{"id":"36c87366-703f-4717-a731-148a2b18d475","arxiv_id":"2502.10326","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"TESS fast-cadence light curves contain a high-frequency blue noise component, likely from spacecraft jitter, that standard linear decorrelation does not fully correct.","lead":"The paper reports a high-frequency noise component, called blue noise, in TESS's 20-second-cadence starlight data, using the star WASP-167 as the main example. It suggests the noise comes from the satellite's tiny pointing jitter and warns that simple position-based corrections do not always remove it.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"PSD estimator and detrending steps are unspecified; the measured γ = −0.401 ± 0.006 may be an artifact of spectral leakage or aliasing rather than a real blue noise component.","rationale":"The reader's weakest_assumption identifies exactly the load-bearing concern: the high-frequency PSD slope may be an estimator artifact, because the PSD method is unspecified and the assumption that only white noise should appear at high frequencies is not guarded by any null test. The strongest_claim (blue noise from pointing jitter) depends on that slope being a faithful measurement. My independent reading confirms this: the paper gives no details of the spectral estimation (window, binning, averaging, detrending) and the error bar cannot be trusted without knowing the number of independent estimates and fit range. The additional segment analysis (Eqs. 7–9) is secondary and does not rescue the central claim; it only supports the mitigation discussion. No internal contradiction or circularity is present, and the data are public, so the concern is testable. The appropriate verdict remains CONDITIONAL: the central detection is interesting and plausible, but it must be validated with a reproducible PSD computation and a null test before the blue noise claim can be accepted. I agree with the reader's assessment, with no new concern rising to the level of REJECT.","tokens_in":4647,"tokens_out":1429,"duration_ms":13025,"concrete_test":"Reproduce the PSD computation from the public TESS SAP light curve (MAST DOI 10.17909/T9-ST5G-3177) using a stated method: apply a Hann window and compute the Lomb-Scargle or FFT periodogram on the 20 s cadence ΦSAP, with and without the trend/transit model removed, then fit γ over the same high-frequency range. If the fitted γ changes by more than 0.1 when the window or detrending is varied, or if a null test on simulated white noise plus a red-noise component reproduces a similar rising tail, the measured blue noise slope is not robust.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim rests entirely on the measured high-frequency PSD slope γ = −0.401 ± 0.006 (Section 3). The paper does not state how the PSD was computed: no windowing, no segment averaging, no frequency binning, no Nyquist handling, and no description of any detrending applied before the transform. This matters because the high-frequency regime (timescales shorter than 5 min) sits near the 20 s cadence Nyquist frequency, where spectral leakage from the strong low-frequency red noise (γ ≈ 0.86) can artificially raise the estimated slope unless a proper window and/or prewhitening are used. An unstated detrending step (e.g., removing a polynomial or median filter) would similarly modify the high-frequency tail. The quoted error bar of ±0.006 implies a precise fit, but with no specification of the number of independent PSD estimates, the number of fit points, or whether the fit accounts for correlated residuals, this uncertainty is not interpretable. If the rising tail is an estimator artifact, the jitter interpretation and the mitigation claims lose their empirical basis. The claim is plausible but currently unfalsifiable from the text alone.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the detection of a high-frequency noise component in the 20-second-cadence TESS SAP light curve of WASP-167. From a power spectral density (PSD) analysis, the authors measure a high-frequency slope of gamma = -0.401 +/- 0.006, which they call \"blue noise\" and attribute to spacecraft pointing jitter acting through subpixel sensitivity variations. They also present a bilinear decorrelation against the photo-center position as a mitigation technique, and argue from three flux-versus-distance segments that the noise properties change over time, so that simple linear corrections may be inadequate.","tokens_in":4873,"tokens_out":4584,"duration_ms":44885,"significance":"If the detection is robust, it identifies a correlated noise component in TESS fast-cadence photometry that standard pink-noise filters (e.g., Carter & Winn 2009) do not model. The analysis has a notable strength: the decorrelation uses independently measured photo-center positions as the regressor, so the mitigation claim is not circular. The paper also usefully reports that similar high-frequency features appear in some other 20-s light curves (WASP-121, KELT-25) but not in HD 189733, suggesting an instrumental rather than astrophysical origin. However, the central quantitative claim rests on a PSD slope whose estimation procedure is not described, and the quoted uncertainty is not justified. The result is plausible but requires a substantially more detailed methods presentation before it can be accepted.","major_comments":[{"comment":"The manuscript does not specify how the PSD in Figure 1 was computed: no windowing, no segment averaging, no frequency binning, no treatment of the 20-s Nyquist frequency, and no description of any detrending applied before the transform. Because the low-frequency PSD has a strong red slope (gamma ~ 0.86) and the high-frequency domain is adjacent to the Nyquist frequency, spectral leakage and aliasing can bias the high-frequency slope. The measured value gamma = -0.401 +/- 0.006 is not interpretable without this information. Please describe the estimator and demonstrate with simulated red noise, using the same sampling and any pre-processing, that the measured slope is not an artifact of the method.","section":"Section 3, PSD computation"},{"comment":"The quoted uncertainty of +/- 0.006 on the high-frequency slope implies a very precise fit, but the paper provides no information on the number of independent spectral estimates, the number of fit points, or whether correlated residuals in the PSD were accounted for. The same applies to the low-frequency slopes (+/- 0.033). Please report the fitting details and provide robust uncertainties, for example from bootstrap resampling or noise-only simulations.","section":"Section 3, error bars"},{"comment":"According to the paper's own definition in Section 1, blue noise corresponds to gamma = -1, while the measured high-frequency slope is gamma = -0.401 +/- 0.006. This value is far from -1 (the difference is many times the quoted uncertainty) and is not blue noise by the stated definition. Calling it \"blue noise\" in the title and abstract overstates the result. Please either adjust the terminology to something like \"blue-noise-like\" or \"rising high-frequency noise\", or explicitly model the slope as a mixture of white and a steeper blue component.","section":"Section 3, labeling"},{"comment":"The claim that \"the jitter correction can mitigate the high-frequency noise component\" is supported only by visual inspection of the lower-right panel of Figure 1. The paper does not report the high-frequency slope after the correction or any quantitative test for the reduction. Please quantify the change, e.g., the post-correction gamma in the high-frequency band or the ratio of integrated PSD power in that band, so the mitigation claim can be evaluated.","section":"Section 3, jitter correction"},{"comment":"The three segments in the Phi_SAP versus log delta plane (Eqs. 7-9) appear to be chosen post hoc, and the paper does not provide a formal test for whether the slope differences are statistically significant or whether a smooth time-varying relation would be preferred. Since the conclusion that the noise properties change over time rests on these segments, please state how the segment boundaries were selected and provide a significance test or model comparison. There is also a typographical error in the second segment bound ('BJD 2640053' should likely be 'BJD 2460053').","section":"Section 3, segmented flux-distance relation"}],"minor_comments":[{"comment":"There is a typo in the abstract: \"jiiter\" should be \"jitter\".","section":"Abstract"},{"comment":"In the paragraph on blue noise, \"blue nose\" should be \"blue noise\", \"shorter then\" should be \"shorter than\", and \"occuring\" should be \"occurring\".","section":"Introduction"},{"comment":"Equation (4) is typeset ambiguously: the factor f appears to be in the wrong position, and the symbol f is not defined before its use. Please define FLFCSAP clearly and write the equation with an unambiguous division by f.","section":"Section 2, Eq. (4)"},{"comment":"The star HD 189733 is written as \"HD 1897333\" in the text; the extra digit should be removed.","section":"Section 3"},{"comment":"The reference to MAST (2021) does not give the specific DOI or a clear pointer to the data product used; please provide the exact archive identifier.","section":"Section 3"},{"comment":"The text uses both \"delta\" and \"d\" for the photo-center distance; please make the notation consistent in the figure caption and the body.","section":"Figure 1"}],"recommendation":"major_revision","confidential_remarks":"The paper is very short and reads like a research note, which may be appropriate for the claimed detection. However, the central measurement is underdocumented to the point that a reader cannot reproduce or validate it. I would recommend asking the authors to provide a complete description of the PSD estimation and noise simulations, and to soften the 'blue noise' label to match the measured slope. The supporting claims about jitter mitigation and time-varying noise also need quantitative support. If the authors can supply those details, the result could be a useful contribution to TESS data analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The punchline: this is a short empirical note claiming that TESS 20-second SAP light curves contain a high-frequency noise component whose PSD rises toward the Nyquist limit (gamma ~ -0.4), attributed to spacecraft jitter acting through subpixel sensitivity variations. That's a new claim as far as I know, and if real it matters for anyone doing transit or variability work with fast-cadence TESS data.\n\nWhat's genuinely good: the detection is based on a real light curve (WASP-167, Sector 64) and the MAST data are public, so the result is in principle checkable. They measure a high-frequency slope that is clearly distinct from white noise, and they show that a standard bilinear decorrelation against photo-center position removes most of the feature. They also demonstrate that a single linear correction is not enough: the relation between flux and photo-center distance changes across three time segments, with significantly different slopes. That last point is a useful warning for anyone using simple decorrelation. They also report that similar features appear in WASP-121 and KELT-25 but not HD 189733, which is a decent sanity check.\n\nThe soft spots are real but not fatal. The big one: the paper doesn't say how the PSD was computed. No windowing, no binning, no averaging, no mention of detrending before the transform, no Nyquist handling. Since the claim rests on the high-frequency slope, that's a load-bearing omission. The quoted error of +/- 0.006 on gamma implies a precise fit, but without details on the number of independent estimates or how the fit was done, that uncertainty isn't interpretable. The stress-test worry--spectral leakage from the strong low-frequency red noise--is plausible enough that the result needs a proper null test (e.g., a synthetic light curve with the same red noise and white noise, run through the same estimator). The 'blue noise' label is loose; gamma = -0.4 is not gamma = -1, though they do say 'hints at,' so that's acceptable for a short note. There's also a typo in the segment boundaries (line 8 has 2640053 instead of 2460053), and the boundaries themselves look post hoc. None of this contradicts the data, but it means the paper is a proof-of-suggestion, not a proof-of-detection.\n\nWho it's for: TESS fast-cadence users, especially exoplanet transit and stellar variability people. It deserves a serious referee, not a desk reject, because the claim is new, checkable, and potentially important for systematic error budgets. A referee should ask for the PSD method, a null test, and uncertainty that accounts for correlated residuals.\n\nNet: I'd engage with it--read it, maybe cite it as a caution--but I wouldn't rely on the quantitative slope until the methods are pinned down.","headline":"A short, useful empirical note on a possible jitter-induced high-frequency noise component in TESS 20s data, but the PSD methods need to be specified before the detection is solid.","tokens_in":5434,"tokens_out":2131,"would_cite":true,"duration_ms":19842,"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":"TESS's fastest photometry of WASP-167 contains a blue-noise component that rises toward high frequencies, attributed to spacecraft pointing jitter.","keywords":["TESS","20-second cadence","SAP photometry","blue noise","power spectral density","pointing jitter","subpixel sensitivity","WASP-167"],"falsifier":"Recompute the power spectrum of the same WASP-167 20-second SAP light curve with a fully documented estimator, such as Welch's method with explicit windowing, binning, and frequency bounds, and check whether the high-frequency slope remains near $\\gamma \\approx -0.4$; if the rising tail disappears once spectral leakage, aliasing, or the detrending choice is accounted for, the blue-noise detection would not survive.","tokens_in":4433,"feed_emoji":"🛰️","tokens_out":10329,"duration_ms":88687,"temperature":0.7,"pith_summary":"This paper reports a new noise component in TESS's fastest light curves. In the 20-second SAP photometry of WASP-167 from Sector 64, the power spectral density rises toward high frequencies with slope $\\gamma = -0.401 \\pm 0.006$, instead of flattening to the white-noise level expected at short timescales. The authors attribute this 'blue noise' to spacecraft pointing jitter acting through subpixel sensitivity variations on the CCD, and they show that decorrelating against the measured photo-center position weakens the excess. They also find that a single bilinear correction is unreliable because the relation between flux and pointing offset changes across three segments of the light curve. If the detection holds, TESS fast-cadence photometry contains a correlated noise term that standard filters designed for $\\gamma=1$ pink noise do not model.","feed_headline":"Blue noise from pointing jitter found in TESS fast light curves","feed_subtitle":"The power spectrum rises at short timescales, so 20-second TESS data carry a noise that standard filters do not model.","key_machinery":"The central object is the exponent $\\gamma$ in the power-law model $S(f) \\propto f^{-\\gamma}$, which classifies noise as white ($\\gamma=0$), pink ($\\gamma=1$), or blue ($\\gamma=-1$). The paper measures $\\gamma$ separately in the low- and high-frequency domains of the WASP-167 PSD and compares the high-frequency value with the white-noise expectation. The explanatory mechanism is spacecraft pointing jitter: the pixel coordinates of the photo-center, tracked by the TESS pipeline, serve as a proxy for attitude, and the paper tests a bilinear decorrelation of flux against those coordinates. To show why that correction is insufficient, it fits $\\Phi_{\\mathrm{SAP}}$ against $\\log \\delta$, the logarithmic distance of the photo-center from its mean position, in three separate time segments and finds significantly different slopes.","core_discovery":"The paper's central claim is that the 20-second SAP light curve of WASP-167 contains a high-frequency noise component whose power spectral density increases toward short timescales. Fitting the high-frequency end of the PSD gives $\\gamma = -0.401 \\pm 0.006$, which the authors interpret as blue noise. They argue that the most plausible source is spacecraft pointing jitter acting through subpixel sensitivity variations: using the photo-center coordinates as a proxy for pointing, a bilinear decorrelation removes much of the high-frequency excess. The paper further demonstrates that the jitter response is not stationary, because linear fits of flux against $\\log \\delta$ in three successive time intervals have markedly different slopes, so a fixed polynomial correction cannot fully clean the data. The feature is also reported in the 20-second light curves of WASP-121 and KELT-25 and absent in that of HD 189733, placing the detection between a universal artifact and a one-off anomaly.","pith_inferences":["Beyond the paper, if subpixel sensitivity variations are the cause, the blue-noise amplitude should depend on where the point-spread function lands on the CCD, so comparing TESS sectors with different pointing dither patterns would provide a direct confirmation of the mechanism.","The measured slope sits between white noise and the $\\gamma = -1$ definition of blue noise, so the 'blue noise' label should be read loosely; a mixed model of white noise plus a jitter-correlated component might reproduce the observed $\\gamma \\approx -0.4$ without invoking a true blue-noise process.","A practical extension suggested by the paper's time-segment analysis is a segmented or time-varying decorrelation whose breakpoints could be matched against spacecraft events such as momentum dumps, which the authors do not explore.","For exoplanet work, the 20-second light curves should undergo a jitter-decorrelation step before searching for very short-timescale astrophysical signals such as occultations or flares; the paper notes this application remains open."],"forward_implications":["Analyses of TESS 20-second SAP light curves that assume only white noise at high frequencies will underestimate the true noise level for stars like WASP-167.","Wavelet-based noise filters that fix the correlated-noise slope at $\\gamma = 1$ will not absorb a component with $\\gamma \\approx -0.4$; allowing the slope to float risks degeneracies and overfitting.","Binning the light curve acts as a low-pass filter and can suppress the blue-noise excess, at the price of losing the short-timescale information that motivated the 20-second cadence.","A single global bilinear decorrelation against photo-center position is not a complete fix, because the flux–offset relation changes between segments of the light curve.","The presence of the feature in WASP-121 and KELT-25 but not in HD 189733 shows that the effect is neither universal nor unique to one target."],"supporting_citations":[{"why":"Provides the TESS mission description and establishes the 20-second cadence data product analyzed here.","marker":"Ricker et al. 2015"},{"why":"Supplies the TESS fast light-curve archive from which the WASP-167 SAP data were obtained.","marker":"MAST 2021"},{"why":"Presents the WASP-167 system, the target whose Sector 64 light curve is analyzed.","marker":"Temple et al. 2017"},{"why":"Introduces the position-decorrelation technique that the paper's bilinear jitter correction adapts.","marker":"Vanderburg & Johnson 2014"},{"why":"Provides the wavelet-based noise filter whose fixed $\\gamma = 1$ assumption makes the blue-noise component problematic.","marker":"Carter & Winn 2009"},{"why":"Establishes the pink-noise framework and its impact on exoplanet light-curve modeling that the paper extends to blue noise.","marker":"Pont et al. 2006"}],"fun_headline_variants":["TESS fast light curves reveal new blue noise from jitter","Pointing jitter leaves blue-noise fingerprint in TESS fast data","Blue noise in TESS 20-sec light curves traced to jitter","TESS fast data carry blue noise; jitter is the source","New blue noise in TESS fast light curves blamed on jitter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The high-frequency PSD slope is a faithful measurement of the noise rather than an artifact of the estimator, since the paper does not specify the windowing, binning, averaging, or exact frequency range used to compute the spectrum.","fun_headline_variants_meta":{"raw":{"variants":["TESS fast light curves reveal new blue noise from jitter","Pointing jitter leaves blue-noise fingerprint in TESS fast data","Blue noise in TESS 20-sec light curves traced to jitter","TESS fast data carry blue noise; jitter is the source","New blue noise in TESS fast light curves blamed on jitter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000706,"raw_usage":{"total_tokens":3141,"prompt_tokens":861,"completion_tokens":2280,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":477,"completion_tokens_details":{"reasoning_tokens":2202}},"tokens_in":477,"tokens_out":2280,"duration_ms":13902,"temperature":1.0,"reasoning_tokens":2202,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T18:32:29.497442+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the power spectrum of the same WASP-167 20-second SAP light curve with a fully documented estimator, such as Welch's method with explicit windowing, binning, and frequency bounds, and check whether the high-frequency slope remains near $\\gamma \\approx -0.4$; if the rising tail disappears once spectral leakage, aliasing, or the detrending choice is accounted for, the blue-noise detection would not survive.","supporting_citations":[],"review_version":1}