{"id":"30b5624e-b1f5-4f65-aa8a-48bbb21a379e","arxiv_id":"2506.18989","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"P-mode oscillations in HD 142091 behave primarily as pure Doppler shifts of the average line profile, with a small, height-dependent amplitude variation across the line that is not chromatic.","lead":"This paper uses the NEID spectrograph to watch p-mode oscillations on the subgiant star HD 142091 for one night, catching stellar vibrations as they shift spectral lines. It shows the oscillations act mostly as clean Doppler shifts, with a subtle depth-dependent amplitude change that matters for separating planets from stellar noise.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 10%/25% amplitude gradient may be an artifact of the fixed-hyperparameter GP decomposition on a ~5-cycle baseline; a null simulation is needed.","rationale":"The reader's conditional verdict is appropriate. I agree that the GP separation is the weakest step, but I locate the failure mode more specifically than granulation leakage: the fixed-hyperparameter GP is validated on the wrong null and could imprint an amplitude gradient even when none exists. This is partially aligned with the reader's weakest assumption. The paper has independent support for the primary claim: raw CCF residuals resemble pure Doppler shifts, the SCALPELS injection test explains the non-detection, and the line-depth and chromatic analyses are internally consistent. However, the headline amplitude gradient and its physical interpretation rest on the GP decomposition, and the paper provides no code or data artifacts. A null simulation would settle whether the gradient is real; until then the conditional verdict stands.","tokens_in":18267,"tokens_out":9349,"duration_ms":111788,"concrete_test":"Run a null-hypothesis simulation with the exact Section 4.1.2 pipeline: generate ~500 realizations of the 181-observation cadence from the Luhn et al. (2023) GP with oscillation and granulation amplitudes set to the bulk values and A_i=1 for all seven CCF flux bins (plus photon noise), then apply the fixed-hyperparameter GP decomposition, Eq. (1) fit, and quadratic-trend significance test. If >=5% of realizations yield a >=7.2-sigma quadratic trend, the observed 10%/25% gradient is not distinguishable from a statistical artifact. As a second arm, refit the GP with free granulation/oscillation amplitudes per bin and recompute A_i; if the gradient disappears, the fixed hyperparameters are the cause.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central 'shift-driven' conclusion is supported by the raw CCF-residual comparison and by the SCALPELS injection test, and I do not object to it. The load-bearing weak point is the quantitative 10%/25% amplitude gradient and its atmospheric-height interpretation. The gradient is obtained through Eq. (1): each CCF-slice (or depth-bin) time series is reduced by a granulation posterior mean and regressed against the bulk-RV oscillation posterior mean, both coming from a two-component GP whose hyperparameters are fixed from Luhn et al. (2023). Appendix A validates the GP decomposition only on simulations drawn from that same model with bulk amplitudes; it does not validate the regime relevant to the gradient: per-bin oscillation amplitudes that differ from the fixed kernel amplitudes, per-bin noise, and a baseline of only ~5 p-mode cycles. If the fixed oscillation kernel amplitude is too high or too low for a given slice, the granulation posterior can absorb part of the true oscillatory signal and bias A_i. The paper's statement that findings are consistent without granulation subtraction is reassuring, but that simpler regression is also against the same bulk RV and does not test the null hypothesis that no gradient exists. A spurious flux/depth-dependent trend could therefore arise from the decomposition itself.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports time-resolved NEID observations of p-mode oscillations in the subgiant HD 142091, taken on a single night at 2-minute cadence with 181 usable spectra. The authors analyze CCF residuals, CCF bisector velocities in seven flux slices, a SCALPELS decomposition, and line-by-line RVs grouped by line depth and wavelength. Their central claim is that the observed p-modes manifest primarily as pure Doppler shifts of the average line profile, with a higher-order amplitude gradient across the CCF: oscillation amplitudes are about 10% larger near the CCF core and 25% smaller in the wings, interpreted as larger oscillation velocities higher in the stellar atmosphere. A similar depth dependence is found in a line-by-line analysis, and no significant chromatic dependence remains after accounting for line-depth differences. The paper also shows that SCALPELS cannot detect the small shape-driven component at the achieved CCF noise level, and it validates this with an injection-recovery test.","tokens_in":18568,"tokens_out":3877,"duration_ms":39009,"significance":"If correct, the paper provides a rare time-domain, epoch-resolved characterization of how p-mode oscillations distort stellar line profiles, with direct implications for shape-vs-shift mitigation techniques in extreme-precision RV surveys. The raw CCF residual maps in Figures 2 and 3 give robust visual evidence that the dominant response is a translational shift, and the SCALPELS injection test is a well-designed, quantitative explanation of the non-detection. The paper is also commendably transparent about the chromatic analysis, showing that the marginal wavelength trend disappears once line depth is restricted. The main quantitative result that motivates the atmospheric-height interpretation — the 10%/25% amplitude gradient — is, however, derived through the authors' GP decomposition, and that decomposition's reliability in the gradient regime is not established by the validation presented. This makes the paper's central new physical conclusion conditional on a specific modeling assumption that needs further testing.","major_comments":[{"comment":"The amplitude gradient A_i in Eq. (1) is obtained by regressing each CCF-slice bisector time series on the bulk-RV oscillation posterior mean, where both the slice-wise granulation subtraction and the bulk oscillation mean come from a two-component GP with hyperparameters fixed from Luhn et al. (2023). Appendix A validates the decomposition on time series generated from that same GP model with a single bulk amplitude; it does not test the regime that matters for the gradient, namely slices whose true oscillation amplitudes differ from the fixed kernel amplitude, combined with per-slice granulation and a baseline of only about five oscillation cycles. The paper itself states that the granulation kernel has appreciable power inside the oscillation envelope and that granulation variability on oscillation timescales is partly attributed to the oscillation component. A slice-dependent leakage of granulation power into the oscillation posterior could therefore produce a spurious A_i trend. I request a null simulation: generate per-slice time series with a constant A_i (no gradient) plus per-slice granulation and realistic noise, run the same pipeline, and show the recovered A_i has no systematic gradient. Alternatively, a joint GP fit in which each slice's oscillation amplitude is a free parameter would avoid the two-step decomposition and directly test the gradient.","section":"4.1.2 and Appendix A"},{"comment":"The line-depth amplitude trend uses the same two-step GP decomposition. The RMS trend shown in the bottom-right panel of Figure 6 is explicitly admitted to be not significant, and the significant 6.5-sigma amplitude-scale-factor trend in Figure 7 is obtained after subtracting the granulation posterior and regressing on the bulk oscillation component. Consequently, the same concern as in my first comment applies: if granulation leakage varies with line depth, the recovered scale factors could show a spurious trend. The paper should either validate the decomposition in the presence of a constant per-depth oscillation amplitude through simulation, or present an independent estimator of the per-depth oscillation amplitude (e.g., a direct fit of a shared oscillation signal with per-depth scale factors).","section":"4.2.1"},{"comment":"The sentence stating that 'our findings are consistent through either method' (with and without granulation subtraction) is presented as a robustness check, but it does not address the null hypothesis of no gradient, because the simpler regression still uses the same bulk-RV oscillation component and the same per-slice time series. This sentence should be clarified so that it is not read as evidence against decomposition-induced bias.","section":"4.1.2"}],"minor_comments":[{"comment":"The header lists 'Received Oct. 31, 2024; Accepted May 15, 2024', which is chronologically inconsistent; please verify the dates.","section":"Header"},{"comment":"The estimated p-mode period is given as '84 min.' in Section 2 but '80 min.' in the Abstract; the values should be reconciled.","section":"Section 2 and Abstract"},{"comment":"The description of the CCF flux bands ('including a gap between bands of 0.025') leaves ambiguous whether the gap is in units of flux or in velocity; please specify the units and the exact bin edges.","section":"Section 4.1.2"},{"comment":"There is a typo: 'the shallowest bin has has a slight negative trend' should read 'has a slight negative trend'.","section":"Section 4.2.1"},{"comment":"The reference list begins with an entry '1997, ESA Special Publication...' with no author; as formatted, it will be difficult for readers to locate. Please check the journal's reference style for this item.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The main reason for major revision is the unvalidated use of the fixed-hyperparameter GP decomposition to derive the quantitative amplitude gradient, which is the paper's headline physical result. The shift-driven conclusion is well supported by the raw residuals and the SCALPELS injection test, and I see no issue on that front. The requested null simulation is a standard and feasible addition, and I expect it to be decisive."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, here's my take.\n\nThe headline is that this is a genuinely new empirical result: the first time-resolved, single-star measurement of p-mode oscillation amplitude varying across the CCF bisector and with line depth, plus a quantitative SCALPELS sensitivity threshold calibrated to that specific signal shape. The raw CCF residual maps (Figures 2 and 3) make a strong case that low-degree p-modes on this subgiant behave mostly as pure Doppler shifts; the shape-driven component is a higher-order effect. That part of the paper is solid and well presented. The SCALPELS injection-recovery test is also well done and directly addresses why a known shape signal is not recovered: the observed CCF noise is about five times above the threshold. The chromatic null result, with the line-depth confounder handled, is clean.\n\nThe soft spot is the quantitative amplitude gradient: 10% larger in the core, 25% smaller in the wings, and the similar 6.5-sigma line-depth trend. These numbers rest on the two-component GP decomposition with hyperparameters fixed from Luhn et al. (2023), applied to a single 6.5-hour night with only about five p-mode cycles. Appendix A validates the decomposition on simulations drawn from the same model with bulk amplitudes; it does not test the regime where per-bin amplitudes differ from the kernel amplitude, which is exactly the regime of the gradient. The authors explicitly concede that granulation has appreciable power inside the oscillation envelope and that granulation variability on oscillation timescales tends to be attributed to the oscillation component. That means the fitted scale factors A_i could absorb a slice-dependent leakage. The statement that results are consistent without granulation subtraction is reassuring, but that simpler regression still uses the same bulk RV oscillation component, so it is not a full null test. A simulation with per-bin amplitudes set equal (no gradient) through the same pipeline would settle this.\n\nThe paper is honest about its limits: it flags line depth as not a one-to-one proxy for height, and defers spectral synthesis to a follow-up. That is good practice. The main gaps are no public data or code and the lack of that null simulation. For the exoplanet RV community, the practical takeaway—SCALPELS should not be expected to beat exposure-time binning for low-degree p-modes on Sun-like stars—is probably robust regardless of the gradient's exact shape.\n\nWho is it for: anyone working on RV mitigation or stellar variability, particularly users of SCALPELS and shape-vs-shift methods. It deserves a serious referee; I would send it out, with the request to add a null simulation and make data/code available. My own verdict would be conditional acceptance; the qualitative shift-driven conclusion is solid, and the amplitude gradient is plausible but needs that validation.\n\nRecommendation: send to peer review.","headline":"Genuinely new empirical result on p-mode line-profile behavior, with a solid shift-driven core and a plausible but GP-dependent amplitude gradient that needs a null simulation before the 10%/25% numbers are taken at face value.","tokens_in":19215,"tokens_out":3703,"would_cite":true,"duration_ms":33387,"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":"This paper shows that p-mode oscillations in the subgiant HD 142091 shift the averaged spectral line profile almost purely as a rigid Doppler shift, with shape-driven changes only a small residual.","keywords":["p-mode oscillations","radial velocities","cross-correlation function","CCF bisector","line depth","stellar granulation","Gaussian process","shape-vs-shift technique"],"falsifier":"A decisive test is a multi-night campaign on HD 142091 (or a similar subgiant) that resolves individual p-mode frequencies in the Fourier domain and measures the CCF core-to-wing amplitude ratio mode by mode; if the roughly 10% larger and 25% smaller gradient does not reproduce for resolved modes, the single-night decomposition has misattributed granulation power to the oscillations.","tokens_in":18081,"feed_emoji":"⭐","tokens_out":9335,"duration_ms":81635,"temperature":0.7,"pith_summary":"Low-degree p-mode oscillations, the acoustic pulsations that dominate short-timescale stellar noise in radial-velocity planet searches, have been assumed to distort spectral line shapes in ways that can be separated from the pure Doppler shift of a planet. This paper tests that assumption on a resolved, single-night time series of p-mode oscillations in the subgiant HD 142091, taken at 2-minute cadence with the NEID spectrograph. It finds that the oscillations appear primarily as pure translational Doppler shifts of the average line profile, measured by the cross-correlation function (CCF), with shape-driven variations only a higher-order effect. It further measures a height-dependent amplitude gradient: oscillation amplitudes are about 10% larger near the CCF core and 25% smaller near the wings, and deeper spectral lines show larger amplitudes. If true, this means p-mode oscillations will not lend themselves to removal by line-shape diagnostics, and existing exposure-time binning remains the appropriate mitigation for planet surveys.","feed_headline":"Stellar p-modes shift lines as a whole, not by warping","feed_subtitle":"Oscillation amplitude runs 10% high at line cores and 25% low in the wings of one subgiant star.","key_machinery":"The load-bearing object is the cross-correlation function (CCF), the average stellar line profile built by cross-correlating each spectrum with a template mask. Its shape is probed two ways: seven horizontal flux slices yield bisector-velocity time series, and a line-by-line toolkit groups about 2662 lines into depth and wavelength bins of equal radial-velocity weight. Each time series is modeled with a two-component Gaussian process whose granulation and oscillation kernels and hyperparameters are fixed from prior work. Equation (1) fits every slice's oscillation as $A_i \\mu^{\\rm osc}_{\\rm RV}(t)$, an amplitude-scaled version of the bulk oscillation component after subtracting the GP granulation component; the fitted $A_i$ values across CCF slices and across line-depth bins are what expose the 10%/25% amplitude gradient. SCALPELS is the negative control: it is shift-invariant by construction, and its failure to recover the known shape changes, with an injection-recovery test showing detection would require CCF noise near 24 ppm versus the observed 130 ppm, brackets how small the shape-driven component is.","core_discovery":"The central result is that, in a 6.5-hour NEID time series of HD 142091, the p-mode oscillations produce CCF residuals that are almost exactly what a pure Doppler shift of a template CCF would produce: the median ratio of residual RMS inside versus outside ±10 km/s is 1.069. When the time series is split into seven CCF flux slices, each slice's bisector velocity oscillation is well described as a constant scale factor times the bulk radial-velocity oscillation component after GP granulation subtraction. The scale factors are not all unity: the deepest (core) slice is about 10% larger than the bulk, the shallowest (wing) slice is about 25% smaller, a trend for which a quadratic is preferred at 7.2 sigma. A line-by-line analysis reproduces the same pattern, with deeper lines having larger oscillation amplitudes, while no phase lag, no change of timescale, and no wavelength dependence beyond line-depth differences are found. The paper reads the amplitude gradient as the atmospheric-height dependence of the oscillation velocity field, with the caveat that CCF slice position and line depth are only proxies for formation height.","pith_inferences":["The measured core-to-wing amplitude gradient predicts that a Sun-as-a-star dataset, which averages over many more modes, should show a similar slice-dependent bisector amplitude; this is directly checkable with existing solar extreme-precision radial-velocity time series.","If the gradient is confirmed with resolved modes, bisector-slice amplitudes become a time-domain probe of line formation height that could complement spectral synthesis, potentially mapping atmospheric depth dependence for many stars at once.","The paper's negative SCALPELS result suggests that other shape-based activity indicators, such as bisector span or related metrics, will carry p-mode oscillations at reduced amplitude when weighted toward line wings, which could subtly affect activity-cycle measurements in some surveys."],"forward_implications":["Exposure-time binning over an integer number of p-mode cycles, the standard mitigation for Sun-like radial-velocity surveys, should continue to work; shape-vs-shift algorithms will not recover additional p-mode signal from that observing strategy.","Radial-velocity measurements that weight deep line cores will see slightly larger p-mode oscillations than bulk RVs, while wing-weighted or shallow-line RVs will see smaller amplitudes, a systematic any sub-meter-per-second planet search should budget for.","Because low-degree modes dominate on this night, p-modes cannot be assumed to always produce CCF asymmetries; the instantaneous mode content decides whether shape-driven distortions are present.","Line-depth-dependent RV binning cannot suppress p-mode oscillations to near zero, since no collection of lines reduces oscillation RMS substantially; depth selection is therefore not a viable p-mode mitigation."],"supporting_citations":[{"why":"Supplies the granulation and oscillation Gaussian-process kernels and fixed hyperparameters used to model the RV time series and extract individual oscillation components.","marker":"Luhn et al. (2023)"},{"why":"Provides the GP posterior decomposition method (their Equations 19 and 20) used to separate oscillation from granulation in every time series.","marker":"Gupta et al. (2022)"},{"why":"Defines SCALPELS, the shift-invariant autocorrelation method used to search for shape-driven RVs and to define the detection-threshold injection tests.","marker":"Collier Cameron et al. (2021)"},{"why":"Establishes the exposure-time-binning mitigation strategy that the paper argues remains sufficient for p-mode removal in RV surveys.","marker":"Chaplin et al. (2019)"},{"why":"Gives the photon-limited velocity uncertainty formula (their Equation A.2) used for the CCF bisector measurements.","marker":"Boisse et al. (2010)"},{"why":"Supplies the line-by-line template-matching algorithm that measures individual line RVs and depths for the depth and wavelength analyses.","marker":"Siegel et al. (2022)"},{"why":"Predicts that radial p-modes produce pure shifts whereas nonradial modes in rotating stars create shape changes, the theoretical baseline the observations are compared against.","marker":"Telting & Schrijvers (1997)"},{"why":"Provides the earlier solar observation of height-dependent oscillation amplitudes between Na I and K I lines that motivates the atmospheric-height interpretation.","marker":"Isaak et al. (1989)"}],"fun_headline_variants":["Subgiant p-modes: Doppler shifts, not line warping","Oscillating star reveals pure Doppler shifts in line profiles","NEID sees p-mode oscillations as clean Doppler shifts","Line-core p-mode amplitudes beat wings by 10% in subgiant","Subgiant's p-mode oscillations: pure Doppler, no shape change"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result stands on the assumption that the two-component Gaussian-process decomposition, with kernels from earlier work, cleanly separates oscillations from granulation in a single roughly 6.5-hour night, so that the fitted $A_i$ factors measure true oscillation amplitudes rather than granulation leakage that varies across CCF slices or line depths.","fun_headline_variants_meta":{"raw":{"variants":["Subgiant p-modes: Doppler shifts, not line warping","Oscillating star reveals pure Doppler shifts in line profiles","NEID sees p-mode oscillations as clean Doppler shifts","Line-core p-mode amplitudes beat wings by 10% in subgiant","Subgiant's p-mode oscillations: pure Doppler, no shape change"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001145,"raw_usage":{"total_tokens":4839,"prompt_tokens":1124,"completion_tokens":3715,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":740,"completion_tokens_details":{"reasoning_tokens":3624}},"tokens_in":740,"tokens_out":3715,"duration_ms":27073,"temperature":1.0,"reasoning_tokens":3624,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T18:40:16.367637+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test is a multi-night campaign on HD 142091 (or a similar subgiant) that resolves individual p-mode frequencies in the Fourier domain and measures the CCF core-to-wing amplitude ratio mode by mode; if the roughly 10% larger and 25% smaller gradient does not reproduce for resolved modes, the single-night decomposition has misattributed granulation power to the oscillations.","supporting_citations":[{"cited_title":"F., Luhn, J., Wright, J","cited_arxiv_id":null,"evidence_quote":"Provides the GP posterior decomposition method (their Equations 19 and 20) used to separate oscillation from granulation in every time series."},{"cited_title":"H., & Schrijvers, C","cited_arxiv_id":null,"evidence_quote":"Predicts that radial p-modes produce pure shifts whereas nonradial modes in rotating stars create shape changes, the theoretical baseline the observations are compared against."},{"cited_title":"R., McLeod, C","cited_arxiv_id":null,"evidence_quote":"Provides the earlier solar observation of height-dependent oscillation amplitudes between Na I and K I lines that motivates the atmospheric-height interpretation."}],"review_version":1}