recalibration cannot model this geometry-dependent isolation efficiency. The heuristic that mZ/6 ~ mH/8 only speaks to average lepton pT, not to isolation-cone occupancy or trigger turn-on behavior. Without a detector-level cross-check, or at least a conservative topology- and mass-dependent efficiency model, the numerical limits in low-ma, low-mV regions are not established.","section":"Section 4 and Section 3.1.1, Tables 1-2"},{"comment":"The large systematic sigma_epsilon = 0.59 in Eq. (3.5) broadens the signal-efficiency nuisance distribution, but it cannot correct a bias in the central value of the transferred efficiency. If the true H -> 8f efficiency at low (ma, mV) is lower than the Z -> 6f-calibrated value by, say, 30-50%, the resulting limit on BR(H -> aa -> VVVV) weakens by a comparable factor, and the quoted range 4e-5 to 1e-4 would be over-optimistic in that region. The paper should either demonstrate that the bias is negligible with an independent check or present limits under a more conservative efficiency assumption, reporting the dependence of the final range on that assumption.","section":"Section 3.2, Eq. (3.5)"},{"comment":"The trigger/cut-flow ordering test in Section 3.1.2 is performed for Z -> 6f at m_hD = 50 GeV only and does not address the H -> 8f topology, where the lepton pT spectrum and the trigger turn-on can differ at low ma. Similarly, the validation against ATLAS's final Z -> 6f limits in Figure 4, while useful, uses the same fitted recalibration factors and therefore cannot validate their extrapolation to H -> 8f. A separate validation of the H -> 8f signal efficiency, at least at benchmark points spanning low and high (ma, mV), is needed before the quoted central limits can be taken at face value.","section":"Section 3.1.2 and Figure 4"}],"minor_comments":[{"comment":"The title page contains 'recast a TLAS', which appears to be a corruption of 'recast ATLAS'; the grammar in the abstract, 'limit on a exotic Higgs decay mode', should also be corrected.","section":"Title/header"},{"comment":"There is a typo in 'PYTHIA 3.811' which should presumably read 'PYTHIA 8.311'; note also that Section 3.1.1 uses PYTHIA 8.308, and the version difference is not discussed.","section":"Section 4"},{"comment":"The word 'Acouting' in 'Acouting both for the statistical uncertainty' is a typo for 'Accounting'.","section":"Section 5"},{"comment":"Reference [17] is incomplete, ending with a comma and no arXiv identifier; reference [47] cites private communication, which is not independently verifiable and should be flagged as such in the text.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the manuscript is well-suited in scope to a hep-ph journal and the authors are unusually candid about the limitations of their simplified detector modeling. The principal issue is load-bearing: the central numeric limits for H -> 8f rely on an unvalidated transfer of efficiency corrections from Z -> 6f to a more collimated final state. If the authors can supply a detector-level sanity check, or show explicitly how the quoted range changes under a conservative efficiency model, the result would be publishable. I saw no citation-pattern or novelty-disclosure concerns."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does something new and useful: it recasts ATLAS's Z->6f four-lepton analysis to constrain H->aa->VVVV->8f, deriving limits around 4e-5 to 1e-4, roughly an order of magnitude stronger than the prior recast by Izaguirre and Stolarski. The reproduction of ATLAS's Monte Carlo study within 8% and the final ATLAS limits within a factor of 1.5-1.9 gives me real confidence in the overall approach. The authors are transparent about their crude treatment of detector effects, and the large log-normal systematic (sigma_epsilon ~ 0.59) makes the limits conservative in a statistical sense. The discussion of why the m4l < mZ cut makes this search applicable to H->8f is clear, and the comparison with the multilepton-counting search is a useful addition.\n\nThe main soft spot is exactly where the reader put it: the transfer of the constant recalibration factors rlep = 0.78 and rtrig = 0.81 from the Z->6f topology at m_hD = 50 GeV to the entire (ma, mV) grid for H->8f. The paper itself acknowledges that at low ma the a bosons are more boosted, making leptons more collimated and more likely to fail isolation. A constant per-lepton efficiency cannot capture this geometry dependence, so the low-ma end of the quoted range is the least secure. The large sigma_epsilon covers the width of the efficiency uncertainty but not a bias in its central value. I don't think this sinks the paper, because the order-of-magnitude conclusion probably survives, but the specific numbers and their ma dependence should be treated with caution until this is revalidated or a convincing argument is provided. A second, smaller issue is that no code, no numeric table of limits, and no expected-limit bands are provided, which limits reproducibility and makes it hard for others to use the results directly.\n\nI read the cited [31] carefully and agree that the authors' limits are indeed stronger, so the claim of \"best limits to date\" is likely correct, though a quick check against the newly appeared ATLAS and CMS searches they mention would strengthen the statement.\n\nThis paper is for pheno colleagues working on exotic Higgs decays, hidden sectors, and dark photons, and for experimentalists looking for new search ideas. It merits a serious referee: the central conclusion is probably robust, but a referee should push for validation or a quantitative justification of the efficiency transfer at low ma, and for the numbers to be tabulated. I would engage with it.","headline":"A genuinely new recast that likely sets the best current limits on H->aa->VVVV, but the quoted numbers rest on an unvalidated constant-efficiency transfer at low ma.","tokens_in":15626,"tokens_out":1723,"would_cite":true,"duration_ms":15376,"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":"By recasting the ATLAS Z→dark-photon search, this paper places 95% confidence limits on the exotic Higgs decay H→aa→VVVV between 4×10⁻⁵ and 1×10⁻⁴, arguing these are the best limits so far on this signal.","keywords":["exotic Higgs decays","hidden valley","dark photon","eight-fermion final state","ATLAS recast","branching fraction limit","H→aa→VVVV","four-lepton analysis"],"falsifier":"Repeat the recast at $(m_a,m_V)=(20,7)\\,\\mathrm{GeV}$ using a full detector simulation, or ATLAS's published lepton reconstruction/identification efficiencies parameterized by $p_T$ and by the separation between nearby leptons, in place of the constant $r_\\mathrm{lep}=0.78$; if the resulting signal efficiency falls below the value used here by more than the quoted systematic uncertainty, the low-$m_a$ limits are too strong.","tokens_in":14327,"feed_emoji":"⚛️","tokens_out":10703,"duration_ms":65806,"temperature":0.7,"pith_summary":"The paper aims to show that a public ATLAS search, originally written for $Z$ bosons decaying to three dark photons, already constrains a different exotic Higgs decay mode: $H \\to aa \\to VVVV$, where two intermediate scalars each produce two spin-one particles that decay to fermion pairs. Because the ATLAS selection accepts four-lepton events with $m_{4\\ell} < m_Z - 5\\,\\mathrm{GeV}$ rather than requiring $m_{4\\ell} = m_H$, it has sensitivity to $H \\to 8f$ decays that dedicated four-lepton Higgs searches would miss. Reproducing the ATLAS efficiencies with two constant recalibration factors, the authors recast the analysis and obtain 95% confidence upper limits on the branching fraction (decay probability) $\\mathrm{BR}(H\\to aa\\to VVVV)$ in the range $4\\times10^{-5}$ to $1\\times10^{-4}$ over the probed $(m_a,m_V)$ grid, assuming $V$ has dark-photon-like leptonic branching fractions and $\\mathrm{BR}(a\\to VV)=1$. If correct, these are the strongest limits obtained so far on this signal, and they narrow the viable parameter space of hidden-sector models coupled to the Higgs.","feed_headline":"Higgs branching to eight fermions capped near one in ten thousand","feed_subtitle":"A recast ATLAS analysis yields the best limit yet on Higgs decays through hidden-vector cascades.","key_machinery":"The carrying object is the ATLAS signal region itself, defined by at least four isolated leptons forming two same-flavor/opposite-charge pairs with $m_{4\\ell} < m_Z - 5\\,\\mathrm{GeV}$, pair-mass ratio $m_{34}/m_{12} > 0.85$, and dilepton masses away from the $\\Upsilon$ resonances and below 5 GeV. Because this window is set below the $Z$ mass rather than at the Higgs mass, it is kinematically open to $H\\to aa\\to VVVV$ events. The efficiency transfer is carried by two fitted constants, $r_\\mathrm{lep}=0.78$ per lepton and $r_\\mathrm{trig}=0.81$, which convert truth-level acceptance into reconstruction-level efficiency; limits are then computed with the CL$_s$ method on the two leading $\\bar m_{\\ell\\ell}\\equiv (m_{\\ell_1\\ell_2}+m_{\\ell_3\\ell_4})/2$ bins per mass point, with log-normal nuisance parameters for the signal efficiency and background.","core_discovery":"The central claim is that the ATLAS $Z\\to 6f$ analysis, after being reproduced with a simplified simulation, can be reinterpreted as a limit on $H\\to aa\\to VVVV\\to 8f$. The reproduction uses truth-level $Z\\to A'h_D\\to A'A'A'$ simulation with $r_\\mathrm{lep}=0.78$ per lepton and $r_\\mathrm{trig}=0.81$ as constant recalibration factors, matching the ATLAS Monte Carlo study's overall signal efficiencies to within about 8% and reproducing ATLAS's limits within a factor of 1.5–1.9. Applying the same tuned efficiencies to simulated Higgs production, with cross sections reweighted to LHC Higgs working group recommendations and a combined $\\sigma_\\epsilon \\simeq 0.59$ systematic uncertainty, the recast yields 95% CL upper limits on $\\mathrm{BR}(H\\to aa\\to VVVV)$ between $4\\times10^{-5}$ and $1\\times10^{-4}$ over the scanned $(m_a, m_V)$ plane. The limits are weakest at small $m_a$, where the boost of $a$ collimates its decay products so that leptons fail isolation cuts and $m_{4\\ell}$ is pushed toward the kinematic ceiling $m_H - 2m_V$, beyond the search's $m_Z$ window. With $V$ treated as a dark photon and $\\mathrm{BR}(a\\to VV)=1$, the paper states that these are, to its knowledge, the best limits obtained so far on such a signal.","pith_inferences":["The closeness of the two-constant-factor reproduction (within 8% on efficiency, 1.5–1.9 on limits) suggests that public LHC searches can often be recast with a very coarse detector model, but the roughly 59% systematic uncertainty in this paper is dominated by that coarseness; a full simulation would likely sharpen the quoted bounds.","The paper's own diagnosis at low $m_a$ implies a concrete improvement: if isolation cones excluded other leptons, the $m_a$ dependence would flatten and the low-mass limits should drop below $4\\times10^{-5}$; this is testable with the same public search.","The same pipeline should transfer to confining hidden-sector models with dark showers, where the paper notes much larger theoretical uncertainties are expected because hadronization in non-QCD-like sectors is poorly understood.","Because the bound applies to any promptly decaying spin-one particle with dark-photon-like leptonic branching fractions, it also constrains composite 'hidden rho' interpretations, not only elementary dark photons."],"forward_implications":["Hidden-sector models predicting $\\mathrm{BR}(H\\to aa\\to VVVV)$ above roughly $10^{-4}$ with two or more leptonic $V$ decays are excluded by existing 139 fb$^{-1}$ ATLAS data.","The kinematic-pairing search improves on the earlier multilepton-counting recast by close to an order of magnitude, moving the bound from the $10^{-3}$ range to the $10^{-5}$–$10^{-4}$ range.","Because the selection uses $m_{4\\ell}