pith. sign in

arxiv: 2605.21991 · v1 · pith:4Z2QHS2Mnew · submitted 2026-05-21 · 🌌 astro-ph.CO · astro-ph.IM

Machine Learning applications to Galaxy Clusters

Gustavo Yepes , Daniel de Andr\'es This is my paper

Pith reviewed 2026-05-22 05:00 UTC · model grok-4.3

classification 🌌 astro-ph.CO astro-ph.IM
keywords galaxy clustersmachine learningmass estimationcosmologysimulationsSZ effectX-raydynamical states
0
0 comments X

The pith

AI methods capture non-linear features, projection effects and complex morphologies to improve galaxy cluster mass estimates from SZ, X-ray, optical and dynamical data.

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

This review surveys how artificial intelligence techniques are applied to galaxy clusters for inferring properties from multi-wavelength observations. It shows that machine learning approaches handle non-linear relationships, projection effects and irregular shapes more effectively than traditional statistical methods in mass estimation tasks. A reader would care because cluster masses serve as key probes for cosmology and large-scale structure. The text also covers AI for emulating baryonic physics in simulations, classifying dynamical states and analyzing intracluster light, while stressing the need for reliable uncertainty estimates.

Core claim

Artificial intelligence techniques applied to SZ, X-ray, optical and dynamical data from galaxy clusters capture non-linear features, projection effects and complex morphologies, yielding mass estimates that go beyond those obtained with classical methods. Simulations supply the training data, yet the accuracy of baryonic modeling in those simulations and the generalization to real observations remain central concerns, together with the requirement for robust uncertainty quantification and interpretability.

What carries the argument

Machine learning models trained on hydrodynamical simulations to extract cluster masses and dynamical properties while incorporating projection effects and morphological complexity from multi-tracer observations.

If this is right

  • Tighter cosmological constraints from improved cluster mass functions derived from next-generation surveys.
  • More reliable identification and characterization of merging and disturbed clusters.
  • Reduced computational cost for including baryonic effects in large-volume N-body runs through emulation.
  • Better recovery of signals from diffuse components such as intracluster light.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Hybrid physics-informed neural networks could mitigate interpretability problems while retaining flexibility.
  • The same projection-handling strategies may transfer to mass estimates for groups or filaments.
  • Validation campaigns that compare models across independent simulation suites would expose hidden biases.

Load-bearing premise

The simulations used for training the models accurately reproduce baryonic physics and the selection effects present in real observational data.

What would settle it

A large sample of clusters with independent weak-lensing masses showing systematic offsets between AI predictions and lensing masses when the training simulations use different feedback prescriptions from the observed clusters.

Figures

Figures reproduced from arXiv: 2605.21991 by Daniel de Andr\'es, Gustavo Yepes.

Figure 28.1
Figure 28.1. Figure 28.1: The “bullet cluster”, galaxy cluster 1E 0657-556. The two pink clumps [PITH_FULL_IMAGE:figures/full_fig_p002_28_1.png] view at source ↗
Figure 28.2
Figure 28.2. Figure 28.2: Illustration of the sensitivity of the cluster mass function to the underlying [PITH_FULL_IMAGE:figures/full_fig_p003_28_2.png] view at source ↗
Figure 28.3
Figure 28.3. Figure 28.3: The comparison of the mass bias b as a function of the overdensity (200,500 and 2500) between the methods based on Hydrostatic Equilibrium hy￾pothesis and Autoencoder+Random Forest. Image taken from [20]. In [2], deep learning techniques were applied to real CMB data for the first time, demonstrating that these methods are not limited to idealized or proof-of￾concept studies. In this work, Convolutional… view at source ↗
Figure 28.4
Figure 28.4. Figure 28.4: The mass bias between the CNN inferred mass and the original mass [PITH_FULL_IMAGE:figures/full_fig_p009_28_4.png] view at source ↗
Figure 28.5
Figure 28.5. Figure 28.5: Comparison between GNN-inferred masses and SZ-based estimates: the [PITH_FULL_IMAGE:figures/full_fig_p011_28_5.png] view at source ↗
Figure 28.6
Figure 28.6. Figure 28.6: Mass map reconstruction from SZ, X-ray and stellar tracers. The residual [PITH_FULL_IMAGE:figures/full_fig_p013_28_6.png] view at source ↗
Figure 28.7
Figure 28.7. Figure 28.7: An example of SZ and X-ray images of merging clusters. White circles [PITH_FULL_IMAGE:figures/full_fig_p016_28_7.png] view at source ↗
Figure 28.8
Figure 28.8. Figure 28.8: Comparison of the pixel-by-pixel ICL mass fraction for two randomly [PITH_FULL_IMAGE:figures/full_fig_p016_28_8.png] view at source ↗
read the original abstract

This chapter reviews the application of Artificial Intelligence (AI) techniques to the study of galaxy clusters, covering both theoretical developments and their use as tools to infer cluster properties from a variety of observational tracers. We discuss recent advances in mass estimation from SZ, X-ray, optical, and dynamical data, highlighting the ability of AI methods to capture non-linear features, projection effects, and complex cluster morphologies beyond more classical approaches. In addition, we present other emerging applications, including the emulation of baryonic physics from N-body simulations, the characterization of dynamical states and mergers, and the analysis of the diffuse components such as the intracluster light. Particular emphasis is placed on the role of simulations in training these models, the impact of baryonic modelling, and the need for a robust uncertainty quantification and interpretability. Finally, we outline current limitations and future prospects, stressing the importance of combining flexible simulation strategies with AI techniques to fully exploit next-generation surveys for precision cosmology.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 2 minor

Summary. This chapter reviews the application of Artificial Intelligence (AI) techniques to the study of galaxy clusters, covering both theoretical developments and their use as tools to infer cluster properties from a variety of observational tracers. We discuss recent advances in mass estimation from SZ, X-ray, optical, and dynamical data, highlighting the ability of AI methods to capture non-linear features, projection effects, and complex cluster morphologies beyond more classical approaches. In addition, we present other emerging applications, including the emulation of baryonic physics from N-body simulations, the characterization of dynamical states and mergers, and the analysis of the diffuse components such as the intracluster light. Particular emphasis is placed on the role of simulations in training these models, the impact of baryonic modelling, and the need for a robust uncertainty quantification and interpretability. Finally, we outline current limitations and future prospects, stressing the importance of combining flexible simulation strategies with AI techniques to fully exploit next-generation surveys for precision cosmology.

Significance. If the summarized literature holds, this review provides a timely synthesis of how ML methods can advance galaxy cluster cosmology by addressing non-linearities and complex morphologies that classical techniques struggle with. The explicit discussion of simulation dependencies, baryonic uncertainties, and the need for interpretability and uncertainty quantification is a strength, as these are load-bearing issues for applying such methods to precision cosmology with upcoming surveys.

minor comments (2)
  1. [Abstract] Abstract: The abstract is clear but could briefly note the primary AI architectures discussed (e.g., convolutional neural networks, graph neural networks, or ensemble methods) to give readers an immediate sense of the methodological scope.
  2. [Section on limitations and future prospects] The review appropriately flags generalization and simulation fidelity as open issues; consider adding a short dedicated subsection on cross-validation strategies or domain adaptation techniques that have been proposed in the cited works to address these.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive evaluation of the review and for recommending minor revision. The referee's summary accurately captures the scope and emphasis of the manuscript on AI applications to galaxy cluster studies, including mass estimation, dynamical state characterization, and the critical role of simulations and uncertainty quantification. We appreciate the recognition of these as load-bearing issues for precision cosmology.

Circularity Check

0 steps flagged

Review paper surveying external literature; no derivations or self-referential claims

full rationale

This is a review chapter that summarizes published work on AI/ML methods for galaxy cluster mass estimation and related tasks. It explicitly frames its content as a survey of existing results from simulations and observations, highlights open issues such as baryonic modeling uncertainties and generalization biases, and does not introduce new equations, fitted parameters, predictions, or uniqueness theorems. No load-bearing step reduces by construction to the paper's own inputs or self-citations; the text instead defers to external studies while stressing limitations. The derivation chain is therefore empty and the paper is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

As this is a review paper summarizing existing research, the authors introduce no new free parameters, axioms, or invented entities. All content rests on cited prior studies.

pith-pipeline@v0.9.0 · 5686 in / 977 out tokens · 44796 ms · 2026-05-22T05:00:25.529341+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

73 extracted references · 73 canonical work pages · 4 internal anchors

  1. [1]

    Abbott T. M. C., Abdalla F. B., Allam S., Amara A., Annis J., Asorey J., et al., 2016, Monthly Notices of the Royal Astronomical Society, 460, 1270

    work page 2016

  2. [2]

    de Andres D., et al., 2022, Nature Astronomy, 6, 1325

    work page 2022

  3. [3]

    de Andres D., Yepes G., Sembolini F., Mart´ınez-Mu˜noz G., Cui W., Robledo F., Chuang C.-H., Rasia E., 2023, Mon. Not. R. Astron. Soc., 518, 111

    work page 2023

  4. [4]

    de Andres D., Cui W., Yepes G., De Petris M., Ferragamo A., De Luca F., Aver- sano G., Rennehan D., 2024, Mon. Not. R. Astron. Soc., 528, 1517

    work page 2024

  5. [5]

    E., Hahn O., 2022, Living Reviews in Computational Astrophysics, 8, 1

    Angulo R. E., Hahn O., 2022, Living Reviews in Computational Astrophysics, 8, 1

    work page 2022

  6. [6]

    Ansarifard S., et al., 2020, A&A, 634, A113

    work page 2020

  7. [7]

    R., Perrott Y

    Arendt A. R., Perrott Y . C., Contreras-Santos A., de Andres D., Cui W., Ren- nehan D., 2024, Monthly Notices of the Royal Astronomical Society, 530, 20

    work page 2024

  8. [8]

    B., Allen S

    Benyas M., Pfeifer J., Mantz A. B., Allen S. W., Darragh-Ford E., 2024, Astro- phys. J., 969, 58

    work page 2024

  9. [9]

    E., et al., 2020, Astrophys

    Bleem L. E., et al., 2020, Astrophys. J. Suppl. Ser., 247, 25

    work page 2020

  10. [10]

    B ¨ohringer H., Dolag K., Chon G., 2012, A&A, 539, A120

    work page 2012

  11. [11]

    CHEX-MATE Collaboration et al., 2021, A&A, 650, A104

    work page 2021

  12. [12]

    J., 980, 245

    Canepa L., Brough S., Lanusse F., Montes M., Hatch N., 2025, Astrophys. J., 980, 245

    work page 2025

  13. [13]

    Caro A., de Andres D., Cui W., Yepes G., De Petris M., Ferragamo A., Schiltz F., Nef A., 2025, RAS Techniques and Instruments, 4, rzaf007 28 Machine Learning applications to Galaxy Clusters 19

    work page 2025

  14. [14]

    D., Battaglia N., 2020, Mon

    Cohn J. D., Battaglia N., 2020, Mon. Not. R. Astron. Soc., 491, 1575

    work page 2020

  15. [15]

    Contreras-Santos A., Knebe A., Cui W., Haggar R., Pearce F., Gray M., De Petris M., Yepes G., 2023, Mon. Not. R. Astron. Soc., 522, 1270

    work page 2023

  16. [16]

    Cui W., et al., 2018, Mon. Not. R. Astron. Soc., 480, 2898

    work page 2018

  17. [17]

    A., Angl ´es-Alc´azar D., Yang X., 2021, Nature Astronomy, 5, 1069

    Cui W., Dav ´e R., Peacock J. A., Angl ´es-Alc´azar D., Yang X., 2021, Nature Astronomy, 5, 1069

    work page 2021

  18. [18]

    Cui W., et al., 2022, Mon. Not. R. Astron. Soc., 514, 977

    work page 2022

  19. [19]

    E., et al., 2008, Astrophys

    Evrard A. E., et al., 2008, Astrophys. J., 672, 122

    work page 2008

  20. [20]

    Ferragamo A., et al., 2023, Mon. Not. R. Astron. Soc., 520, 4000

    work page 2023

  21. [21]

    Fluri J., Kacprzak T., Lucchi A., Refregier A., Hofmann T., 2019, Physical Review D, 100, 063514

    work page 2019

  22. [22]

    P., et al., 2006, Space Science Reviews, 123, 485

    Gardner J. P., et al., 2006, Space Science Reviews, 123, 485

    work page 2006

  23. [23]

    Ghirardini V ., et al., 2024, A&A, 689, A298

    work page 2024

  24. [24]

    Gianfagna G., Rasia E., Cui W., De Petris M., Yepes G., Contreras-Santos A., Knebe A., 2023, Mon. Not. R. Astron. Soc., 518, 4238

    work page 2023

  25. [25]

    L., 2020, Astrophys

    Gupta N., Reichardt C. L., 2020, Astrophys. J., 900, 110

    work page 2020

  26. [26]

    S., 2018, Physical Review D, 97, 103515

    Gupta A., Lanusse F., Mandelbaum R., Ribli D., Li Y ., Szalay A. S., 2018, Physical Review D, 97, 103515

    work page 2018

  27. [27]

    Herbonnet R., et al., 2020, Mon. Not. R. Astron. Soc., 497, 4684

    work page 2020

  28. [28]

    D., Levasseur L

    Hezaveh Y . D., Levasseur L. P., Marshall P. J., 2017, Nature, 548, 555

    work page 2017

  29. [29]

    Hilton M., et al., 2018, Astrophys. J. Suppl. Ser., 235, 20

    work page 2018

  30. [30]

    M., Ntampaka M., Farahi A., Trac H., P´oczos B., 2019, Astro- phys

    Ho M., Rau M. M., Ntampaka M., Farahi A., Trac H., P´oczos B., 2019, Astro- phys. J., 887, 25

    work page 2019

  31. [31]

    M., Trac H., 2021, Astrophys

    Ho M., Farahi A., Rau M. M., Trac H., 2021, Astrophys. J., 908, 204

    work page 2021

  32. [32]

    M., Chen M., Lansberry A., Ruehle F., Trac H., 2022, Nature Astronomy, 6, 936

    Ho M., Ntampaka M., Rau M. M., Chen M., Lansberry A., Ruehle F., Trac H., 2022, Nature Astronomy, 6, 936

    work page 2022

  33. [33]

    Ho M., Soltis J., Farahi A., Nagai D., Evrard A., Ntampaka M., 2023, Mon. Not. R. Astron. Soc., 524, 3289

    work page 2023

  34. [34]

    Huertas-Company M., Lanusse F., 2023, Publ. Astron. Soc. Australia, 40, e001

    work page 2023

  35. [35]

    W., de Andres D., Melin J.-B., Cui W., 2025, A&A, 704, A334

    Iqbal A., Majumdar S., Rasia E., Pratt G. W., de Andres D., Melin J.-B., Cui W., 2025, A&A, 704, A334

    work page 2025

  36. [36]

    M., Tyson J

    Ivezi ´c ˇZ., Kahn S. M., Tyson J. A., Abel B., Acosta E., Allsman R., Alonso D., et al., 2019, Astrophysical Journal, 873, 111

    work page 2019

  37. [37]

    Kaiser N., 1986, Mon. Not. R. Astron. Soc., 222, 323

    work page 1986

  38. [38]

    Klypin A., Yepes G., Gottl ¨ober S., Prada F., Heß S., 2016, Mon. Not. R. As- tron. Soc., 457, 4340

    work page 2016

  39. [39]

    Kodi Ramanah D., Wojtak R., Ansari Z., Gall C., Hjorth J., 2020, Mon. Not. R. Astron. Soc., 499, 1985

    work page 2020

  40. [40]

    Kodi Ramanah D., Wojtak R., Arendse N., 2021, Mon. Not. R. Astron. Soc., 501, 4080

    work page 2021

  41. [41]

    V ., Borgani S., 2012, Annu

    Kravtsov A. V ., Borgani S., 2012, Annu. Rev. Astron. Astrophys., 50, 353

    work page 2012

  42. [42]

    Krippendorf S., et al., 2024, A&A, 682, A132

    work page 2024

  43. [43]

    Laureijs R., Amiaux J., Arduini S., Augu `eres J.-L., Brinchmann J., Cole R., Cropper M., et al., 2011, Euclid Definition Study Report (arXiv:1110.3193) 20 Gustavo Yepes and Daniel de Andr ´es

    work page internal anchor Pith review Pith/arXiv arXiv 2011

  44. [44]

    Levi M., Bebek C., Beers T., Blum R., Cahn R., Dawson K., Finkbeiner D., et al., 2016, The Dark Energy Spectroscopic Instrument (DESI) (arXiv:1611.00036)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  45. [45]

    Liu A., et al., 2022, A&A, 661, A2

    work page 2022

  46. [46]

    In: Makinson, D., Mali- nowski, J., Wansing, H

    Lovisari L., Maughan B. J., 2022, in Bambi C., Santangelo A., eds, , Handbook of X-ray and Gamma-ray Astrophysics. Springer, Singapore, doi:10.1007/978- 981-16-4544-0˙118-1

    work page doi:10.1007/978- 2022

  47. [47]

    L., Ragone-Figueroa C., Taffoni G., 2022, Mon

    Marini I., Borgani S., Saro A., Murante G., Granato G. L., Ragone-Figueroa C., Taffoni G., 2022, Mon. Not. R. Astron. Soc., 514, 3082

    work page 2022

  48. [48]

    Marini I., Saro A., Borgani S., Boi M., 2024, A&A, 689, A181

    work page 2024

  49. [49]

    Mayet F., et al., 2020, in mm Universe @ NIKA2 - Observing the mm Universe with the NIKA2 Camera. p. 00017 (arXiv:1911.03145), doi:10.1051/epjconf/202022800017

    work page doi:10.1051/epjconf/202022800017 2020

  50. [50]

    Mordvintsev A., Olah C., Tyka M., 2015, Inceptionism: Going Deeper into Neural Networks,https://research.googleblog.com/2015/ 06/inceptionism-going-deeper-into-neural.html

    work page 2015

  51. [51]

    F., Frenk C

    Navarro J. F., Frenk C. S., White S. D. M., 1995, Mon. Not. R. Astron. Soc., 275, 720

    work page 1995

  52. [52]

    J., Battaglia N., P´oczos B., Schneider J., 2015, Astrophys

    Ntampaka M., Trac H., Sutherland D. J., Battaglia N., P´oczos B., Schneider J., 2015, Astrophys. J., 803, 50

    work page 2015

  53. [53]

    J., Fromenteau S., P ´oczos B., Schneider J., 2016, Astrophys

    Ntampaka M., Trac H., Sutherland D. J., Fromenteau S., P ´oczos B., Schneider J., 2016, Astrophys. J., 831, 135

    work page 2016

  54. [54]

    J., 876, 82

    Ntampaka M., et al., 2019, Astrophys. J., 876, 82

    work page 2019

  55. [55]

    Old L., et al., 2018, Mon. Not. R. Astron. Soc., 475, 853

    work page 2018

  56. [56]

    D., Wechsler R

    Perreault Levasseur L., Hezaveh Y . D., Wechsler R. H., 2017, The Astrophys- ical Journal Letters, 850, L7

    work page 2017

  57. [57]

    Planck Collaboration et al., 2014, A&A, 571, A20

    work page 2014

  58. [58]

    Planck Collaboration et al., 2016, A&A, 594, A27

    work page 2016

  59. [59]

    B., Ntampaka M., ZuHone J., 2022, Mon

    Rothschild T., Nagai D., Aung H., Green S. B., Ntampaka M., ZuHone J., 2022, Mon. Not. R. Astron. Soc., 513, 333

    work page 2022

  60. [60]

    R., Cogswell M., Das A., Vedantam R., Parikh D., Batra D., 2020, International Journal of Computer Vision, 128, 336

    Selvaraju R. R., Cogswell M., Das A., Vedantam R., Parikh D., Batra D., 2020, International Journal of Computer Vision, 128, 336

    work page 2020

  61. [61]

    Simonyan K., Vedaldi A., Zisserman A., 2014, Deep Inside Convolu- tional Networks: Visualising Image Classification Models and Saliency Maps (arXiv:1312.6034),https://arxiv.org/abs/1312.6034

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  62. [62]

    J., Geach J

    Smith M. J., Geach J. E., 2023, Royal Society Open Science, 10, 221454

    work page 2023

  63. [63]

    Spergel D., Gehrels N., Baltay C., Bennett D., Breckinridge J., Donahue M., Dressler A., et al., 2015, Wide-Field InfrarRed Survey Telescope (WFIRST) Fi- nal Report (arXiv:1503.03757)

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  64. [64]

    B., Rasia E., Zu Y ., 2025, A&A, 700, A87

    Srivastava A., Cui W., de Andres D., Golden-Marx J. B., Rasia E., Zu Y ., 2025, A&A, 700, A87

    work page 2025

  65. [65]

    A., Zeldovich Y

    Sunyaev R. A., Zeldovich Y . B., 1972, Comments on Astrophysics and Space Physics, 4, 173

    work page 1972

  66. [66]

    J., Xiong L., P ´oczos B., Schneider J., 2012, arXiv e-prints, p

    Sutherland D. J., Xiong L., P ´oczos B., Schneider J., 2012, arXiv e-prints, p. arXiv:1202.0302 28 Machine Learning applications to Galaxy Clusters 21

    work page arXiv 2012

  67. [67]

    J., 692, 1060

    Vikhlinin A., et al., 2009, Astrophys. J., 692, 1060

    work page 2009

  68. [68]

    J., 915, 71

    Villaescusa-Navarro F., et al., 2021, Astrophys. J., 915, 71

    work page 2021

  69. [69]

    V ogelsberger M., Marinacci F., Torrey P., Puchwein E., 2020, Nature Reviews Physics, 2, 42

    work page 2020

  70. [70]

    Wadekar D., et al., 2023, Proceedings of the National Academy of Science, 120, e2202074120

    work page 2023

  71. [71]

    J., Van Waerbeke L., Hinshaw G., McCarthy I

    Yan Z., Mead A. J., Van Waerbeke L., Hinshaw G., McCarthy I. G., 2020, Mon. Not. R. Astron. Soc., 499, 3445

    work page 2020

  72. [72]

    G., et al., 2000, Astronomical Journal, 120, 1579

    York D. G., et al., 2000, Astronomical Journal, 120, 1579

    work page 2000

  73. [73]

    Zwicky F., 1933, Helvetica Physica Acta, 6, 110

    work page 1933