{"id":"1a294d48-f76b-4760-92aa-acb4523e6b2a","arxiv_id":"2507.13884","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A narrow dip in the hard X-ray coherence of Cygnus X-1 at about 0.05 Hz reveals a previously undetected, low-amplitude QPO-like component that appears only in the hard-intermediate state.","lead":"Using AstroSat X-ray data, the authors find a previously unseen narrow dip in the timing coherence of the black hole binary Cygnus X-1 at about 0.05 Hz, present only in a hard-intermediate state with strong jet activity. The finding suggests a hidden, QPO-like variability component in a source long thought to lack quasi-periodic oscillations, offering a new probe of the jet and the Comptonizing corona.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reported 0.05 Hz coherence dip rests on an unvalidated cross-LAXPC estimator; if deadtime/background coupling or estimator bias creates the dip, the central discovery does not hold.","rationale":"I read the paper as attempting to establish a new empirical detection: a narrow drop in the 3-5 vs 6-40 keV coherence at ~0.05 Hz, present in two hard-intermediate observations, with a Lorentzian rms of ~1% that is nearly invisible in the power spectra. For that claim to be true, the cross-detector coherence estimator used to produce the spectra must be measuring source coherence rather than detector coupling. That is the least secure link. Appendix B gives the estimator but no bias or variance derivation; it defers those to a future paper and borrows the Vaughan & Nowak error formula, whose applicability to this averaged ratio estimator is not demonstrated. The 1180 detection is already marginal, so the discovery rests almost entirely on 1210. The fact that the estimator returns unity coherence in the soft-state observations is reassuring but not decisive, because those observations have different count rates and deadtime and do not exercise the same part of the parameter space. I therefore agree with the reader's weakest assumption. An alternative concern about model non-uniqueness (a single partially coherent component could mimic the dip) would affect the hidden-QPO interpretation, not the detection itself, so it is secondary. The proposed single-detector low-frequency coherence check would directly settle the estimator concern. The conditional verdict remains appropriate, so I recommend no change.","tokens_in":19824,"tokens_out":11263,"duration_ms":144549,"concrete_test":"Use LAXPC unit 20 alone for observation 1210 to compute the 3-5 vs 6-40 keV coherence over 0.002-0.3 Hz after subtracting the standard Poisson level, a correction expected to be reliable below ~1 Hz where deadtime effects are small. If the ~0.05 Hz dip persists at comparable depth and frequency, the cross-detector estimator is not responsible for the signal; if it disappears or changes substantially, the central discovery is an artifact of the Appendix B estimator.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The single most load-bearing condition is that the cross-detector Fourier products defined in Appendix B are unbiased estimates of the true coherence. The paper computes coherence from ratios such as |<X1*Y2>|^2 / (Re<X1*X2> Re<Y1*Y2>), averaged with the swapped-detector term, assuming Poisson noise and deadtime in LAXPC units 10 and 20 are uncorrelated. No derivation, bias bound, or simulation of this estimator is provided; the errors are taken from the standard coherence formula (Eq. 9 of Vaughan & Nowak 1997), which was derived for the ordinary single-detector estimator. Appendix B itself defers the formal properties and uncertainties to García et al. (in prep.). If inter-unit deadtime or background variations are slightly correlated, or if the ratio estimator is biased when the two units have different effective areas and background levels, the result is a frequency-dependent feature in the coherence. The weaker 1180 detection (F~3.2, p~6e-3) cannot independently support the claim, so the discovery stands or falls on 1210. This is not an accusation that the dip is artificial; it is a statement that the central result is not yet independently secured because the estimator has not been validated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes ten AstroSat/LAXPC observations of Cygnus X-1 covering the 2016-2017 hard-to-soft transition, using a frequency-segmented multi-Lorentzian fitting technique applied simultaneously to power spectra, cross-spectra, coherence, and phase-lag spectra. It identifies five main variability components whose frequencies and rms amplitudes evolve with spectral state, and additional components in the hard-intermediate (HI) state. The principal claim is the discovery of a narrow dip in the 3-5 keV vs. 6-40 keV coherence function at about 0.05 Hz, most prominently in observation 1210 and more weakly in observation 1180, accompanied by a broad drop in the phase-lag spectrum at the same frequency. The authors interpret this as a previously unseen, QPO-like variability component with rms of about 1% in the subject band, present only in the HI state, and discuss possible links to jet emission and the Comptonizing medium.","tokens_in":20099,"tokens_out":10488,"duration_ms":129673,"significance":"If the detection is robust, the result is a genuinely interesting new timing feature: a coherence dip at energies above 3 keV and at a lower frequency than the previously reported NICER dips, observed during a state with high radio variability and significant hard X-ray polarization. The main detection in observation 1210 is statistically strong (F-statistic ~16.3, p ~ 1.4e-15; ~10 sigma rms in the subject band), and the paper takes care to address deadtime by using cross-detector products between LAXPC units 10 and 20. The frequency-segmented fitting approach is a useful methodological contribution, and the paper explicitly identifies which quantities are fitted and which are predicted. However, the central discovery rests on a cross-detector coherence estimator whose bias and variance properties are not derived or simulated, and the second detection (1180) is marginal. The result is potentially important but not yet independently secured.","major_comments":[{"comment":"The central discovery rests on the cross-detector coherence estimator defined in Appendix B. The paper assumes that deadtime and Poisson noise in LAXPC units 10 and 20 are uncorrelated and therefore eliminated from the cospectrum-based products, but it provides no derivation, bias bound, or simulation for this estimator. The uncertainties are taken from Equation 9 of Vaughan & Nowak (1997), which was derived for the conventional single-detector coherence estimator. If inter-unit deadtime correlations, background differences, or the ratio form of the estimator introduce a frequency-dependent bias, the reported coherence dip at ~0.05 Hz could be spurious. The authors should provide either a formal derivation of the estimator's bias and variance or Monte Carlo simulations with injected known coherence levels and realistic LAXPC deadtime, including different effective areas or background levels between units, demonstrating that no narrow spurious dip is produced.","section":"Appendix B; Section 2"},{"comment":"The phase-lag drop at ~0.05 Hz is described as predicted, but it is not out-of-sample. The real and imaginary parts of the cross spectrum are fitted over the full 0.002-100 Hz range (Eq. A2), and the phase-lag spectrum is then computed from those fitted quantities (Eq. A3). The phase-lag feature at the dip frequency is therefore a deterministic function of the fitted cross-spectrum parameters rather than an independent confirmation. The genuinely out-of-sample quantities are the subject-band power spectrum below 0.3 Hz and the coherence above 0.3 Hz. The authors should reframe the phase-lag agreement as a consistency check, or alternatively fit the model while excluding the narrow-Lorentzian contribution to the cross spectrum and verify that the phase-lag drop is nevertheless predicted.","section":"Section 2; Appendix A"},{"comment":"The second detection in observation 1180 is marginal: the F-statistic is ~3.2 with p ~ 6.2e-3, and the component is not significant in the 3-5 keV reference band (1 sigma). Because the analysis searches over ten observations, multiple Lorentzians, two energy bands, and a range of frequencies, this p-value is unlikely to survive a multiple-trials correction. The paper should explicitly characterize the 1180 feature as tentative, non-independent support rather than a second significant detection, or provide a trial-corrected significance level.","section":"Section 3.3; observation 1180"},{"comment":"The interpretation that the coherence dip requires a new hidden Lorentzian component relies on the model assumption that each additive variability component is perfectly coherent between the two energy bands and has a constant phase lag gi = 2*pi*ki. A frequency-dependent partial coherence of an existing broad component could in principle produce a similar narrow coherence dip without requiring a new additive component. The authors test a partially coherent single-component model only for the soft-state observation 1592 (Appendix D), not for the 0.05 Hz dip in the HI state. A robustness test allowing frequency-dependent phase lags or free partial coherence for the five main components, applied to observation 1210, would substantially strengthen the claim that the dip is caused by a distinct QPO-like component.","section":"Appendix A; Section 4"}],"minor_comments":[{"comment":"The phrase \"the phase lags reaches ~ -0.4 rad\" should be corrected to \"the phase lag reaches\" or \"the phase lags reach.\"","section":"Section 3.3"},{"comment":"The definition of cross-coherence is ambiguous as to whether the two detector permutations are averaged before or after forming the ratio of squared cross-spectrum to the product of cospectra; the exact estimator should be written explicitly (e.g., average of numerators divided by average of denominators, or average of ratios).","section":"Appendix B"},{"comment":"The phrase \"for the first time\" in describing the frequency-segmented fitting approach should be softened, since Mendez et al. (2024) already fit combinations of the six spectra and predict the remaining spectra; the specific segmentation of frequency ranges is the new element, not the general simultaneous-fitting idea.","section":"Section 2"},{"comment":"For the narrow Lorentzian producing the coherence dip, the paper reports only nu_max and rms; reporting the FWHM and quality factor would help readers assess how narrow the proposed component is and how it relates to the observed dip width.","section":"Section 3.3; observation 1210"}],"recommendation":"major_revision","confidential_remarks":"The main detection in obs 1210 is statistically strong, but the discovery claim depends on the cross-LAXPC coherence estimator described in Appendix B, whose formal properties are explicitly deferred to Garcia et al. (in prep.). I recommend requiring estimator validation (derivation or simulations with injected signals and deadtime) before acceptance. The marginal 1180 detection and the overstatement of the phase-lag prediction are also fixable within the manuscript's scope. If the estimator validation reveals inter-unit correlations or bias, the central discovery would not hold; this is a major-revision situation rather than a rejection, because the issue can be addressed with additional analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things worth knowing. The 0.05 Hz coherence dip in Cyg X-1, seen in the 3–5 keV vs 6–40 keV bands, is a genuinely new result that is not in the earlier NICER literature, and the primary detection (obs 1210) is statistically strong (F-test p ~1e-15, ~10 sigma in the subject band). The paper also does some things well: the frequency-segmented fitting is a useful new tool, the CZTI state classification places the feature in the hard-intermediate state without hand-waving, and the spectral decomposition shows the reference band has <5% disk flux, which rules out the soft-band explanation that applied to previous dips.\n\nThe soft spots are real but not fatal. The whole claim rests on the cross-LAXPC coherence estimator in Appendix B. Since Poisson noise and deadtime are assumed uncorrelated between the two units, the estimator should cancel them, but no bias calculation, simulation, or formal uncertainty is given; Appendix B defers to a future paper. If there is residual inter-unit coupling, the dip could be instrumentally produced. This is the main uncertainty.\n\nSecond, the phase-lag drop at the same frequency is presented as a prediction, but it is computed from the fitted real and imaginary parts of the cross spectrum, so it is largely in-sample. The genuinely out-of-sample check is the subject-band power spectrum at 0.002–0.3 Hz, and that is consistent, which is good, but it is not as independent as the text implies.\n\nThird, the second detection (obs 1180) is marginal: F~3.2, p~6e-3, and the component is not significant in the reference band. The paper is honest about this, but the discovery still stands on 1210 alone.\n\nFourth, no alternative model with a single, intrinsically partially coherent component is tested for the dip. The multi-Lorentzian model assumes each component is perfectly coherent across bands, and that assumption is doing real work.\n\nFor the right reader—anyone working on X-ray timing, coherence, QPOs, or jet–disk coupling—this paper is worth a careful look. I would send it to a serious referee rather than desk-reject it. My own recommendation would be a conditional accept: ask the authors to validate the cross-detector estimator with simulations or a derivation, to correct the overstatement about the out-of-sample phase-lag prediction, and to temper the 1180 claim. If the estimator holds up, this is a real discovery.","headline":"A genuinely new 0.05 Hz coherence dip in Cyg X-1, strong in one observation but resting on an unvalidated cross-detector estimator; worth serious review.","tokens_in":20753,"tokens_out":4001,"would_cite":false,"duration_ms":41459,"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":"The paper reports the discovery of a narrow dip in the 3–5 keV versus 6–40 keV coherence function at about 0.05 Hz in Cygnus X-1, identifying a faint quasi-periodic variability component that is nearly invisible in the power spectrum but…","keywords":["Cygnus X-1","coherence dip","hidden quasi-periodic oscillation","X-ray timing","cross spectrum","phase lags","hard-intermediate state","AstroSat LAXPC"],"falsifier":"Re-analyse observation 1210 with the roles of LAXPC units 10 and 20 swapped and with simulated light curves that include detector deadtime but no intrinsic 0.05 Hz signal; if a coherence dip of similar depth appears in the deadtime-only simulations, or if the true uncertainty on $\\gamma^2$ at 0.05 Hz turns out to be large enough to make the observed dip consistent with unity, the discovery would be refuted.","tokens_in":2172,"feed_emoji":"🔭","tokens_out":3473,"duration_ms":90914,"temperature":0.7,"pith_summary":"The paper aims to establish that Cygnus X-1, a black hole binary that has never shown clear quasi-periodic oscillations in its power spectrum, nevertheless contains a narrow, weakly coherent variability component near 0.05 Hz. This component shows up as a sharp dip in the X-ray coherence function and a drop in phase lags, rather than as a peak in the power spectrum. Using ten AstroSat/LAXPC observations covering the source's 2017 hard-to-soft transition, the authors fit a multi-Lorentzian model simultaneously to the power spectra, cross spectra, and low-frequency coherence. The coherence dip appears only in the hard-intermediate state, the same phase in which the source shows its strongest radio variability and a high hard X-ray polarization degree attributed to jet synchrotron emission. If correct, the result shows that coherence spectra can expose variability components that power spectra miss, and it connects such hidden components to jet activity and the geometry of the Comptonizing medium.","feed_headline":"AstroSat finds a hidden 0.05-Hz pulse in Cygnus X-1","feed_subtitle":"A dip in X-ray coherence and phase lags exposes a faint QPO-like signal tied to the jet phase.","key_machinery":"The central object is the set of six Fourier statistics relating two energy bands: the power spectrum in each band, the real and imaginary parts of the cross spectrum, the phase lags, and the coherence function. Under the assumption that each Lorentzian variability component is coherent across bands but incoherent with the other components, the cross spectrum becomes a linear combination of the same Lorentzians with linked normalizations, so phase lags and coherence can be predicted rather than freely fitted. The new procedure is a frequency-segmented simultaneous fit: the subject-band power spectrum is fitted only above 0.3 Hz, the coherence only below 0.3 Hz, and the reference-band power spectrum plus both cross-spectrum parts over the full range, which maximizes sensitivity to weak components that are more visible in coherence than in power. Deadtime is handled by forming all Fourier products as cross statistics between LAXPC units 10 and 20, so uncorrelated Poisson noise and deadtime-induced correlations cancel in the cospectrum.","core_discovery":"Using cross spectra computed between two LAXPC detector units to remove deadtime effects, the authors find that in two hard-intermediate-state observations the coherence between 3–5 keV and 6–40 keV drops from near unity to about 0.8 (observation 1180) and about 0.6 (observation 1210) in a narrow feature at roughly 0.047–0.053 Hz, accompanied by a phase-lag fall to about −0.4 rad. A simultaneous frequency-segmented fit with Lorentzians shows that this feature is produced by a narrow component with an rms amplitude of about 1% in the 6–40 keV band, significant at about 5–10 sigma in the subject band but not in the 3–5 keV reference band: a QPO-like component that is almost invisible in the power spectrum. This is the first coherence dip detected in Cygnus X-1 with both energy bands above 3 keV; earlier dips observed with NICER required a band below 2 keV and appeared at about 1–6 Hz. The authors interpret the dip as interference between the ordinary Comptonizing medium and a transient extra component, possibly Comptonization at the base of the jet.","pith_inferences":["Beyond the paper: if the coherence dip is a genuine oscillation, it may become detectable in the power spectrum with much longer uninterrupted exposures, since a 1% rms component accumulates signal while broadband noise averages down.","Beyond the paper: the same cross-detector cospectrum technique could be applied to other multi-module X-ray instruments to search for hidden low-frequency components in other persistent black hole binaries.","Beyond the paper: an independent check would be to observe the same state with a different instrument or a different pair of LAXPC units and verify that the 0.05 Hz coherence dip persists when the assumed uncorrelated-deadtime hypothesis is varied."],"forward_implications":["If the dip is real, Cygnus X-1 hosts a hidden QPO-like component at about 0.05 Hz that standard power-spectrum searches would miss, so the absence of QPO peaks in the power spectrum does not rule out coherent narrow oscillations.","The component appears only in the hard-intermediate state, where radio variability and hard X-ray polarization peak, empirically tying the feature to jet activity and suggesting the jet base can act as an additional Comptonizing medium.","Because the component is significant only in the 6–40 keV band and its rms increases with energy, any model must place the extra variability in the harder, Comptonized emission rather than in the disk seed photons.","The frequency-segmented cross-spectral fitting method can be applied to other sources and to existing archival data to search for low-frequency signals that are visible in coherence but not in power.","The roughly 0.05 Hz LAXPC dip and the roughly 2 Hz NICER dip, occurring at the same position in the hardness-intensity diagram, imply two hidden QPO-like components at different frequencies, which would argue against a single broadband process."],"supporting_citations":[{"why":"Establishes the multi-Lorentzian cross-spectral fitting method and the equations for predicting phase lags and coherence from the fitted components.","marker":"Méndez et al. 2024"},{"why":"Introduces cross-detector Fourier products to remove deadtime effects, the technique adapted here for all six spectra.","marker":"Bachetti et al. 2015"},{"why":"Defines the coherence function, its interpretation as linear correlation between bands, and the statistical uncertainties used in the fits.","marker":"Vaughan & Nowak 1997"},{"why":"Provides the standard definition of coherence used to build the cross-spectral estimator.","marker":"Bendat & Piersol 2011"},{"why":"Divides Cygnus X-1's accretion states into six modes used to classify each observation, placing the dip in the hard-intermediate state.","marker":"Lubiński et al. 2020"},{"why":"Reports the 23±4% hard X-ray polarization in the hard-intermediate state that links the dip's phase to jet synchrotron emission.","marker":"Chattopadhyay et al. 2024"},{"why":"Reports earlier coherence dips in Cygnus X-1 at about 1–6 Hz with NICER requiring a soft band below 2 keV, the pattern this paper contrasts with.","marker":"König et al. 2024"},{"why":"Shows that overlapping variability components with different amplitudes and lags reduce coherence, the interpretation adopted for the dip.","marker":"Nowak et al. 1999"},{"why":"Provides the earlier RXTE decomposition into the five main Lorentzians that this paper's state-dependent fit extends.","marker":"Pottschmidt et al. 2003"},{"why":"Models how Comptonization parameters can lower coherence, used to argue that such effects explain broadband drops but not the narrow 0.05 Hz dip.","marker":"Hua et al. 1997"}],"fun_headline_variants":["Hidden 0.05-Hz pulse in Cygnus X-1 found by AstroSat","Coherence dip exposes new timing signal in Cygnus X-1","First coherence dip above 3 keV in Cygnus X-1","New signal in Cygnus X-1 hints at jet activity","Cygnus X-1's hidden pulse: a jet-linked signal"],"cache_read_input_tokens":22656,"weakest_assumption_plain":"The central discovery rests on the assumption that deadtime and Poisson noise in LAXPC units 10 and 20 are uncorrelated, so the cross-detector coherence estimator is unbiased; the formal properties, uncertainties, and possible residual deadtime coupling of this estimator are not derived in the paper and are left to a future publication.","fun_headline_variants_meta":{"raw":{"variants":["Hidden 0.05-Hz pulse in Cygnus X-1 found by AstroSat","Coherence dip exposes new timing signal in Cygnus X-1","First coherence dip above 3 keV in Cygnus X-1","New signal in Cygnus X-1 hints at jet activity","Cygnus X-1's hidden pulse: a jet-linked signal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001175,"raw_usage":{"total_tokens":4911,"prompt_tokens":1051,"completion_tokens":3860,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":667,"completion_tokens_details":{"reasoning_tokens":3759}},"tokens_in":667,"tokens_out":3860,"duration_ms":31816,"temperature":1.0,"reasoning_tokens":3759,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:15:28.592398+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-analyse observation 1210 with the roles of LAXPC units 10 and 20 swapped and with simulated light curves that include detector deadtime but no intrinsic 0.05 Hz signal; if a coherence dip of similar depth appears in the deadtime-only simulations, or if the true uncertainty on $\\gamma^2$ at 0.05 Hz turns out to be large enough to make the observed dip consistent with unity, the discovery would be refuted.","supporting_citations":[{"cited_title":"S., & Piersol, A","cited_arxiv_id":null,"evidence_quote":"Provides the standard definition of coherence used to build the cross-spectral estimator."},{"cited_title":"A., et al","cited_arxiv_id":null,"evidence_quote":"Provides the earlier RXTE decomposition into the five main Lorentzians that this paper's state-dependent fit extends."}],"review_version":1}