{"id":"2bd85698-0b61-4246-9134-c691d40544aa","arxiv_id":"2608.13068","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"The apparent energy-dependent frequency shift of the ~6 Hz QPO in GRS 1915+105 is better reproduced by two fixed-frequency Lorentzian components, a QPO and a QPO shoulder, with distinct fractional rms and phase-lag energy spectra.","lead":"Using an AstroSat observation of the black hole binary GRS 1915+105, the authors show that the QPO frequency appears to rise with photon energy in some data segments. They argue the apparent shift is not a real frequency change but the signature of two overlapping variability components with different energy-dependent amplitudes and phase lags.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-component preference rests on an uncalibrated Δχ² between non-nested models; at equal DOF the reported 25.6/43.7 improvements need a null distribution before the central claim is supported.","rationale":"The paper is a careful application of the Méndez et al. (2024) two-component framework, and it is transparent about weaker constraints in Segment 1, where the rms contrast is only 0.8σ and the phase-lag contrast 0.9σ. The central claim, however, depends on Section 3.3's demonstration that two tied Lorentzians outperform one energy-dependent Lorentzian. That model comparison is the only step where 'two components' enters; the subsequent frms and phase-lag contrasts assume the two-component model and therefore cannot, by themselves, validate it. Because the two models have equal degrees of freedom and are non-nested, the reported Δχ² has no known reference distribution, and the strong overlap of the inferred components means a skewed single Lorentzian could plausibly produce the same improvement. The manuscript should provide a Monte Carlo calibration of Δχ² under the single-Lorentzian null, or an independent prediction that distinguishes the two interpretations, before the claim can be regarded as established. The reader's conditional verdict already identifies this as the weakest assumption, and my read does not shift it.","tokens_in":15834,"tokens_out":7123,"duration_ms":83558,"concrete_test":"Use the best-fit single-Lorentzian energy-dependent model for Segment 2 (centroid shifting from 6.59 Hz to 7.27 Hz across the four bands) as the null. Simulate at least 1000 realizations of the four energy-resolved PDSs with the same exposure, binning, dead-time noise, and BBN parameters; fit each realization with both the single-Lorentzian and the two-Lorentzian tied models; and record Δχ². If a substantial fraction of simulations (e.g., more than 5%) yield Δχ² ≥ 43.7, the reported improvement is not significant. Repeat the same procedure for Segment 1 using the threshold 25.6.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.3 reports that replacing the independent single-Lorentzian fits (χ² 471.1/509, 714.7/660) with two Lorentzians whose centroids and FWHMs are tied across energy bands gives equal-DOF Δχ² of 25.6 and 43.7. These models are non-nested, and an equal-DOF Δχ² has no standard χ² null distribution; a single asymmetric or skewed line shape can be mimicked by two overlapping Lorentzians. The inferred components are strongly blended: their separations are 0.68 Hz (Segment 1) and 1.28 Hz (Segment 2), while the shoulder FWHM is roughly 1.2–1.7 Hz, so the two-component shape is not resolved in any single energy band. The later frms and phase-lag contrasts (2.8σ and 3.6σ in Segment 2) are computed after accepting the two-component model, and the phase-lag decomposition additionally assumes the two Lorentzians are mutually incoherent; they therefore do not independently establish the existence of two components. No simulation, bootstrap, or information-criterion calibration is presented to show that Δχ² ≈ 44 is unlikely when the true signal is a single energy-dependent Lorentzian. Without that calibration, the central claim remains conditional.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an energy-resolved timing analysis of the low-frequency quasi-periodic oscillation (QPO) in the black hole X-ray binary GRS 1915+105 using one AstroSat/LAXPC observation (Orbit 02360). Power density spectra are constructed in four energy bands and fitted first with a single Lorentzian QPO whose centroid shifts with energy, and then with a two-Lorentzian model (QPO plus QPO shoulder) in which the centroid frequencies and widths are tied across energy bands. The two-component model yields a lower chi-square at the same number of degrees of freedom in the two segments where a frequency-energy trend is seen. The authors also fit the PDS together with the real and imaginary parts of the cross-spectrum using a constant-phase-lag model, reporting distinct fractional rms amplitude and phase-lag energy spectra for the two components, and conclude that the apparent energy-dependent evolution of the QPO feature may result from more than one variability component. A dynamic PDS analysis is used to argue against temporal frequency drift as the cause of the apparent energy dependence.","tokens_in":16029,"tokens_out":3650,"duration_ms":39362,"significance":"The paper is a careful observational study that applies a recent modeling approach (Méndez et al. 2024) to a new AstroSat/LAXPC observation, extending the two-component QPO interpretation to a different instrument with potentially better high-energy sensitivity. The data reduction follows standard LAXPC procedures, the fits are transparent with chi-square and degrees of freedom reported, and the interpretation is hedged in the abstract. The dynamic PDS analysis is a useful check against a trivial time-drift explanation. However, the statistical case for two components over a single energy-dependent Lorentzian is incomplete: the reported improvements in chi-square are between non-nested models and are not calibrated by simulations, bootstrap, or an information criterion. The later fractional-rms and phase-lag contrasts are derived within the assumed two-component, mutually incoherent framework, so they do not independently establish the existence of two components. The result is plausible and important if it holds, but the current evidence is not yet decisive.","major_comments":[{"comment":"The central claim rests on the comparison between a single energy-dependent Lorentzian and two energy-independent Lorentzians. In Segment 1 the chi-square improves from 471.1 to 445.5 for 509 degrees of freedom, and in Segment 2 from 714.7 to 671 for 660 degrees of freedom, i.e. equal degrees of freedom in each case. Because these models are non-nested, an equal-DOF Δχ² of 25.6 or 43.7 has no standard chi-square null distribution. The two Lorentzians are also strongly blended (separations of 0.68 and 1.28 Hz versus shoulder FWHM of roughly 1.2–1.7 Hz), so a single asymmetric or skewed line shape could plausibly mimic their sum. To support the conclusion that the QPO feature really comprises two components, please provide a calibration of the Δχ² distribution, for example by parametric bootstrap simulations of a single energy-dependent Lorentzian fitted with the two-Lorentzian model, or by using an information criterion / cross-validation on held-out energy bands. Without such a calibration, the superiority of the two-component model is not statistically established.","section":"§3.3 and Abstract"},{"comment":"The 'reproduction' of the observed phase-lag spectrum by the derived model is a self-consistency check rather than an independent validation of the two-component model. The model phase lags are computed from the same fitted cos and sin amplitudes that were used to model the real and imaginary parts of the cross-spectrum, under the explicit assumption that the variability consists of mutually incoherent Lorentzian components. Agreement between the derived and observed phase-lag spectra therefore confirms internal consistency of the assumed model class, but it cannot on its own discriminate between the two-component hypothesis and a single oscillator whose phase behavior changes with energy.","section":"§3.5 and Figure 9"},{"comment":"The physical interpretation assumes that the QPO feature consists of two stationary, mutually incoherent Lorentzian oscillators with energy-independent centroid frequencies and widths, and frequency-independent phase lags. The fractional-rms contrast in Figure 6 and the phase-lag contrast in Figure 10 are computed after accepting this decomposition, and they do not independently confirm that two components exist. If the true variability is a single oscillator whose frequency, coherence, or phase behavior genuinely changes with energy, the two-Lorentzian model could still improve chi-square without having physical reality. Please address this alternative explicitly, either by simulation-based model comparison of a single energy-dependent Lorentzian against the two-component model, or by identifying a measurable discriminant that is not constructed from the two-component fit. The current text does not rule out the single-oscillator alternative.","section":"§§3.3–3.5 (modeling assumptions)"},{"comment":"The dynamic PDS analysis used to argue against temporal drift has unquantified sensitivity. With a 38-s segment length and 0.08 Hz frequency resolution, the ability to detect a frequency drift of the magnitude and character that could produce the apparent energy-dependent shift is not demonstrated with injected signals or Monte Carlo simulations. Since this argument is used to support the 'intrinsic rather than time-dependent' statement in §4.1, it would strengthen the paper to report a sensitivity estimate or an upper limit on the drift amplitude that the dynamic PDS can exclude.","section":"§3.2 and Figure 4"}],"minor_comments":[{"comment":"In the text describing Segment 2, '20-40 eV' should read '20-40 keV'.","section":"§3.4"},{"comment":"The observation identification string 'T01 030T01 9000000358' appears with irregular spacing (also in the Introduction); use a consistent formatting convention, preferably with the standard AstroSat proposal/observation syntax.","section":"§2"},{"comment":"The phase-lag formula phi(ν) = tan^{-1}(Im[CS]/Re[CS]) is not numbered; consider numbering it for ease of reference.","section":"§3.5"},{"comment":"Figure axes and captions use inconsistent notation for fractional rms ('frms' in the text, 'RMS' in Figure 6) and for phase lag ('Phase Lag (rad)' vs 'Phase Lag (radians)'); unify these labels.","section":"Figures 2, 6, 9, 10"},{"comment":"The text alternates between 'PDS' and 'PDSs' inconsistently (e.g., 'the PDS are generated' in §2); this is a stylistic issue but should be made consistent.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The central two-component interpretation is potentially important and the data handling appears sound, but the statistical foundation is currently too weak to support the headline claim. The missing calibration of the Δχ² between non-nested models, and the derived rather than independent nature of the frms/phase-lag contrasts, are the key issues. I would look favorably on a revision that adds a simulation-based null distribution for the Δχ² improvement, or at least an information-criterion comparison, and that softens the language accordingly. No concerns about citation practice or scope; the paper fits an observational timing journal well."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi—\n\nThis is a careful, honest application of the Mendez et al. (2024) two-component framework to an AstroSat/LAXPC observation of GRS 1915+105. What's new is the LAXPC confirmation at higher energies (up to 40 keV) of the QPO-plus-shoulder structure, with clean measurements of the fractional rms and phase-lag energy spectra in two segments. The authors are straightforward about where the evidence is weak: Segment 1 shows only a 0.8σ rms difference and 0.9σ phase-lag separation, and they say so.\n\nThe main statistical gap: the two-Lorentzian model is compared to a single energy-dependent Lorentzian with equal DOF, and the Δχ² of 25.6 and 43.7 are quoted without a significance. These are non-nested models, and no bootstrap or simulation is provided to show how often a single skewed line shape would produce that improvement. Given the heavy blending (separations 0.68 and 1.28 Hz versus FWHM 1.2–1.7 Hz), the decomposition is not unique. The 2.8σ and 3.6σ contrasts in Segment 2 are parameters of the two-component model, so they don't by themselves prove the components exist. They do, however, match the qualitative pattern from Mendez et al., and the phase-lag difference in Segment 2 is a real feature if the model is accepted.\n\nThe dynamic PDS check against temporal drift is reasonable, but its sensitivity is not quantified; that's a minor point.\n\nBottom line: the paper is worth peer review. A referee should request a calibration of the model comparison—simulations or bootstrap under a single-Lorentzian null—and perhaps a short note on the dynamic PDS sensitivity. The central claim is plausible, appropriately hedged in the abstract, and potentially useful as an extension to higher energies. I'd send it back with a request for that analysis rather than desk reject. I wouldn't cite it as proof of the two-component model, but I would cite it as evidence that the phenomenon persists at LAXPC energies.","headline":"Careful, transparent LAXPC extension of the Mendez et al. two-component framework, but the two-component claim needs a calibrated model comparison before it can be taken at face value.","tokens_in":16731,"tokens_out":4179,"would_cite":false,"duration_ms":44564,"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":"The apparent energy-dependent shift of the QPO frequency in GRS 1915+105 is better explained by two overlapping variability components than by a single oscillator whose frequency changes with energy.","keywords":["quasi-periodic oscillations","GRS 1915+105","black hole X-ray binaries","energy-dependent QPO frequency","power density spectra","phase lags","AstroSat LAXPC","Lorentzian decomposition"],"falsifier":"A decisive test would be to generate simulated light curves using a single Lorentzian whose centroid frequency genuinely increases with photon energy and then run the same simultaneous fitting pipeline; if the two-component model still improves $\\chi^2$ there, the statistical preference alone cannot discriminate the interpretations, and one would instead look for direct evidence of two independent oscillators, such as a coherence dip between their frequencies or a flat phase lag within each component.","tokens_in":15526,"feed_emoji":"🔭","tokens_out":11040,"duration_ms":105370,"temperature":0.7,"pith_summary":"This paper tries to establish that the apparent rise of the quasi-periodic oscillation (QPO) frequency with X-ray energy in GRS 1915+105 is not an intrinsic property of a single oscillator. It argues, from AstroSat/LAXPC data in one orbit, that the QPO feature is better described by two closely spaced Lorentzian variability components: a narrow QPO and a broader \"shoulder,\" whose centroid frequencies and widths stay fixed across energy bands while their relative strengths change. The case rests on simultaneous fits to four energy-resolved power spectra, on the absence of temporal frequency drift in dynamic power spectra, and on distinct fractional-rms and phase-lag energy dependences of the two components, most clearly in the second segment. If right, it implies that energy-dependent QPO frequency shifts seen in black hole X-ray binaries can arise from unresolved multiple variability components rather than from energy-dependent accretion geometry.","feed_headline":"Black hole's energy-dependent oscillation may really be two signals","feed_subtitle":"AstroSat data favor two stationary variability components with distinct root-mean-square and phase-lag spectra.","key_machinery":"The central machinery is a simultaneous multi-Lorentzian fit. A Lorentzian is a bell-shaped peak in the power spectrum defined by a centroid frequency, a width (inverse quality factor), and a normalization. The paper ties the centroid frequency and width of each Lorentzian across the four energy-resolved power spectra, allowing only the normalizations (hence fractional rms) to vary; this forces any apparent frequency shift to be explained by changing relative amplitudes rather than by a moving centroid. This is combined with the constant-phase-lag cross-spectrum model, in which each component is internally coherent and mutually incoherent with others, and the real and imaginary parts of the cross-spectrum are fitted as the same Lorentzian profile multiplied by $\\cos\\phi$ and $\\sin\\phi$, with $\\phi$ a frequency-independent phase lag per component. Dynamic power spectra computed in short windows serve as a control, testing whether the apparent shift could be a time-dependent drift rather than an energy effect.","core_discovery":"In the 4.0–40.0 keV band of a single AstroSat/LAXPC orbit, the QPO feature appears as a single peak in each of three light-curve segments, with centroids near 5.64, 6.64, and 4.61 Hz. When power spectra are made in four energy bands, the peak seems to shift upward with energy in Segments 1 and 2, but not in Segment 3. The paper shows that this apparent shift disappears under simultaneous modeling: a model with two Lorentzian components, with centroid frequencies and widths tied across energy bands, fits better than one drifting Lorentzian in both Segment 1 ($\\chi^2$ reduction from 471.1 to 445.5) and Segment 2 (from 714.7 to 671) at equal degrees of freedom. The two components have centroids 5.52 and 6.20 Hz in Segment 1 and 6.66 and 7.94 Hz in Segment 2, and they show distinct fractional-rms energy spectra, with the shoulder exceeding the QPO by $2.8\\sigma$ at 20–40 keV in Segment 2, and distinct phase-lag spectra, with a $3.6\\sigma$ difference in the same band. Dynamic power spectra with 38-second windows show no clear temporal evolution of the frequency, supporting the interpretation that the energy dependence is a projection of two stationary components rather than a time-dependent drift.","pith_inferences":["A direct test beyond this paper would be to re-fit RXTE or Insight-HXMT observations of other black hole X-ray binaries that show energy-dependent QPO frequencies, such as XTE J1550–564 and Swift J1727.8–1613, with the same tied-frequency two-Lorentzian model; if the two-component preference repeats, the interpretation becomes general rather than source-specific.","The model assumes the two components are mutually incoherent with flat phase lags; one could measure the coherence function across the QPO feature, expecting lower coherence between the two component frequencies if they are truly independent oscillators.","A simulation study could inject a single energy-dependent Lorentzian into realistic LAXPC noise and run the identical fitting procedure; if the two-component model still wins, the chi-square improvement alone is not decisive, and the physical discriminator must be the distinct rms and phase-lag energy spectra.","The shoulder's low quality factor and steep rms rise resemble a Type-B-like companion in another source; tracking the shoulder across spectral states or longer observations could reveal whether it is a separate QPO type or a different variability process."],"forward_implications":["Energy-dependent QPO centroid shifts reported in other black hole X-ray binaries should be re-checked with tied-frequency two-component fits, since what looks like a drift may be a changing mix of two stationary oscillators.","The QPO shoulder emerges as a distinct variability component with its own rms and phase-lag spectra, not merely an asymmetric tail of the main QPO, particularly above 20 keV.","The constant-phase-lag model provides a self-consistent description of the phase-lag spectrum around both component frequencies, so no intrinsic energy-dependent frequency is required by these data.","High-energy band coverage is what separates the two components in this observation, meaning future wide-band timing instruments can test this decomposition in other sources."],"supporting_citations":[{"why":"Supplies the AstroSat/LAXPC observation of GRS 1915+105 used here and first reported the apparent QPO frequency rise from about 6.55 to 7.48 Hz with energy.","marker":"Yadav et al. 2016a"},{"why":"Provides the simultaneous PDS plus cross-spectrum fitting method, the constant-phase-lag model, and the QPO-plus-shoulder interpretation that this paper applies to a new dataset.","marker":"Méndez et al. 2024"},{"why":"Documents energy-dependent QPO frequencies in GRS 1915+105 with RXTE, the phenomenon whose single-frequency interpretation this paper challenges.","marker":"Qu et al. 2010"},{"why":"Cited as one of the original justifications for decomposing the QPO feature into a QPO and a shoulder component.","marker":"Belloni et al. 1997"},{"why":"Supports the presence of a QPO shoulder as a separate variability component in the two-component interpretation.","marker":"Jonker et al. 2000"},{"why":"Reports a similar shoulder-like two-component scenario in Swift J1727.8-1613, used as a comparison for the coexistence of Type-C-like and shoulder variability.","marker":"Jin et al. 2026"},{"why":"Proposes differential Lense-Thirring precession as an alternative explanation for energy-dependent QPO frequencies, against which the two-component result is contrasted.","marker":"van den Eijnden et al. 2016"}],"fun_headline_variants":["Black hole QPO 'energy shift' is really two signals","AstroSat reveals black hole QPO as two distinct components","Two signals, not drift, behind black hole's QPO mystery","Black hole's oscillation: one QPO, two Lorentzians","Energy-dependent QPO? AstroSat says it's two features"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the variability really consists of two steady, independent oscillators whose frequencies and widths are the same at every photon energy; if a single oscillator actually changes frequency, coherence, or phase with energy, the better two-component fit could be a modeling artifact.","fun_headline_variants_meta":{"raw":{"variants":["Black hole QPO 'energy shift' is really two signals","AstroSat reveals black hole QPO as two distinct components","Two signals, not drift, behind black hole's QPO mystery","Black hole's oscillation: one QPO, two Lorentzians","Energy-dependent QPO? AstroSat says it's two features"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000945,"raw_usage":{"total_tokens":4097,"prompt_tokens":1067,"completion_tokens":3030,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":683,"completion_tokens_details":{"reasoning_tokens":2942}},"tokens_in":683,"tokens_out":3030,"duration_ms":21609,"temperature":1.0,"reasoning_tokens":2942,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:30:51.823481+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test would be to generate simulated light curves using a single Lorentzian whose centroid frequency genuinely increases with photon energy and then run the same simultaneous fitting pipeline; if the two-component model still improves $\\chi^2$ there, the statistical preference alone cannot discriminate the interpretations, and one would instead look for direct evidence of two independent oscillators, such as a coherence dip between their frequencies or a flat phase lag within each component.","supporting_citations":[],"review_version":1}