{"id":"88182a42-354a-44b2-bf26-9ef0b3f15e47","arxiv_id":"2412.19269","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"For a benchmark B-L extension with a 3 TeV Z' and 420 GeV right-handed neutrinos, the HL-LHC could reach discovery-level significance in three final states using BDT-based event selection.","lead":"This paper simulates the discovery potential of a proposed extension of the Standard Model in which a heavy Z' boson decays into right-handed neutrinos. Using boosted decision trees, it finds that the High-Luminosity LHC could reach significant excesses in three final states.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The >3σ claim for the 2ℓ+4j channels rests on an unvalidated 782.5 fb fake-lepton background; a factor-two error in the fake rate or a 20% systematic uncertainty erases the signal.","rationale":"The paper is a clear, standard fast-simulation study of the well-known B-L extension of the Standard Model. The model implementation via SARAH/SPheno, event generation with MadGraph, and detector simulation with Delphes are standard tools, and the benchmark point is chosen to avoid current dilepton bounds. The strongest claim is the HL-LHC discovery potential, stated as >3σ in each of three channels. The weakest link is the fake-lepton background in the two 2ℓ+4j final states: it is the dominant background, it is estimated only from Delphes with no data-driven validation, and the significance calculation is purely statistical. This is exactly the weakest assumption identified by the reader. I agree with the reader that this does not invalidate the analysis but renders the results an optimistic projection, so the conditional verdict is appropriate. A single concrete check—varying the fake background normalization and adding a systematic uncertainty—would settle whether the 3σ claim survives. No other concern appears more load-bearing: the model equations are consistent, the benchmark satisfies current constraints, and the BDT methodology is standard, though the absence of systematic uncertainties compounds the fake-background issue.","tokens_in":14189,"tokens_out":3907,"duration_ms":37300,"concrete_test":"Recompute the FS1 opposite-sign 2ℓ+4j significance after scaling the 782.5 fb W+jets fake-lepton background by 0.5, 1.0, and 2.0, and also after adding a flat 20% systematic uncertainty via S/sqrt(S+B+(0.2B)^2). If the peak significance drops below 3σ for either the ×2 scaling or the systematic-inclusive calculation, the paper's central conclusion is not robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In FS1 (Table 3) and FS2 (Table 4), the W+jets+fake-lepton background is 782.5 fb, i.e. about 97% of the total background (806 fb for opposite-sign, 938 fb for same-sign), while the signal is 0.18/0.17 fb. At L=3000 fb^-1 this corresponds to roughly 2.4 million background events. To obtain the displayed S/sqrt(S+B) ~ 3–5 after a BDT cut, the fake background must be suppressed by a factor of O(1000). This estimate relies entirely on the Delphes fast simulation of fake leptons; the paper states only that the background was 'identified in our simulation' (Section 4.1.1) with no validation against ATLAS/CMS fake-rate measurements. If the true fake rate is a factor of two larger, or if the fake-lepton kinematic distributions differ from Delphes, the optimized BDT boundary and the resulting significances change substantially. Moreover, all quoted significances use the statistical-only formula S/sqrt(S+B). Even a 1% systematic uncertainty on the dominant fake background corresponds to more than 2×10^4 events, far exceeding the ~500 signal events expected at 3000 fb^-1. The central claim of >3σ in every channel is therefore contingent on the unvalidated fake-lepton background and on the absence of systematic uncertainties, neither of which is demonstrated in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the B-L extension of the Standard Model (BLSM) with a heavy Z' boson and right-handed neutrinos, focusing on LHC phenomenology at 14 TeV. For a benchmark point with MZ' = 3 TeV, MnuR = 420 GeV, gB-L = 0.42, and g~ = -0.55, the authors simulate the processes pp -> Z' -> nuR nuR with subsequent nuR -> l W, using MadGraph5, Pythia, and Delphes. They analyze three final states: 2l + 4j (opposite- and same-sign), 4l + MET, and 3l + 2j + MET, applying Boosted Decision Trees and XGBOOST to suppress SM backgrounds. The central claim is that at the HL-LHC with 3000 fb^-1, a signal significance exceeding 3 sigma can be achieved in each of the considered channels.","tokens_in":14455,"tokens_out":4474,"duration_ms":40812,"significance":"The paper presents a complete and standard simulation chain (SARAH, SPheno, MadGraph, Delphes) for a well-motivated benchmark, and it applies multivariate classification in a clean way, with feature ranking and ROC curves for each channel. If the claimed significances were robust, the study would provide a useful guide for HL-LHC searches for Z' -> nuR nuR in the BLSM. However, the central claim currently rests on a statistical-only significance formula and on an unvalidated fake-lepton background for the 2l+4j channels; those issues must be addressed before the conclusions can be accepted.","major_comments":[{"comment":"The significance is computed as S/sqrt(S+B) with no systematic uncertainties. In FS1, the dominant background is the W+jets fake-lepton background with 782.5 fb (Tables 3 and 4), while the signal is 0.18/0.17 fb. At 3000 fb^-1, a 1% systematic uncertainty on that background corresponds to about 2.3e4 events, far exceeding the expected signal of about 5e2 events. Please include a systematic-uncertainty treatment for each channel, or demonstrate that such uncertainties are negligible; otherwise the claim of 'exceeding 3 sigma' in every channel is not supported.","section":"Section 4.1.1, Figures 4, 8, 12, 16"},{"comment":"The W+jets background with one fake lepton is assigned a cross section of 782.5 fb with no validation against ATLAS/CMS fake-rate measurements; the text only states that such processes were 'identified in our simulation'. Because this background constitutes roughly 97% of the total background in the 2l+4j final states, the reported significance depends critically on the Delphes modeling of fake leptons. Please provide a data-driven validation, or a conservative variation of the fake rate (e.g., a factor of two), and show the effect on the optimized BDT threshold and the resulting significance.","section":"Section 4.1.1 and Tables 3-4"},{"comment":"The BDT and XGBOOST hyperparameters (number of trees, maximum depth, learning rate, minimum child weight, etc.) and the training sample sizes are not stated, and the BDT score threshold used to obtain the final significances is not quoted numerically. Without these details the analysis is not reproducible, and the reported significances cannot be independently checked. Please provide the hyperparameters, the number of events used in training, and the cut values corresponding to the results in Figures 4, 8, 12, and 16.","section":"Section 3 and Section 4"}],"minor_comments":[{"comment":"The model name is written inconsistently as 'BLSM' and 'B-L', and 'XGBOOST' and 'XGBoost' are used interchangeably throughout the text.","section":"General"},{"comment":"The notation g~ is used before it is defined in Eq. (2.6); please define it earlier in the text.","section":"Section 2, after Eq. (2.9)"},{"comment":"The sentence introducing Figure 10 refers to the 'same-sign lepton case', but FS2 is a four-lepton final state; this appears to be a copy-paste error.","section":"Section 4.2 and Figure 10"},{"comment":"Please clarify that the quoted cross sections are after the parton-level and baseline selection cuts, and specify the PDF set and the renormalization/factorization scales used in the generation.","section":"Tables 3-6"},{"comment":"Reference [20] gives only the ATLAS dilepton search; please also cite the corresponding CMS high-mass dilepton resonance search for completeness.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the standard scope of a phenomenology journal and uses a defensible simulation chain. The central '>3 sigma' claim, especially for the 2l+4j channels, is currently contingent on an unvalidated fake-lepton background and on a significance definition that ignores systematics. If the authors can provide a systematic-uncertainty treatment and a robustness check of the fake-rate assumption, the paper would be a solid contribution; otherwise, the conclusions should be tempered. There is no indication of any ethical issue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear Colleague,\n\nThis is a solid but incremental simulation study of the B-L extended SM with a 3 TeV Z' and 420 GeV right-handed neutrino. The model and production/decay chains are known from refs [11,12], but the specific benchmark with nonzero kinetic mixing (g~=-0.55), which suppresses the dilepton channel, and the BDT/XGBOOST analysis of three final states are new numerical results. The MC chain is standard and the cross sections are plausible. The paper does a good job of explaining why these channels could be competitive.\n\nThe soft spot is the fake-lepton background in the 2ℓ+4j final states. At 782.5 fb it's about 97% of the total background in both opposite-sign and same-sign cases, while the signal is only ~0.18 fb. The paper states only that it was 'identified in our simulation,' with no validation against ATLAS or CMS fake-rate measurements. If the true fake rate is a factor of two higher, or the kinematic distributions differ from Delphes, the optimized BDT cut changes and the quoted S/sqrt(S+B) significances for FS1 drop substantially. These channels also have no systematic uncertainties in the significance calculation; even a 1% systematic on the dominant background exceeds the expected signal. So the Conclusions claim of >3 sigma in 'each' channel is solid for the 4ℓ+MET and 3ℓ+2j+MET channels, but contingent for 2ℓ+4j. This is a real weakness, but it does not invalidate the paper.\n\nThe other limitations are minor: single benchmark point, not a scan, and BDT hyperparameters not reported. For a phenomenological projection, these are acceptable.\n\nOverall, the paper is a useful, clearly written projection for a specific BLSM benchmark. I'd send it to peer review. The referee should ask for validation of the fake-lepton background or softened FS1 claims, and an explicit statement that significances are statistical-only.\n\nBest,\n[Your name]","headline":"Solid but incremental BLSM study with a fragile 2ℓ+4j claim due to unvalidated fake-lepton background; still worth peer review.","tokens_in":15069,"tokens_out":3232,"would_cite":true,"duration_ms":27376,"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":"A benchmark BLSM with a 3 TeV $Z'$ and a 420 GeV right-handed neutrino would be discoverable at the High-Luminosity LHC in all three final states studied, with signal significance above $3\\sigma$ in each channel.","keywords":["B-L extension","right-handed neutrinos","Z' boson","LHC phenomenology","HL-LHC discovery potential","boosted decision trees","seesaw mechanism","same-sign dileptons"],"falsifier":"Search the $2\\ell+4j$ final state at the HL-LHC with the same lepton $p_T>200$ GeV and $M_{\\ell\\ell}>250$ GeV cuts; if the observed event yield matches the Standard Model prediction and no peak near 3 TeV appears in $M(2\\ell+4j)$, or no lepton-plus-dijet peak near 420 GeV appears, the benchmark point is falsified.","tokens_in":13950,"feed_emoji":"⚛️","tokens_out":9140,"duration_ms":78015,"temperature":0.7,"pith_summary":"The paper works out how the High-Luminosity LHC could see a heavy $Z'$ boson and right-handed neutrinos in the $B-L$ extension of the Standard Model. It studies the chain $pp\\to Z'\\to\\nu_R\\nu_R$ with each $\\nu_R$ decaying to a charged lepton and a $W$ boson, which produces three final states: two leptons plus four jets, four leptons plus missing energy, and three leptons plus two jets plus missing energy. For a benchmark with a 3 TeV $Z'$ and a 420 GeV right-handed neutrino, machine-learned boosted decision trees separate the signal from Standard Model backgrounds using mass-sensitive kinematic variables. The central claim is that with $3000\\text{ fb}^{-1}$ of 14 TeV proton-proton data, each of the three channels reaches a significance above $3\\sigma$, making the model testable at the HL-LHC.","feed_headline":"Three LHC channels can expose a 3 TeV Z' and its right-handed neutrino","feed_subtitle":"Boosted-tree analysis promises >3 sigma in all three final states with 3000 fb-1 of data.","key_machinery":"The load-bearing object is the decay chain $Z'\\to\\nu_R\\nu_R$, $\\nu_R\\to\\ell^\\pm W^\\mp$, with the $W$s decaying hadronically, leptonically, or semi-leptonically to give the three final states. The $Z'$ is the gauge boson of an extra $U(1)_{B-L}$ symmetry whose breaking by a TeV-scale scalar vev generates a Majorana mass for right-handed neutrinos through the Yukawa term $\\lambda_{\\nu_R}\\,\\chi\\,\\overline{\\nu_R^c}\\,\\nu_R$, implementing the seesaw mechanism. The classifier that carries the argument is a boosted decision tree trained on kinematic variables sensitive to the two masses, especially the invariant mass of the full final state, which reconstructs the $Z'$ peak, and lepton-plus-dijet invariant masses, which reconstruct the $\\nu_R$ peak.","core_discovery":"On the paper's own terms, the discovery is that the $B-L$ extension's characteristic decay chain, $Z'\\to\\nu_R\\nu_R$ followed by $\\nu_R\\to\\ell W$, produces experimentally accessible signatures whose backgrounds can be tamed. For the benchmark point $M_{Z'}=3$ TeV, $M_{\\nu_R}=420$ GeV, $g_{B-L}=0.42$, and $\\tilde{g}=-0.55$, the effective $Z'$-lepton coupling is only $0.13$, which suppresses the usual dilepton-resonance search but leaves the multi-lepton plus jets channels carrying the discovery potential. After training boosted decision trees on observables such as the dilepton invariant mass, the full final-state invariant mass, and lepton-plus-dijet masses, the paper reports that all three channels exceed $3\\sigma$ at $3000\\text{ fb}^{-1}$. The reconstructed mass distributions peak near the input masses, although missing energy shifts the four-lepton and three-lepton peaks below the true $Z'$ and $\\nu_R$ masses.","pith_inferences":["If the simulated fake-lepton $W+$jets background is underestimated in real detectors, the projected significances in the $2\\ell+4j$ channels would shrink, so the headline >3-sigma reach carries a systematic caveat in that channel.","The mass-sensitive classifier variables are largely model-independent, so the same search strategy could be applied to other $U(1)$ extensions with different $Z'$ couplings and different right-handed neutrino masses.","A recast of existing Run 2 searches in the same-sign dilepton and multilepton final states could already constrain this benchmark before HL-LHC data accumulate."],"forward_implications":["Each of the three final states can independently reach above $3\\sigma$ at $3000\\text{ fb}^{-1}$, so a discovery could be corroborated in more than one topology.","The same-sign dilepton plus four-jets channel has a small Standard Model background, mainly vector-boson-scattering $W^\\pm W^\\pm jj$ and fake-lepton $W+$jets, and the BDT suppresses both.","Mass reconstruction works even with missing energy: $M(2\\ell+4j)$ and $M(3\\ell+2j)$ peak near the 3 TeV $Z'$ mass, while lepton-plus-dijet and dilepton masses peak near the right-handed neutrino mass, allowing mass extraction.","Because the benchmark's effective $Z'$-lepton coupling is small, standard dilepton resonance searches are weakened, so these channels are a necessary complement rather than an alternative.","The same classification workflow can be rerun for other mass points and for the leptonically decaying $W$ mode to map out the model's exclusion and discovery reach."],"supporting_citations":[{"why":"Defines the B-L extension of the Standard Model with right-handed neutrinos and the seesaw origin of neutrino masses.","marker":"[6-10]"},{"why":"Earlier proposals for pair-producing heavy right-handed neutrinos through a Z' at hadron colliders; the approach this work extends.","marker":"[11, 12]"},{"why":"Supply the effective Z'-fermion coupling formula that makes the benchmark dilepton rate small and sets the quoted mass bound.","marker":"[17, 18]"},{"why":"Establishes the low-scale B-L breaking scenario with a TeV-scale vev that fixes the right-handed neutrino mass relation.","marker":"[19]"},{"why":"Provide the boosted decision tree algorithm used for signal-background classification.","marker":"[13-16]"},{"why":"Supplies the model implementation, spectrum calculation, event generation, and detector simulation that produce all quoted cross sections.","marker":"[25]-[28]"}],"fun_headline_variants":["B-L Z' decays to right-handed neutrinos: three LHC signatures","Three LHC channels probe B-L Z' and heavy neutrinos","Z' and right-handed neutrino discovery at HL-LHC via BDTs","Z' decays to neutrino pairs: three LHC discovery channels","BLSM Z' and heavy neutrinos: HL-LHC reach via boosted trees"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The projected discovery reach depends on the simulated $W+$jets background with one fake lepton being accurate at 782.5 fb after the tight cuts; if real detectors misidentify jets as leptons more often, the $2\\ell+4j$ significances drop.","fun_headline_variants_meta":{"raw":{"variants":["B-L Z' decays to right-handed neutrinos: three LHC signatures","Three LHC channels probe B-L Z' and heavy neutrinos","Z' and right-handed neutrino discovery at HL-LHC via BDTs","Z' decays to neutrino pairs: three LHC discovery channels","BLSM Z' and heavy neutrinos: HL-LHC reach via boosted trees"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00054,"raw_usage":{"total_tokens":2654,"prompt_tokens":1072,"completion_tokens":1582,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":688,"completion_tokens_details":{"reasoning_tokens":1484}},"tokens_in":688,"tokens_out":1582,"duration_ms":12405,"temperature":1.0,"reasoning_tokens":1484,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T00:45:48.339356+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Search the $2\\ell+4j$ final state at the HL-LHC with the same lepton $p_T>200$ GeV and $M_{\\ell\\ell}>250$ GeV cuts; if the observed event yield matches the Standard Model prediction and no peak near 3 TeV appears in $M(2\\ell+4j)$, or no lepton-plus-dijet peak near 420 GeV appears, the benchmark point is falsified.","supporting_citations":[],"review_version":1}