{"id":"b0c0fc67-5bd5-48b9-82ce-62728d07ff1c","arxiv_id":"2411.18046","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"The authors find strong Bayesian evidence for orbital decay in the QPE source GSN 069 and report a likely T ~ M^0.8 scaling among low-eccentricity QPE sources.","lead":"This paper models the orbits of compact objects crashing through the hot gas disks around supermassive black holes, which produces repeating X-ray flares called quasi-periodic eruptions. It finds evidence that two stable sources are slowly losing orbital energy and suggests these flare sources split into two populations based on how circular their orbits are.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed T_obt ∝ M_bullet^0.8 population relation is not robust: it depends on six sources, two post-hoc exclusions, heterogeneous mass estimates, and no reported slope uncertainty; the correlation may vanish under a proper error treatment.","rationale":"The reader's weakest assumption is the idealized disk geometry. I agree that is a limitation, but it affects mainly interpretation of the two-source fits; the T–M relation is the claim with the least evidential support and is presented as a new population result. It is the single most load-bearing because the abstract and Sec. 4.1 elevate it to a 'two populations' conclusion. The necessary condition for this claim is that the six-point correlation survives a statistically honest treatment. That condition is currently unchecked: no slope uncertainty or significance is reported, exclusions are post hoc, mass measurements are not homogeneous, and two of the six points come from the timing model being proposed. This does not undermine the osculating-trajectory method or the GSN 069 decay evidence, but it should prevent the population relation from being used as a constraint on EMRI formation. The verdict should remain CONDITIONAL; the paper's main methodological and single-source results stand, but the population claim needs to be downgraded to a tentative trend unless the proposed test supports it. Because the reader's detailed rationale already flags the population slope, my concern partially overlaps the reader's overall assessment, though it is not the same as the stated weakest assumption.","tokens_in":22787,"tokens_out":6534,"duration_ms":62889,"concrete_test":"Recompute Fig. 3 with a Bayesian log-log linear regression including intrinsic scatter and measurement errors, using all 9 sources (no eccentricity cut) and with AT2019qiz's four mass estimates marginalized into one posterior rather than plotted four times; report the 95% credible interval of the slope and the Bayes factor for a nonzero slope. Then repeat excluding GSN 069 and eRO-QPE2 masses (i.e., using only external M-σ/TDE-disk masses) and with RX J1301 and eRO-QPE4 included. If the slope credible interval contains 0 in either variant, the claimed T_obt ∝ M^0.8 relation is not established by the current data.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central population claim (Abstract; Sec. 4.1, Fig. 3) — T_obt ∝ M_bullet^n with n≈0.8 for low-eccentricity QPEs — is load-bearing for the paper's \"two populations / multiple formation channels\" conclusion, but it is the least secure part of the analysis. The support is six sources after excluding RX J1301 and eRO-QPE4 because they are high-eccentricity; eccentricity is measured with the same timing model, so the cut is not independent of the plotted variables. The slope is quoted without uncertainty; individual masses come from heterogeneous estimators: M-σ scaling relations with ~0.3–0.5 dex errors, TDE disk modeling, and, for GSN 069 and eRO-QPE2, the present timing model, which already incorporates the M-σ prior. Several periods are proxies 2⟨Trec⟩ with unquantified errors; AT2019qiz has four correlated mass estimates and Swift J0230 is dashed as an uncertain QPE identification. Previous works using similar diagrams found no clear correlation; the improvement is attributed to two timing-inferred masses, so the apparent n≈0.8 may largely be generated by the new mass constraints and by deleting the two outlying sources. A robust slope and significance test is therefore required before this becomes a claim about EMRI formation channels.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an osculating-trajectory method for evolving EMRI orbits under disk collision losses, and applies it to QPE timing data for GSN 069 and eRO-QPE2, claiming evidence for orbital decay in GSN 069 and evidence for orbital decay or disk alignment in eRO-QPE2. It then collects eight QPE sources and argues for a low-eccentricity population with T_obt proportional to M_bullet^0.8, which it interprets as evidence for two EMRI formation channels. The central technical contribution is a large speed-up in trajectory computation via analytic Kerr geodesic elements with slowly varying constants of motion. The paper is clearly written and the Bayesian framework is appropriate, but the population scaling relation and the conditional nature of the eRO-QPE2 disk constraints need substantial strengthening before the broad claims can be accepted.","tokens_in":23102,"tokens_out":6061,"duration_ms":59537,"significance":"If the GSN 069 timing evidence and the population scaling relation both hold, QPE timing would become a genuinely new probe of low-mass SMBH masses, accretion-disk physics, and EMRI formation channels. The osculating method is a practical contribution: it reduces the cost of long-baseline QPE timing analyses by orders of magnitude and is already used to obtain multi-epoch fits that were previously impractical. The GSN 069 orbital-decay measurement, once carefully calibrated against the stated Jeffreys scale, is a strong result. The eRO-QPE2 analysis is valuable but presently underdetermined: two physically different models are nearly equally favored, so the disk surface density and viscosity constraints derived from one of the two models are conditional. The population relation is interesting but not yet statistically established; the paper should be credited for explicitly acknowledging the external mass prior and the proxy recurrence times, but the scaling claim needs quantitative error treatment.","major_comments":[{"comment":"The paper reports log B_1^0 = 4.4 +/- 0.2 and calls this 'decisive evidence' for a nonzero orbital decay rate. However, the Jeffreys scale defined in Section 2 states that logB in (3.5, 4.6) is 'very strong' and only logB > 4.6 is 'decisive.' The point estimate is therefore one category below the claimed level, and the 2-sigma lower bound is even lower. Please correct the wording throughout (abstract, Section 3.1, and Section 4.1) and, if 'decisive' is needed, justify it by a different evidence criterion or by additional data.","section":"Section 3.1, Eq. (16)"},{"comment":"For eRO-QPE2, the two alternatives H1 (forced orbit plus fast disk alignment) and H2 (forced orbit plus precessing disk) are nearly equally favored, with log B_1^0 = 15.0 and log B_2^0 = 14.4; the posterior quantities such as p, e, and M_bullet are consistent, but the inferred decay rates differ between the models. Equations (35) and (36) then use H1 alone to derive E_col, the QPE radiation efficiency, and by extension the disk surface density and viscosity. These derived quantities should be explicitly labeled as conditional on H1, and ideally the same calculations should be repeated under H2 so that the reader can see the model-dependent range. As written, the paper gives the impression that the disk constraints from eRO-QPE2 are model-independent.","section":"Section 3.2, Eqs. (31)-(36)"},{"comment":"The claimed T_obt proportional to M_bullet^0.8 relation is the main population-level conclusion, but it is not supported by a quantitative statistical analysis. The slope is quoted without an uncertainty; the fit uses six sources with heterogeneous mass estimators, several of which carry 0.3-0.5 dex errors; two of the six masses come from the same timing model with an external M-sigma prior; and the two excluded sources are rejected on the basis of high eccentricity, which is itself measured with the same timing model. Multiple correlated mass estimates are plotted for AT2019qiz and Swift J0230, and Swift J0230 is marked as an uncertain identification. Please report a proper regression with uncertainties on both axes, a slope and scatter with error bars, and a robustness check that excludes the two timing-derived masses and that varies the inclusion/exclusion of the high-eccentricity sources. Without such a test, the two-population and multiple-formation-channel conclusion is not yet load-bearing.","section":"Section 4.1, Fig. 3"},{"comment":"In the forced-trajectory system, the equation for dq_t/dlambda is written as Upsilon_z(E,L,C), while the corresponding geodesic equation (7) has dq_t/dlambda = Upsilon_t(E,L,C). If this is a typographical error, it should be corrected because it appears in the methodological core of the paper. If the implementation actually uses Upsilon_z for the time phase, the computed flare arrival times would be wrong, and the timing results would need to be rechecked.","section":"Section 2.1.2, Eq. (8)"},{"comment":"The disk model is a rigidly precessing slab with a fixed height H = 1.5 M_bullet, and the alignment process is restricted to two extreme cases: immediate alignment after the first observation, or no alignment at all. The eRO-QPE2 degeneracy between H1 and H2 shows how sensitive the interpretation is to these choices. The paper should at least discuss how a warped disk, a continuously varying disk height, or a finite alignment timescale would alter the inferred decay rate and SMBH mass, and why the two extreme cases bracket the likely behavior rather than merely representing convenient limits.","section":"Section 2.2 and Section 3.2"}],"minor_comments":[{"comment":"Aside from the dq_t/dlambda typo noted above, the notation Hdf/Hds for the fast and slow alignment hypotheses is confusing: in Section 3.2, H1 is 'fast alignment' and H2 is 'precessing disk,' but the subscripts df/ds are not defined explicitly in Section 2.2. Please spell out the correspondence in one place.","section":"Section 2.1.2, Eq. (8)"},{"comment":"The statement that QPE timing gives a much tighter SMBH mass constraint than the external M-sigma prior is reasonable, but the paper should clarify in Fig. 3 and in the text that the plotted masses for GSN 069 and eRO-QPE2 are posterior values obtained with the M-sigma prior included, rather than independent measurements.","section":"Section 3.1 and Eq. (19)"},{"comment":"The proxy T_obt = 2 <T_rec> is used for several sources, but the uncertainty in this proxy is not shown or discussed, especially for AT2019vcb where the text says the recurrence time is most uncertain. Please either add error bars for T_obt or state explicitly why the proxy is reliable for the sources included in the fit.","section":"Fig. 3 and Section 4.1"},{"comment":"The comparison with Miniutti et al. (2024) is informative, but the sentence about the 19-day period coincidence would benefit from a brief statement of whether the current model was tested against a model with an additional sinusoidal component, rather than only comparing residuals by eye.","section":"Section 4.2"},{"comment":"The abstract says 'informative constraints on ... the QPE radiation efficiency' for eRO-QPE2, but the efficiency estimate in Eq. (36) has a very wide 95% interval (roughly 9%-45% times (m/M_sun)^-1) and is conditional on H1. Please soften the abstract language accordingly.","section":"Abstract and Section 1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well within the scope of the journal and the osculating method is a real practical advance. The main reason for major revision is that the population scaling relation is presented as a central result without the statistical support needed to sustain it, and the eRO-QPE2 disk constraints are reported as if they were model-independent despite the near-degeneracy between the two favored hypotheses. Both issues are fixable with additional analysis and reframing, so I do not recommend rejection. I would also ask the editor to ensure that the Jeffreys-scale terminology is corrected before publication, since the paper's own scale contradicts its 'decisive' label."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: solid and useful extension of the QPE timing program, worth sending to a good referee, but the T~M^0.8 population scaling is not yet a result you should hang a formation-channel claim on.\n\nWhat is new: an osculating-trajectory implementation for disk-driven secular evolution of QPE EMRIs. The method is borrowed from LISA EMRI waveform work, but the application here is appropriate and the speed gain is real; the reported ~50x for GSN069 is a floor, and the longer eRO-QPE2 baseline would be impractical with direct integration. Applying the method with an added Chandra epoch gives logB1_0=4.4±0.2 for orbital decay in GSN069 against the geodesic equatorial-disk model; that is very strong on the paper's own Jeffreys scale, and the reduced systematic scatter in the H1 residuals supports it. The eRO-QPE2 analysis is handled honestly: the data strongly reject the vanilla model, but H1 and H2 are nearly equally favored, so the authors correctly stop short of claiming they know which disk process is at work.\n\nWhat is soft. First, the paper calls logB=4.4 \"decisive,\" but their stated threshold for decisive is >4.6. That is a labeling error, not a fatal one; very strong evidence for decay is still a real result. Second, the population relation in Fig. 3 is weakly grounded. The slope n~0.8 comes from six sources after excluding two high-eccentricity objects, with no quoted uncertainty on n, heterogeneous mass estimators with 0.3-0.5 dex errors, and several periods used as 2<Trec> proxies. The two timing-inferred masses for GSN069 and eRO-QPE2 already ingest the M-sigma prior that anchors most other points, so those two points are not independent in the way an ideal correlation test requires. Earlier papers using similar diagrams found no clear correlation, so the apparent slope may be driven largely by the new mass constraints and the deletions. A proper likelihood fit treating all mass and period uncertainties, with a stated slope uncertainty and a sensitivity test to the exclusions, is needed before this becomes a claim about multiple EMRI formation channels. Third, the disk model is treated as a flat rigid slab with constant height H=1.5M and either immediate or absent alignment; the derived disk surface density, viscosity, and radiation efficiency inherit that assumption. The authors acknowledge parts of this, but the constraint should be read as model-dependent.\n\nCredit where due: the work uses public data, gives the data processing and priors, and compares explicitly with Miniutti et al. 2024 rather than ignoring the discrepancy. No code is released, which limits immediate reproducibility of the timing pipeline, though the method is described in enough detail to re-implement.\n\nWho it is for: QPE timing analysts, EMRI-disk theorists, and people using low-mass SMBH demographics. It deserves peer review. My recommendation: send it, with a requested major revision focused on the Bayes-factor label and a real statistical treatment of the population correlation. I would cite the timing method and the GSN069 decay evidence, not the T~M^0.8 scaling.","headline":"Strong timing analysis for GSN069 and eRO-QPE2; the T~M^0.8 population claim needs real error treatment before it carries weight.","tokens_in":23669,"tokens_out":4361,"would_cite":true,"duration_ms":40142,"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":"Quasi-periodic eruptions carry a measurable orbital-decay signal that turns X-ray burst timing into a probe of low-mass black hole masses and accretion disks.","keywords":["quasi-periodic eruptions","extreme mass ratio inspirals","orbital decay","disk precession and alignment","supermassive black hole mass scaling","osculating trajectory method","GSN 069","eRO-QPE2"],"falsifier":"Continue X-ray monitoring of eRO-QPE2 through 2024-2025: the fast-alignment hypothesis predicts a constant orbital decay rate after alignment, while the precessing-disk hypothesis predicts $\\dot{T}_{\\rm obt}$ oscillating with a period near $\\tau_p\\sim10$ d, so a measured oscillation or constancy rules one of them out. For GSN 069, search the residual timing after subtracting the best-fit forced-EMRI model for a third period: the model predicts only the orbital period and apsidal precession, so detection of a coherent modulation near 19 d with a consistent phase would favor the rival external-source interpretation.","tokens_in":22522,"feed_emoji":"🕳️","tokens_out":15231,"duration_ms":113759,"temperature":0.7,"pith_summary":"The paper tries to establish that the repeating soft X-ray bursts known as quasi-periodic eruptions are not strictly periodic: the orbiting stellar-mass object (an extreme-mass-ratio inspiral, EMRI) that plows through the accretion disk is slowly losing energy, so its orbital period decays, and the disk itself can precess or align. To capture this, the authors introduce an osculating-trajectory scheme that evaluates the Kerr geodesic analytically and updates the orbit only on the slow decay timescale, accelerating the calculation by orders of magnitude. Applied to GSN 069, the model finds decisive Bayesian evidence for orbital decay and yields a tight SMBH mass, a QPE radiation efficiency near $10\\%\\,(M_\\odot/m)$, and disk surface-density and viscosity estimates; applied to eRO-QPE2 it finds strong evidence for either orbital decay or a fast disk alignment. Across the eight QPE sources with reliable periods and masses, the six low-eccentricity systems follow $T_{\\rm obt}\\propto M_\\bullet^{0.8}$, suggesting two EMRI populations. If correct, timing alone becomes a precision probe of low-mass black holes, disk physics, and EMRI formation channels.","feed_headline":"X-ray burst clocks reveal orbital decay and a black-hole mass law","feed_subtitle":"Modeling the burst timings weighs low-mass black holes and points to two distinct stellar-inspiral populations.","key_machinery":"The load-bearing object is the osculating trajectory method for forced Kerr geodesics. Instead of integrating the equations of motion orbit by orbit, it uses the analytic Mino-time solution of the Kerr geodesic, with orbital elements $p$, $e$, $\\theta_{\\min}$ updated on the long timescale $T_{\\rm obt}/|\\dot{T}_{\\rm obt}|$ over which disk collisions dissipate energy; this is why the calculation is orders of magnitude faster. The disk is modeled as a flat slab of constant height $H=1.5M_\\bullet$ that precesses rigidly at period $\\tau_p$ and either does not align or aligns immediately, and the computed disk-crossing time, with light-travel and Shapiro corrections, is identified as the flare start time.","core_discovery":"On the paper's own terms, the central discovery is that QPE flare timings encode secular orbital evolution, not just a fixed geodesic. Modeling the stellar-mass object (SMO) as a forced EMRI that loses energy each time it crosses a disk, with the disk treated as a flat rigidly precessing slab of height $H=1.5M_\\bullet$, yields decisive evidence for a nonzero orbital decay rate $\\dot{T}_{\\rm obt}\\simeq -6.5\\times10^{-5}$ in GSN 069 (log Bayes factor 4.4 against a free geodesic) and nearly equally strong evidence for eRO-QPE2 under two hypotheses: uniform decay with fast disk alignment (log Bayes factor 15.0) or nonuniform decay modulated by disk precession (log Bayes factor 14.4). The fit gives $\\log_{10}(M_\\bullet/M_\\odot)=5.6\\pm0.1$ from the main orbital-size peak of GSN 069 ($6.1\\pm0.1$ from the minor peak) and about 4.7-4.8 for eRO-QPE2, both much tighter than host-galaxy scaling relations. Combining all available sources, the paper reports a likely correlation $T_{\\rm obt}\\propto M_\\bullet^{0.8}$ among six low-eccentricity QPE EMRIs, with RX J1301 and eRO-QPE4 standing apart as high-eccentricity systems.","pith_inferences":["If the $T_{\\rm obt}\\propto M_\\bullet^{0.8}$ relation survives more QPE discoveries, recurrence times alone could become a rough SMBH mass estimator for low-mass galactic nuclei, independent of host-galaxy scaling relations; this is an extension beyond the paper's claim.","The eRO-QPE2 degeneracy between disk alignment and disk precession might be broken by flare morphology: if alignment is real, the collision geometry changes systematically over the alignment timescale, which should leave a correlated trend in flare durations or spectral temperatures.","The same osculating-trajectory machinery could be carried over to other long-lived EMRI observables, where environmental torques from a disk are usually ignored, potentially changing predicted inspiral rates and phases."],"forward_implications":["QPE timing becomes an independent precision probe of low-mass SMBH masses: GSN 069 is constrained to $\\log_{10}(M_\\bullet/M_\\odot)=5.6\\pm0.1$ (or $6.1\\pm0.1$ for the minor peak), far tighter than the roughly 1 dex $M_\\bullet$-$\\sigma_\\star$ relation.","The measured energy loss per collision, combined with flare energies, gives a QPE radiation efficiency of about $10\\%\\,(M_\\odot/m)$ for GSN 069 and about $24\\%\\,(M_\\odot/m)$ for eRO-QPE2, which is hard to reconcile with a stellar-mass black hole heavier than about $30M_\\odot$.","The near-circular orbits ($e\\sim10^{-2}$, semi-major axis $\\sim10^2M_\\bullet$) favor the wet EMRI formation channel over the dry or Hills channels, and the $T_{\\rm obt}\\propto M_\\bullet^{0.8}$ correlation among low-eccentricity sources points to two distinct QPE populations.","The osculating method cuts the cost of full Bayesian timing fits by orders of magnitude (for example roughly a factor 52 for GSN 069) and allows joint fits of eRO-QPE2 observations spanning about $10^4$ orbital periods, making long-term QPE monitoring tractable."],"supporting_citations":[{"why":"It established the QPE timing model and the EMRI+disk interpretation that this work extends.","marker":"Zhou et al. 2024"},{"why":"It supplies the analytic Kerr geodesic solution in Mino-time phases that the osculating method uses.","marker":"van de Meent 2020"},{"why":"It provides the analytic solution method for Kerr geodesics on which the osculating trajectories are built.","marker":"Fujita & Hikida 2009"},{"why":"It converts the orbital parameters $p$, $e$, $\\theta_{\\min}$ into the integrals of motion used to initialize each trajectory.","marker":"Schmidt 2002"},{"why":"It reports the XMM-Newton observations of eRO-QPE2 showing nonuniform orbital decay that motivates the disk alignment and precession hypotheses.","marker":"Arcodia et al. 2024a"},{"why":"It supplies GSN 069 X-ray light curves and flare energy estimates used to derive the QPE radiation efficiency.","marker":"Miniutti et al. 2023b"},{"why":"It provides the host-galaxy SMBH mass measurements that enter the likelihood through the mass term of the total likelihood.","marker":"Wevers et al. 2022"},{"why":"It gives analytic $\\alpha$- and $\\beta$-disk surface density profiles used to convert the measured energy loss into a viscosity estimate.","marker":"Kocsis et al. 2011"},{"why":"It quantifies the stellar-mass black hole collision scenario whose too-low radiation efficiency disfavors a heavy SMO.","marker":"Franchini et al. 2023"},{"why":"It defines the $\\alpha$-disk model used to interpret the inferred surface density and viscosity.","marker":"Shakura & Sunyaev 1973"}],"fun_headline_variants":["QPE burst timing reveals orbital decay and two inspiraling families","Secular QPE evolution yields black hole masses and a mass-period law","QPEs trace orbital decay, weigh central black holes, and split into two classes","How QPE flare timings reveal orbital decay and a black hole mass relation","Two QPE inspiraling populations from burst timings and a mass-period law"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The timing model treats the accretion disk as a flat, rigidly precessing slab with a fixed height of $1.5M_\\bullet$ and assumes the disk either stays misaligned or snaps to the equator immediately; if the real disk is warped, changes height, or aligns continuously, the inferred decay rate, precession period, and black hole mass would shift.","fun_headline_variants_meta":{"raw":{"variants":["QPE burst timing reveals orbital decay and two inspiraling families","Secular QPE evolution yields black hole masses and a mass-period law","QPEs trace orbital decay, weigh central black holes, and split into two classes","How QPE flare timings reveal orbital decay and a black hole mass relation","Two QPE inspiraling populations from burst timings and a mass-period law"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001,"raw_usage":{"total_tokens":4308,"prompt_tokens":1097,"completion_tokens":3211,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":713,"completion_tokens_details":{"reasoning_tokens":3112}},"tokens_in":713,"tokens_out":3211,"duration_ms":18626,"temperature":1.0,"reasoning_tokens":3112,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T11:34:56.009926+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Continue X-ray monitoring of eRO-QPE2 through 2024-2025: the fast-alignment hypothesis predicts a constant orbital decay rate after alignment, while the precessing-disk hypothesis predicts $\\dot{T}_{\\rm obt}$ oscillating with a period near $\\tau_p\\sim10$ d, so a measured oscillation or constancy rules one of them out. For GSN 069, search the residual timing after subtracting the best-fit forced-EMRI model for a third period: the model predicts only the orbital period and apsidal precession, so detection of a coherent modulation near 19 d with a consistent phase would favor the rival external-source interpretation.","supporting_citations":[],"review_version":1}