Rethinking Total Absorption Gamma Spectroscopy Deconvolution: Supervised Machine Learning vs Response-Matrix Methods

C. Fonseca-Vargas; E. N{\'a}cher; J. Balibrea-Correa; J. L. Tain

REVIEW 4 major objections 5 minor 39 references

Rethinking Total Absorption Gamma Spectroscopy Deconvolution: Supervised Machine Learning vs Response-Matrix Methods

T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read In a simulated TAGS decay with a high level density, supervised machine learning recovers individual β-feeding intensities about ten times more accurately than response-matrix unfolding—provided the training and test conditions match.

desk verdict The paper's order-of-magnitude ML superiority claim is an in-distribution artifact, and the authors' own caveats concede the point — but the benchmark itself is new and worth a referee's time. read the letter →

arxiv 2608.00090 v1 pith:3YY5JUBW submitted 2026-07-30 physics.data-an cs.LGnucl-ex

J. Balibrea-Correa , E. N{ácher , C. Fonseca-Vargas , J. L. Tain This is my paper · ORCID

classification physics.data-ancs.LGnucl-ex

keywords totalabsorptiongamma-rayspectroscopybeta-feedingdistributionunfoldinginverseproblemsupervisedmachinelearningconvolutionalneuralnetworkgradientboosteddecisiontreesresponsematrix

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper confronts the ill-posed problem of extracting β-feeding distributions from total absorption gamma-ray spectra, where many detector responses overlap. It argues that the choice of optimization target matters more than the algorithm family: response-matrix methods minimize the discrepancy between measured and reconstructed spectra, while supervised machine-learning models are trained to minimize the error on the feeding vector directly. Using a realistic simulated decay of a high-level-density nucleus, the authors find that all three supervised models (two convolutional networks and a gradient-boosted tree ensemble) outperform the three response-matrix methods by more than an order of magnitude on the L2 feeding-accuracy metric, cutting per-level uncertainty by roughly a factor of ten. They conclude that neither family should be used alone and recommend a hybrid workflow: a response-matrix method supplies the initial feeding estimate, and a supervised model trained around it refines the result. The findings would matter because TAGS is a main experimental route to reliable feeding data in the very nuclei where unfolding is hardest.

What carries the argument

The central object is the response-matrix inverse problem d = N R f, in which the measured TAS spectrum is a Poisson-noised superposition of per-level response functions; because adjacent level responses are nearly degenerate, direct inversion of R is unstable. The argument turns on the distinction between two optimization targets: response-matrix methods (differential evolution, expectation-maximization/Richardson–Lucy, and non-negative least squares) minimize the residual between measured and reconstructed spectra, while supervised machine-learning estimators (one-dimensional convolutional networks and gradient-boosted trees) minimize a loss (L1 or L2) directly on the feeding vector. This

What would settle it

Train the CNN and GBDT models exactly as described, then generate a validation set with a response matrix perturbed by a few percent (e.g., recomputed with a different detector geometry or physics list) or with feedings drawn from a different prior, and compare L2 errors. If the supervised models' advantage over NNLS/DIFEVO/BAYES collapses under either change, the order-of-magnitude claim is an artifact of matching training and test conditions rather than a general property of the methods.

Watch

Extended reading notes

Core claim

The central claim, in the authors' terms, is that 'a good spectral fit does not guarantee an accurate feeding reconstruction.' Response-matrix algorithms optimize the fit to the measured spectrum and therefore develop oscillatory biases in the feeding distribution that grow with statistics, whereas supervised machine-learning estimators are trained to minimize the error on the feeding vector itself and are statistically compatible with one another and consistently superior on the test set: on the L2 metric, all three supervised methods outperform the response-matrix methods by more than an order of magnitude, with individual feeding uncertainties reduced by up to an order of magnitude. The p

Load-bearing premise

The evaluation is in-distribution: the 1000-case validation set is generated with the same detector response matrix, the same ±10% perturbation around the same nominal feeding vector, and the same Poisson sampling used to train the models, so the claimed ML superiority assumes that this simulated setup transfers to real TAGS data with response-model errors, background, contaminants, and different feeding priors.

Editorial extensions

If this is right

If the simulated results transfer to experiment, TAGS analyses of complex decays can push per-level feeding uncertainties down by about an order of magnitude at high statistics.
The hybrid workflow—response-matrix initial estimate, then supervised ML refinement—becomes the paper's practical recommendation for handling real data.
The decoupling result implies that reporting only reconstructed-spectrum χ² is insufficient; feeding-level accuracy must be assessed directly, and the authors plan a systematic uncertainty evaluation next.
For experiments with 10^7–10^8 counts, supervised ML gains roughly two orders of magnitude in feeding accuracy over the low-statistics regime, while response-matrix methods stall.
Supervised ML requires a representative training set and prior knowledge of the solution; without those, the response-matrix methods remain the fallback.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the validation set is generated in-distribution (same response matrix, same ±10% prior, same Poisson sampling as training), the reported order-of-magnitude advantage is best read as an interpolation result; a validation set built with an independently perturbed response matrix would test whether the advantage survives response-model mismatch, which is the likely real-world failure mode.
The same optimization-target insight should generalize to other ill-posed spectroscopy inverses: any deconvolution benchmark that scores only spectral goodness-of-fit can reward methods that fit noise, so ML-trained-on-truth may appear superior partly because it is scored on its own loss.
The hybrid workflow suggests an iterative closed loop the paper does not explore: use the ML refinement as a new reference, generate a fresh training set around it, and retrain; convergence of this loop would be a testable extension.
The authors' choice to train with ±10% perturbations around a single nominal feeding vector means the models never see qualitatively different feeding patterns; a test with a different decay or a different nominal vector would reveal how much of the learned mapping is decay-specific.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The central claim — that supervised ML beats response-matrix methods by an order of magnitude on TAGS feeding reconstruction — is an in-distribution result, and the paper's own caveat that ML "requires good prior knowledge of the solution" basically concedes the point. The validation feedings come from the same ±10% perturbation around the same nominal vector used for training, and L2 is the training loss for the CNN and GBDT models, so the superior L2 numbers are partly by construction. The RM methods, meanwhile, get no prior or truth information.

What is actually new and worthwhile: this is the first systematic comparison of CNNs and GBDTs against three established RM methods on a realistic high-level-density TAGS decay. The simulation is carefully done — a GEANT4 response matrix for the Lucrecia detector, 2M training samples, 1000 validation samples, and all methods run on the same data. The observation that spectral χ²_BC does not track feeding accuracy is useful and cleanly illustrated in the L2–χ² scatter plot. The paper is also transparent: it defers uncertainty quantification, lists contaminants/background/pileup as future work, and explicitly notes the need for prior knowledge.

The soft spots are substantial. The reduced-statistics test rescales 10^4–10^7-count spectra up to 10^8, which nonphysically amplifies the Poisson fluctuations and muddies the χ² comparisons. The ML "uncertainty reduction" is measured as dispersion over the validation ensemble, not as a posterior uncertainty, so calling it an order-of-magnitude reduction is premature. Most importantly, the recommended hybrid workflow — RM estimate to seed the training distribution, then ML refinement — is never tested, even though it is the stated practical takeaway. On a smaller note, code and data are not released, so the simulation pipeline can't be independently inspected.

None of this negates the value of the benchmark. Within its own matched conditions, the comparison is internally consistent and the decoupling of spectral fit and feeding accuracy is a legitimate caution for the field. But the conclusions overreach: "clear superiority" should be scoped to in-distribution simulation performance, and the hybrid claim should be marked untested.

Worth sending to peer review — it's a serious piece of work with real simulation machinery behind it. But a referee should require a domain-shift test (perturbed response, wider feeding prior, background) and a proper uncertainty treatment before the general claim is accepted.

Referee Report

4 major / 5 minor

Summary. The paper presents a systematic comparison of supervised machine-learning estimators (CNN-MSE, CNN-MAE, GBDT) against conventional response-matrix unfolding methods (DIFEVO, BAYES, NNLS) for the TAGS deconvolution problem. The authors use a realistic GEANT4-based response matrix for a synthetic benchmark, 152Tb†/152Gd†, generate two million training samples by perturbing a reference feeding vector by ±10% and applying Poisson statistics, and evaluate on one thousand independently generated but identically distributed validation samples. They report that the supervised ML methods achieve substantially lower L2 error on feeding-vector reconstruction, reduce per-feeding uncertainties by about an order of magnitude, and behave differently from RM methods as counting statistics vary. The paper proposes a hybrid workflow in which an RM method provides an initial feeding estimate that is then refined by a supervised ML model. The central quantitative claim is stated in Sec. 4: on the L2 metric, all three ML methods outperform the RM methods by more than one order of magnitude.

Significance. If the claimed superiority held outside the specific simulation setup, the paper would be a useful step toward applying supervised ML to TAGS, a problem where conventional unfolding is known to be ill-conditioned. The paper's strengths include the use of a realistic detector response, a large training set, a validation set that is independent of training, and the simultaneous reporting of spectral-fit (χ2_BC/Nb) and feeding-reconstruction (L2) metrics. The authors also clearly separate the optimization targets of the two method families, which is an important conceptual point. However, the evidence presented is confined to an in-distribution evaluation: the validation data are generated by the same procedure as the training data, and the primary metric L2 is the training objective for CNN-MSE and GBDT. The paper itself acknowledges that supervised ML requires good prior knowledge of the solution and defers systematic uncertainty, background, contaminants, and pileup to future work. Therefore, the practical significance is currently conditional on the transfer of the in-simulation ranking to real TAGS data, which is not demonstrated. The proposed hybrid strategy is plausible but unt

major comments (4)

[Sec. 2, Eq. (3); Sec. 4] The headline 'more than one order of magnitude' L2 advantage is largely forced by the evaluation design. The validation set is generated with the same procedure as the training set (same response matrix R, same ±10% flat perturbation around the same f_ref, same Poisson sampling; Sec. 2, 'ensuring consistency between training and evaluation conditions'), and the L2 metric of Eq. (3) is exactly the training loss for CNN-MSE and GBDT. RM methods minimize spectral residuals, not L2, so their poorer L2 is expected. To make the central claim load-bearing, the paper needs a domain-shift test: evaluate on spectra produced with a perturbed response matrix (e.g., energy-scale or efficiency changes), with a different reference feeding vector or wider perturbation range, and with added background/contaminants, and report both L2 and χ2_BC/Nb. Without such a test, the conclusions in Sec. 4 should be
[Sec. 3.2] The low-statistics comparison mixes training-distribution mismatch with algorithmic capability. The supervised models were trained exclusively at N=10^8; for N=10^4–10^7 the validation spectra are rescaled to 10^8 counts before being fed to the ML models, while the RM methods are applied at the original statistics. This means the ML inputs at low statistics have noise levels far outside the training distribution, and no noise augmentation or retraining at each N is described. Consequently, the conclusion that 'only the supervised ML approaches are able to translate the reduction of statistical fluctuations into a more accurate reconstruction' is not a fair head-to-head comparison. Please clarify the exact preprocessing for RM methods and retrain or noise-train the ML models at each statistics level, or reinterpret the result as a sensitivity analysis rather than a comparative claim.
[Sec. 3.1; Sec. 4 (last paragraph)] The claim that ML reduces 'the uncertainty on individual feeding values by up to an order of magnitude' overstates what is quantified. The reported standard deviations are only the Monte Carlo dispersion over validation samples; systematic uncertainties from response-model errors, missing transitions, continuum-model assumptions, and background are not evaluated. The paper itself defers 'a systematic evaluation of the uncertainties' to future work. Please rephrase the uncertainty claim to refer to the in-simulation statistical dispersion, and add a caveat that the absolute uncertainties in a real TAGS analysis will include response-model systematics that are not present in this benchmark.
[Sec. 4] The central practical recommendation—use an RM estimate as the training prior for a supervised ML model—is not tested. If the initial RM estimate is inaccurate, the training distribution will be offset from the test distribution, potentially erasing the ML advantage. At minimum, the paper should include a proof-of-principle experiment: train a model on spectra generated around an RM estimate (rather than around f_ref) and evaluate on the true validation distribution. Without this, the hybrid strategy remains a conjecture.

minor comments (5)

[General] The phrase 'continium' in the Table 1 caption should be 'continuum'. Also, the author affiliation contains 'Fonseca-V argas' with an extra space.
[Sec. 2] Please specify the training/validation split more precisely: how many of the two million samples are used for training versus validation, and whether any early stopping or hyperparameter selection was performed on the validation set.
[Fig. 5 and Sec. 3.1] The axis label in Fig. 5 is 'χ2_BC–L2' but the text discusses χ2_BC/Nb; the figure should be labeled consistently with the metric actually shown.
[References] Several reference names are garbled, e.g., 'T y oshida' and 'V etterling'; please proofread the bibliography against the original sources.
[Sec. 2, Eq. (4)] The definition of the DIFEVO loss omits the factorial term from the Poisson log-likelihood; this is fine as a constant, but the text should state that the constant has been dropped.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the comparison is a transparent, held-out benchmark rather than a self-referential derivation.

full rationale

The paper reports an empirical benchmark between supervised ML estimators and response-matrix unfolding methods, not a derivation from first principles. Its central claim—that ML methods trained to map spectra directly onto feeding vectors achieve lower L2 on a held-out validation set than RM methods that optimize spectral discrepancy—is supported by a standard train/validation split: the 1000 validation pairs were not used in training. The fact that Eq. 3 is both the training loss for CNN-MSE and GBDT and the evaluation metric does not constitute circularity; the paper explicitly acknowledges this by design, stating that supervised ML methods 'by construction, minimize the error on the feeding vector directly, and it is precisely this difference in optimization target that drives their superior performance.' This is a transparent explanation of mechanism, not a hidden reduction. CNN-MAE, which is trained with L1 loss, also achieves low L2 values, showing the result is not a simple tautology of training on the same metric. The in-distribution nature of the validation data and the untested hybrid workflow are genuine external-validity limitations, and the authors explicitly defer such generality to future work ('Establishing the generality of the trends reported here across decays of different level densities and cascade complexities is left for future work'), but these are limitations, not circular steps. Self-citations to Refs. [1,12] are used to construct the benchmark decay scheme (statistical-model branching ratios), but no load-bearing conclusion is imported from them. No step in the analysis reduces to its own inputs by definition or by fitted-parameter renaming.

Assumptions & free parameters 8 free parameters · 5 assumptions · 1 invented entities

The central comparison depends on several hand-chosen simulation parameters (perturbation width, continuum onset and binning, ad hoc feedings) and on the unvalidated assumption that the training distribution matches real data. The ML advantage is tied to these choices; a wider perturbation interval or a different continuum model could change the ranking.

free parameters (8)

Feeding perturbation interval (±10% of nominal) = ±10%
Training and validation feedings are sampled uniformly within ±10% of the nominal ENSDF/continuum values; this defines the prior width and the difficulty of the problem. The claimed ML advantage depends on this narrow prior.
Continuum onset energy = 2.9 MeV
Continuum is assumed to start at Ex=2.9 MeV to account for missing levels; changes the level scheme and the response matrix.
Continuum bin width = 40 keV
Discretization of the continuum region; affects the conditioning of R and the granularity of the recovered feeding distribution.
Nominal continuum feeding where ENSDF has no info = 0.1%
Ad hoc assignment for continuum bins without ENSDF information; defines the reference feeding vector around which all samples are drawn.
Training sample size = 2,000,000
Computational choice; enough samples for the ML models to memorize the generative distribution, but no scaling study is shown.
Training/validation total statistics = 10^8 decays
Sets the noise level in the training set; the reduced-statistics study rescales spectra to this total, creating a mismatch.
CNN architecture hyperparameters
Filters 128/64/32, kernel 5, stride 2, FC 128/128/64; chosen without systematic optimization, performance may depend on these choices.
GBDT hyperparameters
200 trees, depth 5, learning rate 0.8, subsample 0.8; chosen without systematic optimization, only L2 loss used.

assumptions (5)

domain assumption The GEANT4 model of the Lucrecia detector (Physics List Option 4, Radioactive Decay module) accurately represents the response of the real detector.
All spectra in the paper are generated from this response matrix; if the simulation is wrong, every comparison is biased. The model is benchmarked on calibration measurements only by reference, not shown in this paper.
domain assumption The statistical-model prescription of Ref. [12] correctly describes the de-excitation branching ratios of the continuum levels.
Used to construct the synthetic 152Tb† level scheme; Ref. [12] is by the same group and is not independently verified here.
domain assumption The ENSDF decay scheme for 152Tb, plus the added continuum, is a realistic and representative case for high-level-density TAGS decays.
The paper explicitly generalizes from this single benchmark; no other decay is tested.
ad hoc to paper Validation samples drawn from the same generative procedure as training samples (same R, same ±10% perturbations, same Poisson noise) are representative of the conditions where the trained model will be used.
This is the load-bearing premise behind the claim that ML outperforms RM; it is stated in Sec. 2 without argument that real data would follow the same distribution.
standard math Poisson statistics adequately describe the counting fluctuations in a TAGS measurement.
Used in Eq. 7 and in the chi2 definitions; standard assumption.

invented entities (1)

152Tb† / 152Gd† synthetic decay benchmark
purpose: Provides a high-level-density test case with a continuum of weakly fed states above 2.9 MeV, including ad hoc 0.1% feedings, used to compare all methods.
The continuum and its feedings are constructed by the authors, not measured; the benchmark is not validated against experimental data.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Rethinking Total Absorption Gamma Spectroscopy Deconvolution: Supervised Machine Learning vs Response-Matrix Methods." pith.science (2026). https://pith.science/paper/3YY5JUBW

@misc{pith2026260800090,
  author       = {Pith},
  title        = {Pith review of: Rethinking Total Absorption Gamma Spectroscopy Deconvolution: Supervised Machine Learning vs Response-Matrix Methods},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3YY5JUBW}},
  note         = {Machine review of arXiv:2608.00090}
}

abstract

The extraction of $\beta$-feeding distributions in Total Absorption $\gamma$-ray Spectroscopy constitutes a challenging inverse problem, particularly in nuclei with complex decay schemes involving a large number of excited states. In such cases, the measured spectrum arises from the superposition of many detector response functions, making the determination of the individual feedings intrinsically ill-posed and highly sensitive to the methodology employed. In this work, we present a systematic comparison between supervised Machine-Learning techniques and Response-Matrix methods using realistic Monte Carlo simulations of an experimental Total Absorption Spectrometer. Supervised Machine-Learning approaches construct a non-parametric estimator that infers level feedings from the measured spectrum after a training stage, whereas Response-Matrix methods determine the feeding distribution by directly minimizing the difference between measured and reconstructed spectra. Our results show that supervised Machine-Learning techniques achieve superior accuracy in the reconstruction of individual feeding intensities, whereas Response-Matrix methods provide robust and physically consistent initial solutions. These findings support a hybrid strategy in which a Response-Matrix method is first used to obtain an initial feeding estimate, which is then refined using a supervised Machine-Learning approach to achieve improved overall accuracy.

Figures

Figures reproduced from arXiv: 2608.00090 by the authors.

**Figure 1.** Schematic representation of the supervised ML estimator workflow. The top part illustrates the training phase, where the model learns from labeled data, while the bottom part depicts the inference phase, in which the trained model is applied to new input spectra to estimate the corresponding feeding distribution. output layer. The final output layer is a fully connected layer whose dimensionality is equal to the num… view at source ↗

**Figure 2.** Schematics of the CNN architecture employed in this work for 152Tb† . On the left the output dimension of each layer is indicated. measures the discrepancy between the observed spectrum, d⃗, and the reconstructed TAS spectrum, d⃗ est from the estimated feeding vector, f⃗ est (Eq. 1). In general, the algorithms are iterative starting from an initial guess or a priori solution. Typically, the loss function incorpora… view at source ↗

**Figure 3.** Top: Input TAS spectrum (black) for 152Tb† decay, obtained from the nominal feeding distribution, and reconstructed spectra (color) obtained from the estimated feedings for all methodologies considered in this work. Bottom: Calculated residuals between input and reconstructed TAS spectrum expressed in units of the statistical uncertainty. See the text for a detailed discussion. ergies above 3.2 MeV, although the osc… view at source ↗

Figures from the paper (3 more)

**Figure 5.** Figure 5: Two-dimensional χ 2 BC-L2 scatter plot for 152Tb† decay studied in this work. Each point corresponds to an individual TAS reconstruction from the test dataset, allowing a simultaneous evaluation of spectral reproduction and feeding recovery. See text for further det…

**Figure 4.** Figure 4: Average difference between the estimated and true feedings for all individual levels 152Gd as a function of the excitation energy for all methodologies. Top: Results for RM methodologies. Bottom: Results for supervised ML methodologies. See the text for more details. …

**Figure 6.** Figure 6: Effect of input spectrum statistics on the 152Tb† reconstruction performance. The horizontal axis shows the average value of ⟨χ 2 BC/Nb⟩ , while the vertical axis displays the corresponding average ⟨L2⟩. The error bars represent the standard deviation of each quantity…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 7 canonical work pages

[1]

J. Tain, D. Cano-Ott, Algorithms for the analysis of β-decay total absorption spec- tra, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equip- ment 571 (3) (2007) 728–738. doi:https: //doi.org/10.1016/j.nima.2006.10.098. URL https://www.sciencedirect.com/ science/article/pii/S0168900206018985

work page doi:10.1016/j.nima.2006.10.098 2007
[2]

W. H. Press, S. A. Teukolsky, W. T. V etterling, B. P . Flannery, Numerical Recipes in FORTRAN: The Art of Scientiﬁc Computing, 2nd Edition, Cambridge University Press, Cambridge, 1992

1992
[3]

A. P . Dempster, N. M. Laird, D. B. Rubin, Maximum likelihood from incomplete data via the em algorithm , Journal of the Royal Statistical Society: Series B (Methodolog- ical) 39 (1) (1977) 1–22. arXiv:https: //rss.onlinelibrary.wiley.com/doi/pdf/ 10.1111/j.2517-6161.1977.tb01600.x, doi:https://doi.org/10.1111/j. 2517-6161.1977.tb01600.x. URL https://rss....

arXiv 1977
[4]

L. A. Shepp, Y . V ardi, Maximum likelihood recon- struction for emission tomography, IEEE Transac- tions on Medical Imaging 1 (2) (1982) 113–122. doi:10.1109/TMI.1982.4307558

arXiv 1982
[5]

W. H. Richardson, Bayesian-based it- erative method of image restoration ∗, J. Opt. Soc. Am. 62 (1) (1972) 55–59. doi:10.1364/JOSA.62.000055. URL https://opg.optica.org/abstract. cfm?URI=josa-62-1-55

doi:10.1364/josa.62.000055 1972
[6]

L. B. Lucy, An iterative technique for the rectiﬁca- tion of observed distributions, Astrophysical Jour- nal 79 (1974) 745. doi:10.1086/111605

doi:10.1086/111605 1974
[7]

G. D’Agostini, A multidimensional unfolding method based on bayes’ theorem , Nuclear In- struments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detec- tors and Associated Equipment 362 (2) (1995) 487–498. doi:https://doi.org/10.1016/ 0168-9002(95)00274-X . URL https://www.sciencedirect.com/ science/article/pii/016890029500274X

arXiv 1995
[8]

Shore, R

J. Shore, R. Johnson, Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross-entropy, IEEE Transactions on Information Theory 26 (1) (1980) 26–37. doi: 10.1109/TIT.1980.1056144

arXiv 1980

Show all 39 references

[9]

Tikochinsky, N

Y . Tikochinsky, N. Z. Tishby, R. D. Levine, Con- sistent inference of probabilities for reproducible experiments, Phys. Rev. Lett. 52 (1984) 1357–

1984
[10]

D. M. Titterington, General structure of regular- ization procedures in image reconstruction, As- tronomy and Astrophysics 144 (1985) 381

1985
[11]

L. Meng, D. Ramsden, An inter-comparison of three spectral-deconvolution algorithms for gamma-ray spectroscopy, IEEE Transactions on Nuclear Science 47 (4) (2000) 1329–1336. doi: 10.1109/23.872973

2000 doi
[12]

J. Tain, D. Cano-Ott, The inﬂuence of the unknown de-excitation pattern in the analy- sis of β-decay total absorption spectra , Nu- clear Instruments and Methods in Physics 12 Research Section A: Accelerators, Spec- trometers, Detectors and Associated Equip- ment 571 (3) (2007...

2007 doi
[13]

C. Duke, P . Hansen, O. Nielsen, G. Rudstam, Strength-function phenomena in electron-capture beta decay , Nuclear Physics A 151 (3) (1970) 609–633. doi:https://doi.org/10.1016/ 0375-9474(70)90400-8 . URL https://www.sciencedirect.com/ science/article/pii/0375947470904008

1970
[14]

Rubio, W

B. Rubio, W. Gelletly, A. Algora, E. Nacher, J. L. Tain, Beta decay studies with total absorp- tion spectroscopy and the lucrecia spectrometer at isolde* , Journal of Physics G: Nuclear and Particle Physics 44 (8) (2017) 084004. doi:10.1088/1361-6471/aa797f. URL https://doi.or...

2017 doi
[15]

Hardy, L

J. Hardy, L. Carraz, B. Jonson, P . Hansen, The es- sential decay of pandemonium: A demonstration of errors in complex beta-decay schemes , Physics Letters B 71 (2) (1977) 307–310. doi:https:// doi.org/10.1016/0370-2693(77)90223-4 . URL https://www.sciencedirect.com/ science/a...

1977
[16]

Dembski, M

C. Dembski, M. Kuchera, S. Liddick, R. Ramanu- jan, A. Spyrou, Two-dimensional total absorption spectroscopy with conditional generative ad- versarial networks , Nuclear Instruments and Methods in Physics Research Section A: Accel- erators, Spectrometers, Detectors and Associa...

2024
[17]

Storn, K

R. Storn, K. Price, Diﬀerential evolution – a simple and e ﬃcient heuristic for global opti- mization over continuous spaces , Journal of Global Optimization 11 (4) (1997) 341–359. doi:10.1023/A:1008202821328. URL https://doi.org/10.1023/A: 1008202821328

1997 doi
[18]

C. L. Lawson, R. J. Hanson, Solving Least Squares Problems , Society for In- dustrial and Applied Mathematics, 1995. arXiv:https://epubs.siam.org/ doi/pdf/10.1137/1.9781611971217, doi:10.1137/1.9781611971217. URL https://epubs.siam.org/doi/abs/ 10.1137/1.9781611971217

1995 doi
[19]

C. M. Bishop, H. Bishop, Deep Learn- ing, 1st Edition, Springer, Cham, 2024. doi:10.1007/978-3-031-45468-4 . URL https://doi.org/10.1007/ 978-3-031-45468-4

2024 doi
[20]

Coadou, Boosted Decision Trees , WORLD SCIENTIFIC, 2022, p

Y . Coadou, Boosted Decision Trees , WORLD SCIENTIFIC, 2022, p. 958. doi:10.1142/9789811234033_0002. URL http://dx.doi.org/10.1142/ 9789811234033_0002

2022 doi
[21]

Algora, J

A. Algora, J. L. Taín, A. B. Perez, B. Ru- bio, J. Agramunt, L. Caballero, E. Nácher, A. Krasznahorkay, M. D. Hunyadi, J. Gulyás, A. Vitéz, J. Äystö, H. Penttilä, S. Rinta-Antila, I. Moore, T. Eronen, A. Jokinen, A. Niemi- nen, J. Hakala, P . Karvonen, A. Kankainen, U. Hager, ...

2007
[22]

C. M. Bishop, Pattern Recognition and Machine Learning (Information Science and Statistics), 1st Edition, Springer, 2007

2007
[23]

K. P . Murphy, Machine learning : a probabilistic perspective, MIT Press, Cambridge, Mass. [u.a.], 2013

2013
[24]

Abadi, A

M. Abadi, A. Agarwal, P . Barham, E. Brevdo, Z. Chen, C. Citro, G. S. Corrado, A. Davis, J. Dean, M. Devin, S. Ghemawat, I. Good- fellow, A. Harp, G. Irving, M. Isard, Y . Jia, R. Jozefowicz, L. Kaiser, M. Kudlur, J. Leven- berg, D. Mané, R. Monga, S. Moore, D. Mur- ray, C. Ol...

2015
[25]

T. Chen, T. He, M. Benesty, V . Khotilovich, Y . Tang, H. Cho, K. Chen, R. Mitchell, I. Cano, T. Zhou, M. Li, J. Xie, M. Lin, Y . Geng, Y . Li, J. Y uan, D. Cortes, xgboost: Extreme Gradient Boosting, r package version 3.2.0.0 (2025). URL https://github.com/dmlc/xgboost

2025
[26]

Rokach, O

L. Rokach, O. Maimon, Decision Trees , Springer US, Boston, MA, 2005, pp. 165–

2005
[27]

Breiman, Random forests , Machine Learn- ing 45 (1) (2001) 5–32

L. Breiman, Random forests , Machine Learn- ing 45 (1) (2001) 5–32. doi:10.1023/A: 1010933404324. URL https://doi.org/10.1023/A: 1010933404324

2001 doi
[28]

A. N. Tikhonov, Solution of incorrectly formu- lated problems and the regularization method, Doklady Akademii Nauk SSSR 4 (1963) 1035– 1038

1963
[29]

Fortin, F.-M

F.-A. Fortin, F.-M. De Rainville, M.-A. Gardner, M. Parizeau, C. Gagné, DEAP: Evolutionary al- gorithms made easy, Journal of Machine Learning Research 13 (2012) 2171–2175

2012
[30]

Bourbeau, Z

J. Bourbeau, Z. Hampel-Arias, Pyunfold: A python package for iterative unfolding , The Journal of Open Source Software 3 (26) (2018)

2018
[31]

Virtanen, R

P . Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P . Peterson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, ˙I. Polat, Y . F...

2020
[32]

Bayes, An essay toward solving a problem in the doctrine of chances, Philosophical Transac- tions of the Royal Society of London (1764)

T. Bayes, An essay toward solving a problem in the doctrine of chances, Philosophical Transac- tions of the Royal Society of London (1764)
[33]

Allison, K

J. Allison, K. Amako, J. Apostolakis, P . Arce, M. Asai, T. Aso, E. Bagli, A. Bagulya, S. Baner- jee, G. Barrand, B. Beck, A. Bogdanov, D. Brandt, J. Brown, H. Burkhardt, P . Canal, D. Cano-Ott, S. Chauvie, K. Cho, G. Cirrone, G. Cooperman, M. Cortés-Giraldo, G. Cosmo, G. Cutt...

2016
[34]

R. Brun, F. Rademakers, ROOT: An object ori- ented data analysis framework, Nucl. Instrum. Meth. A 389 (1997) 81–86. doi:10.1016/ S0168-9002(97)00048-X

1997
[35]

Baker, R

S. Baker, R. Cousins, Clariﬁcation of the use of chi-square and likelihood functions in ﬁts to histograms , Nuclear Instruments and Methods in Physics Research 221 (2) (1984) 437–442. doi:https://doi.org/10.1016/ 0167-5087(84)90016-4 . URL https://www.sciencedirect.com/ scienc...

1984
[36]

M. J. Martin, Nuclear data sheets for a = 152, Nu- clear Data Sheets 114 (2013) 1497–1847. 14

2013
[192]

URL https://doi.org/10.1007/ 0-387-25465-X_9

doi:10.1007/0-387-25465-X_9 . URL https://doi.org/10.1007/ 0-387-25465-X_9

doi
[741]

URL https://doi.org/10.21105/joss

doi:10.21105/joss.00741. URL https://doi.org/10.21105/joss. 00741

doi
[1360]

URL https://link.aps.org/doi/10.1103/ PhysRevLett.52.1357

doi:10.1103/PhysRevLett.52.1357. URL https://link.aps.org/doi/10.1103/ PhysRevLett.52.1357

doi

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

J. Tain, D. Cano-Ott, Algorithms for the analysis of β-decay total absorption spec- tra, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equip- ment 571 (3) (2007) 728–738. doi:https: //doi.org/10.1016/j.nima.2006.10.098. URL https://www.sciencedirect.com/ science/article/pii/S0168900206018985

work page doi:10.1016/j.nima.2006.10.098 2007

[2] [2]

W. H. Press, S. A. Teukolsky, W. T. V etterling, B. P . Flannery, Numerical Recipes in FORTRAN: The Art of Scientiﬁc Computing, 2nd Edition, Cambridge University Press, Cambridge, 1992

1992

[3] [3]

A. P . Dempster, N. M. Laird, D. B. Rubin, Maximum likelihood from incomplete data via the em algorithm , Journal of the Royal Statistical Society: Series B (Methodolog- ical) 39 (1) (1977) 1–22. arXiv:https: //rss.onlinelibrary.wiley.com/doi/pdf/ 10.1111/j.2517-6161.1977.tb01600.x, doi:https://doi.org/10.1111/j. 2517-6161.1977.tb01600.x. URL https://rss....

arXiv 1977

[4] [4]

L. A. Shepp, Y . V ardi, Maximum likelihood recon- struction for emission tomography, IEEE Transac- tions on Medical Imaging 1 (2) (1982) 113–122. doi:10.1109/TMI.1982.4307558

arXiv 1982

[5] [5]

W. H. Richardson, Bayesian-based it- erative method of image restoration ∗, J. Opt. Soc. Am. 62 (1) (1972) 55–59. doi:10.1364/JOSA.62.000055. URL https://opg.optica.org/abstract. cfm?URI=josa-62-1-55

doi:10.1364/josa.62.000055 1972

[6] [6]

L. B. Lucy, An iterative technique for the rectiﬁca- tion of observed distributions, Astrophysical Jour- nal 79 (1974) 745. doi:10.1086/111605

doi:10.1086/111605 1974

[7] [7]

G. D’Agostini, A multidimensional unfolding method based on bayes’ theorem , Nuclear In- struments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detec- tors and Associated Equipment 362 (2) (1995) 487–498. doi:https://doi.org/10.1016/ 0168-9002(95)00274-X . URL https://www.sciencedirect.com/ science/article/pii/016890029500274X

arXiv 1995

[8] [8]

Shore, R

J. Shore, R. Johnson, Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross-entropy, IEEE Transactions on Information Theory 26 (1) (1980) 26–37. doi: 10.1109/TIT.1980.1056144

arXiv 1980