Pith. sign in

REVIEW 4 major objections 3 minor 39 references

AI-Assisted analysis of $^{28}$Si$^*$ $\rightarrow$ 7$\alpha$ break-up data

T0 review · 4 major / 3 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that Gaussian Mixture Model analysis of the 7-alpha excitation spectra from the 28Si + 12C reaction at 35 MeV/A reveals reproducible components near 114 and 138 MeV in both published experimental datasets, indicating…

desk verdict New cHαC collision results and a clear GMM write-up, but the toroidal claim rests on a method never tested against a smooth background, and Fig. 8(b) admits arbitrary peak placement. read the letter →

arxiv 2505.23920 v1 pith:A2Z6QFKE submitted 2025-05-29 nucl-th

Theodoros Depastas , Aldo Bonasera , Joe Natowitz This is my paper
classification nucl-th PACS 25.70.-z21.60.-n25.70.Pq
keywords 28SitoroidalstatesalphaclusteringGaussianMixtureModel7-alphabreakupexcitationenergyspectraheavy-ionreactionsmachinelearninginnuclearphysics
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

Two published experiments on the peripheral 28Si + 12C reaction at 35 MeV/A disagreed about whether the 7-alpha breakup spectrum of 28Si* contains resonant peaks: one reported candidate toroidal resonances near 114, 126 and 138 MeV, while the other saw only statistical fluctuations over a smooth background. This paper argues that the disagreement is an artifact of how the background was subtracted. It develops a Gaussian Mixture Model analysis that decomposes the same spectra into Gaussian components without assuming a background shape, and applies it to both experimental datasets, to rotating-silicon model spectra, and to new collision calculations. The extracted components cluster near the toroidal-state predictions, and the centroids stay stable when the second experiment's events are split into eight time-ordered subsets. If the analysis is correct, it resolves the dispute in favor of real underlying structure close to toroidal excitations.

What carries the argument

The central object is the Gaussian Mixture Model used as an unsupervised analytical tool: a histogram of counts or cross section versus excitation energy is normalized to a probability distribution, sampled with $10^6$ synthetic points drawn from normal distributions centered on each bin, and decomposed by expectation-maximization into a sum of Gaussians. The initial means are set by the paper's Eq. (2), $\mu_{I_G}^{(0)} = (x_{\max}-x_{\min})\,I_G(I_G+1)/(N_G(N_G-1)) + x_{\min}$, and the number of Gaussians is chosen by requiring reduced chi-square $\chi^2_\nu$ close to 1. The same machinery applied to time-ordered partitions of the second experiment's data serves as the statistical-significance test. The HαC and cHαC models provide the theoretical spectra: a molecular-dynamics approach with $\alpha$ particles as semiclassical degrees of freedom, in which 7-$\alpha$ events are selected by two filters and excitation energies are computed from kinetic, Coulomb, and Q-value terms via Eq. (5).

What would settle it

Take the full statistics of the second experiment but replace the measured excitation energies with samples from a smooth phase-space or mixed-event distribution that matches the same total yield and errors, then run the same Gaussian Mixture Model pipeline on eight time-ordered partitions. If stable Gaussian components appear near 114 and 138 MeV in this background-only control, the claimed underlying structure is an artifact of the method rather than a resonance.

Watch

Extended reading notes

Core claim

The central claim is that all examined datasets — the two experiments and the new cHαC collision model — contain excitation-energy structure compatible with toroidal states of 28Si* around 114 MeV and 138 MeV. The authors do not claim to reproduce every predicted resonance: the 126 MeV peak, for instance, is not recovered as an independent Gaussian but falls inside a wider component. The supporting demonstration is statistical: dividing the 186,097 unbinned events of the second experiment into eight time-ordered partitions and re-running the decomposition yields centroid shifts of only about 2.6 percent, which the paper presents as evidence that the extracted peaks are not random fluctuations. The paper further shows that a 9th-order polynomial background subtraction creates spurious maxima and can erase real peaks, and that the same GMM procedure recovers seven of ten known Gaussian components in the rotating-silicon model spectra where the polynomial-root method recovers only four of twelve.

Load-bearing premise

The load-bearing premise is that decomposing a histogram into a few Gaussians with reduced chi-square close to one identifies physically meaningful resonances, rather than merely being a flexible way to approximate any smooth distribution.

Editorial extensions

If this is right

  • If the central claim is right, the null result of the second experiment reflects the polynomial background-subtraction procedure rather than the absence of resonances; the six Gaussian components found in its data align with the 114 and 138 MeV toroidal predictions.
  • The new cHαC collision calculations converge to the experimental data at high excitation energy, implying strong alpha clustering in the most energetic fragments and giving an upper-limit cross section because non-alpha-conjugate channels are absent from the model.
  • The about 2.6 percent centroid variation across time-ordered subsets offers a practical protocol for using GMM stability as a statistical-significance test for resonance claims in breakup spectra.
  • The widths and lifetimes of the extracted peaks being comparable to giant E0/E1/E2 resonances of 28Si places the claimed structure in a collective-excitation regime rather than a narrow-resonance regime.

Reading between the lines

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

  • The authors do not run the GMM on a purely smooth, resonance-free spectrum; that control would show how often the method finds stable Gaussians by construction, and it is the most direct test of the claim.
  • Because the rotating-silicon training data are themselves a known sum of Gaussians, recovering seven of ten components is a check on the fitting machinery rather than an external validation of physical content; the time-slice stability on real data carries more weight.
  • A natural extension is to apply the same pipeline to simulated background-only spectra with the same statistics as the experiments to map the false-positive rate as a function of counts per bin.
  • If the claim is right, higher-statistics future experiments should resolve the 126 MeV peak as an independent Gaussian rather than a shoulder inside a wider component, which is a concrete, testable prediction.
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

4 major / 3 minor

Summary. The paper develops a Gaussian Mixture Model (GMM) analysis of 7-alpha breakup excitation-energy spectra and applies it to the Cao and Hannaman experimental datasets, to the rotating-silicon HαC training data of Ref. [21], and to new collision HαC (cHαC) calculations performed by the authors. The central claim, stated in the abstract and conclusion, is that in all examined datasets the GMM decomposition reveals Gaussian components whose centroids lie close to the energies predicted for toroidal states of 28Si, notably near 114 MeV and 138 MeV, and that the time-ordered partitions of the Hannaman data show that these components are statistically stable. The paper also criticizes the 9th-order polynomial background-subtraction method used by Hannaman et al. and argues that the GMM approach resolves the dispute between Refs. [15] and [17] in favor of real resonant structure.

Significance. If the central claim were fully established, the paper would be a significant contribution to the toroidal-state controversy for 28Si and would offer a reusable analysis tool for sparse excitation-energy spectra. The new cHαC collision calculations are a useful theoretical addition, and the time-partition stability check in Figs. 6 and 7 is a commendable attempt to address statistical significance. However, the paper's diagnostic engine, the GMM decomposition, is validated only on data that are themselves sums of Gaussians, and no null-background test is provided. Because the interpretation of stable Gaussian components as physical resonances is the load-bearing assumption, the paper currently overclaims its conclusion.

major comments (4)
  1. [Sec. II, Eq. (3)] The criterion that the reduced chi-square be close to unity is necessary but not sufficient to identify resonances; a flexible Gaussian mixture can also achieve chi-square near unity for a featureless continuum. The paper does not include a null-hypothesis test in which the same pipeline (binning, Np resampling, EM initialization, and N_G selection) is applied to a smooth background-only spectrum with similar statistics. Because the abstract and Section V conclude 'underlying structure' from this decomposition, the absence of such a test is load-bearing. Please add at least one null test, for example a phase-space or smoothed polynomial continuum with the same binning and errors, and report the resulting centroid stability and proximity to the x=y line.
  2. [Sec. IV, Fig. 8(b)] The explicit statement that peaks outside the prediction limits are 'positioned arbitrarily close to the x=y line' means that part of the apparent agreement in panel (b) is by construction. This directly weakens the cross-dataset comparison with the toroidal predictions. Please report the unadjusted centroid values for all datasets, show the comparison without arbitrary repositioning, and state explicitly how many components fall outside the prediction limits and at what energies.
  3. [Sec. II, Fig. 1] The method validation on the rotating-silicon HαC data of Ref. [21] is not an external or blind test because that spectrum is itself generated as a sum of Gaussian components plus a phase-space factor. Recovering those injected Gaussians demonstrates that the algorithm can decompose a mixture of Gaussians, but it does not test the central diagnostic assumption that stable GMM components in an experimental histogram correspond to physical resonances rather than arbitrary smooth structure. Add at least one validation case with a known non-Gaussian background and, ideally, an independent experimental spectrum with confirmed resonances.
  4. [Sec. II, Eq. (1)] The resampling step generates Np=10^6 points 'from a normal distribution around the center of each bin', but the standard deviation of these normal distributions is not specified. If the width is tied to the bin size, the procedure can artificially broaden or smooth the histogram and can introduce smooth Gaussian-like components into a featureless spectrum. Please specify the resampling width, justify the choice, and report the sensitivity of the extracted centroids and of N_G to Np and to the resampling width.
minor comments (3)
  1. [General] There are several typographical and formatting errors, including 'Stasczak' for Staszczak, 'a only variation' for 'only a variation', 'un-binned' for 'unbinned', and inconsistent spacing in 'F AUST' and '7αcross section'.
  2. [Sec. III and Fig. 3] The scaling of the Hannaman data by a factor of 2x10^-6 (mb/MeV) to the low-energy region of the Cao data is described only briefly; the units, the fitting procedure, and the uncertainty of this normalization should be stated, since the quantitative comparison of calculated and experimental cross sections depends on it.
  3. [Sec. I] The phrase 'novel Artificial Intelligence (AI)-based machine learning method' overstates the degree of novelty; GMM with expectation-maximization is a standard unsupervised technique, and the paper would be clearer if it described the specific new elements beyond applying scikit-learn with a chosen initialization and chi-square selection.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the GMM centroid extraction does not fit the toroidal predictions, and the central comparison is not forced by construction.

full rationale

The paper's central claim is that Gaussian Mixture Model decomposition of experimental and theoretical excitation-energy spectra yields Gaussian centroids near 114 MeV and 138 MeV, consistent with toroidal-state predictions. No circular reduction is present in the derivation chain. In Eqs. (1)-(3), the Gaussian means, widths, and weights are free parameters; the initial means are set by Eq. (2) using only the x-range of the data (x_min, x_max), not by the toroidal-state energies. The number of Gaussians is selected by reduced chi-square near unity, and the extracted centroids are subsequently compared with, not fitted to, the predicted toroidal resonances of Refs. [15,16,21]. The validation on the rotating-silicon HαC data of Ref. [21] is indeed a self-citation involving coauthor Bonasera, and that dataset is itself built from known Gaussian components, so it demonstrates only that the algorithm can recover injected Gaussian structure. However, this validation is not the load-bearing step for the paper's experimental conclusion; the application to the Cao and Hannaman data is an independent analysis, and Fig. 8(b) uses the external benchmark of Ref. [16]. The time-split stability test of Figs. 6-7 shows reproducibility of the decomposition within the Hannaman data, which is not a substitute for a proper null-hypothesis test against a smooth background; this is a methodological limitation, not a circular derivation. No uniqueness theorem is imported from the authors' prior work, no fitted parameter is renamed as a prediction, and no known result is merely renamed. The arbitrary placement of out-of-range peaks in Fig. 8(b) weakens the visual comparison but does not make the conclusion equivalent to the input by construction. Therefore, under the requirement that circularity be exhibited by a specific reduction, the paper contains no significant circular step.

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

The paper introduces no new physical entities. Its central claim rests on a set of model assumptions from the HαC framework, on the ad hoc assumption that Gaussian decompositions correspond to physical resonances, and on several free parameters including the number of Gaussians and the chi-square threshold. The parameter count is moderate for a phenomenological analysis, but the lack of a null model makes the extracted peaks difficult to interpret.

free parameters (6)
  • Number of Gaussian components N_G = 6 for experimental spectra, 10 for HαC training data
    Selected by requiring reduced chi-square close to 1; controls how many peaks are extracted and has no independent physical prior.
  • Reduced chi-square threshold close to 1 = Not specified numerically
    Hand-chosen model selection criterion; different thresholds would change the number and properties of the extracted Gaussians.
  • Uniform 5% error assumption for HαC training data = 5%
    Used in the chi-square computation for the method test; arbitrary and directly affects the fit quality and N_G.
  • Scaling factor for Hannaman data = 2 x 10^-6 (mb/MeV)
    Arbitrary normalization to match the low-energy tail of Cao data; affects the comparison in Fig. 3 but not the peak extraction.
  • Binning widths = 2 MeV for model data, 1.25 MeV for Hannaman data
    Binning affects the histogram shape and the resampled distribution, and therefore the extracted Gaussian components.
  • Time partition size = 8 groups
    The stability test depends on dividing the 186097 unbinned points into eight equal time-ordered groups; the choice is not motivated quantitatively.
assumptions (5)
  • domain assumption The HαC model with Coulomb, Bass, and effective Fermi repulsion is a valid description of alpha-cluster dynamics.
    Invoked in Section III; the new cHαC spectra depend entirely on this model, and no independent evidence is supplied beyond the cited papers.
  • domain assumption The selected 7alpha events (Filter 1 or Filter 2) represent the projectile-like 28Si* decay relevant to toroidal states.
    Section III, Eqs. (5)-(10); the two filters give somewhat different spectra, so the choice of filter affects the extracted peaks.
  • ad hoc to paper A Gaussian mixture decomposition of a normalized histogram represents underlying physical resonances.
    Section II, Eq. (1), and Section IV; no null model or uniqueness guarantee is provided, so the decomposition may be purely mathematical.
  • domain assumption The HαC Q-values and ground-state energies are accurate within 2%.
    Section III, Eq. (10); the excitation energy scale of all model results relies on this model calibration.
  • ad hoc to paper Time-ordered partition stability implies statistical significance.
    Section IV, Figs. 6-7; no formal hypothesis test is performed, and a stable instrumental background would also produce stable Gaussian fits.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AI-Assisted analysis of $^{28}$Si$^*$ $\rightarrow$ 7$\alpha$ break-up data." pith.science (2026). https://pith.science/paper/A2Z6QFKE

@misc{pith2026250523920,
  author       = {Pith},
  title        = {Pith review of: AI-Assisted analysis of $^28$Si$^*$ $\rightarrow$ 7$\alpha$ break-up data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A2Z6QFKE}},
  note         = {Machine review of arXiv:2505.23920}
}
abstract

Mid-weight $\alpha$-conjugate nuclei are predicted to possess exotic toroid-like resonances with high angular momenta. The search for these states in $^{28}$Si$^*$ is the main point of two published experimental investigations of the peripheral $^{28}$Si + $^{12}$C reaction by Cao and collaborators and by Hannaman and collaborators. In this work, we develop a novel Artificial Intelligence (AI)-based machine learning method utilizing the Gaussian Mixture Model (GMM) to analyze available experimental and theoretical data. We additionally study the reaction with the Hybrid $\alpha$-Cluster (H$\alpha$C) model. In all the examined data our results suggest the presence of underlying structure which is close to that predicted for toroidal states.

Figures

Figures reproduced from arXiv: 2505.23920 by the authors.

Figure 1
Figure 1. (Color online) Panel (a): Original HαC toroid data and their Gaussian components, according to the key (repro￾duction of [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. (Color online) Partial and total differential cross sections for different ranges of [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. (Color online) Differential Cross Sections calculated [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: (Color online) Panel (a): Results for the Cao [15] and panel (b) for the Hannaman [17] experimental data. The [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: (Color online) The polynomial fit of Ref. [17] (full [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: (Color online) AI calculations of the time-ordered partitions from the Hannaman data [17]. The excitation energies [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: (Color online) Evolution of AI Gaussians along the time-ordered partitions from the Hannaman data [17]. Each panel [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: (Color online) Collective machine learning excitation energy results for the theoretical and experimental datasets [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: (Color online) Collective machine learning Full [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

39 extracted references · 35 canonical work pages

  1. [21]

    The Cao experimen- tal data obtained by Ref

    corresponds to the green curve. The Cao experimen- tal data obtained by Ref. [15] are shown in green and the Hannaman data in magenta. The latter are taken from Fig. 2 of Ref. [17], where they are binned with a width of 1.25 MeV and include standard Poisson errors. The transformation from counts to differential cross section occurs via the scaling with a ...

    work page

  2. [15]

    Our observations of similarity between the results for all the available datasets, are also shown in Fig

    and 143.18 MeV [14] are found between the resulting peaks. Our observations of similarity between the results for all the available datasets, are also shown in Fig. 9. There we plot the FWHM against the Excitation Energy cen- troids for each M.L. Gaussian, following the same process as in Fig. 6. The arrows show areas of common excita- tion energy peaks i...

    work page

  3. [17]

    Hannaman, B

    A. Hannaman, B. Harvey, A. B. McIntosh, K. Hagel, A. Abbott, A. Fentress, J. Gauthier, T. Hankins, Y.-W. Lui, L. McCann, L. A. McIntosh, S. Regener, R. Rider, S. Schultz, M. Q. Sorensen, J. Tobar, Z. N. Tobin, and S. J. Yennello, Phys. Rev. C109, 054615 (2024)

    work page 2024

  4. [1]

    Basunia and A

    M. Basunia and A. Chakraborty, Nucl. Data Sheets186 (2022)

    work page 2022

  5. [2]

    Ikeda, N

    K. Ikeda, N. Takigawa, and H. Horiuchi, Progress of The- oretical Physics SupplementE68, 464 (1968)

    work page 1968

  6. [3]

    Hoyle, Astrophys

    F. Hoyle, Astrophys. J. Suppl.1, 121 (1954)

    work page 1954

  7. [4]

    AI-Assisted analysis of $^{28}$Si$^*$ $\rightarrow$ 7$\alpha$ break-up data

    and rotational bands of deformed nuclei [2], to the equation of state ofα-clustered Nuclear Matter [5, 6]. The presence of toroid-like excited states was first pre- dicted by Bohr, Mottelson, Wheeler, Nilsson, Rainwater and later Wong [7–11]. More specifically, Wong stud- ied these structures in light to mid-weight nuclei with high angular momenta from th...

    work page Pith review arXiv 2024

  8. [5]

    Brink and J

    D. Brink and J. Castro, Nuclear Physics A216, 109 (1973)

    work page 1973

Show all 39 references
  1. [6]

    Zheng and A

    H. Zheng and A. Bonasera, Phys. Rev. C83, 057602 (2011)

    2011

  2. [7]

    Borderie, A

    B. Borderie, A. Raduta, E. De Filippo, E. Geraci, N. L. Neindre, G. Cardella, G. Lanzalone, I. Lombardo, O. Lopez, C. Maiolino, A. Pagano, M. Papa, S. Pirrone, F. Rizzo, and P. Russotto, Symmetry13(2021)

    2021

  3. [8]

    Bohr and B

    A. Bohr and B. R. Mottelson, Phys. Rev.89, 316 (1953)

    1953

  4. [9]

    J. J. Griffin and J. A. Wheeler, Phys. Rev.108, 311 (1957)

    1957

  5. [10]

    Bohr, Phys

    A. Bohr, Phys. Rev.81, 331 (1951)

    1951

  6. [11]

    C. Y. Wong, Physics Letters B41, 446 (1972)

    1972

  7. [12]

    C. Y. Wong, Annals of Physics77, 279 (1973)

    1973

  8. [13]

    Rainwater, Phys

    J. Rainwater, Phys. Rev.79, 432 (1950)

    1950

  9. [14]

    Staszczak and C

    A. Staszczak and C. Y. Wong, Physics Letters B738, 401 (2014)

    2014

  10. [16]

    C. Y. Wong, Phys. Rev. C17, 331 (1978)

    1978

  11. [18]

    X. G. Cao, E. J. Kim, K. Schmidt, K. Hagel, M. Barbui, J. Gauthier, S. Wuenschel, G. Giuliani, M. R. D. Ro- driguez, S. Kowalski, H. Zheng, M. Huang, A. Bonasera, R. Wada, N. Blando, G. Q. Zhang, C. Y. Wong, A. Staszczak, Z. X. Ren, Y. K. Wang, S. Q. Zhang, J. Meng, and J. B. ...

    2019

  12. [19]

    Z. Ren, P. Zhao, S. Zhang, and J. Meng, Nuclear Physics A996, 121696 (2020)

    2020

  13. [20]

    Natowitz and R

    J. Natowitz and R. Wada, inInterna- tional Workshop on Nuclear Dynamics in Heavy-Ion Reactions(Zhuhai, China, 2024) https://docs.google.com/presentation/d/e/ 2PACX-1vSZOTyNl3woLdLc_AMIYKJahYGcyNUsu8JO_ 7bVd-07oOr7w_MPU-lJc1o7yCQRkwEGyTrcIKGieT8i/ embed?slide=id.p1[Last Accesse...

    2024

  14. [22]

    Hannaman, A., Harvey, B., McIntosh, A.B., Hagel, K., Bills, L., Abbott, A., Fentress, A., Gauthier, J., Hankins, T., Lui, Y.-W., Bills, L., McIntosh, L.A., Regener, S., Rider, R., Schultz, S., Sorensen, M.Q., Tobar, J., Tobin, Z., and Yennello, S., EPJ Web Conf.304, 01001 (2024)

    2024

  15. [23]

    Wada, IL NUOVO CIMENTO 48 C48(2025)

    R. Wada, IL NUOVO CIMENTO 48 C48(2025)

    2025

  16. [24]

    Zheng and A

    H. Zheng and A. Bonasera, Symmetry13, 1777 (2021)

    2021

  17. [25]

    Depastas, S

    T. Depastas, S. T. Sun, H. Zheng, and A. Bonasera, Phys. Rev. C108, 035806 (2023)

    2023

  18. [26]

    Depastas, Theodoros, Sun, Shuting, He, Hongbin, Zheng, Hua, and Bonasera, Aldo, EPJ Web Conf.304, 02004 (2024)

    2024

  19. [27]

    Depastas, S

    T. Depastas, S. Sun, H. He, H. Zheng, and A. Bonasera, Physics Letters B860, 139180 (2025)

    2025

  20. [28]

    Bass, Phys

    R. Bass, Phys. Rev. Lett.39, 265 (1977)

    1977

  21. [29]

    J. D. Sanctis, M. Masotti, M. Bruno, M. D’Agostino, E. Geraci, G. Vannini, and A. Bonasera, Journal of Physics G: Nuclear and Particle Physics36, 015101 (2008)

    2008

  22. [30]

    J. J. Hopfield, Proceedings of the National Academy of Sciences79, 2554 (1982)

    1982

  23. [31]

    D. H. Ackley, G. E. Hinton, and T. J. Sejnowski, Cogni- tive Science9, 147 (1985)

    1985

  24. [32]

    Jumper, R

    J. Jumper, R. Evans, A. Pritzel, T. Green, M. Fig- urnov, O. Ronneberger, K. Tunyasuvunakool, R. Bates, A. ˇZ ´ ıdek, A. Potapenko, A. Bridgland, C. Meyer, S. Kohl, A. Ballard, A. Cowie, B. Romera-Paredes, S. Nikolov, R. Jain, J. Adler, T. Back, S. Petersen, D. Reiman, E. Clan...

    2021

  25. [33]

    Jothilakshmi and V

    S. Jothilakshmi and V. Gudivada, inCognitive Comput- ing: Theory and Applications, Handbook of Statistics, Vol. 35, edited by V. N. Gudivada, V. V. Raghavan, V. Govindaraju, and C. Rao (Elsevier, 2016) pp. 301– 340

    2016

  26. [34]

    Pedregosa, G

    F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cour- napeau, M. Brucher, M. Perrot, and E. Duchesnay, Jour- nal of Machine Learning Research12, 2825 (2011)

    2011

  27. [35]

    Arthur and S

    D. Arthur and S. Vassilvitskii, inProceedings of the Eigh- teenth Annual ACM-SIAM Symposium on Discrete Al- gorithms, SODA ’07 (Society for Industrial and Applied Mathematics, USA, 2007) p. 1027–1035

    2007

  28. [36]

    Andrae, T

    R. Andrae, T. Schulze-Hartung, and P. Melchior, (2010), arXiv:1012.3754 [astro-ph.IM]

    2010 arXiv

  29. [37]

    M. Papa, T. Maruyama, and A. Bonasera, Phys. Rev. C 64, 024612 (2001)

    2001

  30. [38]

    Bonasera, M

    A. Bonasera, M. Colonna, M. Di Toro, F. Gulminelli, and H. Wolter, Physics Letters B244, 169 (1990)

    1990

  31. [39]

    X. Chen, Y. W. Lui, H. L. Clark, Y. Tokimoto, and D. H. Youngblood, Phys. Rev. C80, 014312 (2009)

    2009

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.