Pith. sign in

REVIEW 4 major objections 6 minor 56 references

Towards a topological data analysis for heavy-ion collisions

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Persistent homology is a workable, stable observable family for heavy-ion collisions, but it does not beat conventional observables in sensitivity to model parameters.

desk verdict Useful, honest TDA benchmark for heavy-ion simulations, but the 'no enhanced sensitivity' claim is built on 20 posterior draws and an asserted comparison, so it is underpowered as stated. read the letter →

arxiv 2509.02339 v1 pith:VS7CXRCC submitted 2025-09-02 nucl-th

Federica Capellino , Andrea Dubla , Silvia Masciocchi , Govert Nijs , Daniel Spitz This is my paper
classification nucl-th PACS 25.75.-q
keywords persistenthomologyheavy-ioncollisionsquark-gluonplasmaBetticurvesalphacomplexesTrajectumBayesianparameterestimationoxygen
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 establishes that persistent homology—a toolset that tracks how many connected clusters and holes appear in a point cloud as a length scale grows—can be applied end-to-end to simulated heavy-ion collisions. Working inside a full relativistic hydrodynamics model, the authors compute Betti curves and persistence distributions for the transverse-momentum point clouds of final-state hadrons in Pb-Pb and O-O collisions, with statistical and systematic uncertainties propagated from a Bayesian parameter fit. The observables reproduce known physics: centrality and mass ordering track particle multiplicities, and comparing real events with azimuthally randomized ones isolates the imprint of elliptic flow. The quantitative message is that these topological observables are practical and stable, but they do not carry more sensitivity to the model's tunable parameters than conventional observables do. The paper therefore positions persistent homology as an alternative perspective and a benchmark tool, with its real constraining power deferred until experimental measurements exist.

What carries the argument

Alpha complexes: the family of simplicial complexes formed by growing disks around each final-state hadron's transverse-momentum position; sweeping the radius r and counting connected components (Betti number beta_0) and holes (beta_1) produces Betti curves whose birth-death structure is summarized by persistence distributions. The paper's discriminating device is the randomized-azimuth counterpart beta^rand: rebuilding the same point clouds with azimuthal angles drawn uniformly from [0,2pi) and subtracting recovers the contribution of angular correlations (elliptic flow) to the topology. Sensitivity to model parameters is quantified by Pearson correlation coefficients between each observabl

What would settle it

Adopt the paper's own suggested test: generate mock Betti-curve data from Trajectum, then run Bayesian fits with and without the topological observables and compare how much the posteriors of parameters such as (zeta/s)T0 shrink. If fits including Betti curves tighten constraints beyond the baseline, the 'no enhanced sensitivity' conclusion is false; if not, it stands.

Watch

Extended reading notes

Core claim

The central claim is that topological data analysis, concretely the Betti curves beta_0(r) and beta_1(r) and the persistence distributions dN_l/dP of alpha complexes built from final-state hadron momenta, constitutes a workable and informative observable family for heavy-ion collisions. In Pb-Pb collisions at 5.02 TeV and O-O collisions at 7 TeV, these observables encode, in a single curve family, features normally separated into multiplicity, mass-ordered radial flow, and anisotropic flow; the azimuthal-randomization ratio beta^Delta/beta exposes the anisotropic-flow component. Correlating the observables with the parameters of the Trajectum Bayesian posterior, the authors find genuine corr

Load-bearing premise

The comparison that supports the paper's main negative conclusion—that topological observables are not more sensitive than conventional ones—uses Pearson correlations computed from just 20 parameter draws from the Bayesian posterior, a sample too small for correlation estimates below roughly 0.6 to be reliable.

Editorial extensions

If this is right

  • Pb-Pb and O-O Betti curves with full statistical and systematic uncertainties are now available as benchmarks for future experimental measurements at the LHC.
  • beta_0(r=0) equals the charged-hadron multiplicity, making the dimension-0 Betti curve a topological embedding of multiplicity plus clustering information.
  • The ratios beta^Delta_0/beta_0 and beta^Delta_1/beta_1 isolate anisotropic-flow effects, with stronger signals in semicentral collisions and in dimension 1, linking topology to flow phenomenology.
  • The correlation between dN/dP_1 and the bulk-viscosity peak temperature (zeta/s)T0 suggests Betti-curve data could help constrain this fluid parameter if the correlation survives larger statistics.
  • The absence of enhanced sensitivity relative to standard observables means the near-term role of these observables is cross-validation and model benchmarking rather than parameter constraint.

Reading between the lines

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

  • If experimental Betti curves become available and disagree with Trajectum, the same observables would switch roles: from parameter constraint to model-discrimination diagnostics, potentially revealing 3D collective structure that n-point correlators miss.
  • The paper's correlation analysis uses 20 posterior draws; with that sample size, Pearson coefficients below roughly 0.6 are only marginally significant, so the negative sensitivity result should be re-checked with a larger posterior sample or a mock-data Bayesian fit before being treated as definitive.
  • Extending the point cloud from the transverse plane to include longitudinal momentum—a future direction the paper names—is a natural test: TDA sensitivity to long-range 3D correlations could differ from the transverse-only case studied here.
  • Because the paper's dictionary ties Betti curves to known observables, the same pipeline could be ported to smaller collision systems such as Ne-Ne or to high-multiplicity proton-proton events, where conventional flow analyses are contested.
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 / 6 minor

Summary. The paper implements persistent homology observables for alpha complexes built from transverse-momentum point clouds in the Trajectum hydrodynamic model, presenting Betti curves and persistence distributions for Pb–Pb at 5.02 TeV and O–O at 7 TeV across several centralities and particle species. The model parameters are taken as 20 random posterior draws from an earlier Bayesian analysis [1], which allows the authors to quote statistical and systematic uncertainties. A dictionary is constructed linking the topological observables to multiplicity, mass ordering from radial flow, and elliptic-flow-induced azimuthal correlations, using azimuthal randomization to isolate the latter. The paper's headline conclusion is that persistent homology observables are viable and robust but do not show enhanced sensitivity to model parameters compared with conventional observables. This conclusion is based on Pearson correlation coefficients computed from the 20 posterior draws and on a qualitative comparison with a previous correlation study [46].

Significance. If the main claims are established, the paper would provide a useful benchmark for TDA in heavy-ion phenomenology: it demonstrates that Betti curves and persistence distributions can be computed in a state-of-the-art event generator, that the results are stable under posterior parameter uncertainty, and that they encode familiar physics (multiplicity ordering, mass ordering from radial flow, and elliptic-flow effects). The O–O predictions and the explicit systematic-error treatment are valuable for future experimental comparisons. The authors are also careful to note that no persistent-homology observable entered the Bayesian fit of Ref. [1], so the predictions are genuine extrapolations rather than in-sample fits. However, the central negative claim—that TDA observables do not show enhanced parameter sensitivity—is not established with the statistical rigor currently presented, because it rests on a very small number of posterior draws and an unquantified comparison to Ref. [46].

major comments (4)
  1. [Sec. IV.D and Sec. III] The Pearson coefficients in Fig. 8 are computed from only n=20 posterior parameter draws, as stated in Sec. III. For n=20, the sampling uncertainty of r under the null hypothesis is approximately 1/sqrt(17) ≈ 0.24, so most of the reported values in the range (-0.6, 0.6) are within roughly 2.5 standard errors of zero. No confidence intervals, p-values, permutation nulls, or bootstrap estimates are given. This is load-bearing because Sec. V uses these correlations to conclude that the persistent-homology observables 'do not show enhanced sensitivity ... compared to conventional observables.' With n=20, the analysis cannot currently distinguish a genuine absence of enhanced sensitivity from insufficient sampling of the posterior. Please provide uncertainty estimates for the correlation coefficients, or increase the number of posterior draws, or both.
  2. [Sec. V, comparison to [46]] The statement 'most of them have not been larger than the correlations observed for standard observables [46]' is not supported by any quantitative side-by-side comparison in this manuscript. Ref. [46] may involve different observables, a different parameter set, and possibly different posterior statistics, so the comparison is not apples-to-apples. To make the headline claim robust, the authors should compute Pearson (or rank) correlations for the same set of conventional observables using the same 20 posterior draws and present the two sets of coefficients in a common table or figure. Without this, the 'no enhanced sensitivity' conclusion is an assertion rather than a demonstrated result.
  3. [Sec. IV.D, selection of homology radii] The correlation analysis is performed at radii chosen by 'qualitatively looking at the spread of the 20 individual calculations ... and selecting the interval in which the spread is the widest.' This data-driven selection, combined with scanning 25 (or 23) parameters and several observables, introduces a multiple-comparisons bias: large |rho| values are more likely to be found when both the radius and the parameter axis are scanned. The same issue applies to the persistence-window choice at P = 0.10 ± 0.04 GeV/c. The authors should either report the correlations as a function of r over a dense grid, or account for the selection effect, or present the analysis as exploratory rather than as evidence for the absence of enhanced sensitivity.
  4. [Sec. IV.D, Figs. 9 and 10] Pearson's r measures only linear association and is highly sensitive to outliers. Fig. 9 explicitly demonstrates this: a strong Pearson correlation between nc and beta0 is driven by outliers, with no visible monotonic ordering in nc. Yet the later comparison with standard observables and the discussion of 'strong correlations' (e.g., with (zeta/s)_T0 in Fig. 10) continue to rely on Pearson coefficients without a linearity diagnostic or robust alternative. I recommend adding rank-based correlation measures (e.g., Spearman rho), scatter plots for the parameters highlighted in the text, and a statement of how many of the apparent correlations survive after removing outliers or using robust estimators.
minor comments (6)
  1. [Sec. III] The sentence 'By performing around 400k calculations for each of these different choices' is ambiguous: does 'calculations' mean simulated events, hydrodynamic runs, or individual observable computations? Please clarify the computational cost and the number of events per posterior draw.
  2. [Sec. IV.C, Fig. 7] The text refers to 'saddle points in the distributions dN0/dP and dN0^rand/dP' when discussing the dimension-1 persistence distributions shown in Fig. 7. The subscripts appear to be typos and should be dN1/dP and dN1^rand/dP.
  3. [Sec. I and Sec. V] The introduction and conclusions mention 'n-point connected correlation functions' as traditional observables, but the dictionary built in this paper mainly connects Betti curves to multiplicity, radial flow, and elliptic flow. The connection to n-point connected correlators is not made explicit; either remove the phrase or spell out the correspondence.
  4. [Fig. 2] The left and right panels use different vertical-axis ranges and different collision energies, which makes direct visual comparison of Pb–Pb and O–O somewhat difficult. Consider adding an inset or a ratio panel if this comparison is central.
  5. [Appendix C] The list of parameters is clear, but there is an inconsistency in the text: the main text says 26 parameters in the Bayesian analysis, while Appendix C explains that 25 are varied for Pb–Pb and 23 for O–O because Norm is doubled and Woods-Saxon parameters are absent for O–O. Please make this count explicit at the first mention in Sec. III to avoid confusion.
  6. [Various] There are several minor typographical issues, including missing spaces before 'Cech' in the footnote of Sec. II A, inconsistent notation for beta^rand_l, and the use of 'an' as a parameter name in Appendix C, which is easily misread as the English article. A careful proofreading pass would help.

Circularity Check

0 steps flagged · score 1.0 of 10

No circularity: TDA observables are out-of-sample predictions; the self-cited model and comparison papers are externally benchmarked, and the only definitional identifications are explicit dictionary entries.

full rationale

The derivation chain is self-contained and non-circular. The TDA observables (Betti curves, persistence distributions, angular-randomization ratios) are computed from Trajectum events at 20 posterior draws from the Bayesian fit of Ref. [1]; the paper states explicitly that 'Crucially for this work, no persistent homology observables were included' in that fit (Sec. III), so the Pb–Pb and O–O predictions are genuine out-of-sample extrapolations and the quoted systematic uncertainties are honestly propagated posterior spreads. The only definitional identification is explicit dictionary building: 'β0(r) at radius r = 0 GeV/c can be identified with the number of produced particles' (Sec. IV.A), used to interpret curves, not to derive a fitted quantity. The central negative claim ('do not show enhanced sensitivity to the model's tunable parameters compared to conventional observables') rests on Pearson correlations from n=20 draws (Sec. IV.D, Eq. 2) compared with published correlations in Ref. [46]; Refs. [1] and [46] share an author with the present paper (G. Nijs), but both are peer-reviewed, externally falsifiable results whose fitted observables exclude the TDA quantities, so under the review rules these citations are real evidence and do not raise the circularity score. The paper itself flags the relevant limitations: the correlation window is chosen 'by qualitatively looking at the spread of the 20 individual calculations' (Sec. IV.D); some correlations are demonstrated to be spurious (the nc case, Fig. 9); and Sec. V concedes that constraining power 'can only be fully assessed through experimental measurements' and that a mock-data Bayesian closure test is 'left for future work.' Small-n statistics and the unquantified side-by-side with Ref. [46] are robustness/correctness risks, not definitional reductions: no observable was fitted to the target output, and no equation reduces to its own input. Score 1 reflects only the minor self-citation overlap, which is not itself circular.

Assumptions & free parameters 3 free parameters · 7 assumptions · 0 invented entities

The predictions and uncertainty quantification are built entirely on the Trajectum model and the posterior parameter ensemble from [1], a self-cited Bayesian fit to Pb-Pb data. No new particles, forces, dimensions, or conserved quantities are introduced. The analysis windows for correlations are chosen ad hoc from the widest spread. Standard topological facts (stability, structure theorem, Cech-alpha equivalence) are sourced from the literature.

free parameters (3)
  • Homology radius evaluation window = r = 0.47 +/- 0.04 GeV/c
    Chosen post hoc as the interval with the widest spread across the 20 posterior runs (Sec. IV.D); all reported Pearson correlations depend on this choice.
  • Persistence evaluation window = P = 0.10 +/- 0.04 GeV/c
    Same post hoc selection for the persistence distributions (Sec. IV.D).
  • Trajectum model parameters = 20 posterior samples from [1]
    The predictions and their systematic uncertainties are inherited from the Pb-Pb Bayesian fit in [1]; key examples are Norm, Tswitch, eta/s, and (zeta/s)_T0. These are fitted values in prior work used as inputs here.
assumptions (7)
  • standard math Alpha complexes and Cech complexes share persistent homology under general conditions
    Sec. II.A footnote, cited to [33]; needed to justify using the 2-skeleton of Cech complexes as alpha complexes.
  • standard math Persistence module structure theorem: tame persistence modules are isomorphic to persistence diagrams
    Appendix A.3, refs [49,50]; needed to interpret birth-death pairs as the observable.
  • standard math Persistent homology is stable under perturbation of the input point cloud
    Sec. II.B, cited [38,39]; justifies the robustness claims.
  • domain assumption The Trajectum model with posterior parameters from the Pb-Pb fit [1] remains valid when extrapolated to O-O at sqrt(s_NN)=7 TeV with NLEFT nuclear configurations
    Sec. III and IV; all O-O predictions and their systematic uncertainties rest on this transfer without refitting.
  • domain assumption Randomizing azimuthal angles produces beta^rand, whose difference from beta isolates the effect of angular (flow) correlations
    Sec. IV.B, Eq. (1); the dictionary between beta^Delta/beta and elliptic flow assumes randomization removes only angular correlations while leaving radial structure intact.
  • domain assumption 20 posterior samples are sufficient to estimate Pearson correlations and the systematic uncertainty via variance additivity
    Sec. III and Sec. IV.D; the sensitivity and negative conclusions rest on n=20 and on sigma_syst = sqrt(sigma_tot^2 - sigma_stat^2).
  • ad hoc to paper The homology radius and persistence windows chosen by widest spread are representative of the observable's information content
    Sec. IV.D: 'selecting the interval in which the spread is the widest'.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards a topological data analysis for heavy-ion collisions." pith.science (2026). https://pith.science/paper/VS7CXRCC

@misc{pith2026250902339,
  author       = {Pith},
  title        = {Pith review of: Towards a topological data analysis for heavy-ion collisions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VS7CXRCC}},
  note         = {Machine review of arXiv:2509.02339}
}
read the original abstract

The collective expansion of the quark-gluon plasma (QGP) created in heavy-ion collisions suggests that geometry-inspired approaches can be useful in extracting information about the QGP. In this work, a systematic study of observables based on topological data analysis is provided for simulations of heavy-ion collisions. Specifically, we implement persistent homology observables for metric-based complexes in the heavy-ion model Trajectum and provide predictions for Pb-Pb and O-O collisions, where the tunable model parameters are taken from a Bayesian analysis performed in Pb-Pb collisions. This, in particular, allows us to compute systematic uncertainties on our observables from the uncertainties in the model parameters. To bridge between new and already established observables, we build a dictionary linking the topological observables to traditional ones, such as particle multiplicities, momentum distributions, and the elliptic flow coefficient. While the persistent homology observables largely reflect known phenomenology and do not show enhanced sensitivity to the model's tunable parameters compared to conventional observables, this study demonstrates the viability and robustness of topological techniques in the context of heavy-ion physics. They may offer alternative perspectives and potential applications in heavy-ion physics.

Figures

Figures reproduced from arXiv: 2509.02339 by the authors.

Figure 1
Figure 1. FIG. 1. Alpha complexes of a point cloud given by transverse momentum space ( [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Dimension-0 and 1 Betti curves for Pb–Pb (left) and O–O collisions (right) at [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Dimension-0 Betti curves for pions, kaons, (anti)protons and all charged hadrons in the 0–5% (left panel) and 30–40% [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Dimension-0 Betti curves for pions, kaons, [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Dimension-1 Betti curves for pions, kaons, (anti)protons and all charged hadrons in the 0–5% centrality class. The [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Dimension-0 persistence distributions for pions, kaons, (anti)protons and all charged hadrons in the 0–5% centrality [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Dimension-1 persistence distributions for pions, kaons, (anti)protons and all charged hadrons in the 0–5% centrality [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8. Pearson correlation coefficients of simulation parameters with different persistent homology observables for central [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: FIG. 9. The dimension-0 Betti curve, [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: FIG. 10. The curve d [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]
Figure 11
Figure 11. Figure 11: FIG. 11. Pearson correlator of simulation parameters and persistent homology observables for semicentral Pb–Pb collisions [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

56 extracted references · 29 canonical work pages

  1. [46]

    Nijs and W

    G. Nijs and W. van der Schee, (2023), arXiv:2304.06191 [nucl-th]

    arXiv 2023

  2. [1]

    Simplicial and chain complexes An ℓ-simplex σ is a set of ℓ+1 points, σ = {x0, . . . , xℓ}. We call a subset of σ consisting of ℓ points a face of σ. For instance, if σ is a triangle, then all its edges are faces of σ. A set C of simplices is called a simplicial complex, if (i ) for all simplices in C all their faces are also included in C, and ( ii ) for...

    work page

  3. [2]

    For instance, a path along edges, i.e., a connected collection of edges, is character- ized as closed, if it has an empty boundary

    Homology groups In order to finally be able to define homology as pro- vided by homology groups, we need to have a notion of closed paths of simplices. For instance, a path along edges, i.e., a connected collection of edges, is character- ized as closed, if it has an empty boundary. If the path would not be closed, it would come with boundary points. Form...

    work page

  4. [3]

    Persistent homology At least in brevity, we now turn to persistent homology groups, whose mathematical constructions are a bit more abstract and involved than those leading to the homol- ogy groups. Let {Cr}r≥0 be a filtration of simplicial com- plexes, i.e., a nested sequence of simplicial complexes so that for all r ≤ s we have Cr ⊆ Cs, for instance the...

    work page

  5. [4]

    Giacalone, G

    G. Giacalone, G. Nijs, and W. van der Schee, Phys. Rev. Lett. 131, 202302 (2023), arXiv:2305.00015 [nucl-th]

    arXiv 2023

  6. [5]

    Busza, K

    W. Busza, K. Rajagopal, and W. van der Schee, Ann. Rev. Nucl. Part. Sci. 68, 339 (2018), arXiv:1802.04801 [hep-ph]

    arXiv 2018

  7. [6]

    Acharya et al

    S. Acharya et al. (ALICE), Eur. Phys. J. C 84, 813 (2024), arXiv:2211.04384 [nucl-ex]

    arXiv 2024

  8. [7]

    Adams et al

    J. Adams et al. (STAR), Nucl. Phys. A 757, 102 (2005), arXiv:nucl-ex/0501009

    arXiv 2005

Show all 56 references
  1. [8]

    Adcox et al

    K. Adcox et al. (PHENIX), Nucl. Phys. A 757, 184 (2005), arXiv:nucl-ex/0410003

    2005 arXiv

  2. [9]

    G. Nijs, W. van der Schee, U. G¨ ursoy, and R. Snellings, Phys. Rev. C 103, 054909 (2021), arXiv:2010.15134 [nucl-th]. 15

    2021 arXiv

  3. [10]

    G. Nijs, W. van der Schee, U. G¨ ursoy, and R. Snellings, Phys. Rev. Lett. 126, 202301 (2021), arXiv:2010.15130 [nucl-th]

    2021 arXiv

  4. [11]

    Schenke, C

    B. Schenke, C. Shen, and P. Tribedy, Phys. Rev. C 102, 044905 (2020), arXiv:2005.14682 [nucl-th]

    2020 arXiv

  5. [12]

    Vermunt, Y

    L. Vermunt, Y. Seemann, A. Dubla, S. Floerchinger, E. Grossi, A. Kirchner, S. Masciocchi, and I. Se- lyuzhenkov, Phys. Rev. C 108, 064908 (2023), arXiv:2308.16722 [hep-ph]

    2023 arXiv

  6. [13]

    P. Lu, R. Kavak, A. Dubla, S. Masciocchi, and I. Se- lyuzhenkov, (2024), arXiv:2407.09207 [hep-ph]

    2024 arXiv

  7. [14]

    M. R. Heffernan, C. Gale, S. Jeon, and J.-F. Pa- quet, Phys. Rev. C 109, 065207 (2024), arXiv:2302.09478 [nucl-th]

    2024 arXiv

  8. [15]

    M. R. Heffernan, C. Gale, S. Jeon, and J.-F. Paquet, Phys. Rev. Lett. 132, 252301 (2024), arXiv:2306.09619 [nucl-th]

    2024 arXiv

  9. [16]

    Everett et al

    D. Everett et al. (JETSCAPE), Phys. Rev. Lett. 126, 242301 (2021), arXiv:2010.03928 [hep-ph]

    2021 arXiv

  10. [17]

    J. E. Parkkila, A. Onnerstad, S. F. Taghavi, C. Mor- dasini, A. Bilandzic, M. Virta, and D. J. Kim, Phys. Lett. B 835, 137485 (2022), arXiv:2111.08145 [hep-ph]

    2022 arXiv

  11. [18]

    Virta, J

    M. Virta, J. Parkkila, and D. J. Kim, Phys. Rev. C 111, 044903 (2025), arXiv:2411.01932 [hep-ph]

    2025 arXiv

  12. [19]

    Giacalone et al., Phys

    G. Giacalone et al., Phys. Rev. Lett. 134, 082301 (2025), arXiv:2405.20210 [nucl-th]

    2025 arXiv

  13. [20]

    Giacalone et al

    G. Giacalone et al. , (2024), arXiv:2402.05995 [nucl-th]

    2024 arXiv

  14. [21]

    Brewer, A

    J. Brewer, A. Mazeliauskas, and W. van der Schee, in Opportunities of OO and pO collisions at the LHC (2021) arXiv:2103.01939 [hep-ph]

    2021 arXiv

  15. [22]

    Otter, M

    N. Otter, M. A. Porter, U. Tillmann, P. Grindrod, and H. A. Harrington, EPJ Data Sci. 6, 1 (2017), arXiv:1506.08903 [math.AT]

    2017 arXiv

  16. [23]

    Chazal and B

    F. Chazal and B. Michel, Front. Artif. Intell. 4 (2021)

    2021

  17. [24]

    Hamilton, T

    G. Hamilton, T. Dore, and C. Plumberg, Phys. Rev. C 106, 064912 (2022), arXiv:2209.15480 [nucl-th]

    2022 arXiv

  18. [25]

    R. Wang, C. Qiu, C.-S. Hu, Z. Li, and Y. Wu, (2024), arXiv:2412.06151 [nucl-th]

    2024 arXiv

  19. [26]

    Thomas, S

    D. Thomas, S. Demers, S. Krishnaswamy, and B. Rieck, in Machine Learning and the Physical Sciences workshop, NeurIPS 2022 (2022)

    2022

  20. [27]

    Gaertner and J

    T. Gaertner and J. Reiten, (2023), arXiv:2312.06596 [hep-ph]

    2023 arXiv

  21. [28]

    Spitz, K

    D. Spitz, K. Boguslavski, and J. Berges, Phys. Rev. D 108, 056016 (2023), arXiv:2303.08618 [hep-ph]

    2023 arXiv

  22. [29]

    Spitz, J

    D. Spitz, J. M. Urban, and J. M. Pawlowski, Phys. Rev. D 107, 034506 (2023), arXiv:2208.03955 [hep-lat]

    2023 arXiv

  23. [30]

    Spitz, J

    D. Spitz, J. M. Urban, and J. M. Pawlowski, (2024), arXiv:2412.09112 [hep-lat]

    2024 arXiv

  24. [31]

    Sehayek and R

    D. Sehayek and R. G. Melko, Phys. Rev. B 106, 085111 (2022), arXiv:2201.09856 [cond-mat.stat-mech]

    2022 arXiv

  25. [32]

    N. Sale, B. Lucini, and J. Giansiracusa, Phys. Rev. D 107, 034501 (2023), arXiv:2207.13392 [hep-lat]

    2023 arXiv

  26. [33]

    Crean, J

    X. Crean, J. Giansiracusa, and B. Lucini, SciPost Phys. 17, 100 (2024), arXiv:2403.07739 [hep-lat]

    2024 arXiv

  27. [34]

    Edelsbrunner, D

    H. Edelsbrunner, D. Kirkpatrick, and R. Seidel, IEEE Trans. Inf. Theory 29, 551 (1983)

    1983

  28. [35]

    Edelsbrunner and E

    H. Edelsbrunner and E. P. M¨ ucke, ACM Trans. Graph. 13, 43 (1994)

    1994

  29. [36]

    Bauer and H

    U. Bauer and H. Edelsbrunner, Trans. Amer. Math. Soc. 369, 3741 (2017)

    2017

  30. [37]

    Hensel, M

    F. Hensel, M. Moor, and B. Rieck, Front. Artif. Intell. 4, 681108 (2021)

    2021

  31. [38]

    Bobrowski and P

    O. Bobrowski and P. Skraba, Sci. Rep. 13, 12274 (2023)

    2023

  32. [39]

    Bobrowski and P

    O. Bobrowski and P. Skraba, (2024), arXiv:2406.05553 [math.PR]

    2024 arXiv

  33. [40]

    Maria, J.-D

    C. Maria, J.-D. Boissonnat, M. Glisse, and M. Yvinec, in Mathematical Software – ICMS 2014 , edited by H. Hong and C. Yap (Springer Berlin Heidelberg, Berlin, Heidel- berg, 2014) pp. 167–174

    2014

  34. [41]

    Cohen-Steiner, H

    D. Cohen-Steiner, H. Edelsbrunner, and J. Harer, Dis- crete Comput. Geom. 37, 103 (2007)

    2007

  35. [42]

    Cohen-Steiner, H

    D. Cohen-Steiner, H. Edelsbrunner, J. Harer, and Y. Mi- leyko, Found. Comput. Math. 10, 127 (2010)

    2010

  36. [43]

    Hiraoka, T

    Y. Hiraoka, T. Shirai, and K. D. Trinh, Ann. Appl. Prob. 28, 2740 (2018), arXiv:1612.08371 [math]

    2018 arXiv

  37. [44]

    Spitz and A

    D. Spitz and A. Wienhard, arXiv:2012.05751 [math.PR]

    2012 arXiv

  38. [45]

    J. S. Moreland, J. E. Bernhard, and S. A. Bass, Phys. Rev. C 92, 011901 (2015), arXiv:1412.4708 [nucl-th]

    2015 arXiv

  39. [48]

    smash-transport/smash: Smash-1.8,

    D. Oliinychenko, V. Steinberg, J. Weil, M. Kretz, J. Stau- denmaier, S. Ryu, A. Sch¨ afer, J. Rothermel, J. Mohs, F. Li, H. E. (Petersen), L. Pang, D. Mitrovic, A. Gold- schmidt, L. Geiger, J.-B. Rose, J. Hammelmann, and L. Prinz, “smash-transport/smash: Smash-1.8,” (2020)

    2020

  40. [49]

    Nijs and W

    G. Nijs and W. van der Schee, Phys. Rev. C 106, 044903 (2022), arXiv:2110.13153 [nucl-th]

    2022 arXiv

  41. [50]

    Hatcher, Algebraic topology (”Cambridge University Press”, ”Cambridge, England”, 2002)

    A. Hatcher, Algebraic topology (”Cambridge University Press”, ”Cambridge, England”, 2002)

    2002

  42. [51]

    J. R. Munkres, Elements of algebraic topology (CRC press, Boca Raton, Florida, 2018)

    2018

  43. [52]

    Edelsbrunner, D

    H. Edelsbrunner, D. Letscher, and A. Zomorodian, in Proceedings 41st Annual Symposium on Foundations of Computer Science (2000) pp. 454–463

    2000

  44. [53]

    Zomorodian and G

    A. Zomorodian and G. Carlsson, in Proceedings of the 20th Annual Symposium on Computational Geometry , SCG ’04 (Association for Computing Machinery, New York, NY, 2004) p. 347–356

    2004

  45. [54]

    J. S. Moreland, J. E. Bernhard, and S. A. Bass, Phys. Rev. C 101, 024911 (2020), arXiv:1808.02106 [nucl-th]

    2020 arXiv

  46. [55]

    J. E. Bernhard, J. S. Moreland, S. A. Bass, J. Liu, and U. Heinz, Phys. Rev. C 94, 024907 (2016), arXiv:1605.03954 [nucl-th]

    2016 arXiv

  47. [56]

    Weil et al

    J. Weil et al. (SMASH), Phys. Rev. C 94, 054905 (2016), arXiv:1606.06642 [nucl-th]

    2016 arXiv

  48. [57]

    Sjostrand, S

    T. Sjostrand, S. Mrenna, and P. Z. Skands, Comput. Phys. Commun. 178, 852 (2008), arXiv:0710.3820 [hep- ph]

    2008 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.