{"id":"16c74778-a515-4d5b-9ffe-0d1a71f7cdcc","arxiv_id":"2505.02429","paper_version":2,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":4,"one_line_summary":"No evidence for emerging jets was found in 51.8 fb-1 of 13.6 TeV ATLAS data; new exclusion limits are set on s-channel Z' and t-channel scalar mediator models producing pairs of emerging jets.","lead":"ATLAS searched for 'emerging jets', jets filled with particles that decay a short distance from the collision point, in 51.8 inverse femtobarns of 2022-2023 LHC data. No excess over background was seen, and the results rule out new mediator particles in previously untested mass ranges.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The load-bearing soft spot is the benchmark-dependence of the quoted exclusions: the signal grid fixes N_c=3, N_f=7 and the m_rhoD=2Lambda_D=2m_qD=4m_piD hierarchy, and Section 8 admits this was not varied.","rationale":"Reader and I identify the same assumption as the least secure. The analysis is well executed and internally consistent: the ABCD closures are checked with A' regions and an ML-inverted VR, the mistag-rate method includes a 61%/23% parameterisation systematic, and the observed yields in all SRs are below or consistent with prediction. I considered the fixed 10 GeV Z' width, the trigger scale factor, and the large ML background systematic; none of these undermines the central null claim. The remaining soft spot is model dependence, and it is load-bearing only for the quoted mass ranges, not for the no-excess observation. Since the paper already flags this limitation and ATLAS searches conventionally present simplified-model limits, I do not change the ACCEPT verdict; no correction or formal revision is required, though a supplemental scan would strengthen the claim.","tokens_in":63265,"tokens_out":7760,"duration_ms":104439,"concrete_test":"Re-run one high-sensitivity signal point with altered dark-sector parameters: Pythia8 Hidden Valley samples for m_Z'=1500 GeV, m_piD=10 GeV, c tau_piD=50 mm, with (N_c,N_f)=(2,4) and with an alternative hierarchy such as m_rhoD=3 Lambda_D=3 m_qD=6 m_piD, keeping all analysis selections and the ML tagger unchanged. Recompute the ML-based SR efficiency and the expected 95% CL limit on sigma x BR. If the limit moves by more than about 20%, the quoted Z' mass range is materially benchmark-dependent; if it moves by less than about 10%, the Section 8 robustness argument is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central physics claim is a null result plus two model-specific exclusions (Z' up to 2.55 TeV, Phi up to 1.375 TeV). The null result is robust: both cut-based and ML strategies see yields consistent with data-driven predictions, and the nonclosure uncertainties are quantified. The exclusions, however, rest on the simulated signal grid of Section 3.1: every sample uses N_c=3, N_f=7, the mass hierarchy m_rhoD=2 Lambda_D = 2 m_qD = 4 m_piD, and dark pions forced to decay to down quarks. These parameters set the dark-pion multiplicity and pT spectrum, which directly feed the observables used for selection: N_vtx, PTF, ECF2, N_subjet, and the transformer tagger's track-level inputs. Section 8 explicitly states that the impact of varying N_c, N_f, and the hierarchy has not been studied, and the argument that only the dark-pion multiplicity changes is qualitative. A different dark-sector point could change signal acceptance by more than the quoted limits' precision, so the headline mass ranges should be read as benchmark-specific rather than universal dark-QCD constraints.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a search for emerging jets—jets containing multiple displaced vertices from decays of long-lived dark mesons—using 51.8 fb⁻¹ of pp collisions at √s = 13.6 TeV recorded by ATLAS in 2022–2023. Two production modes are targeted: s-channel production of a Z′ boson decaying to a dark-quark pair, and t-channel exchange of a scalar mediator Φ producing dark quarks plus up to two SM quarks. Two complementary strategies are used: a cut-based selection on jet observables (PTF, ECF2, N_vtx, N_subjet) with a data-driven ABCD background estimate, and an ML-based strategy using a GN2-style transformer tagger with a data-driven mistag-rate background estimate. Both strategies are split into low- and high-m_jj regions with distinct triggers, including a dedicated emerging-jet trigger introduced for Run 3. No significant excess is observed: the yields in the four signal regions (Table 8) agree with the data-driven predictions (e.g., cut-based high-m_jj: 8 observed vs 7.5 ± 1.1 ± 1.1 predicted; ML-based low-m_jj: 24 vs 31.8 ± 0.8 ± 7.5). The background methods are validated in control regions and signal-adjacent validation regions (Tables 5–7). Using profile-likelihood fits and the CLs procedure, 95% CL exclusions are set for the specified benchmark dark sector: Z′ masses 600–2550 GeV for g_q = 0.01, g_qD = 0.1 with cτ between 5 and 50 mm, and Φ masses 600–1375 GeV for κ = 0.1.","tokens_in":63557,"tokens_out":19256,"duration_ms":224822,"significance":"The core null result is robust and model-independent: the observed yields agree with fully data-driven background predictions in all four signal regions, and the background uncertainties are quantified through ABCD closure tests, ML-inverted and VR-tag validation regions, and alternative mistag-rate parameterizations. The search is methodologically significant: it is the first application of a GN2-type transformer tagger to a beyond-the-Standard-Model signature in ATLAS, the first ATLAS ML tagger to exploit the large-impact-parameter tracking pass, and the first ML-based identification of displaced vertices in an ATLAS LLP search; the tagger demonstrably generalizes to the t-channel signal, which is absent from its training. The statistical model (profile likelihood, CLs, pyhf) is standard, and the asymptotic approximation is checked against pseudo-experiments. If the results stand, they provide the first direct constraints on emerging-jet pair production via an s-channel Z′ mediator and the first search for t-channel scalar-mediated emerging jets.","major_comments":[{"comment":"The headline exclusions (Z′ up to 2550 GeV, Φ up to 1375 GeV) are computed for a single dark-sector benchmark with N_c = 3, N_f = 7, the mass hierarchy m_rhoD = 2Λ_D = 2m_qD = 4m_piD, dark pions forced to decay to down quarks, and no dark baryons, yet the Abstract and Conclusions quote these ranges as conditions only on the mediator couplings and on cτ. Section 8 states that the impact of varying N_c, N_f, and the mass hierarchy “has not been explicitly studied” and that these parameters are “primarily expected to affect the dark pion multiplicity” (Ref. [26]), which is a qualitative argument. Since the dark-pion multiplicity and pT spectrum directly determine N_vtx, displaced-track counts, PTF, ECF2, N_subjet, and the transformer tagger's track-level features, alternative dark-sector parameter points could shift the signal acceptance and hence the quoted mass ranges by an amount the paper does not quantify. I do not regard this as an internal inconsistency, because the benchmark is fully specified in Table 1 and Section 3.1, but the manuscript should carry the benchmark condition explicitly whenever the 2550 GeV/1375 GeV numbers are quoted in the Abstract and Conclusions, and should reference the m_piD scan of Figure 12 as partial robustness evidence or soften the final claim of Section 8 that the results “are therefore expected to remain sensitive across a broader range of dark QCD parameter choices.”","section":"§3.1, §8, Table 1, Abstract/Conclusions"},{"comment":"In the ML-based strategy, the n_tag < 2 control region used to measure the mistag rate is not included as Poisson terms in the profile likelihood of Section 8; the robustness of the background extraction therefore rests on the <10% signal-contamination check and on the 23–61% parameterization systematics rather than on a simultaneous CR fit. Given the small SR yields this is adequate, but explicitly stating why the CR was not incorporated into the fit would strengthen the statistical presentation.","section":"§6.2 and §8"}],"minor_comments":[{"comment":"A passage of heavily corrupted text beginning “figure QZ a diagram illustrati¯g…” appears in Section 1 near Figures 1–2 and is repeated, including duplicated paragraphs on the trigger regions and paper structure; this makes part of the introduction unreadable and must be corrected in the published version.","section":"§1"},{"comment":"In the first paragraph of the Conclusions, “Φ masses up to 1350 GeV are excluding” should read “are excluded.”","section":"§9"},{"comment":"The performance statement for the ECF2/pT > 40 GeV requirement (approximately 40% background rejection at over 90% signal efficiency) is not tied to a displayed distribution; a small table or figure of the signal and background efficiencies as a function of the ECF2/pT threshold would aid the reader.","section":"§5.1.1"}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is well within the scope of RPP and, in my judgment, is a publishable ATLAS search: the background estimation is carefully validated, the null result is robust, and the transformer-tagger novelty is genuine. The point I would keep visible is the benchmark-specificity of the quoted exclusion ranges; it is disclosed in Section 8 but should also appear at the abstract/conclusion level to avoid the 2550 GeV and 1375 GeV numbers being cited as generic dark-QCD bounds. The garbled passage in Section 1 should be checked as a possible production defect. These are local fixes; no reanalysis is required."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things about this one. First, it is a well-executed null search: observed yields match data-driven predictions across all four signal regions, with nonclosure uncertainties quantified and validated in control regions. Second, the headline exclusions are explicitly tied to a specific dark-sector benchmark (N_c=3, N_f=7, the m_rhoD=2Lambda_D=2m_qD=4m_piD hierarchy, dark pions forced to down quarks). The stress-test note gets this right: Section 8 admits the impact of varying those parameters was not studied, and the argument that only dark-pion multiplicity changes is qualitative. So read the 2550 GeV Z' and 1375 GeV Phi limits as constraints on that benchmark, not as universal dark-QCD bounds. That does not sink the paper, but it should be stated plainly in any summary.\n\nWhat is genuinely new: this is the first search for s-channel Z' mediated emerging jets and the first for t-channel scalar mediated emerging jets. The prior CMS searches targeted bi-fundamental mediator pair production, so the coverage is complementary and real. The transformer-based tagger, adapted from GN2 and using large-impact-parameter tracks, is also a legitimate technical contribution; it can be reused in other displaced-signature searches. The analysis is careful about triggers (the dedicated emerging-jet trigger is a nice Run 3 addition), and the background estimation is data-driven with multiple closure tests. The ML-based background has a large systematic uncertainty in the high-mjj region (61%), but they validate with a signal-free region and it does not undermine the null result.\n\nThe soft spots beyond the benchmark dependence are minor. The paper does not provide supplementary data or code for reinterpretation, which limits its utility for model-builders, but that is standard for ATLAS. The full text I was sent has some rendering artifacts in the figures and a duplicated section, but those are clearly presentation issues, not physics. The authors cite the relevant prior searches and do not oversell their novelty; they explicitly frame the limits as benchmark-specific, which is honest.\n\nWho is this for? Anyone working on long-lived particles, dark sectors, or jet tagging will find it useful. It deserves a serious referee: the analysis is internally consistent, the new channels are meaningful, and the tagger is a reusable tool. I would support acceptance, and if I were the editor I would send it to review without hesitation.","headline":"A clean ATLAS null search that closes two unprobed emerging-jet channels and debuts a reusable transformer tagger, with the caveat that the quoted exclusions are benchmark-specific rather than universal dark-QCD bounds.","tokens_in":64106,"tokens_out":1390,"would_cite":true,"duration_ms":18444,"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":"No emerging jets found: Z' dark mediator excluded to 2.6 TeV","keywords":["emerging jets","dark QCD","long-lived particles","displaced vertices","Z' mediator","t-channel scalar mediator","transformer jet tagger","LHC search"],"falsifier":"Recompute the 95% confidence-level exclusion limits after re-running the same selections on signal samples with a different dark-sector mass hierarchy or a different number of dark quark flavours; if the boundary moves by more than the quoted uncertainties, the reported numbers depend on the unstudied parameter choice.","tokens_in":63038,"feed_emoji":"🔭","tokens_out":4407,"duration_ms":55368,"temperature":0.7,"pith_summary":"This search aims to establish whether dark-sector “emerging jets”—jets crowded with displaced vertices from long-lived dark pions decaying back to Standard Model particles—are produced in proton collisions through two channels not previously targeted: an s-channel Z' vector mediator and a t-channel scalar mediator. Using 51.8 inverse femtobarns of 13.6 TeV proton–proton collision data, it finds no significant excess over the Standard Model background and reports the first direct exclusions for both production modes. If correct, Z' mediators between 600 and 2550 GeV are ruled out for quark and dark-quark couplings of 0.01 and 0.1 (for dark pion decay lengths of 5–50 mm), and scalar mediators between 600 and 1375 GeV are ruled out for a quark–dark-quark coupling of 0.1.","feed_headline":"No emerging jets found: Z' dark mediator excluded to 2.6 TeV","feed_subtitle":"First search for this dark-QCD production channel sets limits far below existing dijet searches.","key_machinery":"The central object is the emerging-jet topology: a jet containing several displaced vertices produced by dark pions that travel macroscopic distances before decaying into Standard Model quarks. Two selection strategies carry the analysis: a cut-based chain using the prompt-track fraction, the number of displaced vertices, the energy-correlation function ECF2, and the number of subjets, with background estimated by the ABCD method; and a transformer-based jet tagger trained on 12 million jets that classifies each jet with a score p_EJ, with background estimated from a measured per-jet mistag rate. A dedicated trigger selecting jets with a very low prompt-track fraction extends sensitivity to lower dijet masses, while a standard single-jet trigger covers the high-mass region.","core_discovery":"The paper claims that, after a fully data-driven background estimate, the observed event counts agree with the Standard Model prediction in every signal region, so no evidence for emerging jets is found. The strongest result is a set of 95% confidence-level exclusion limits on mediator production: for a dark pion proper decay length between 5 and 50 mm, Z' masses from 600 to 2550 GeV are excluded at couplings g_q = 0.01 and g_qD = 0.1, and for a quark–dark-quark coupling of 0.1, Phi masses from 600 to 1375 GeV are excluded. The paper also claims that at m_Z' = 1500 GeV with a 50 mm dark pion decay length, quark couplings above 0.003 are excluded when the dark coupling exceeds 0.03, a sensitivity more than twenty times stronger than existing dijet resonance searches. These constitute the first direct constraints on emerging-jet pair production through an s-channel Z' mediator and the first search for t-channel scalar-mediated emerging jets.","pith_inferences":["The transformer tagger's auxiliary tasks—classifying track origin and grouping tracks into vertices—could be transferred directly to other long-lived-particle searches at the same experiment, potentially sharpening their sensitivity without retraining from scratch.","The analysis is limited by the 61% systematic uncertainty attached to the mistag-rate parameterisation in the high-mass region; a finer parameterisation in terms of more jet observables could bring a notable gain in the reported limits.","With only a fraction of the full Run 3 dataset used here, simply doubling the integrated luminosity would push the Z' exclusion boundary beyond 2550 GeV, assuming the background estimates scale as expected.","The quoted exclusion numbers rest on one specific dark-sector benchmark (N_f = 7, N_c = 3, and the stated mass hierarchy); a reanalysis using alternative benchmark points would be a natural test of how much of the reported region actually survives."],"forward_implications":["The excluded Z' mass range, up to 2550 GeV, is the first direct bound on this specific emerging-jet production channel, where previously only pair-produced bi-fundamental mediators had been constrained.","The ML-based strategy is more powerful than the cut-based one, excluding Z' masses up to 2550 GeV versus 2150 GeV and giving an order-of-magnitude stronger limits on the t-channel scalar mediator.","The search probes quark–dark-quark couplings down to about 0.003, more than twenty times smaller than what dijet resonance searches reach, opening a previously inaccessible part of dark-sector parameter space.","The limits stay nearly flat for dark pion decay lengths between 1 and 100 mm and only weaken above 100 mm, when a growing fraction of dark pions decay beyond the inner tracker.","The results are expected to remain sensitive to a broader set of dark QCD parameters, since variations in the dark colour and flavour numbers mostly change the dark pion multiplicity rather than the core signature."],"supporting_citations":[{"why":"Defines the emerging-jet signature and supplies the dark-sector parameter choices adopted by the signal simulation.","marker":"[26]"},{"why":"Motivates the choice of N_f = 7 and N_c = 3 by matching the observed dark matter relic abundance.","marker":"[22]"},{"why":"Provides the darkqcd_fv_down model used to compute t-channel scalar-mediated signal matrix elements.","marker":"[37]"},{"why":"Is the previous dedicated emerging-jet search whose assumed production mode this paper extends beyond.","marker":"[33]"},{"why":"Provides the dijet resonance search limits against which the improved coupling sensitivity of this search is compared.","marker":"[105]"}],"fun_headline_variants":["First dark-QCD search: no emerging jets, Z' excluded to 2.55 TeV","No dark-meson jets: first limits on Z' and Phi mediators","ATLAS: no emerging jets, dark sector excluded up to 2.55 TeV","No signal in first ATLAS emerging-jet search; limits set on Z' and Phi","ATLAS excludes Z' to 2.55 TeV and Phi to 1.375 TeV in dark-sector search"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The quoted mass limits assume the simulated dark sector is configured with seven dark quark flavours, three dark colours, the mass hierarchy m_rhoD = 2 Lambda_D = 2 m_qD = 4 m_piD, and dark pions forced to decay to down quarks, and the paper states that the impact of varying these parameters was not explicitly studied.","fun_headline_variants_meta":{"raw":{"variants":["First dark-QCD search: no emerging jets, Z' excluded to 2.55 TeV","No dark-meson jets: first limits on Z' and Phi mediators","ATLAS: no emerging jets, dark sector excluded up to 2.55 TeV","No signal in first ATLAS emerging-jet search; limits set on Z' and Phi","ATLAS excludes Z' to 2.55 TeV and Phi to 1.375 TeV in dark-sector search"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00095,"raw_usage":{"total_tokens":4106,"prompt_tokens":1048,"completion_tokens":3058,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":664,"completion_tokens_details":{"reasoning_tokens":2937}},"tokens_in":664,"tokens_out":3058,"duration_ms":21102,"temperature":1.0,"reasoning_tokens":2937,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:51:55.580646+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the 95% confidence-level exclusion limits after re-running the same selections on signal samples with a different dark-sector mass hierarchy or a different number of dark quark flavours; if the boundary moves by more than the quoted uncertainties, the reported numbers depend on the unstudied parameter choice.","supporting_citations":[],"review_version":1}