Pith. sign in

REVIEW 2 major objections 7 minor 47 references

Bayesian evidence adaptive pursuit to identify neutron sources with scatter-based spectrometers

T0 review · 2 major / 7 minor · reviewed 2026-08-01 · deepseek-v4-flash

Pith's one-line read BEAP: Bayesian evidence pruning identifies mixed neutron sources from recoil spectra, with decisive >4σ support and no exhaustive enumeration.

desk verdict Solid Bayesian source identification with MC templates, but the adaptive pruning that the paper advertises is never actually exercised in the validations. read the letter →

arxiv 2607.21543 v1 pith:N2DAMFHL submitted 2026-07-23 physics.ins-det physics.app-phphysics.data-an

classification physics.ins-detphysics.app-phphysics.data-an PACS 29.30.Hs29.40.Mc02.50.Tt
keywords Bayesianevidenceneutronsourceidentificationrecoilspectroscopymodelselectionadaptivepursuitscintillatordetectorsnuclearsafeguardsmixed-sourcefields
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 attempts to establish that a Bayesian model-selection search, BEAP, can identify neutron sources—including mixtures of two or three—from recoil spectra of scatter-based spectrometers without enumerating every possible source combination and without experimentally trained forward models. It replaces learned response models with physics-based templates generated by Monte Carlo transport simulation, and prunes the combinatorial space of candidate source mixtures using an evidence-based Occam window. The authors validate the method on controlled californium-252 and deuterium-deuterium generator measurements and on synthetic spectra. If correct, the approach gives quantitatively rigorous source identification for nuclear safeguards and emergency response with acquisition times that shrink as detector efficiency grows. A sympathetic reader would take the central claim to be that evidence-guided adaptive search makes Bayesian full-spectrum source identification practical for realistic source libraries.

What carries the argument

The load-bearing object is the Bayesian evidence Z_S = ∫ L(θ;y,M_S) π(θ|S) dθ, computed for each candidate subset S of the source library using a negative-binomial likelihood with a dispersion parameter and weakly informative priors. Carrying the argument is the adaptive Occam-window criterion (Eq. 5): after ranking mixtures by log-evidence at order k, a mixture is retained if its log-evidence does not drop faster than the average decay across the current window, and the next iteration's source indices are the union of indices in retained mixtures. This single criterion balances greediness (keeping only winners) against robustness (keeping nearly competitive alternatives), and it is what con

What would settle it

As a concrete check, build a three-source mixture whose weakest component has a singleton log-evidence far below the Occam-window cutoff but whose inclusion dominates all three-source models (e.g., a weak fusion component on top of two overlapping fission-like sources), run BEAP with the paper's B=15, D=3 settings, and compare against exhaustive evaluation of all 2^10−1 subsets; missing the weak source would show the pruning assumption fails.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that Bayesian evidence—the marginal likelihood of the measured recoil spectrum under a candidate source-mixture model—can serve both as the ranking statistic and as the pruning rule that makes source-ensemble search tractable. BEAP starts with all single sources, evaluates their evidence, retains a competitive window of mixtures, and builds higher-order mixtures only from source indices that survive. In experiments and simulations spanning fission, (α,n), and fusion sources, the true source mixture is the highest-evidence model with >4σ separation from alternatives, and the required number of detected recoil events ranges from about ten for distinct

Load-bearing premise

The load-bearing premise is that the true source mixture can always be reached from the retained low-order mixtures: a source with poor singleton evidence but a decisive role in a higher-order mixture could be pruned at the first step and never re-enter the search.

Editorial extensions

If this is right

  • Because templates come from Monte Carlo transport rather than dedicated measurement campaigns, BEAP can be re-targeted to new detectors and source classes without new training data.
  • For a ten-source library with up to three sources, BEAP evaluates 175 candidate mixtures instead of 1023, so larger libraries become feasible without exhaustive enumeration.
  • The framework can flag weak secondary components: it recovered a ~14.1 MeV deuterium-tritium contamination at 0.25 of total emission in the DD-generator data.
  • Event-count thresholds from synthetic studies translate, for the 21.6 cm³ validation spectrometer, to acquisition times of roughly 2×10² s, 2×10³ s, and 2×10⁴ s for the hardest single-, two-, and three-source cases; larger detectors shrink these to minutes.
  • All inference is statistically calibrated by Bayes factors, so outputs are interpretable as decisive, strong, or weak support rather than qualitative spectral metrics.

Reading between the lines

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

  • The Occam-window pruning assumes that a source weak at first order will re-enter through a higher-order mixture, but this is not proven; a natural stress test is to compare BEAP against exhaustive search on small libraries with deliberately constructed adversarial spectra.
  • The same evidence-search machinery could be applied to gamma-ray spectroscopy or to detector arrays with position-dependent response, where the combinatorial source-ensemble problem has the same structure.
  • Because the templates encode a known source–detector geometry, the method could be extended to jointly infer source position or intervening shielding by treating them as additional model parameters or by expanding the template library, a direction the paper flags but does not implement.
  • The reported event-count thresholds are detector-agnostic, so a reader could use them to predict acquisition time for any scatter-based spectrometer of known efficiency, not just the validation instrument.
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.

Referee Report

2 major / 7 minor

Summary. The paper introduces Bayesian Evidence Adaptive Pursuit (BEAP), a model-selection algorithm for identifying neutron source ensembles from scatter-based recoil-spectroscopy measurements. The forward model is a linear superposition of Monte Carlo–generated source templates, and model comparison is performed via nested-sampling Bayesian evidence. BEAP is claimed to reduce the combinatorial search space by iteratively ranking, retaining, and pruning source subsets according to an adaptive Occam-window criterion. The authors validate the framework with three laboratory experiments (single Cf-252, single DD, and mixed Cf-252+DD), supported by synthetic spectra spanning single-, two-, and three-source mixtures with varying event counts and emission-rate imbalances. The central claims are that BEAP recovers the true source mixture with decisive statistical evidence and that its adaptive pruning enables scalable inference without exhaustive enumeration.

Significance. If the pruning claim were validated, the paper would be a valuable practical contribution to neutron source identification, since existing methods are largely qualitative and exhaustive model comparison scales exponentially. The strengths include the use of physics-based templates rather than experimentally trained surrogate models, a full-spectrum Bayesian evidence treatment, and a systematic synthetic study covering a wide event-count range and emission-rate ratios. The posterior predictive checks in Figs. 3 and 4 are informative. However, the core novelty—adaptive Occam-window pruning—is never actually exercised in any validation: with N=6, D=3, and B=15, the search is exhaustive for all subsets up to size 3. The scalability claim therefore rests on an unproven completeness property. The paper also contains an internal inconsistency in Fig. 3 regarding both the display ordering and the reported best-vs-runner-up Bayes factors. These issues are fixable, and the underlying inference pipeline appears sound, so the work is of interest to the nuclear nonproliferation and instrumentation community once the pruning component is either demonstrated on a larger library or explicitly de-e

major comments (2)
  1. [Section 2.2, Algorithm 1, Eq. (5)] The central claim that BEAP identifies source ensembles “without exhaustive enumeration” is not supported by the validations. With N=6 and D=3, the algorithm evaluates exactly all nonempty subsets up to size 3: C(6,1)=6 singletons, C(6,2)=15 pairs, and C(6,3)=20 triples. Because B=15 and there are exactly 15 pairs, the retention step in Eq. (5) cannot remove any pair, and the candidate source set for k=3 is the full library. Thus the experimental and synthetic results are identical to exhaustive search. The only quantitative pruning example (N=10, reducing 1023 to 175 candidates) is hypothetical and not validated. Moreover, no proof or worst-case bound is provided that a source pruned at order k could not be needed in the true optimal mixture at order k+1. Please add a validation with N>6 where pruning is active, provide a completeness guarantee for the Occam-window step, or explicitly r
  2. [Section 3.1, Fig. 3] The caption states that models are ordered by decreasing evidence, and the text claims the preferred model was favored over the runner-up with logB>9.0 in all three experiments. In panel (a), however, the diagonal logZ values increase from -357.8(3) (model 1) to -346.0(2) (model 7), i.e., the ordering is reversed. For the Cf-252 single-source experiment, the best and second-best displayed models differ by only about 7.1 log units, not the claimed >9.0. This discrepancy undermines the headline “>4σ” experimental-support claim for that configuration. Please correct the display/ordering convention and report the actual best-vs-runner-up Bayes factors for each panel.
minor comments (7)
  1. [Section 4] The paragraph beginning “These performance estimates highlight…” is duplicated nearly verbatim. Please remove one copy.
  2. [Fig. 3 caption] The text refers to “anti-diagonal entries” for the logZ values; these are the main-diagonal entries. Please correct the terminology.
  3. [Section 2.3] The text says the experiments correspond to “the complete power set” of the source mixture; since the empty set is not measured, this should be “all nonempty subsets”.
  4. [Figs. 4 and 6 captions] The shaded uncertainty is described as “3-sigma”, but the plotted quantity p(M_true|y) is a probability confined to [0,1]. It would be clearer to show posterior intervals or to define how the sigma is computed for this bounded quantity.
  5. [Section 2.5] It is unclear how “scaled by the number of detected neutron events rather than by live time” is reconciled with Eq. (1), where t appears explicitly. Please define the normalization used for the synthetic spectra.
  6. [Algorithm 1, line 16] The code should explicitly handle the case l=0 (for example if D>N). As written, m=min(B,0)=0 and the loop is skipped, leaving Lambda_k empty. A guard would make the pseudocode robust.
  7. [Section 2.1] The likelihood and prior definitions are deferred to Ref. [2]. For a self-contained article, the negative-binomial likelihood and the exact prior hyperparameters should be stated in an appendix or in the main text.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: BEAP's derivation is a self-contained Bayesian model-selection scheme with external experimental and Monte Carlo anchors; the only notable issue is an unproven pruning-completeness claim, which is a scalability gap rather than a circular reduction.

full rationale

The paper's derivation chain is not circular in any of the enumerated senses. The forward model in Eq. (1) is a linear superposition of independently generated Monte Carlo recoil templates; the Bayesian evidence in Eq. (2) is a standard integral over likelihood and priors; and the Occam-window criteria in Eqs. (3)-(5) are algebraic rearrangements of a retention rule, not a restatement of the conclusion. The experimental validations use real Cf-252 and DD measurements, giving an external anchor independent of the identification output. The Monte Carlo templates are generated from MCNPX transport simulations and are not fitted to the experimental spectra used for validation. No fitted parameter is renamed as a prediction: the reported Bayes factors, posterior model probabilities, and inferred emission rates are computed by nested sampling from the stated likelihood and priors. The self-citations to Ref. [2] supply likelihood/prior parameterization, detector calibration, and previous inference protocols, but they are not load-bearing uniqueness theorems or ansatzes and do not by construction force the posterior to the true mixture. The most serious concern in the paper is an omitted proof, not circularity: the pruning step is never exercised in the reported N=6 validations because with D=3 and B=15 all subsets through order 3 are evaluated exhaustively (6 singletons, 15 pairs, 20 triples), and the paper's only quantitative pruning example, 'reduces the candidate space from 1023 to 175,' is a hypothetical count with no completeness guarantee for the Occam-window retention. That gap affects the scalability claim, but it does not make the reported identification results equivalent to their inputs.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard Bayesian machinery plus several modeling choices: negative-binomial likelihood, MC template accuracy, fixed geometry, and an unproven pruning completeness assumption. No new physical entities are introduced. The prior hyperparameters and BEAP thresholds are hand-set but weakly informative.

free parameters (5)
  • Emission-rate prior mean/std = 1e8 s^-1
    Truncated normal prior on each source rate; chosen by hand (Sec. 2.1).
  • Dispersion prior mean/std = 1
    Truncated normal prior on alpha_NB; chosen by hand (Sec. 2.1).
  • BEAP source-order threshold D = 3
    User-set max number of sources; if the true mixture has more than 3 sources, BEAP cannot find it (Sec. 2.2).
  • BEAP retention threshold B = 15
    User-set minimum retained mixtures per iteration; controls pruning aggressiveness (Sec. 2.2).
  • Fusion source Doppler broadening = 2%
    Modeling choice for DD/DT angular spectra (Sec. 2.4).
assumptions (6)
  • domain assumption Negative binomial likelihood with dispersion parameter alpha_NB describes recoil count overdispersion.
    Adopted from [2]; no independent validation in this paper.
  • domain assumption MCNPX-PoliMi mass model of detector, room, and generator reproduces the measured response.
    Backed by prior benchmark [42], but template error is not propagated into evidence.
  • domain assumption Source-detector geometry is known and fixed; templates are generated at that geometry.
    Explicitly acknowledged in Conclusion; unknown position/shielding introduces degeneracies.
  • standard math Equal model priors in Bayesian model comparison.
    Standard noncommittal choice (Sec. 2.1).
  • domain assumption dynesty nested sampling converges to accurate logZ within reported uncertainties.
    No convergence diagnostics reported beyond least-significant-figure uncertainties.
  • ad hoc to paper BEAP Occam-window pruning preserves any source index needed by the true higher-order mixture.
    Algorithm 1 Eq. (5); no proof or counterexample; tested cases only.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bayesian evidence adaptive pursuit to identify neutron sources with scatter-based spectrometers." pith.science (2026). https://pith.science/paper/N2DAMFHL

@misc{pith2026260721543,
  author       = {Pith},
  title        = {Pith review of: Bayesian evidence adaptive pursuit to identify neutron sources with scatter-based spectrometers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N2DAMFHL}},
  note         = {Machine review of arXiv:2607.21543}
}
abstract

Reliable neutron source identification is essential for nuclear nonproliferation, safeguards, and homeland security, but remains challenging because neutron spectral inversion is often ill-conditioned, especially for mixed-source fields with overlapping spectral signatures. Here, we present a scalable Bayesian framework for neutron source identification from recoil spectroscopy measurements using evidence-based model selection. The method introduces a Bayesian Evidence Adaptive Pursuit (BEAP) algorithm that efficiently searches the combinatorial space of candidate source ensembles by iteratively ranking, retaining, and pruning source mixtures according to their Bayesian evidence. We validate the framework experimentally with controlled Cf-252 and deuterium--deuterium neutron-generator measurements, complemented by high-fidelity Monte Carlo simulations spanning representative fission, $(\alpha,\text{n})$, and fusion sources with varying emission rates and mixture complexities. BEAP correctly identifies single- and multi-source mixtures with decisive statistical support ($>\!4\sigma$), requiring between $\mathcal{O}(10^1)$ and $\mathcal{O}(10^6)$ detected recoil events depending on source-mixture complexity, spectral similarity, and emission-rate imbalance. These findings establish BEAP as a practical, scalable, and robust tool for quantitative source identification in mixed neutron fields, significantly extending the operational capabilities of scatter-based neutron spectrometers in nuclear security and emergency response applications.

Figures

Figures reproduced from arXiv: 2607.21543 by the authors.

Figure 1
Figure 1. Experimental setup used for the validation measurements showing the Cf-252 spontaneous fission source and the DD neutron generator together with their respective photon shielding, and the organic-glass scintillator (OGS)-based neutron spectrometer. Spectrometer DD generator Cf-252 source Power supply Digitizer Pb shield Sn-Cu shield [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Monte Carlo mass model of the experimental setup used for spectral template generation. The model includes the benchmarked organic-glass scintillator spectrometer together with the dominant surrounding experimental components and laboratory structures relevant for neutron transport and scattering. Pu-240), (𝛼, n) neutron sources (PuBe & AmBe), as well as fusion-based deuterium-deuterium (DD) and deuterium￾tritium (D… view at source ↗
Figure 3
Figure 3. Bayesian inference results for the neutron recoil spectroscopy experiments performed using single- and mixed￾source configurations. (a–b) Single-source Cf-252 experiment. (c–d) Single-source DD experiment. (e–f) Two-source Cf-252 & DD experiment. Panels (a), (c), and (e) summarize the Bayesian model comparison results for the seven highest-ranking source mixture models evaluated by the BEAP algorithm. Shown are the … view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Bayesian inference results for synthetic neutron recoil spectroscopy datasets for varying source mixtures and neutron events: (a–b) Single-source mixtures {Cf-252}, {Pu-240}, {PuBe}, {AmBe}, {DD}, and {DT}; (c–d) Two-source mixtures {Cf-252, Pu-240}, {PuBe, AmBe}, and …
Figure 5
Figure 5. Figure 5: Posterior distribution of the {Cf-252,Pu-240,DT} source set for 105 neutron events. The true neutron emission rates (𝜉Cf-252,𝜉Pu-240,𝜉DT) and the true dispersion parameter (𝛼NB) are indicated by the red dashed lines. For the two￾dimensional marginal posteriors, we indi…
Figure 6
Figure 6. Figure 6: Bayesian inference results for synthetic neutron recoil spectroscopy datasets for varying neutron emission rate ratios 𝜉𝑖∕𝜉𝑗 for to-source mixtures {𝑖, 𝑗} and varying neutron events: (a–b) Two-source mixture {Cf-252, Pu-240}; (c–d) Two-source mixture {PuBe, AmBe}; (e–f…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

47 extracted references · 15 canonical work pages

  1. [1]

    Al Hamrashdi, S

    H. Al Hamrashdi, S. D. Monk, D. Cheneler, Passive Gamma-Ray and Neutron Imaging Systems for National Security and Nuclear Non- Proliferation in Controlled and Uncontrolled Detection Areas: Review of Past and Current Status, Sensors 19 (2019) 2638. doi:10.3390/ s19112638

  2. [2]

    Breitenmoser, R

    D. Breitenmoser, R. Lopez, S. D. Clarke, S. A. Pozzi, Identifying neutron sources using recoil and time-of-flight spectroscopy, Phys. Rev. Appl. 25 (2026) 064013. doi:10.1103/v6j6-f4rx.arXiv:2512.10044

  3. [3]

    Wallenius, K

    M. Wallenius, K. Lützenkirchen, K. Mayer, I. Ray, L. A. de las Heras, M. Betti, O. Cromboom, M. Hild, B. Lynch, A. Nicholl, H. Ottmar, G. Rasmussen, A. Schubert, G. Tamborini, H. Thiele, W. Wagner, C. Walker, E. Zuleger, Nuclear forensic investigations with a focus on plutonium, Journal of Alloys and Compounds 444–445 (2007) 57–62. doi:10.1016/j.jallcom.2...

  4. [4]

    K.Mayer,M.Wallenius,Z.Varga, NuclearForensicScience:CorrelatingMeasurableMaterialParameterstotheHistoryofNuclearMaterial, Chem. Rev. 113 (2013) 884–900. doi:10.1021/cr300273f

  5. [5]

    H. Meng, Z. Jianyu, W. Jun, L. Rui, A Passive Method for the Detection of Explosives and Weapons-Grade Plutonium in Nuclear Warheads, Sci. Glob. Secur. 26 (2018) 57–69. doi:10.1080/08929882.2018.1517431

  6. [6]

    E.Lepowsky,J.Jeon,A.Glaser, Confirmingtheabsenceofnuclearwarheadsviapassivegamma-raymeasurements, Nucl.Instrum.Methods Phys. Res. A 990 (2021) 164983. doi:10.1016/j.nima.2020.164983

  7. [7]

    S. P. LaMont, S. E. Glover, R. H. Filby, Determination of plutonium-240/239 ratios in low activity samples using high resolution alpha- spectrometry, J Radioanal Nucl Chem 234 (1998) 195–199. doi:10.1007/BF02389771

  8. [8]

    Trotta, Bayes in the sky: Bayesian inference and model selection in cosmology, Contemp

    R. Trotta, Bayes in the sky: Bayesian inference and model selection in cosmology, Contemp. Phys. 49 (2008) 71–104. doi:10.1080/ 00107510802066753

Show all 47 references
  1. [9]

    von Toussaint, Bayesian inference in physics, Rev

    U. von Toussaint, Bayesian inference in physics, Rev. Mod. Phys. 83 (2011) 943–999. doi:10.1103/RevModPhys.83.943

  2. [10]

    Veitch, Parameter estimation for compact binaries with ground-based gravitational-wave observations using the LALInference software library, Phys

    J. Veitch, Parameter estimation for compact binaries with ground-based gravitational-wave observations using the LALInference software library, Phys. Rev. D 91 (2015). doi:10.1103/PhysRevD.91.042003

  3. [11]

    Ashton, M

    G. Ashton, M. Hübner, P. D. Lasky, C. Talbot, K. Ackley, S. Biscoveanu, Q. Chu, A. Divakarla, P. J. Easter, B. Goncharov, F. H. Vivanco, J.Harms,M.E.Lower,G.D.Meadors,D.Melchor,E.Payne,M.D.Pitkin,J.Powell,N.Sarin,R.J.E.Smith,E.Thrane,Bilby:AUser-friendly Bayesian Inference Lib...

  4. [12]

    R. J. E. Smith, G. Ashton, A. Vajpeyi, C. Talbot, Massively parallel Bayesian inference for transient gravitational-wave astronomy, MNRAS 498 (2020) 4492–4502. doi:10.1093/mnras/staa2483

  5. [13]

    R. J. MacDonald, N. Madhusudhan, HD 209458b in new light: Evidence of nitrogen chemistry, patchy clouds and sub-solar water, MNRAS 469 (2017) 1979–1996. doi:10.1093/mnras/stx804

  6. [14]

    Pinhas, B

    A. Pinhas, B. V. Rackham, N. Madhusudhan, D. Apai, Retrieval of planetary and stellar properties in transmission spectroscopy with Aura, MNRAS 480 (2018) 5314–5331. doi:10.1093/mnras/sty2209

  7. [15]

    Hünnefeld, R

    M. Hünnefeld, R. Abbasi, M. Ackermann, J. Adams, J. A. Aguilar, M. Ahlers, M. Ahrens, C. Alispach, A. A. Alves, N. M. Amin, R. An, K. Andeen, T. Anderson, G. Anton, C. Argüelles, Y. Ashida, S. Axani, X. Bai, A. V. Balagopal, A. Barbano, S. W. Barwick, B. Bastian, V.Basu,S.Baur...

  8. [16]

    Salinas, V

    D. Salinas, V. Flunkert, J. Gasthaus, T. Januschowski, DeepAR: Probabilistic forecasting with autoregressive recurrent networks, Int. J. Forecast. 36 (2020) 1181–1191. doi:10.1016/J.IJFORECAST.2019.07.001.arXiv:1704.04110

  9. [17]

    J. O. Lloyd-Smith, Maximum Likelihood Estimation of the Negative Binomial Dispersion Parameter for Highly Overdispersed Data, with Applications to Infectious Diseases, PLOS ONE 2 (2007) e180. doi:10.1371/JOURNAL.PONE.0000180

  10. [18]

    Praszalowicz, Negative Binomial Distribution and the multiplicity moments at the LHC, Phys

    M. Praszalowicz, Negative Binomial Distribution and the multiplicity moments at the LHC, Phys. Lett. B 704 (2011) 566–569. doi:10.1016/j.physletb.2011.09.101

  11. [19]

    S. V. Tezlaf, Significance of the negative binomial distribution in multiplicity phenomena, Phys. Scr. 98 (2023) 115310. doi:10.1088/ 1402-4896/acfead

  12. [20]

    L. A. Perez, S. Malhotra, J. E. Rhoads, V. Tilvi, Void Probability Function of Simulated Surveys of High-redshift Ly𝛼Emitters, ApJ 906 (2021) 58. doi:10.3847/1538-4357/abc88b

  13. [21]

    J.N.Fry,S.Colombi, Voidstatisticsandhierarchicalscalinginthehalomodel, MNRAS433(2013)581–590.doi:10.1093/mnras/stt745

  14. [22]

    Hurtado-Gil, V

    L. Hurtado-Gil, V. J. Martínez, P. Arnalte-Mur, M.-J. Pons-Bordería, C. Pareja-Flores, S. Paredes, The best fit for the observed galaxy counts-in-cell distribution function, A&A 601 (2017) A40. doi:10.1051/0004-6361/201629097

  15. [23]

    Hameeda, A

    M. Hameeda, A. Plastino, M. C. Rocca, Generalized Poisson distributions for systems with two-particle interactions, IOPSciNotes 2 (2021) 015003. doi:10.1088/2633-1357/abec9f

  16. [24]

    doi:10.1093/mnras/staa278

    J.S.Speagle,Dynesty:AdynamicnestedsamplingpackageforestimatingBayesianposteriorsandevidences,MNRAS493(2020)3132–3158. doi:10.1093/mnras/staa278

  17. [25]

    R. E. Kass, A. E. Raftery, Bayes Factors, J. Am. Stat. Assoc. 90 (1995) 773–795. doi:10.1080/01621459.1995.10476572

  18. [26]

    Skilling, Nested sampling for general Bayesian computation, Bayesian Anal

    J. Skilling, Nested sampling for general Bayesian computation, Bayesian Anal. 1 (2006) 833–859. doi:10.1214/06-BA127

  19. [27]

    Feroz, M

    F. Feroz, M. P. Hobson, M. Bridges, MultiNest: An efficient and robust Bayesian inference tool for cosmology and particle physics, Mon. Not. R. Astron. Soc. 398 (2009) 1601–1614. doi:10.1111/J.1365-2966.2009.14548.X/2/M_MNRAS0398-1601-M32.GIF. arXiv:0809.3437

  20. [28]

    Buchner, UltraNest - a robust, general purpose Bayesian inference engine, J

    J. Buchner, UltraNest - a robust, general purpose Bayesian inference engine, J. Open Source Softw. 6 (2021) 3001. doi:10.21105/JOSS. 03001.arXiv:2101.09604

  21. [29]

    Ashton, N

    G. Ashton, N. Bernstein, J. Buchner, X. Chen, G. Csányi, A. Fowlie, F. Feroz, M. Griffiths, W. Handley, M. Habeck, E. Higson, M. Hobson, A. Lasenby, D. Parkinson, L. B. Pártay, M. Pitkin, D. Schneider, J. S. Speagle, L. South, J. Veitch, P. Wacker, D. J. Wales, D. Yallup, Nest...

  22. [30]

    B. E. Nelson, E. B. Ford, J. Buchner, R. Cloutier, R. F. Díaz, J. P. Faria, N. C. Hara, V. M. Rajpaul, S. Rukdee, Quantifying the Bayesian Evidence for a Planet in Radial Velocity Data, AJ 159 (2020) 73. doi:10.3847/1538-3881/ab5190

  23. [31]

    K.Perrakis,I.Ntzoufras,E.G.Tsionas, Ontheuseofmarginalposteriorsinmarginallikelihoodestimationviaimportancesampling, Comput. Stat. Data Anal. 77 (2014) 54–69. doi:10.1016/j.csda.2014.03.004

  24. [32]

    Metodiev, M

    M. Metodiev, M. Perrot-Dockès, S. Ouadah, N. J. Irons, P. Latouche, A. E. Raftery, Easily Computed Marginal Likelihoods from Posterior Simulation Using the THAMES Estimator, Bayesian Anal. -1 (2024) 1–28. doi:10.1214/24-BA1422

  25. [33]

    Llorente, L

    F. Llorente, L. Martino, D. Delgado, J. López-Santiago, Marginal Likelihood Computation for Model Selection and Hypothesis Testing: An Extensive Review, SIAM Rev. 65 (2023) 3–58. doi:10.1137/20M1310849. :Preprint submitted to Elsevier Page 15 of 16 Bayesian evidence adaptive pursuit

  26. [34]

    T. H. Prettyman, J. J. Hagerty, R. C. Elphic, W. C. Feldman, D. J. Lawrence, G. W. McKinney, D. T. Vaniman, Elemental composition of the lunar surface: Analysis of gamma ray spectroscopy data from Lunar Prospector, J. Geophys. Res. Planets 111 (2006). doi:10.1029/ 2005JE002656

  27. [35]

    T. H. Prettyman, W. C. Feldman, H. Y. McSween, R. D. Dingler, D. C. Enemark, D. E. Patrick, S. A. Storms, J. S. Hendricks, J. P. Morgenthaler, K. M. Pitman, R. C. Reedy, Dawn’s gamma ray and neutron detector, Space Sci. Rev. 163 (2011) 371–459. doi:10.1007/ s11214-011-9862-0

  28. [36]

    Space Sci

    P.N.Peplowski, Theglobalelementalcompositionof433Eros:FirstresultsfromtheNEARgamma-rayspectrometerorbitaldataset, Planet. Space Sci. 134 (2016) 36–51. doi:10.1016/j.pss.2016.10.006

  29. [37]

    Breitenmoser, A

    D. Breitenmoser, A. Stabilini, M. M. Kasprzak, S. Mayer, Quantitative mobile gamma-ray spectrometry through Bayesian inference, 2025. doi:10.48550/arXiv.2512.18769.arXiv:2512.18769

  30. [38]

    Breitenmoser, G

    D. Breitenmoser, G. Butterweck, M. M. Kasprzak, E. G. Yukihara, S. Mayer, Experimental and Simulated Spectral Gamma-Ray Response of a NaI(Tl) Scintillation Detector used in Airborne Gamma-Ray Spectrometry, Adv. Geosci. 57 (2022) 89–107. doi:10.5194/ ADGEO-57-89-2022

  31. [39]

    S. A. Pozzi, S. D. Clarke, W. J. Walsh, E. C. Miller, J. L. Dolan, M. Flaska, B. M. Wieger, A. Enqvist, E. Padovani, J. K. Mattingly, D. L. Chichester, P. Peerani, MCNPX-PoliMi for nuclearnonproliferation applications, Nucl. Instrum. Methods Phys. Res.A 694 (2012) 119–125. doi...

  32. [40]

    R. F. Lang, J. Pienaar, E. Hogenbirk, D. Masson, R. Nolte, A. Zimbal, S. Röttger, M. L. Benabderrahmane, G. Bruno, Characterization of a deuterium–deuterium plasma fusion neutron generator, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectromet...

  33. [41]

    doi:10.1016/S0092-640X(73)80081-6

    H.Liskien,A.Paulsen, NeutronproductioncrosssectionsandenergiesforthereactionsT(p,n)3He,D(d,n)3He,andT(d,n)4He, AtomicData and Nuclear Data Tables 11 (1973) 569–619. doi:10.1016/S0092-640X(73)80081-6

  34. [42]

    R.Lopez,O.Pakari,C.Ballard,S.Clarke,S.Pozzi, Asimulationpipelineforfastneutronimagingandspectroscopyusingquantifieddetector attributes, Radiat. Meas. 184 (2025) 107440. doi:10.1016/j.radmeas.2025.107440

  35. [43]

    R.J.McConn,C.J.Gesh,R.T.Pagh,R.A.Rucker,CompendiumofMaterialCompositionDataforRadiationTransportModeling,Technical Report, Pacific Northwest National Laboratory, Richland, 2011

  36. [44]

    Jeffreys, Theory Of Probability, 2 ed., Oxford University Press, 1948

    H. Jeffreys, Theory Of Probability, 2 ed., Oxford University Press, 1948

  37. [45]

    Sellke, M

    T. Sellke, M. J. Bayarri, J. O. Berger, Calibration of𝜌Values for Testing Precise Null Hypotheses, Am. Stat. 55 (2001) 62–71. doi:10.1198/000313001300339950

  38. [46]

    F. E. Cecil, E. B. Nieschmidt, Production of 14 MeV neutrons from D-D neutron generators, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 16 (1986) 88–90. doi:10.1016/0168-583X(86)90230-2

  39. [47]

    Foreman-Mackey, Corner.py: Scatterplot matrices in Python, J

    D. Foreman-Mackey, Corner.py: Scatterplot matrices in Python, J. Open Source Softw. 1 (2016) 24. doi:10.21105/joss.00024. :Preprint submitted to Elsevier Page 16 of 16

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.