{"id":"3d2d1047-de75-479d-9a0a-b1e1fa6d2bef","arxiv_id":"2504.17866","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"The LArQL model for ionization and scintillation in liquid argon was re-fitted globally to existing charge and light data, yielding updated parameters with reported uncertainties and slightly better overall agreement.","lead":"A physics collaboration re-fitted the parameters of LArQL, an existing model that describes how liquid argon converts particle energy into electric charge and scintillation light. The updated parameters fit published ionization and light data slightly better, which matters for large neutrino detectors that use liquid argon time projection chambers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reported SSR reductions are computed on the same data used to set the weights and search ranges, so the claimed improvement over original LArQL parameters may be an in-sample fitting artifact rather than a robust parameter update.","rationale":"The reader's weakest assumption is the adequacy of the two phenomenological functional forms. I agree that this limits the physical interpretation of the parameters, but the paper's central claim is a parameter update within the LArQL model, not a validation of the functional forms; a re-fit can reduce residuals even if the form is imperfect. The more load-bearing issue is statistical: the WSSR weights in Eq. (3.1) are computed from intermediate LArQL fits to the same data, the parameter search is restricted to a neighborhood of the original values, and the quoted SSR reductions are in-sample. This does not make the fit invalid, but it means the paper does not yet rule out that the improvement is an artifact of the fitting procedure. A leave-one-dataset-out or fixed-weight refit would directly test whether the updated parameters generalize. The reader did flag the model-derived weights in the rationale, but not as the weakest assumption, so my agreement is partial. The existing CONDITIONAL verdict is appropriate and should be retained until such a check is reported.","tokens_in":56,"tokens_out":7953,"duration_ms":151119,"concrete_test":"Recompute the Table 1 fit with WSSR weights fixed to inverse published point-to-point variances instead of RMS_i_best from individual LArQL fits, and compare SSR_i^fit/SSR_i^original for every dataset. If any dataset's relative SSR is no longer reduced, or if parameter shifts exceed the quoted uncertainties, the claimed all-dataset improvement depends on the model-derived weighting and is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Sec. 3 the WSSR is minimized with weights 1/[n_i (RMS_i_best)^2], where RMS_i_best is obtained from an intermediate LArQL fit to each dataset individually. The same ICARUS and ARIS datasets are then used to quote the SSR_i^fit/SSR_i^original reductions. Because the weights and the parameter ranges ('narrowed down ... around the values originally adopted') are derived from the same data and model, the claim that every dataset improves is not an independent statement; a 9-parameter fit with model-tuned weights will typically lower in-sample WSSR. The paper reports no cross-validation, no bootstrap, and no uncertainty on the SSR ratios, and the only external check (MicroBooNE dQ/dx, Fig. 2) is visual and unquantified. The functional-form concern raised by the reader is real but secondary: re-fitting within a fixed phenomenological model is legitimate. What would actually support the central claim is evidence that the residual reduction survives when weights are fixed independently or when the fit is evaluated on data not used to determine them.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports a re-fit of the parameters of the LArQL phenomenological model, which describes ionization charge and scintillation light in liquid argon as a function of electric field and dE/dx. The authors minimize a weighted sum of squared residuals (Eq. 3.1) over the Birks parameters (A_B, k_B) and the four parameters of the escape-electron correction (alpha, beta, A, B, C, D) using ICARUS charge data [7] and ARIS/scintillation data [5,8]. They obtain a new parameter set (Table 1) and report that the sum of squared residuals for each individual dataset is reduced by less than 10% relative to the original LArQL parameters. They also show a qualitative comparison with MicroBooNE dQ/dx data, where the updated LArQL curve lies closer to the data than the Birks model alone.","tokens_in":4785,"tokens_out":2746,"duration_ms":29178,"significance":"If the updated parameters are robust, they provide a practical improvement for LArTPC simulations, especially because LArQL is already used in the LArSoft framework. The paper is transparent that the parameters are fitted to data, and the MicroBooNE comparison is an external check that was not part of the fit, which strengthens the empirical relevance. The main limitation is that the reported improvement is evaluated on the same data used to set the weights and to narrow the parameter ranges, so the central claim of a better description is currently supported only in-sample.","major_comments":[{"comment":"The weights in Eq. (3.1) are derived from RMS_i^best obtained from an intermediate fit of the same LArQL model to each individual dataset, and the parameter ranges were narrowed using the same datasets. Consequently, the reported reductions in SSR_i^fit relative to SSR_i^original are in-sample quantities and do not by themselves demonstrate that the updated parameters generalize better. I request a cross-validation or holdout analysis, or at minimum bootstrap confidence intervals on the SSR ratios, to support the claim that the new parameters are a robust improvement rather than an artifact of fitting the calibration data.","section":"Sec. 3, Eq. (3.1)"},{"comment":"The text states that the global fit considers all datasets 'with same weight,' but Eq. (3.1) assigns per-dataset weights 1/[n_i (RMS_i^best)^2], which depend on the dataset and on an intermediate fit. If the intended meaning is that all measurements receive equal weight, the formula should be clarified; if the intended meaning is per-dataset weighting, the wording is misleading. This distinction matters for interpreting the relative importance of the charge and light datasets in the fit.","section":"Sec. 3, Eq. (3.1) and Table 1"},{"comment":"No global goodness-of-fit statistic, per-dataset chi^2, p-value, or number of data points per dataset is reported. For a nine-parameter fit with model-tuned weights, the statement that minimization provides a 'satisfactory estimate' is unsupported without a quantitative check of fit quality. Please report the actual SSR_i values, the number of measurements n_i, and a global chi^2 or equivalent statistic, together with the parameter covariance matrix.","section":"Sec. 3, Table 1"},{"comment":"The claimed improvement is conditional on the assumed functional forms f_corr = exp(-E/(alpha ln(dE/dx)+beta)) and chi0 = A/[B + exp(C + D dE/dx)]. The manuscript does not test alternative forms or examine residual structure as a function of E and dE/dx. If these phenomenological forms are inadequate in some region of the stated validity range, the re-fitted parameters are still an improvement only within the chosen ansatz. Showing residual plots for all datasets at all fields, or a brief comparison with a simple alternative form, would substantially strengthen the conclusion.","section":"Sec. 2, Eq. (2.4)"},{"comment":"The MicroBooNE comparison is presented as a visible improvement, but no quantitative metric is given for the agreement between the updated LArQL curve and the MicroBooNE data. Since this is the only external validation in the paper, I ask for a numerical measure (e.g., chi^2, RMS, or mean absolute difference over the plotted points) and a statement of whether the displayed data were used in any stage of the fit.","section":"Sec. 3, Fig. 2"}],"minor_comments":[{"comment":"There is a typo: 'stablish' should be 'establish'.","section":"Sec. 3"},{"comment":"The right panel is described as 'S1/S10' in the text but the caption uses 'S'; please define the ratio and its normalization consistently.","section":"Fig. 1 caption"},{"comment":"The stated validity range is 0-0.75 kV/cm and 2-40 MeV/cm, but the manuscript does not list the electric-field values and dE/dx ranges actually covered by the fitted datasets. A table of datasets with their field configurations and point counts would help readers assess coverage and the possibility of extrapolation.","section":"Sec. 1 and Sec. 2"},{"comment":"In the caption, it would be helpful to state explicitly that the original A, B, C, D values had no uncertainties, whereas the original A_B and k_B uncertainties are from [7]; the current wording is ambiguous.","section":"Table 1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a concise parameter-update paper, appropriate for a technical journal like JINST. The main concern is the in-sample nature of the validation; if the authors can add a simple cross-validation or bootstrap analysis and a global goodness-of-fit statistic, the paper would meet the standard for acceptance. I would also encourage the authors to make the C++/ROOT fitting code publicly available, since the paper does not currently provide it and the fit is the central result."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this paper re-fits the parameters of the LArQL phenomenological model for charge and light in liquid argon, using the same functional forms and mostly the same data as the original. It gives a new parameter table with uncertainties and claims a modest (<10%) reduction in summed squared residuals across all datasets. That claim is likely true as an in-sample statement, but it is not yet supported as a robust update.\n\nWhat's actually new is the parameter set and the correlations. The authors are honest: they say the search was restricted around the original values, and the weights in Eq. (3.1) come from intermediate individual fits of the same model to the same datasets. So the 'global' fit is a local refinement, and the reported SSR reduction is exactly what you'd expect from fitting nine parameters to data that already informed the original choice. The MicroBooNE dQ/dx comparison is external and not used in the fit, which is good, but it's only shown visually with no quantified metric. That's the soft spot, and it's the main one.\n\nThe paper does push in the right direction: it makes the fit procedure explicit, gives uncertainties for all parameters, notes the strong AB–kB correlation, and flags that the AB/kB values should not be interpreted as Birks parameters in isolation. That restraint is creditable.\n\nThe bigger question — whether the chosen functional forms for f_corr and chi0 are adequate over the full 0–0.75 kV/cm and 2–40 MeV/cm range — is not addressed, but that's a limitation stated in the model's foundations, not a new flaw introduced here. The paper is a calibration update, not a claim about the correct parametric form.\n\nBottom line: if you're working with LArQL in LArSoft and need current parameter values with tolerances, this is useful and worth knowing about. As a standalone physics paper it's minor but legitimate. I'd send it to a competent referee for the instrumentation community, mainly because the parameters are embedded in simulation software and independent verification of the fit weights and uncertainties matters. I would not require new data for acceptance, but I would ask the authors to state clearly that the SSR reduction is in-sample and to quantify the MicroBooNE comparison, even with a simple chi2 or residual plot.","headline":"A transparent, incremental re-fit of LArQL parameters; the modest improvement is plausible but in-sample, and the external MicroBooNE check needs quantification.","tokens_in":5280,"tokens_out":2020,"would_cite":true,"duration_ms":20320,"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":"A global re-fit of the LArQL model's parameters lowers residuals for every tested liquid-argon charge and light dataset and improves the match to MicroBooNE dQ/dx over Birks alone.","keywords":["liquid argon time projection chamber","LArQL model","ionization charge yield","scintillation light yield","recombination factor","escaping electrons","charge-light anti-correlation","global parameter fit"],"falsifier":"Measure the recombination factor $R$ and the scintillation ratio $S/S_0$ in liquid argon at field values not well covered by the training data, such as $\\mathcal{E} = 0.1$ kV/cm at $dE/dx = 30$ MeV/cm, with percent-level uncertainties; if the updated LArQL parameters do not reproduce the measured points within the residuals quoted for the original datasets, the functional forms are the limiting assumption. Alternatively, re-fit the same data with different two-parameter forms for $\\chi_0$ and $f_{corr}$; a meaningfully lower $WSSR$ would show the claimed improvement depends on the chosen parametrization.","tokens_in":4328,"feed_emoji":"⚛️","tokens_out":10061,"duration_ms":89605,"temperature":0.7,"pith_summary":"LArQL is a phenomenological model that describes, from one master equation, how energy deposited in liquid argon splits into ionization charge and scintillation light, including electrons that escape recombination even at zero field. This paper re-fits the model's six phenomenological parameters plus the two Birks parameters, within their known uncertainties, to published ICARUS charge data and ARIS scintillation data in a single weighted global fit. The result is a parameter table that reduces the sum of squared residuals for every dataset compared with the original LArQL parameters, with relative reductions all under ten percent, and that agrees visually with MicroBooNE charge-per-length data better than the Birks model alone. The goal is a single internally consistent charge-and-light description for LArTPCs over $\\mathcal{E}$ from 0 to 0.75 kV/cm and $dE/dx$ from 2 to 40 MeV/cm, which matters for energy reconstruction in neutrino detectors.","feed_headline":"Re-fitted LArQL model beats original on all datasets","feed_subtitle":"New global fit lowers residuals for all ICARUS, ARIS, and MicroBooNE comparisons used.","key_machinery":"The load-bearing object is the LArQL charge-light master equation $Q + L = N_i + N_{ex}$ together with the modified Birks formula of Eq. (2.3): $Q = \\frac{A_B/W_{ion}}{1 + \\frac{k_B}{\\rho_{LAr}}\\frac{1}{\\mathcal{E}}\\frac{dE}{dx}} + \\chi_0(dE/dx)\\, f_{corr}(\\mathcal{E},dE/dx)\\, Q_\\infty$. The first term is standard Birks recombination; the second adds escaping electrons, with $\\chi_0$ the fraction of electrons that escape recombination at zero field and $f_{corr}$ the empirical field dependence, both defined in Eq. (2.4) as $f_{corr} = e^{-\\mathcal{E}/(\\alpha \\ln(dE/dx)+\\beta)}$ and $\\chi_0 = A/[B+e^{C+D\\,dE/dx}]$. The argument is carried by the weighted residual objective of Eq. (3.1), which forces a single parameter set to account for charge and light together, and by the correlation structure among the fitted parameters, especially the strong $A_B$--$k_B$ correlation and the coupling of $\\alpha$ to both.","core_discovery":"On its own terms, the paper establishes that all the charge and light datasets used can be described simultaneously with a single LArQL parameter set, and that this set beats the original one on every dataset. Using a weighted sum of squared residuals ($WSSR$) and random sampling of parameter space around the original values, the fit gives $A_B = 0.808(2)$, $k_B = 49.7(4)\\ \\mathrm{g\\,V\\,MeV^{-1}\\,cm^{-3}}$, $\\alpha = 0.0387(8)\\ \\mathrm{cm/kV}$, $\\beta = 0.0128(6)\\ \\mathrm{cm/kV}$, $A = 3.61(5)\\times 10^{-3}$, $B = -5.7(1)$, $C = 1.74(2)$, and $D = 2.01(3)\\times 10^{-4}\\ \\mathrm{cm/MeV}$. Every dataset's $SSR_i$ is lower than with the original parameters and satisfies $|1 - SSR_i^\\mathrm{fit}/SSR_i^\\mathrm{original}| < 10\\%$. The paper is explicit that the fitted $A_B$ and $k_B$ are not standalone Birks values; they are only meaningful inside the full LArQL expression. In the MicroBooNE $dQ/dx$ comparison, both LArQL versions track the data while the Birks model alone slightly underestimates $dQ/dx$ at higher $dE/dx$.","pith_inferences":["A sharper test of the model would be to check whether the same two functional forms also fit liquid-xenon charge and light data; if they do, the phenomenological core generalizes beyond argon, and if they do not, the forms are argon-specific.","The reported residual reductions are all below ten percent, so the practical payoff may lie more in the MicroBooNE $dQ/dx$ comparison, a dataset outside the fit; the shape agreement there is the stronger evidence of genuine improvement.","Because the optimizer was a random sampling rather than a more systematic minimizer, a repeat fit with a gradient-based or Bayesian sampler on the same objective could confirm that the reported 'most probable' parameters are a true minimum and not just the best point found.","Computing energy resolution from the fitted parameters versus the original set would make the impact concrete: if the charge-light anti-correlation is sharper, the benefit for neutrino oscillation analyses becomes quantitative."],"forward_implications":["LArTPC simulations that adopt the updated table describe the anti-correlated charge and light channels with one parameter set instead of separate charge and light fits.","Because every individual $SSR_i$ decreased, the global fit did not improve the overall picture by sacrificing one dataset for another.","The strong $A_B$--$k_B$ correlation means the parameters should be used as a block; varying one alone damages the model's electric-field behavior.","The fitted $A_B$ and $k_B$ cannot be transplanted into standalone Birks recombination calculations; the paper restricts them to use within LArQL.","The same functional forms can be re-fitted as new charge and light datasets become available, which is the paper's stated next step."],"supporting_citations":[{"why":"Supplies the ICARUS recombination-factor ($R$) charge data used as one of the two global-fit datasets.","marker":"[7]"},{"why":"Supplies the ARIS scintillation-light ratio data used alongside the charge data in the weighted global fit.","marker":"[8]"},{"why":"Provides absolute scintillation-yield data including zero-field yields from which the escape-electron fraction $\\chi_0$ is inferred.","marker":"[5]"},{"why":"Defines the original LArQL model and its parameter set, which is the baseline that the global fit must beat.","marker":"[6]"},{"why":"Provides the MicroBooNE $dQ/dx$ data used as an external comparison to show improved agreement over Birks alone.","marker":"[9]"}],"fun_headline_variants":["Global refit of LArQL improves every dataset fit","All argon charge-light data fit with one LArQL set","Refitted LArQL beats original on every dataset","One global LArQL fit lowers residuals for all data"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the two chosen formulas—an exponential field correction and a logistic-type escape fraction—are the correct shapes for the whole 0–0.75 kV/cm and 2–40 MeV/cm range; if real physics takes different forms, the new parameter table only re-fits a wrong model.","fun_headline_variants_meta":{"raw":{"variants":["Global refit of LArQL improves every dataset fit","All argon charge-light data fit with one LArQL set","Refitted LArQL beats original on every dataset","One global LArQL fit lowers residuals for all data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000382,"raw_usage":{"total_tokens":2033,"prompt_tokens":962,"completion_tokens":1071,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":578,"completion_tokens_details":{"reasoning_tokens":1004}},"tokens_in":578,"tokens_out":1071,"duration_ms":8171,"temperature":1.0,"reasoning_tokens":1004,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T10:30:16.515421+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the recombination factor $R$ and the scintillation ratio $S/S_0$ in liquid argon at field values not well covered by the training data, such as $\\mathcal{E} = 0.1$ kV/cm at $dE/dx = 30$ MeV/cm, with percent-level uncertainties; if the updated LArQL parameters do not reproduce the measured points within the residuals quoted for the original datasets, the functional forms are the limiting assumption. Alternatively, re-fit the same data with different two-parameter forms for $\\chi_0$ and $f_{corr}$; a meaningfully lower $WSSR$ would show the claimed improvement depends on the chosen parametrization.","supporting_citations":[{"cited_title":"Marinho, L","cited_arxiv_id":null,"evidence_quote":"Supplies the ICARUS recombination-factor ($R$) charge data used as one of the two global-fit datasets."},{"cited_title":"Amoruso et al., Study of electron recombination in liquid argon with the ICARUS TPC , https://doi.org/DOI Nucl","cited_arxiv_id":null,"evidence_quote":"Supplies the ARIS scintillation-light ratio data used alongside the charge data in the weighted global fit."},{"cited_title":"Doke et al., Let dependence of scintillation yields in liquid argon, Nucl","cited_arxiv_id":null,"evidence_quote":"Provides absolute scintillation-yield data including zero-field yields from which the escape-electron fraction $\\chi_0$ is inferred."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the original LArQL model and its parameter set, which is the baseline that the global fit must beat."}],"review_version":1}