{"id":"6e881297-ba1d-4410-af25-fff3c4976f7c","arxiv_id":"2505.00478","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A 1 TeV ILC with polarized beams could discover GeV-scale ALP-portal dark matter through mono-photon plus missing energy at 7.6 sigma and measure the ALP-photon coupling to about 1%.","lead":"This paper asks whether a future electron-positron collider can see dark matter through an axion-like particle that acts as a messenger, using a photon plus missing energy signal. It reports that a polarized 1 TeV ILC could discover the benchmark model at high significance and measure the messenger coupling to about one percent.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The analytic cross-section in Eq. (B.32) is roughly a factor 3–4 below the signal yield in Table 1, so the quoted 7.6σ significance and the χ2-based gaγγ precision rest on an unreported normalization.","rationale":"The reader's weakest assumption concerned Eq. (15) and the invisible branching ratio. A quick estimate shows that this concern, while a genuine typo, is not numerically fatal: the correct invisible width is Γ(a→ΨΨbar) = cΨ^2 mamΨ^2/(8πfa^2) (1−4mΨ^2/ma^2)^{3/2}. At the benchmark with cΨ = 1, this is ~9.7×10−8 GeV, still ~10^3 times the diphoton width ~9.7×10−11 GeV, so BR(a→invisible) ≈ 0.999. The missing cΨ^2 and phase-space power therefore do not dilute the quoted significance enough to threaten the central claim. The more serious issue is the factor 3–4 discrepancy between the analytic cross-section Eq. (B.32) and the event count in Table 1. Since the discovery significance scales roughly as S/√B for this background-dominated search, reducing S by a factor 3–4 changes a claimed 7.6σ discovery into a sub-5σ fluctuation-level excess. It is possible that Eq. (B.32) simply has a normalization typo (the known ALP-strahlung cross-section is typically a factor 4 larger, σ ~ 4.6×10−4 pb), and that Table 1 was produced with the correct matrix element. But the paper does not state this, and Section 5 says Eq. (B.32) is used in the χ2 analysis. The manuscript therefore contains an unresolved internal inconsistency in a quantity that is directly load-bearing for the headline claims. This justifies keeping the reader's CONDITIONAL verdict: the central projections can be accepted only after the authors reconcile the printed cross-section formula with the simulated event rate and, if needed, correct the χ2 extraction.","tokens_in":14989,"tokens_out":28461,"duration_ms":326000,"concrete_test":"Recompute σ(e+e−→γa) at √s = 1 TeV for ma = 5 GeV and fa = 186 GeV with the same FeynRules/Madgraph model used for Table 1, and compare the integrated cross-section after the basic cuts with the numerical integral of Eq. (B.32). If the simulated cross-section exceeds Eq. (B.32) by a factor ~4, the table is internally consistent with a correct matrix element but the printed formula must be corrected and the χ2 extraction in Section 5 rerun with the corrected normalization; if the ratio is ~1, the signal yields and the 7.6σ claim must be scaled down accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Integrating Eq. (B.32) with gaγγ = 1.33×10−5 GeV−1 at √s = 1 TeV gives σ(e+e−→γa) ≈ 1.16×10−4 pb, or about 580 events at L = 5 ab−1 before acceptance. After the basic cuts (pTγ > 10 GeV, |ηγ| ≤ 2.5), which retain nearly all events, this predicts roughly 500–580 signal events, whereas Table 1 reports S = 1750 events, i.e. σ ≈ 3.5×10−4 pb. The factor is about 3–4. If the Madgraph/Delphes number is correct, Eq. (B.32) is wrong by this factor, and the Section 5 χ2 analysis, which explicitly refers to Eq. (B.32), is not reproducible; the claimed ~1% accuracy on gaγγ is then unsupported. If Eq. (B.32) is instead correct, the signal has been overestimated by a factor 3–4, and the central significance falls from Z = 7.64 to Z ≈ 2–3 (with unchanged background), below the 5σ discovery threshold. This is more load-bearing than the reader's Eq. (15) concern: with the proper phase-space factor (1−4mΨ2/ma2)^{3/2} and cΨ = 1, Γ(a→ΨΨbar)/Γ(a→γγ) ≈ 10^3 at the benchmark, so BR(a→invisible) ≈ 0.999 and the significance is not materially diluted.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a model in which a GeV-scale ALP mediates between the Standard Model and a Dirac fermion dark-matter candidate, with interactions limited to ALP couplings to electroweak gauge bosons and a derivative ALP-DM coupling. The authors find that the observed relic abundance can be obtained near the resonant condition m_a ≈ 2 m_Ψ, choosing a benchmark m_a = 5 GeV, m_Ψ = 2.47 GeV, f_a = 186 GeV. They then analyse e+e− → γ a, a → ΨΨbar at a 1 TeV ILC with 5 ab^-1, using mono-photon plus missing-energy events. After basic cuts, E_miss < 510 GeV, |η_γ| > 1, and beam polarization {P_e+ : P_e−} = {−20% : +80%}, they report a discovery significance Z = 7.64 (Table 1) and, through a binned χ² analysis of the differential cross-section, claim a ~1% determination of g_aγγ (Section 5).","tokens_in":15280,"tokens_out":10023,"duration_ms":106044,"significance":"If the reported signal size is correct, the paper would demonstrate a nice complementarity between the DM relic-density constraint and a clean lepton-collider mono-photon search, with a distinctive E_miss double-peak discrimination between the ALP signal and the SM neutrino background. The use of explicit MadGraph/Pythia/Delphes event generation, the explicit Lagrangian, and the cross-check against the relic contour of ref. [22] are strengths. However, the central collider claim depends on a normalization that is internally inconsistent: the analytic cross-section in Eq. (B.32) is a factor of roughly three below the event yield reported in Table 1. This discrepancy directly affects both the 7.6σ significance and the claimed 1% coupling precision, so the quantitative conclusions cannot be accepted until it is resolved.","major_comments":[{"comment":"The analytic and simulated signal normalizations are inconsistent. Integrating Eq. (B.32) with g_aγγ = 1.33×10^-5 GeV^-1, √s = 1 TeV, and m_a = 5 GeV gives σ(e+e−→γa) ≈ (π α/12) g_aγγ² ≈ 1.3×10^-4 pb, i.e. about 650 events at L = 5 ab^-1 before acceptance. Table 1 reports S = 1750 events after the same basic cuts for the unpolarized case, and 2050 for the polarized case. The basic cuts p_T^γ > 10 GeV and |η_γ| ≤ 2.5 retain essentially all events for this benchmark, so acceptance cannot explain the factor of about three. If Eq. (B.32) is the correct cross-section, the post-cut signal is roughly 500–650 events and the polarized significance falls from 7.64 to approximately 2–3, below the 5σ discovery threshold; if the MadGraph/Delphes number is correct, then Eq. (B.32) is missing a factor of about three and the Section 5 χ² analysis, which explicitly uses Eq. (32) as the differential cross-section, inherits the error. The authors should show a direct numerical comparison of the analytic formula with the generator-level cross-section and confirm which normalization is correct.","section":"Appendix B, Eq. (B.32) and Table 1"},{"comment":"The invisible decay width in Eq. (15) is missing both the c_Ψ² factor and the threshold phase-space factor. For the derivative coupling used in the paper, one expects Γ(a→ΨΨbar) = c_Ψ² m_a m_Ψ²/(8π f_a²) (1 − 4m_Ψ²/m_a²)^{3/2}. At the benchmark m_a = 5 GeV, m_Ψ = 2.47 GeV, the threshold factor is O(3×10^-3), not O(1). Because Γ(a→γγ) is much smaller than Γ(a→ΨΨbar) even after this suppression, the invisible branching ratio remains ~1 for c_Ψ = O(1), so the reader's concern that the signal would be diluted by a small invisible branching ratio is not realized. Nevertheless, Eq. (15) as printed is incorrect and affects the subtraction term in Eq. (16) and the quantitative width used in the Boltzmann equation, so it must be corrected.","section":"Section 3, Eq. (15)"}],"minor_comments":[{"comment":"The abstract states that the effect of early kinetic decoupling is discussed in the resonant regime, but I could not find any discussion of kinetic decoupling in the main text or appendices; the authors should either add the promised discussion or remove the sentence.","section":"Abstract and Section 3"},{"comment":"The text in Section 4 describes the W-mediated t-channel as one background and the Z-mediated s-channel as the other, but the caption of Fig. 4 labels the left panel as Z-mediated and the right panel as W-mediated; this is reversed relative to the text.","section":"Figure 4 and Section 4"},{"comment":"The cut row is written as '|η_γ| > 1 GeV'; pseudorapidity is dimensionless, so the 'GeV' unit should be removed.","section":"Table 1"},{"comment":"The sentence 'Based on the ILC snowmass report [84]' cites ref. [84], which is the CMB-HD Snowmass white paper; presumably the ILC Snowmass report [43] is the intended reference.","section":"Section 4, reference [84]"},{"comment":"The momentum symbol k is used both for the off-shell photon momentum and for the photon polarization vector, which makes the amplitude notation ambiguous; a distinct symbol such as q or k_γ for the on-shell photon should be introduced.","section":"Appendix B, Eq. (31)"}],"recommendation":"major_revision","confidential_remarks":"The normalization discrepancy is the load-bearing issue: the paper's quantitative claims depend on resolving it in favour of the Monte Carlo normalization. The relic-side formulas also need a minor correction, but the invisible branching ratio is not threatened. The study is otherwise a reasonable phenomenological exercise, and the missing kinetic-decoupling discussion and miscited reference should be fixed in revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick read: this is a straightforward phenomenological projection for the electroweak axion-portal DM model of Ref. [22], applied to a 1 TeV ILC mono-photon search. The central logic is sound: the final-state photon from e+e- -> a gamma has a missing-energy peak near half the beam energy, well separated from the SM neutrino background, and beam polarization helps. The relic-density contour matches [22] and the benchmark (ma=5 GeV, fa=186 GeV, mPsi~2.47 GeV) is a legitimate target. What is new is the ILC cutflow, the Z versus luminosity and mass curves, and the chi2-based estimate of gaγγ accuracy. That is a useful service for the community, though it is not a new framework.\n\nThe soft spots, in order of severity. First, the analytic cross section in Eq. (B.32) does not agree with the simulated event numbers. Integrating Eq. (B.32) at sqrt(s)=1 TeV with the benchmark coupling gives sigma(e+e- -> gamma a) about 1.4e-4 pb, which at 5 ab^-1 yields roughly 700 events before cuts. Table 1 quotes 1750 signal events after basic cuts (unpolarized), implying sigma about 3.5e-4 pb, a factor around 2.5 higher. For the polarized case the discrepancy is closer to 3.5. The paper does not address this. This matters because the quoted Z=7.64 and the 1.1% gaγγ precision rest on the simulation yield; if the analytic formula is right, the significance drops below 5 sigma, and if the simulation is right, Eq. (B.32) is wrong and the chi2 analysis in Sec. 5, which explicitly uses that formula, is not reproducible. A referee needs this reconciled.\n\nSecond, Eq. (15) for Gamma(a -> Psi Psi-bar) is missing both c_Psi^2 and the (1-4m_Psi^2/ma^2) factor in front of the square root. As written it overestimates the width; with the correct factors the invisible branching ratio is still about 0.999, so the collider signal is not materially diluted, but the formula should be fixed.\n\nThird, the abstract promises that the effect of early kinetic decoupling is discussed in the resonant regime, but no such discussion appears in the body. Either add it or cut the sentence.\n\nFourth, there is no public code, UFO file, or detector cards, so the event count cannot be checked externally. That is not fatal for a phenomenological paper, but it makes the normalization question harder to resolve.\n\nWho is this for? ILC phenomenologists and anyone working on ALP-portal DM; the Emiss double-peak discrimination is a nice observation and likely transferable.\n\nMy recommendation: send to peer review, but flag the cross-section discrepancy as a major revision. The paper is useful and the core idea is not broken, but the current numbers cannot be taken at face value.","headline":"A competent ILC projection for an established ALP-portal DM model, but the simulated signal looks a factor 2.5–3.5 above its own analytic cross-section and the abstract oversells a kinetic-decoupling discussion that isn't in the body.","tokens_in":15897,"tokens_out":10994,"would_cite":true,"duration_ms":101011,"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 GeV-scale axion-like particle that mediates dark matter and decays invisibly is discoverable at a 1 TeV electron-positron collider, with 7.64σ significance and a 1%-level measurement of its photon coupling.","keywords":["axion-like particles","dark matter mediator","mono-photon search","missing energy","electron-positron collider","ALP-photon coupling","relic density","beam polarization"],"falsifier":"Count diphoton events from $a\\to\\gamma\\gamma$ in the same 1 TeV, 5 ab$^{-1}$ sample: if more than a few percent of produced ALPs decay visibly, the invisible-branching assumption fails and the significance drops below $5\\sigma$. A calculation of $\\Gamma(a\\to\\Psi\\bar{\\Psi})$ with the exact phase-space factor $(1-4m_\\Psi^2/m_a^2)^{3/2}$ and the $c_\\Psi^2$ coupling would show whether the invisible width really dominates $\\Gamma(a\\to\\gamma\\gamma)$.","tokens_in":14716,"feed_emoji":"⚛️","tokens_out":12084,"duration_ms":110591,"temperature":0.7,"pith_summary":"This paper argues that a GeV-scale axion-like particle (ALP) can act as the portal between the Standard Model and a Dirac fermion dark matter particle, and that the same ALP-photon interaction which sets the dark matter relic abundance is directly testable at a 1 TeV electron-positron collider. In the near-resonant regime, with the ALP mass just above twice the dark matter mass ($m_a/m_\\Psi = 2.02$), thermal freeze-out through $\\Psi\\bar{\\Psi}\\to a\\to \\gamma\\gamma$ produces the observed relic density while evading gamma-ray line searches. The ALP then decays almost entirely into the dark matter pair, so the collider signature is a single hard photon plus missing energy from $e^+e^-\\to \\gamma a \\to \\gamma+\\Psi\\bar{\\Psi}$. With 5 ab$^{-1}$ and polarized beams, the paper finds a significance of $7.64\\sigma$ after the cuts $E_{\\rm miss}<510$ GeV and $|\\eta_\\gamma|>1$, and a binned $\\chi^2$ analysis can determine the ALP-photon coupling to about 1%. The result matters because it puts dark matter cosmology and a clean lepton-collider observable on one testable footing.","feed_headline":"Axion mediator hits 7.6-sigma mono-photon sensitivity","feed_subtitle":"Polarized e+e− beams let the ALP-photon coupling that fixes dark matter's abundance be measured to about 1%.","key_machinery":"The engine of the argument is the ALP-photon coupling $g_{a\\gamma\\gamma}=\\alpha_{\\rm EM}/(\\pi f_a)$, which enters both the relic-abundance process $\\Psi\\bar{\\Psi}\\to a\\to\\gamma\\gamma$, through a near-resonant $s$-channel ALP, and the collider production $e^+e^-\\to\\gamma a$. The near-resonant mass ratio $r=m_a/m_\\Psi=2.02$ lets thermal freeze-out produce $\\Omega h^2\\simeq0.12$ while evading gamma-ray line constraints, and the same coupling appears in the differential production cross section, so cosmology and collider reach are tied to one parameter. A second load-bearing piece is the invisible decay $a\\to\\Psi\\bar{\\Psi}$, assumed to dominate the total width; it converts every produced ALP into missing energy and creates the mono-photon signal. The kinematic handle that separates signal from background is the missing-energy peak at $\\sqrt{s}/2$: the ALP carries half the beam energy in associated production, while the neutrino backgrounds populate higher and distinctly shaped peaks.","core_discovery":"On the paper's own terms, the central discovery is that mono-photon plus missing-energy events cleanly separate ALP-portal dark matter from the Standard Model neutrino background at a 1 TeV $e^+e^-$ collider. For the benchmark $m_a=5$ GeV, $f_a=186$ GeV, and $m_\\Psi=2.47$ GeV, associated ALP production followed by the invisible decay $a\\to\\Psi\\bar{\\Psi}$ produces a missing-energy distribution peaking at 500 GeV, whereas the dominant $W$-mediated $\\nu\\bar{\\nu}$ background peaks near $\\sqrt{s}$ and the subdominant $Z$-mediated piece sits at a lower, distinct position; the upper cut $E_{\\rm miss}<510$ GeV removes roughly 97% of the background while leaving the signal intact. Adding $|\\eta_\\gamma|>1$ and the beam polarization $\\{P_{e^+}:P_{e^-}\\}=\\{-20\\%:+80\\%\\}$ leaves about 1810 signal events against about 56,189 background events, corresponding to $Z=7.64$ at $\\sqrt{s}=1$ TeV with 5 ab$^{-1}$, and the same setup measures $g_{a\\gamma\\gamma}\\simeq 1.33\\times10^{-5}$ GeV$^{-1}$ to about 1% precision.","pith_inferences":["The claimed significance scales linearly with the invisible branching ratio ${\\rm Br}(a\\to\\Psi\\bar{\\Psi})$; if visible decays such as $a\\to\\gamma\\gamma$ compete at the few-percent level, the quoted $Z$ is diluted, and a first-principles width including the phase-space factor $(1-4m_\\Psi^2/m_a^2)^{3/2}$ and the $c_\\Psi^2$ coupling is needed to certify the benchmark.","The abstract announces a discussion of early kinetic decoupling in the resonant regime, but the main text as provided does not contain that analysis; including it could shift the relic-satisfied contour and, with it, the benchmark collider predictions.","Because the discriminating power relies on the missing-energy peak at $\\sqrt{s}/2$, other future $e^+e^-$ machines would need re-optimized cuts, but the mono-photon strategy transfers wherever polarized beams are available.","The EFT validity condition $\\sqrt{s}<4\\pi f_a$ holds at the benchmark, but near the lower-$f_a$ edge of the relic contour the effective scale approaches the collider energy, so a propagator-level computation would be a useful cross-check of the quoted rates."],"forward_implications":["A 5-sigma discovery of the ALP portal is within reach at a 1 TeV $e^+e^-$ collider: the paper quotes roughly 1 ab$^{-1}$ for 3-sigma and 3 ab$^{-1}$ for 5-sigma with the chosen polarization.","The ALP-photon coupling that fixes the dark matter relic abundance can be extracted at the collider to about 1% precision, turning a cosmological target into a laboratory measurement.","The mono-photon signature remains effective for ALP masses up to roughly 100 GeV, beyond which phase-space suppression lowers the significance.","With unpolarized beams the same analysis reaches only $Z\\simeq2.86$, so beam polarization is essential to the claimed sensitivity.","Because the ALP decays invisibly, the collider events directly produce dark matter pairs, so the observed missing-energy events would be production of the same particle that populates the relic density."],"supporting_citations":[{"why":"supplies the near-resonance condition $m_a/m_\\Psi\\approx 2$ as the way to reach the relic density while evading indirect-detection bounds.","marker":"[14]"},{"why":"provides the invisible-ALP search constraints and the resonance-region motivation for the benchmark scenario.","marker":"[20]"},{"why":"supplies the electroweak ALP portal framework, the relic-density treatment, and the indirect-detection constraints used in the parameter scan.","marker":"[22]"},{"why":"gives the effective ALP Lagrangian and future-collider signal definitions that the mono-photon search is built on.","marker":"[47]"},{"why":"fixes the observed relic density $\\Omega h^2\\simeq0.12$ that the freeze-out scan targets.","marker":"[61]"},{"why":"computes the relic-density-satisfied parameter space from which the benchmark point is taken.","marker":"[62]"},{"why":"fixes the beam-polarization benchmark $\\{P_{e^+}:P_{e^-}\\}=\\{-20\\%:+80\\%\\}$ used in the collider analysis.","marker":"[84]"},{"why":"defines the significance formula $Z$ used to quote the signal observability.","marker":"[85]"}],"fun_headline_variants":["Mono-photon signal cleanly isolates ALP dark matter","7.6-sigma reach for axion-mediated dark matter","ALP-photon coupling pinned to 1% at 1 TeV collider","Dark matter relic abundance tied to mono-photon excess","Probing fermionic dark matter via axion-photon portal"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Almost every produced ALP must decay invisibly into the dark matter pair; if visible decays such as $a\\to\\gamma\\gamma$ compete at even the few-percent level, the missing-energy signal and the quoted $7.64\\sigma$ significance shrink correspondingly.","fun_headline_variants_meta":{"raw":{"variants":["Mono-photon signal cleanly isolates ALP dark matter","7.6-sigma reach for axion-mediated dark matter","ALP-photon coupling pinned to 1% at 1 TeV collider","Dark matter relic abundance tied to mono-photon excess","Probing fermionic dark matter via axion-photon portal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000687,"raw_usage":{"total_tokens":3141,"prompt_tokens":1001,"completion_tokens":2140,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":617,"completion_tokens_details":{"reasoning_tokens":2051}},"tokens_in":617,"tokens_out":2140,"duration_ms":15403,"temperature":1.0,"reasoning_tokens":2051,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:43:17.343752+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Count diphoton events from $a\\to\\gamma\\gamma$ in the same 1 TeV, 5 ab$^{-1}$ sample: if more than a few percent of produced ALPs decay visibly, the invisible-branching assumption fails and the significance drops below $5\\sigma$. A calculation of $\\Gamma(a\\to\\Psi\\bar{\\Psi})$ with the exact phase-space factor $(1-4m_\\Psi^2/m_a^2)^{3/2}$ and the $c_\\Psi^2$ coupling would show whether the invisible width really dominates $\\Gamma(a\\to\\gamma\\gamma)$.","supporting_citations":[],"review_version":1}