{"id":"58668db6-4b13-4cdc-ae34-3cc6e40f4c14","arxiv_id":"2502.04831","paper_version":3,"verdict":"REJECT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":6,"one_line_summary":"An empirical model splits projectile ions into four atomic-number groups, fits a logistic curve to measured mean charge states per group, and uses a Gaussian with an earlier width formula to produce charge state distributions.","lead":"This paper fits four logistic curves to existing measurements of mean charge states of ions passing through solid foils, then combines them with a Gaussian spread to predict charge state distributions. The reported agreement is computed on the same data used for the fit, so the result is an interpolation rather than an independent prediction.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The model's precision claim is unsupported because all fitted parameters are calibrated and validated on the same Shima et al. data; no holdout test establishes transferability to the claimed full Z1-Z2-energy range.","rationale":"The reader identified the same core weakness: the model is calibrated and validated on the same Shima et al. data, so the reported agreement is expected and does not establish predictive accuracy. My reading of the manuscript confirms this concern is load-bearing. The four-group logistic forms (Eqs. 16–19) are fits to the carbon-target mean charge states in Figs. 5–6; the Table I corrections are then adjusted post hoc to bring non-carbon targets into agreement in Fig. 8. Every subsequent CSD comparison, including the heavy-ion cases in Figs. 15–16, uses these same fitted data. Thus the paper demonstrates interpolation, not prediction. I considered other potential issues, such as the unnormalized Gaussian in Eq. 12 and discontinuities at the piecewise boundaries in Table I, but those are secondary: even if the functional forms were physically motivated and smooth, the absence of any out-of-sample test would still leave the central precision claim unsupported. Therefore I keep the reader's REJECT verdict unchanged. A straightforward holdout cross-validation, such as dropping one projectile species from the fit and predicting its CSDs, would directly settle whether the fitted parameters generalize or merely memorize the database.","tokens_in":15218,"tokens_out":5545,"duration_ms":56358,"concrete_test":"Perform a leave-one-projectile-out cross-validation on the Shima et al. dataset: remove all data for one projectile, e.g. 35Cl, refit Eqs. 16–19 and all Table I correction coefficients on the remaining data, then predict the excluded CSDs and compare with experiment. If the held-out RSM for F(q) remains comparable to the in-sample RSM, transferability is supported; if it degrades substantially, the corrections are memorizing the training data.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing assumption is that Eqs. 16–19 and the 22 piecewise corrections in Table I, whose free parameters are fixed by the same experimental data later used for validation, remain accurate for unmeasured projectile-target-energy combinations. The central claim—precise CSDs over the entire 1–4 MeV/u range with Z1 up to 92 and Z2 up to 92—therefore rests entirely on out-of-sample transferability, and the paper supplies no holdout test. In particular, Eq. 19 for group IV (54 < Z1 <= 92) is acknowledged to be fitted to very few points, and the Table I corrections are introduced only after Fig. 8 shows systematic deviations; the subsequent agreement in Figs. 12–16 is then an in-sample check. The reported RSM values (e.g., 0.0017 for C/Si) quantify fit residuals, not predictive accuracy, and no error bars or uncertainty propagation are given. Without a held-out set, the apparent precision could be pure overfitting, so the claim that the model predicts CSDs independently of experimental charge fractions is unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an empirical model to compute mean charge states and charge-state distributions (CSDs) of projectile ions passing through solid targets in the 1-4 MeV/u energy range. The model divides projectiles into four atomic-number groups, fits a logistic function of the Schiwietz-Grande reduced parameter x0 to data from Shima et al. (Refs. [9,21]), adds 22 piecewise correction terms for heavy targets (Z2 >= 12) in Table I, and combines the resulting mean charge with the Novikov-Teplova width in a Gaussian to obtain F(q). The authors compare the resulting mean charges, widths, and CSDs with experimental values from the same Shima et al. data and report excellent agreement.","tokens_in":15461,"tokens_out":4930,"duration_ms":52227,"significance":"If the model could be validated out-of-sample, it would be practically useful: it offers an explicit procedure for estimating full CSDs without running a charge-exchange code, covers heavy ions up to uranium, and makes a concrete falsifiable claim that a single Gaussian with the Novikov-Teplova width reproduces F(q) for the full Z1 range. The paper also clearly identifies the limitation of the previous single-formula models for Z1 > 16. However, the current evidence is fit-quality evidence, not prediction evidence: no independent data set, holdout split, or cross-validation is used, and the four-group structure plus the many correction terms are chosen ad hoc from the same data. No uncertainty analysis is provided, so the claimed precision over the entire Z1-Z2-energy range is not established.","major_comments":[{"comment":"The validation is circular. The logistic parameters in Eqs. (16)-(19) and the 22 correction entries in Table I were chosen by fitting to the same Shima et al. experimental data (Refs. [9,21]) that are later compared in Figs. 7, 9, and 11-16. The reported agreement is therefore an in-sample residual metric, not a test of predictive accuracy. The authors should either compare with independent experimental data not used in the fitting or perform a holdout/cross-validation study (e.g., train on a subset of projectile-target-energy combinations and test on the remaining ones) before claiming precise prediction over the full range.","section":"Section IV and Section V"},{"comment":"The paper acknowledges that the group IV fit (54 < Z1 <= 92) is based on very few points: only Au at several energies and a few points for Pb and U. A three-parameter logistic fitted to such a sparse set cannot support the claim of precise prediction for all Z1 up to 92. Please provide parameter uncertainties, indicate the actual number of data points used, and either add more data or restrict the claimed validity range for group IV.","section":"Section IV, Eq. (19)"},{"comment":"The RSM values (e.g., 0.0017 for C/Si) are computed on the same data used for fitting, so they quantify training residuals rather than prediction error. Moreover, no error bars are shown for the experimental data in Figs. 1, 7, and 9-16, and there is no propagation of uncertainties from the fitted parameters to the final F(q) values. Without such information, the stated precision cannot be evaluated.","section":"Section V, Eq. (10)"},{"comment":"The Novikov-Teplova width formula introduces extra empirical parameters (alpha = 0.23, beta = 0.32, and C as a function of Z1 and Z2) that themselves are fitted to experimental data. The paper does not state the fitting database for this formula or test whether it remains accurate for heavy targets and the heaviest projectile groups. This is load-bearing because the final CSD claim depends on both the new mean-charge model and this width formula.","section":"Section II, Eq. (7)"}],"minor_comments":[{"comment":"There are spacing typos in the title and affiliations: 'throu gh' and 'In dia' should be corrected.","section":"Title page"},{"comment":"Equation (9) has an unbalanced parenthesis: the expression contains an extra closing bracket after (Z1^{−0.45} v1/v')^{-1/0.6}. Please check and correct the formula.","section":"Section II, Eq. (9)"},{"comment":"The exponent of Z2 in Eq. (14) is ambiguous ('Z_2^{-0.019Z - 0.52 v1/v0} / 1.68'); please clarify the intended power-law form and define all symbols in the equation.","section":"Section II, Eq. (14)"},{"comment":"In Eq. (15), x' is called the midpoint, but the fitted constants in Eqs. (16)-(19) are not identified as x' values; please state explicitly which fitted parameter corresponds to x'.","section":"Section IV, Eq. (15)"},{"comment":"Many panels are very small and the projectile/target/energy labels are difficult to read, especially in Figs. 11-16. Please enlarge fonts and ensure each panel's labels are legible.","section":"Section V, Figs. 3-4 and 11-16"},{"comment":"The abstract states the range as 1 MeV/u < E < 4 MeV/u, but several validation data points lie below 1 MeV/u (e.g., Cl on C at 0.688 MeV/u in Fig. 3 and Cu on C at 0.558 MeV/u in Fig. 4). Please reconcile the claimed energy range with the data used.","section":"Abstract and Section IV"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is essentially an empirical fitting exercise. The lack of any holdout or independent validation is a serious deficiency for the central predictive claim. If the authors add a proper out-of-sample test, provide the fitting code, and present uncertainty estimates, a resubmission could be considered."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the paper gives a simple closed-form recipe for mean charge states and Gaussian charge-state distributions across a wide Z1 range. That is genuinely useful as an interpolation tool for accelerator people. But the evidence for \"precise\" prediction is in-sample only. Eqs. (16)-(19) and Table I are fit to the Shima et al. data that Figs. 7-16 compare against. No holdout set, no error bars, no code. The abstract's claim overreaches.\n\nWhat is actually new: the four-group logistic parametrization of q_m/Z1 against the SGM reduced parameter x_o, plus the piecewise target corrections. Going back to the older x_o is a reasonable move, and the four-band pattern in Fig. 5(a) is visually convincing. Combining the fitted q_m with the Novikov-Teplova width and a Gaussian is not new, but the complete package is a practical tool. The RSM values are honest fit residuals; they just do not support the conclusion's predictive language.\n\nSoft spots: the validation design is the main one. With 12 logistic parameters plus 22 correction cells, the model will reproduce its training data. The non-carbon agreement is expected because Table I was adjusted post hoc after Fig. 8 showed systematic deviations. Eq. (19) for group IV rests on very few points (Au, Pb, U), so it is fragile. There are no uncertainty estimates, so we cannot tell whether the quoted RSM values are stable. Minor: the authors say they plan to develop a code, so the method is not directly usable yet, and there is no quantitative head-to-head against ETACHA or GLOBAL, only a qualitative statement of their limitations. The reference list looks appropriate; the main prior models are cited.\n\nWho this is for: accelerator physicists who need rough CSD estimates for stripper design and do not have experimental charge fractions. For that purpose the formula is probably better than nothing, and I would not call it a waste. But I would not cite it as a validated predictive model.\n\nRecommendation: send it to peer review only if the journal is willing to require a genuine out-of-sample test. If the authors can show even one projectile-target-energy combination held out from the fit, the paper becomes worthwhile. As is, it is a useful technical note, not a validated predictive model.","headline":"A useful four-group logistic interpolation of published mean charge states, but the paper overclaims predictive precision because all parameters are fit and validated on the same Shima et al. data.","tokens_in":16027,"tokens_out":2757,"would_cite":false,"duration_ms":31686,"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 claims that using four separate logistic curves for mean charge, one per projectile atomic-number group, yields charge-state distributions close to measured values for all solid targets in the 1-4 MeV/u range.","keywords":["charge-state distribution","mean charge state","empirical model","logistic function","solid targets","ion-atom collisions","accelerator physics","Novikov-Teplova width"],"falsifier":"A concrete check: refit Equations 16-19 and Table I on half of the data in Shima et al. and Ball et al., then compare predictions on the other half; if the held-out mean charges and widths scatter by more than the deviations shown in Fig. 9, the claimed precision is a fitting artifact.","tokens_in":14965,"feed_emoji":"⚛️","tokens_out":8155,"duration_ms":82808,"temperature":0.7,"pith_summary":"This paper proposes an empirical recipe for predicting the charge-state distribution of an ion beam after it passes through a solid target in the 1-4 MeV/u range used by tandem accelerators. The authors claim that grouping projectile ions by atomic number into four classes, fitting each class with its own logistic curve for the mean charge, and spreading the distribution with a Gaussian whose width comes from the Novikov-Teplova formula reproduces measured distributions for projectiles from carbon to uranium ($Z_1$ up to 92). If correct, the recipe would deliver the full distribution $F(q)$ without requiring experimental charge fractions as input, something the current standard mean-charge formulas cannot do. The practical value is in estimating usable beam intensities at any charge state after a stripper foil, which is exactly what accelerator and ion-atom collision planning needs.","feed_headline":"Four fitted curves predict ion charge states from carbon to uranium","feed_subtitle":"Predicting the full charge-state spectrum without experimental input would help design stripper foils and heavy-ion beam lines","key_machinery":"The engine of the model is a four-branch logistic description of the mean charge state. Using the Schiwietz-Grande reduced parameter $x_o$, a scaled projectile velocity that absorbs $Z_1$ and $Z_2$ dependence, the paper fits $q_o^m/Z_1$ to the logistic form $L/(1+\\exp(-K(x_o-x')))$ separately for $Z_1 \\le 10$, $10 < Z_1 \\le 18$, $18 < Z_1 \\le 54$, and $54 < Z_1 \\le 92$. A table of 22 piecewise corrections in $Z_1$, $Z_2$, and energy adjusts these predictions for non-carbon targets. The resulting $q_o^m$ centers a Gaussian charge-state distribution whose width is set by the Novikov-Teplova formula $\\Gamma(x_1)$ with $x_1 = q_o^m/Z_1$. This combination is what carries the argument: it converts a mean-charge estimate into a complete $F(q)$ without any experimental charge fractions.","core_discovery":"The central claim is that a single universal mean-charge formula is the wrong level of description for the intermediate-energy solid-target regime: the reduced parameter $x_o$ from Schiwietz and Grande organizes the data into four distinct bands, and each band needs its own logistic fit. The paper asserts that the four fitted curves (Eqs. 16-19) give mean charges $q_o^m$ in agreement with experiment on carbon targets, and that the addition of 22 piecewise target corrections (Table I) extends the agreement to heavy targets. With $q_o^m$ from this construction and the distribution width of Novikov and Teplova, a Gaussian ansatz for $F(q)$ yields charge-state distributions the authors find very close to the measured ones across the whole projectile range. The paper claims this supplies complete CSDs for $Z_1$ up to 92 in the 1-4 MeV/u energy range without using experimental charge fractions as input.","pith_inferences":["Not reported in the paper: the four logistic curves and the 22 target corrections are fitted to the same experimental set used for validation, so a leave-one-out or holdout test on unseen projectile-target-energy combinations would decide whether the model predicts or interpolates.","A natural extension, not pursued here, is to test the same four-group logistic structure just outside the stated 1-4 MeV/u window; the piecewise target corrections would reveal whether the energy bands are physical or just fitting artifacts.","The same Gaussian-plus-width construction could be applied to gaseous targets or to thinner foils before charge equilibrium is reached, domains where the structure of the corrections may need to change."],"forward_implications":["Accelerator users could read off the full charge-state spectrum after a stripper foil for ions from carbon to uranium in the intermediate-energy range, not just the mean charge.","Beams of non-dominant charge states, often needed for specific experiments, could be estimated without repeated calibration runs.","The model offers an empirical benchmark against which fully theoretical CSD codes can be tested in the regime where heavier-than-Ni-like ions are hard to treat.","Because the recipe is algebraic, it is suitable for online beam-dynamics and code-based accelerator planning."],"supporting_citations":[{"why":"Supplies the reduced parameter $x_o$ (Eq. 14) whose four-band structure motivates splitting projectiles into four groups.","marker":"[11]"},{"why":"The ISGM formula whose failure for $Z_1 > 16$ is the problem the paper sets out to solve; also the baseline for RSM comparisons.","marker":"[12]"},{"why":"Provides the experimental charge-state fraction data on carbon and the Gaussian distribution form the model uses for $F(q)$.","marker":"[9]"},{"why":"Provides experimental mean charge states for heavier projectiles and targets used in fitting and comparison.","marker":"[21]"},{"why":"Supplies the Novikov-Teplova width formula $\\Gamma(x_1)$ used to set the Gaussian width.","marker":"[22]"},{"why":"Alternative mean-charge and width formulas whose larger RSM led the authors to choose ISGM and NAT.","marker":"[19]"},{"why":"SRIM code used to convert incident beam energies to emergent energies for all comparisons.","marker":"[23]"}],"fun_headline_variants":["Four fitted curves predict ion charge states up to uranium","Replace one universal formula with four for better charge states","Ion charge states in solids depend on four distinct fits","New model yields precise charge states for solid targets","Four empirical formulas cover carbon to uranium charge states"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The model assumes that four logistic curves plus 22 target corrections, all fitted to the Shima et al. carbon-target data, transfer to every solid target and every energy in the 1-4 MeV/u range without being tested on data that were not used in fitting.","fun_headline_variants_meta":{"raw":{"variants":["Four fitted curves predict ion charge states up to uranium","Replace one universal formula with four for better charge states","Ion charge states in solids depend on four distinct fits","New model yields precise charge states for solid targets","Four empirical formulas cover carbon to uranium charge states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000295,"raw_usage":{"total_tokens":1706,"prompt_tokens":930,"completion_tokens":776,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":546,"completion_tokens_details":{"reasoning_tokens":701}},"tokens_in":546,"tokens_out":776,"duration_ms":9109,"temperature":1.0,"reasoning_tokens":701,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T21:17:32.745337+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check: refit Equations 16-19 and Table I on half of the data in Shima et al. and Ball et al., then compare predictions on the other half; if the held-out mean charges and widths scatter by more than the deviations shown in Fig. 9, the claimed precision is a fitting artifact.","supporting_citations":[{"cited_title":"Specifically, the width parameter here makes it obvious that the model cannot predict F(q) independently","cited_arxiv_id":null,"evidence_quote":"Supplies the reduced parameter $x_o$ (Eq. 14) whose four-band structure motivates splitting projectiles into four groups."},{"cited_title":"To examine which one of these agrees better in this comparison, we have tested the data by using a quantity called relative sum metrics (RSM)","cited_arxiv_id":null,"evidence_quote":"The ISGM formula whose failure for $Z_1 > 16$ is the problem the paper sets out to solve; also the baseline for RSM comparisons."},{"cited_title":"4 - (h), (i), (k), the other 17 data sets show a clear mismatch between the empirical and experimen- tal CSD","cited_arxiv_id":null,"evidence_quote":"Provides the experimental charge-state fraction data on carbon and the Gaussian distribution form the model uses for $F(q)$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides experimental mean charge states for heavier projectiles and targets used in fitting and comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Novikov-Teplova width formula $\\Gamma(x_1)$ used to set the Gaussian width."},{"cited_title":"9 that the 9 0 0.2 0.4 Present 0 0.2 0.4 Expt","cited_arxiv_id":null,"evidence_quote":"Alternative mean-charge and width formulas whose larger RSM led the authors to choose ISGM and NAT."},{"cited_title":"Sharma and T","cited_arxiv_id":null,"evidence_quote":"SRIM code used to convert incident beam energies to emergent energies for all comparisons."}],"review_version":1}