{"id":"2afa586c-4038-47e1-84ae-4ebaca8c5782","arxiv_id":"2501.03333","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"A self-consistent X-ray plus UV analysis of GSN 069 finds a compact, viscously expanding TDE disk whose inferred properties in 2014 and 2018 challenge both disk-instability and orbiter-collision models of quasi-periodic eruptions.","lead":"This paper analyzes Hubble and XMM-Newton data of the tidal disruption event candidate and quasi-periodic eruption source GSN 069, showing its ultraviolet emission is a point source with a disk-like spectrum, and that a compact accretion disk cooled and expanded between 2014 and 2018. The result challenges current models for quasi-periodic eruptions because the disk was already large enough in 2014 to interact with an orbiting body, yet no eruptions were seen.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed 2018 disk expansion may be an artifact: §3.1 finds the 2018 Rout/Rin posterior is flat (lower limit only), and §3.2 does not demonstrate how kerrSED restores a two-sided constraint when the spectral break has moved out of the UV band.","rationale":"The reader's weakest_assumption identifies exactly the issue I consider most load-bearing: the 2018 outer radius may be a lower limit, making the 'expansion' a lower-limit shift rather than a measured physical growth. The paper's own §3.1 states the 2018 Rout/Rin posterior is flat above ≈ 330, and §3.2 does not explain how relativistic effects break this degeneracy. Because the expansion is the cornerstone of the TDE-origin claim and the subsequent QPE-model stress tests, this concern must be resolved. The proposed test—fitting the 2018 epoch alone or fixing R2018out = R2014out—directly determines whether the 1.15 ± 0.02 ratio is data-driven. I agree with the reader's conditional verdict: the paper is valuable and likely correct in its qualitative picture, but the quantitative expansion claim needs clarification or revision. No other concern (e.g., the blanket statement about disk instability models) is as decisive for the central claim, and the point-source identification, variability, and cooling are well supported.","tokens_in":32376,"tokens_out":11906,"duration_ms":115948,"concrete_test":"Run the kerrSED fit on the 2018 epoch alone, with the hyperparameters (spin, inclination, Rin) fixed to the medians from the joint fit, and inspect the marginal posterior of Rout/Rin. If this posterior is flat above 300 (i.e., only a lower limit), the tight R2018/R2014 ratio is not driven by the 2018 data. As an additional check, rerun the full joint fit with R2018out forced equal to R2014out and compare the Bayesian evidence; if Δln Z < 5, the expansion is not required by the data.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that GSN 069's disk expanded between 2014 and 2018, with P(R2018out > R2014out | data) = 1.0, depends entirely on the kerrSED fit returning a tight ratio R2018out/R2014out = 1.15 ± 0.02 (Table 2). But §3.1 explicitly reports that in the Newtonian diskSED fit the 2018 Rout/Rin posterior is 'essentially flat for Rout/Rin ≳ 330', so only a lower limit is obtained. The stated reason is that the disk cools, shifting the spectral break between the ν^4/3 mid-frequency range and the Rayleigh-Jeans tail to frequencies below the sampled UV band. This is a physical degeneracy: when the observed UV frequencies are well above kT_out/h, the integrated disk spectrum is independent of Rout, and no relativistic ray-tracing can recover information that is not in the data. The paper offers no closed-form or numerical demonstration that the kerrSED model changes this. The 1.15 ± 0.02 ratio is therefore suspicious: it could simply reflect that the 2018 posterior is a lower limit and the 2014 value is ≈ 280–800, making the inequality P(R2018out > R2014out) = 1.0 trivially satisfied without a real two-sided constraint. Since the expansion is the key evidence for a viscously expanding TDE disk, this is the most load-bearing risk to the paper's central conclusion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes archival HST UV/optical imaging and STIS UV spectra of GSN 069 together with two epochs of XMM-Newton X-ray spectra, and fits the broad-band data with thin-disk models that include a finite outer radius. The authors decompose the nuclear emission into a dominant point source and a minor stellar component, measure a ~10% FUV flux decline between 2014 and 2018, and show that the X-ray and UV spectra can be described self-consistently by a compact disk with outer radius of order 10^3 Rg. They further argue that the disk cooled and expanded between the two epochs and use the inferred parameters to test quasi-periodic eruption (QPE) models, concluding that published disk-instability models cannot explain the absence of QPEs in 2014 and their presence in 2018, and that orbiter/disk interaction models face a similar timing problem.","tokens_in":32741,"tokens_out":3337,"duration_ms":36827,"significance":"If the central claims hold, this is a valuable and unusually complete multi-wavelength study of a QPE host: it ties the UV excess to the accretion disk, independently recovers host extinction consistent with the Balmer decrement, and turns two epochs of X-ray and UV data into sharp constraints on QPE model parameters. The analysis is carefully executed in its imaging decomposition, extinction treatment, and simultaneous X-ray/UV fitting, and the use of public data with reproducible Bayesian tools is a clear strength. The cooling of the disk is robust and consistent with earlier X-ray-only work, and the comparison of the two epochs against instability and orbiter models is a useful stress test. However, the expansion claim, which is central to the TDE interpretation and to the QPE-model discussion, rests on a 2018 outer-radius constraint that the Newtonian fit itself shows to be only a lower limit; the relativistic fit does not yet demonstrate that it restores a two-sided constraint. The significance of the paper is therefore high if that gap is closed, but the expansion result is currently not fully supported.","major_comments":[{"comment":"The paper does not establish that the kerrSED model recovers a two-sided constraint on the 2018 outer radius, which is required for the expansion claim. Section 3.1 states that in the diskSED fit the 2018 Rout/Rin posterior is 'essentially flat for Rout/Rin ≳ 330', so only a lower limit is obtained. Section 3.2 then reports in Table 2 a tight ratio R2018out/R2014out = 1.15 ± 0.02 and P(R2018out > R2014out | data) = 1.0, but no posterior for Rout/Rin from the kerrSED fit is shown, and no argument is given for how relativistic ray tracing breaks the physical degeneracy when the spectral break has moved below the observed UV band. I ask the authors to present the 2018 Rout/Rin posterior for kerrSED explicitly and to demonstrate, analytically or numerically, that the UV data constrain both sides of the posterior. Without this, the apparent 1.15 ratio may simply reflect that the 2018 posterior is a lower limit, making the inequality P(R2018out > R2014out) = 1.0 trivially satisfied.","section":"§3.1 and §3.2"},{"comment":"The claimed probability P(R2018out > R2014out | data) = 1.0 is quoted as 'overwhelming evidence' for expansion, but this probability is computed from posteriors in which the 2018 parameter may be unconstrained from above. If the 2018 Rout is only a lower limit, the posterior comparison reports a prior-dependent and essentially uninformative probability; the same caveat applies to the ratio R2018out/R2014out in Table 2. The authors should either demonstrate that the full 2018 posterior is proper and two-sided, or soften the expansion conclusion and rephrase the evidence as a one-sided constraint consistent with, but not requiring, expansion.","section":"§3.2, Fig. 6, Table 2"},{"comment":"The conclusion that no disk-instability model can explain stability in 2014 and instability in 2018 depends directly on the 2018 outer-radius and Eddington-ratio values, and through Equation (7) on the assumed p0 scaling. The argument is internally consistent for the quoted parameters, but because the 2018 Rout lower-limit issue feeds into the same parameter set, the strongest version of this conclusion should be presented conditionally on the kerrSED Rout constraint being two-sided. The instability analysis itself is a useful stress test, but its headline claim inherits the uncertainty from the expansion evidence.","section":"§4.2.1"}],"minor_comments":[{"comment":"The phrase 'The reader is refereed to' should read 'The reader is referred to'.","section":"§1"},{"comment":"In the caption of Figure 1, 'F6060W' should be 'F606W'.","section":"§2.1"},{"comment":"The footnote markers and text around 'calSTIS3' and 'HST Geo-Coronal Airglow4' appear as raw footnote placeholders; these should be formatted as proper footnotes.","section":"§2.2"},{"comment":"The sentence 'This bimodal posterior can be improve' should read 'can be improved'.","section":"§3.2"},{"comment":"In the conclusions bullet, 'though not value of the parameter space can be excluded' appears to be a typo for 'though no value of the parameter space can be excluded'.","section":"§5"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is competently executed and the cooling result is solid, but the expansion claim, which is load-bearing for the TDE interpretation and for the QPE-model conclusions, is not yet demonstrated because the 2018 outer radius appears to be a lower limit in the Newtonian fit. The key request is a direct presentation and justification of the 2018 kerrSED Rout posterior; if that posterior is genuinely two-sided, the paper would be a strong accept candidate. If it is not, the expansion and the related QPE-model conclusions need to be substantially reframed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First things first: this is a genuinely useful paper, and the main reason to read it is the first self-consistent joint fit of X-ray and UV spectra of GSN 069. The HST imaging decomposition is careful, and the argument that the point source cannot be a nuclear star cluster is convincing—the Sigma_SFR estimate being two orders of magnitude above anything in MaNGA is a strong point. The ~10% FUV variability is solid, and the cooling of the disk between 2014 and 2018 is robust; P(T2014p > T2018p)=1.0 is well supported. The new inclination constraint (31-63 degrees) is a real step forward, and the bimodal spin solutions are handled honestly. I don't think the models being developed by the same group is a circularity issue—they are physical models applied to new data. Now the soft spot. The headline claim is that the disk expanded, with P(R2018out > R2014out | data)=1.0 and a ratio 1.15 +/- 0.02 in the kerrSED fit. But in Section 3.1 the authors themselves report that the 2018 Rout/Rin posterior is flat for Rout/Rin > 330, i.e., only a lower limit. This is the expected degeneracy when the spectral break moves out of the observed UV band. The jump to a tight two-sided constraint in Section 3.2 is not justified in the text: they don't show how relativistic ray tracing lifts that degeneracy, and no simulation or closed-form argument demonstrates that the 2018 UV data actually pin down the outer edge. If the 2018 constraint is in fact a lower limit, then P(R2018out > R2014out)=1.0 is trivially true and the expansion claim collapses to a limit-shift driven by the 10% cooling. That is the load-bearing piece for the TDE interpretation and for the challenge to QPE models, so it deserves a direct answer. This is fixable. The authors could run a posterior predictive check showing that the kerrSED model's 2018 posterior on Rout/Rin is genuinely two-sided when fit to simulated data, or they could reframe the claim as a lower limit and discuss what follows. Without that, I'd take the expansion claim with a grain of salt, even though I think the TDE-origin conclusion is likely correct on other grounds. One more minor point: the blanket statement that no published disk instability model can explain the 2014 stability and 2018 instability is stronger than what the analysis supports—they test Kaur et al. carefully, but Pan et al. is dismissed for lacking explicit stability criteria, which is fair but not a disproof. A softer phrasing would help. Overall: this deserves a serious referee. It's a solid observational paper with one questionable inference that is addressable. I'd send it to review and ask for the demonstration above.","headline":"Solid multi-wavelength analysis with a genuine soft spot in the claimed disk expansion—worth a serious referee, but the expansion claim needs a direct demonstration.","tokens_in":823,"tokens_out":1771,"would_cite":true,"duration_ms":40927,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Joint X-ray and ultraviolet spectra show that GSN 069 hosts a compact accretion disk that cooled and expanded between 2014 and 2018, evidence for a tidal disruption origin that challenges current quasi-periodic eruption models.","keywords":["Accretion","High energy astrophysics","Supermassive black holes","X-ray transient sources","Time domain astronomy","Quasi-periodic eruptions","Tidal disruption events","GSN 069"],"falsifier":"A measurement of the 2018 outer disk radius that does not rely on the relativistic model—for example, UV/optical photometry that brackets the predicted spectral break, or a time-dependent disk fit—would settle whether $R_{\\rm out}^{2018}$ is truly about 15 percent larger than $R_{\\rm out}^{2014}$ or is instead a lower limit. If the 2018 radius is only a lower limit, the expansion claim weakens to a cooling-driven shift in the SED break.","tokens_in":32144,"feed_emoji":"🔭","tokens_out":9473,"duration_ms":78091,"temperature":0.7,"pith_summary":"GSN 069, the first source known to show quasi-periodic eruptions (QPEs), was quiet in 2014 and erupting in 2018; this paper asks what its accretion disk was doing at both times. Modeling the ultraviolet and X-ray spectra together, the authors show that the nuclear UV light is dominated by a point-like source, not stars, and that the full spectrum is described by a finite, thin accretion disk with an outer radius of order $10^3$ gravitational radii. Between 2014 and 2018 the disk cooled by about 10 percent and expanded by about 15 percent, behavior predicted for a viscously spreading disk formed in a tidal disruption event. The paper then uses the measured disk properties to stress-test QPE models: no published disk-instability model can make the disk stable in 2014 and unstable in 2018, and orbiter/disk collision models struggle because the disk was already large enough to be hit in 2014, when no eruptions were seen.","feed_headline":"GSN 069's disk cooled and expanded before its QPEs began","feed_subtitle":"Joint X-ray and UV fits date a compact tidal-disruption disk and stress-test eruption models.","key_machinery":"The central object is the finite-disk spectral model \\texttt{diskSED} and its relativistic extension \\texttt{kerrSED}: a standard thin-disk spectrum in which the outer radius $R_{\\rm out}$ is a free parameter alongside the peak temperature $T_p$, inner radius, spin, and inclination. \\texttt{kerrSED} adds numerical ray tracing through Kerr spacetime, which breaks the degeneracy between inclination and spin present in the Newtonian limit and lets the paper convert the fitted $R_{\\rm out}/R_{\\rm in}$ into $R_{\\rm out}$ in units of gravitational radii. The UV spectrum carries the argument: the observed $\\nu L_\\nu \\propto \\nu^{4/3}$ mid-frequency disk shape means the data sit near the outer cutoff, so comparing the two epochs tracks how the disk cooled and expanded.","core_discovery":"On the paper's own terms, the discovery is that the broad-band 2014 and 2018 spectra of GSN 069 are jointly described by a color-corrected thin accretion disk with a finite outer radius: $R_{\\rm out}^{2014} = 1208^{+424}_{-250}\\,R_g$ (high-spin mode) and $R_{\\rm out}^{2014} = 4096^{+1548}_{-1280}\\,R_g$ (low-spin mode), with $R_{\\rm out}^{2018}/R_{\\rm out}^{2014} = 1.15 \\pm 0.02$ and $P(R_{\\rm out}^{2018} > R_{\\rm out}^{2014}\\,|\\,{\\rm data}) = 1.0$. The peak disk temperature drops by about 10 percent over the same interval, with $P(T_p^{2014} > T_p^{2018}\\,|\\,{\\rm data}) = 1.0$. This simultaneous cooling and expansion is the signature of a viscously spreading, non-steady accretion flow fed close to the black hole, which the paper reads as strong evidence that the disk formed in a tidal disruption event. The same fits place the disk inclination between about $31^\\circ$ and $63^\\circ$, leave spin bimodal, and give black hole masses of roughly $7.5\\times10^6\\,M_\\odot$ or $1.5\\times10^6\\,M_\\odot$ depending on the spin mode.","pith_inferences":["Inference: If the 2018 $R_{\\rm out}$ is only a lower limit, the expansion ratio $1.15 \\pm 0.02$ may be an artifact of the cooling spectrum pushing the SED break out of the observed band; later ultraviolet-to-optical coverage bracketing the break would directly test this.","Inference: The photon-starvation rescue for orbiter models predicts that 2014 collisions would emit in the extreme ultraviolet near 10 eV; archival EUV or soft-X-ray limits could falsify or support that fine-tuning.","Inference: Applying the same joint X-ray/UV fitting to other QPE sources with multi-epoch data would show whether a stable-then-unstable disk pattern is common to the class.","Inference: A fully time-dependent disk model fit to the whole GSN 069 light curve, not just two epochs, could break the spin-inclination bimodality and independently measure the viscous timescale that sets the expansion rate."],"forward_implications":["The ultraviolet point source in GSN 069 is dominated by the accretion disk, with the host stellar population contributing at most 5 percent of the inner FUV flux.","An outer radius of order $10^3\\,R_g$ is far smaller than a long-lived AGN disk and is consistent with a tidal disruption disk observed years after formation.","The inferred cooling and expansion are the signature of a non-steady disk whose mass is decreasing with time, supporting the TDE interpretation.","No published disk-instability model for QPEs can satisfy both a stable disk in 2014 and an unstable disk in 2018 with the observed eruption recurrence.","Orbiter/disk collision models can accommodate the 2018 eruptions only if the disk surface density evolved so that collisions became photon-starved by 2018, a fine-tuning the paper does not rule out but cannot confirm."],"supporting_citations":[{"why":"Discovered GSN 069's quasi-periodic eruptions and defines the recurrence time and eruption temperature that any QPE model must reproduce.","marker":"Miniutti et al. 2019"},{"why":"Supplies the time-dependent thin-disk solution predicting simultaneous cooling and viscous expansion of a TDE disk.","marker":"Mummery & Balbus 2020"},{"why":"Gives the canonical viscously expanding disk solution $R_{\\rm out} \\propto t^{3/8}$ that the inferred growth is compared with.","marker":"Cannizzo et al. 1990"},{"why":"Provides the diskSED and kerrSED models with finite outer radius, the fitting machinery of the paper.","marker":"Guolo & Mummery 2024"},{"why":"Provides the numerical Kerr ray-tracing algorithm used by kerrSED to incorporate relativistic effects.","marker":"Mummery et al. 2024a"},{"why":"Defines the magnetized-disk instability criterion and recurrence-time relations tested against both epochs.","marker":"Kaur et al. 2023"},{"why":"The orbiter/disk collision model whose $R_{\\rm orb} \\le R_{\\rm out}$ condition is confronted with the measured disk sizes.","marker":"Linial & Metzger 2023"},{"why":"Provides the Balmer-decrement color excess used to correct for host-galaxy extinction and cross-check the SED fit.","marker":"Wevers et al. 2024"},{"why":"Earlier UV spectral decomposition whose power-law slope and stellar interpretation is revised by this paper.","marker":"Sheng et al. 2021"}],"fun_headline_variants":["Tidal disruption disk cools and swells in GSN 069","GSN 069's disk grew and cooled, puzzling QPE models","Expanding, cooling disk in GSN 069 defies QPE theories","Quasi-periodic eruption source shows viscously spreading disk","GSN 069 disk's growth and cooling hint at tidal disruption"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the observed ultraviolet bands actually constrain the outer disk radius in both epochs, so the 2018 disk is measured to be larger rather than merely being unconstrained above a lower limit.","fun_headline_variants_meta":{"raw":{"variants":["Tidal disruption disk cools and swells in GSN 069","GSN 069's disk grew and cooled, puzzling QPE models","Expanding, cooling disk in GSN 069 defies QPE theories","Quasi-periodic eruption source shows viscously spreading disk","GSN 069 disk's growth and cooling hint at tidal disruption"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000953,"raw_usage":{"total_tokens":4220,"prompt_tokens":1253,"completion_tokens":2967,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":869,"completion_tokens_details":{"reasoning_tokens":2874}},"tokens_in":869,"tokens_out":2967,"duration_ms":19166,"temperature":1.0,"reasoning_tokens":2874,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:53:51.393470+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A measurement of the 2018 outer disk radius that does not rely on the relativistic model—for example, UV/optical photometry that brackets the predicted spectral break, or a time-dependent disk fit—would settle whether $R_{\\rm out}^{2018}$ is truly about 15 percent larger than $R_{\\rm out}^{2014}$ or is instead a lower limit. If the 2018 radius is only a lower limit, the expansion claim weakens to a cooling-driven shift in the SED break.","supporting_citations":[],"review_version":1}