{"id":"862c10f4-9cc9-42c9-923b-22cef0fcd9a2","arxiv_id":"2412.05524","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"The birth mass function of neutron stars is a steep power law peaking at about 1.27 solar masses, strongly favored over the double-Gaussian model.","lead":"Using mass measurements of 90 neutron stars, the authors infer that neutron stars are born with a narrow range of masses peaking around 1.27 solar masses, with a steep decline toward higher masses. This contradicts the widely used double-peaked model and sharpens the link between supernova explosions and the fate of massive stars.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Claimed 3-sigma preference for the turn-on power law over the double-Gaussian model is inconsistent with the paper's own false-alarm rate of 2.5%, which corresponds to about 2 sigma.","rationale":"The reader's weakest assumption is the lack of a full selection function, and that is a legitimate concern. However, I identify a more direct, internally testable issue: the paper's own false-alarm simulation yields 2.5%, which contradicts the '3 sigma' wording in the abstract and main text. This is not a matter of external consensus but of internal inconsistency, and it strikes at a headline quantitative claim. The paper otherwise deserves credit: it provides reproducible code and data, cross-checks two accreted-mass prescriptions, handles the correlated DNS likelihoods explicitly, and discusses selection effects in Section G. Those strengths mean the shape of the inferred birth-mass function may be robust, but the significance should be reported consistently with the simulation-based false-alarm rate. Because the reader's verdict is already CONDITIONAL and my concern reinforces the need for a condition (revising or better substantiating the significance claim), I recommend no change to the verdict category, though the condition should now include the false-alarm discrepancy.","tokens_in":29998,"tokens_out":11339,"duration_ms":107414,"concrete_test":"Reproduce the simulation behind Extended Data Figure 4 using the released code (https://github.com/GW-BNUZ/NSbirthMass) with 1000 realizations of synthetic 2G datasets (with the same sample size and noise properties). Count the fraction of realizations with ln BF_TOP,2G > 5.8. If this false-alarm fraction is approximately 2.5%, the '3 sigma' claim should be revised to about 2 sigma; if it is below 0.3%, the original claim stands. As a second check, rerun the model comparison with alternative hyperpriors (e.g., wider 2G means, narrower TOP alpha priors) to test whether Bayes-factor values above 300 are robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim includes a quantitative significance statement: the TOP model is 'strongly favoured against the widely-adopted empirical double-Gaussian model at the 3 sigma level' (abstract and main text), with a Bayes factor greater than 300 and odds of 0.003 described as 'equivalently ~3 sigma'. Yet the paper's own calibration in Section E and Extended Data Figure 4 reports a false-alarm probability of 2.5% for the observed ln BF = 5.8 under the null (2G) model, based on 200 simulated datasets. A false-alarm rate of 2.5% corresponds to about 1.96 sigma (one-sided) or about 2.2 sigma (two-sided), not 3 sigma. The 'odds 0.003' conversion treats the posterior probability of the 2G model, under an equal-prior assumption, as if it were a Gaussian p-value; this is not a calibrated frequentist statement. The simulation-based false-alarm probability is the paper's own repeated-sampling calibration and is an order of magnitude larger than the 0.3% implied by a 3-sigma claim. This is an internal inconsistency: the same paper reports both the 3-sigma claim and the 2.5% false-alarm probability. Since the 3-sigma statement is a headline result and a key reason the paper is influential, the strength of evidence for the unimodal turn-on power law is overstated. The inferred shape may still be correct, but the significance claim requires revision.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper compiles mass measurements for 90 neutron stars, separates them into recycled and non-recycled subclasses, and applies probabilistic accreted-mass corrections to the recycled systems using an analytical prescription and a phenomenological prescription. The corrected birth-mass posteriors are then fed into a hierarchical Bayesian analysis comparing fifteen parametric models. The authors report that a turn-on power-law (TOP) model is strongly preferred over the empirical two-Gaussian model, with a Bayes factor greater than 300, and infer a birth-mass function that turns on near 1.1 Msun, peaks near 1.27 Msun, and declines as a steep power law with index about 6.5. They further connect this shape to supernova progenitor physics, arguing that stars with initial masses above roughly 18 Msun do not form neutron stars.","tokens_in":30307,"tokens_out":9241,"duration_ms":102676,"significance":"If the central claim is correct, the paper overturns the widely used double-Gaussian description of the neutron-star birth-mass function and replaces it with a unimodal, steeply falling distribution. Such a result would have broad implications for supernova explosion modelling, double-star evolution, and the neutron-star equation of state. The paper has notable strengths: it uses joint posterior information for double neutron stars, tests two independent accreted-mass prescriptions, compares a large family of parametric models, and includes a repeated-simulation false-alarm calibration. The data and analysis scripts are publicly available. However, the quantitative significance attached to the headline model comparison is not supported by the paper's own calibration, and the selection-function treatment is lighter than the strength of the population-level claims.","major_comments":[{"comment":"The '3 sigma' claim is inconsistent with the paper's own false-alarm calibration. In Section E, the authors report a false-alarm probability of 2.5% for the observed ln Bayes factor of 5.8 against the two-Gaussian null, based on 200 simulated datasets. A 2.5% one-sided or two-sided false-alarm rate corresponds to roughly 1.96 sigma, not 3 sigma. The alternative conversion quoted in the main text, from posterior odds of 0.003 to '~3 sigma', treats an equal-prior posterior probability as a Gaussian p-value and is not a calibrated frequentist statement. The abstract's 'strongly favoured at the 3 sigma level' therefore overstates the evidence by the paper's own metric. The authors should either report the calibrated false-alarm significance or revise the significance language throughout.","section":"Abstract and Section E, Extended Data Figure 4"},{"comment":"The reported Bayes factor of 'greater than 300' uses the TOPmax variant rather than the TOP model as defined in the text. For the ANA data set, Extended Data Table 3 gives ln BF(TOP vs uniform) = 15.4 and ln BF(2G vs uniform) = 10.3, giving ln BF = 5.1 and BF about 164, whereas the value 5.8 quoted in Extended Data Figure 4 corresponds to the TOPmax row (16.1 - 10.3 = 5.8). Since the paper says it does not distinguish TOP from TOPmax, the model whose evidence is being quoted is not uniquely defined. The distinction matters because the Bayes factor is a headline quantity, and the pure TOP versus 2G comparison is weaker than the quoted value.","section":"Extended Data Table 3 and Section E"},{"comment":"The inference treats the compiled sample as representative of the underlying neutron-star population with respect to mass, and this assumption is load-bearing for the inferred steep power-law tail. The checks in Section G only demonstrate the absence of obvious correlations between measured mass and spin period or luminosity for a subset of the sample; they do not quantify a selection function. If high-mass neutron stars are preferentially detectable through timing precision, orbital geometry, or survey selection, the slope of the high-mass tail would be biased. The closing statement that the results are 'robust against potential selection effects' is stronger than the presented evidence supports, and the authors should either add a selection model or explicitly weaken this claim.","section":"Section G, Extended Data Figure 10"},{"comment":"The agreement between the semi-analytic supernova model and the TOP distribution is obtained after adjusting model parameters (beta = 3.1, alpha_turb = 1.10, zeta = 0.85, tau_1.5 = 1.2 s) specifically to reproduce the observed steep power-law shape. This is a useful consistency check, but it should be presented as an illustrative demonstration rather than independent confirmation of the progenitor-mass interpretation, since the parameters are tuned to the target distribution.","section":"Section F"}],"minor_comments":[{"comment":"The text contains a typo: 'evolvd' should be 'evolved'.","section":"Methods Section B"},{"comment":"The phrase 'were G is the gravitational constant' should read 'where G is the gravitational constant'.","section":"Equation (1)"},{"comment":"The word 'logrithmic' should be 'logarithmic'.","section":"Section E"},{"comment":"The caption says the simulation uses 87 neutron star mass measurements while the main text describes 90 compiled neutron stars; the authors should state explicitly that the three recycled neutron stars without measured spin parameters are excluded from the ANA dataset used in the simulation.","section":"Extended Data Figure 4 caption"},{"comment":"The models 'TOPmax' and 'TOPG' are named and compared in Extended Data Table 3 but not defined in Section D; since the TOP model in Equation (20) already contains an upper cutoff through H(mmax - m), the difference between TOP and TOPmax should be clarified.","section":"Section D and Section E"},{"comment":"The sentence describing the maximum-mass cutoff as supported with a Bayes factor of 2 should clarify that this is marginal evidence, and the abstract should avoid giving the impression that the maximum mass is a firmly constrained feature of the model.","section":"Section E"}],"recommendation":"major_revision","confidential_remarks":"The paper is well executed and reproducible, and the Bayesian model comparison is likely to survive in broad shape, but the headline significance statement needs to be recalibrated. The discrepancy between the quoted 3 sigma claim and the paper's own 2.5% false-alarm probability is an internal inconsistency rather than a mere presentational issue, and the treatment of selection effects is not sufficient to support the strength of the population-level claims. I therefore recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read. The substantive result is new and likely important: they fit a turn-on power law to the birth masses of 90 neutron stars, correcting recycled pulsar masses for accretion, and find strong model preference over the usual double Gaussian. The analysis is thorough: joint posteriors for double neutron stars, two accreted-mass prescriptions, many parametric models, and a simulation-based false-alarm check. They ship code and data. The inferred shape—steep decline from a peak at 1.27 Msun, minimum near 1.1 Msun—is consistent across data subsets and prescriptions, and the radio-only subset reproduces it. That is real evidence.\n\nThe main worry is the significance statement. The abstract and text say '3 sigma' based on Bayes factor >300 and odds 0.003. But their own simulation in Section E and Extended Data Figure 4 gives a 2.5% false-alarm probability for the observed ln BF=5.8 under the double-Gaussian null. That is about 2 sigma, not 3. The odds-to-sigma conversion implicitly treats an equal-prior posterior probability as a Gaussian p-value, which is not calibrated. This is not a fatal flaw in the shape inference—the TOP model still wins—but the headline number is overstated and should be corrected.\n\nSelection effects are the other soft spot. The sample is a heterogeneous compilation of pulsars, X-ray binaries, and GW events. They check mass versus spin period and luminosity and see no correlation, and they discuss binary survival bias, but they do not construct a full selection function. If high-mass neutron stars are preferentially measured, the steep tail could be biased. This is a legitimate concern, though it is not obvious it would produce the specific TOP shape; I would treat it as a caveat, not a refutation.\n\nSection F is labeled as tuning: they adjust semi-analytic supernova model parameters to reproduce the power-law. That is fine as an illustration of plausibility, but it should not be read as independent support.\n\nOverall: the central argument holds up. The paper deserves serious peer review and will likely be influential. The needed revision is to fix the significance claim and be more explicit about the unmodeled selection function. I would send it to a good journal with those conditions.","headline":"A careful, reproducible inference that the neutron-star birth mass function is a turn-on power law, but the headline 3-sigma claim is not supported by the paper's own false-alarm calibration.","tokens_in":30922,"tokens_out":1506,"would_cite":true,"duration_ms":14953,"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":"Neutron stars are born near 1.3 solar masses, not in two peaks","keywords":["neutron star birth mass function","recycled pulsars","accretion correction","turn-on power law","double Gaussian","Bayesian model selection","supernova progenitors","neutron star masses"],"falsifier":"A larger, independently selected sample of slow (non-recycled) neutron stars with precise masses, especially in double neutron star systems, would settle the shape: if such a sample shows a clear excess near 1.8 $M_\\odot$ or a flat high-mass tail instead of a steep decline, the turn-on power law is wrong.","tokens_in":29783,"feed_emoji":"🌟","tokens_out":2253,"duration_ms":24694,"temperature":0.7,"pith_summary":"This paper tries to establish that the birth masses of neutron stars follow a single, smoothly rising and steeply falling distribution, not the widely assumed double-peaked (two-Gaussian) shape. Using mass measurements of 90 neutron stars and correcting for mass gained by recycled pulsars during accretion, the authors find a minimum birth mass near 1.1 $M_\\odot$, a peak near 1.27 $M_\\odot$, and a steep power-law decline. If correct, the result changes how astronomers link supernova progenitors to neutron stars and refines predictions for gravitational-wave signals from merging neutron stars. The paper also argues that the dearth of massive neutron stars implies single stars heavier than roughly 18 $M_\\odot$ do not leave neutron stars.","feed_headline":"Neutron stars are born near 1.3 solar masses, not in two peaks","feed_subtitle":"Re-analysis of 90 neutron star masses finds a single peak and a steep high-mass tail, overturning the double-Gaussian picture.","key_machinery":"The central object is the turn-on power-law (TOP) distribution, a parametric mass function that rises from a minimum mass over a smooth transition width and then declines as $m^{-\\alpha}$. The argument runs on two supporting mechanisms: (1) a probabilistic accretion correction that subtracts the mass each recycled pulsar gained during spin-up, derived from accretion spin-up theory and calibrated against binary evolution calculations; and (2) a hierarchical Bayesian model-selection framework that compares TOP against single- and multi-Gaussian, power-law, log-normal, gamma, and skewed Student-t models using Bayes factors. The accretion correction is what converts observed masses of recycled pulsars into birth masses, and the model comparison is what establishes TOP as the preferred shape.","core_discovery":"The central claim is that the birth mass function of neutron stars is unimodal and well described by a turn-on power law: the distribution rises smoothly from a minimum mass of $1.10^{+0.04}_{-0.05}\\,M_\\odot$, peaks at $1.27^{+0.03}_{-0.04}\\,M_\\odot$, and then declines with a power-law index of $6.5^{+1.3}_{-1.2}$. This model is strongly favoured over the empirical two-Gaussian model, with a Bayes factor greater than 300, corresponding to roughly $3\\sigma$ significance. The paper further claims that the steep tail implies that stars with initial masses above approximately 18 $M_\\odot$ rarely form neutron stars, consistent with the observed absence of massive red supergiant progenitors of supernovae.","pith_inferences":["If the TOP shape holds, future neutron star mass measurements should show a dearth of objects near 1.8 $M_\\odot$ compared with what the two-Gaussian model predicts; a single well-measured cluster near that mass would challenge the conclusion.","The same accretion-correction machinery could be applied to the growing sample of neutron star-black hole merger events to test whether the inferred birth mass function is consistent across formation channels.","The steep slope implies that binary neutron star merger rates and mass ratios are dominated by low-mass primaries, which may sharpen predictions for the electromagnetic counterparts and post-merger emission of nearby mergers."],"forward_implications":["The neutron star birth mass function has a single peak near 1.27 $M_\\odot$, so the double-peaked structure inferred from observed masses is largely an artifact of accretion and small samples.","Single stars with initial masses above about 18 $M_\\odot$ rarely produce neutron stars, narrowing the allowed progenitor mass range for core-collapse supernovae.","The steep power-law decline means most neutron stars are born near the low end of the mass range, which shifts predicted post-merger gravitational-wave frequencies for binary neutron star mergers.","Marginal evidence for a maximum mass cutoff around $2.36^{+0.29}_{-0.17}\\,M_\\odot$ leaves room for a small population of very massive neutron stars but does not require one.","The TOP shape may be inherited from the initial mass function of massive stars, but only if explosion physics allows substantial net accretion after shock revival in more massive progenitors."],"supporting_citations":[{"why":"Provides the empirical two-Gaussian model and its parameter priors that the paper compares against and ultimately disfavours.","marker":"ref.16"},{"why":"Supplies the accretion spin-up theory that underlies the analytical model for mass gained by recycled pulsars.","marker":"ref.24"},{"why":"Gives the standard recycling framework, spin-up line formalism, and accretion-mass relations used to estimate birth masses.","marker":"ref.27"},{"why":"Provides numerical simulations of the recycling process used to calibrate the phenomenological accreted-mass prescription.","marker":"ref.28"},{"why":"Supplies the semi-analytic supernova models used to connect the inferred power-law neutron star mass function to progenitor and explosion physics.","marker":"ref.21"},{"why":"Provides the observed spin-up line for millisecond pulsars, which anchors the analytical birth-mass correction.","marker":"ref.110"},{"why":"Supplies high-precision mass measurements for many recycled pulsars in the compiled sample.","marker":"ref.74"},{"why":"Establishes the joint-posterior treatment for correlated masses in double neutron star systems, which the paper adopts.","marker":"ref.133"}],"fun_headline_variants":["Neutron star births: single peak at 1.27 solar masses","Birth masses of neutron stars: one peak, steep drop","Neutron star birth masses overturn double-peak model","Neutron stars born near 1.27 solar masses, not in two peaks","Single birth mass peak for neutron stars: 1.27 solar masses"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The inference assumes that the 90 compiled neutron stars with measured masses fairly represent the underlying neutron star population, so that no strong mass-dependent selection effect distorts the shape of the distribution.","fun_headline_variants_meta":{"raw":{"variants":["Neutron star births: single peak at 1.27 solar masses","Birth masses of neutron stars: one peak, steep drop","Neutron star birth masses overturn double-peak model","Neutron stars born near 1.27 solar masses, not in two peaks","Single birth mass peak for neutron stars: 1.27 solar masses"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000186,"raw_usage":{"total_tokens":1300,"prompt_tokens":898,"completion_tokens":402,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":514,"completion_tokens_details":{"reasoning_tokens":307}},"tokens_in":514,"tokens_out":402,"duration_ms":4215,"temperature":1.0,"reasoning_tokens":307,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T20:38:02.053406+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A larger, independently selected sample of slow (non-recycled) neutron stars with precise masses, especially in double neutron star systems, would settle the shape: if such a sample shows a clear excess near 1.8 $M_\\odot$ or a flat high-mass tail instead of a steep decline, the turn-on power law is wrong.","supporting_citations":[],"review_version":1}