{"id":"9205ea59-f2e6-4384-a722-2697b9c44fa2","arxiv_id":"2506.17422","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A 157-object homogeneous Hβ AGN sample is standardizable via the radius-luminosity relation across six cosmological models, yielding a flatter slope than expected.","lead":"The authors tested whether 157 reverberation-mapped active galactic nuclei follow a standard radius-luminosity relation that does not depend on the assumed cosmology. They find the relation is consistent across six cosmological models, supporting the use of such AGNs as distance indicators, though larger samples are needed to confirm this.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed standardizability is not yet established: the sample is dominated by low-redshift repeated epochs of a few AGNs, so the six-model invariance in Table V may reflect weak cosmological sensitivity, not an intrinsic R-L property.","rationale":"The strongest claim is that the sample is standardizable. The paper demonstrates consistency of R-L parameters across six cosmologies, but as the authors themselves note, the AGN data provide weak cosmological constraints ('significantly less restrictive' than H(z)+BAO). I identified the logical consequence: consistency in a low-power test is not evidence for standardizability. The repeated low-z observations compound this by giving maximal weight to the redshift range where cosmology cannot change the luminosity distance. This is not an internal inconsistency in the fitting code—the method is the community-standard one—but a limitation on what the data can establish. The reader's weakest_assumption mentioned both repeated observations and weak sensitivity; I agree partially, emphasizing the weak-sensitivity aspect as the more fundamental issue because it is what makes the Table V test non-informative. A mock injection test directly measures the power of the dataset and would settle whether the claimed invariance is meaningful. Because the central claim is currently unverified rather than contradicted, the appropriate verdict is UNVERDICTED pending this sensitivity check.","tokens_in":36638,"tokens_out":9758,"duration_ms":106693,"concrete_test":"Run a mock-injection test with the same MCMC pipeline and the same observed redshifts, fluxes, and lag uncertainties: generate a synthetic sample from an R-L relation whose intercept is deliberately cosmology-dependent, e.g., β_true = 1.37 + 0.5 z, or shift β by 0.3 dex between low- and high-redshift halves. Repeat the six-model analysis; if the inferred β varies by less than 1σ across models despite the injected non-standardizability, the dataset has insufficient power to prove standardizability. If the analysis instead detects the injected shift, the weak-sensitivity concern is refuted and the paper's claim stands.","verdict_should_be":"UNVERDICTED","load_bearing_attack":"The paper's central claim is that the 157 Hβ RM AGN measurements are standardizable because the R-L parameters γ, β, σ_int vary by only 0.06σ–0.53σ across six cosmological models (Table V). This test has low power. The sample's median redshift is 0.065 and its 84th percentile is only 0.234 (Sec. III A, Fig. 1); at these redshifts the luminosity distance differs by only a few percent among very different cosmologies. Moreover, Table I shows that many of the lowest-redshift entries are repeated multi-epoch observations of the same AGN (NGC5548 appears 13 times, Mrk335 at least 6, Mrk590 4), and these repeated sources are exactly the points for which DL(z) is effectively cosmology-independent. The handful of high-z RM sources (z≈0.2–0.8) that could discriminate cosmologies have large lag uncertainties and cannot anchor the fit. Consequently, the near-invariance of γ and β across models (Table V) is almost guaranteed by the redshift distribution, regardless of whether the true R-L relation is standardizable. If the underlying relation had a redshift- or Eddington-ratio-dependent intercept, this analysis would very likely still return small Δγ and Δβ. The conclusion that the R-L relation can be used as a distance indicator is therefore not supported by the presented test.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript compiles the 157-measurement \"Best\" Hβ reverberation-mapping sample of Wang and Woo (2024) and fits the BLR radius–luminosity relation, log(τ/day) = β + γ log(L5100/10^44 erg/s), simultaneously with the cosmological parameters of six Friedmann models (flat and nonflat ΛCDM, XCDM, and φCDM). Using an MCMC likelihood with an intrinsic-scatter term (Eqs. 10–12), the authors report that the R–L slope, intercept, and intrinsic scatter are nearly invariant across the six models (Table V), interpret this as standardizability, and derive weak cosmological constraints from the AGN sample alone as well as slightly shifted constraints when the AGN data are combined with H(z)+BAO data. They also split the sample by Eddington ratio and find a mild dependence of the R–L parameters on accretion rate (Sec. V.A). The central conclusion is that Hβ RM AGNs are standardizable and can in principle serve as cosmological distance indicators.","tokens_in":36851,"tokens_out":5819,"duration_ms":63206,"significance":"The analysis is careful and transparent in its construction: the priors are explicit (Table II), the likelihood is the standard one used for this type of simultaneous fit, and the comparison of the inferred slope with the Wang and Woo (2024) value of γ = 0.42 ± 0.02 is a useful cross-check. The enlarged homogeneous sample is a genuine observational asset and represents progress over the previous 41-source sample. However, the standardizability test is much weaker than the conclusion requires. The redshift distribution is strongly concentrated at low z (median 0.065, 84th percentile 0.234), and the sample is dominated by repeated epochs of a few nearby AGNs, so the six-model comparison has little chance of detecting cosmology-dependent R–L parameters. The paper's own Eddington-ratio split (Eqs. 14–15) also indicates that the single-power-law model is incomplete. If the R–L parameters drifted with redshift or Eddington ratio, the reported Δγ and Δβ in Table V would likely still be small. The claim that the sample is standardizable therefore needs additional support before Hβ AGNs can be treated as reliable distance indicators.","major_comments":[{"comment":"The standardizability test has low statistical power because of the sample's redshift distribution. With a median redshift of 0.06458 and an 84th percentile of 0.23371, the luminosity distance DL(z) changes by only a few percent among very different cosmological models; indeed, the Hβ-only cosmological constraints in Table IV are mostly one-sided lower limits (for example Ωm0 > 0.319 at 2σ in flat ΛCDM). The near-constancy of γ and β in Table V is therefore almost a foregone conclusion even if the R–L relation were intrinsically cosmology-dependent. I ask the authors to quantify the sensitivity of the test, for example by injecting a redshift-dependent intercept into simulated data generated from the same redshift and lag-error distribution and reporting how often the six-model Δγ/Δβ would exceed the observed values, or by fitting the R–L relation separately for low- and high-redshift subsamples (for instance z < 0.1 and z > 0.1).","section":"Sec. III.A, Fig. 1, Table V"},{"comment":"Repeated observations of a few nearby AGNs are treated as independent data points. NGC5548 appears about thirteen times and Mrk335 at least six times, all at z ≈ 0.02–0.03; these epochs carry almost no cosmological information, and treating them as independent overweights a small number of objects and can bias the R–L fit if the repeated epochs are correlated. I request that the authors report results with one randomly chosen or averaged epoch per object, or adopt a hierarchical model with an object-level random effect, and state the effective number of independent AGNs in the sample.","section":"Table I, Sec. III.A"},{"comment":"The Eddington-ratio split shows that the simple single-power-law R–L relation is not the complete description of the sample: the high-accreting subsample has γ = 0.48 ± 0.04 and β = 1.31 ± 0.03, while the low-accreting subsample has γ = 0.43 ± 0.05 and β = 1.42 ± 0.04. If λEdd correlates with redshift or luminosity, the constant-γ,β fit used in Eqs. (8)–(12) will absorb a systematic, redshift-dependent bias that the six-model invariance test cannot reveal, because the low-redshift points dominate the likelihood. I request a direct test of whether the R–L residuals correlate with redshift and with L5100 after the global fit, together with a discussion of how the lag-quality cuts described in Sec. III.A might select a biased subsample.","section":"Sec. V.A, Eqs. (14)–(15), Fig. 6"}],"minor_comments":[{"comment":"The tabulated L5100 values are computed for a fixed flat ΛCDM model (H0 = 72 km/s/Mpc, Ωm0 = 0.3), whereas Eq. (9) in Sec. IV recomputes L5100 for each cosmological model in the fit; the paper should clarify that the Table I luminosities are only illustrative and are not the values used directly in the likelihood.","section":"Table I note and Sec. IV"},{"comment":"The phrase \"six spatially flat and nonflat cosmological models\" is confusing; there are three flat and three nonflat models, and the wording should be adjusted accordingly.","section":"Abstract"},{"comment":"The statement that the Hβ mono dataset \"favors a currently decelerating cosmological expansion\" is stronger than the constraints warrant, since the posteriors peak at the prior boundary (e.g., Ωm0 > 0.319 in flat ΛCDM); the result is better described as an unconstraining lower limit.","section":"Sec. V, Fig. 4"},{"comment":"The zero-acceleration lines are said to be computed for the third cosmological parameter set to the H(z)+BAO best-fit values only in panels (d) and (f); the caption should either specify the same procedure for all panels or explain why those two panels are special.","section":"Fig. 3 caption"},{"comment":"Reference [72] contains a typo (\"netron stars\" instead of \"neutron stars\").","section":"References"},{"comment":"The total uncertainty in Eq. (12) includes the fitted parameter γ in the term γ²σ²_logF,i; this parameter dependence is standard but should be explicitly noted so that readers do not misinterpret the error budget as fixed.","section":"Eq. (12)"}],"recommendation":"major_revision","confidential_remarks":"The paper is competent and the data compilation is valuable, but the headline claim is stronger than the test can support. I would be willing to accept after the authors either provide a sensitivity or power analysis demonstrating that the six-model comparison can detect plausible non-standardizable R–L evolution, or substantially soften the standardizability claim and reframe the result as consistency with standardizability. The repeated-epoch issue and the Eddington-ratio dependence also need to be addressed quantitatively. The paper fits the journal's scope and should be revisable within a moderate revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know about arXiv:2506.17422. First, it delivers a genuinely bigger and cleaner Hβ RM sample: 157 sources from Wang & Woo with uniform lag analysis, 3.8× the 41-source sample used before, and the empirical R-L slope comes out flatter than 0.5 (0.428 ± 0.025) with a mild dependence on Eddington ratio. Second, the paper's headline claim—that this sample is standardizable—is not established by the test they run. The invariance of γ and β across six cosmologies is nearly guaranteed by the redshift distribution, not by any property of the AGNs.\n\nWhat's good: the sample construction is careful, with host-galaxy contamination handled upstream, quality cuts explicit, and peculiar-velocity corrections. The MCMC treatment is standard, priors are listed, and the simultaneous fit of R-L and cosmological parameters is the right way to avoid circularity. The comparison with the previous 41-source sample is informative: the flatter slope tracks the inclusion of higher-accreting sources, and the Eddington-ratio subsample fits (low- and high-λ_Edd) give slopes of 0.43 and 0.48 with a smaller intrinsic scatter in the high-accreting half. Those are useful empirical results, and the authors are appropriately cautious that the sample's Eddington-ratio range is limited.\n\nWhere it's soft: the standardizability test has very low power. The median redshift is 0.065; the 84th percentile is 0.234. Over that range, luminosity distances among the six models differ by a few percent, so the R-L fit parameters are barely affected by the cosmology. The problem is compounded by treating multi-epoch observations of the same AGN (NGC5548 appears about ten times in Table I) as independent data points; these repeats sit at the lowest redshifts where the cosmological signal is weakest and allow a handful of objects to dominate the fit. The near-constancy of γ, β, and σ_int in Table V therefore does not discriminate between standardizable and non-standardizable R-L relations. A redshift-dependent intercept would survive this test. The Eddington-ratio analysis also leans on λ_Edd computed from L5100 and τ themselves, which the authors acknowledge, but they don't propagate the additional uncertainties from line widths and bolometric corrections. And there is no public code or data table outside the paper, which makes the analysis harder to reuse.\n\nBottom line: this is a real advance in the empirical R-L relation for Hβ AGNs, and the sample will be used. The cosmological standardizability conclusion deserves a serious referee because the dataset is important, but the current test doesn't support it. I'd send it to review with a request to add a sensitivity check—e.g., allow a redshift-dependent intercept, or fit with duplicate sources collapsed—and to tone down the standardizability wording.","headline":"Bigger, cleaner Hβ RM sample gives a flatter R-L slope and a mild Eddington-ratio trend, but the standardizability claim is undercut by a nearly cosmology-insensitive test.","tokens_in":37509,"tokens_out":2976,"would_cite":true,"duration_ms":31427,"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 157-source AGN sample standardizes via the Hβ radius–luminosity relation.","keywords":["active galactic nuclei","reverberation mapping","broad-line region","radius-luminosity relation","standardizable distance indicators","Eddington ratio","Hβ emission line","cosmological parameter constraints"],"falsifier":"Refit the $R$–$L$ relation after keeping only one observation per AGN; if the slope, intercept, or intrinsic scatter shifts by more than the quoted uncertainties, the independent-point assumption fails. Alternatively, split the sample by luminosity at fixed Eddington ratio: a slope change larger than $\\sim0.025$ would indicate that the standardization is not universal.","tokens_in":36361,"feed_emoji":"🔭","tokens_out":8106,"duration_ms":76484,"temperature":0.7,"pith_summary":"The paper assembles a homogeneous sample of 157 Hβ reverberation-mapped active galactic nuclei (AGN)—galaxies whose accreting supermassive black holes power broad emission lines—about 3.8 times larger than the previous high-quality homogeneous sample, and asks whether the broad-line region radius–luminosity ($R$–$L$) relation can standardize them as cosmological distance indicators. Fitting the $R$–$L$ relation and cosmological parameters simultaneously in six flat and nonflat dark-energy models, it finds that the relation's slope, intercept, and intrinsic scatter vary by at most 0.06σ to 0.53σ across models. That near-invariance is the paper's evidence that the sample is standardizable. The inferred slope, $\\gamma = 0.428 \\pm 0.025$ in flat $\\Lambda$CDM, is flatter than the photoionization expectation of 0.5 and flatter than the slope from the earlier 41-source sample, an effect attributed to the larger number of high-accreting sources. A mild dependence of the relation on Eddington ratio is also found, so the authors caution that a yet larger sample spanning a wider range of luminosities and Eddington ratios is needed to confirm the result.","feed_headline":"Radius-luminosity relation turns 157 AGNs into standard candles","feed_subtitle":"Hβ time delays plus 5100 Å luminosities stay stable across six cosmological models, making the sample usable as a distance probe.","key_machinery":"The load-bearing object is the broad-line region radius–luminosity relation, written as $\\log(\\tau_{\\rm H\\beta}/{\\rm day}) = \\beta + \\gamma \\log(L_{5100}/10^{44}\\,{\\rm erg\\,s^{-1}})$, where $\\tau_{\\rm H\\beta}$ is the rest-frame Hβ time delay and $L_{5100}$ is the monochromatic luminosity at 5100 Å derived from the flux and the cosmological luminosity distance. The argument works by fitting $\\beta$, $\\gamma$, and an intrinsic scatter $\\sigma_{\\rm int}$ simultaneously with the parameters of six flat and nonflat $\\Lambda$CDM, XCDM, and $\\phi$CDM models; if the $R$–$L$ parameters are insensitive to which cosmology is assumed, the relation can serve as a standardizable distance indicator without circularity. The likelihood includes $\\sigma_{\\rm int}$ as a free Gaussian scatter added to the measurement errors, and asymmetric lag errors are folded in through the total variance in Eq. (12).","core_discovery":"The central claim is that the Hβ mono reverberation-mapping dataset of 157 measurements is standardizable through the $R$–$L$ relation: when the intercept $\\beta$, slope $\\gamma$, and intrinsic scatter $\\sigma_{\\rm int}$ are fitted simultaneously with the cosmological parameters of six models, the $R$–$L$ parameters stay essentially fixed. In the flat $\\Lambda$CDM model the relation is $\\log(\\tau_{\\rm H\\beta}/{\\rm day}) = 1.368 \\pm 0.022 + (0.428 \\pm 0.025)\\log(L_{5100}/10^{44}\\,{\\rm erg\\,s^{-1}})$ with $\\sigma_{\\rm int}=0.202^{+0.015}_{-0.017}$, and the slope is $2.8$–$2.9\\sigma$ shallower than the simple photoionization value of 0.5. The 1D cosmological constraints from the AGN sample alone agree within $2\\sigma$ with those from Hubble-parameter and baryon-acoustic-oscillation data, with the two nonflat models that other data already disfavor showing the largest shifts. The paper also reports a mild Eddington-ratio dependence: splitting the sample at the median Eddington ratio gives a steeper slope ($\\gamma\\simeq0.48$) and smaller intercept for high-accreting sources, consistent with shortened lags in high-accretion AGNs.","pith_inferences":["The 157 entries include repeated observations of the same AGNs (NGC5548 appears roughly ten times), so the effective number of independent objects is smaller than 157; re-fitting with one epoch per object would test how much of the precision comes from duplicated sources.","If the mild Eddington-ratio dependence correlates with luminosity or redshift, the near-constancy of the $R$–$L$ parameters across cosmologies could be a property of this particular sample rather than of the underlying population; a luminosity-stratified split at fixed Eddington ratio would expose that.","A sample spanning more extreme Eddington ratios could reveal whether the slope for low-accreting sources approaches 0.5 while high-accreting sources continue to pull the global slope below it, possibly requiring a second parameter such as the shape of the ionizing spectral energy distribution."],"forward_implications":["If correct, the 157-source Hβ sample becomes a standardizable distance indicator that can be combined with other probes at intermediate redshifts.","The flatter slope implies that the simple photoionization expectation $R\\propto L^{1/2}$ does not hold for the full AGN population; high-accreting sources systematically have shorter lags.","The small intrinsic scatter ($\\sigma_{\\rm int}\\simeq0.20$ dex) and narrow parameter uncertainties make the sample competitive for future cosmological fits, though the AGN-only constraints remain weaker than those from H(z)+BAO data.","Because the $R$–$L$ parameters shift by less than about $0.5\\sigma$ across models, adding Hβ mono data to H(z)+BAO changes cosmological constraints by only $\\lesssim0.2\\sigma$, so the AGN data are consistent but not yet powerful.","The Eddington-ratio dependence means future samples must control accretion-rate selection to avoid biasing the slope and scatter."],"supporting_citations":[{"why":"Supplies the 157-measurement Best sample, the uniform lag re-analysis, and the lag quality cuts used throughout.","marker":"[90]"},{"why":"Provides the canonical Hβ R–L relation with slope near 0.5 that the paper compares against.","marker":"[74]"},{"why":"The previous 41-source homogeneous sample whose steeper slope and larger scatter are the baseline for the new results.","marker":"[89]"},{"why":"Early evidence that high-Eddington-ratio sources have shortened Hβ lags, motivating the Eddington-ratio analysis.","marker":"[79]"},{"why":"Adds high-redshift reverberation-mapped measurements and discusses how monitoring cadence affects the R–L scatter.","marker":"[82]"},{"why":"Earlier 118-source Hβ standardizability analysis that found about 2σ tension and motivated the homogeneous compilation.","marker":"[84]"},{"why":"Supplies the H(z)+BAO dataset and the likelihood functions used for the joint cosmological fits.","marker":"[9]"},{"why":"Establishes the simultaneous-fit approach for testing whether a correlation is independent of assumed cosmology.","marker":"[55, 56]"}],"fun_headline_variants":["157 Hβ AGNs become standard candles via R-L relation","Standard candles from 157 RM AGNs","Hβ lags standardize 157 AGNs for cosmology","R-L relation yields distance probes from 157 AGNs","Flatter R-L slope lets 157 AGNs join standard candles"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result stands on the assumption that after the quality cuts, one power-law $R$–$L$ relation with a single constant Gaussian intrinsic scatter describes all 157 measurements, and that repeated observations of the same AGN (NGC5548 appears about ten times in the table) can be treated as independent data points.","fun_headline_variants_meta":{"raw":{"variants":["157 Hβ AGNs become standard candles via R-L relation","Standard candles from 157 RM AGNs","Hβ lags standardize 157 AGNs for cosmology","R-L relation yields distance probes from 157 AGNs","Flatter R-L slope lets 157 AGNs join standard candles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000997,"raw_usage":{"total_tokens":4328,"prompt_tokens":1156,"completion_tokens":3172,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":772,"completion_tokens_details":{"reasoning_tokens":3090}},"tokens_in":772,"tokens_out":3172,"duration_ms":22872,"temperature":1.0,"reasoning_tokens":3090,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T19:09:52.140236+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Refit the $R$–$L$ relation after keeping only one observation per AGN; if the slope, intercept, or intrinsic scatter shifts by more than the quoted uncertainties, the independent-point assumption fails. Alternatively, split the sample by luminosity at fixed Eddington ratio: a slope change larger than $\\sim0.025$ would indicate that the standardization is not universal.","supporting_citations":[],"review_version":2}