{"id":"22a3653a-927d-469b-96d8-24775c01db42","arxiv_id":"2505.10135","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A search of 55,232 Kepler stars yields 88 candidate non-transiting substellar companions with periods below 2.3 days, mostly in the short-period dearth zone previously inferred from transiting planets.","lead":"Astronomers searched Kepler light curves for brightness variations caused by very close-in planets that do not cross in front of their star, and found 88 candidate systems with orbital periods under 2.3 days. The candidates cluster in a region where star-planet tidal and magnetic interactions should be strongest, so they may help calibrate how close-in planets spiral toward their hosts.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 88-candidate claim rests on an unquantified stellar-activity null: no injection-recovery or control sample shows that the sharp coherent PSD peaks are companions, and activity false positives would bias the dearth-zone claim.","rationale":"The reader's weakest assumption is the same one I find most load-bearing. The central claim is a catalogue of 88 non-transiting candidates; if a large fraction are actually active-longitude or spot-coherent signals, the catalogue and the dearth-zone population statement both lose their physical meaning. The paper is careful in contamination checks and openly states caveats, including unphysical fits, which supports a conditional rather than a reject verdict. My proposed injection-recovery test with an activity control directly quantifies the false-positive rate and would settle whether the concern lands. Without it, the candidate status is the honest reading, but the dearth-zone claim should not be cited as evidence for star-planet interactions until the contamination rate is measured. Hence the reader's CONDITIONAL verdict is unchanged.","tokens_in":20593,"tokens_out":10065,"duration_ms":103662,"concrete_test":"Run an injection-recovery experiment on ~1,000 stars drawn from the same Santos et al. (2019, 2021) parent sample that did not pass the Section 2.1 threshold. For each star, inject a synthetic companion phase curve using Eq. (1) or Eq. (4) with parameters drawn from the 88 candidates' posterior distributions, at a period drawn from the candidate period distribution; as a control, inject an activity-only signal (a sinusoid at a period distinct from P_rot and its harmonics, mimicking an active longitude) with amplitude matching the candidate amplitudes. Process all light curves through the same automated pipeline (PSD threshold, harmonic exclusion, catalogue cross-checks, wavelet stability; blind the visual inspection if feasible). Measure the recovery fraction of injected companions and the fraction of activity-only injections flagged as candidates.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 2.1, the detection threshold s_det/s~ ~ 260 corresponds to p_det = 1e-100 under a pure-noise chi-squared-2 null (Woodard 1984; Appourchaux et al. 2000). This null does not include stellar activity: persistent active longitudes or spot groups can produce sharp, coherent, high-amplitude PSD peaks at periods distinct from the catalogued P_rot, especially in fast rotators whose short rotation periods bring rotation harmonics into the searched band (>5 microHz). The Section 2.3 visual checks (phase-folding, wavelet stability, catalogue cross-checks, Gaia and 2MASS contamination) are useful filters but are not a statistical control. Section 2.1 explicitly retains peaks inside the rotation harmonic pattern when the morphology is 'clearly distinct', and Appendix D keeps three candidates whose P_orb equals a rotation harmonic based on a subjective stability assessment. The candidates concentrate at P < 1 day (55/88) and in the dearth zone below the McQuillan et al. (2013) lower envelope (Fig. 5, Section 4), precisely the fast-rotator regime where magnetic activity is strongest. If a substantial fraction of the 88 are activity false positives, the population statement 'located mostly within the dearth zone' becomes an artifact of contamination, not evidence about star-planet interactions. Section 3 internal checks do not resolve this: the three-component fit to KIC 5697777 gives a companion mass of ~1.3 Msun, and three other systems yield inconsistent inclinations; the paper acknowledges these as model failures, but they also show that non-companion signals can pass all selection steps and, when fitted, produce converged but physically meaningless parameters. No injection-recovery or activity-only control is presented, so the false-positive rate among the 88 is completely unquantified.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper searches 55,232 Kepler FGKM main-sequence and subgiant stars from the Santos et al. (2019, 2021) rotation catalogues for photometric phase-curve signatures of close-in non-transiting substellar companions. The detection pipeline computes the power spectral density of each light curve, applies an empirically chosen high-amplitude peak threshold (s_det/s_bar ≈ 260), removes candidate peaks that coincide with the stellar rotation period or its harmonics unless the morphology is judged clearly distinct, cross-matches against binary and false-positive catalogues, and performs phase-folding, wavelet, Gaia, and 2MASS visual checks. The final list contains 88 candidates with orbital periods below 2.3 days, 55 of which have P_orb < 1 day. The paper then fits sinusoidal or three-component phase-curve models to 86 of these candidates, derives companion radius/mass constraints for some of them, and places the 88 systems in the P_rot versus P_orb plane, concluding that they lie mostly inside the McQuillan et al. (2013) dearth zone. The authors repeatedly and appropriately use the term 'candidates' and explicitly condition the dearth-zone statement on confirmation.","tokens_in":20977,"tokens_out":5339,"duration_ms":56876,"significance":"If the candidate sample is acceptably pure, the paper would provide a substantial new set of very short-period non-transiting companion candidates that transit searches systematically miss, and it would strengthen the empirical case that such systems populate the dearth zone predicted by tidal and magnetic star-planet interaction models. The work is valuable as a survey plus candidate catalogue: the pipeline is clearly described, the starting sample is large and public, the catalogue cross-matches are documented, and the phase-curve fitting uses Bayesian posterior sampling with reported uncertainties. The paper is also transparent about some model limitations, notably the inconsistent physical parameters obtained for several three-component fits. However, the central population claim rests on the assumption that the sharp coherent PSD peaks are produced by orbiting companions rather than by stellar activity, and the current manuscript does not quantify that false-positive rate, so the significance of the dearth-zone conclusion is not yet established.","major_comments":[{"comment":"The detection threshold s_det/s_bar ≈ 260 is calibrated against a pure-noise chi-squared two-degrees-of-freedom null (Woodard 1984; Appourchaux et al. 2000) after 'some empirical tests.' The relevant null for this search is not white noise but the stellar-activity signal: coherent active longitudes and persistent spot groups can produce sharp, high-amplitude, quasi-coherent PSD peaks at periods distinct from the catalogued rotation period and its harmonics, especially among the fast rotators that dominate the final sample. The absence of an injection-recovery test or an activity-control sample means that the paper does not currently quantify the false-positive rate of the 88-candidate list. This is load-bearing because the dearth-zone population statement in §4 would be biased if a substantial fraction of the candidates are activity artefacts. Please add quantitative false-positive and recovery metrics, for example by injecting synthetic companions into real Kepler light curves of inactive stars and by comparing candidate rates as a function of rotation period or activity level.","section":"§2.1"},{"comment":"Several selection steps are subjective and are not given quantitative criteria: peaks within the rotation harmonic pattern are retained when they show a 'clearly distinct morphology,' and the phase-folding, wavelet-stability, and 2MASS-halo judgements are made by visual inspection. Appendix D explicitly keeps three candidates whose P_orb equals a rotation harmonic based on a stability assessment. These choices may be appropriate, but they currently lack reproducibility metrics. Please provide either a scoring rule with thresholds or an inter-rater/audit table, and quantify how many candidates at each stage (4,788 → 362 → 283 → 245 → 153 → 142 → 88) were removed by each visual criterion. Since a stricter rotation-harmonic veto would remove three of the final candidates, this directly affects the dearth-zone comparison.","section":"§2.3 and Appendix D"},{"comment":"The three-component phase-curve fits do not provide independent confirmation of the companion interpretation. For KIC 5697777 the inferred companion mass is about 1.3 M_sun, which is manifestly inconsistent with a substellar companion, and for KIC 4373708, KIC 11702835, and KIC 5622796 the inferred inclination values are inconsistent. The manuscript attributes these results to model breakdown, but this means the §3 'constraints' cannot be used as supporting evidence for the candidate nature of these systems. Please state more prominently that the candidate status rests on the PSD and photometric-morphology selection alone until radial-velocity or other follow-up is available, and consider removing or clearly flagging the systems whose fitted parameters are unphysical.","section":"§3 and Table B.2"},{"comment":"The comparison with the McQuillan et al. (2013) dearth envelope is not corrected for the survey selection function. The search is restricted to frequencies above 5 µHz and the method strongly favours very short orbital periods, so the clustering of candidates at P_orb < 1 day and below the M13 lower envelope is partly a selection effect. The paper acknowledges this qualitatively, but the claim that the candidates 'populate a previously deserted area' requires a detection-completeness map: the recovery fraction as a function of P_orb, P_rot, and signal amplitude. Without such a map, the dearth-zone statement cannot be separated from the sensitivity of the search method.","section":"§4 and Fig. 5"}],"minor_comments":[{"comment":"There is a typo in the threshold sentence: 'we decided to consider as det/s value' should read 'we decided to consider a s_det/s value of approximately 260,' and 'Nyquist frequence' should be 'Nyquist frequency.'","section":"§2.1"},{"comment":"The target name is inconsistent: the text begins by discussing KIC 2852669 but concludes that KIC 3233612 was excluded, and the figure caption refers to KIC 3233612. Please correct the identifiers.","section":"Appendix A.3"},{"comment":"The statement that for all 86 datasets 'the posterior distribution of each parameter as well as the residuals of the fit follow a normal distribution' is too strong given the reported asymmetric credible intervals and the mild asymmetries visible in Figures B.1 and B.2; please rephrase to 'are unimodal and show no strong systematic residuals' or provide a normality test.","section":"§3"},{"comment":"The heading 'Data avaibility' contains a typo, and Table A.1 uses 'correponding' for 'corresponding.'","section":"Data availability"},{"comment":"The package name UltraNest appears as 'theUltraNestsampler' with missing spaces, and the software versions used for the pipeline are not listed; please add the version numbers or a software environment statement.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The main technical issue is the unquantified stellar-activity false-positive rate, which is a standard concern for this kind of blind PSD survey. The requested injection-recovery and control-sample tests are feasible with the same public Kepler light curves and would materially strengthen the central population claim. I therefore recommend major_revision rather than rejection. I would also encourage the editor to ask the authors to make the candidate-by-candidate audit trail (including the visual-selection decisions) available as supplementary material."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Candidate catalogue with real value, but the false-positive rate is the open wound.\n\nThe paper has one clear product: a vetted list of 88 non-transiting companion candidates with orbital periods below 2.3 days, 55 below one day, none overlapping with Millholland & Laughlin. That is a meaningful extension of the phase-curve search into the ultra-short-period regime, and it gives the tidal and magnetic star-planet interaction community a set of targets worth follow-up.\n\nThe search design is mostly sound. Using the Santos rotation catalogue to remove rotation harmonics is the right first cut, and the contamination checks against KOI, Gaia, and 2MASS are thorough. The authors also deserve credit for transparency: they explicitly flag the empirical nature of their PSD threshold, they exclude stars with non-convergent fits, and they report the embarrassing 1.3 solar-mass 'companion' for KIC 5697777 and inconsistent inclinations for three other systems rather than hiding them. That honesty makes the catalogue more credible. The citation pattern is also reasonable - the rotation catalogue is the natural resource, and the G23 model is the thing under test.\n\nThe soft spot is real and is the difference between a candidate list and a population statement. The Section 2.1 detection threshold is calibrated on a pure noise chi-square null, not on stellar activity. Active longitudes and spot groups in fast rotators can produce sharp, coherent PSD peaks at periods distinct from the catalogued rotation period, and the visual checks in Section 2.3 are filters, not a statistical control. Three candidates in Appendix D are kept even though their periods are rotation harmonics, on the strength of morphology. No injection-recovery test or activity-only control sample is presented, so the false-positive rate among the 88 is unquantified. If a substantial fraction are activity artifacts, the dearth-zone concentration in Section 4 becomes a contamination story rather than an interaction story. The authors themselves remind us that the method favors short periods, so part of the dearth-zone clustering is selection.\n\nNone of this kills the paper. The candidates are openly labeled as candidates, and the fitting failures are acknowledged. What it means is that the central population claim should be treated as a hypothesis, not a result, until the null is addressed.\n\nThe paper deserves a serious referee. I would send it to review but require the authors to add an injection-recovery test or an equivalent activity-control analysis, and to soften the dearth-zone language if the correction is not applied. It is a useful paper for the star-planet interaction community and for anyone building follow-up target lists.","headline":"A useful candidate catalogue of 88 ultra-short-period non-transiting companions, but the unquantified stellar-activity false-positive rate keeps the dearth-zone claim at hypothesis level.","tokens_in":21607,"tokens_out":3359,"would_cite":true,"duration_ms":33510,"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":"The paper identifies 88 Kepler stars whose power spectra contain a sharp, stable, rotation-distinct periodicity consistent with a close non-transiting substellar companion, most with orbital periods under a day and located inside the…","keywords":["non-transiting exoplanets","close-in planets","star-planet interactions","phase-curve photometry","Kepler light curves","dearth zone","tidal dissipation","power spectral density"],"falsifier":"Run high-resolution radial-velocity monitoring of the 88 candidates at the predicted orbital periods and amplitudes; if a large fraction show no radial-velocity variation with the photometric period and phase, the peaks are not caused by companions. The fastest version is to start with the nine stars whose multi-component fits give self-inconsistent masses, such as the solution implying a roughly solar-mass object for one candidate; spectroscopy there can reject or confirm the companion model within a few nights.","tokens_in":20414,"feed_emoji":"🪐","tokens_out":12048,"duration_ms":113207,"temperature":0.7,"pith_summary":"The paper reports a systematic search for close-in non-transiting substellar companions in 55,232 Kepler light curves of main-sequence and subgiant stars with measured surface rotation periods, targeting orbital periods below 2.3 days. After requiring a sharp, high-amplitude power-spectral peak distinct from the rotation period and its harmonics, checking against known binaries, transiting planets, false positives, and field contaminants, and verifying that the modulation is stable in phase-folded and wavelet analyses, the authors are left with 88 candidate hosts. More than half of the candidates (55) have periods under one day, a regime where transiting planets are rare. The paper argues that if confirmed, these systems would sit mostly inside the dearth zone in the rotation-period versus orbital-period plane, the region that tidal and magnetic star-planet interaction models predict to be depleted, and would thereby provide a population for calibrating those models.","feed_headline":"88 Kepler stars may host close-in non-transiting planets","feed_subtitle":"Most orbit in under a day, filling the tidal dearth zone where transit searches find almost nothing.","key_machinery":"The carrying tool is a power-spectral-density peak search over calibrated light curves, in which a candidate is a sharp, high-amplitude peak at a frequency corresponding to a period below 2.3 days, distinct from the star's known rotation period and its broad harmonics. A statistical detection threshold suppresses noise peaks, visual inspection removes rotation-harmonic chains, and phase-folding plus wavelet analysis tests whether the modulation is coherent across the four-year observing baseline. For the surviving candidates, fits using the three physical phase-curve components (atmospheric reflection and emission, tidal ellipsoidal distortion, and Doppler boosting) convert the observed modulation amplitude into constraints on companion radius and mass. The dearth zone, the region where fast rotators are thought to have lost their close-in planets to tides and magnetic braking, is the interpretive target that gives the detected population its wider significance.","core_discovery":"The central claim is that 88 of 55,232 stars carry a detectable photometric signature consistent with a close non-transiting substellar companion, with orbital periods below 2.3 days and 55 of them below one day. After removing known binaries, transiting systems, false positives, and field contaminants, the authors fit the phase-folded light curves of 86 candidates: for 77 the modulation is quasi-sinusoidal and is modelled as reflected starlight plus thermal emission, while for nine the fit includes tidal ellipsoidal distortion and Doppler boosting. The paper then shows that, in the rotation-period versus orbital-period plane, the candidates fall mostly below the lower envelope of known close-in planets, that is, inside the dearth zone that tidal and magnetic interaction models predict to be depleted. The paper repeatedly stresses that the detections are candidates, not confirmed planets, and that ground-based radial-velocity follow-up is required.","pith_inferences":["If even a fraction of the candidates survive radial-velocity follow-up, the dearth zone would look less empty than the transiting-planet record suggests; the depletion may be partly an inclination and detection-geometry effect of the transit method rather than a purely dynamical one.","The wavelet-stability cut assumes a companion signal is coherent for all four years of Kepler; a genuine companion whose phase-curve amplitude is modulated by magnetic activity could be rejected by that cut. Injecting synthetic non-transiting signals into real Kepler light curves and re-running the pipeline would quantify how often this happens.","Because non-transiting geometry bounds the orbital inclination from above, the photometric fits can be combined with radial-velocity masses to test the assumed albedo and heat-recirculation behaviour: any systematic inconsistency would point to missing physics in the phase-curve model."],"forward_implications":["If confirmed, the 88 candidates would substantially enlarge the known population of non-transiting short-period companions, which currently numbers only a handful of systems.","The 55 sub-day candidates give tidal and magnetic interaction models a concrete sample, because their short orbits imply fast orbital evolution and hence measurable decay or migration over years of monitoring.","The three candidates with a dominant second harmonic are systems where tidal ellipsoidal distortion outweighs the other phase-curve terms, making them the best targets for extracting companion masses from photometry alone.","The same peak-search and phase-curve method can be re-run on K2, TESS, and PLATO data, where brighter target stars will make ground-based radial-velocity confirmation more efficient."],"supporting_citations":[{"why":"Supplies the 55,232-star sample of FGKM main-sequence and subgiant stars with measured rotation periods; the rotation priors let the search exclude rotation and harmonic peaks.","marker":"Santos et al. (2019, 2021)"},{"why":"Sets the dearth-zone context with a tidal-plus-magnetic model and supplies the comparison sample of confirmed planet hosts used in the rotation-period versus orbital-period diagram.","marker":"García et al. (2023)"},{"why":"Identified the depletion of close-in planets around fast rotators and provides the lower-envelope fit that the candidates are shown to lie below.","marker":"McQuillan et al. (2013)"},{"why":"Supplies the three phase-curve components (reflection/emission, ellipsoidal distortion, Doppler boosting) and their amplitude equations used in the fits.","marker":"Shporer (2017)"},{"why":"The previous automated Kepler search for non-transiting short-period companions that this work is compared against, and the source of the ellipsoidal and boosting parameterisation.","marker":"Millholland & Laughlin (2017)"},{"why":"Defines the KEPSEISMIC light-curve calibration and the 20-day high-pass filter that is the starting point of the peak search.","marker":"García et al. (2011); Pires et al. (2015)"},{"why":"Provides the statistical detection-threshold calculation used to set the power-spectral peak selection level.","marker":"Appourchaux et al. (2000)"},{"why":"Identifies binary systems among the candidates, removing fifteen stars whose peaks are likely not planetary.","marker":"Berger et al. (2018)"}],"fun_headline_variants":["88 Kepler stars may host hidden close-in planets","Close-in non-transiting planet candidates found in Kepler data","Kepler survey finds 88 likely hidden planets orbiting in under 2.3 days","88 non-transiting exoplanet candidates may explain the tidal dearth zone","Hidden worlds: 88 close-in non-transiting candidates from Kepler"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that any sharp, high-amplitude, stable peak in a star's brightness variations at a period different from the star's rotation period and its harmonics is caused by an orbiting companion, not by stellar activity, active longitudes, or an unseen neighbouring star.","fun_headline_variants_meta":{"raw":{"variants":["88 Kepler stars may host hidden close-in planets","Close-in non-transiting planet candidates found in Kepler data","Kepler survey finds 88 likely hidden planets orbiting in under 2.3 days","88 non-transiting exoplanet candidates may explain the tidal dearth zone","Hidden worlds: 88 close-in non-transiting candidates from Kepler"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001275,"raw_usage":{"total_tokens":5223,"prompt_tokens":959,"completion_tokens":4264,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":575,"completion_tokens_details":{"reasoning_tokens":4173}},"tokens_in":575,"tokens_out":4264,"duration_ms":28978,"temperature":1.0,"reasoning_tokens":4173,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:15:27.050839+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run high-resolution radial-velocity monitoring of the 88 candidates at the predicted orbital periods and amplitudes; if a large fraction show no radial-velocity variation with the photometric period and phase, the peaks are not caused by companions. The fastest version is to start with the nine stars whose multi-component fits give self-inconsistent masses, such as the solution implying a roughly solar-mass object for one candidate; spectroscopy there can reject or confirm the companion model within a few nights.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the 55,232-star sample of FGKM main-sequence and subgiant stars with measured rotation periods; the rotation priors let the search exclude rotation and harmonic peaks."},{"cited_title":"& Laughlin , G","cited_arxiv_id":null,"evidence_quote":"The previous automated Kepler search for non-transiting short-period companions that this work is compared against, and the source of the ellipsoidal and boosting parameterisation."},{"cited_title":"2000, , 538, 401","cited_arxiv_id":null,"evidence_quote":"Provides the statistical detection-threshold calculation used to set the power-spectral peak selection level."}],"review_version":1}