{"id":"1e771914-50ab-43e2-99c5-6a31fe99a855","arxiv_id":"2411.08577","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":7,"one_line_summary":"From MCMC fits to Fermi-LAT and MAGIC data on B2 2234+28A and 3C 454.3, the paper reports best-fit ALP-photon couplings of 3.05 and 7.40 x 10^-11 GeV^-1 with masses near 5 x 10^-8 eV, plus 95% CL upper limits.","lead":"This paper fits axion-like particle parameters to gamma-ray spectra of two distant active galaxies, reporting a coupling around 3 to 7 x 10^-11 GeV^-1 and a mass near 5 x 10^-8 eV. The analysis tests whether axion-photon mixing can explain very-high-energy photons surviving intergalactic absorption, but the main B2A result depends on treating upper limits as detections.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The B2A gaγ best-fit is not supported because seven MAGIC upper limits enter Eq. (16) as detections; the paper's own Section V.A concedes only an upper threshold, so the abstract overstates a fit.","rationale":"The reader's primary weakest assumption is exactly right: the seven B2A upper limits are entered into the ordinary chi-square of Eq. (16) as though they were measured fluxes. The manuscript itself flags this in Section V.A, noting that the result can only serve as an upper threshold, yet the abstract promotes the B2A MCMC value as a best-fit coupling. This is not an external consistency quarrel; it is an internal statistical inconsistency between the stated data type and the likelihood used. Treating upper limits as detections systematically penalizes models that fall below the limits, which is precisely the situation in Fig. 5 where the no-ALP prediction lies below the red points. The claimed improvement from ALP oscillations is therefore manufactured by the likelihood, not demanded by the data. The concrete test would settle the matter quickly: a censored-likelihood refit should remove the artificial pull toward high flux and very likely destroy the narrow gaγ peak. I agree with the reader's rejection of the paper's headline result. The CLs upper-limit analysis may be salvageable, but the current manuscript's central numerical claims are not supported as stated, so I see no reason to change the reader's REJECT verdict.","tokens_in":13604,"tokens_out":5580,"duration_ms":55506,"concrete_test":"Re-fit the B2A SED replacing Eq. (16) with a proper censored likelihood for the seven MAGIC upper limits, e.g. L_UL = Phi((Phi_UL - Phi_theo)/sigma_UL) for each upper limit while keeping Gaussian terms for the six lower-energy detections, or alternatively drop the seven upper limits entirely and fit only the six low-energy points. Report the resulting posterior on gaγ and the delta-chi^2 between the no-ALP and ALP models. If the gaγ posterior broadens or becomes consistent with zero, the abstract's gaγ = 3.05e-11 best-fit is an artifact of treating upper limits as detections.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing claim is that the MCMC fit to B2 2234+28A yields gaγ = 3.05e-11 GeV^-1 with narrow uncertainties, and that the ALP scenario fits the high-energy SED better than the null hypothesis. This claim fails at the level of the likelihood. Equation (16) defines chi-square as a sum over all N data points of (Phi_theo - Phi_obs)^2 / delta_i^2, treating every point as a Gaussian measurement. Section V.A explicitly states that the last 7 B2A SED points are not observed fluxes but upper limits, and that the result can therefore serve only as an upper threshold for gaγ. For a true upper limit, a theoretical spectrum lying well below the limit is perfectly consistent, whereas Eq. (16) penalizes the model for falling below the nominal flux value used for the upper limit. The apparent deficit of the no-ALP model above 0.1 TeV in Fig. 5 is therefore an artifact of treating 95% CL upper limits as central detections; it does not constitute evidence for ALP-induced enhancement. The abstract presents the B2A MCMC value as a measurement, not as the upper-limit-oriented statement the authors themselves attach to it. This statistical flaw alone invalidates the abstract's headline B2A coupling, independent of the additional systematic uncertainty from the hand-set source magnetic field. The 3C 454.3 fit, while not sharing the upper-limit issue, gives a best-fit gaγ above the CAST limit and is not a common-ALP consistency check with B2A; however, the B2A upper-limit treatment is the more direct and decisive defect.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a model-dependent source-selection criterion based on the ALP-photon conversion critical energy and a redshift-energy window, and applies it to two flat-spectrum radio quasars, B2 2234+28A and 3C 454.3, using Fermi-LAT and MAGIC data. The authors perform Markov Chain Monte Carlo fits with a power-law-with-exponential-cutoff intrinsic spectrum and a cellular magnetic-field model, reporting best-fit ALP parameters gaγ = 3.05+0.51-0.31 × 10^-11 GeV^-1 and ma = 5.25+2.35-2.65 × 10^-8 eV for B2A, and gaγ = 7.40+2.65-2.74 × 10^-11 GeV^-1 and ma = 5.50+1.69-2.17 × 10^-8 eV for 3C 454.3, together with a CLs exclusion region. The central claim is that the ALP scenario provides a better description of the observed high-energy spectral energy distributions than the no-ALP null hypothesis.","tokens_in":14011,"tokens_out":9005,"duration_ms":78948,"significance":"The paper introduces a useful idea: selecting high-redshift AGNs in a specific redshift-energy window to maximize sensitivity to ALP-photon oscillations. The authors use publicly available Fermi-LAT and MAGIC data, provide a clear description of the propagation model, and explicitly compare their CLs exclusion region with CAST, HESS, and Fermi limits. The MCMC machinery and the CLs scan are standard and reproducible in principle. If the statistical treatment of upper limits and the magnetic-field systematics were corrected, the method could contribute meaningful constraints in the small-mass ALP parameter space. The authors' explicit admission in Section V.A that the B2A result is only an upper threshold is a candid limitation, but it is not reflected in the abstract.","major_comments":[{"comment":"The last seven B2A SED points are upper limits, as stated in Section V.A, but Eq. (16) enters all points as symmetric Gaussian measurements. A model prediction below an upper limit is statistically consistent, yet the chi-square term penalizes it; entering the nominal upper-limit value as a central value biases the fit and the quoted gaγ. The paper itself concedes that the B2A result can only serve as an upper threshold, so the abstract's report of gaγ = 3.05+0.51-0.31 × 10^-11 GeV^-1 as a fit result is not supported. The fit should be redone with a censored likelihood or one-sided terms, and the B2A results should be presented only as upper limits.","section":"V.A and Eq. (16)"},{"comment":"The best-fit couplings for the two sources, gaγ = 3.05 × 10^-11 GeV^-1 and 7.40 × 10^-11 GeV^-1, do not agree within their quoted 68% intervals. Since ALP parameters are universal, two independent measurements should be consistent; the paper's statement that the difference 'may arise from our assumptions regarding the magnetic field' is not quantified. A joint fit or an explicit treatment of source-by-source systematics is needed before the results can be interpreted as constraints on a single ALP.","section":"V.B and Table I"},{"comment":"The 3C 454.3 best-fit gaγ = 7.40 × 10^-11 GeV^-1 lies above the CAST limit gaγ < 6.6 × 10^-11 GeV^-1 quoted later in the paper, yet the paper does not discuss this tension. A best-fit value in a region excluded by a laboratory experiment requires explanation; at minimum, the 3C 454.3 result should be reported as an upper limit with the CAST constraint overlaid.","section":"V.B and V.C"},{"comment":"The conversion probability depends sensitively on the assumed magnetic field, with Pγ→a ~ (gaγ BT)^2 in the small-mixing regime, but BT = 5 μG, r = 3 kpc, and L = 10r are set by hand without source-specific data. The MCMC uncertainties therefore include only statistical errors and not the dominant systematic uncertainty from the magnetic-field model. The authors should provide a systematic error budget or scan over plausible BT, r, and L values.","section":"III.A-B"},{"comment":"The CLs procedure uses a threshold of χ²/d.o.f > 2.7 for 95% CL, but the justification is not given. For two fitted parameters, the 95% confidence region is usually defined by Δχ² = 5.99 relative to the minimum; the 'half-χ² distribution' in Eq. (18) is nonstandard and needs a derivation or a proper citation. Without this, the exclusion region in Fig. 10 may be incorrect.","section":"IV.B"},{"comment":"The reduced chi-square for B2A is 3.38, which indicates a poor fit even with the ALP contribution. The paper does not report the null-hypothesis chi-square or a likelihood-ratio statistic, so the claim that the ALP scenario fits the high-energy data 'better' is not quantitatively supported.","section":"V.A and Table I"}],"minor_comments":[{"comment":"The selection interval is z ∈ [0.6, 0.8], but 3C 454.3 has z = 0.859; the paper should explain why this source is included outside the stated interval.","section":"III.B"},{"comment":"Uncertainties are given in the text but not in the table; include the 68% intervals for all fitted parameters.","section":"Table I"},{"comment":"Figure 10 caption and text disagree on which line corresponds to which source (B2A vs 3C 454.3); please check and make consistent.","section":"V.C and Fig. 10"},{"comment":"There are numerous typos, including 'sectioin' (Section III.A), 'threotical' (Section III.A), 'F ermi-LAT' (multiple), 'souce's' (Section II.C), '3C 354.3' (Section V.C), and 'confidential level' (Section IV.A).","section":"Throughout"},{"comment":"The notation L = lp - χ²/2 in Eq. (17) should be defined; if lp is the log-prior, this is the log-posterior, but the text calls it the likelihood.","section":"IV.A"},{"comment":"The sentence about not incorporating the term for unabsorbed photons converting to ALP before reception is unclear; 'reception area' should be 'Milky Way', and the statement that B_G = 0.5 μG is 'much less significant' should be quantified.","section":"II.C"}],"recommendation":"major_revision","confidential_remarks":"The paper's abstract overstates the B2A result relative to the authors' own caveat in Section V.A. The central quantitative claims need a corrected statistical treatment of upper limits and an explicit systematic-error budget for the magnetic-field model. The source-selection idea is interesting and may be publishable after substantial revision, but the current form should not be accepted as is."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the headline B2A coupling is not a reliable measurement, and the stress-test note is right. Eq. (16) treats the last seven MAGIC points as ordinary Gaussian measurements, but Section V.A says they are upper limits, and the fit's apparent 'deficit' of the no-ALP model above 0.1 TeV is an artifact of that. The authors themselves only claim an upper threshold, so the abstract overstates a fit.\n\nWhat's genuinely new: these two FSRQs haven't been used for ALP bounds before, and the source-selection method — picking z and energy ranges where the transmission ratio is large, using critical-energy contours — is a reasonable heuristic that could help future surveys. The paper also gives a clean one-sided CLs procedure (Eq. 18) that is appropriate for upper-limit data, and that part of the result may be salvageable.\n\nThe soft spots beyond the stress-test: the two sources' best-fit couplings, 3.05 and 7.40 in units of 1e-11 GeV^-1, are inconsistent with a single ALP. The paper notes this and blames magnetic-field assumptions, but the field BT=5 uG and scales r=3 kpc, L=10r are simply assumed without source-specific data. Since the conversion probability scales roughly as (ga-gamma * BT)^2, that is a real unquantified systematic. The MCMC chi2/dof for B2A is 3.38, which already signals a poor fit.\n\nWho should read this: people working on gamma-ray spectral ALP searches. The CLs exclusion and selection method deserve referee time; the headline MCMC numbers do not. I'd ask for a major revision that either drops the B2A best-fit or reframes it as an upper-limit-oriented scan, and quantifies the magnetic-field systematics. If that happens, the paper could be a useful addition to the crowded ALP-constraints plot.","headline":"The abstract's B2A best-fit is statistically unsupported — the paper's own upper-limit caveat undercuts it — but the CLs exclusion and source-selection heuristic are worth a look.","tokens_in":14540,"tokens_out":2127,"would_cite":false,"duration_ms":18808,"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":"The paper reports that axion-photon oscillation improves the fit to very-high-energy gamma-ray spectra of two distant blazars, giving couplings near 1e-11 GeV^-1 and masses near 5e-8 eV.","keywords":["axion-like particles","ALP-photon oscillation","very-high-energy gamma rays","flat-spectrum radio quasars","Markov chain Monte Carlo","extragalactic background light","magnetic field cellular model","blazar SED fitting"],"falsifier":"Re-fit the B2A SED treating its last seven MAGIC points as upper limits in a censored likelihood instead of the ordinary chi-square, or re-fit either source with an independently measured kpc-scale magnetic field; if the best-fit $g_{a\\gamma}$ leaves the quoted $1\\sigma$ intervals, the reported values are artifacts of those assumptions.","tokens_in":13401,"feed_emoji":"🔭","tokens_out":16407,"duration_ms":126535,"temperature":0.7,"pith_summary":"The paper aims to show that the very-high-energy $\\gamma$-ray spectra of two distant flat-spectrum radio quasars, B2 2234+28A at $z=0.790$ and 3C 454.3 at $z=0.859$, carry a measurable imprint of axion-like particles. It combines Fermi-LAT and MAGIC data with a cellular model for magnetic fields and Markov chain Monte Carlo fitting, claiming that the ALP scenario fits the high-energy SEDs better than the no-ALP null hypothesis. The reported best-fit parameters are $g_{a\\gamma}=3.05^{+0.51}_{-0.31}\\times10^{-11}\\,\\mathrm{GeV}^{-1}$ and $m_a=5.25^{+2.35}_{-2.65}\\times10^{-8}\\,\\mathrm{eV}$ for B2A, and $g_{a\\gamma}=7.40^{+2.65}_{-2.74}\\times10^{-11}\\,\\mathrm{GeV}^{-1}$ and $m_a=5.50^{+1.69}_{-2.17}\\times10^{-8}\\,\\mathrm{eV}$ for 3C4. If the claim stands, these two sources also demonstrate a practical selection rule for finding other AGN that can tighten ALP limits, especially at low mass.","feed_headline":"Blazar spectra point to axion-photon coupling of order 1e-11 GeV^-1","feed_subtitle":"Fits to Fermi-LAT and MAGIC data favor ALP masses near 5e-8 eV and tighten small-mass limits.","key_machinery":"The central object is the photon–axion conversion probability computed under the cellular model of galactic magnetic fields, in which each galaxy is divided into cells of scale $r$ with equal field strength but random directions. The mixing is governed by the matrix entries $\\Delta_{a\\gamma}$, $\\Delta_a$, and $\\Delta_{pl}$, giving the per-cell probability $P_0=(\\Delta_{a\\gamma}r)^2\\,\\sin^2(\\Delta_{\\mathrm{osc}}r/2)/(\\Delta_{\\mathrm{osc}}r/2)^2$ and the total conversion probability $P_{\\gamma\\to a}=\\frac{1}{3}(1-e^{-3P_0L/2r})$. The detected fraction of a source's flux is $P_{\\gamma\\to\\gamma}=P^S_{\\gamma\\to a}P^G_{a\\to\\gamma}+(1-P^S_{\\gamma\\to\\gamma})e^{-\\tau_{\\gamma\\gamma}}$, where the first term is the axion-regeneration channel and $e^{-\\tau_{\\gamma\\gamma}}$ is the EBL absorption. A source-selection criterion uses the critical energy for conversion and demands photon energies near 0.5 TeV from sources at redshift 0.6–0.8, where the ALP transmission gain exceeds a factor of ten.","core_discovery":"The paper's central claim is that the high-energy excess seen in the SEDs of B2 2234+28A and 3C 454.3 is naturally accounted for by axion-photon conversion, with both sources yielding ALP masses near $5\\times10^{-8}$ eV and couplings of a few $10^{-11}$ GeV$^{-1}$. It further claims that the 95% confidence-level exclusion regions derived from these sources improve existing bounds in the small-mass part of the ALP parameter space. The paper presents the results as model-dependent constraints rather than a detection, and explicitly notes that the B2A result should be read as an upper threshold because its high-energy points are MAGIC upper limits.","pith_inferences":["Going beyond the paper: because the conversion probability scales roughly as $(g_{a\\gamma}B_T)^2$, a direct measurement of the kpc-scale magnetic field in either source would rescale the fitted couplings; for a $1\\,\\mu\\mathrm{G}$ field instead of $5\\,\\mu\\mathrm{G}$, the implied coupling would be roughly five times larger.","Going beyond the paper: treating the seven B2A upper limits with a censored likelihood rather than ordinary flux points would test whether the quoted best-fit $g_{a\\gamma}$ survives; the paper itself says the result can only serve as an upper threshold.","Going beyond the paper: the same redshift–energy selection rule could be applied to a population of Fermi-LAT flat-spectrum radio quasars to build a stacked ALP constraint that averages over unknown source magnetic fields.","Going beyond the paper: if the ALP interpretation is correct, the predicted axion regeneration in the Milky Way should imprint a characteristic energy-dependent hardening that could be searched for in the spectra of all $z>0.6$ blazars under a single magnetic-field model."],"forward_implications":["If the fits are right, ALP masses near $5\\times10^{-8}$ eV and couplings of a few $10^{-11}$ GeV$^{-1}$ can explain why very-high-energy photons from $z\\simeq0.8$ blazars reach Earth despite EBL absorption.","The 95% CL exclusion regions from B2A and 3C4 extend existing bounds in the low-mass part of the ALP parameter space, where laboratory and other gamma-ray limits are weaker.","The selection rule (redshift 0.6–0.8, photon energy near 0.5 TeV) gives a concrete recipe for choosing additional AGN targets for ALP searches.","For 3C4, the ALP enhancement dominates the high-energy SED so strongly that the intrinsic cutoff energy $E_c$ is poorly constrained; more VHE detections would separate the two effects.","For B2A, the quoted coupling should be read as an upper threshold because the high-energy data are upper limits; actual detections would turn it into a two-sided measurement."],"supporting_citations":[{"why":"Supplies the photon–axion mixing formalism (mixing matrix and propagation equations) on which the whole analysis is built.","marker":"[13]"},{"why":"Supplies the MAGIC very-high-energy observations of flat-spectrum radio quasars that provide the B2A upper-limit data points.","marker":"[45]"},{"why":"Supplies early Fermi-LAT observations of 3C 454.3 used in the SED fit.","marker":"[46]"},{"why":"Provides the cellular-model formula for photon–axion conversion probability in a randomly oriented magnetic field.","marker":"[57]"},{"why":"Provides the extragalactic background light attenuation model used to compute expected observed fluxes.","marker":"[58]"},{"why":"Supplies the reference ALP parameter point and the half-chi-square CLs method used for the 95% exclusion limits.","marker":"[63]"},{"why":"Gives the critical-energy criterion and the earlier use of AGN spectra for ALP constraints that the source-selection method builds on.","marker":"[41]"},{"why":"Supplies numerical values of the mixing terms and the magnetic-field scales adopted for the sources.","marker":"[56]"},{"why":"Supplies the typical AGN magnetic-field strength and length scales behind the assumed 5 microgauss, 3-kiloparsec cells.","marker":"[64]"}],"fun_headline_variants":["Axion-photon conversion fits blazar excess, hints at 5e-8 eV mass","Blazar SEDs favor 5e-8 eV axion and tighten coupling","AGN gamma-ray spectra tighten axion mass and coupling","Axion mass near 5e-8 eV from blazar gamma-ray fits","Blazar data constrain axion-photon coupling near 1e-11 GeV^-1"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the last seven B2A spectral points can be fitted as ordinary measured fluxes even though they are only upper limits, and that the magnetic field in each source is 5 microgauss with 3-kiloparsec cells; if either is wrong, the fitted coupling shifts.","fun_headline_variants_meta":{"raw":{"variants":["Axion-photon conversion fits blazar excess, hints at 5e-8 eV mass","Blazar SEDs favor 5e-8 eV axion and tighten coupling","AGN gamma-ray spectra tighten axion mass and coupling","Axion mass near 5e-8 eV from blazar gamma-ray fits","Blazar data constrain axion-photon coupling near 1e-11 GeV^-1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001251,"raw_usage":{"total_tokens":5111,"prompt_tokens":913,"completion_tokens":4198,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":529,"completion_tokens_details":{"reasoning_tokens":4091}},"tokens_in":529,"tokens_out":4198,"duration_ms":30010,"temperature":1.0,"reasoning_tokens":4091,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:31:18.784850+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-fit the B2A SED treating its last seven MAGIC points as upper limits in a censored likelihood instead of the ordinary chi-square, or re-fit either source with an independently measured kpc-scale magnetic field; if the best-fit $g_{a\\gamma}$ leaves the quoted $1\\sigma$ intervals, the reported values are artifacts of those assumptions.","supporting_citations":[{"cited_title":"Mixing of the photon with low-mass particles","cited_arxiv_id":null,"evidence_quote":"Supplies the photon–axion mixing formalism (mixing matrix and propagation equations) on which the whole analysis is built."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies early Fermi-LAT observations of 3C 454.3 used in the SED fit."},{"cited_title":"Ef- fects of initial axion production and photon–axion oscillation on type ia supernova dimming","cited_arxiv_id":null,"evidence_quote":"Provides the cellular-model formula for photon–axion conversion probability in a randomly oriented magnetic field."},{"cited_title":"Modeling the extragalactic background light and the cosmic star formation history","cited_arxiv_id":null,"evidence_quote":"Provides the extragalactic background light attenuation model used to compute expected observed fluxes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the reference ALP parameter point and the half-chi-square CLs method used for the 95% exclusion limits."},{"cited_title":"Probing µ ev alps with future lhaaso ob- servations of agn γ-ray spectra","cited_arxiv_id":null,"evidence_quote":"Gives the critical-energy criterion and the earlier use of AGN spectra for ALP constraints that the source-selection method builds on."},{"cited_title":"Axion-photon conversion of grb221009a","cited_arxiv_id":null,"evidence_quote":"Supplies numerical values of the mixing terms and the magnetic-field scales adopted for the sources."},{"cited_title":"Magnetic fields in galaxies","cited_arxiv_id":null,"evidence_quote":"Supplies the typical AGN magnetic-field strength and length scales behind the assumed 5 microgauss, 3-kiloparsec cells."}],"review_version":1}