{"id":"9c93d5a0-3fe8-4ba8-be25-8fc3f4f91eff","arxiv_id":"1908.02373","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A variable-star search in K2 observations of M4 delivers 4,554 light curves, 76 newly discovered variables plus 67 suspected variables, and precise RR Lyrae and asteroseismic data.","lead":"This paper turns K2 space telescope images of the globular cluster M4 into light curves for 4,554 stars, finding 76 new variable stars and 67 suspected variables. It makes the light curves public, giving astronomers the longest continuous dataset yet for a globular cluster.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The catalog's reliability rests on SNR thresholds chosen without false-alarm control and a subjective by-eye veto; no per-object false-positive rate is given, so counts of low-amplitude variables may be inflated.","rationale":"The reader's weakest-assumption identification and my own stress-test converge on the lack of a quantitative false-positive control in the detection pipeline. The threshold calibration in Section 2.7 uses known variables as labeled positives, which is reasonable for setting a detection boundary, but without labeled negatives or a null sample it cannot establish the false-positive rate. The by-eye step in the same section is explicitly acknowledged by the authors as 'less than ideal' and 'likely to lead to incorrect determinations in some of the marginal cases.' Appendix A provides concrete evidence that the automatic thresholds admit systematic signals, with many stars removed at the resaturation period, and that several objects near that period were retained only because their signals were judged strong. Those retained objects include a headline candidate X-ray binary (W4490). The independent support in the paper is real: recovery of known RR Lyrae variables, consistency with Kuehn et al. (2017) and Miglio et al. (2016), and public release of light curves. However, these do not resolve the false-positive question for new low-amplitude variables. An injection/recovery plus null test is the natural way to settle whether the concern lands. Until such a test is run, the appropriate verdict remains the reader's CONDITIONAL: the catalog is credible and useful, but individual low-amplitude and systematic-adjacent detections should be treated as needing confirmation.","tokens_in":43876,"tokens_out":6972,"duration_ms":84615,"concrete_test":"Take a random subset of the 4554 final light curves and run the Section 2.7 automated period search and SNR thresholds on the unmodified light curves as nulls, and on the same light curves with injected synthetic signals spanning periods 0.1-30 days and amplitudes 0.05-50 mmag. Measure the fraction of nulls flagged as variables (false-positive rate) and the recovery fraction for injections as a function of period and amplitude. If the false-positive rate is non-negligible, especially near the 1.962-day resaturation period or at amplitudes below ~1 mmag, per-object false-alarm probabilities are needed before trusting the low-amplitude and systematic-adjacent entries such as W4490 and the newly reported HB 'mh' variables.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is a catalog of 66 cluster-member variables, 52 new. Its reliability depends on the step in Section 2.7 where period-dependent SNR thresholds are 'determined as a function of period by comparing the SNR values for objects and periods with previously determined variability and (for BLS) injected transits with the rest of the detected periods.' That procedure separates known true variables from the bulk of all detected periods, but it never measures the false-positive rate among candidates above threshold. The final safety net is a by-eye veto that reduces 1310 automatically 'robust' periods to 161 objects. Appendix A demonstrates that the automatic stage still produced many systematic detections (e.g., 26 stars at the ~1.962-day resaturation period), and the authors state in Section 4 that relying 'so heavily on a manual and qualitative final vetting step is less than ideal and likely to lead to incorrect determinations in some of the marginal cases.' Objects retained near that systematic (W92, W4268, W4490) are justified by signal strength, not by a quantitative false-alarm criterion. Since many new entries are very low-amplitude (0.1-0.4 mmag 'mh' HB/giant variables), the unquantified false-positive rate directly threatens the headline numbers.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a variability search in the K2 Campaign 2 superstamp observations of the globular cluster M4. The authors extract light curves for 4554 Gaia DR1 sources with G < 19, process them with image subtraction, roll decorrelation, and TFA, and search for periodic variability with GLS, PDM, and BLS, using a custom deblending routine and period-dependent SNR thresholds. They report 66 cluster-member variables (52 claimed new), 24 non-member variables (20 new), 4 ambiguous-membership variables, and 67 suspected variables, including asteroseismic detections in horizontal-branch and red-giant stars, several eclipsing binaries, and a candidate X-ray binary. All light curves are made publicly available.","tokens_in":44155,"tokens_out":6469,"duration_ms":67434,"significance":"If the catalog is reliable, this is a valuable resource: it provides the longest continuous high-precision light curves for a globular cluster, recovers previously known variables with sensible periods, and adds many low-amplitude variables and asteroseismic targets. The public data release, the detailed pipeline description, and the candid discussion of the method's limitations in Section 4 are genuine strengths. The main risk is that the headline counts depend on detection thresholds and a manual vetting step whose false-positive rate is not quantified, and the paper overcounts two previously reported mmRR stars as new discoveries.","major_comments":[{"comment":"The detection thresholds are calibrated using previously known variables and injected transits, but this procedure provides no estimate of the false-positive rate among candidates above the threshold. The final step reduces 1310 automated robust periods to 161 objects by eye, and Section 4 explicitly concedes that this manual step 'is less than ideal and likely to lead to incorrect determinations in some of the marginal cases.' Because many of the new variables in Table 3 have amplitudes at or below 0.5 mmag (e.g., W837 at 0.1 mmag and W491 at 0.3 mmag), the unquantified contamination rate directly affects the central claim of 66 cluster-member variables and 52 new discoveries. I request a quantitative false-alarm analysis (e.g., injection-recovery tests on the same cadences with the same noise properties, or empirical false-alarm probabilities from shuffled light curves) and per-object detection statistics.","section":"Section 2.7 and Figure 4"},{"comment":"The retention of W92, W4268, and W4490 near the 1.962-day resaturation systematic is justified only by 'strength of signal,' with no quantitative criterion, while 26 other stars at the same period are discarded as systematics. For W4490, the adopted period of 1.959 d is only 0.15% away from the systematic period, and this period underpins the candidate X-ray binary classification. Please supply a statistical comparison of these light curves to the systematic template or to the rejected stars demonstrating that the signals cannot be produced by the resaturation artifact; otherwise these objects should be moved to the suspected-variable category with the systematic ambiguity stated.","section":"Appendix A, Section 3.4, and Table 5"},{"comment":"The claimed number of new discoveries is inconsistent with the paper's own attribution. Section 3.3 identifies W2015 and W2386 as mmRR variables previously reported in Wallace et al. (2019a), but Section 4 states that '52 cluster members (when including the two mmRRs of Wallace et al. 2019a) ... are new discoveries.' If these two objects were already announced, they cannot be counted as new in this paper; the abstract's '52 are new discoveries' should be 50, with corresponding adjustments to the total new-variable count, unless the authors explicitly justify the double counting.","section":"Abstract and Section 4"}],"minor_comments":[{"comment":"The keyword 'Detatched binary stars' contains a typo and should be 'Detached binary stars.'","section":"Keywords"},{"comment":"The phrase 'all the the 67 suspected variables' contains a duplicated article and should be corrected.","section":"Section 5"},{"comment":"The note 'There is no W1873 in this table' is confusing; please explain the numbering gap or remove the note, as it reads like an artifact.","section":"Table 1 note"},{"comment":"The phrase 'the theG magnitudes' contains a typo; it should read 'the G magnitudes.'","section":"Section 2.2"},{"comment":"The caption states 'the blue points show the values...' but the blue points are not described in the main text; a brief definition of what is plotted would improve readability.","section":"Figure 4 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is a good fit for ApJS and the public data release is valuable. The main barriers to acceptance are the unquantified false-positive rate and the inconsistent counting of previously reported mmRR stars as new discoveries; both need to be addressed before the catalog can be used at face value."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Straight to it: this paper is a solid, honest survey paper, not a breakthrough. It gives the first general variable-star search of a K2 cluster superstar, and the light curves are public. That alone is worth a lot. The headline numbers—66 cluster variables, 52 new, 57 suspected—are probably close to right for the strong signals, but I would not trust the low-amplitude tail without more calibration.\n\nWhat's actually good: the reduction is detailed and the known variables (RR Lyrae, eclipsing) are recovered with sensible periods, which is a good internal check. The asteroseismic detections on the HB and giants are a real addition, and the candidate X-ray binary W4490 is interesting. The paper is unusually transparent about its limitations; the list in Section 4 is honest.\n\nThe soft spot is exactly what the stress-test note says. In Section 2.7, the SNR thresholds are set by comparing known variables and injected transits against the bulk detections. That can set a sensible threshold, but it does not measure the false-positive rate among candidates above the threshold. The final veto is by eye, and the authors themselves say this is 'less than ideal.' Appendix A shows many systematic detections at the 1.962-day resaturation period. They keep W4490, W92, and W4268 near that period because the signals are strong. That is a reasonable judgment call, but it is not a quantitative criterion. The consequence is that some of the 0.1–0.4 mmag HB and giant variables may be systematics, so the counts are uncertain at the margin. I would not hang a strong scientific claim on the exact number of new variables.\n\nThere are smaller issues: period uncertainties are bootstrap confidence intervals, not formal errors; amplitudes have no uncertainties; classification is often qualitative. These are minor for a survey paper.\n\nThe central catalog holds up. The strong detections—the EBs, the mmRRs, the candidate X-ray binary—are likely real. The paper is a resource for the community. My recommendation: send to review. It deserves a serious referee, and a revision should add a false-alarm characterization, e.g., injection-recovery or a fit of the noise distribution, and flag which variables survive a strict quantitative cut. That would push this from 'useful with caveats' to 'benchmark.'","headline":"A solid, honest survey paper that delivers the first general variable-star catalog from a K2 cluster superstamp with public light curves; the strong detections are credible, but the low-amplitude tail needs quantitative false-alarm control.","tokens_in":44694,"tokens_out":2324,"would_cite":true,"duration_ms":26643,"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":"Continuous 78-day K2 observations of M4, the nearest globular cluster, produce 4554 light curves and reveal 66 variable stars among cluster members, 52 of them new.","keywords":["globular clusters","variable stars","K2 mission","RR Lyrae","eclipsing binaries","asteroseismology","image subtraction","M4"],"falsifier":"Take the public light curves, remove all cadences within one resaturation period of each thruster event, and re-run the period search; if the 1.959-day signal in W4490 and the other near-1.962-day candidates disappear, the detections are systematics rather than stellar variability.","tokens_in":43690,"feed_emoji":"✨","tokens_out":9345,"duration_ms":99292,"temperature":0.7,"pith_summary":"The paper claims that the K2 superstamp of the globular cluster M4 can be turned into a high-precision variable-star catalog despite Kepler's coarse 4-arcsecond pixels. Using image subtraction and systematics correction, the authors produce light curves for 4554 sources with $9<G<19$, reaching sub-millimagnitude precision on the horizontal branch. Among 3784 proper-motion cluster members they find 66 variables, 52 new, plus 57 suspected; among non-members they find 24 variables, 20 new, plus 10 suspected and 4 ambiguous. The detections include asteroseismic oscillations in six horizontal-branch stars outside the instability strip, pulsations in 19 red giants, several eclipsing binaries, and a candidate X-ray binary with 20 mmag periodic optical variation. If correct, M4's known variable-star inventory in the observed area grows substantially, and a 78-day public light-curve resource becomes available for deeper follow-up.","feed_headline":"Nearest globular cluster yields 76 new variable stars from K2","feed_subtitle":"78 days of continuous light curves reveal pulsating horizontal-branch stars, eclipsing binaries, and a candidate X-ray binary.","key_machinery":"The machinery that carries the argument is a multi-stage photometric and classification pipeline. Stitched K2 target-pixel images are subtracted against a median reference using image subtraction to suppress crowding; the residuals are corrected for spacecraft roll via position decorrelation and cleaned of common trends with a trend filter. Periodicity is searched with three algorithms (generalized Lomb-Scargle, phase dispersion minimization, and box-fitting least squares), and candidate periods are accepted only if their periodogram signal-to-noise exceeds period-dependent thresholds set empirically by comparing known variables and injected transits against noise. A Fourier-amplitude comparison identifies which star in a 12-pixel neighborhood actually produces the signal, with a recursion that unmixes blended variables, and the surviving candidates are vetted by eye. The final periods and uncertainties come from fine-grid searches with bootstrap resampling.","core_discovery":"The central discovery claimed is a diverse variable-star sample extracted from continuous 78-day K2 monitoring of M4: 66 cluster-member variables (52 new), 24 non-member variables (20 new), 57 cluster-member suspected variables, 10 non-member suspected variables, and four variables of ambiguous membership. The most striking results are the horizontal-branch oscillators: six stars outside the instability strip and one inside it show asteroseismic variability at amplitudes near or below 1 mmag, extending a continuum between RR Lyrae pulsators and the previously reported millimagnitude RR Lyrae candidates. The paper also identifies a 4.6-day slightly eccentric detached eclipsing binary, an EW-class contact binary, a 22-day detached binary outside the cluster, and a 1.959-day periodic optical variation in a cluster member positionally coincident with a ROSAT X-ray source, interpreted as a candidate X-ray binary. One RR Lyrae, V7, is flagged as a candidate Blazhko variable with a period longer than the observing window.","pith_inferences":["The paper does not test this, but the same pipeline should transfer to the other K2 cluster superstamps the authors list, with recalibrated thresholds, since crowding and distance differ.","The apparent continuation of the RR Lyrae period-luminosity relation to low-amplitude multiharmonic variables suggests that once modes are identified, oscillation periods could serve as density diagnostics across the horizontal branch and red giant branch in a single cluster; this goes beyond the paper's breadth-first presentation.","A quick test of the resaturation-systematics hypothesis would be to stack the residuals of all 4554 light curves at 1.962 days and look for a common waveform; if such a common signal persists, any single-star period within a few percent of it deserves extra scrutiny.","The box-shaped brightenings of W1834 and W2127 are suggestive of self-lensing by compact companions, a scenario the paper raises but does not commit to; targeted monitoring would test whether the events repeat."],"forward_implications":["M4's census of certain variables within the superstamp area grows substantially, giving a fuller inventory of the cluster's pulsators and binaries.","The sub-millimagnitude horizontal-branch light curves make M4 a testbed for studying asteroseismic oscillations outside the classical instability strip, connecting the new mmRR class to both ordinary RR Lyrae stars and non-pulsating HB stars.","The 4.6-day detached eclipsing binary and the EW contact binaries give future radial-velocity programs concrete targets for precise masses and radii of metal-poor, old-population stars.","The released 78-day light curves let other groups search for long-period variability that this paper's spline fitting demonstrably removes, and for transiting planets the paper did not focus on.","The candidate X-ray binary W4490 gives an X-ray observatory a specific target whose 1.959-day optical period predicts when quiescent accretion variability should appear."],"supporting_citations":[{"why":"Supplies the image-subtraction method used to create difference images that mitigate crowding in the coarse pixels.","marker":"Alard & Lupton (1998)"},{"why":"Defines the K2 mission and campaign structure, including the superstamp observations that provide the continuous M4 time series.","marker":"Howell et al. (2014)"},{"why":"Provides the roll-decorrelation approach the authors adapt to remove position-dependent systematics from the raw light curves.","marker":"Vanderburg & Johnson (2014)"},{"why":"Supplies the trend-filtering algorithm used to remove common systematics across the light curves.","marker":"Kovacs et al. (2005)"},{"why":"Supplies the standard catalog of previously known M4 variables whose members define the baseline that new detections are measured against.","marker":"Clement et al. (2001)"},{"why":"Provides the astrometric and photometric reference catalog used to define source positions and G magnitudes.","marker":"Gaia Collaboration et al. (2016a,b)"},{"why":"Provides DR2 proper motions used to assign cluster membership probabilities.","marker":"Gaia Collaboration et al. (2018)"},{"why":"Provides the prior RR Lyrae analysis of the same K2 data, used for Blazhko comparisons.","marker":"Kuehn et al. (2017)"},{"why":"Provides the prior asteroseismic giant analysis of the same data, used for comparison of red-giant detections.","marker":"Miglio et al. (2016)"},{"why":"Reports the two millimagnitude RR Lyrae candidates that this paper extends to a broader horizontal-branch asteroseismic sample.","marker":"Wallace et al. (2019a)"}],"fun_headline_variants":["K2 finds 76 new variable stars in nearby globular cluster M4","New M4 variables: 76 discovered, plus asteroseismic horizontal-branch stars","M4's hidden pulsators: 76 new variables from K2 include sub-millimagnitude stars","M4: 76 new variables, eclipsing binaries, and a candidate X-ray binary"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The catalog's reliability rests on the assumption that the period-dependent signal-to-noise thresholds, blend-removal rules, and by-eye vetting cleanly separate true stellar variability from K2 spacecraft systematics, most perilously near the 1.962-day resaturation period.","fun_headline_variants_meta":{"raw":{"variants":["K2 finds 76 new variable stars in nearby globular cluster M4","New M4 variables: 76 discovered, plus asteroseismic horizontal-branch stars","M4's hidden pulsators: 76 new variables from K2 include sub-millimagnitude stars","M4: 76 new variables, eclipsing binaries, and a candidate X-ray binary"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000975,"raw_usage":{"total_tokens":4198,"prompt_tokens":1056,"completion_tokens":3142,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":672,"completion_tokens_details":{"reasoning_tokens":3046}},"tokens_in":672,"tokens_out":3142,"duration_ms":22609,"temperature":1.0,"reasoning_tokens":3046,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:46:37.393862+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the public light curves, remove all cadences within one resaturation period of each thruster event, and re-run the period search; if the 1.959-day signal in W4490 and the other near-1.962-day candidates disappear, the detections are systematics rather than stellar variability.","supporting_citations":[{"cited_title":"A., Moskalik , P., & Drury , J","cited_arxiv_id":null,"evidence_quote":"Provides the prior RR Lyrae analysis of the same K2 data, used for Blazhko comparisons."}],"review_version":1}