{"id":"682ab6fe-9ab8-41a2-a8ed-ce70a36597f2","arxiv_id":"2506.22131","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":10,"one_line_summary":"A catalog of 61 short-period quasi-periodic pulsations in 20-second TESS stellar flares is presented, along with a reported scaling where pulse period grows with flare duration to the 0.33 power in a selected branch.","lead":"Using 20-second TESS data, this study found 61 quasi-periodic pulsations in stellar flares with periods of 42 to 193 seconds, and reports that a selected branch of these pulsations has periods that grow with flare duration. It is the largest short-period QPP catalog from TESS so far and may strengthen the case that stellar and solar flare oscillations behave similarly.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The period-duration scaling relies on a branch selected post hoc from the same P-tau plane being regressed, and the AFINO detection window itself can create a spurious positive trend, so Eq. 16 is not yet supported.","rationale":"Good-faith reading: the paper is primarily a catalog paper, and its flare/QPP detections in 20 s TESS data fill a real gap; the AFINO method is established and the occurrence rate near 1.6 percent is plausible relative to prior work. The flare catalog power laws (alpha = 1.65 for energy, 2.62 for equivalent duration) are broadly consistent with literature. Credit: no machine-checked proofs or released code, only tables on GitHub, so the analysis relies entirely on the description in the text. The main conclusion, however, is the scaling law in Eq. 16, and that is the weakest link. The reader identified the same concern: post-hoc visual branch selection and lack of a detection-bias correction. I sharpen it by noting that the K-means input variables are identical to the regression variables, making the p-values circular, and that the AFINO search window imposes P less than or similar to a fraction of the flare duration as a natural truncation; both effects can generate a spurious positive slope. The authors themselves caution that the scaling should be interpreted with caution and needs more study, which supports treating the catalog as sound and the scaling as unconfirmed. Eq. 15 also appears to have a sign error in the intercept (negative 2.59 rather than positive 2.59 given the plotted log-duration units), a minor but concrete inconsistency. Verdict unchanged: CONDITIONAL.","tokens_in":19431,"tokens_out":6517,"duration_ms":74625,"concrete_test":"Permutation/recovery test: under the null that P and tau are independent, assign the 61 observed QPP periods at random to the 61 observed flare durations, then rerun the full Sect. 4.2 branch selection (K-means K=4, merge the two clusters with similar slopes) and extract the slope/Spearman coefficient of the selected branch. Repeat at least 1000 times and compare the observed slope 0.33 and Spearman 0.42 to the permutation distribution; report the fraction of null trials that produce a branch slope at least 0.25. If that fraction exceeds about 5 percent, the headline correlation is a selection artifact. An even stronger version would inject synthetic oscillations with independent P and tau into simulated 20 s-cadence flares and pass them through AFINO before clustering, so that the detection-boundary effect is included.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The full sample shows no P-duration correlation (r = -0.042, p = 0.767), so the headline result (Sect. 4.2, Eq. 16, P proportional to tau^(0.33 +/- 0.08)) comes only from a branch that the authors identify 'on visual inspection' of Fig. 7(b)/Fig. 9 and then isolate with K-means clustering (K=4). The cluster labels are computed from the same two variables (log P and log tau) that are later regressed, and the p-values reported for the selected clusters do not include this selection step. Under a null model with no physical P-tau relation, the same K-means plus 'steepest pair of adjacent clusters' recipe will still find a positive slope, so the significance is overstated. A second, independent selection effect compounds this: AFINO requires several resolved cycles inside the flare window, and the search window is bounded between the 40 s Nyquist period and 300 s. For short flares only short periods are recoverable, so even with independent P and tau the detected subset will show P increasing with tau. The robustness check that removes QPPs below 60 s (Eq. 18) repeats the same branch-selection procedure and therefore does not break the circularity. The catalog of 61 QPPs is valuable, but Eq. 16 has not been shown to be astrophysical.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper analyzes 20-second cadence TESS light curves from sectors 27 to 80. It uses an ARIMA-based automated flare detector to report 3878 flares on 1285 stars and then applies the AFINO Fourier model-comparison method to identify 61 quasi-periodic pulsations in 57 stars, with periods between 42 and 193 seconds. The paper's central quantitative claim is a positive scaling between QPP period and flare duration, P ∝ τ^(0.33±0.08), found in a visually identified branch of the period-duration diagram and interpreted as analogous to solar-flare QPP scaling.","tokens_in":19741,"tokens_out":6378,"duration_ms":66280,"significance":"If validated, the claimed period-duration scaling would connect short-period stellar QPPs to the well-studied solar-flare QPP phenomenology and would provide a new observational constraint on QPP mechanisms. The catalog of 61 short-period stellar QPPs is itself a useful addition to the literature, and the paper is transparent about several limitations, including the 4-hour flare-duration cutoff and AFINO's conservative detection behavior. The availability of the detection tables in a public repository is a strength. However, the headline scaling relation is currently supported only by a post hoc selected subset of the data, while the full sample shows no correlation; the statistical demonstration therefore does not yet match the strength of the astrophysical claim.","major_comments":[{"comment":"The scaling law in Eqs. (15)-(16) is derived from a branch that the authors identify 'on visual inspection' and then isolate with K-means clustering on the same two variables (log P and log τ_flare) that are later regressed. The full sample shows no correlation (r = -0.042, p = 0.767), so the positive result exists only after selection. The p-values quoted for the branch (Spearman p = 0.0112, Pearson p = 0.0000) do not include the selection step, and under a null model with no physical P-τ relation the same 'visual branch plus K-means' recipe will still produce positive slopes. Please supply a pre-specified selection rule, a simulation-based null that includes the clustering step, or an out-of-sample validation before Eq. (16) can be taken as an astrophysical scaling law.","section":"4.2, Fig. 7(b), Fig. 9, Eq. (15)-(16)"},{"comment":"AFINO's search is restricted to periods between 40 s (Nyquist) and 300 s and requires several resolved cycles inside the flare window. This creates a detection selection effect: short flares can only contain short-period oscillations, so even if P and τ_flare are independent, the detected subset will show P increasing with τ_flare. The period cut at 60 s and the K=2 re-clustering leading to Eq. (18) do not remove this effect, because the same branch-selection recipe is applied and the search-window constraint is unchanged. Please quantify the expected null trend by injecting synthetic oscillations with known periods into the actual flare duration distribution and running the full AFINO plus clustering pipeline.","section":"4.1 and 4.2"},{"comment":"The choice K=4 is asserted as 'optimal' without a quantitative criterion, and clusters 1-2 and 3-4 are grouped after inspecting the slopes; the reported Spearman and Pearson p-values are not corrected for this model selection. The stability of the grouping should be assessed (e.g., silhouette or gap statistic, bootstrap resampling), and the two-branch regression should be reported together with the full-sample regression so that the sensitivity of Eq. (16) to the cluster definition is visible to the reader.","section":"4.2, K-means procedure"}],"minor_comments":[{"comment":"The intercept in Eq. (15) is printed with a minus sign (-2.59), while Eq. (18) has +2.67 and the plotted range in Fig. 9 requires a positive intercept when τ_flare is in days; please correct the sign in Eq. (15) or state the units explicitly.","section":"Eq. (15) vs Eq. (18)"},{"comment":"The sentence 'we restricted the frequency fp to be ≤ 300 s' should read 'period' rather than 'frequency' and the units should be seconds; similarly, 'below the Nyquist frequency of 0.025 Hz or equivalently 40 s' should be phrased as 'periods shorter than 40 s' for clarity.","section":"4.1"},{"comment":"The Pearson p-value is reported as 0.0000; this should be reported as p < 0.0001.","section":"4.2"},{"comment":"The statement that the stellar slope 0.33 is similar to the solar slope 0.67 (Hayes et al. 2020) is not supported by overlap of the quoted values; please qualify this comparison or provide a quantitative test of similarity.","section":"5 and 4.2"},{"comment":"The 4-hour duration cut is acknowledged to risk rejecting real megaflares; please state how many candidate flares are removed by this cut and whether any of the 61 QPP detections are near the cut, since this affects the interpretation of the duration distribution.","section":"3.1.2 and 4.2"},{"comment":"The data availability statement points to a GitHub repository, but no version or DOI is given; archiving the exact tables and pipeline version would improve reproducibility.","section":"Data availability"}],"recommendation":"major_revision","confidential_remarks":"The catalog contribution and the AFINO application to 20-second TESS data are likely publishable, but Section 4.2's scaling-law claim should be either substantially re-analyzed with selection-free statistics or downgraded from the headline. If the authors can show through injection-recovery simulations that the branch survives the AFINO selection effect and the K-means selection step, a major revision is appropriate; otherwise the scaling claim should be removed or presented only as a tentative, selection-sensitive trend."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me cut to the chase. This is a useful paper for the catalog it ships: 61 QPPs with periods 42–193 s in 20-s TESS data is a real step beyond the 9 and 17 QPPs in the earlier TESS samples. The flare detection pipeline is standard (ARIMA + thresholding), but the authors are honest about its biases — the 4-hour cut, the 160-s minimum flare duration, the low occurrence rate. The QPP detection uses AFINO, a well-tested method. The paper also reports null correlations between QPP period and stellar parameters, which is a useful negative result. I trust the catalog.\n\nWhere I part ways with the authors is the headline scaling law. The full sample shows no period–duration correlation (r=-0.042, p=0.767). The positive relation, P ∝ τ^0.33, only appears after the authors visually identify a branch in the P–τ plane and isolate it with K-means. That selection step is not folded into the significance. Under a null model, the same recipe (K-means + pick the steepest adjacent clusters) will produce a positive slope. The stress-test note is right about this.\n\nThere is also a second, independent selection effect the paper does not address: AFINO searches periods between 40 s (Nyquist) and 300 s, and a short flare simply does not have enough cycles for a long period to be detected. So even if P and τ were physically independent, the detected subset will show P increasing with τ. The robustness check that drops QPPs below 60 s repeats the same branch-selection procedure and therefore does not break the circularity. The authors themselves caution against overinterpreting the scaling, which I appreciate, but the caution is in the conclusions, not in the method.\n\nSo my verdict: the catalog is sound and worth publishing; the scaling-law claim should be clearly labeled as tentative or removed from the abstract. The comparison to solar flare scaling (Hayes et al. 2020, slope 0.67) is premature when the stellar slope 0.33 comes from a post hoc subset plus a detection bias.\n\nWho should read this? Anyone working on stellar flare QPPs, especially the short-period regime. The catalog alone makes it worth referee time. This paper should go to peer review; a good referee will ask for a simulation-based null test of the branch selection and a correction for the detectability window. But it deserves that attention.","headline":"A genuinely useful short-period QPP catalog, but the P–tau scaling law is not established; the branch is selected post hoc and the detection window itself can create the trend.","tokens_in":20346,"tokens_out":1737,"would_cite":true,"duration_ms":17748,"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":"TESS data show stellar flare pulsations scale with flare duration, mirroring the Sun.","keywords":["quasi-periodic pulsations","stellar flares","TESS 20-second cadence","AFINO","ARIMA flare detection","QPP period–duration scaling","solar–stellar flare analogy"],"falsifier":"Run an injection-recovery experiment: embed synthetic QPPs with known periods and randomly chosen flare durations into real TESS 20-second light curves, run the same ARIMA+AFINO+K-means pipeline, and check whether a $P \\propto \\tau^{0.33}$ branch emerges when no true period–duration correlation exists. If it does, the observed branch is a selection effect; if not, the correlation is supported.","tokens_in":19192,"feed_emoji":"⭐","tokens_out":15130,"duration_ms":138092,"temperature":0.7,"pith_summary":"Using TESS 20-second cadence data from sectors 27–80, this study identifies 3878 flares on 1285 stars and, within 61 of those flares on 57 stars, quasi-periodic pulsations (QPPs) with periods of 42–193 seconds — a short-period regime that earlier Kepler- and TESS-based stellar surveys largely missed. The central claim is that a branch of these QPPs follows a power-law relation between oscillation period and flare duration, $P \\propto \\tau_{\\mathrm{flare}}^{0.33 \\pm 0.08}$, meaning longer-duration flares host longer-period pulsations. The authors argue that this mirrors the scaling seen in solar-flare QPPs, supporting the idea that stellar and solar QPPs share a common physical mechanism. If correct, the result makes QPP period a usable diagnostic of flare size and opens a route to seismic probing of stellar flare loops.","feed_headline":"Longer stellar flares host longer-period pulsations","feed_subtitle":"TESS 20-second data find 61 pulsations of 42–193 s; longer flares host longer periods, as on the Sun.","key_machinery":"The argument is carried by AFINO (Automated Flare Inference of Oscillations), a Fourier-domain model-comparison test that fits three models to a flare's power spectrum — a single power law, a power law with a Gaussian bump, and a broken power law — and requires a Bayesian information criterion difference of $\\Delta\\mathrm{BIC} > 10$ to accept the bump model. Because AFINO assumes oscillations whose amplitude and period stay roughly constant over the window, the detected QPPs are 'stationary' by construction, and short-period oscillations are favored. Flare candidates come from an ARIMA residual search that flags excursions above $3\\sigma$ lasting at least eight 20-second points and validates them with fast-rise/exponential-decay shape checks. The branch correlation is produced by K-means clustering in the period–duration plane followed by Pearson and Spearman tests and a Bayesian linear regression in log–log space.","core_discovery":"On the paper's own terms, the central discovery is that short-period stellar QPPs can be surveyed statistically in 20-second TESS observations, and that the recovered period–duration diagram contains a distinct branch — selected after visual inspection and isolated with K-means clustering — in which the QPP period grows with flare duration. For that branch the paper reports a Bayesian linear fit in log–log space of $\\log P = (0.33 \\pm 0.08)\\log\\tau_{\\mathrm{flare}} - (2.59 \\pm 0.17)$, i.e. $P \\propto \\tau_{\\mathrm{flare}}^{0.33 \\pm 0.08}$, with Pearson $r = 0.69$ ($p < 0.0001$) and Spearman $\\rho = 0.42$ ($p = 0.011$). Restricting the analysis to QPPs with periods above 60 seconds leaves 34 events and reproduces the trend with slope $0.34 \\pm 0.10$. The authors interpret the similarity of this slope to the solar-flare scaling reported in earlier work as evidence that stellar and solar QPPs are governed by analogous processes, while cautioning that the correlation appears for a branch rather than the full sample.","pith_inferences":["The slope should be reproducible in later TESS sectors; applying the same ARIMA+AFINO pipeline to untapped sectors provides a genuine out-of-sample test of the branch.","If the scaling is physical, higher-cadence observations such as 2-second data should reveal the same branch extending below the 40-second Nyquist limit of this survey.","The two ~60-second QPPs on the B-type subdwarf motivate radial-velocity monitoring to distinguish an intrinsic oscillator from a companion-driven signal.","A natural extension the paper implies but does not perform is to inject synthetic QPPs with known periods and durations into real TESS light curves to map the detection efficiency and quantify selection effects."],"forward_implications":["If the scaling is real, future 20-second TESS surveys should find that longer flares systematically host longer-period QPPs, making QPP period a predictor of flare duration.","The 42–193 second period range becomes accessible to routine optical stellar-flare studies, while surveys at 2-minute cadence are expected to miss most of these events.","The solar-like slope supports treating stellar and solar QPPs as the same phenomenon, which would extend solar flare-loop seismology to other stars.","The 61-event catalog offers a benchmark for comparing QPP detection methods, since AFINO is conservative and detects different events than wavelet- or network-based searches.","The repeated ~60-second period in two flares of one B-type subdwarf points to a stable oscillator worth targeted follow-up."],"supporting_citations":[{"why":"Introduces the AFINO Fourier model-comparison method used to detect QPPs in flare power spectra.","marker":"Inglis et al. (2015)"},{"why":"Refines AFINO and reports solar QPP periods that the paper compares its 42–193 s stellar periods against.","marker":"Inglis et al. (2016)"},{"why":"Blind test across QPP detection methods showing AFINO has the lowest false-alarm probability, motivating its use here.","marker":"Broomhall et al. (2019)"},{"why":"Supplies the flare template and fast-rise/exponential-decay asymmetry checks used to validate flare candidates.","marker":"Davenport et al. (2014)"},{"why":"Provides the solar QPP period–duration scaling that the paper's stellar branch is compared with and whose Bayesian regression approach it follows.","marker":"Hayes et al. (2020)"},{"why":"GOES X-ray solar flare studies that first reported the QPP period–flare duration correlation the paper mirrors.","marker":"Pugh et al. (2017, 2019)"},{"why":"Kepler QPP survey whose period range and occurrence rate are the main comparison for the short-period stellar QPPs found here.","marker":"Pugh et al. (2016)"},{"why":"Previous TESS QPP study using 20-second cadence data whose detections and period range this survey extends.","marker":"Howard & MacGregor (2022)"},{"why":"Describes the TESS mission whose 20-second cadence mode provides the data analyzed in this work.","marker":"Ricker et al. (2014)"}],"fun_headline_variants":["20-second TESS data spot sub-minute flare pulsations","Stellar QPPs scale with flare duration, akin to solar","61 flare pulsations found, periods 42-193 s","Sub-minute pulsations in stellar flares discovered"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central claim assumes that the branch picked out visually and by K-means is a real physical subgroup rather than a chance arrangement, and that the period–duration trend is not a detection bias in which short flares naturally yield short periods because they contain only a few oscillation cycles.","fun_headline_variants_meta":{"raw":{"variants":["20-second TESS data spot sub-minute flare pulsations","Stellar QPPs scale with flare duration, akin to solar","61 flare pulsations found, periods 42-193 s","Sub-minute pulsations in stellar flares discovered"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00082,"raw_usage":{"total_tokens":3656,"prompt_tokens":1081,"completion_tokens":2575,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":697,"completion_tokens_details":{"reasoning_tokens":2508}},"tokens_in":697,"tokens_out":2575,"duration_ms":21366,"temperature":1.0,"reasoning_tokens":2508,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:10:27.932999+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run an injection-recovery experiment: embed synthetic QPPs with known periods and randomly chosen flare durations into real TESS 20-second light curves, run the same ARIMA+AFINO+K-means pipeline, and check whether a $P \\propto \\tau^{0.33}$ branch emerges when no true period–duration correlation exists. If it does, the observed branch is a selection effect; if not, the correlation is supported.","supporting_citations":[{"cited_title":"R., Ireland, J., & Dominique, M","cited_arxiv_id":null,"evidence_quote":"Introduces the AFINO Fourier model-comparison method used to detect QPPs in flare power spectra."},{"cited_title":"R., Ireland, J., Dennis, B","cited_arxiv_id":null,"evidence_quote":"Refines AFINO and reports solar QPP periods that the paper compares its 42–193 s stellar periods against."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Blind test across QPP detection methods showing AFINO has the lowest false-alarm probability, motivating its use here."},{"cited_title":"A., Inglis, A","cited_arxiv_id":null,"evidence_quote":"Provides the solar QPP period–duration scaling that the paper's stellar branch is compared with and whose Bayesian regression approach it follows."},{"cited_title":"E., Armstrong, D","cited_arxiv_id":null,"evidence_quote":"Kepler QPP survey whose period range and occurrence rate are the main comparison for the short-period stellar QPPs found here."},{"cited_title":"R., Winn, J","cited_arxiv_id":null,"evidence_quote":"Describes the TESS mission whose 20-second cadence mode provides the data analyzed in this work."}],"review_version":1}