Pith. sign in

REVIEW 4 major objections 5 minor 28 references

Soft Checksums to Flag Untrustworthy Machine Learning Surrogate Predictions and Application to Atomic Physics Simulations

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Adding one extra output node—a soft checksum—lets a neural-network surrogate flag its own unreliable predictions in the same forward pass.

desk verdict Novel and cheap checksum-based OOD flagging that deserves peer review, but the central claim outruns the evidence: the motivating interior-hole case is never tested. read the letter →

arxiv 2412.03497 v1 pith:C2QV4IMD submitted 2024-12-04 cs.LG physics.atom-ph

Casey Lauer , Robert C. Blake , Jonathan B. Freund This is my paper · ORCID
classification cs.LGphysics.atom-ph
keywords softchecksumout-of-distributiondetectionsurrogatemodelingscientificmachinelearninguncertaintyquantificationatomicphysicssimulationssingleforwardpasserror
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

The paper proposes that the reliability of a neural-network surrogate can be audited by attaching one extra output node—a soft checksum—that the network is trained to satisfy as a function of its own predicted outputs. For any prediction, the mismatch between the check node and the checksum function is computed in the same forward pass that produces the prediction, so flagging untrustworthy out-of-distribution inputs costs almost nothing. The authors demonstrate on a high-dimensional non-local thermodynamic equilibrium atomic physics dataset that this checksum error separates trustworthy in-distribution predictions from out-of-distribution ones, and that deliberately exposing training to out-of-distribution points sharpens the separation. If the method holds, simulation workflows could automatically fall back to expensive physics calculations only when the surrogate flags itself.

What carries the argument

The soft checksum is a continuous analogue of a transmission checksum: one extra output node $\hat{C}_y$ is trained to match a chosen function $C(\hat{y})$ of the network's own predicted outputs, and the checksum error $L(\hat{C}_y, C(\hat{y}))$ is computed during the same forward pass. The loss combines the usual prediction loss with a checksum-matching term, an in-distribution term that penalizes checksum violations on training data, and an out-of-distribution term that rewards violations on points sampled outside the hypercube bounding the training set. The checksum error then acts as the reliability score, with a user-set threshold deciding which predictions are trusted.

What would settle it

Train a surrogate on data that contains a known empty hole inside the bounding hypercube—inputs that are physically possible, between training samples, but never seen. If the checksum error on points inside that hole stays at in-distribution levels while the prediction error rises, the method's central ID/OOD assumption fails for realistic gaps.

Watch

Extended reading notes

Core claim

The central claim is that a neural network trained with an auxiliary output node constrained to satisfy a user-chosen checksum function of its own predictions will violate that relation on out-of-distribution inputs, and the degree of violation—the soft checksum error—is a usable reliability signal. On the NLTE surrogate, with a linear and a sinusoid checksum, the method reaches false-negative rates at a 99% true-negative threshold as low as 1.64% for out-of-distribution points, and the checksum error correlates with prediction error for out-of-distribution data. The paper presents this as a general method requiring no ground-truth labels at inference, no extra models, and only one additional output node, so it can be added to existing architectures with negligible time and memory overhead.

Load-bearing premise

The method assumes that out-of-distribution inputs force the network to violate the learned checksum relation, while in-distribution inputs let the network satisfy it; this behavior is not guaranteed by the training loss, and no baseline comparison establishes it.

Editorial extensions

If this is right

  • Surrogate predictions can be tagged unreliable on the fly, letting a simulation restart the expensive physics solver only when flagged.
  • The method generalizes across regression tasks with no prior on data shape, and an existing physical conservation law can play the role of the checksum.
  • The checksum error can double as a continuous proxy for prediction error, since the two correlate on out-of-distribution data.
  • Training with randomly sampled out-of-distribution points improves separation without biasing toward a particular OOD region, because the samples lie outside the bounding hypercube.
  • Multiple check nodes could make the flag harder to fool, since memorizing one checksum is more plausible than memorizing several.

Reading between the lines

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

  • The threshold for flagging is chosen from validation data at a user-set true-negative rate; a natural extension is to calibrate the threshold per application cost asymmetry between false positives and false negatives.
  • Because OOD sampling is confined to outside the bounding hypercube, interior holes in the training distribution remain a blind spot; testing on deliberately constructed gaps inside the cube would reveal how much this limits the method.
  • The surprising result that the in-distribution penalty term hurts separation suggests the two checksum targets (true output vs predicted output) pull the check node in different directions; an analysis of that conflict could guide loss-design improvements.
  • A soft checksum might also serve as a cheap early-warning signal in online learning or active data acquisition, where the model itself indicates which regions to sample next.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes 'soft checksums' for detecting untrustworthy predictions of neural-network surrogate models in scientific regression. The idea is to add an extra output node that predicts a user-chosen checksum function of the network's own outputs; the discrepancy between this node and the checksum of the predicted outputs serves as a continuous OOD flag. The loss function includes terms that encourage low checksum error on in-distribution data and high checksum error on out-of-distribution data, where OOD points are sampled outside the bounding hypercube of the training set. The method is demonstrated on a high-dimensional NLTE atomic-physics surrogate (87 input, 85 output), reporting FNR99 values of 4.76% for a linear checksum and 1.64% for a sinusoid checksum when the OOD-reward loss is included, together with a positively correlated relationship between checksum error and prediction error.

Significance. If the central claim holds, the method is attractive: it requires a single model, a single forward pass, negligible extra memory, and no ground-truth values at test time. The authors are explicit that the method is general and can be layered onto existing architectures. However, the significance is tempered by the absence of any comparison to existing OOD-detection and uncertainty-quantification baselines, and by the fact that the demonstrated regime (exterior extrapolation) does not cover the internal-gap regime that motivates the work. The paper also honestly acknowledges several of these limitations, which is a strength in presentation but does not by itself resolve them.

major comments (4)
  1. [Section 5 (Discussion) and Section 4 (Numerical Experiment)] The motivating scenario—holes inside the bounding hypercube of the training data—is never tested. The Introduction argues that gaps inside the training-data hypercube are a likely failure mode for high-dimensional physical problems, and Section 5 explicitly concedes that 'sampling outside of a bounding hypercube ... also misses potential OOD regions within the hypercube and holes within the training dataset.' Both the training-time OOD loss (Section 4, LOOD sampled 20–25% outside the hypercube) and the evaluation OOD set (Figure 2, an arbitrary dividing line in the density–temperature plane) are exterior extrapolations. Therefore the FNR99 values in Table 1 support the Section 3.1 claim only for exterior OOD points, not for the interior-hole case that the paper itself identifies as the key practical problem. For an interior hole, the input lies inside the coordinate range of the training data, so the network has no incentive from LOOD to violate the checksum relation, and the method could fail to flag exactly the predictions that are most dangerous.
  2. [Section 4 (Hyperparameter Selection) and Table 1] The hyperparameters λID, λOOD, and w were selected by a 'limited parameter sweep' that, in the authors' own words, 'depended on the chosen OOD dataset.' Since the same OOD dataset is used to evaluate the FNR99 values in Table 1, the reported numbers are optimistically biased by selection on the test set. The paper acknowledges this, but the Table 1 results are the main quantitative evidence for the method's effectiveness, and without an independent OOD validation set or a sensitivity analysis, the reported FNR99 values cannot be taken as unbiased estimates of performance on unseen OOD data.
  3. [Section 5 (Discussion, 'we must also conduct benchmark comparisons')] No comparisons to existing OOD-detection or uncertainty-quantification methods are provided. The paper states that 'we must also conduct benchmark comparisons to establish the relative effectiveness,' but the abstract and Section 1 claim that soft checksums can 'effectively separate' ID and OOD predictions. Without baselines such as deep ensembles, MC dropout, or simple distance-to-training-data measures, the reader cannot judge whether the FNR99 numbers represent a practical improvement or merely a particular result on one dataset.
  4. [Table 1 and Section 4 (training protocol)] The paper does not state the number of independent training runs or report any variance or confidence intervals for the FNR99 values. If each entry in Table 1 comes from a single model, the differences between loss functions (e.g., 4.76% vs. 1.64%) may not be robust, and the central claim that the OOD-reward term improves separation is not statistically supported. The authors should either report multiple seeds with means and spreads or clearly state that the results are from a single run and should be interpreted accordingly.
minor comments (5)
  1. [Figure 3] The claimed 'linear correlation' between checksum error and prediction error is assessed only visually; the paper should report a quantitative correlation coefficient (e.g., Pearson or Spearman r) or R² for each panel.
  2. [Section 3.1 (threshold selection)] The threshold is defined as the 99% true-negative value on Dvalidation, but the practical procedure for choosing this threshold when validation labels are unavailable or when the cost of false negatives is not known is not discussed.
  3. [Section 4 (OOD sampling details)] The description of the LOOD sampling ('a subset of DOOD with values between 20% to 25% outside the hypercube') does not specify how many points are sampled, whether they are drawn uniformly, or whether any of these synthetic points overlap with the evaluation OOD set defined by the dividing line in Figure 2; these details affect both the loss computation and the evaluation.
  4. [Section 4 (parameter sweep)] The paper states that λID and λOOD were set to 0.01 after a sweep, but gives no information about the range of values considered or the criterion used to choose them; providing this context would help readers assess the sensitivity of the method.
  5. [Section 3.1 and Section 5] The claim that the method 'makes no a priori assumptions about the data' is too strong, since the user must choose a checksum function and an OOD sampling distance, both of which require assumptions about the geometry of the input space.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial evaluation leakage: LOOD is computed from the same OOD set used for FNR99, so the demonstrated improvement is partly by construction; the core checksum mechanism itself is not self-definitional.

  1. fitted input called prediction [Section 4, Eq. (4d), Table 1]
    "We generate the dataset by running many ICF simulations in Cretin, and manually divide ID data for training and validation, and OOD data for evaluating the soft checksum metric, as shown in Figure 2. ... To calculate LOOD, we sample a subset of DOOD with values between 20% to 25% outside of the hypercube bounding Dtraining. ... we conducted a limited parameter sweep to determine the optimal hyperparameters for the given experiment. Importantly, this was not a general method of selecting the hyperparameters and depended on the chosen OOD dataset."

    Eq. (4d) defines the LOOD reward on inputs drawn from DOOD, and Section 4 identifies DOOD as the same set used 'for evaluating the soft checksum metric.' The FNR99 values in Table 1 are computed on this DOOD set. Thus, for the rows that include LOOD, the model was explicitly trained to produce large checksum errors on a subset of the evaluation OOD points, and the hyperparameters were selected using that same OOD dataset. The improved FNR99 values are therefore not an independent out-of-sample prediction; they are in part a check that the training objective was optimized on the evaluation set. The baseline row without LOOD is not affected, so the core checksum idea retains independent empirical content.

full rationale

Most of the paper's derivation chain is not circular. The checksum error is defined as L(Ĉy, C(ŷ)), a function of network outputs alone, while OOD is defined by input-space distribution, not by the checksum error. The base loss Lprediction + Lchecksum, without LOOD, already yields nontrivial FNR99 values (8.93% for the linear checksum and 3.84% for the sinusoid), so the central claim that checksum violations correlate with prediction error has independent empirical content. The main circular element is confined to the demonstration of the LOOD improvement: Eq. (4d) computes the OOD reward on a subset of DOOD, and Section 4 explicitly states that DOOD is the set used for evaluating the soft checksum metric, with hyperparameters chosen by a sweep that 'depended on the chosen OOD dataset.' Consequently, the improved FNR99 numbers in the LOOD rows of Table 1 are partly a measure of fitting the training objective to the evaluation set, not a prediction on unseen OOD data. This is evaluation leakage rather than definitional circularity. The Section 5 admission that sampling outside a bounding hypercube 'misses potential OOD regions within the hypercube and holes within the training dataset' is a generalization/validity limitation, not a circularity, because it concerns whether the learned checksum-violation behavior transfers to untested interior gaps. No load-bearing self-citations, imported uniqueness theorems, or ansatz-smuggling citations are present.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The method rests on a small set of hand-chosen hyperparameters (two loss weights, one sinusoid frequency, an OOD sampling distance) and three behavioral assumptions about how neural networks generalize. None of the free parameters is derived from theory; all were set by a sweep that the paper states depended on the chosen OOD dataset. No new physical entities are proposed.

free parameters (5)
  • lambda_ID = 0.01
    Weight on the ID checksum penalty in Eq. 4c; chosen by a limited parameter sweep that depended on the chosen OOD dataset (Section 4).
  • lambda_OOD = 0.01
    Weight on the OOD checksum reward in Eq. 4d; selected by the same sweep-dependent procedure (Section 4).
  • sinusoid frequency w = 0.0001
    Frequency in Eq. 2; set in Section 4 to keep the checksum learnable but nonlinear.
  • OOD sampling distance = 20% to 25% outside the bounding hypercube
    Region from which LOOD training points are drawn (Section 4); chosen ad hoc and not optimized.
  • FNR99 threshold = 99% true-negative value on Dvalidation
    User-set threshold in Section 3.1; determines the reported false negative rates. It is data-dependent but not a model parameter.
assumptions (4)
  • domain assumption Checksum error on OOD inputs is large and on ID inputs is small.
    Section 3.1: 'If the model is unable to produce a small enough checksum error, then the model is likely predicting on OOD data.' This behavioral assumption is the foundation of the detector.
  • ad hoc to paper Random points outside the bounding hypercube of Dtrain are valid OOD data for training.
    Section 3.2 introduces this sampling to compute LOOD; Section 5 admits it misses OOD regions inside the hypercube and holes in the training data.
  • ad hoc to paper The chosen checksum functions are learnable on ID data but not memorized for OOD inputs.
    Section 5 item 1 states that an ideal checksum is complex enough that the model cannot memorize it, but simple enough to learn on ID; the frequency w was chosen by hand.
  • domain assumption The validation set accurately represents the in-distribution domain.
    The 99% true-negative threshold is calibrated on Dvalidation (Section 3.1), so the reported FNR99 depends on the validation set being representative of ID.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Soft Checksums to Flag Untrustworthy Machine Learning Surrogate Predictions and Application to Atomic Physics Simulations." pith.science (2026). https://pith.science/paper/C2QV4IMD

@misc{pith2026241203497,
  author       = {Pith},
  title        = {Pith review of: Soft Checksums to Flag Untrustworthy Machine Learning Surrogate Predictions and Application to Atomic Physics Simulations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C2QV4IMD}},
  note         = {Machine review of arXiv:2412.03497}
}
read the original abstract

Trained neural networks (NN) are attractive as surrogate models to replace costly calculations in physical simulations, but are often unknowingly applied to states not adequately represented in the training dataset. We present the novel technique of soft checksums for scientific machine learning, a general-purpose method to differentiate between trustworthy predictions with small errors on in-distribution (ID) data points, and untrustworthy predictions with large errors on out-of-distribution (OOD) data points. By adding a check node to the existing output layer, we train the model to learn the chosen checksum function encoded within the NN predictions and show that violations of this function correlate with high prediction errors. As the checksum function depends only on the NN predictions, we can calculate the checksum error for any prediction with a single forward pass, incurring negligible time and memory costs. Additionally, we find that incorporating the checksum function into the loss function and exposing the NN to OOD data points during the training process improves separation between ID and OOD predictions. By applying soft checksums to a physically complex and high-dimensional non-local thermodynamic equilibrium atomic physics dataset, we show that a well-chosen threshold checksum error can effectively separate ID and OOD predictions.

Figures

Figures reproduced from arXiv: 2412.03497 by the authors.

Figure 1
Figure 1. Adding a check node to the neural network allows the user to encode a checksum function into the output [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. We generated the training, validation and out-of-distribution (OOD) datasets from trusted [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Relationship between checksum error and prediction error with an optimized loss function, and either a [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 21 canonical work pages

  1. [1]

    Verification, validation, and predictive capability in computa- tional engineering and physics,

    W. L. Oberkampf, T. G. Trucano, and C. Hirsch, “Verification, validation, and predictive capability in computa- tional engineering and physics,” Applied Mechanics Reviews, vol. 57, pp. 345–384, Dec. 2004

    work page 2004

  2. [2]

    Unified Framework and Survey for Model Verifica- tion, Validation and Uncertainty Quantification,

    S. Riedmaier, B. Danquah, B. Schick, and F. Diermeyer, “Unified Framework and Survey for Model Verifica- tion, Validation and Uncertainty Quantification,” Archives of Computational Methods in Engineering , vol. 28, pp. 2655–2688, June 2021

    work page 2021

  3. [3]

    Fundamental concepts and terminology,

    C. J. Roy and W. L. Oberkampf, “Fundamental concepts and terminology,” in Verification and Validation in Scientific Computing, pp. 21–82, Cambridge: Cambridge University Press, 2010

    work page 2010

  4. [4]

    Accelerating Chemical Kinetics Calculations With Physics Informed Neural Networks,

    A. Almeldein and N. Van Dam, “Accelerating Chemical Kinetics Calculations With Physics Informed Neural Networks,” Journal of Engineering for Gas Turbines and Power, vol. 145, July 2023

    work page 2023

  5. [5]

    Surrogate modelling of VLE: Integrating machine learning with thermodynamic constraints,

    A. Carranza-Abaid, H. F. Svendsen, and J. P. Jakobsen, “Surrogate modelling of VLE: Integrating machine learning with thermodynamic constraints,” Chemical Engineering Science: X, vol. 8, p. 100080, Nov. 2020

    work page 2020

  6. [6]

    Design Space Exploration of Turbulent Multiphase Flows Using Machine Learning-Based Surrogate Model,

    H. Ganti, M. Kamin, and P. Khare, “Design Space Exploration of Turbulent Multiphase Flows Using Machine Learning-Based Surrogate Model,” Energies, vol. 13, p. 4565, Jan. 2020. Number: 17 Publisher: Multidisci- plinary Digital Publishing Institute. 7

    work page 2020

  7. [7]

    Deep learning for NLTE spectral opacities,

    G. Kluth, K. D. Humbird, B. K. Spears, J. L. Peterson, H. A. Scott, M. V . Patel, J. Koning, M. Marinak, L. Divol, and C. V . Young, “Deep learning for NLTE spectral opacities,” Physics of Plasmas , vol. 27, p. 052707, May 2020

    work page 2020

  8. [8]

    Physics-informed neural networks: A deep learning frame- work for solving forward and inverse problems involving nonlinear partial differential equations,

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics-informed neural networks: A deep learning frame- work for solving forward and inverse problems involving nonlinear partial differential equations,” Journal of Computational Physics, vol. 378, pp. 686–707, Feb. 2019

    work page 2019

Show all 28 references
  1. [9]

    Energy-based Out-of-distribution Detection,

    W. Liu, X. Wang, J. Owens, and Y . Li, “Energy-based Out-of-distribution Detection,” in Advances in Neural Information Processing Systems, vol. 33, pp. 21464–21475, Curran Associates, Inc., 2020

    2020

  2. [10]

    Deep Anomaly Detection with Outlier Exposure,

    D. Hendrycks, M. Mazeika, and T. Dietterich, “Deep Anomaly Detection with Outlier Exposure,” Jan. 2019. arXiv:1812.04606 [cs, stat]

    2019 arXiv

  3. [11]

    Weight Uncertainty in Neural Networks,

    C. Blundell, J. Cornebise, K. Kavukcuoglu, and D. Wierstra, “Weight Uncertainty in Neural Networks,” May

  4. [12]

    Bayesian Neural Network Priors Revisited,

    V . Fortuin, A. Garriga-Alonso, S. W. Ober, F. Wenzel, G. R¨atsch, R. E. Turner, M. van der Wilk, and L. Aitchison, “Bayesian Neural Network Priors Revisited,” Mar. 2022. arXiv:2102.06571 [cs, stat]

    2022 arXiv

  5. [13]

    R. M. Neal, Bayesian Learning for Neural Networks , vol. 118 of Lecture Notes in Statistics . New York, NY: Springer, 1996

    1996

  6. [14]

    Bayesian Deep Learning and a Probabilistic Perspective of Generalization,

    A. G. Wilson and P. Izmailov, “Bayesian Deep Learning and a Probabilistic Perspective of Generalization,” Mar

  7. [15]

    Output- Constrained Bayesian Neural Networks,

    W. Yang, L. Lorch, M. A. Graule, S. Srinivasan, A. Suresh, J. Yao, M. F. Pradier, and F. Doshi-Velez, “Output- Constrained Bayesian Neural Networks,” May 2019. arXiv:1905.06287 [cs, stat]

    2019 arXiv

  8. [16]

    Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning,

    Y . Gal and Z. Ghahramani, “Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning,” Oct. 2016. arXiv:1506.02142 [cs, stat]

    2016 arXiv

  9. [17]

    Fast dropout training,

    S. Wang and C. Manning, “Fast dropout training,” in Proceedings of the 30th International Conference on Ma- chine Learning, pp. 118–126, PMLR, May 2013. ISSN: 1938-7228

    2013

  10. [18]

    Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles,

    B. Lakshminarayanan, A. Pritzel, and C. Blundell, “Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles,” Nov. 2017. arXiv:1612.01474 [cs, stat]

    2017 arXiv

  11. [19]

    Single Model Uncertainty Estimation via Stochastic Data Centering,

    J. J. Thiagarajan, R. Anirudh, V . Narayanaswamy, and P.-T. Bremer, “Single Model Uncertainty Estimation via Stochastic Data Centering,” Dec. 2022. arXiv:2207.07235 [cs, stat]

    2022 arXiv

  12. [20]

    PAGER: Accurate Failure Characterization in Deep Regression Models,

    J. J. Thiagarajan, V . Narayanaswamy, P. Trivedi, and R. Anirudh, “PAGER: Accurate Failure Characterization in Deep Regression Models,” June 2024

    2024

  13. [21]

    Generalized Out-of-Distribution Detection: A Survey,

    J. Yang, K. Zhou, Y . Li, and Z. Liu, “Generalized Out-of-Distribution Detection: A Survey,” International Journal of Computer Vision, June 2024

    2024

  14. [22]

    An Arithmetic Checksum for Serial Transmissions,

    J. Fletcher, “An Arithmetic Checksum for Serial Transmissions,”IEEE Transactions on Communications, vol. 30, pp. 247–252, Jan. 1982. Conference Name: IEEE Transactions on Communications

    1982

  15. [23]

    Efficient error estimating coding: feasibility and applications,

    B. Chen, Z. Zhou, Y . Zhao, and H. Yu, “Efficient error estimating coding: feasibility and applications,” SIG- COMM Comput. Commun. Rev., vol. 40, pp. 3–14, Aug. 2010

    2010

  16. [24]

    mEEC: A novel error estimation code with multi-dimensional feature,

    Z. Zhang and P. Kumar, “mEEC: A novel error estimation code with multi-dimensional feature,” inIEEE INFO- COM 2017 - IEEE Conference on Computer Communications, pp. 1–9, May 2017

    2017

  17. [25]

    Soft Checksum Method for Error-tolerant Multi-hop Transmission in Wireless Sen- sor Networks,

    M.-S. Lee and S. Bahk, “Soft Checksum Method for Error-tolerant Multi-hop Transmission in Wireless Sen- sor Networks,” in 2021 International Conference on Information and Communication Technology Convergence (ICTC), pp. 1895–1897, Oct. 2021. ISSN: 2162-1233

    2021

  18. [26]

    Cretin—a radiative transfer capability for laboratory plasmas,

    H. A. Scott, “Cretin—a radiative transfer capability for laboratory plasmas,” Journal of Quantitative Spec- troscopy and Radiative Transfer, vol. 71, pp. 689–701, Oct. 2001. 8

    2001

  19. [2015]

    arXiv:1505.05424 [cs, stat]

    arXiv

  20. [2022]

    arXiv:2002.08791 [cs, stat]

    2002 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.