{"id":"5c5a2b7d-c6f4-4fe6-ad75-2ec8a0615cd7","arxiv_id":"2506.07212","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"Multi-epoch Swift SED fits of NGC 6814 find the accretion disk inner edge at 58 to 270 gravitational radii, far outside the ISCO, suggesting a non-standard inner disk.","lead":"Using ten years of Swift satellite data, the authors show that the Seyfert galaxy NGC 6814's low X-ray states have different causes in different years, and that its accretion disk appears to be truncated much farther out than standard theory predicts. General readers should watch because the result suggests ordinary AGN may harbor non-standard disk structures, which would revise how accretion and black hole spin are inferred from spectra.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2012 epoch's R_in=270 Rg rests on one effective UV filter with tied mdot; a sparse-coverage bias could invalidate the 'all epochs' claim.","rationale":"The paper is careful and the large-radius signal in 2022 (six UVOT filters) and 2016 (four filters) is internally consistent and tested against alternative continuum models (agnsed, diskbb+nthcomp, and others). I do not see a defect that would overturn those two epochs. However, the abstract and Section 8 claim that 'all epochs are better fit' with R_in much larger than the ISCO, and the 2012 epoch is both the most extreme and the least constrained. The authors themselves flag the one-filter limitation in Section 5.3, and they respond by tying mdot; that makes R_in dependent on the tied value and on other parameters imported from 2022. This is precisely the kind of self-identified limitation that should control the verdict. The fixed f_col=1.7 is a weaker concern: if the true AGN color correction were larger (e.g., 2.2), the inferred R_in would move upward, not toward the ISCO, so the qualitative claim is conservative with respect to f_col. The reader's weakest assumption partly overlaps with this (the 2012 fragility is listed), so I mark partial agreement. The proposed recovery test on 2022 data with 2012-like coverage is a direct way to decide whether the 2012 result is a selection effect or a real physical state.","tokens_in":17919,"tokens_out":12777,"duration_ms":153372,"concrete_test":"Simulate the best-fit 2022 average spectrum (R_in=58 Rg, mdot=0.013, full six-filter UVOT) with realistic noise, then delete all filters except V and UVW1 and re-fit with the exact 2012 procedure: tie mdot to the true value, freeze M, Balmer continuum, and host-galaxy normalization. If the recovered R_in is within ~20% of 58 Rg, the sparse UV coverage is not intrinsically biased and the 2012 value deserves credence; if the recovered R_in shifts toward ~200-300 Rg, the 2012 R_in is an artifact of having only one UV filter and the 'all epochs' claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing premise is that the 2012 epoch independently determines R_in = 270 +/- 30 Rg (Table 2, Section 4.5). This epoch was observed in only V and UVW1, and Section 5.3 states that 'the V band is dominated by host galaxy emission, thus we effectively have only one filter to constrain the disk parameters, for this reason we favour the model with the tied accretion rate.' With one effective UV photometric point, R_in is not measured from a spectral shape; it is mapped one-to-one from the UVW1 flux once M, mdot, the Balmer-continuum normalization, and the host-galaxy template are fixed or tied to the 2022 values. Any epoch-dependent error in those tied components (Balmer continuum strength, C iii]/Mg ii contamination, host-galaxy subtraction) translates directly into R_in. The 2012 value is also the most extreme of the three epochs and drives the Section 7.2 claim that the inner radius moved ~200 Rg in four years. If sparse coverage biases the 2012 measurement upward, the central 'all epochs' conclusion and the moving-radius narrative are no longer supported, leaving only 2022 (R_in=58) and 2016 (R_in=100) as robust large-radius epochs.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"NGC 6814 is observed with Swift UVOT/XRT over 2012, 2016, 2019, 2021, and 2022; the paper combines DCF, PCA, and fractional-variability diagnostics with broadband SED fitting using a custom ntdisk + nthcomp continuum plus warm/cold absorbers, Balmer continuum, host galaxy, and photoionized emitter. The three high-cadence epochs are fit individually and simultaneously; Table 2 reports Rin = 58 +/- 7 Rg (2022), 100 +/- 10 Rg (2016), and 270 +/- 30 Rg (2012), all much larger than the ISCO, with enormous AICc improvements over fixing Rin at the ISCO. The authors conclude that all epochs favor a disk inner radius far beyond the ISCO, possibly indicating a non-standard disk or inflated central structure, while noting the corona is compact and consistent with eclipse and time-delay constraints.","tokens_in":18359,"tokens_out":9716,"duration_ms":92848,"significance":"The result, if correct, is significant for accretion-disk models in type 1 Seyferts because it challenges the usual assumption that the disk extends to the ISCO and ties together observed compact coronae, eclipses, and changing-look behavior. The paper's strengths include the use of publicly available Swift data, the explicit model-independent variability analysis (DCF, PCA, F_var), the simultaneous multi-epoch fitting, and the reported tests of alternate continuum models. The main limitations are that the key ntdisk model is unpublished, fixed model parameters (spin, inclination, f_col) are not varied in a sensitivity study, and the 2012 epoch with the largest Rin is constrained by only one effective UV filter with a tied accretion rate. These are addressable with additional analysis and clearer caveats.","major_comments":[{"comment":"The central claim that Rin >> ISCO in all epochs relies on the custom ntdisk model (Gonzalez et al., in prep) with maximal spin, inclination fixed to 60 degrees, and f_col fixed to 1.7. No derivation, validation, or code release for this model is provided, and no sensitivity analysis shows how Rin changes when spin, inclination, or f_col are varied over reasonable ranges. Because Table 2's Rin values (58, 100, and 270 Rg) and the Section 7.2/8 conclusions are direct outputs of this model, the result is not independently checkable as written. Please supply the ntdisk model description or reference and a systematic grid demonstrating that Rin remains above the ISCO under plausible variations of these fixed parameters.","section":"§4.1, Table 2, §7.2"},{"comment":"The 2012 epoch has the largest Rin (270 +/- 30 Rg in Table 2; 282 +/- 56 Rg in Table 3) but is constrained by only V and UVW1; since Section 5.3 states the V band is host-dominated, the disk parameters are effectively constrained by one UV photometric point. The accretion rate is tied across segments because the free fit produces disk values inconsistent with the low observed flux. With mdot tied and with M, host-galaxy normalization, and Balmer-continuum normalization linked to the 2022 fit, Rin is mapped almost one-to-one from the UVW1 flux rather than from a spectral shape; any epoch-dependent error in host-galaxy subtraction, Balmer continuum, or the assumed UV slope shifts Rin directly. The statistical errors in the tables do not capture these systematics. Please add a systematic test for 2012 (e.g., varying host normalization, Balmer normalization, and alpha_u within plausible ranges and showing Rin remains > ISCO with the 2012 data alone), or explicitly state in Section 8 that the 2012 Rin is not an independent measurement.","section":"§5.3, Table 3"},{"comment":"The assertion in Section 7.2 that the large Rin values are 'robust and independent of the continuum model used' is not supported by reported numbers. The text lists agnsed, diskbb+nthcomp, broken power law+ntdisk, and simpl+ntdisk as tested alternatives, but gives no Rin values, fit statistics, or ΔAICc values for these models. Given that the AICc improvements over fixing the ISCO are enormous (>2000, 5642, and 6059 in Sections 4.2–4.4), it is important to show whether these alternatives also prefer Rin >> ISCO and by how much. Please include a summary table of the alternative-model Rin values and their ΔAICc relative to the ISCO-frozen version.","section":"§4.2–4.4, §7.2"}],"minor_comments":[{"comment":"The text says the PCA input data set uses 'the daily spectra as the input data set for 2016 and 2021'; given the epoch list, this should be 2016 and 2022.","section":"§3.3"},{"comment":"The green right-pointing triangle is labelled '2018' in the caption, but the text and Table 1 list the single observation as 2019; please correct the label.","section":"Figure 2 caption"},{"comment":"The text states 90 and 95 percent significance contours on the DCF plots, while the Figure 1 caption says 90 and 99 percent; please harmonize the significance levels.","section":"Figure 1 caption and §3.1"},{"comment":"The base model expression ends with four 'agauss' components that are not described in the text; the Fe K-alpha line is already included as 'zgauss'. Please clarify what these Gaussians represent or remove the unused terms.","section":"§4.1"},{"comment":"The 2012 errors are the standard deviation of the segmented fits, while the 2022 and 2016 errors are propagated fit uncertainties; these are different quantities and should not be compared as equivalent 1-sigma errors. Please state this distinction explicitly or use a common estimator.","section":"Table 3 caption"},{"comment":"The alpha_ox measurements use alpha_u = -2 measured from the 2022 average for all other epochs, and the 2019/2021 points use a 2022-based count-rate-to-flux conversion; the systematic uncertainty from these assumptions should be propagated into the alpha_ox errors or discussed as a caveat in Section 3.2.","section":"§3.2, Eq. (1)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is suitable in principle for MNRAS; the topic and data are appropriate. The main concern is verifiability: the central model is described only as 'Gonzalez et al. in prep', and the strongest outlier epoch is one-filter. I do not see grounds for rejection, but the authors should make the model description or reference available and add the sensitivity tests described in the major comments. The paper would also benefit from a conclusion statement that the 2012 Rin is model-dependent and not independently measured."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Readable and honest paper. The genuinely new thing is the multi-epoch optical-to-X-ray SED modeling of NGC 6814 with the inner disk radius left free: three epochs of Swift data, daily and segmented fits, and consistent evidence that R_in is larger than the ISCO. The model-independent part (DCF, alpha_ox, PCA) is competently done and sets up the SED fitting nicely. I was also impressed by the transparency: the authors explicitly note that the V band is host-dominated in 2012 and that they effectively have one filter to constrain the disk, and they favor the tied-accretion-rate model for that reason. That is the right kind of caveat to state.\n\nThe soft spots are the ones you'd expect. The ntdisk model is unpublished (in prep by Gonzalez et al.), so the mapping from UVOT fluxes to R_in is hard to verify. The fixed maximal spin, inclination 60°, and f_col=1.7 could all shift R_in. The 2012 epoch is the weakest: with one effective UV point and mdot tied to the average, R_in=270 Rg is essentially a one-to-one translation of UVW1 flux into radius. If the Balmer continuum or host-galaxy subtraction is slightly off for that epoch, the value could move a lot. The stress-test note is right that the 'all epochs' claim and the ~200 Rg movement in four years rest heavily on 2012.\n\nThat said, I would not throw out the large-radius conclusion. For 2022, with all six UVOT filters, R_in=58±7 Rg is robust. For 2016, with four filters, R_in=100±10 Rg is still far from the ISCO. So the non-standard disk claim survives even if 2012 is treated as tentative. The authors should be asked to present 2012 as a tentative measurement, to tone down the moving-radius narrative, and to make ntdisk public or at least describe it in enough detail to be checked.\n\nMy take: this deserves a serious referee. The core result is interesting and the analysis is careful. The conditional verdict is fair. I'd recommend sending it out.","headline":"The large inner disk radius is real for 2022 and 2016, but the 2012 measurement is too thin to carry the 'all epochs' and moving-radius story; still worth publishing with caveats.","tokens_in":18811,"tokens_out":5368,"would_cite":true,"duration_ms":51362,"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":"Using a decade of Swift optical-to-X-ray data, this paper finds that the accretion disk in NGC 6814 is consistently better fit with an inner radius of roughly 60–270 gravitational radii, far beyond the innermost stable circular orbit…","keywords":["galaxies: active","galaxies: nuclei","galaxies: individual: NGC 6814","X-rays: galaxies","accretion disks","spectral energy distribution","AGN variability"],"falsifier":"A decisive test would be a high-signal-to-noise X-ray observation of NGC 6814 that detects blurred reflection or a reverberation lag requiring the reflector to be at the ISCO, which would contradict inner radii of 60–270 $R_g$; alternatively, additional UV filters in a re-observation of the 2012 state that recover $R_{\\rm in}$ near the ISCO would falsify the large-radius claim for that epoch.","tokens_in":17752,"feed_emoji":"🕳️","tokens_out":11721,"duration_ms":101699,"temperature":0.7,"pith_summary":"The paper analyzes ten years of Swift optical-to-X-ray observations of the nearby Seyfert 1.5 galaxy NGC 6814, covering three high-cadence monitoring campaigns in 2012, 2016, and 2022 plus two single-epoch snapshots. Its central finding is that at every epoch, the spectral energy distribution is significantly better described by a standard accretion disk whose inner edge sits far outside the innermost stable circular orbit, at roughly 60, 100, and 270 gravitational radii in 2022, 2016, and 2012 respectively, than by a disk reaching down to the ISCO. If correct, this means the usual assumption that the disk extends to the last stable orbit fails for this AGN, and the inner region may instead be a hot, inflated flow or a truncated disk. The authors also show that the three epochs differ in the cause of X-ray weakness: eclipses in 2016, continuum changes with mild cold absorption in 2012, and largely unabsorbed continuum variability in 2022.","feed_headline":"Accretion disk in NGC 6814 may end far beyond the last stable orbit","feed_subtitle":"SED fits over three epochs put the disk edge 60 to 270 gravitational radii out, far beyond the last stable orbit.","key_machinery":"The load-bearing machinery is the inner radius parameter $R_{\\rm in}$ of the ntdisk model, a custom xspec implementation of the Novikov–Thorne general-relativistic disk temperature profile. ntdisk takes black hole mass, accretion rate, spin, inner and outer disk radii, inclination, a color-temperature correction factor, and the source distance, and it is paired with nthcomp Comptonization for the corona, absorbers, a Balmer continuum, a photoionized emitter, an Fe K-$\\alpha$ line, and a host galaxy template. The argument proceeds by comparing fits with $R_{\\rm in}$ free versus fixed at the ISCO using the corrected Akaike information criterion (AICc), where a difference of 6 is taken as significant; leaving $R_{\\rm in}$ free improves the fit by hundreds to thousands of AICc units in every epoch. In the simultaneous three-epoch fit, the black hole mass, warm absorber, Balmer continuum, Fe K-$\\alpha$ line, and host galaxy are tied across epochs to constrain the shared parameters.","core_discovery":"On its own terms, the paper's discovery is that the optical-to-X-ray spectral energy distribution of NGC 6814 is, at every observed epoch, significantly better described by an accretion disk whose inner edge is far outside the innermost stable circular orbit. In a simultaneous fit to the 2012, 2016, and 2022 average spectra, the free inner radius converges to $R_{\\rm in}=58\\pm 7\\,R_g$ in 2022, $100\\pm 10\\,R_g$ in 2016, and $270\\pm 30\\,R_g$ in 2012, with an Eddington-scaled accretion rate of about $0.01$–$0.1$ and a black hole mass of $\\log(M_{\\rm BH}/M_\\odot)\\approx 7.66$. Fixing $R_{\\rm in}$ at the ISCO worsens the fit by thousands of AICc units, and the large-radius preference survives across four alternative continuum models (agnsed, diskbb+nthcomp, a broken power law plus ntdisk, and simpl plus ntdisk). The authors interpret the result as evidence for a non-standard accretion disk or an inflated central structure, consistent with the absence of strong blurred reflection and with the measured X-ray/UV lags.","pith_inferences":["Inference: If large inner disk radii turn out to be common in Seyferts, then SED fits that naively assume $R_{\\rm in} = R_{\\rm ISCO}$ may systematically underestimate black hole spin or misestimate accretion rates, so the present result is a caution for the wider population.","Inference: The 2012 measurement of $R_{\\rm in} \\approx 270\\,R_g$ is strongly anti-correlated with the cold absorber column density, so the two may be partially degenerate; a re-analysis with better UV band coverage could break that degeneracy and is a concrete next step.","Inference: An inflated inner structure suggests that a future high-resolution X-ray observation during an eclipse, analogous to the 2016 XMM-Newton campaign, could spatially map the inner disk edge and directly test the truncation geometry.","Inference: The idea that the disk edge moves with changing accretion rate mirrors state-transition behavior seen in X-ray binaries; if the relation between $\\dot{m}$ and $R_{\\rm in}$ in NGC 6814 follows a similar pattern, that would strengthen the analogy between AGN and stellar-mass black holes."],"forward_implications":["If the large inner radii are real, the standard assumption that the disk reaches the ISCO fails for NGC 6814, and the innermost region may be a truncated disk or an inflated flow rather than a standard thin disk.","The measured radii link the SED result to independent data: they are consistent with the lack of strong blurred reflection and with the observed X-ray/UV lags, so the disk-edge inference is not an isolated fit artifact.","The inner edge appears to move by roughly 200 $R_g$ between 2012 and 2022, implying that the truncation radius can change on timescales much shorter than the standard viscous time unless the disk is far thicker than the usual $h/r \\sim 0.01$.","An inflated inner disk provides a natural physical setting for NGC 6814's recurring eclipses and changing-look behavior, potentially connecting the SED finding to the source's known long-term phenomenology.","Continued high-cadence, multi-band UV/optical monitoring could track $R_{\\rm in}$ within a single campaign and test whether the disk edge migrates on dynamical timescales."],"supporting_citations":[{"why":"Sets the color-correction factor f_col = 1.7 and reports the 2022 X-ray/UVW2 lag consistent with a large inner disk radius.","marker":"Gonzalez et al. 2024"},{"why":"Captured the 2016 eclipse, measured the compact corona size (~20 Rg) and 60-degree inclination, and showed the absence of strong blurred reflection that the large-radius result must accommodate.","marker":"Gallo et al. 2021"},{"why":"Provides the general-relativistic disk temperature profile used by the ntdisk model to translate fluxes into radii.","marker":"Novikov & Thorne 1973"},{"why":"Gives the relativistic disk flux profile that ntdisk adopts alongside Novikov–Thorne.","marker":"Page & Thorne 1974"},{"why":"Defines the standard thin-disk paradigm whose ISCO-truncation assumption the paper tests and finds wanting.","marker":"Shakura & Sunyaev 1973"},{"why":"Measured the 2012 X-ray/UVW1 time delay cited as independent support for a large inner disk radius.","marker":"Troyer et al. 2016"},{"why":"Identified additional possible eclipses in 2016 that shape the partial-coverer model and the interpretation of the 2016 variability.","marker":"Pottie et al. 2023"},{"why":"Provides the adopted distance to NGC 6814, a fixed input to the ntdisk SED fits.","marker":"Bentz et al. 2019"}],"fun_headline_variants":["Disk edge in NGC 6814 lies far beyond last stable orbit","NGC 6814's disk inner edge is far outside ISCO","SED fits reveal disk inner radius 60-270 Rg in NGC 6814","Non-standard accretion disk in NGC 6814 inner edge beyond ISCO","Eclipsing AGN NGC 6814: disk ends far from black hole"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the ntdisk model, with spin fixed to maximal, inclination to 60 degrees, and color-correction factor to 1.7, correctly maps the observed UVOT fluxes to an inner disk radius; this is especially fragile for 2012, where only V and UVW1 filters were available, V is host-dominated, and the accretion rate had to be tied across segments because the free fit gave inconsistent values.","fun_headline_variants_meta":{"raw":{"variants":["Disk edge in NGC 6814 lies far beyond last stable orbit","NGC 6814's disk inner edge is far outside ISCO","SED fits reveal disk inner radius 60-270 Rg in NGC 6814","Non-standard accretion disk in NGC 6814 inner edge beyond ISCO","Eclipsing AGN NGC 6814: disk ends far from black hole"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000765,"raw_usage":{"total_tokens":3459,"prompt_tokens":1075,"completion_tokens":2384,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":691,"completion_tokens_details":{"reasoning_tokens":2283}},"tokens_in":691,"tokens_out":2384,"duration_ms":18692,"temperature":1.0,"reasoning_tokens":2283,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T05:39:33.537708+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test would be a high-signal-to-noise X-ray observation of NGC 6814 that detects blurred reflection or a reverberation lag requiring the reflector to be at the ISCO, which would contradict inner radii of 60–270 $R_g$; alternatively, additional UV filters in a re-observation of the 2012 state that recover $R_{\\rm in}$ near the ISCO would falsify the large-radius claim for that epoch.","supporting_citations":[{"cited_title":"G., Gallo L","cited_arxiv_id":null,"evidence_quote":"Sets the color-correction factor f_col = 1.7 and reports the 2022 X-ray/UVW2 lag consistent with a large inner disk radius."},{"cited_title":"M., Bentz M","cited_arxiv_id":null,"evidence_quote":"Measured the 2012 X-ray/UVW1 time delay cited as independent support for a large inner disk radius."},{"cited_title":"C., Gonzalez A","cited_arxiv_id":null,"evidence_quote":"Identified additional possible eclipses in 2016 that shape the partial-coverer model and the interpretation of the 2016 variability."}],"review_version":1}