{"id":"6f6dfeb6-597d-4f46-9a56-689d03fd9dd2","arxiv_id":"2603.13746","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"At the CEPC Z-pole (91.2 GeV, 100 ab^-1), displaced-vertex searches for photophobic ALPs via Z→aγ could probe g_aWW from about 10^-3 to 0.68 TeV^-1 for m_a=1-9 GeV.","lead":"Researchers simulated searches at the proposed CEPC electron-positron collider for a hypothetical particle called the photophobic axion-like particle, which would be produced with a photon and then decay a short distance from the collision point. They find the collider could see such particles in two channels, covering regions of the particle's mass and coupling that other experiments cannot reach.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Projected reach rests on an unvalidated zero-background assumption: the paper explicitly defers detector-level instrumental backgrounds as 'beyond the scope,' yet the 3-event 95% C.L. sensitivity depends on them being negligible.","rationale":"The reader's weakest-assumption identification matches mine: the paper's central sensitivity estimate is valid only if instrumental backgrounds are truly negligible, and the paper explicitly places that quantification outside its scope. I considered the other candidate concern—the quoted upper edge g_aWW≈0.68 TeV−1 appearing inconsistent with 1/g^2 lifetime scaling—but on closer inspection that does not clearly land: at larger g the production cross section grows as g^2 while the geometric acceptance shrinks exponentially, so a high upper edge can arise from the competition between these effects; the reader flagged it as a 'tension' rather than a demonstrated error. The fixed f_a=10 TeV assumption affects lifetimes only logarithmically through Eq. (5), so it is secondary. The instrumental-background assumption, by contrast, is the one step where the paper itself concedes a missing calculation, and it directly controls the 3-event benchmark. Thus the CONDITIONAL verdict already given is appropriate; this stress-test reinforces it without moving it.","tokens_in":15933,"tokens_out":31538,"duration_ms":303473,"concrete_test":"Run a GEANT4-based CEPC detector simulation (e.g., CEPCSW) on a statistically large sample of e+e−→μ+μ−γ events at √s=91.2 GeV, overlaid with expected beam-induced/pile-up backgrounds, and reconstruct displaced vertices with exactly the paper's cuts: |d0|>2 mm, 0.1 m25 GeV. Extrapolate the surviving background yield to L=100 ab−1. If the expected background is ≥1 event, the zero-background 3-event sensitivity is invalid and the sensitivity contours must be recomputed; if the extrapolated yield is <0.1 event, this concern is resolved and the conditional can be lifted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that after optimized cuts the CEPC can exclude photophobic ALPs down to a few signal events, with the SM and instrumental backgrounds reduced to a negligible level. The paper simulates only e+e−→μ+μ−γ and e+e−→τ+τ−γ as SM backgrounds, and these are removed trivially at generator level because their production vertex is the IP: the displacement cuts (|d0|>2 mm, 0.1 m Z -> a gamma as dominant, and simulate the displaced-vertex signals from a -> mu+ mu- and a -> tau+ tau- (with hadronic tau decays), using MadGraph5_aMC@NLO, PYTHIA8, and a simplified fast simulation. Optimized cuts on displacement variables, angular separation, and the reconstructed ALP transverse momentum are applied. With the assumption that SM and instrumental backgrounds are negligible after all cuts, the authors use a 3-event benchmark for 95% C.L. sensitivity and report reachable regions: g_aWW in [1.27e-3, 6.80e-1] TeV^-1 for m_a in [1,4] GeV from the dimuon channel, and g_aWW in [7.00e-4, 9.40e-3] TeV^-1 for m_a in [4,9] GeV from the ditau channel. The paper argues these regions complement existing LEP, LHC, LHCb, and CHARM bounds and the projected HL-LHC sensitivity.","tokens_in":16339,"tokens_out":3695,"duration_ms":37426,"significance":"If the projected reach is correct, this would be a useful and nontrivial result: the CEPC would probe photophobic ALP parameter space not covered by current experiments, and the analysis demonstrates a concrete DV search strategy for a future Z-factory. The study uses state-of-the-art Monte Carlo tools (FeynRules/MadGraph/PYTHIA/MadAnalysis5), presents production cross sections and cut-flow efficiencies, and compares with a broad set of existing constraints. The main weakness is that the central sensitivity claim rests on an unvalidated zero-background assumption: after the displacement cuts, the SM prompt background is eliminated by construction, and the paper explicitly assumes instrumental backgrounds are negligible without a detector-level study. The projected low-edge couplings are therefore conditional on this assumption and cannot be taken as demonstrated without further quantification. This is a correctable issue, but it is load-bearing for the headline numbers.","major_comments":[{"comment":"The paper states that after all optimized cuts 'the SM background is expected to be suppressed to a negligible level,' but no background event counts or cross sections after each cut are reported. Only signal cross sections and efficiencies are given in Tables II and III. This matters quantitatively: with L = 100 ab^-1, a background cross section of 10^-11 pb gives O(1) background events, so the 3-event 95% C.L. benchmark requires that the residual background is below roughly 10^-11 pb. The authors must report the expected number of SM background events (for e+e- -> mu+mu- gamma and e+e- -> tau+tau- gamma) at each cut step, or at least give the final background count, to justify the zero-background assumption.","section":"Sec. III A, Tables II and III"},{"comment":"The manuscript explicitly states that instrumental backgrounds such as pile-up interactions, vertices from dense detector regions, and random-track crossings are 'beyond the scope of this study' and are assumed negligible. This is a load-bearing assumption for the quoted sensitivity: the displacement cuts reject prompt SM background by construction, so the only remaining protection against fake displaced vertices is the assumed negligible instrumental rate. With O(10^12) Z decays and 100 ab^-1, even a fractional fake rate of ~10^-9 yields O(1) events. The authors need a quantitative estimate based on CEPC occupancy, vertexing efficiency, and track-fake probabilities, or a conservative background uncertainty, before the 3-event sensitivity can be considered robust.","section":"Sec. III A, paragraph on instrumental backgrounds"},{"comment":"The sensitivity boundaries in Fig. 9 depend on the ALP lifetime, but the paper never gives the explicit formula for the total ALP width Gamma_a entering Eq. (9). The branching fractions shown in Fig. 1 are insufficient to reproduce the results; the reader needs the sum of the loop-induced partial widths (a -> f fbar, a -> gamma gamma, etc.). In addition, the effective fermion couplings in Eq. (5) depend on the assumed cutoff scale Lambda = 4 pi f_a with f_a fixed to 10 TeV. Since the sensitivity regions are quoted as functions of (m_a, g_aWW) only, the dependence of the lifetime and hence of the projected boundaries on the choice of f_a should be stated explicitly.","section":"Sec. II, Eq. (5) and Eq. (9); Fig. 9"},{"comment":"The SFS framework does not include full detector-level reconstruction of displaced vertices: there is no material interaction model, no vertex-fitter efficiency, and no track-fake reconstruction. The signal efficiencies in Tables II and III are obtained by applying geometric cuts on generator-level quantities (|d0|, v0, vz). Since the signal is concentrated near the tracker boundaries (e.g., 0.1 m < v0 < 1.8 m, vz < 2.35 m), detector resolution, material effects, and reconstruction inefficiencies can significantly modify these efficiencies. The authors should validate the key efficiencies with a more realistic detector simulation, or at least provide an efficiency correction/uncertainty estimate.","section":"Sec. III, simplified fast simulation (SFS)"}],"minor_comments":[{"comment":"'owning to' should be 'owing to' in the sentence about the clean experimental environment.","section":"Sec. I"},{"comment":"The second relation in Eq. (4) appears typeset incorrectly ('c2W/c2W'); if intended, it should be gaZZ = (c_W^2 / s_W^2) g_aWW. Please correct the rendering.","section":"Eq. (4)"},{"comment":"The phrase 'with the number cuts being taken as the first step filter' is unclear; rephrase.","section":"Sec. III A"},{"comment":"The arXiv rendering of Figs. 1 and 9 is badly garbled (e.g., the legend text in Fig. 9 is scrambled). The published-quality figures need to be readable, with all curves and regions clearly labeled.","section":"Figs. 1 and 9"},{"comment":"The statement that for m_a above the b bbar threshold 'no accessible parameter space' remains for long-lived ALPs is asserted without a quantitative lifetime/cut-efficiency demonstration. A short quantitative explanation would strengthen the argument.","section":"Sec. III B and IV"}],"recommendation":"major_revision","confidential_remarks":"The comparison with the HL-LHC projection [72] is not an independent check, since it uses the same authors' model implementation and event-generation framework; the 'complementary coverage' claim should be framed as a model-dependent projection comparison. The core issue, however, is the unquantified zero-background assumption: without a detector-level background estimate, the low-g_aWW boundaries in Fig. 9 are not yet established. This is fixable but requires substantial additional work, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The two-minute version: this is a clean, standard LLP-projection paper, and for once the authors tell you where the body is buried. What is new is the specific application — photophobic ALPs at the CEPC Z pole, produced via Z→aγ, decaying in the inner tracker to μμ or ττ. That displaced-vertex treatment for the CEPC does not appear in the cited literature; the closest prior paper [55] did prompt leptons and mono-photon. The MC chain is ordinary but competently executed, the cross-section comparison among single-production channels is informative, and the comparison plot against LEP/LHC/LHCb/CHARM/HL-LHC is useful. Credit where due: the authors explicitly state, in Sec. III A, that instrumental backgrounds (pile-up, fake tracks, material interactions) are assumed negligible and are not simulated. That is the load-bearing assumption, because the quoted 95% C.L. regions are built on a 3-event benchmark. If a handful of background events survive, the lower edges of the claimed regions move up. For a Z factory with ~10^12 Z's, a fractional fake-DV rate of 10^-9 already gives O(1) events, so the assumption deserves more than a one-sentence disclaimer.\n\nThere is a second, quieter problem that the reader's report did not emphasize enough: the upper edge of the μ+μ−γ reach, g_aWW < 6.8×10^-1 TeV^-1, is not credible given the lifetime scaling for the masses in question. At that coupling the ALP is so short-lived that the v0>0.1 m cut should kill the signal; the number looks like a typo for 6.8×10^-2. The tau-channel upper edge is more reasonable. That should be checked before quoting the paper.\n\nMinor: the paper never writes the total-width formula, though it uses it to draw the sensitivity boundaries; a reader cannot reproduce the curves without hunting down Refs. [57,58]. That is easily fixed.\n\nOverall: this is a serious, honest projection study, not a revolutionary one. The physics motivation is standard, the analysis is reproducible in outline, and the weaknesses are addressable in revision rather than fatal. It deserves a referee. I would not cite the numbers as they stand, but I would follow the paper and would bring it to a reading group as an example of how easy it is for a clean projection to overstate sensitivity by assuming away the hardest part of a displaced-vertex search.","headline":"A useful but unvalidated CEPC sensitivity projection for photophobic ALPs via displaced vertices; the zero-background assumption is explicit and the dimuon upper edge looks off by an order of magnitude.","tokens_in":16818,"tokens_out":5552,"would_cite":false,"duration_ms":53068,"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":"At a future Z-pole electron-positron collider, long-lived 'photophobic' axion-like particles could be discovered through displaced vertices produced in Z→aγ decays, covering unexplored regions of mass and coupling.","keywords":["axion-like particles","photophobic ALP","long-lived particles","displaced vertices","Z pole","CEPC","lepton collider","electroweak couplings"],"falsifier":"Run a detector-level simulation of the Z-pole environment, including pile-up and track reconstruction, and count background events passing the full cut set (|d0| > 2 mm, 0.1 m < v0 < 1.8 m, vz < 2.35 m, ΔR < 1.0, and dilepton pT > 25 GeV for muons or > 30 GeV for taus); if any background event survives, the projected 95% CL regions are optimistic.","tokens_in":15805,"feed_emoji":"🔭","tokens_out":10589,"duration_ms":95142,"temperature":0.7,"pith_summary":"The paper argues that a Z-pole electron-positron collider running at 91.2 GeV with very high integrated luminosity can discover light, long-lived axion-like particles that have no direct couplings to photons or fermions. Such 'photophobic' ALPs, if lighter than about 10 GeV, decay through loop-induced couplings and thus travel macroscopic distances before decaying. By focusing on the dominant production channel Z→aγ followed by a→μ+μ− or a→τ+τ−, the displaced vertex inside the inner tracker becomes a nearly background-free signature. With 100 ab−1 of data, the proposed collider could probe ALP–W couplings between roughly 10−4 and 10−1 TeV−1 for ALP masses of 1–9 GeV, a parameter region that existing collider searches do not cover.","feed_headline":"CEPC displaced-vertex search could expose photophobic axions","feed_subtitle":"If right, the Z-pole experiment would cover a light, weakly coupled axion window that hadron colliders cannot reach.","key_machinery":"The central object is the photophobic ALP: an axion-like particle whose tree-level couplings to photons and fermions vanish, leaving only couplings to electroweak gauge bosons (W, Z, Zγ). After electroweak symmetry breaking, the scenario reduces to two parameters: the ALP mass m_a and the coupling g_aWW. The effective fermion and photon couplings are generated by one-loop renormalization-group running, which suppresses the total decay width and lengthens the proper decay length. The search exploits the observables |d_0| (transverse impact parameter), v_0 and v_z (vertex displacement), ΔR between the two leptons, and the transverse momentum of the dilepton system to separate the boosted, long","core_discovery":"At the heart of the paper is the observation that photophobic ALPs — axion-like particles with no tree-level coupling to photons or fermions — can be produced at the Z pole via Z→aγ, then decay into charged leptons through loop-induced couplings. Because those couplings are small, the ALP travels a macroscopic distance before decaying, creating a displaced vertex that is essentially background-free after moderate cuts. The authors simulate the μ+μ−γ and τh+τh− E_T^miss γ signals and derive 95% CL sensitivities: g_aWW ∈ [1.27×10−3, 0.68] TeV−1 for m_a ∈ [1,4] GeV and g_aWW ∈ [7×10−4, 9.4×10−3] TeV−1 for m_a ∈ [4,9] GeV. These regions are presented as new ground, uncovered by existing collider","pith_inferences":["A full detector-level simulation including pile-up and random-track crossings is the natural next step; any surviving background would shrink the quoted regions, especially at the weak-coupling (long-lifetime) boundary where the signal is diluted across larger volumes.","Because the effective fermion couplings are computed at a fixed cutoff scale f_a = 10 TeV, the sensitivity bands are tied to that choice; scanning over f_a would show how the boundaries move as the lifetime changes.","Extending the vertex search from the inner tracker to the calorimeters or muon system would add sensitivity to shorter and longer lifetimes, respectively; the paper explicitly leaves those options open.","The results suggest that a dedicated displaced-vertex search at the Z pole could be mounted with relatively modest detector requirements, which may influence design choices for future electron-positron colliders."],"forward_implications":["A Z-pole electron-positron collider with 100 ab−1 can probe photophobic ALP couplings down to about 10−4 TeV−1 for masses 4–9 GeV and about 10−3 TeV−1 for masses 1–4 GeV.","The dimuon and ditau channels together give continuous mass coverage from 1 to 9 GeV, with no branching-ratio gaps below the b-quark threshold.","The sensitivity is achieved using only inner-tracker information, so the search does not rely on calorimeter or muon-system capabilities.","The projected regions extend into parameter space that is complementary to existing hadron-collider and rare-meson-decay searches, so a positive signal would be a new discovery rather than a confirmation.","The same production and displacement logic applies to any proposed Z-pole lepton collider, so the method provides a blueprint for comparable facilities."],"fun_headline_variants":["CEPC Z-pole could unmask photophobic axions","Displaced vertex signals probe photophobic ALPs at CEPC","CEPC extends reach for light photophobic axions","Z-pole decays of photophobic axions visible at CEPC","CEPC to hunt photophobic ALPs via displaced vertices"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The entire sensitivity projection rests on the assumption that after the chosen cuts, pile-up, random-track crossings, and other instrumental backgrounds are negligible; if even one such event survives in 100 ab−1, the 3-event benchmark for the 95% CL limit no longer holds.","fun_headline_variants_meta":{"raw":{"variants":["CEPC Z-pole could unmask photophobic axions","Displaced vertex signals probe photophobic ALPs at CEPC","CEPC extends reach for light photophobic axions","Z-pole decays of photophobic axions visible at CEPC","CEPC to hunt photophobic ALPs via displaced vertices"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000736,"raw_usage":{"total_tokens":3241,"prompt_tokens":971,"completion_tokens":2270,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":715,"completion_tokens_details":{"reasoning_tokens":2179}},"tokens_in":715,"tokens_out":2270,"duration_ms":14916,"temperature":1.0,"reasoning_tokens":2179,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T18:13:18.083010+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a detector-level simulation of the Z-pole environment, including pile-up and track reconstruction, and count background events passing the full cut set (|d0| > 2 mm, 0.1 m < v0 < 1.8 m, vz < 2.35 m, ΔR < 1.0, and dilepton pT > 25 GeV for muons or > 30 GeV for taus); if any background event survives, the projected 95% CL regions are optimistic.","supporting_citations":[],"review_version":1}