{"id":"b0c1b0e3-130a-4ed6-a695-7a91dbbbcc68","arxiv_id":"2411.09042","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"Resonant conversion of axion-like particles into photons in stochastic magnetic fields can produce an f^-2 radio excess matching ARCADE-2 and deepen the 21cm absorption trough seen by EDGES.","lead":"This paper proposes that axion-like particles converting into photons in cosmic magnetic fields can explain both the ARCADE-2 radio excess and the EDGES 21-centimeter absorption signal with one mechanism. A smart generalist might read it because it ties two unexplained low-frequency radio anomalies to dark radiation and primordial magnetism, with new predictions for upcoming experiments.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central benchmark assumes a frequency-flat ALP abundance that is not self-consistently normalized with the total dark-radiation density γ=0.06; under the invoked decoupled-production picture, the predicted radio excess is orders of magnitude smaller, so the claimed simultaneous fit is not…","rationale":"The reader's weakest assumption—that the EDGES trough must be a genuine astrophysical signal—is a legitimate condition on the word 'simultaneously', and the paper itself concedes the 21 cm data are inconclusive. However, I do not think it is the single most load-bearing concern, because even if EDGES is confirmed, the radio-excess half of the central claim depends on a frequency-flat ALP spectrum whose normalization is handled inconsistently with the total dark-radiation ratio γ=0.06. Equation (5) uses Ω_ALP(ω0)/Ωγ0 = γ, but γ is an integral over logarithmic frequency; for a flat spectrum the per-log value is γ/Δlnω. Over the 78 MHz–10 GHz band this is a factor of about 4.8, and a thermal decoupled population with the same total γ would suppress the low-frequency abundance by orders of magnitude. This concern is internal to the paper's own formulas and does not rely on external data status. It strengthens the case for keeping the verdict CONDITIONAL rather than moving to ACCEPT or REJECT: the mechanism may still survive with a renormalized spectrum or a nonthermal production model, but the benchmark amplitudes and the 5σ/r overlap in Fig. 2 are not reliable as stated. The paper does have real value in the testable predictions (f^-2 spectrum above 0.5 GHz and a second trough below 30 MHz), so the appropriate verdict is unchanged conditional acceptance pending the normalization check and future 21 cm data.","tokens_in":19653,"tokens_out":16641,"duration_ms":175605,"concrete_test":"Recompute the ARCADE-2 fit and the r contours of Fig. 2 using the correct normalization for a flat spectrum, γ = ∫_{ln(0.078 GHz)}^{ln(10 GHz)} dlnω Ω_ALP(ω)/Ωγ = 0.06, so Ω_ALP(ω0)/Ωγ = 0.06/ln(10/0.078) ≈ 0.0124 in Eqs. (5) and (7), and rerun the χ² analysis and the r=0.02–0.2 overlap. If the best-fit B0 shifts by more than a factor √(ln(10/0.078)) ≈ 2.2, or if the overlap with the r benchmarks changes by more than the quoted range, the stated benchmark (B0≈0.1 nG, gaγ=5×10^-13 GeV^-1) is not self-consistent and the simultaneous-fit claim needs revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Equation (5) evaluates T_b0 using Ω_ALP(ω0)/Ωγ0 = γ = 0.06, treating the per-logarithmic-frequency ALP density as equal to the total ALP-to-photon energy-density ratio. But γ is defined as the total ratio, γ = Ω_ALP/Ωγ = ∫ dlnω Ω_ALP(ω)/Ωγ. For a frequency-flat spectrum extending over the relevant band, say 78 MHz to 10 GHz, this integral contributes a factor Δlnω ≈ 4.8, so the per-log value used in Eq. (5) should be γ/Δlnω, not γ. Appendix B explicitly writes 'ΩALP = ∫ d lgω ΩALP(ω) ≃ ΩALP(ω)', which is valid only if the spectrum is supported over about one e-fold; but the same flat spectrum is used across the ARCADE band and the 21 cm rest-frame band, so the approximation fails. The result is an overestimate of T_b0 and T_b(z) by roughly Δlnω, equivalent to overestimating B0 or gaγ by √Δlnω ≈ 2.2. A related, physical version of the same problem is that a decoupled thermal ALP population normalized to total γ=0.06 has a blackbody spectrum, not a flat one; at 78 MHz and 1 GHz the energy density per log is suppressed by powers of x=hν/kT_ALP, making the low-frequency radio excess negligible. The paper does not supply a production mechanism that populates only radio frequencies. Thus the central amplitude of the shared-origin explanation rests on an unstated and internally inconsistent spectral ansatz, independent of whether the EDGES trough is eventually confirmed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper attributes the ARCADE-2 radio excess (22 MHz–10 GHz) and the EDGES 21 cm absorption trough at 78 MHz to a single source: resonant ALP–photon conversion in a stochastic primordial magnetic field during the post-recombination epoch. Relativistic ALPs with a frequency-flat abundance, mass ma ≈ 3×10^-14–3×10^-13 eV, photon coupling gaγ = 5×10^-13 GeV^-1, and energy-density ratio γ = 0.06 (close to the Planck ΔNeff limit) convert in a field of amplitude B0 ≈ 0.1 nG with a two-scale power spectrum, producing a present-day brightness spectrum T_b ∝ f^-2. The model is reported to improve the radio fit by Δχ² = 65 (≈5σ) over a synchrotron-only model and to yield fractional contributions r ≈ 0.02–0.2 at 78 MHz. The 21 cm analysis adds the extra background TR(z) = T0(1+z) + rC(1+z)^3 Θ(zres−z) to the standard simulation and finds trough depths compatible with EDGES-2/3 for star-formation efficiencies f* ≈ 1–10%. The paper explicitly labels the 21 cm conclusion provisional because the EDGES signal is disputed (SARAS-3 rejected it at 95.3% CL), and it predicts a spectral slope change above 0.5 GHz and a second trough below 30 MHz for the heavier-mass benchmark.","tokens_in":20219,"tokens_out":38591,"duration_ms":344375,"significance":"The proposed framework is attractive in principle: if calibrated correctly it would unify two long-standing radio anomalies within one dark-sector mechanism, and it yields cleanly falsifiable predictions — the f^-2 scaling above 0.5 GHz, the second absorption trough below 30 MHz for the heavier mass, and a small y-type distortion (Appendix F) consistent with COBE/FIRAS. The conversion-probability calculation in Appendix A is a genuine strength: the semi-steady discretization, the linearized detuning, and the stochastic-resonance suppression factor in Eq. (24) go beyond domain-like Landau–Zener treatments and appear internally consistent. The main limitation is that the amplitude calibration is not supportable as written: the assumed flat ALP spectrum with γ = 0.06 cannot simultaneously satisfy the total-energy normalization and the Planck ΔNeff bound (Major Comment 1), and the stated thermal-production picture would give a near-blackbody spectrum, not a flat radio-band one. Combined with the contested status of the EDGES detection, the paper currently demonstrates a plausible mechanism and a proof of principle rather than an established simultaneous explanation.","major_comments":[{"comment":"The conversion amplitude is over-normalized. The paper defines γ = Ω_a/Ω_γ as the total ALP-to-photon energy-density ratio and then uses γ as the per-logarithmic-frequency ratio Ω_ALP(ω0)/Ω_γ0 in Eq. (5) (and in Eqs. (7) and (30) for the 21 cm background). For a frequency-flat spectrum, Ω_ALP(ω) = const, the total is ∫ dlnω Ω_ALP(ω) = Ω_ALP(ω) × Δlnω, so Appendix B's assertion 'Ω_ALP = ∫ d lg ω Ω_ALP(ω) ≃ Ω_ALP(ω) due to the logarithmic integral manner' is incorrect by Δlnω ≈ 5–6 over the 22 MHz–10 GHz band (and by more if the spectrum extends to THz as in Appendix F). Consequently T_b^AP in Eq. (5) is overestimated by this factor, equivalent to overestimating B0 or gaγ by √Δlnω ≈ 2.3, and the same error propagates into the r contours of Fig. 2 that feed the 21 cm analysis. The inconsistency has a physical counterpart: the only production mechanism mentioned (thermal decoupling near 0.1 GeV, Appendix B) produces a quasi-blackbody ALP spectrum with T_ALP ≈ 0.4 T0 and peak near 60–70 GHz, whose per-log energy density at 78 MHz is suppressed by (0.078 GHz/60 GHz)^3 ≈ 2 × 10^-9, making the radio signal negligible. The three statements 'γ = 0.06 (total)', 'flat spectrum over the ARCADE band', and 'ΔNeff ≤ 0.33' cannot hold simultaneously; the analysis must be redone either with per-log normalization γ/Δlnω or with a non-thermal production mechanism for a radio-frequency population, and the viable regions in Figs. 2, 6, and 7 must be recomputed.","section":"Eq. (5), Eqs. (7)/(30), and Appendix B"},{"comment":"The central 'simultaneous explanation' claim is conditional on the reality of the EDGES trough, which the authors themselves treat as inconclusive ('Despite the inconclusive results on 21cm signal, in this paper we concentrate on the anomaly data reported from EDGES'; 'This conclusion should be considered provisional for reference only'). SARAS-3 (Ref. [14]) rejected the EDGES best-fit profile at 95.3% confidence, and the supporting EDGES-3 results cited in the main text are laboratory memos (Refs. [62–64]), not peer-reviewed publications. If the trough is a systematic, the 21 cm half of the claim cannot be sustained. The abstract's phrasing ('resolve both anomalies simultaneously') and the Conclusion should therefore be revised to present the 21 cm connection as a conditional application of the radio model, not as an established resolution.","section":"Introduction and Conclusion"},{"comment":"The statistical support for the radio half is weaker than the '5σ' framing suggests. The best fit has χ²_min = 49 with 12 degrees of freedom (χ²_ν ≈ 4), so the ALP-plus-minimal-EGR model is still a formally poor fit to the 14 data points (p ≈ 10^-6 for χ²_ν = 4); the quoted 'more than 5σ' is the Δχ² = 65 improvement over the null model, i.e., a relative statement. Given that the low-frequency excess data are themselves subject to unmodeled systematics (as the introduction notes), the paper should report the absolute goodness of fit and present the radio explanation as a relative improvement that is promising but not yet a high-quality fit.","section":"Radio Excess and 21cm Trough (χ² analysis)"}],"minor_comments":[{"comment":"Units and typos: 'costant' (§ALP-Photon Mixing), 'discritizing' (Appendix A), 'invarinat' (Appendix F), 'affects' → 'effects' (Appendix C), and the inconsistent field-strength notation 'nGs' (Fig. 1 caption) vs 'nG' (text).","section":"Throughout"},{"comment":"The sentence 'the scale invariant spectrum is insensitive to both UV and IR cutoffs' is misleading: it is the differential spectrum Ω_ALP(ω) that is cutoff-insensitive, while the total energy ∫ dlnω Ω_ALP(ω) grows with the bandwidth and is centrally relevant to the ΔNeff normalization.","section":"Appendix F"},{"comment":"The 21 cm agreement in Fig. 3 is obtained by fitting: the paper varies r, z_res (through ma), and f*, and scans fα and fX in Appendix C, so the matching trough is not a free prediction; the genuinely predictive elements are the f^-2 slope above 0.5 GHz and the sub-30 MHz second trough, and the abstract should distinguish these.","section":"Abstract and Fig. 3"},{"comment":"The benchmark coupling gaγ = 5×10^-13 GeV^-1 is acknowledged to lie 'slightly beyond the region excluded by Chandra' (Ref. [37]); because the normalization correction of Major Comment 1 pushes the required coupling or field strength upward, the paper should quantify how much of the viable region in Figs. 2, 6, and 7 survives existing astrophysical exclusion bounds.","section":"Radio Excess and 21cm Trough"},{"comment":"The χ² statement '14 experimental data' with '12 d.o.f.' should specify which parameters are varied in the fit (presumably B0 and λB) and how the 22, 45, 408, and 1420 MHz survey points are weighted relative to ARCADE-2.","section":"Radio Excess and 21cm Trough (χ² analysis)"},{"comment":"The datasets labeled EDGES-2/3 in the right panel should be marked as preliminary (lab memos Refs. [62–64]) in the figure caption, and the main text should restate that the EDGES-3 'consistency' has not passed peer review.","section":"Fig. 3 caption"}],"recommendation":"major_revision","confidential_remarks":"The normalization issue in Eq. (5) and Appendix B is the paper's central technical defect: it is an internal inconsistency (the total and per-log ratios are conflated), not a matter of taste, and it directly inflates the headline signal. I want the authors to have the chance to fix it — the conversion calculation itself is careful and the predictions are testable — but the revised version must recompute the amplitude, the r parameter, and the viable-region plots, and either provide a production mechanism for a flat radio-band spectrum or adopt the (much weaker) blackbody normalization, which would likely eliminate the radio fit. Separately, the paper leans on unpublished lab memos for EDGES-3 and on the contested EDGES detection; the editor may want to ensure the abstract does not overclaim. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take on 2411.09042. The genuinely new piece is the stochastic-resonance conversion treatment in Appendix A: they discretize the post-recombination era, derive a conversion probability for a turbulent two-scale magnetic field, and get a clean f^-2 brightness spectrum with a suppression factor that domain-like models miss. That is a real technical step. The two explicit predictions — f^-2 scaling above 0.5 GHz and a second absorption trough below 30 MHz — are testable and give the paper value.\n\nThe simultaneous-fit claim, though, has two soft spots. First, the EDGES signal remains disputed; SARAS-3 rejects the EDGES best-fit at 95.3% and the authors themselves call the data inconclusive, then proceed anyway. If the trough is a systematic, the 21cm half of the argument collapses. Second, the amplitude normalization in Eq. (5) is off. They set Omega_ALP(omega0)/Omega_gamma = gamma = 0.06, but gamma is defined as the total ALP-to-photon energy-density ratio. For a frequency-flat spectrum over the 78 MHz–10 GHz band, the per-log value should be gamma/Delta ln omega ≈ gamma/4.8. Appendix B's claim that ∫ d ln omega Omega_ALP(omega) ≈ Omega_ALP(omega) is only valid if the spectrum is supported over about one e-fold, which is not the case here. Correcting this lowers the predicted radio excess by roughly Delta ln omega, and recovering the fit would require boosting ga_gamma or B0 by about sqrt(Delta ln omega) ≈ 2.2 — while gamma already sits at the Planck bound and the benchmark coupling is already above the Chandra exclusion. A thermally decoupled ALP population normalized to gamma=0.06 would also have a blackbody spectrum, not a flat one, so no production mechanism for the assumed low-frequency flat spectrum is supplied. That is load-bearing, not cosmetic.\n\nThe 5 sigma improvement is against a simplified fixed astrophysical null model; I would not lean on it. The 21cm fit also uses r taken from the ARCADE-2 fit, so it is more a parameter scan than a prediction. The intended audience is ALP phenomenologists and 21cm/radio observers; the derivation and forecasts are worth their time.\n\nBottom line: this is a serious, well-written paper with a useful derivation and two real forecasts, but the central simultaneous explanation currently rests on an unnormalized spectrum plus a disputed dataset. I would send it to a knowledgeable referee who can check the normalization and demand either a production mechanism or a rescaled parameter region. I would not cite the central amplitude claim as it stands.","headline":"A genuinely new stochastic-resonance treatment and two testable forecasts, but the shared-origin claim rests on a spectral normalization error and a disputed 21cm dataset.","tokens_in":20678,"tokens_out":4570,"would_cite":false,"duration_ms":47120,"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":"Resonant conversion of axion-like particles into photons in a weak primordial magnetic field can simultaneously explain the ARCADE-2 radio excess and the EDGES 21cm absorption trough.","keywords":["axion-like particles","ALP-photon conversion","ARCADE-2 radio excess","EDGES 21cm anomaly","primordial magnetic field","cosmic microwave background","21cm cosmology","dark radiation"],"falsifier":"A future global 21cm experiment that sets a 95%-confidence upper limit excluding a trough deeper than about $-0.3$ K at 78 MHz would disprove the simultaneous explanation, since the model needs $r\\gtrsim0.02$ to explain the ARCADE-2 excess and that $r$ range is what deepens the trough.","tokens_in":19447,"feed_emoji":"📡","tokens_out":14564,"duration_ms":199306,"temperature":0.7,"pith_summary":"This paper tries to show that one new-physics mechanism can explain two unexplained radio anomalies at once: the extra radio brightness measured by ARCADE-2 and other low-frequency surveys between 22 MHz and 10 GHz, and the surprisingly deep 21cm absorption trough reported by EDGES at 78 MHz. The mechanism is resonant conversion of axion-like particles (ALPs) into photons in a stochastic primordial magnetic field during the post-recombination universe. If correct, the model also predicts a specific spectral shape and a second low-frequency absorption trough, so upcoming radio experiments can test it. The authors are careful to note that the 21cm anomaly remains inconclusive, but they concentrate on it because the newest EDGES-phase data still support a deep trough.","feed_headline":"Axion-photon conversion could solve two radio anomalies","feed_subtitle":"Resonant conversion of dark axion-like particles would explain both the extra radio glow and the deep 21cm absorption.","key_machinery":"The load-bearing object is the ALP–photon mixing matrix in an expanding universe, driven by the interaction $-\\frac{1}{4}\\sqrt{-g}\\,g_{a\\gamma}F_{\\mu\\nu}\\tilde{F}^{\\mu\\nu}a$. Conversion is negligible until the plasma frequency $\\omega_{\\rm pl}(z)$ equals the ALP mass $m_a$, the mass-equal resonance; at that redshift the comoving oscillation length $l_{\\rm osc}(z)=2(1+z)/|\\Delta_{\\rm pl}-\\Delta_a|$ grows sharply and the conversion probability peaks. The probability is computed perturbatively with a semi-steady approximation over redshift bins, yielding $P(z_i,z_{i+1})\\propto g_{a\\gamma}^2\\int k^2 P_B(k)\\,W(t_1,t_2;k)$ for a stochastic Gaussian magnetic field, and the resulting present-day brightness temperature is $T_{b0}^{\\rm AP}(\\omega_0)=(\\pi^4\\gamma/15)\\,T_0^4\\,\\omega_0^{-3}\\,P_{\\rm tot}(\\omega_0,0)$. For 21cm physics the extra background is parameterized as $T_R(z)=T_0(1+z)+rC(1+z)^3\\Theta(z_{\\rm res}-z)$ with $C=2.698\\,\\mathrm{K}$, so only the resonance redshift and the fraction $r=T^{\\rm AP}_{b0}/T^{\\rm obs}_{b0}|_{f_{78}}$ control the trough depth.","core_discovery":"The central claim is that relativistic ALPs with masses in the range $10^{-14}$ to $10^{-12}\\,\\mathrm{eV}$, an ALP–photon coupling near $g_{a\\gamma}=5\\times10^{-13}\\,\\mathrm{GeV}^{-1}$, and an ALP-to-photon energy-density ratio of $\\gamma=0.06$ can resonantly convert into photons when the effective photon mass matches the ALP mass in the post-recombination plasma. With a present-day magnetic field strength around $B_0=0.1\\,\\mathrm{nG}$, the converted radiation produces a brightness temperature scaling as $f^{-2}$ that improves the fit to the ARCADE-2 excess by more than $5\\sigma$ over astrophysical synchrotron emission alone. The same photons raise the radiation background at $z\\simeq17$ and deepen the 21cm absorption trough into the range reported by EDGES, giving a single source for both anomalies. For a higher ALP mass with resonance at $z\\simeq100$, the model predicts an additional absorption trough below 30 MHz.","pith_inferences":["One consequence the authors do not draw out: if a future global 21cm experiment rules out a 78 MHz trough deeper than about $-0.3$ K, the two halves of the model separate, and the ARCADE-2-only fit could still survive even though the unified explanation would not.","The paper's appendix explores red, flat, and blue ALP spectra but stops short of a full scan; a red spectrum $\\Omega_{\\rm ALP}\\sim\\omega^{-0.5}$ already fits the radio data better, so a systematic spectral-shape search would sharpen the prediction.","A laboratory detection of an ALP in the $10^{-14}$–$10^{-12}$ eV mass range with a coupling near $5\\times10^{-13}\\,\\mathrm{GeV}^{-1}$ would independently corroborate the astrophysical scenario, since the benchmark sits just above current quasar X-ray bounds."],"forward_implications":["ALP-to-photon conversion replaces part of the extragalactic synchrotron explanation for the ARCADE-2 excess, improving the fit from $\\chi^2=114$ to $\\chi^2=49$ (12 d.o.f.) for the best benchmark.","The injected radiation deepens the global 21cm absorption line to the EDGES-observed level for $0.02\\lesssim r\\lesssim0.2$, with the trough depth governing the allowed ALP and magnetic-field parameters.","For a resonance at $z_{\\rm res}\\simeq100$ (higher ALP mass), the model predicts a second absorption trough around $f\\sim15$ MHz, below the standard 21cm feature.","Above about 100 GHz the brightness-temperature scaling steepens from $f^{-2}$ to $f^{-3}$, a signature absent in domain-like treatments of the magnetic field.","Jointly fitting both anomalies converts the radio anomalies into a probe of the ALP mass and coupling and of the primordial magnetic field's strength, correlation length, and spectral index."],"supporting_citations":[{"why":"It supplies the ARCADE-2 brightness-temperature excess data that the model fits from 3 to 90 GHz.","marker":"[5]"},{"why":"It reports the EDGES 21cm absorption trough at 78 MHz that the model targets.","marker":"[12]"},{"why":"It describes the SARAS-3 measurement that rejected the EDGES best-fit profile, the main caution on the 21cm anomaly.","marker":"[14]"},{"why":"It provides the global 21cm simulation with Lyman-$\\alpha$ and X-ray heating used to compare model predictions with EDGES data.","marker":"[28]"},{"why":"It gives the ALP-photon mixing kernel in an expanding universe on which the conversion-probability calculation is based.","marker":"[32]"},{"why":"It supplies the stochastic primordial magnetic-field spectra and the $B_0$–$\\lambda_B$ parameter plane used to scan viable regions.","marker":"[34]"},{"why":"It identifies stochastic-resonance conversion in the same magnetic-field setup and is used for the spectral comparison.","marker":"[35]"},{"why":"It bounds the ALP-to-photon energy-density ratio through extra relativistic species, fixing $\\gamma\\simeq0.06$.","marker":"[36]"},{"why":"It sets the X-ray constraint on the ALP-photon coupling that the benchmark choice slightly exceeds.","marker":"[37]"},{"why":"It provides the CMB distortion constraint that the viable parameter combination $g_{a\\gamma}B$ is checked against.","marker":"[46]"}],"fun_headline_variants":["Axion-photon conversion ties ARCADE2 and EDGES anomalies together","Dark axions could explain both extra radio glow and 21cm dip","Single axion mechanism resolves two radio anomalies","Axion-photon resonance links excess radio and deep 21cm trough","One axion model fits ARCADE2 excess and EDGES absorption"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 21cm half of the central claim assumes the EDGES absorption trough is a genuine astrophysical signal, which the paper itself calls inconclusive and which a SARAS-3 measurement rejected at 95.3% confidence.","fun_headline_variants_meta":{"raw":{"variants":["Axion-photon conversion ties ARCADE2 and EDGES anomalies together","Dark axions could explain both extra radio glow and 21cm dip","Single axion mechanism resolves two radio anomalies","Axion-photon resonance links excess radio and deep 21cm trough","One axion model fits ARCADE2 excess and EDGES absorption"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000241,"raw_usage":{"total_tokens":1511,"prompt_tokens":921,"completion_tokens":590,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":537,"completion_tokens_details":{"reasoning_tokens":499}},"tokens_in":537,"tokens_out":590,"duration_ms":6326,"temperature":1.0,"reasoning_tokens":499,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:08:34.202937+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A future global 21cm experiment that sets a 95%-confidence upper limit excluding a trough deeper than about $-0.3$ K at 78 MHz would disprove the simultaneous explanation, since the model needs $r\\gtrsim0.02$ to explain the ARCADE-2 excess and that $r$ range is what deepens the trough.","supporting_citations":[{"cited_title":"COSMIC WISPers","cited_arxiv_id":null,"evidence_quote":"It provides the global 21cm simulation with Lyman-$\\alpha$ and X-ray heating used to compare model predictions with EDGES data."}],"review_version":1}