{"id":"024321b8-3330-4bce-b1fc-c0aefe7b4a78","arxiv_id":"1908.04555","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Sub-millimeter light curves of IRC+10216 and o Ceti show the same periods as their optical variability, and IRC+10216's 850-micron peak lags its optical peak by about 540 days.","lead":"Astronomers used seven years of telescope calibration snapshots of two well-known giant stars to track their brightness changes at sub-millimeter wavelengths and found that one star's sub-millimeter peak arrives about 540 days after its optical peak. The result gives a new way to watch dust form and fade around dying stars, though it leaves the main driver of the lag unexplained.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Phase-lag claim is not robust to the period's large FWHM uncertainty; an 85-day period error can shift the folded peak by up to ~0.5 cycles, so the 540-day lag is unconstrained without a stability check.","rationale":"The reader's weakest assumption correctly flags the missing propagated uncertainty and the few-cycle baseline, but my concern is more specific and quantitative: the period's FWHM-based uncertainty (σ_FWHM ≈ 85 d) is so large that the phase of the folded peak is unstable across the allowed period range. The paper's dismissal of period-choice effects (Sec. 4.2, only 11 d between methods) is misleading because it ignores the FWHM. This directly threatens the 540-day phase-lag claim, which is the novelty of the paper. However, the core period measurements for o Ceti and IRC+10216 at 850 µm are likely sound, and the modeling sections are honestly caveated. Therefore the appropriate action is to keep the reader's CONDITIONAL verdict, with the condition strengthened to explicitly require a stability check of the lag across the period uncertainty. I agree with the reader that the paper should be accepted only after this check is added or the lag is appropriately downgraded to an upper limit.","tokens_in":13914,"tokens_out":5580,"duration_ms":55230,"concrete_test":"Recompute the 850 µm phase-folded light curve and the resultant optical-to-sub-mm lag for a grid of periods spanning the 1σ range, e.g., P = 678 ± 85 d (593, 640, 678, 763 d), using the same T0 and rebinning. If the recovered lag changes by more than ~0.2 cycles (≈140 d) across this grid, the 540-day lag is not robust and should be reported with a large systematic uncertainty or as unconstrained. For a sharper test, perform a Monte Carlo: generate 1000 synthetic light curves with the actual sampling and noise, inject a known lag, and verify the recovery; if the scatter of recovered lags exceeds ±200 d, the claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the ~540-day phase lag between IRC+10216's 850 µm and optical light curves. The lag is obtained by phase-folding the 850 µm data with the Astropy Lomb-Scargle period (678 d) and reading off the peak phase relative to the optical T0. This is not robust because the period is only constrained to σ_Tot = 85 d (Table 2; dominated by the periodogram FWHM). Over the ~2557-day baseline (~3.8 cycles), a ±85 d period error shifts the phases of individual observations by up to ±2557×(85/678²) ≈ ±0.47 cycles. Thus the folded light curve is heavily smeared for periods within 1σ, and the bin containing the nominal peak can move by a large fraction of a period; the derived 540-day lag (0.79 phase) could change by hundreds of days. The paper's statement in Sec. 4.2 that the four methods' periods differ by only 11 days (1% level) addresses method scatter, not the FWHM uncertainty, and therefore does not justify using a single period. Furthermore, if the true sub-mm period is closer to the optical 640 d (which is within ~0.4σ of the sub-mm value), folding with 678 d introduces a phase drift of ~0.22 cycles over the baseline, further shifting the peak. No uncertainty is propagated to the lag, so the headline result is effectively unconstrained by the current analysis.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents sub-mm (450 and 850 micron) light curves of two AGB stars, IRC+10216 and o Ceti, obtained from JCMT/SCUBA-2 pointing calibration observations over seven years. Periods are derived with four methods: two Lomb-Scargle implementations, Gatspy Supersmoother, and P4J. For IRC+10216 at 850 micron, the periods range from 667 to 678 days with total uncertainties of about 80-95 days, while the 450 micron data are noisier and partly aliased. After phase-folding the 850 micron data with the Astropy Lomb-Scargle period of 678 days, the authors report a phase offset of about 0.79 relative to the optical peak, corresponding to a ~540 day lag. They then explore possible origins: light travel time, molecular-line contamination, free-free emission, and a dust formation/destruction cycle modeled with Hyperion radiative-transfer snapshots. They conclude that the dust cycle can contribute only partially and that a second, unknown mechanism is needed to explain the sub-mm variability and the phase lag.","tokens_in":14244,"tokens_out":6251,"duration_ms":63140,"significance":"The paper makes clever use of a large volume of calibration data to probe long-timescale sub-mm variability in two benchmark AGB stars. The multi-method period analysis is careful, and the public release of scripts and photometry tables is a strength. If the phase-lag measurement were robust, it would provide a new observational constraint on the connection between stellar pulsation and dust formation in the inner circumstellar envelope. However, the headline ~540 day lag is currently presented without a propagated uncertainty, and the wide periodogram peak (sigma_Tot = 85 days) directly threatens the stability of the folded phase. The central observational claim therefore needs additional analysis before the result can be considered established.","major_comments":[{"comment":"The ~540-day phase lag is not robust to the period uncertainty. The phase folding in Sec. 4.2 uses a single period of 678 days, but the reported total uncertainty is sigma_Tot = 85 days, dominated by the periodogram FWHM. Over the roughly 2557-day baseline (about 3.8 cycles), a period error of 85 days corresponds to a phase drift of approximately 0.47 cycles at the end of the baseline, which will substantially smear the folded light curve and can shift the apparent peak by a large fraction of a period. The authors correctly note in Sec. 4.2 that the data cover only about three cycles and that the periodogram peak is wide, but they do not propagate this uncertainty into the lag. The statement that the four methods agree to within 11 days measures method scatter, not the FWHM uncertainty, so it does not justify the use of a single period. To support the 540-day claim, the authors should provide a sensitivity analysis or Monte Carlo procedure that folds the data with periods drawn from the sigma_Tot distribution and reports the resulting distribution of peak phases and lags. Without this, the phase lag and the subsequent comparison with radio data and with the dust-cycle interpretation are effectively unconstrained.","section":"Sec. 4.2, Table 2"}],"minor_comments":[{"comment":"There is a typo: \"sqaure root\" should be \"square root\".","section":"Sec. 3.2"},{"comment":"The statement that recovering only ~4% of the observed flux \"is not an issue for our analysis\" is under-justified. A model that underestimates the 850 micron flux by a factor of ~25 may not correctly capture the radial distribution of emission even if the peak-to-trough ratio is similar; a brief caveat about this assumption would strengthen the discussion.","section":"Sec. 5.3.1"},{"comment":"The caption says \"the peak is shifted by delta-phi = 0.45 to better present the shape\" but does not explain whether the phase axis has been offset for display or whether this shift affects the reported phase lag. Please clarify.","section":"Fig. A1 caption"},{"comment":"The citation \"SciPy (Jones et al. 01)\" is incomplete; a full reference with year and DOI or arXiv identifier should be provided.","section":"Acknowledgements"}],"recommendation":"major_revision","confidential_remarks":"The paper is appropriate for MNRAS and the data-release practices are commendable. The main issue is the unsupported uncertainty on the phase lag. If the authors can add a rigorous propagation of the period uncertainty into the lag, with a Monte Carlo or bootstrap analysis, the paper could be acceptable. I would also encourage the editor to ensure the phase-lag claim is not presented in the abstract until that analysis is done."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper gives a solid, well-documented sub-mm photometric dataset for two archetypal AGB stars and derives periods that agree with independent optical/far-IR values. The headline result—a ~540 day lag between IRC+10216's 850 micron and optical light curves—does not hold up as presented, because the period uncertainty alone can shift the folded peak by hundreds of days. The authors are honest about many limitations, but they do not propagate the period's 85-day FWHM into the lag.\n\nWhat is new and good: this is the first multi-year 850 micron monitoring of IRC+10216 and o Ceti from SCUBA-2 calibration observations. The use of four period-finding methods (two non-parametric) is careful, and the sigma_Tot combination of MC and periodogram FWHM is a reasonable way to express how broad the period peak is. The period results for o Ceti (332-350 days at 450 micron, 310-336 days at 850 micron) are consistent with the well-known 333-day period. For IRC+10216, the 850 micron periods (667-678 days) agree with the 640 day optical period within the stated uncertainty, and the 450 micron data are convincingly shown to be noise/aliasing limited. The paper also does a good job ruling out molecular-line contamination and free-free emission as dominant causes, and it makes the data and scripts public on figshare and VizieR.\n\nThe soft spot is the lag. The phase-folded peak is read off after folding with a single period, 678 days, from the Lomb-Scargle periodogram. The full uncertainty on that period is 85 days, dominated by the periodogram FWHM. Over the ~2557-day baseline, a 1-sigma period error changes phases by about ±0.47 cycles; even a shift to the optical 640 day period (well within 1 sigma) would drift phases by ~0.22 cycles. The paper's statement that the four methods differ by only 11 days (1%) only addresses method scatter, not the true period uncertainty. So the claimed 0.79-phase lag could be off by hundreds of days, or could disappear. The fix is straightforward: refold with periods at the edges of the 1-sigma range and show the peak movement, or fit period and phase jointly and marginalize. The authors do note that data cover only about 3 cycles and the peak is wide, but that makes a stability check essential, not optional.\n\nThe radiative-transfer section is exploratory and clearly labeled as such; static Hyperion models cannot produce a lag, and the authors admit the dominant mechanism is unknown. That is acceptable, but it means the interpretation section is a discussion, not a demonstration. The paper deserves a serious referee because the dataset and period measurements are solid and useful. I would send it to review with a clear request to either add the lag robustness analysis or soften the claim to a tentative offset consistent with the uncertainties.\n\nWho this is for: AGB and circumstellar-dust people, and anyone working with calibration time series. Worth citing for the periods and light curves, not for the lag.","headline":"Solid sub-mm period dataset for two AGB stars, but the 540-day phase lag is not robust to the 85-day period uncertainty and needs a stability check before it can be taken as real.","tokens_in":14826,"tokens_out":3379,"would_cite":true,"duration_ms":35460,"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":"By folding seven years of 850-micron observations of the carbon-rich star IRC+10216, the paper finds the sub-mm brightness peaks about 540 days after the optical peak and argues that the dust formation-destruction cycle is partially…","keywords":["asymptotic giant branch stars","submillimetre variability","IRC+10216","o Ceti","phase lag","dust formation and destruction","pulsation periods","periodogram analysis"],"falsifier":"A longer 850-micron time series covering at least three more cycles, folded at both the paper's 667 to 678 day period and the Herschel 640-day period, should reproduce a roughly 0.79 phase shift for the lag to count as real; if the peak shifts to align with the optical light curve, or if the rebinned folded shape changes when the period is varied within the FWHM of the periodogram peak, the 540-day lag is an artifact of sparse sampling.","tokens_in":1862,"feed_emoji":"🌠","tokens_out":1921,"duration_ms":88057,"temperature":0.7,"pith_summary":"Two well-studied evolved stars, the carbon-rich IRC+10216 and the oxygen-rich Mira variable o Ceti, were monitored at 450 and 850 microns for seven years using routine pointing-calibration observations. The paper finds that four independent period-finding methods agree on sub-mm periods of 667 to 678 days for IRC+10216 and about 332 to 336 days for o Ceti, the latter matching a century of optical data. The striking result is that the 850-micron light curve of IRC+10216 peaks about 540 days, or a phase of roughly 0.79, after the optical peak even though the period is the same at both wavelengths. Radiative transfer models show the dust condensation radius moving inward and outward by about one stellar radius with the pulsation cycle, which can explain part of the variability and lag but not all of it. An unidentified second mechanism, likely in the inner envelope, must be doing the rest, and the paper leaves that question for future work.","feed_headline":"Sub-mm light lags optical peak by 540 days in carbon star","feed_subtitle":"Seven years of 850-micron monitoring tie the lag to the dust cycle, with a second driver still unknown.","key_machinery":"The workhorse is phase folding: rebinning the unevenly sampled sub-mm time series at a candidate period and comparing the phase of peak brightness with the phase of the optical peak after aligning both to a common zero-point. To avoid assuming a sine shape for the light curve, two non-parametric period estimators were used alongside two parametric sine-based methods, and the consistency of all four at 850 microns is what makes the lag claim interpretable. The physical mechanism is probed with a sequence of static radiative-transfer models built along the bolometric luminosity light curve; these models locate 99 percent of the beam flux in the inner 2-arcsecond region, show the dust condensation radius changing by about one stellar radius between minimum and maximum light, and recover the observed amplitude of variation.","core_discovery":"The central discovery is a measured phase lag at submillimetre wavelengths: when the 850-micron light curve of IRC+10216 is folded at the common period of about 678 days and aligned to the same zero-point as the optical light curve, its peak occurs at a phase difference of roughly 0.79, corresponding to about 540 days. The same period is recovered by four independent methods at 850 microns, with values from 667 to 678 days, while the 450-micron data are noisier and consistent only within their larger uncertainties. Light-travel time across the envelope is far too short to explain the lag, and molecular-line contamination and free-free emission can account for at most a few percent and about ten percent of the flux respectively, so the lag must arise in the dust or in some other mechanism tied to the pulse. Static radiative transfer snapshots along the stellar luminosity cycle reproduce the observed fractional amplitude, with a peak-to-trough ratio of about 1.4 observed versus 1.6 in the model, and show the dust condensation radius shifting by roughly one stellar radius, from about 2.5 to 3.5 stellar radii between minimum and maximum light. The paper concludes that the dust formation and destruction cycle contributes to the variability and lag, but a second, dominant mechanism is still required.","pith_inferences":["If the lag is real, sub-mm light curves could become a practical tracer of dust formation and destruction timescales in AGB stars, since the lag would encode the delay between the stellar pulse and the dust response.","The same phase-lag analysis could be applied to other AGB stars that appear in pointing-calibration archives, potentially revealing how the lag depends on mass-loss rate and dust composition.","The static radiative-transfer models show no lag even though they match the amplitude, which suggests that time-dependent dust formation and destruction, rather than simple geometry, is essential to explaining the observations.","If the dominant mechanism is shock-driven, the lag should vary with wavelength and might be resolved with high-resolution sub-mm imaging; that is a testable prediction the paper does not make explicitly."],"forward_implications":["o Ceti's sub-mm periods agree with its well-established optical period, so sub-mm continuum variability traces the same stellar pulsation in a dust-poor, optically thin case.","IRC+10216's sub-mm period matches optical-to-far-IR periods, meaning the pulsation period is stable across wavelengths and the 540-day lag is a phase shift rather than a different period.","The close phase agreement between the 850-micron and radio light curves suggests that one physical mechanism may drive variability at both long wavelengths.","The modelled inward and outward shift of the dust condensation radius by about one stellar radius over the pulsation cycle provides a partial explanation of the lag, linking it to dust formation and destruction.","Because the sub-mm data cover only about three cycles and the periodogram peak is wide, continued sub-mm monitoring is needed to tighten the periods and confirm the lag."],"supporting_citations":[{"why":"Supplies the Catalina optical light curve that defines the peak epoch (T0) against which the sub-mm phase lag is measured.","marker":"Drake et al. 2014"},{"why":"Derives the optical period and phase zero-point used in computing the 0.79 phase offset.","marker":"Kim et al. 2015"},{"why":"Provides the prior far-IR period and the 402-day bow-shock lag that the paper compares with its own 540-day lag.","marker":"Groenewegen et al. 2012"},{"why":"Gives the radio period, phase, and free-free flux levels used to argue that free-free emission contributes only about ten percent of the 850-micron flux.","marker":"Menten et al. 2006"},{"why":"Supplies the PSF photometry method, composite image profile, and dust parameters on which the light curves and radiative-transfer models are built.","marker":"Dharmawardena et al. 2018"},{"why":"Presents the self-consistent wind model showing dust density increasing out of phase with the pulse, the basis for the dust-formation-cycle explanation.","marker":"Bladh 2019"},{"why":"Provides the P4J information-theoretic period estimator, one of the two non-parametric methods used in the study.","marker":"Huijse et al. 2018"},{"why":"Supplies the gatspy implementations of Supersmoother and Lomb-Scargle period finding used for comparison.","marker":"VanderPlas & Ivezić 2015"},{"why":"Provides the century-long optical period of o Ceti used as ground truth for the sub-mm period agreement.","marker":"Templeton & Karovska 2009"}],"fun_headline_variants":["Sub-mm lag of 540 days in carbon star tied to dust cycle","Dust cycle partially explains 540-day sub-mm lag in IRC+10216","Carbon star's sub-mm pulse lags optical: dust plus unknown driver","540-day sub-mm lag in carbon star points to dust and unsolved mechanism","Sub-mm light from IRC+10216 lags optical by 540 days, dust implicated"],"cache_read_input_tokens":16896,"weakest_assumption_plain":"The 540-day lag rests on the assumption that IRC+10216's sub-mm brightness varies with the same period as its optical brightness, and that the peak seen in only about three cycles of sparse, calibration-limited 850-micron data is real rather than a product of noise or the chosen period.","fun_headline_variants_meta":{"raw":{"variants":["Sub-mm lag of 540 days in carbon star tied to dust cycle","Dust cycle partially explains 540-day sub-mm lag in IRC+10216","Carbon star's sub-mm pulse lags optical: dust plus unknown driver","540-day sub-mm lag in carbon star points to dust and unsolved mechanism","Sub-mm light from IRC+10216 lags optical by 540 days, dust implicated"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000715,"raw_usage":{"total_tokens":3267,"prompt_tokens":1051,"completion_tokens":2216,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":667,"completion_tokens_details":{"reasoning_tokens":2111}},"tokens_in":667,"tokens_out":2216,"duration_ms":15698,"temperature":1.0,"reasoning_tokens":2111,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:39:14.296469+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A longer 850-micron time series covering at least three more cycles, folded at both the paper's 667 to 678 day period and the Herschel 640-day period, should reproduce a roughly 0.79 phase shift for the lag to count as real; if the peak shifts to align with the optical light curve, or if the rebinned folded shape changes when the period is varied within the FWHM of the periodogram peak, the 540-day lag is an artifact of sparse sampling.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Derives the optical period and phase zero-point used in computing the 0.79 phase offset."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the prior far-IR period and the 402-day bow-shock lag that the paper compares with its own 540-day lag."},{"cited_title":"M., Reid M","cited_arxiv_id":null,"evidence_quote":"Gives the radio period, phase, and free-free flux levels used to argue that free-free emission contributes only about ten percent of the 850-micron flux."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Presents the self-consistent wind model showing dust density increasing out of phase with the pulse, the basis for the dust-formation-cycle explanation."}],"review_version":1}