{"id":"64158118-8ced-4145-857c-990ac199ec3c","arxiv_id":"2501.04925","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"First observational delay time distribution for classical novae in M31, with detections at stellar ages of 2-3.2 Gyr and 7.9-14.1 Gyr.","lead":"The paper combines a century-long catalog of novas in the Andromeda galaxy with a Hubble star-formation map to measure the ages of the stellar populations that produce them. It finds two age groups, one old and one intermediate, that host nova progenitors, giving the first observational handle on how nova production depends on stellar age.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2-3.2 Gyr detection is not robust: Padova is an upper limit, not a detection, and the unspecified combination method could make the claimed significant detection disappear.","rationale":"The paper's most valuable contribution is the first observational nova DTD in M31, and the oldest bin (7.9-14.1 Gyr) is recovered consistently by all four isochrone sets. The reader correctly flags the 2-3.2 Gyr bin as marginal, and our stress-test confirms this is the principal load-bearing weakness of the central claim. However, the reader's 'weakest assumption' field emphasizes extinction bias in the catalog; we consider the model-combination ambiguity and internal inconsistency about Padova to be more directly decisive, because it is a testable flaw within the paper's own presented data rather than an indirect external bias. The concrete check would settle whether the claimed second detection survives proper combination. If it does not, the paper should be revised to downgrade the second detection or present it as a candidate, which is consistent with a CONDITIONAL verdict. The reader already recommended CONDITIONAL, so our read does not change that verdict; it only sharpens the required correction. The argument is not fundamentally unsound, and no ad hominem or overstatement is intended.","tokens_in":17591,"tokens_out":4876,"duration_ms":44632,"concrete_test":"Recompute the combined 2-3.2 Gyr posterior using an explicit, reproducible combination rule (e.g., equal-weight mixture of the MIST, PARSEC, and Padova posteriors, or a hierarchical model with a systematic offset term) and check whether the 68.27% HPD interval excludes zero. If the interval includes zero, revise the abstract and Table 1 to report only an upper limit for this bin. Independently cross-check the Section 4 sentence claiming a Padova detection against the Table 1 entry '< 2.0e-8'.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of two robust detections depends on the 2-3.2 Gyr bin. Table 1 lists Padova as '< 2.0e-8' (a 2-sigma upper limit) and BaSTI as '< 3.9e-8' in this bin, yet Section 4 states that 'the other three isochrone sets (MIST, Padova, and PARSEC) also yield a detection'. The quoted combined DTD (excl. BaSTI) reports a detection at (3.7^{+6.8}_{-3.5} +/- 2.1)e-9. The combination procedure is only described as 'averaging only the results of the PARSEC, Padova, and MIST posteriors', with no explicit rule for incorporating Padova's non-detection posterior. Averaging an upper-limit posterior with two detection posteriors will shift the combined posterior toward zero, and the reported significance is undefined without a precise scheme. If the 68.27% HPD of the properly combined posterior includes zero, the robust-detection claim for this bin fails and the headline result reduces to one secure detection (7.9-14.1 Gyr) plus upper limits.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the first attempt to measure a delay-time distribution (DTD) for classical nova progenitors in M31, using spatially resolved star formation histories from the PHAT survey (Williams et al. 2017) and the historical M31 nova catalog of Pietsch et al. (2007). The method follows Maoz & Badenes (2010) and Badenes et al. (2015): nova counts in 826 spatial cells are modeled as a convolution of the stellar mass formed in seven age bins with a common DTD, and the DTD amplitudes are inferred with dynamic nested sampling. The analysis is repeated with four isochrone sets to assess systematic uncertainties. The paper reports two statistically significant DTD detections, at 2–3.2 Gyr and at 7.9–14.1 Gyr, with upper limits in other bins, and compares completeness-corrected rates to theoretical DTDs from Kemp et al. (2021) and to the Type Ia supernova DTD.","tokens_in":17867,"tokens_out":6084,"duration_ms":62814,"significance":"If the results hold, this is a valuable first empirical nova DTD for an external galaxy: the oldest-bin detection appears robust across all four isochrone sets, the use of four independent stellar evolution models is a genuine strength, and the comparison with BPS predictions is made without tuning the data to the models. The derived upper limit on the fraction of nova progenitors that become Type Ia supernovae is also an interesting constraint. The significance of the paper is reduced, however, by the fragility of the 2–3.2 Gyr detection, which is one of the two headline claims and is not consistently recovered by the Padova isochrone set. The central methodology is sound, but the reporting and combination scheme need to be corrected before the results can be accepted as stated.","major_comments":[{"comment":"The claim that the 2–3.2 Gyr bin is a robust detection is internally inconsistent with Table 1. The table lists Padova as an upper limit (< 2.0e-8) and BaSTI as an upper limit (< 3.9e-8) in this bin, yet Section 4 states that 'the other three isochrone sets (MIST, Padova, and PARSEC) also yield a detection.' This is a direct contradiction of the paper's own detection criterion in Section 3, which defines a non-detection when the lower edge of the 68.27% HPD region reaches the edge of the parameter space. Moreover, the 'combined (excl. BaSTI)' entry averages only the PARSEC, Padova, and MIST posteriors, but no exact rule is given for averaging a posterior that is an upper limit with two detection posteriors. Averaging an upper-limit posterior will shift the combined posterior toward zero, and the reported 68% HPD of (3.7^{+6.8}_{-3.5})e-9 is not reproducible without that rule. Please specify the exact combination procedure, report the actual Padova HPD for this bin, and state whether the properly combined HPD excludes zero. If it does not, the abstract and the conclusion that there are 'two statistically significant detections' must be revised.","section":"Section 4 / Table 1"},{"comment":"The spatial completeness of the nova catalog against extinction in the dusty disk is load-bearing for the young-age bins and for the overall DTD shape. The paper acknowledges this concern and argues against a large missing disk population using the PNe distribution from Shafter & Irby (2001), but that is an indirect argument. Because the PHAT survey already provides a resolved extinction map (Dalcanton et al. 2015) that is used in the SAD fitting, the analysis could directly test robustness by repeating the DTD fit with cells weighted by extinction or by excluding high-AV regions, or by modeling an incompleteness term as a function of cell extinction. As written, the possibility that disk novae are undercounted is acknowledged but not quantitatively bounded, and this directly affects the credibility of the 2–3.2 Gyr detection and the interpretation of a declining DTD.","section":"Section 5 / Section 2.1"},{"comment":"The absolute-rate comparisons, including the statement that the completeness-corrected nova DTD is about four orders of magnitude higher than the SN Ia DTD, depend on the effective survey length of 38 years. That number is derived from the assumption that the final 3000-day bin is complete, while the paper itself notes that Darnley et al. (2006) infer a rate about twice as high. The effective survey length should carry an uncertainty and that uncertainty should be propagated into the completeness-corrected DTDs and into the derived SN Ia fraction. Without this propagation, the claimed 'strong upper limit' on the SN Ia fraction is not fully supported.","section":"Section 5 / Figure 6"}],"minor_comments":[{"comment":"Equation (4) uses Ẍ_{i,j} in the numerator, but the model has only defined M_{i,j}; this appears to be a typographical artifact and should be corrected to M_{i,j} for consistency.","section":"Section 3, Eq. (4)"},{"comment":"The text says the authors 'calculated the edges of the 5% highest probability density (HPD) region' and then took the midpoint as the maximally likely rate; this is likely intended to be a different percentile (e.g., 50% or the posterior mode), and the wording should be clarified.","section":"Section 3"},{"comment":"The caption singles out BaSTI for lacking a detection in the 2–3.2 Gyr bin, but Table 1 shows that Padova also has only an upper limit in that bin; the caption should be updated to match Table 1.","section":"Figure 3 caption"},{"comment":"The rolling average of the nova rate in panel (c) is shown without uncertainty bands, which makes the claim of consistency with Capaccioli et al. (1989) and Shafter & Irby (2001) difficult to judge; adding uncertainties would strengthen the completeness discussion.","section":"Figure 1"},{"comment":"The paper states that data and code 'will be shared upon a reasonable request'; for a Bayesian analysis whose central claim depends on the exact posterior combination scheme, public release of the code and posterior samples would be much more appropriate and should be considered.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The referee report identifies an internal contradiction between Section 4 and Table 1 regarding the Padova 2–3.2 Gyr result. This is the key issue: the headline claim of two robust detections depends on a properly defined combination that includes Padova's upper limit, and the current text does not provide that scheme. I would recommend asking the authors to either provide a reproducible combination rule and a defensible combined HPD, or revise the abstract and conclusions to report one secure detection plus an upper limit/marginal signal in the 2–3.2 Gyr bin. A quantitative extinction-completeness test using the PHAT dust map would also materially strengthen the paper."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the headline: this is a genuine new result—the first empirical delay time distribution for classical novae, using 253 unique novae and the PHAT stellar age maps. The oldest-bin detection (7.9–14.1 Gyr) is robust across all four isochrone sets and is the core of the paper. The second claimed detection, at 2–3.2 Gyr, is much less secure.  The method is a sensible application of the Maoz & Badenes DTD recovery to novae. The authors treat the historical catalog carefully (recurrent nova removal, completeness corrections, coordinate uncertainties) and use four isochrone sets to quantify systematic error. The comparison to Kemp et al. and the SN Ia DTD is useful and appropriately cautious.  The soft spots are real. Section 4 says Padova “also yields a detection” in the 2–3.2 Gyr bin, but Table 1 lists Padova as an upper limit (<2.0e-8) and BaSTI as an upper limit too. That is a factual error in the text. The combination of the three posteriors (PARSEC, Padova, MIST) is not specified well enough to know whether the reported detection survives. Averaging an upper-limit posterior with two detection posteriors can shift the result toward zero, and the quoted 68.27% HPD lower bound (0.2e-9) is close enough to zero that a slightly different combination could cross it. The stress-test note is on target. Also, the paper says code is available upon reasonable request—given the inference machinery, a public pipeline would help. The dust incompleteness concern is real but handled as well as possible with existing data; the planetary nebula argument is indirect, and I'd call it a caveat rather than a fatal flaw.  Who is this for? People working on novae, binary evolution, and DTD methods. It deserves review—the oldest-bin measurement alone is a useful contribution—but the authors need to correct the Padova statement, clarify the combination scheme, and downgrade the 'two robust detections' language to one robust detection plus marginal evidence. I'd send it to referees with the expectation of major revision.","headline":"First observational nova DTD for M31 is a real step forward, but the claimed second detection is marginal and the text overstates Padova; deserves revision, not rejection.","tokens_in":18501,"tokens_out":3740,"would_cite":true,"duration_ms":33662,"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 derives the first statistically significant delay time distribution for classical novae in M31, finding two progenitor populations: stars aged 2-3.2 Gyr and stars older than 7.9 Gyr.","keywords":["Classical novae","Delay time distribution","M31","Stellar age distribution","Binary stellar evolution","PHAT survey","Bayesian inference","Nova progenitors"],"falsifier":"A near-infrared or radio nova survey across the PHAT footprint that is complete through dust and finds a large population of disk novae concentrated in young stellar cells would falsify the spatial-completeness assumption; depending on the size of that population, the 2-3.2 Gyr detection could disappear when those systems are included. A more direct test is to re-run the full delay time distribution recovery using the southern PHAT extension once published: if the 2-3.2 Gyr signal is absent in the doubled sample, the published detection was likely statistical noise.","tokens_in":17329,"feed_emoji":"💥","tokens_out":8764,"duration_ms":79396,"temperature":0.7,"pith_summary":"This paper derives the first delay time distribution for classical novae in Andromeda (M31), connecting where novae explode to the ages of the stars around them. A delay time distribution measures how many nova-producing systems form per unit stellar mass as a function of how long after a burst of star formation the systems turn on. The result has two statistically significant detections: a progenitor population with ages 2-3.2 Gyr and another with ages 7.9-14.1 Gyr, together with upper limits at other ages. The distribution is consistent with either a constant production efficiency or a higher production efficiency at earlier delay times. If correct, this gives the first observational handle on the evolutionary timescales and formation efficiencies of nova progenitors in a large galaxy.","feed_headline":"First nova delay-time map of Andromeda finds two age peaks","feed_subtitle":"Novae trace 2-3 Gyr and ancient stars; rates tighten the single-degenerate Type Ia supernova channel.","key_machinery":"The central object is the delay time distribution itself: the rate per unit stellar mass at which nova-producing binaries form as a function of delay after a star-formation burst. The machinery recovers it by treating it as the common vector $\\Psi_j$ in the discrete convolution $R_i(t_0)=\\sum_j M_{i,j}\\Psi_j$, where $M_{i,j}$ is the stellar mass formed in spatial cell $i$ at age bin $j$ from the Panchromatic Hubble Andromeda Treasury (PHAT) stellar age distribution map and $R_i(t_0)$ is the present-day nova rate in that cell. The inversion uses dynamic nested sampling to explore the seven-bin parameter space and a modified chi-squared likelihood ($\\chi^2_\\gamma$ from Mighell) that folds the variance of the stellar mass map into the variance of the predicted nova counts. The 826 spatial cells with different stellar compositions serve as simultaneous constraints on one galaxy-wide delay time distribution.","core_discovery":"The paper's central claim is that the spatial correlation between 253 unique historical novae and the spatially resolved stellar age distribution of the M31 disk, measured from HST photometry, can be inverted to recover the delay time distribution of nova production. In the recovery, the delay time distribution $\\Psi_j$ is common to all 826 spatial cells and is convolved with the stellar mass formed in each age bin $M_{i,j}$ to predict each cell's nova count; the predicted counts are fit to observed counts with a Bayesian nested-sampling routine. The recovered delay time distribution has statistically significant signal in two age bins: $(3.7^{+6.8}_{-3.5}\\pm 2.1)\\times 10^{-9}$ events per solar mass of formed stars for delay times of 2-3.2 Gyr and $(4.8^{+1.0}_{-0.9}\\pm 0.2)\\times 10^{-9}$ events per solar mass for 7.9 Gyr to the age of the universe, with the quoted uncertainties being statistical and systematic in that order. The result is consistent across three of four isochrone models, with BaSTI an outlier; all bins together are consistent with a decaying, constant, or mildly non-monotonic formation efficiency.","pith_inferences":["If the southern extension of the PHAT footprint doubles the nova sample, a disappearing 2-3.2 Gyr signal would point to small-sample noise, while a sharpening signal would support a merger-driven burst of nova progenitors.","Carrying the same inversion to M33, the LMC, or the SMC, where resolved stellar age maps and transient catalogs exist, would test whether the nova delay time distribution is universal or shaped by each galaxy's star formation history.","The less-than-0.1 percent single-degenerate conversion fraction implied by the comparison with the Type Ia supernova delay time distribution is a direct test for population synthesis models: models that route more than that fraction of nova binaries into Type Ia explosions are in conflict with these data.","A larger, extinction-insensitive nova sample could resolve the predicted roughly 40 Myr turn-on delay when the first white dwarfs form, a regime the current seven-bin map cannot probe."],"forward_implications":["The nova population of M31 is not dominated exclusively by very old progenitors: a statistically significant population arises from 2-3.2 Gyr old stars, implying that intermediate-age stellar populations can produce a meaningful share of classical novae.","The delay time distribution and its upper limits are consistent with either a constant or a declining production efficiency with delay time, so a uniform nova formation rate cannot yet be ruled out.","Completeness-corrected, the nova delay time distribution is about four orders of magnitude above the observed Type Ia supernova delay time distribution, implying a strong upper limit of $\\lesssim 0.1\\%$ on the fraction of nova-producing binaries that later explode as Type Ia supernovae in star-forming galaxies like M31.","The 2-3.2 Gyr signal could reflect a burst of nova progenitor formation associated with the 2-3 Gyr old galactic merger in M31, though the wide uncertainties caution against over-interpretation.","Extending the analysis to the southern half of M31, which the paper notes is underway, would roughly double the nova sample and improve the time resolution of the delay time distribution."],"supporting_citations":[{"why":"It establishes the convolution-based delay time distribution recovery formalism that the paper applies to M31.","marker":"Maoz & Badenes 2010"},{"why":"It demonstrates delay time distribution recovery from spatially resolved stellar populations and sets the posterior and upper-limit conventions used here.","marker":"Badenes et al. 2015"},{"why":"It is the source of the historical M31 nova catalog matched to the stellar age map.","marker":"Pietsch et al. 2007"},{"why":"It supplies the spatially resolved stellar age distribution map in 826 spatial cells that serves as the mass matrix for the inversion.","marker":"Williams et al. 2017"},{"why":"It provides the PHAT survey photometry from which the resolved stellar age map is built.","marker":"Dalcanton et al. 2012"},{"why":"It derives the modified chi-squared statistic used in the likelihood, which stays unbiased at low nova counts.","marker":"Mighell 1999"},{"why":"It gives the theoretical binary population synthesis nova delay time distribution used for comparison and for interpreting accretor and donor populations in the detected age bins.","marker":"Kemp et al. 2021"},{"why":"It provides the observed Type Ia supernova delay time distribution whose comparison yields the small single-degenerate fraction from nova systems.","marker":"Maoz et al. 2012"},{"why":"It is the planetary nebula spatial distribution argument the paper uses to claim that dust does not dominate the bulge-to-disk nova ratio.","marker":"Shafter & Irby 2001"},{"why":"It supplies the dynamic nested sampling package used for the Bayesian posterior exploration of the delay time distribution parameters.","marker":"Speagle 2020"}],"fun_headline_variants":["M31 nova delay times show two distinct progenitor ages","Andromeda novae trace stellar ages: 2-3 Gyr and ancient","First nova delay-time distribution in M31 has two peaks","Nova birth ages in Andromeda: youngish and very old stars"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the historical nova catalog is not strongly biased against faint or dust-hidden novae in the disk, so the measured spatial correlation between novae and younger stellar populations is real rather than artificially weakened.","fun_headline_variants_meta":{"raw":{"variants":["M31 nova delay times show two distinct progenitor ages","Andromeda novae trace stellar ages: 2-3 Gyr and ancient","First nova delay-time distribution in M31 has two peaks","Nova birth ages in Andromeda: youngish and very old stars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000156,"raw_usage":{"total_tokens":1265,"prompt_tokens":1038,"completion_tokens":227,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":654,"completion_tokens_details":{"reasoning_tokens":151}},"tokens_in":654,"tokens_out":227,"duration_ms":3316,"temperature":1.0,"reasoning_tokens":151,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:21:27.581419+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A near-infrared or radio nova survey across the PHAT footprint that is complete through dust and finds a large population of disk novae concentrated in young stellar cells would falsify the spatial-completeness assumption; depending on the size of that population, the 2-3.2 Gyr detection could disappear when those systems are included. A more direct test is to re-run the full delay time distribution recovery using the southern PHAT extension once published: if the 2-3.2 Gyr signal is absent in the doubled sample, the published detection was likely statistical noise.","supporting_citations":[{"cited_title":"2007, A&A, 465, 375, doi: 10.1051/0004-6361:20066812","cited_arxiv_id":null,"evidence_quote":"It is the source of the historical M31 nova catalog matched to the stellar age map."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It derives the modified chi-squared statistic used in the likelihood, which stays unbiased at low nova counts."}],"review_version":1}