{"id":"941958a1-6597-4664-a8b7-a4af8bc64a2f","arxiv_id":"2607.18911","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":2,"one_line_summary":"Bayesian fits of a deformed Woods–Saxon WKB model to fourteen proton-decay half-lives show widening parameter-uncertainty bands near Z=82, which the paper reads as a signature of potential softening.","lead":"This paper uses Bayesian statistics to fit a deformed proton-decay model to one measured half-life for each of fourteen exotic nuclei, extracting ten fitted parameters per nucleus and their uncertainties. It interprets widening uncertainty bands near Z=82 as evidence that statistical error profiles can reveal nuclear shell-structure softening.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Prior-dominated posterior widths likely cause the Z≈82 σ-band expansion; absent prior hyperparameters, the structural-softening mapping is unsupported.","rationale":"The reader's weakest_assumption identifies the same load-bearing issue: one half-life per nucleus cannot identify 10 parameters, and with unspecified prior widths the posterior widths can reflect priors, not structure. My stress test sharpens this into a decisive falsifiable check: a synthetic-data experiment with structureless deformations should reproduce the claimed Z≈82 widening if the paper's interpretation is wrong. Since the manuscript does not provide the prior hyperparameters or any contraction-ratio or prior-sensitivity diagnostic, the current evidence for the central claim is insufficient. This warrants the same REJECT verdict as the reader, not because the method is necessarily invalid, but because the central spectroscopic interpretation is not yet supported. The proposed synthetic-data rerun is inexpensive and would settle whether the concern actually lands; if the widening disappears in the synthetic data, the paper could be reconsidered.","tokens_in":27385,"tokens_out":3813,"duration_ms":40485,"concrete_test":"Generate synthetic half-life data for the same 14 nuclei from the same WKB model using fixed, Z-independent true deformation parameters (e.g., β2=0.3, β4=0.05 with no near-Z=82 softening), after first reporting the actual prior hyperparameters used in Eqs. (2)–(3). Run the identical MCMC pipeline on these synthetic data and compare posterior contraction ratios σ_post/σ_prior for β2 and β4 as functions of Z. If the β2/β4 σ bands still widen systematically near Z=82 (contraction ratios approaching 1), the observed widening is an identifiability artifact rather than structural softening. This single simulation directly distinguishes the prior-dominated explanation from the paper's physical interpretation.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim requires that the σ bands in Figs. 7 and 9 carry information about the nuclear potential, not about the prior. That condition is not established. Eq. (2) supplies one experimental half-life per nucleus while 10 parameters are sampled, so each likelihood is a single scalar constraint. Eq. (3) defines independent Gaussian priors, but the hyperparameters {σ_αi} and {ᾱ_i} are never listed, and no prior-sensitivity analysis appears. In a 10-parameter model constrained by one datum, any direction in which the half-life is insensitive will have posterior = prior. Near a magic shell, β2 and β4 variations can have small net effect on the WKB half-life; the posterior then simply inherits the prior width. The observed widening at Z≈82 is therefore exactly what an unidentifiable, prior-dominated fit would produce, without any softening of the physical potential. The signed sensitivity index in Eq. (21) uses posterior percentiles to set parameter perturbations, so it cannot separate how much a parameter matters from how broad its prior is. The statement in Sec. III.C that multidimensional degeneracy inflates the marginalized bands is asserted rather than demonstrated: the reported correlation matrices for heavy nuclei do not show the strong β2–β4 or β2–r0 correlations that such compensation would require. The abstract's mapping claim is thus not supported by the presented evidence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript develops a Bayesian uncertainty-quantification framework for the deformed Woods–Saxon potential with a semi-classical WKB description of odd-A proton radioactivity in the Z = 50–82 region. For each of the 14 proton emitters, ten model parameters (V0, a0, r0, P0, Rc0, β2, β4, Vso0, Rso, aso) are calibrated to the single experimental half-life through MCMC (emcee), using an independent Gaussian prior per parameter (Eq. 3) and a likelihood containing only experimental uncertainties (Eq. 2). The paper reports posterior medians and credible intervals, Pearson correlation matrices, a signed relative sensitivity index (Eq. 21), and posterior predictive half-life bands (Fig. 9). The central claim, stated in the abstract and conclusion, is that the systematic expansion and volatility of the σ bands near the Z = 82 shell closure reflect a structural softening of the nuclear potential, so that parametric error profiles can be mapped onto potential energy configurations.","tokens_in":27781,"tokens_out":14183,"duration_ms":119527,"significance":"The methodological program—full Bayesian posterior sampling for a decay-model calibration, with covariance diagnostics and a quantitative sensitivity index—is timely and appropriate for a field that has largely relied on deterministic point fits, and the figures are informative. If the central claim were established, it would constitute a new spectroscopic use of proton-decay systematics. Credit is due for providing MCMC diagnostics and 68% credible intervals for all nuclei, and for flagging sloppiness in 113Cs. However, the paper's significance rests entirely on the proposition that the posterior widths carry physical information about the nuclear potential. That proposition is not established: the priors are unspecified, the model is underdetermined with one datum per nucleus, the predictive check is circular, and the reported correlation matrices contradict the accompanying narrative. As presented, the mapping claim is unsupported.","major_comments":[{"comment":"The prior hyperparameters {ᾱ_i, σ_{α_i}} of Eq. (3) are never given. Each nucleus contributes exactly one experimental half-life to the likelihood (Eq. 2), so ten parameters are constrained by a single scalar; any direction in which the half-life is insensitive has posterior equal to prior, and no prior-sensitivity analysis is supplied. Tables I–II indicate this limit is reached: in Table II the spin-orbit parameters are nearly identical for all 14 nuclei (V_so0 ≈ 6.2 ± 1.0 MeV, a_so = 0.75 ± 0.05 fm), and several β2/β4 credibles in Table I span the full plausible range (e.g., 171Au: β2 = 0.14(+0.21/−0.44); 121Pr: β4 = 0.08(+0.33/−0.35)). The Z ≈ 82 widening of the deformation bands in Fig. 7 could therefore be an identifiability/prior artifact rather than a physical softening—precisely the distinction the central claim relies on.","section":"Sec. II A (Eq. 3); Tables I–II; Fig. 7"},{"comment":"Fig. 9 is presented as validation: the posterior predictive median and 1σ/2σ bands 'tightly encompass' the experimental half-lives. But those half-lives are the same data that enter the likelihood (Eq. (2) via Eq. (20)); for a flexible model with wide posteriors, containing the fitted points is unsurprising. The further statement in Sec. III D that this 'demonstrates that the framework does not suffer from parameter overfitting' does not follow and is not tested: there is no held-out nucleus, no leave-one-out analysis, and no comparison with an independent observable (e.g., a second decay branch or a spectroscopic factor from a different reaction). A genuine out-of-sample check is required before the predictive bands can be claimed to carry structural information.","section":"Sec. III D, Fig. 9"},{"comment":"The sensitivity index S_i in Eq. (21) is the half-life response to shunting the parameter between the 16th and 84th marginals of the posterior. When the marginal is prior-dominated, those percentiles are fixed by the prior, not by the data; S then cannot separate 'the parameter matters for the half-life' from 'the prior is broad'. The Sec. III C account of the near-Z=82 regime—that degeneracy 'geometrically inflates the marginalized posterior uncertainty bands'—is asserted, but the reported heavy-nucleus correlations do not show the compensating channels: for 171Au (Fig. 6) β2–β4 = 0.038 and β2–r0 = 0.069. The hierarchy in Fig. 8 should be recomputed with explicit priors and with a data-driven perturbation scale (e.g., credible intervals of a profile likelihood).","section":"Sec. III C, Eq. (21)"},{"comment":"The text in Sec. III A claims that the off-diagonal correlation elements 'remain consistently small for almost all parameter pairs' and that deformation–geometry correlations are 'near-zero'. The figures contradict this: for 109I (Fig. 4) β2–r0 = 0.798 and β2–a0 = 0.611; for 145Tm (Fig. 5) a0–r0 = −0.519 and r0–P0 = −0.439. These are substantial correlations, and the same manuscript uses strong degeneracy in 113Cs as a central example. The presence or absence of parameter compensation is used both to exclude (Sec. III A) and to invoke (Sec. III C) degeneracy; this internal inconsistency must be resolved.","section":"Sec. III A vs Figs. 4–6"},{"comment":"The experimental input—half-lives, Q_p values, parent/daughter spins and parities, and the assigned uncertainties σ_i—is never tabulated; Fig. 9 displays only the ratio log10(T_Bayes/T_exp). The likelihood of Eq. (2) is therefore non-reproducible, and the claimed agreement cannot be independently checked. The data table (or a direct reference to it) is required.","section":"Sec. III (dataset)"}],"minor_comments":[{"comment":"Typo in the first sentence: 'poseterior' should be 'posterior'.","section":"Sec. II A"},{"comment":"The radial kernel Kλ is typeset with garbled exponents ('rλ(λ−2) −1 Rλ−2c'); rewrite in standard notation and check dimensions.","section":"Eq. (13)"},{"comment":"Eq. (21) defines a 'signed' index, but Fig. 8 plots |S_i|; state explicitly whether the sign information is discarded or retained.","section":"Eq. (21), Fig. 8"},{"comment":"The abstract and text refer to the 'Z = 50–82 region', but the fitted sequence is Z = 53–81 (no Z = 50, 52, or 82 nucleus appears); the abstract should match the data.","section":"Abstract"}],"recommendation":"reject","confidential_remarks":"For the editor: the reader's negative assessment is, on my reading, accurate. The load-bearing weakness is that the central claim (σ-band widening near Z=82 as a structural-softening signature) cannot be separated from prior choice and model underdetermination with the information currently in the paper, and the claimed validation is circular. The internal contradictions between Sec. III A's 'no degeneracy' claim and Figs. 4–5, and between that claim and the sloppiness discussion for 113Cs, reinforce the concern. These are not presentation-level fixable; the analysis would need an explicit prior specification, prior-sensitivity or profile-likelihood checks, and an out-of-sample validation to be considered. I recommend rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nQuick take: this paper does something new and useful—first MCMC calibration of the deformed Woods–Saxon WKB model for odd-A proton emitters, with posteriors, correlation matrices, and a sensitivity ranking. That part is competently executed. But the headline claim, that σ-band widening near Z=82 maps onto structural softening, is not supported by the evidence. The fit is 10 parameters per nucleus against a single half-life, the prior hyperparameters in Eq. (3) are never stated, and the 'posterior predictive' agreement in Fig. 9 is evaluated on the same data used in the likelihood. Any widening of posterior bands under those conditions is exactly what an unidentifiable, prior-dominated fit produces; it does not require the potential to soften.\n\nWhat is genuinely good: the author is transparent about sloppiness in 113Cs, the correlation matrices are shown in full, and the sensitivity index is clearly defined. The Bayesian methodology itself is standard, and the application to this decay mode is a legitimate extension of ref. [47]. The paper also honestly notes use of AI tools for polishing; no issue there.\n\nThe soft spots are not tiny. The sensitivity index in Eq. (21) sets perturbations from posterior percentiles, so a parameter with a wide prior automatically looks important—it cannot separate physical sensitivity from prior width. The claimed degeneracy-driven band inflation near Z=82 is contradicted by the paper's own correlation matrices: most off-diagonal elements are near zero, so there is no evidence that β2–β4 or β2–r0 compensation inflates the marginalized bands. The text asserts this in Sec. III C but the figures don't show it.\n\nWould I recommend engaging with it? Yes, for a referee. The application is a first, the code appears reproducible, and the issues are fixable in revision: specify prior hyperparameters, run a prior-sensitivity analysis, include a held-out nucleus or cross-validation, and temper the conclusion to 'posterior widths are currently dominated by prior assumptions' rather than a structural probe. As it stands, the central claim is circular and should not be published as is.","headline":"Useful Bayesian application to deformed proton decay, but the central mapping claim rests on unidentified priors and one data point per nucleus.","tokens_in":28197,"tokens_out":2131,"would_cite":false,"duration_ms":20322,"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":"Bayesian uncertainty bands in proton-decay half-lives widen near the Z=82 shell closure, and the paper reads this as a map of nuclear potential softness.","keywords":["proton radioactivity","Bayesian uncertainty quantification","deformed Woods–Saxon potential","WKB approximation","parameter sensitivity","shell structure","half-life prediction","MCMC"],"falsifier":"Generate synthetic half-lives from the same WKB model with fixed deformation parameters (no shell-dependent softening), run the same MCMC calibration with identical priors, and check whether the σ bands still widen near Z=82; if they do, the widening is an inference artifact. Alternatively, repeat the fit with prior widths varied by a factor of 10 and see whether the band expansion persists.","tokens_in":27260,"feed_emoji":"☢️","tokens_out":3464,"duration_ms":31759,"temperature":0.7,"pith_summary":"This paper argues that the widths and volatility of Bayesian uncertainty bands in calculated proton-decay half-lives carry physical information about nuclear shell structure. Calibrating a deformed Woods–Saxon potential plus WKB tunneling model to one experimental half-life per odd-A emitter in the Z=50–82 region, the author extracts ten-parameter posteriors for fourteen nuclei. The study reports that posterior predictive medians reproduce experimental half-lives across orders of magnitude, that sensitivity hierarchies reorder with mass, and that near the Z=82 shell closure the σ bands of deformation parameters systematically expand. The proposed reading is that this expansion reflects a softened nuclear potential with shape coexistence, so parametric error profiles might be mapped onto potential energy configurations. If correct, Bayesian UQ becomes a spectroscopic tool that reads shell structure from decay systematics.","feed_headline":"Bayesian error bands expose potential softening at Z=82","feed_subtitle":"Fitting one half-life per nucleus, posterior σ widths expand where the potential softens; decay errors become structure probes.","key_machinery":"The machinery is a deformed Woods–Saxon potential (radius and diffuseness with β2, β4 angular dependence), a multipole-expanded Coulomb field, an orientation-averaged WKB tunneling width, and a Bayesian calibration using MCMC with independent Gaussian priors. The signed relative sensitivity index (half-life shift from 16th to 84th percentile of each marginal posterior) converts posterior widths into fractional half-life sensitivities. The central object doing the interpretive work is the Z-dependence of the 1σ/2σ credible-interval widths of the deformation parameters.","core_discovery":"Constrained by experimental half-lives, the posterior distributions of the deformed potential parameters show little pairwise correlation for most nuclei, with the exception of 113Cs where sloppiness appears. The posterior predictive half-lives fall within the 1σ and 2σ bands for every nucleus without systematic bias. The signed relative sensitivity indices reveal that radius r0 and quadrupole deformation β2 dominate, with β4 active only in two windows near Z=59 and Z=75–79. As Z approaches 82, the 1σ and 2σ bands of β2 and β4 expand and become volatile, which the author interprets as evidence that the nuclear potential softens near the shell closure; the mapping between error-profile struct","pith_inferences":["The widening of σ bands near Z=82 could be an identifiability artefact of fitting ten parameters to one datum; an independent check would re-fit with synthetic data from a flat potential and see whether similar band structure emerges.","If the mapping is physical, the same pipeline applied to alpha decay or cluster emission should show analogous band expansion at other magic numbers (Z=50, N=82, N=126).","A hierarchical Bayesian model that pools information across the Z sequence could tighten the posteriors and directly test whether the per-nucleus widths reflect structure rather than prior spread.","Sensitivity indices computed at posterior medians may miss nonlinear compensation; a global variance-based sensitivity analysis on the posterior could refine the hierarchy claims."],"forward_implications":["If the error-profile mapping holds, posterior σ-band widths become an observable that can be compared across models and used to locate shell closures.","Bayesian calibration turns a single half-life per nucleus into ten parameter posteriors, and the posterior predictive distributions provide half-life uncertainties that can be propagated to other decay observables.","Mass-dependent re-ordering of sensitivities means no fixed parameter ranking can be assumed; future fits must allow hierarchy changes along the Z sequence.","Detection of sloppiness in 113Cs warns that some nuclei admit degenerate parameter combinations that point values cannot reveal."],"fun_headline_variants":["Proton decay errors expose soft shell at Z=82","Why error bands balloon near Z=82 in proton decay","Mapping nuclear softness from proton radioactivity errors","Error profile structure mirrors deformed potential at Z=82","Decay data: Bayesian bands show potential softening at Z=82"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The assumption that a single experimental half-life per nucleus, combined with independent Gaussian priors whose widths are not stated, is enough to identify ten model parameters per nucleus, so the posterior widths and their Z-dependence reflect nuclear structure rather than prior choices or unidentifiability.","fun_headline_variants_meta":{"raw":{"variants":["Proton decay errors expose soft shell at Z=82","Why error bands balloon near Z=82 in proton decay","Mapping nuclear softness from proton radioactivity errors","Error profile structure mirrors deformed potential at Z=82","Decay data: Bayesian bands show potential softening at Z=82"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000212,"raw_usage":{"total_tokens":1261,"prompt_tokens":759,"completion_tokens":502,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":503,"completion_tokens_details":{"reasoning_tokens":423}},"tokens_in":503,"tokens_out":502,"duration_ms":5527,"temperature":1.0,"reasoning_tokens":423,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T13:57:27.600001+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Generate synthetic half-lives from the same WKB model with fixed deformation parameters (no shell-dependent softening), run the same MCMC calibration with identical priors, and check whether the σ bands still widen near Z=82; if they do, the widening is an inference artifact. Alternatively, repeat the fit with prior widths varied by a factor of 10 and see whether the band expansion persists.","supporting_citations":[],"review_version":1}