Pith. sign in

REVIEW 3 major objections 4 minor 31 references

The QPO in REJ1034+396 originates in the hot corona

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Analyzing seven XMM-Newton observations of RE J1034+396, this paper argues that its quasi-periodic oscillation is produced entirely in the hot corona, with no measurable contribution from the accretion disk.

desk verdict A promising corona-localization analysis whose main quantitative claim is undercut by a mis-scaled soft-band PSD model. read the letter →

arxiv 2506.04347 v1 pith:ZSLEY7TB submitted 2025-06-04 astro-ph.HE

Chloe Taylor , Daniel Wilkins , Steven Allen This is my paper
classification astro-ph.HE
keywords AGNquasi-periodicoscillationX-raytiminghotcoronaaccretiondiskREJ1034+396timelagspowerspectraldensity
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

RE J1034+396 is one of the few active galactic nuclei with a reliable quasi-periodic oscillation, a roughly 3800-second X-ray flicker. This paper analyzes seven long XMM-Newton observations and argues that the oscillation is generated entirely in the hot corona, the compact X-ray-emitting plasma near the black hole, rather than in the accretion disk. The soft-band detection is explained as leakage of the corona's roughly 10% contribution to that band. If true, the result isolates an AGN QPO to a single spectral component for the first time, with direct consequences for which physical mechanisms can produce these oscillations.

What carries the argument

The load-bearing tool is Model A, a tied Lorentzian: the QPO Lorentzian appears in the hard band and in the soft band only with amplitude scaled by $F_{\rm PL}/F_{\rm comp}$, the ratio of coronal power-law flux to Comptonized disk flux in the soft band, fixed by earlier spectral fits. Model B adds an independent disk Lorentzian, and Bayesian Information Criterion selection between A and B with a threshold of $\Delta\mathrm{BIC}>10$ is what separates a corona-only origin from a disk-plus-corona origin. The cross-spectrum machinery for coherence, time lags, and covariance then tests whether the variability at the QPO frequency is consistent with coronal fluctuations and disk reprocessing.

What would settle it

Re-fit the same periodograms with the soft-band coronal fraction varied over its full spectral-fit uncertainty: if any observation then prefers an independent disk Lorentzian (Model B) by $\Delta\mathrm{BIC}>10$, the no-disk-QPO conclusion fails. A direct check would bin the soft band below 0.3 keV, where disk contamination is even smaller, and test whether the QPO amplitude still tracks the coronal fraction.

Watch

Extended reading notes

Core claim

Using a spectral decomposition of RE J1034+396, the authors treat the 0.3–0.5 keV band as about 90% disk emission and 10% coronal contamination, and the 2–7 keV band as essentially pure coronal emission. Fitting the hard and soft periodograms simultaneously, a model with a single tied Lorentzian scaled by the coronal flux fraction is preferred over a model that adds an independent disk Lorentzian in all seven observations. The QPO is significantly detected in five of seven observations, and the best-fitting disk contribution to the QPO power is consistent with zero, with a 99.9% upper limit of 73%. The coherence peak at the QPO frequency, the covariance spectrum peaking near 0.7 keV, and the phase-wrapped roughly 2000-second soft lag are all consistent with a coronal origin and with the disk merely reprocessing the coronal signal.

Load-bearing premise

The soft-band flux split (about 90% disk and 10% corona) is taken from a single earlier spectral fit, and the conclusion that no disk QPO exists stands or falls with that split.

Editorial extensions

If this is right

  • The QPO in RE J1034+396 must be generated by processes in the hot corona, so models that place the oscillator in the accretion disk are disfavored.
  • The roughly 2000-second intrinsic soft lag implies that the disk reprocesses coronal emission at all frequencies, and at the QPO frequency this lag phase-wraps, explaining the observed lag sign reversals.
  • The covariance spectrum at the QPO frequency is consistent with a varying coronal photon index, indicating the oscillation involves spectral softening when brighter rather than a separate soft spectral component.
  • If the QPO is precession of a hot inner flow or corona, its period constrains the corona radius or height to a few to tens of gravitational radii; if it is a magnetically choked accretion flow, it implies a black hole spin of about 0.5.
  • This is the first AGN QPO isolated to a single spectral component, providing a new benchmark for QPO models connecting AGN to black hole X-ray binaries.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same tied-Lorentzian test could be applied to the only other AGN QPO candidate and to high-frequency QPOs in black hole X-ray binaries to see whether a coronal origin is a general feature of these oscillations.
  • Because the conclusion depends on the $F_{\rm PL}/F_{\rm comp}$ ratio from one spectral model, future spectral fits that alter this ratio would predict a different soft-band QPO amplitude, providing a direct way to test the spectral decomposition.
  • The magnetically choked accretion flow interpretation predicts a spin near $a \sim 0.5$; an independent spin measurement, for example from the Fe K$\alpha$ profile, would discriminate between that model and the precession scenarios.
  • The phase-wrapping interpretation predicts that the measured lag sign at the QPO frequency should flip when the QPO frequency crosses roughly $2.6 \times 10^{-4}$ Hz, a pattern already present in the seven observations and testable with future monitoring.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper analyzes seven ~90 ks XMM-Newton observations of the narrow-line Seyfert 1 galaxy RE J1034+396, fitting the power spectral densities (PSDs) in a soft (0.3-0.5 keV) band and a hard (2-7 keV) band. The authors compare a model in which the QPO is a single Lorentzian seen in the hard band and appearing in the soft band only through coronal contamination (Model A) with a model that adds an independent soft-band Lorentzian (Model B). They report that Model A is preferred in five of seven observations, that the best-fitting disk QPO contribution is consistent with zero, and that the soft-band QPO is therefore entirely attributable to coronal contamination. Supporting evidence is drawn from a coherence peak at the QPO frequency, a phase-wrapping interpretation of the hard/soft lag, and a covariance spectrum at the QPO frequency that is described by a varying photon-index power law. The paper concludes that the QPO originates in the hot corona and discusses implications for Lense-Thirring precession, failed-jet, and magnetically choked accretion flow models.

Significance. If the central claim survives scrutiny, this would be the first convincing isolation of an AGN QPO to a single spectral component, and it would provide an important observational constraint for theoretical models of QPO generation. The paper has several strengths: the two-band PSD decomposition is a well-motivated way to separate disk and coronal emission, the significance simulations using Timmer & K\"onig realizations are a good practice, the analysis uses public XMM-Newton data and standard tools, and the coherence, lag, and covariance diagnostics are examined with appropriate care. However, the quantitative basis for the 'no disk QPO' conclusion is currently undermined by a scaling error in the soft-band coronal contamination term and by a very loose upper limit that is inconsistent with the abstract's wording. The result is potentially important but needs a corrected analysis before it can be accepted.

major comments (3)
  1. [Section 3, Eq. (A)] The soft-band coronal contamination term is mis-scaled for an (rms/mean)^2-normalized PSD. If the soft band has coronal flux F_PL and disk flux F_comp, a coherent coronal QPO with fractional rms described by L1 contributes (F_PL/(F_PL+F_comp))^2 L1 to the soft PSD, not (F_PL/F_comp) L1. For F_PL/F_comp ~ 0.13, the correct factor is ~ 0.013, about ten times smaller than the value used in the model. Because Model A is the reference model and Model B can only add positive power L2, this overprediction biases the BIC comparison against Model B, so the conclusion that there is no disk QPO contribution is not established by the current analysis. The 73% upper limit on the disk QPO contribution is also derived from this mis-scaled model and is therefore not reliable. The authors should rerun the PSD fits and the upper-limit derivation with the corrected scaling factor.
  2. [Section 3, Table 2 and text] Even under the authors' own model, the 99.9% upper limit on the disk QPO contribution is 73%. This is not consistent with the abstract's and conclusions' statements that the QPO is 'entirely attributable' to coronal emission with 'no additional contribution from the disk.' The data permit a disk contribution up to 73%; the correct statement is that no significant disk QPO is required, with the upper limit quoted explicitly. The wording should be softened in the abstract, Section 3, and Section 6.
  3. [Section 3, Model A] The fixed scaling F_PL/F_comp is taken from the spectral decomposition in Taylor et al. (2025), but the uncertainty in this ratio is not propagated into the PSD model comparison or the upper limit on L2. Since the soft-band prediction depends on this ratio (and, with corrected units, on its square), the conclusion could change if the ratio is different. The authors should propagate the uncertainty in F_PL/F_comp, or at minimum show how the BIC comparison and the L2 upper limit depend on the assumed value of this ratio.
minor comments (4)
  1. [Section 3, Table 2] The F_PL/F_comp values in Table 2 (0.120-0.137) imply a coronal fraction of roughly 11-12% in the soft band, not the ~10% stated in the text. Please make these numbers consistent.
  2. [Section 3, significance simulations] The significance simulations are generated using Obs 1's count rate, variance, and number of bins, but they are used to claim >3.5 sigma significance for all five QPO detections (Obs 1-5). Please justify that the single-observation simulation is representative of the other observations, or run simulations for each observation.
  3. [Section 3, model comparison] The paper states that Model B is never preferred over Model A but does not report the Delta BIC values for this comparison. Reporting these values would make the model-selection step more transparent and reproducible.
  4. [Section 4.2, phase wrapping] The phase-wrapping interpretation of the lag sign changes is plausible and consistent with Table 2, but the model has free parameters tau0 and R and no formal fit statistic is reported for the lag-frequency spectrum. A quantitative comparison with the data would strengthen this part of the argument.

Circularity Check

1 steps flagged · score 6.0 of 10

Model A's soft-band QPO scaling uses F_PL/F_comp linearly, but in (rms/mean)^2 units the correct factor is (F_PL/(F_PL+F_comp))^2, so the ~10x overprediction forces the BIC test to reject disk QPO by construction.

  1. fitted input called prediction [Section 3, Model A (Eq. A), after the PSD normalization statement]
    "The power spectral density function (PSD) was estimated from the periodogram using an [rms/mean]2 normalization ... PS(ν) = con. + FPL/Fcomp L1, (A) ... FPL/Fcomp is the ratio of coronal power law flux to the Comptonized disk flux in the soft band ... is set by the spectral fits performed in Taylor et al. (2025)."

    With [rms/mean]^2 normalization, a varying component of mean FPL in a band of total mean FPL+Fcomp has fractional rms diluted by FPL/(FPL+Fcomp), so its PSD contribution is (FPL/(FPL+Fcomp))^2 times its intrinsic PSD. The paper instead scales by FPL/Fcomp. For the tabulated FPL/Fcomp~0.13, this overpredicts the coronal soft-band QPO power by ~10x. Because Model B only adds a positive Lorentzian L2, an overpredicting Model A cannot be beaten, so the BIC comparison is biased against Model B. The conclusion 'no evidence for QPO power originating in the disk' is thus an artifact of the mis-scaled model, i.e., forced by the fitted flux ratio rather than by the data.

full rationale

The paper's central claim that the QPO originates in the corona is supported by several independent lines of evidence: the coherence peak at the QPO frequency, the covariance spectrum shape, and the lag-energy spectrum with phase wrapping. However, the PSD-based test for a separate disk QPO (Model A vs Model B) is compromised by a unit/scaling error in the soft-band coronal contamination term. In (rms/mean)^2 normalization, the correct scaling factor is the square of the flux fraction, not the flux ratio; using the flux ratio overpredicts the coronal soft-band QPO by an order of magnitude. Since Model B can only add positive power, it can never improve on an overpredicting Model A, so the BIC test cannot actually detect a disk QPO. The 'no disk QPO' conclusion is therefore forced by the model construction and the fitted flux ratio from the authors' prior work (Taylor et al. 2025), which is a self-cited value whose uncertainty is not propagated. Other parts of the analysis (coherence, covariance, lag) are not circular because they compare forward models based on independent spectral parameters to the observed data, rather than deriving the conclusion from the same fitted quantity. The central claim is not fully equivalent to the input, but the principal quantitative evidence for zero disk QPO reduces to a mis-scaled construction, warranting a score of 6.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central claim rests on a spectral decomposition from the authors' previous paper, on the assumed linearity of the disk response, and on the assumption that any disk QPO would add a second Lorentzian. No new physical entities are introduced; the hot corona and Comptonized disk are established components from prior literature.

free parameters (4)
  • F_PL/F_comp (soft-band coronal-to-disk flux ratio) = 0.120-0.137 per observation
    Used in Model A to set the soft-band QPO amplitude as a fixed multiple of the hard-band Lorentzian. The ratio comes from spectral fits in Taylor et al. (2025), the authors' prior paper, and is not fitted in this work; the conclusion that the soft QPO is coronal contamination depends on this ratio.
  • Intrinsic soft lag tau0 = ~1920-2000 s
    Chosen to match the phase-wrapping frequency of ~2.6e-4 Hz inferred from the sign changes of the lag at the QPO frequency. It is an input to the phase-wrapped lag model.
  • Dilution factor R = ~1
    Chosen so the phase-wrapped 2000 s soft lag model matches the observed lag amplitude above ~2e-4 Hz.
  • Covariance model photon index range = 2.05-2.3
    Chosen to reproduce the QPO-frequency covariance spectrum with a softer-when-brighter power law; not fitted to the covariance data, but selected by hand from plausible variability ranges.
assumptions (6)
  • domain assumption The soft 0.3-0.5 keV band is ~90% Comptonized disk emission and ~10% coronal power law, and the hard 2-7 keV band is dominated by the hot corona.
    Stated in Section 3 and based on the spectral analysis of Taylor et al. (2025). Model A's scaling of the soft QPO amplitude uses this decomposition.
  • domain assumption The QPO in the hard and soft bands is the same signal, so a tied Lorentzian model (same centroid and width) is appropriate.
    Implicit in Model A/B; if the soft and hard QPOs were independent signals, the model comparison would not test coronal vs disk origin.
  • domain assumption Any disk contribution to the QPO would appear as an additive Lorentzian in the soft band.
    Model B assumes a second Lorentzian, L2, captures disk QPO power; this assumes a linear superposition of independent components.
  • domain assumption The disk response to coronal irradiation is a linear reprocessing with a constant time delay of about 2000 s, so phase wrapping describes the lag-frequency spectrum.
    Used in Section 4.2 to interpret the lag sign changes as phase wrapping; the physical picture is that the disk responds to the corona at all frequencies.
  • domain assumption The lag energy spectrum prediction uses a disk response radius of ~100 rg and a reflection fraction that matches R ~ 1.
    Section 4.2: 'impose a height to simulate the soft disk emission responding at disk radii of ~100rg, and a reflection fraction to match the R~1 dilution.' This follows Taylor et al. (2025) and future work in prep.
  • standard math Standard Fourier/statistical tools (Whittle likelihood, BIC, coherence definitions) are valid for these periodograms.
    The analysis relies on Vaughan (2010), Uttley et al. (2014), and standard X-ray timing practice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The QPO in REJ1034+396 originates in the hot corona." pith.science (2026). https://pith.science/paper/ZSLEY7TB

@misc{pith2026250604347,
  author       = {Pith},
  title        = {Pith review of: The QPO in REJ1034+396 originates in the hot corona},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZSLEY7TB}},
  note         = {Machine review of arXiv:2506.04347}
}
read the original abstract

REJ1034+396 is one of the few active galactic nuclei with a significant quasi-periodic oscillation (QPO). The QPO has been observed in over 1 Ms of XMM-Newton observations spanning over a decade. We investigate the power spectral density function (PSD) of 7 long (~90 ks) XMM-Newton observations of the active galactic nucleus REJ1034+396 in two energy bands. The soft (0.3-0.5 keV) band targets emission from the disk, while the hard (2-7 keV) band isolates the primary X-ray continuum emission from the corona. The QPO is significantly detected in the hard band of 5 of the 7 observations. The best fitting models indicate that the QPO detection in both bands is entirely attributable to the coronal emission with no additional contribution from the disk. This explains the strong coherence between the hard and soft bands at the QPO frequency. The covariance spectrum is consistent with this picture as the variability at QPO frequencies is attributed solely to fluctuations in the hot corona. The time lag as a function of energy is well described by a ~2000 s intrinsic soft lag, resulting from the disk responding to emission from the corona, that undergoes phase wrapping at approximately the QPO frequency. By demonstrating that in this system the QPO arises in the corona, we provide new insights into the mechanisms generating QPOs.

Figures

Figures reproduced from arXiv: 2506.04347 by the authors.

Figure 2
Figure 2. The stacked peiodogram of all 7 observations in the hard (red) and soft (blue) bands. The coherence is shown in grey. where con. represents the preferred continuum model, FPL Fcomp is the ratio of coronal power law flux to the Comptonized disk flux in the soft band, and L1 is the Lorentzian. FPL Fcomp is set by the spectral fits performed in Taylor et al. (2025). Model A assumes the QPO orig￾inates entirely in the c… view at source ↗
Figure 1
Figure 1. The light curves for the 7 observations in [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. (a) The hard band periodogram of Obs 1. The best fitting model (Model A) is shown in dashed black. The Poisson noise is indicated in dotted black. (b) The soft band periodogram of Obs 1. The best fitting model (Model A) is shown in dashed black. For visualization purposes, above f > 10−3 Hz we overlay the binned data in six log spaced bins. The Poisson noise is indicated in dotted black and re￾flects the expected le… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: a) The stacked lag frequency spectrum of all seven observations. Predicted phase wrapping due to a di￾luted ∼ 2000 s soft lag is shown in light grey. A typical low frequency hard lag dominates below ∼ 2 × 10−4 Hz. Above this frequency, the phase wrapped soft lag is a g…
Figure 5
Figure 5. Figure 5: The stacked lag energy spectrum of all seven ob￾servations in three frequency bins below the QPO frequency. The frequency bins are 2 − 2.6 × 10−4 (black), 1 − 2 × 10−4 (red), and 0.1 − 1 × 10−4 (blue). The soft Comptonized disk emission lag is superimposed on an overal…
Figure 6
Figure 6. Figure 6: The individual lag energy spectra of four of the observations. Observations 2 and 7 show the soft Comp￾tonized disk emission lag while observations 5 and 6 show the phase wrapped soft lag. The slight asymmetry at low energies (∼ 700s soft lag vs ∼ 1000s hard lag) is du…
Figure 8
Figure 8. Figure 8: Lense-Thirring precession period as a function of black hole spin and the radius of the hot inner flow for the precessing hot inner flow QPO model. Contours indicating the upper (4000 s) and lower (3570 s) limits of the observed QPO period are indicated. The upper band…
Figure 9
Figure 9. Figure 9: Lense-Thirring precession period as a function of black hole spin and the height of the corona for the failed jet QPO model. Contours indicating the upper (4000 s) and lower (3570 s) limits of the observed QPO period are indicated. The upper band assumes the minimum ma…
Figure 10
Figure 10. Figure 10: Schematic of the inner regions of RE J1034+396 (in the MCAF scenario). The turbulent accretion flow is il￾lustrated in orange with the warm (∼ 0.2 keV) Comptonizing plasma spanning the surface shown in gold. Disk photons are Compton scattered to soft X-ray energies as…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

31 extracted references · 20 canonical work pages

  1. [1]

    2014, MNRAS, 445, L16

    Middleton, M. 2014, MNRAS, 445, L16

    work page 2014

  2. [2]

    N., Vaughan, S., & Uttley, P

    Alston, W. N., Vaughan, S., & Uttley, P. 2013, MNRAS, 435, 1511 Ar´ evalo, P., & Uttley, P. 2006, MNRAS, 367, 801

    work page 2013

  3. [3]

    Arnaud, K. A. 1996, in Astronomical Society of the Pacific Conference Series, Vol. 101, Astronomical Data Analysis Software and Systems V, ed. G. H. Jacoby & J. Barnes, 17

    1996

  4. [4]

    2010, MNRAS, 401, 507

    Bian, W., & Huang, K. 2010, MNRAS, 401, 507

    work page 2010

  5. [5]

    2014, Class

    Chakraborty, C., & Majumdar, P. 2014, Class. Quantum Grav., 31, 075006

    work page 2014

  6. [6]

    2016, A&A, 594, A102 De Marco, B., Ponti, G., Cappi, M., et al

    Czerny, B., You, B., Kurcz, A., et al. 2016, A&A, 594, A102 De Marco, B., Ponti, G., Cappi, M., et al. 2013, MNRAS, 431, 2441

    work page 2016

  7. [7]

    2012, MNRAS, 420, 1848 Gierli´ nski, M., Middleton, M., Ward, M., & Done, C

    Done, C., Davis, S., Jin, C., Blaes, O., & Ward, M. 2012, MNRAS, 420, 1848 Gierli´ nski, M., Middleton, M., Ward, M., & Done, C. 2008, Nature, 455, 369–371

    work page 2012

  8. [8]

    R., & Motta, S

    Ingram, A. R., & Motta, S. E. 2019, NewAR, 85, 101524

    work page 2019

Show all 31 references
  1. [9]

    2001, A&A, 365, L1

    Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1

    2001

  2. [10]

    2020, MNRAS, 495, 3538

    Jin, C., Done, C., & Ward, M. 2020, MNRAS, 495, 3538

    2020

  3. [11]

    N., Fabian, A

    Kara, E., Alston, W. N., Fabian, A. C., et al. 2016, MNRAS, 462, 511, doi: 10.1093/mnras/stw1695

    2016 doi

  4. [12]

    E., & Raftery, A

    Kass, R. E., & Raftery, A. E. 1995, J. Amer. Statist. Assoc., 90, 773, doi: 10.2307/2291091

    1995 doi

  5. [13]

    C., Tchekhovskoy, A., & Blandford, R

    McKinney, J. C., Tchekhovskoy, A., & Blandford, R. D. 2012, MNRAS, 423, 3083

    2012

  6. [14]

    2010, MNRAS, 403, 9

    Middleton, M., & Done, C. 2010, MNRAS, 403, 9

    2010

  7. [15]

    2009, MNRAS, 394, 250

    Schurch, N. 2009, MNRAS, 394, 250

    2009

  8. [16]

    2011, MNRAS, 417, 250, doi: 10.1111/j.1365-2966.2011.19185.x

    Middleton, M., Uttley, P., & Done, C. 2011, MNRAS, 417, 250, doi: 10.1111/j.1365-2966.2011.19185.x

    2011

  9. [17]

    F., Done, C., & Pounds, K

    Mushotzky, R. F., Done, C., & Pounds, K. A. 1993, ARA&A, 31, 717

    1993

  10. [18]

    C., & Miller, J

    Reis, R. C., & Miller, J. M. 2013, ApJL, 769, L7, doi: 10.1088/2041-8205/769/1/L7 R´ o˙ za´ nska, A., Malzac, J., Belmont, R., Czerny, B., &

    2013 doi

  11. [19]

    Petrucci, P. O. 2015, A&A, 580, A77

    2015

  12. [20]

    1973, A&A, 24, 337

    Shakura, N., & Sunyaev, R. 1973, A&A, 24, 337

    1973

  13. [21]

    L., Mushotzky, R

    Smith, K. L., Mushotzky, R. F., Boyd, P. T., & Wagoner, R. V. 2018, ApJL, 860, L10

    2018

  14. [22]

    L., Tandon, C

    Smith, K. L., Tandon, C. R., & Wagoner, R. V. 2021, ApJ, 906, 92 Str¨ uder, L., Briel, U., Dennerl, K., et al. 2001, A&A, 365, L18

    2021

  15. [23]

    S., Wilkins, D

    Taylor, C. S., Wilkins, D. R., & Allen, S. W. 2025, ApJ, 979, 101, doi: 10.3847/1538-4357/ad9f3e

    2025 doi

  16. [24]

    D., Uttley, P., & McHardy, I

    Taylor, R. D., Uttley, P., & McHardy, I. M. 2003, MNRAS, 342, L31, doi: 10.1046/j.1365-8711.2003.06742.x

    2003

  17. [25]

    1995, A&A, 300, 707

    Timmer, J., & K¨ onig, M. 1995, A&A, 300, 707

    1995

  18. [26]

    Wilkins, D. R. 2014, A&A Rv, 22, 72

    2014

  19. [27]

    A., & Nowak, M

    Vaughan, B. A., & Nowak, M. A. 1997, ApJL, 474, L43

    1997

  20. [28]

    2010, MNRAS, 402, 307

    Vaughan, S. 2010, MNRAS, 402, 307

    2010

  21. [29]

    R., & Fabian, A

    Wilkins, D. R., & Fabian, A. C. 2013, MNRAS, 430, 247

    2013

  22. [30]

    2009, MNRAS, 397, 666, doi: 10.1111/j.1365-2966.2009.15008.x

    Wilkinson, T., & Uttley, P. 2009, MNRAS, 397, 666, doi: 10.1111/j.1365-2966.2009.15008.x

    2009

  23. [31]

    2024, ApJL, 961, L32

    Xia, R., Liu, H., & Xue, Y. 2024, ApJL, 961, L32

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.