{"id":"f025cde8-fa0d-4992-a40e-4aa9525a9982","arxiv_id":"2411.08949","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"AT 2021hdr shows repeating 60-90 day brightenings in optical, UV, and X-rays, proposed to be the tidal disruption of a gas cloud by an unresolved binary supermassive black hole.","lead":"Astronomers report a galaxy nucleus that has been flashing every 60 to 90 days for three years in optical, UV, and X-ray light. They propose the flashes are a gas cloud being torn apart by a pair of supermassive black holes, a rare chance to study a merging black hole system.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The derived binary separation and merger time rest entirely on the assumption that the 60-90 d recurrences are strictly periodic with P≈130 d, yet the periodogram in Appendix D.1 is only marginal and quasi-periodicity is admitted, so the binary parameters lack a secure foundation.","rationale":"The reader's verdict identifies the periodicity assumption as the weakest link, and I agree with that assessment. I did not find an additional concern that outweighs it: the exclusion of standard TDEs, QPEs, changing-look AGN, and jet precession is reasonable given the small amplitudes, hard X-ray spectrum, and VLBA non-detection, and the multiwavelength correlated oscillations are a solid observational result. The paper is explicit that the model comparison to Goicovic et al. (2016) is by visual inspection and that the 2024 brightening is not reproduced, so the candidate status is properly caveated. The decisive issue is that all quantitative binary parameters (separation 0.83 mpc, merger time 7e4 yr, cloud mass 0.3-3 M_sun) are obtained from P_b ≈ 130 d, which in turn is obtained by doubling a recurrence interval that the paper's own periodogram does not firmly establish. If the oscillations prove quasi-periodic, the binary interpretation loses its quantitative basis. A focused periodicity test on the existing and continued light curves would settle this, so I would keep the verdict CONDITIONAL rather than raising or lowering it.","tokens_in":28098,"tokens_out":3779,"duration_ms":39186,"concrete_test":"Epoch-fold the ZTF g-band difference flux at the candidate periods 97.7, 130, and 195 d over the full 2018-2024 light curve, and compute false-alarm probabilities from 10^6 Monte Carlo light curves simulated from the best-fit DRW model of Appendix D.2. If the 97.7 d peak has FAP > 5% or if the folded profile loses phase coherence when the post-cutoff 2024-2025 ZTF/Swift data are added, then the strict periodicity assumption fails and the derived binary parameters should be treated as unconstrained.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the mapping from observed brightenings to a binary period. The paper's quantitative conclusions (a ≈ 0.83 mpc, t_merge ≈ 7e4 yr, cloud mass 0.3-3 M_sun) all follow from taking the recurrence time of roughly 60-90 d as half the orbital period, P_b ≈ 130 d (Sect. 5.2). This requires both that the oscillations be strictly periodic and that the correct harmonic interpretation is two accretion peaks per binary orbit. Neither is established. In Appendix D.1, the P4J periodogram of the ZTF data (t > MJD 59500) gives a best period of 97.7 d whose significance the authors describe only as \"could be significant\"; they explicitly state that \"quasi-periodicity cannot be ruled out at this moment.\" The light curve shows only five or six peaks with variable spacing (e.g., Table 1 intervals of roughly 61, 95, 115, and 202 d), and the 2024 rebrightening is not produced by the Goicovic et al. (2016) models that motivate the half-period interpretation. If the recurrences are quasi-periodic or stochastic (which the CARMA fits in Appendix D.2 cannot robustly exclude and the structure function cannot rule out), then P_b is not determined, and a, t_merge, and the cloud mass are unconstrained. The multiwavelength correlated oscillations are a solid observational result; the binary scenario is an interesting candidate, but its quantitative claims rest on the least secure ingredient in the analysis.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"AT 2021hdr is a Seyfert 1 nucleus whose ZTF light curve since late 2021 shows repeated brightening episodes with ~0.2 mag amplitudes on timescales of 60-90 days, with correlated variability in Swift/UVOT and XRT. The paper argues that the source is not a standard TDE, a changing-look AGN, a disk/jet instability, or a QPE-like event, and proposes instead that the behavior results from the tidal disruption of a gas cloud by an unresolved binary supermassive black hole. Assuming the observed recurrence is half the binary orbital period (~130 d) and adopting a virial black hole mass (~4e7 Msun), the authors derive a binary separation of ~0.83 mpc and a gravitational-wave merger time of ~7e4 yr, and infer an initial cloud mass of 0.3-3 Msun. The paper also spectroscopically confirms a companion LINER galaxy 9 kpc away, reporting the galaxy pair as a new merger.","tokens_in":28439,"tokens_out":6568,"duration_ms":55694,"significance":"The observational dataset is a valuable contribution: multiwavelength coverage, forced photometry, VLBA non-detection, and a new companion-galaxy redshift. The proposed scenario, if confirmed, would be the first gas-cloud TDE by a BSMBH and a rare case of a binary in the GW-driven regime, making it of substantial astrophysical interest. However, the quantitative conclusions (binary separation, merger time, cloud mass) are all conditional on a periodicity that the paper itself cannot firmly establish; the periodogram's strongest peak is at 97.7 d and is only 'could be significant', and quasi-periodicity is admitted. The significance of the paper is therefore that of a well-presented candidate with qualitative model support, not a confirmed detection.","major_comments":[{"comment":"The load-bearing assumption that the observed ~60-90 d recurrence is half the binary orbital period is not established. Section 5.2 sets P_b≈130 d and derives a≈0.83 mpc and t_merge≈7e4 yr, but Appendix D.1's P4J periodogram of the same data yields a best period of 97.7 d with significance described only as 'could be significant', and the authors explicitly state that quasi-periodicity cannot be ruled out. Table 1 shows starting-date intervals of ~207, 95, 115, 202, and 61 d between successive brightenings, inconsistent with a strict half-orbit clock. Because all of the binary's quantitative parameters are obtained from P_b via Kepler's law and Peters' formula, the current data do not support the quoted precision of a≈0.83 mpc and t_merge≈7e4 yr. The paper should either provide a rigorous periodicity test (with false-alarm probability and comparison against stochastic models) or present the binary parameters as illustrative values with the dependence on the assumed harmonic explicitly propagated.","section":"5.2, Appendix D.1, Table 1"},{"comment":"The proposed match to the Goicovic et al. (2016) cloud-disruption model is qualitative: the paper relies on 'visual inspection' of their Fig. 4, and explicitly acknowledges that other cloud configurations could also work. Moreover, the 2024 rebrightening is not addressed in the comparison with the Goicovic et al. (2016) models; the paper only notes this problem for the stellar-TDE models of Vigneron et al. (2018). Since the same issue may apply to the cloud-disruption scenario, and since the observed peak spacing is variable rather than the constant half-orbit separation predicted by the model, the current qualitative match is incomplete. A quantitative comparison—e.g., synthesizing light curves from the simulations with the same cadence and comparing peak times, amplitudes, and colors—is necessary before the proposed scenario can be considered more than a plausible candidate.","section":"5.2, Appendix D.3"},{"comment":"The cloud mass estimate is not robust. The authors attribute the full change in bolometric luminosity ΔL≈1.7e44 erg/s between the pre-oscillation (2010) and oscillation (2022-2024) states to accretion of the cloud, using ε≈0.06 and an assumed standard accretion regime. However, the pre- and post-oscillation luminosities are measured in different bands (BAT hard X-ray vs XRT 0.5-10 keV plus optical/UV) and at epochs separated by 12 years; intrinsic AGN variability of the order of the Eddington-ratio change can easily be present. The resulting cloud mass range 0.3-3 Msun should therefore be treated as an order-of-magnitude estimate at best, and the paper should state this limitation explicitly.","section":"5.2"}],"minor_comments":[{"comment":"There is a typo: 'AT 2120hdr' should read 'AT 2021hdr'.","section":"3.2.1"},{"comment":"The phrase 'see Appendix 3.2.2' should cite Section 3.2.2, since the spectral information is presented in the main text, not in an appendix.","section":"3.2.2"},{"comment":"The periodogram finds best periods of 97.7 and 355.7 d, but the paper does not reconcile the 97.7 d peak with the adopted P_b≈130 d in Section 5.2. The authors should explain why the observed recurrence time is preferred over the periodogram peak, or whether 97.7 d corresponds to a different harmonic.","section":"D.1"},{"comment":"Adding the intervals between successive starting dates would help the reader evaluate the periodicity claim; the current table lists only dates, magnitudes, and amplitudes.","section":"Table 1"},{"comment":"The statement that 'the same shape is observed in X-rays and UV' is based on a short overlapping Swift baseline; the paper should note that the X-ray/UV coverage begins only in late 2022, so the first two optical peaks are not covered by the Swift data.","section":"4"},{"comment":"The DHO fit parameters in Table D.1 have extremely large uncertainties and several entries are formatted inconsistently; a cleaner presentation would help support the conclusion that stochastic models fail.","section":"D.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is honest about the periodicity uncertainty, but the abstract and Section 5.2 present quantitative binary parameters without this caveat. If the authors can strengthen the periodicity analysis or reframe the numbers as illustrative, the paper could be acceptable. The stress-test concern about the periodicity is on target and is the main reason for the major revision recommendation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a careful observational paper about a source worth knowing. AT 2021hdr shows repeated optical/UV/X-ray brightenings every ~60-90 days, starting in late 2021, in a Seyfert 1 nucleus. The multiwavelength coverage is real and well-reduced, and the paper reports two genuinely new things: the first spectrum of the companion galaxy, confirming a 9 kpc merger, and a VLBA nondetection that rules out a compact jet. The proposal that this is a gas cloud tidally disrupted by an unresolved binary SMBH is a reasonable candidate, and to my knowledge the first application of the Goicovic et al. (2016) model to a real source. The paper also does something right: it is unusually honest about its own weaknesses. It explicitly says the periodogram peak 'could be significant,' that quasi-periodicity cannot be ruled out, that the CARMA fits do not constrain stochastic models well, that the structure function is consistent with AGN variability, and that the 2024 rebrightening is not reproduced by the simulation. That is the right way to present a candidate. The soft spot is the load-bearing step. The derived binary separation (0.83 mpc), merger time (7e4 yr), and cloud mass (0.3-3 Msun) all follow from assuming the 60-90 day recurrence is strictly periodic and equals half the binary orbital period, giving P ~ 130 days. The periodogram's best period is 97.7 days, not 130, the few observed peaks have uneven spacing, and the 2024 rebrightening does not fit. If the recurrences are quasi-periodic or stochastic, those quantitative claims lose their foundation. The paper acknowledges this, which is why the headline is a candidate rather than a detection, but the mismatch between the adopted 130-day period and the 97.7-day periodogram peak deserves sharper treatment. Also, the model is for a cloud falling onto a bare binary, while this source has a pre-existing accretion disk; the paper mentions this but does not quantify how it changes the expected light curve. None of this kills the paper. The observational phenomenon is solid, the interpretation is plausible enough to test, and the data are public. I would send this to a serious referee and would expect it to be published with the interpretation clearly labeled as a candidate, which the authors already do. For anyone working on nuclear transients or binary SMBHs, this is a useful reference and a good reading-group discussion about how much periodicity a handful of peaks can really establish. I would cite it as a candidate source, not as a confirmed detection. Send it to review.","headline":"A genuinely odd, well-observed nuclear transient, with an interesting binary-SMBH gas-cloud interpretation that the data do not yet firmly support.","tokens_in":29218,"tokens_out":1867,"would_cite":true,"duration_ms":20216,"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":"Repeating brightenings in AT 2021hdr point to a gas cloud torn apart by binary supermassive black holes","keywords":["AT 2021hdr","tidal disruption events","binary supermassive black holes","gas cloud disruption","AGN variability","gravitational-wave-driven binaries","galaxy mergers","multi-wavelength monitoring"],"falsifier":"Observe AT 2021hdr at high cadence in X-rays and UV for several more years and compare every brightening time with a strict 130-day period. If the peaks drift in phase, skip cycles, or the 97.7-day periodogram peak fails to reappear as more data accumulate, the periodicity assumption—and with it the binary separation and merger time—is falsified, even though the gas-cloud disruption idea could still be the right physical picture.","tokens_in":27891,"feed_emoji":"🕳️","tokens_out":13124,"duration_ms":112981,"temperature":0.7,"pith_summary":"The paper argues that AT 2021hdr, a Seyfert nucleus whose light curve began oscillating every roughly 60–90 days in late 2021, is a candidate for the tidal disruption of a gas cloud by an unresolved binary supermassive black hole. The same oscillating shape appears in optical, UV, and X-ray data, and the paper shows that standard explanations—a stellar tidal disruption event, a single changing-look active nucleus, or disk and jet instabilities—do not fit the observations. If correct, this is the first candidate for a gas-cloud tidal disruption by a binary supermassive black hole, with the binary in the gravitational-wave-driven regime: separation about 0.83 milliparsec and merger in about $7\\times10^4$ years. The work also reports, for the first time, that the host galaxy is in an early-stage merger with a companion 9 kiloparsec away, though the oscillations are not related to that companion.","feed_headline":"Binary black holes may be shredding a gas cloud in AT 2021hdr","feed_subtitle":"First candidate gas-cloud tidal disruption by a supermassive black hole pair, merging in about 70,000 years.","key_machinery":"The load-bearing mechanism is the periodic intersection of a gas cloud with a bound pair of black holes: each black hole accretes from the cloud as it passes, producing two accretion-rate peaks per binary orbit. The paper converts the observed peak spacing into a binary period by assuming the spacing is half the orbit, then uses Kepler's third law with a total mass of about $4\\times10^7\\,M_\\odot$ to set the separation at about 0.83 mpc, and the gravitational-wave decay formula for a circular equal-mass binary to set the merger time at about $7\\times10^4$ years. The comparison template is a numerical model in which the disrupted cloud is several times larger than the binary; depending on the cloud's trajectory and impact parameter, the model produces one or two peaks per orbit, and the authors identify the perpendicular-approach geometry with an impact parameter comparable to the binary radius as the configuration closest to the observed light curve.","core_discovery":"The central claim is that AT 2021hdr's oscillating brightenings are the accretion signature of a gas cloud being tidally disrupted by a close binary supermassive black hole. In this picture each black hole crosses the cloud once per binary orbit, so the recurrence time of the peaks is half the orbital period; taking the observed 60–90 day recurrence as half a period of about 130 days, and using the virial black-hole mass of about $4\\times10^7\\,M_\\odot$, the binary separation is about 0.83 mpc. The gravitational-wave decay formula for a circular equal-mass binary then yields a merger time of about $7\\times10^4$ years. The authors argue that standard stellar TDEs, a single binary without a cloud, changing-look AGN, and disk or jet instabilities cannot reproduce the observations, and that the broad shape of the optical–UV–X-ray light curves, including the color behavior and the absence of line-profile changes, matches the gas-cloud disruption simulations. The accreted mass so far would be about $0.1\\,M_\\odot$, implying an initial cloud mass between 0.3 and $3\\,M_\\odot$.","pith_inferences":["A testable extension the paper leaves implicit: if the 130-day period is real, the peak timings should be phase-coherent, so a phase-coherence search on the combined ZTF, Swift, and future light curves is a clean discriminator between periodic and quasi-periodic interpretations.","The same two-peaks-per-orbit accretion pattern should appear whenever a dense cloud falls onto a tight binary, so wide-field surveys with daily cadence should find a population of similar low-amplitude, well-defined-onset oscillating AGN.","If the 2024 rebrightening comes from cloud debris interacting with the pre-existing accretion disk, as the authors suggest, later cycles should show altered amplitudes or damping; dedicated numerical simulations including a pre-existing disk would make this prediction quantitative.","Were quasi-periodicity to win over strict periodicity, the derived binary parameters would lose their basis, but the more general gas-cloud-capture scenario might survive, so the longer monitoring baseline decides between these two levels of the claim."],"forward_implications":["If the scenario is right, AT 2021hdr becomes the first reported candidate for a gas-cloud tidal disruption by a binary supermassive black hole.","The inferred binary is in the gravitational-wave-driven regime and will merge within roughly $7\\times10^4$ years, making it a useful laboratory for studying hierarchical supermassive-black-hole growth.","The model predicts that the oscillating pattern should persist with roughly stable phase at 60–90 day intervals until the cloud is consumed, giving a concrete schedule for continued monitoring.","Because the binary separation is far smaller than the broad-line region, the model naturally accounts for the observed single set of optical lines with no velocity shifts."],"supporting_citations":[{"why":"Supplies the numerical model of gas-cloud disruption by a binary SMBH that predicts two accretion peaks per orbit; the observed light curve is matched to this model.","marker":"Goicovic et al. 2016"},{"why":"Supplies the gravitational-wave orbital-decay formula used to convert the inferred separation into the ~7×10^4 yr merger time.","marker":"Peters 1964"},{"why":"Supplies binary-stellar-TDE simulations that predict periodic troughs on a decaying TDE light curve; used as the comparison the observations are argued to fail.","marker":"Vigneron et al. 2018"},{"why":"Provides the ZTF forced-photometry service data from which the optical oscillating light curve is built.","marker":"Masci et al. 2023"},{"why":"Provides the alert broker that discovered and classified AT 2021hdr and the conversion used for the alert light curve.","marker":"Förster et al. 2021"},{"why":"Supplies the single-epoch virial mass estimator (applied to Hα) used to derive the 4×10^7 M⊙ black-hole mass.","marker":"Reines & Volonteri 2015"},{"why":"Supplies the bolometric correction used to turn the X-ray luminosity into L_bol, from which the accretion rate and cloud mass are estimated.","marker":"Marconi et al. 2004"},{"why":"Supplies the measured mass distribution of molecular cloud cores in the Galactic center, used to argue the inferred 0.3–3 M⊙ cloud is physically plausible.","marker":"Lu et al. 2020"}],"fun_headline_variants":["Binary black holes shred gas cloud in AT 2021hdr","AT 2021hdr: Gas cloud shredded by black hole pair","Binary black hole pair disrupts gas cloud: AT 2021hdr","AT 2021hdr: Black hole binary shreds gas cloud, merge in 70k yrs","First candidate: gas cloud tidal disruption by binary black holes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"Everything about the binary—the 130-day period, the 0.83 mpc separation, and the 70,000-year merger time—rests on the assumption that the 60–90 day brightenings are strictly periodic and equal to half the binary period; the paper's own periodogram analysis finds only a marginal 97.7-day peak and states that quasi-periodicity cannot be ruled out.","fun_headline_variants_meta":{"raw":{"variants":["Binary black holes shred gas cloud in AT 2021hdr","AT 2021hdr: Gas cloud shredded by black hole pair","Binary black hole pair disrupts gas cloud: AT 2021hdr","AT 2021hdr: Black hole binary shreds gas cloud, merge in 70k yrs","First candidate: gas cloud tidal disruption by binary black holes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000626,"raw_usage":{"total_tokens":2999,"prompt_tokens":1147,"completion_tokens":1852,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":763,"completion_tokens_details":{"reasoning_tokens":1752}},"tokens_in":763,"tokens_out":1852,"duration_ms":25775,"temperature":1.0,"reasoning_tokens":1752,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:13:12.588450+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Observe AT 2021hdr at high cadence in X-rays and UV for several more years and compare every brightening time with a strict 130-day period. If the peaks drift in phase, skip cycles, or the 97.7-day periodogram peak fails to reappear as more data accumulate, the periodicity assumption—and with it the binary separation and merger time—is falsified, even though the gas-cloud disruption idea could still be the right physical picture.","supporting_citations":[{"cited_title":"2018, , 476, 5312","cited_arxiv_id":null,"evidence_quote":"Supplies binary-stellar-TDE simulations that predict periodic troughs on a decaying TDE light curve; used as the comparison the observations are argued to fail."}],"review_version":1}