Pith. sign in

REVIEW 4 major objections 4 minor 45 references

Active galactic nucleus time-variability analysis and its caveats

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Flare-biased sampling distorts the power spectra of AGN light curves, and standard simulations that reuse the observed time stamps cannot reproduce the distortion.

desk verdict Useful caveats review with two simulation-based findings, but the flare-bias claim needs error bars and a null ensemble before it carries weight. read the letter →

arxiv 2412.08192 v1 pith:UBVH7QJX submitted 2024-12-11 astro-ph.GA

Sofia Kankkunen , Merja Tornikoski , Talvikki Hovatta This is my paper · ORCID
classification astro-ph.GA
keywords galaxies:activequasars:generalmethods:dataanalysispowerspectraldensityperiodogramred-noiseleaklight-curvesimulationssurrogate
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

The paper argues that common statistical tools for measuring AGN variability timescales have hidden failure modes, and demonstrates two with simulations. First, when a monitoring campaign observes more densely during flaring states, interpolating the unevenly sampled light curve changes its measured power spectral density in a way that standard PSRESP-style simulations do not capture: simulated light curves that reuse the observed time stamps have flares at random times, so interpolation smooths them more than it smooths the real flare-biased light curve. Second, the popular EMP13 algorithm for generating light curves with a given probability distribution is often misused by cutting the simulated series short before the iterative step, which randomizes phase information, ruins the red-noise-leak correction, and flattens periodograms of steep-spectrum sources. If these findings hold, previously reported PSD slopes, bend timescales, and quasiperiodicity claims from such analyses may contain a systematic bias that error bars do not reflect.

What carries the argument

The load-bearing object is the periodogram computed after linear interpolation of an unevenly sampled light curve, together with the PSRESP simulation procedure that copies the observed time stamps onto simulated light curves, and the EMP13/IAAFT surrogate-data iteration that is supposed to match a specified probability density function. Interpolation smooths over gaps, reducing high-frequency power and steepening the periodogram; the paper's simulations show that when dense sampling is aligned with flares, this steepening is partly cancelled because the flares themselves are well resolved, whereas randomly placed dense-sampling windows in simulations are not. The EMP13 algorithm's iterative step also acts as a phase-randomising operation, and the paper shows that cutting the simulated series before that step mixes the red-noise leak into the periodogram in a way that windowing cannot undo.

What would settle it

Measure the correlation between observing-cadence density and flux density in a monitoring data set, then compare the observed periodogram with the PSRESP-simulated mean periodogram: if a source with strong flare-biased sampling shows no systematic offset, or if a source with zero sampling-flux correlation still shows the mismatch, the flare-bias explanation would be wrong. A simpler simulation check is to generate a light curve with a known PSD, apply a deliberately flare-biased sampling pattern, and verify the predicted sign and size of the offset; removing the flare-sampling correlation should make the offset disappear.

Watch

Extended reading notes

Core claim

The central discovery is a mismatch between observed and simulated periodograms that traces to the correlation between sampling density and source brightness. In the PSRESP method, the observed sampling pattern is copied onto simulated light curves so that sampling artefacts appear in both, and the goodness of fit is judged by comparing the observed periodogram with the simulated mean periodogram. The authors simulate a flare-biased observing pattern and find that the real flare-biased light curve, after interpolation, yields a periodogram that is steeper than the evenly sampled case but not as steep as the mean periodogram of simulations carrying the same time stamps. The reason is that in the simulations the dense-sampling windows fall randomly with respect to flares, whereas in the actual flare-biased light curve they preferentially fall on the flares, suppressing the smoothing effect of interpolation. The same mechanism, the paper argues, can make a single observed light curve look significantly different from its own simulations even when the underlying PSD model is correct. The paper's second discovery is that the EMP13 surrogate algorithm loses phase information when its long initial simulation is cut to the observing-window length before the iterative step, which prevents Hann-window or end-matching corrections from removing red-noise leak and flattens the periodogram for steep PSD slopes.

Load-bearing premise

The claim depends on real monitoring campaigns being genuinely denser during flaring states, coupled with the absence of that coupling in simulations that merely reuse the time stamps; the paper illustrates the effect with simulated flare-biased sampling and notes that its own radio data shows much less prominent flare bias, but does not measure the correlation's strength in actual AGN data.

Editorial extensions

If this is right

  • For sources whose monitoring cadence is correlated with flaring, PSRESP model fits will systematically compare the observed periodogram to a simulation ensemble that is biased in the opposite direction, so inferred PSD slopes and bend timescales carry an error not captured by the fit's scatter.
  • A large parameter space of good fits is to be expected in sparsely sampled periodogram fitting; slope and characteristic timescale are strongly correlated and cannot be pinned down from such data.
  • The EMP13 shortcut of cutting the light curve before the iterative step flattens periodograms of steep-spectrum sources, making it impossible to put reliable upper limits on PSD slopes when windowing is used.
  • The Hann window and end-matching correct red-noise leak only when phase information is preserved; quasiperiodicity searches that use windowing on phase-randomised surrogates can produce false detections or false non-detections.
  • The Lomb-Scargle false-alarm probability applies only to white noise and should not be used to claim quasiperiodicities in red-noise AGN data.

Reading between the lines

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

  • The flare-bias mechanism should apply to any stochastic astronomical source observed with a target-of-opportunity or alert-driven cadence, not just AGNs; X-ray and gamma-ray transients and microlensing events could show the same observed-versus-simulated periodogram mismatch.
  • A direct empirical test would be to measure the cross-correlation between observing density and flux state in a real monitoring data set and check whether the observed-minus-simulated periodogram offset scales with that correlation.
  • Papers that report bend timescales from PSRESP on sparsely sampled radio light curves may need re-analysis with flare-bias-aware simulations; the authors note their own radio data has much less prominent flare bias, so the effect may be most relevant to X-ray and optical campaigns with flare-triggered follow-up.
  • The EMP13 phase-randomization issue suggests that published surrogate-based significance estimates for quasiperiodicities, especially for steep PSD slopes, could shift if the full-length iterative step is used; re-running such analyses would be a cheap check.
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

4 major / 4 minor

Summary. The paper presents a practical review of statistical caveats in AGN variability analysis, supported by simulations. It reproduces known issues (red-noise leak, aliasing, interpolation effects, structure function biases) and claims two previously overlooked caveats: (1) interpolation of light curves whose sampling is biased toward flaring states affects the measured PSD differently than applying the same sampling times to independent simulated light curves, thereby biasing PSRESP-style fits; and (2) the EMP13 surrogate algorithm is often misused when the light curve is cut to a shorter segment before the iterative step, because this randomizes phase information and prevents windowing from correcting red-noise leak. The paper also discusses periodogram normalisation, timescale estimation, and quasiperiodicity detection, and it summarises lessons learned from the authors' companion Paper I.

Significance. If the flare-bias claim holds, it would affect a broad class of PSD fitting studies of AGN light curves, particularly in radio monitoring where sampling density may correlate with flaring activity. The EMP13 caveat is a useful practical warning for surrogate-data users. The paper is valuable as a consolidated review with simulation-based demonstrations, and the authors are appropriately cautious about some limitations of their own tests. However, the statistical evidence for the main new claim is currently insufficient, so the central novelty is not yet established.

major comments (4)
  1. [§4.1.1, Fig. 3] The central claim that flare-biased sampling changes the effect of interpolation on the PSD is supported only by a visual comparison between a single periodogram (black) and the mean of 1000 simulated periodograms (purple), with no error bars, quantile ranges, or significance test. Because red-noise periodograms have large realization-to-realization scatter, the apparent difference could be a fluctuation. Please provide the distribution of the simulated periodograms (e.g., 5–95% percentiles) or, ideally, the distribution of single-realization periodograms under the null hypothesis that the sampling times are independent of the flux values, and state explicitly whether the observed black periodogram lies outside that null distribution.
  2. [§4.1.1] The construction of the flare-biased sampling is described only qualitatively ('dividing the time stamps of flux densities into two arrays based on their values'), which makes it impossible to reproduce or assess the magnitude of the bias. Please specify the exact algorithm, including the selection rule, the density contrast between flaring and non-flaring states, the interpolation scheme used, and the parameter values for the simulation shown in Fig. 3. Without this, the generality of the claimed effect cannot be evaluated.
  3. [§4.1.1 and §7] The paper notes that the MRO light curves have 'much less prominent' flare bias but never defines or measures flare bias in real data. To make the practical claim actionable, the authors should propose a quantitative metric for the correlation between sampling density and flux level, report it for the simulations, and, if possible, for the MRO data. Otherwise the relevance of the effect to actual AGN monitoring campaigns remains speculative.
  4. [§5.4.2, Figs. 11–12] The claim that cutting the TK95 light curve before the EMP13 iterative step randomizes phase and hence disables windowing-based red-noise-leak correction is demonstrated only by example light curves and periodograms. Please add a quantitative comparison across multiple realisations (e.g., mean and scatter of measured PSD slopes with and without phase preservation) and state the range of β for which the effect is significant. As written, the reader cannot tell whether the effect is large enough to affect typical AGN analyses.
minor comments (4)
  1. [§5.1, Eq. (8)] Please define ΔPsim(ν) explicitly; from the text it is presumably the standard deviation of the simulated periodograms, but the notation is ambiguous.
  2. [§4.1.2, Fig. 5] The comparison of normalisation schemes is based on one example; it would be helpful to state over how many simulations the 'closer' agreement was assessed.
  3. [§5.4.1] The statement that TK95 variance does not depend on N is imprecise; the variance of the simulated series does depend on the number of points through the normalization of the Fourier components. Please clarify the intended meaning, particularly in relation to the red-noise-leak simulation lengths.
  4. [§6.1, Fig. 13] The correlation between β and the bend timescale is shown visually but not quantified; a simple measure of overlap of the periodogram distributions or a summary statistic would strengthen the claim.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; the paper is a transparent simulation study whose caveat claims do not reduce to fitted inputs or self-citations.

full rationale

The paper's central claims are produced by controlled simulations with known input PSD/PDF models, not by fitting parameters and then predicting the same data. The flare-bias finding (Sect. 4.1.1, Fig. 3) is constructed by generating timestamps correlated with flux density in one light curve and applying the same timestamps to independent realizations; the paper explicitly explains that the difference arises because 'the replicated periods of denser sampling only randomly occur during flares in simulated light curves.' This is a transparent, by-design demonstration of a mechanism rather than a hidden circularity: no quantity is fitted from one part of the data and then reported as a prediction of another part. The EMP13 phase-randomization caveat (Sect. 5.4.2, Figs. 11-12) is likewise demonstrated by comparing simulation choices; the conclusion follows from the algorithm description but is independently exhibited numerically. Self-citations to Paper I provide model parameters, slopes, and context (Sects. 1, 2.2, 4.1.1, 6), but none of these citations are load-bearing for the abstract's new caveats; the new results stand on the simulations reported in this paper. The absence of error bars on the mean simulated periodogram in Fig. 3 is a statistical robustness concern, not a circularity concern. Overall, the derivation chain is empirical and self-contained; no equation reduces to its own input, and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper's claims are qualitative and rest on simulation choices rather than fitted parameters. The most consequential unstated choice is the flare-bias construction in Section 4.1.1, which is not quantified. No new physical entities are introduced, and no parameters are fitted to observational data.

free parameters (3)
  • Flare-bias strength and selection rule = not quantified
    The paper constructs flare-biased sampling by 'dividing time stamps of flux densities into two arrays based on their values' and 'heavily favouring higher flux densities' (Section 4.1.1). The threshold and degree of bias are not specified, yet the central interpolation claim depends on this setup.
  • Simulated PSD parameters (beta and xb) = e.g., beta = 2, xb = 500 days
    The demonstrations use chosen input PSDs, such as a bending power law with beta = 2 and xb = 500 in Figure 3. These are illustrative inputs, not fitted values, but they set the regime in which the claims are demonstrated.
  • Red-noise-leak simulation length multiplier = 10 (up to 10,000 tested)
    To simulate red-noise leak, Paper I used initial light curves 10 times longer than the target segment (Section 5.2.1). The paper states this choice 'appeared to be sufficient', a hand-chosen practical value.
assumptions (4)
  • domain assumption AGN light curves can be modeled as realizations of stochastic processes with simple or bending power-law PSDs.
    Used throughout the simulations (Sections 3.2, 3.3, and 5) to generate mock light curves with known properties. This is standard practice in the field but is an assumption about the true underlying process, not a demonstrated fact.
  • standard math The periodogram is an approximately unbiased estimator of the underlying PSD when binned or averaged.
    The paper interprets simulated mean periodograms as the true PSD when comparing models (Sections 4.1 and 5.1). This is a standard result from Fourier analysis.
  • domain assumption The TK95 and EMP13 algorithms produce valid realizations of the target PSD and PDF when used according to their original prescriptions.
    The paper tests the failures of these algorithms, so their baseline validity is assumed. The EMP13 claim in Section 5.4.2 relies on the original algorithm's described phase behavior to explain the observed windowing failure.
  • ad hoc to paper Observations of AGNs can be biased toward flaring states, meaning sampling density correlates with flux density.
    The central new claim in Section 4.1.1 and Figure 3 requires this correlation to exist in real data and to be absent when the same time stamps are applied to independent stochastic realizations. The paper states this assumption but does not measure the bias in the MRO data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Active galactic nucleus time-variability analysis and its caveats." pith.science (2026). https://pith.science/paper/UBVH7QJX

@misc{pith2026241208192,
  author       = {Pith},
  title        = {Pith review of: Active galactic nucleus time-variability analysis and its caveats},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UBVH7QJX}},
  note         = {Machine review of arXiv:2412.08192}
}
read the original abstract

In this study, we demonstrate some of the caveats in common statistical methods used for analysing astronomical variability timescales. We consider these issues specifically in the context of active galactic nuclei (AGNs) and use a more practical approach compared to mathematics literature, where the number of formulae may sometimes be overwhelming. We conducted a thorough literature review both on the statistical properties of light-curve data, specifically in the context of sampling effects, as well as on the methods used to analyse them. We simulated a wide range of data to test some of the known issues in AGN variability analysis as well as to investigate previously unknown or undocumented caveats. We discovered problems with some commonly used methods and confirmed how challenging it is to identify timescales from observed data. We find that interpolation of a light curve with biased sampling, specifically with bias towards flaring events, affects its measured power spectral density in a different manner than those of simulated light curves. We also find that an algorithm aiming to match the probability density function of a light curve has often been used incorrectly. These new issues appear to have been mostly overlooked and not necessarily addressed before, especially in astronomy literature.

Figures

Figures reproduced from arXiv: 2412.08192 by the authors.

Figure 1
Figure 1. Comparison between observed data of 3C454.3 at 37 GHz (error bars not included) and simulated data on the right with similar modelling parameters obtained from the analysis of Paper I. The simulated data is evenly sampled, and the observed data has an uneven cadence. The naming conventions of stochastic processes vary be￾tween disciplines and also within astronomical literature as noted by Scargle (2020). Here we di… view at source ↗
Figure 2
Figure 2. Plot showing four PSDs of varying parameters and with a slope of β = 2. In the blue is the simple power law, in red is the bending power law with a characteristic timescale of 20 days, in purple is the bending power law with a characteristic timescale of 200 days, and last in green is the bending power law with a characteristic timescale of 2000 days. The dashed grey line shows the principle of how a simple power la… view at source ↗
Figure 3
Figure 3. Comparison between the PSDs of an evenly sampled light curve and flare-biased light curve. Left-side plot: Light curve we generated using the bending power-law PSD with parameters β = 2 and xb = 500. The grey light curve has been evenly sampled, and the black light curve has flare bias. Right-side plot: Obtained periodograms. The periodogram in grey shows the PSD of the light curve with even sampling. The periodogra… view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: Mean periodograms plotted using simulated light curves constructed from two simple power laws with β = 2 and β = 3 with a varying number of data points. The first plot includes 48 data points, the second plot 146 data points, and the third plot is evenly sampled with t…
Figure 5
Figure 5. Figure 5: Examples of scaling the periodogram. The black and orange pe￾riodograms use the normalisation presented in Eq. 5, whereas the grey and purple periodograms are shifted to have a mean of zero in logarith￾mic space. mock light curve. The result is desirable as it removes …
Figure 6
Figure 6. Figure 6: Plots illustrating the cause of red-noise leak in Fourier analysis. The top left-side plot shows a sine wave with two full periods resulting in no red-noise leak, depicted by the continuous sine wave transfer from blue to orange. The right-side plot shows a non-integer…
Figure 7
Figure 7. Figure 7: Comparison between periodograms with no red-noise leak and with red-noise leak present. The blue periodogram is of a light curve without red-noise leak, and the orange periodogram is of a light curve with red-noise leak. Both of the light curves were simulated using th…
Figure 8
Figure 8. Figure 8: Effect of windowing on steep slopes in periodogram analysis. The left-side plot shows examples of periodograms with varying simple power-law slopes and red-noise leak present. On the right-side plot are the equivalent power-law slopes, but this time the same time serie…
Figure 9
Figure 9. Figure 9: Comparison between simulated light curves when varying the simulation length in the TK95 algorithm. The left-side light curve is a 40-year segment from a 400 year-long simulated light curve using the TK95 algorithm and slope β = 2. The right-side light curve has the sa…
Figure 10
Figure 10. Figure 10: Distributions of flux densities from simulated data using the TK95 method with the simple power law and the slopes β = 0, β = 1, and β = 2, respectively [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Plots illustrating how randomising the phase information affects the resulting light curve in the EMP13 algorithm. The top row is the entire simulated light curve with slope β = 3 using the TK95 algorithm. The bottom left-side plot shows the light curve obtained from …
Figure 12
Figure 12. Figure 12: Resulting periodograms when using a window function on light curves with phase information randomised and with phase information retained. The light curves have been convolved with the Hann window to show the effect of phase randomisation in the iterative step of the …
Figure 13
Figure 13. Figure 13: Mean periodograms of sparsely sampled light curves showing how small the differences are between the different parameter combina￾tions. variance. We do not expect the pulses to have equal durations, and thus the bend frequency should refer to the longest pulse duratio…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 34 canonical work pages

  1. [1]

    & Hasinger, G

    Belloni, T. & Hasinger, G. 1990, A&A (ISSN 0004-6361), vol. 227, no. 2, Jan. 1990, p. L33-L36., 227, L33

    work page 1990

  2. [2]

    Deeming, T. J. 1975, Ap&SS, 36, 137

    1975

  3. [3]

    & Nandra, K

    Edelson, R. & Nandra, K. 1999, ApJ, 514, 682

    1999

  4. [4]

    2013, MNRAS, 433, 907

    Emmanoulopoulos, D., McHardy, I., & Papadakis, I. 2013, MNRAS, 433, 907

    work page 2013

  5. [5]

    M., & Uttley, P

    Emmanoulopoulos, D., McHardy, I. M., & Uttley, P. 2010, MNRAS, 404, 931

    work page 2010

  6. [6]

    Fougere, P. F. 1985, J. Geophys. Res.: Space Physics, 90, 4355

    work page 1985

  7. [7]

    1968, Proc

    Halford, D. 1968, Proc. IEEE, 56, 251

    work page 1968

  8. [8]

    2002, Max Plank Inst, 12

    Heinzel, G., Rüdiger, A., & Schilling, R. 2002, Max Plank Inst, 12

    work page 2002

Show all 45 references
  1. [9]

    2007, A&A, 469, 899

    Hovatta, T., Tornikoski, M., Lainela, M., et al. 2007, A&A, 469, 899

    2007

  2. [10]

    1992, ApJ, Part 1 (ISSN 0004-637X), vol

    Hughes, P., Aller, H., & Aller, M. 1992, ApJ, Part 1 (ISSN 0004-637X), vol. 396, no. 2, Sept. 10, 1992, p. 469-486., 396, 469

    1992

  3. [11]

    Hunt, G. A. 1956, Trans. Am. Math. Soc, 81, 294

    1956

  4. [12]

    C., Bechtold, J., & Siemiginowska, A

    Kelly, B. C., Bechtold, J., & Siemiginowska, A. 2009, ApJ, 698, 895

    2009

  5. [13]

    C., Becker, A

    Kelly, B. C., Becker, A. C., Sobolewska, M., Siemiginowska, A., & Uttley, P. 2014, ApJ, 788, 33

    2014

  6. [14]

    Keshner, M. S. 1982, Proc. IEEE, 70, 212 Kozłowski, S. 2016, MNRAS, 459, 2787 Lähteenmäki, A., Järvelä, E., Hovatta, T., et al. 2017, A&A, 603, A100

    1982

  7. [15]

    A., & Abbt, M

    Limpert, E., Stahel, W. A., & Abbt, M. 2001, Biosci., 51, 341

    2001

  8. [16]

    J., Türler, M., Valtaoja, E., et al

    Lindfors, E. J., Türler, M., Valtaoja, E., et al. 2006, A&A, 456, 895

    2006

  9. [17]

    2017, MNRAS, 467, 4565

    Liodakis, I., Pavlidou, V ., Hovatta, T., et al. 2017, MNRAS, 467, 4565

    2017

  10. [18]

    Lomb, N. R. 1976, Ap&SS, 39, 447

    1976

  11. [19]

    2008, Surrogate Data - A Qualitative and Quantitative Analysis, ed

    Maiwald, T., Mammen, E., Nandi, S., & Timmer, J. 2008, Surrogate Data - A Qualitative and Quantitative Analysis, ed. R. Dahlhaus, J. Kurths, P. Maass, & J. Timmer (Berlin, Heidelberg: Springer Berlin Heidelberg), 41–74

    2008

  12. [20]

    2002, arXiv preprint physics/0204033

    Milotti, E. 2002, arXiv preprint physics/0204033

    2002 arXiv

  13. [21]

    J., Chakraborty, N., & Cotter, G

    Morris, P. J., Chakraborty, N., & Cotter, G. 2019, MNRAS, 489, 2117

    2019

  14. [22]

    & Lawrence, A

    Papadakis, I. & Lawrence, A. 1993, MNRAS, 261, 612

    1993

  15. [23]

    & Trippe, S

    Park, J. & Trippe, S. 2017, ApJ, 834, 157

    2017

  16. [24]

    Press, W. H. 1978, Comments on Modern Physics, Part C-Comments on Astro- physics, vol. 7, no. 4, 1978, p. 103-119., 7, 103

    1978

  17. [25]

    1978, Proc

    Rutman, J. 1978, Proc. IEEE, 66, 1048

    1978

  18. [26]

    2002, A&A, 394, 851

    Savolainen, T., Wiik, K., Valtaoja, E., Jorstad, S., & Marscher, A. 2002, A&A, 394, 851

    2002

  19. [27]

    Scargle, J. D. 1979, Studies in astronomical time series analysis: Modeling ran- dom processes in the time domain, V ol. 81148 (National Aeronautics and Space Administration, Ames Research Center)

    1979

  20. [28]

    Scargle, J. D. 1982, ApJ, Part 1, vol. 263, Dec. 15, 1982, p. 835-853., 263, 835

    1982

  21. [29]

    Scargle, J. D. 2020, ApJ, 895, 90

    2020

  22. [30]

    2018, in Circuits at the Nanoscale (CRC Press), 95–115

    Schmid, H. 2018, in Circuits at the Nanoscale (CRC Press), 95–115

    2018

  23. [31]

    & Schmitz, A

    Schreiber, T. & Schmitz, A. 2000, Phys. D: Nonlinear Phenom., 142, 346

    2000

  24. [32]

    1984, ApJ, Part 1 (ISSN 0004-637X), vol

    Simonetti, J., Cordes, J., & Spangler, S. 1984, ApJ, Part 1 (ISSN 0004-637X), vol. 284, Sept. 1, 1984, p. 126-134. Research supported by the National As- tronomy and Ionosphere Center., 284, 126

    1984

  25. [33]

    M., Takahashi, T., et al

    Tanihata, C., Urry, C. M., Takahashi, T., et al. 2001, ApJ, 563, 569

    2001

  26. [34]

    S., & Rubin, D

    Theiler, J., Linsay, P. S., & Rubin, D. M. 1993, arXiv preprint comp-gas/9302003

    1993 arXiv

  27. [35]

    Thorne, K. S. & Blandford, R. D. 2021, Statistical Physics: V olume 1 of Modern Classical Physics (Princeton University Press)

    2021

  28. [36]

    & Koenig, M

    Timmer, J. & Koenig, M. 1995, A&A, v. 300, p. 707, 300, 707

    1995

  29. [37]

    2000, A&A, 361, 850

    Turler, M., Courvoisier, T.-L., & Paltani, S. 2000, A&A, 361, 850

    2000

  30. [38]

    Uhlenbeck, G. E. & Ornstein, L. S. 1930, Phys. Rev., 36, 823

    1930

  31. [39]

    2002, MNRAS, 332, 231

    Uttley, P., McHardy, I., & Papadakis, I. 2002, MNRAS, 332, 231

    2002

  32. [40]

    VanderPlas, J. T. 2018, ApJS, 236, 16

    2018

  33. [41]

    2005, A&A, 431, 391

    Vaughan, S. 2005, A&A, 431, 391

    2005

  34. [42]

    2003, MNRAS, 345, 1271

    Vaughan, S., Edelson, R., Warwick, R., & Uttley, P. 2003, MNRAS, 345, 1271

    2003

  35. [43]

    2016, MNRAS, 461, 3145

    Vaughan, S., Uttley, P., Markowitz, A., et al. 2016, MNRAS, 461, 3145

    2016

  36. [44]

    Whittle, P. 1957, J. R. Stat: Series B (Methodological), 19, 38

    1957

  37. [45]

    & Kürster, M

    Zechmeister, M. & Kürster, M. 2009, A&A, 496, 577 Article number, page 14 of 14

    2009

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.