arxiv: 2605.06400 · v1 · submitted 2026-05-07 · 🌌 astro-ph.CO

Recognition: unknown

Galaxy clusters in the LoTSS-DR3: Catalogues and detection pipeline for diffuse radio emission

C. Stuardi , G. Di Gennaro , A. Botteon , F. Braga , C. Gheller , F. Vazza , M. Balboni , N. Biava

show 14 more authors

A. Bonafede M. Br\"uggen G. Brunetti R. Cassano M. Cianfaglione V. Cuciti F. De Gasperin F. Gastaldello M.J. Hardcastle M. Hoeft H.J.A. Rottgering N. Sanvitale T. W. Shimwell R.J. van Weeren

Authors on Pith no claims yet

Pith reviewed 2026-05-08 05:33 UTC · model grok-4.3

classification 🌌 astro-ph.CO

keywords galaxy clustersdiffuse radio emissionLOFARLoTSSconvolutional neural networkimage segmentationradio haloscluster catalogues

0 comments

The pith

A convolutional neural network pipeline automatically detects diffuse radio emission in 3822 galaxy clusters from the LoTSS-DR3 survey.

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

The paper develops an automated pipeline to find galaxy clusters that host diffuse radio emission using deep learning on large radio survey images. Radio U-Net performs pixel-level segmentation to flag the emission and then matches the results to positions and properties from existing X-ray and Sunyaev-Zeldovich cluster catalogues. This produces both a probability score for each cluster and shape maps that can guide follow-up work. A demonstration extracts 357 high-confidence detections at 76 percent network accuracy and confirms that the overall detection rate rises with cluster mass and redshift. The approach replaces manual visual checks that cannot scale to current and future survey volumes.

Core claim

We employed Radio U-Net, a convolutional neural network optimised for image segmentation of diffuse radio emission. To associate detected emission with individual clusters, we combined the network output with positional, mass, and redshift information from four X-ray- and Sunyaev-Zeldovich-selected cluster catalogues, resulting in a merged sample of 3822 clusters covered by the LoTSS-DR3. We produced a pixel-level segmentation map of the full LoTSS-DR3 and a quantitative indicator for the presence of diffuse emission in each cluster. As a demonstration, we identified a sub-sample of 357 clusters selected at the highest network accuracy (76 percent) and verified that the detection fraction of

What carries the argument

Radio U-Net, a convolutional neural network for pixel-level segmentation of diffuse radio emission in 144 MHz images, which generates maps that are then cross-matched with cluster catalogues to yield detection probabilities and morphological information.

Load-bearing premise

The neural network trained on a labeled subset of images accurately identifies genuine diffuse radio emission across the full survey without significant contamination from noise, artifacts, or unrelated sources.

What would settle it

If visual inspection or targeted follow-up of the 357 highest-accuracy candidates reveals that more than about 20 percent are false positives from non-cluster sources, the pipeline's claimed reliability would not hold.

Figures

Figures reproduced from arXiv: 2605.06400 by A. Bonafede, A. Botteon, C. Gheller, C. Stuardi, F. Braga, F. De Gasperin, F. Gastaldello, F. Vazza, G. Brunetti, G. Di Gennaro, H.J.A. Rottgering, M. Balboni, M. Br\"uggen, M. Cianfaglione, M. Hoeft, M.J. Hardcastle, N. Biava, N. Sanvitale, R. Cassano, R.J. van Weeren, T. W. Shimwell, V. Cuciti.

**Figure 1.** Figure 1: Mass versus redshift plot of the 3822 galaxy clusters in the LoTSS-DR3 area after the merging of the four catalogues (ACT-DR5, 1eRASS, MCXC2, PSZ2). Following our selection criteria, only clusters with a mass above 1014 M⊙ were considered. In Appendix A, we show the plot of the individual catalogues before the merging. 2024). This catalogue contains 1,3341 clusters in the full sky and spans the redshift r… view at source ↗

**Figure 2.** Figure 2: Position of the 3822 galaxy clusters in the merged catalogue described in Sec. 2 overlaid on the Global Sky Model at 144 MHz derived by Zheng et al. (2017). The colour and shape of the markers represent different galaxy cluster catalogues, as explained by the legend. The black contours show the LoTSS-DR3 footprint. 3.1. The LOFAR Two-metre Sky Survey third data release The LoTSS is a wide-area 120-168 MHz … view at source ↗

**Figure 3.** Figure 3: Representative example of quality 2 (top and bottom left panels), quality 1 (top and bottom central panels), and quality 0 (top and bottom right panels) mosaics. The top row shows the LoTSS-DR3 mosaics with a common colour-scale. The bottom row shows the corresponding segmented maps produced by Radio U-Net. While quality class 2 mosaics show clear large-scale patterns, quality class 1 mosaics are character… view at source ↗

**Figure 4.** Figure 4: Position of the 2551 pointings of the LoTSS-DR3 overlaid on the Global Sky Model at 144 MHz derived by Zheng et al. (2017). The colour and shape of the markers represent the quality class given in Sec. 3.3, as explained by the legend. In particular, red empty diamonds are quality 2 pointings, which were excluded by this analysis mainly due to the contamination of the Milky Way emission. Black lines show th… view at source ↗

**Figure 5.** Figure 5: Binary classification metrics for the test sample of 246 LoTSSDR2/PSZ2 galaxy clusters (Botteon et al. 2022) as a function of the value of R used as a threshold, considering its uncertainties. threshold Rt are E[TP(Rt)] = X i yi pi(Rt), E[FP(Rt)] = X i (1 − yi) pi(Rt), E[TN(Rt)] = X i (1 − yi) (1 − pi(Rt)), E[FN(Rt)] = X i yi (1 − pi(Rt)). From these, we computed the expected accuracy (fraction of correct… view at source ↗

**Figure 6.** Figure 6: Confusion matrix obtained for the test sample of 174 galaxy clusters with a good classification flag using R ∗ t = 0.021 as a detection threshold. The number in brackets is the total number in each box. considered reliable if its measured R value lies more than one standard deviation from the threshold: |Ri − R∗ t | > σi . (2) This approach ensures that both threshold selection and individual measurement … view at source ↗

**Figure 7.** Figure 7: Distribution of the R values computed for the 3367 galaxy clusters in the merged catalogue outside quality 2 regions. The vertical line represents the value R ∗ t = 0.021. the position of galaxy clusters contained in the merged catalogue created in Sec. 2. Therefore, as explained in Sec. 3.4, for each of the 3822 galaxy clusters in our merged catalogue, we found the reference pointing and, for those in qu… view at source ↗

**Figure 8.** Figure 8: Intrinsic fraction of clusters with diffuse radio emission over the total number of clusters in each mass and redshift bin in the PSZ2 cluster catalogue computed following Eq. 3. The total number of clusters in each bin (N0+N1) is also shown in brackets. However, this number must be corrected for the classification performance of our detection pipeline. As discussed in Sec. 3.4, the threshold R ∗ t = 0.02… view at source ↗

read the original abstract

The third data release of the LOFAR Two-metre Sky Survey provides an unprecedented view of the northern sky at 144 MHz. While compact sources can be efficiently identified with automated software packages, the detection of diffuse radio emission associated with galaxy clusters still requires dedicated processing and visual inspection. Given the scale of current and forthcoming radio surveys, automated approaches based on artificial intelligence are becoming essential to the identification of the most interesting targets. We aim to develop an automated pipeline to construct a catalogue of galaxy clusters hosting diffuse radio emission from LoTSS-DR3 20arcsec images. The pipeline is designed to provide both the probability that a cluster hosts diffuse radio emission and an interpretable image of its shape and morphology. We employed Radio U-Net, a convolutional neural network optimised for image segmentation (i.e. pixel-level identification) of diffuse radio emission. To associate detected emission with individual clusters, we combined the network output with positional, mass, and redshift information from four X-ray- and Sunyaev-Zeldovich-selected cluster catalogues, resulting in a merged sample of 3822 clusters covered by the LoTSS-DR3. We produced a pixel-level segmentation map of the full LoTSS-DR3 and a quantitative indicator for the presence of diffuse emission in each cluster. This enables the selection of sub-samples with specific properties for targeted follow-up or statistical studies. As a demonstration of the first application, we identified a sub-sample of 357 clusters selected at the highest network accuracy (76%), and we showed some examples of newly detected systems. For the second, using a larger statistical sample, we verified that the detection fraction of diffuse radio sources in the four catalogues increases with the mass and redshift of the clusters. [Abridged]

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.

Desk Editor's Note private letter to a colleague

The paper delivers a U-Net pipeline that produces probability scores and segmentation maps for diffuse emission across 3822 LoTSS-DR3 clusters, with a 357-cluster high-accuracy subset and a check on mass/redshift trends.

read the letter

The core output is a practical pipeline that runs Radio U-Net on the 20-arcsec LoTSS-DR3 images, then folds in positions, masses, and redshifts from four external cluster catalogs to assign a detection probability and a pixel map to each of 3822 systems. They pull out 357 clusters at the top end of the network score and show that the fraction of detections rises with cluster mass and redshift, which matches what is already expected from the literature on radio halos and relics. That trend check is a minimal but useful sanity test. The segmentation maps are also a step up from simple yes/no flags because they let users see the morphology directly. The work is aimed squarely at people who need to sift large radio surveys for extended cluster emission without eyeballing every field. For that audience the catalog and the code-like pipeline are the main value. The soft spot is the validation. The abstract gives 76% network accuracy on the selected sub-sample but does not spell out whether that number is pixel accuracy, intersection-over-union, or precision at a fixed threshold, nor does it report a separate held-out test set or a false-positive rate measured on real survey noise and artifacts. Without those numbers it is difficult to judge how much contamination is carried into the larger 3822-cluster list. If the full methods section supplies the training-set size, the exact metric definition, and a direct comparison against human inspection on a blind test sample, that concern shrinks; otherwise it remains the main thing a referee would press on. The paper is worth sending to review because the underlying problem is real and the approach is a clear step toward scalable processing for SKA-era data volumes. A referee can tighten the validation section without changing the overall contribution.

Referee Report

3 major / 3 minor

Summary. The paper introduces Radio U-Net, a convolutional neural network for pixel-level segmentation of diffuse radio emission in LoTSS-DR3 20-arcsec images. It combines network outputs with positional, mass, and redshift data from four X-ray- and SZ-selected cluster catalogues to produce a merged sample of 3822 clusters, selects a high-accuracy sub-sample of 357 clusters (reported 76% accuracy), provides examples of new detections, and verifies that the detection fraction of diffuse sources increases with cluster mass and redshift.

Significance. If the network generalizes reliably, this pipeline offers a scalable, interpretable tool for identifying diffuse radio emission in large surveys, reducing dependence on visual inspection and enabling targeted follow-up and statistical analyses of cluster radio halos and relics. The alignment of results with expected astrophysical trends (higher detection at higher mass and redshift) provides supporting evidence for the method's utility.

major comments (3)

Abstract: The central claim of 76% network accuracy for the 357-cluster sub-sample is not supported by details on training set size/composition, train/validation/test splits, the precise accuracy metric (pixel accuracy, IoU, or precision at threshold), or false-positive rates measured on real survey data containing noise, artifacts, and unrelated extended sources. This information is required to bound contamination in the parent 3822-cluster sample.
Methods/Results: No quantitative comparison of network outputs to independent human visual inspection is reported for a held-out test set or the selected sub-sample, leaving the generalization performance and equivalence to traditional methods unquantified.
Results: The reported increase in detection fraction with mass and redshift relies on network-derived selections, yet lacks completeness/contamination corrections or error bars derived from the network's validation metrics, weakening the physical interpretation.

minor comments (3)

Abstract: '20arcsec' should be written as '20-arcsec' for standard formatting.
Consider adding a dedicated table or subsection summarizing Radio U-Net hyperparameters, dataset splits, and all performance metrics (including precision, recall, and IoU) to improve reproducibility.
The catalogue-merging procedure would benefit from explicit discussion of overlaps, duplicate handling, and selection biases between the four input catalogues.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed report. We address each major comment below and indicate where revisions will be made to strengthen the manuscript.

read point-by-point responses

Referee: Abstract: The central claim of 76% network accuracy for the 357-cluster sub-sample is not supported by details on training set size/composition, train/validation/test splits, the precise accuracy metric (pixel accuracy, IoU, or precision at threshold), or false-positive rates measured on real survey data containing noise, artifacts, and unrelated extended sources. This information is required to bound contamination in the parent 3822-cluster sample.

Authors: We agree that additional methodological details are required to support the accuracy claim and to allow readers to assess contamination. In the revised manuscript we will expand the Methods section with the training set size and composition (number of images and labeling procedure), the train/validation/test split ratios, and the exact definition of the 76% accuracy metric (pixel-level agreement with human labels on the validation set). We will also report the false-positive rate measured on the validation data and discuss its implications for the parent sample, including a qualitative estimate of contamination. revision: yes
Referee: Methods/Results: No quantitative comparison of network outputs to independent human visual inspection is reported for a held-out test set or the selected sub-sample, leaving the generalization performance and equivalence to traditional methods unquantified.

Authors: The 76% figure already reflects quantitative agreement with human visual inspection on the validation set. However, we acknowledge the value of an explicit held-out test-set evaluation and of metrics for the high-confidence sub-sample. We will add a dedicated subsection reporting precision, recall and IoU on the test set, together with the level of agreement between network predictions and independent visual inspection for the 357-cluster sub-sample. revision: yes
Referee: Results: The reported increase in detection fraction with mass and redshift relies on network-derived selections, yet lacks completeness/contamination corrections or error bars derived from the network's validation metrics, weakening the physical interpretation.

Authors: We agree that error bars and a discussion of selection biases would improve the robustness of the trend. In the revision we will add binomial or bootstrap error bars to the detection-fraction plots, derived from the network validation metrics. We will also include a short discussion of completeness and contamination, noting that the observed trends are consistent with prior expectations but that full statistical corrections would require a larger labeled dataset; this limitation will be stated explicitly. revision: partial

Circularity Check

0 steps flagged

No circularity: application of trained segmentation network to independent survey data

full rationale

The paper trains Radio U-Net on a labeled subset of images for pixel-level segmentation of diffuse emission, then applies the fixed model to the full LoTSS-DR3 20-arcsec images and merges outputs with four external X-ray/SZ cluster catalogues (yielding 3822 clusters). The 357-cluster high-accuracy subsample and the mass/redshift trend verification are downstream empirical results, not quantities defined by or fitted to the final catalogue itself. No equation or claim reduces by construction to its own inputs; the network output is an independent forward pass, and accuracy is reported as a measured performance figure rather than a self-referential definition. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Based solely on the abstract, the central claim rests on the trained neural network generalizing to new data and the input cluster catalogues being reliable for association. No explicit numerical free parameters are stated beyond the implicit model training process.

axioms (2)

domain assumption The Radio U-Net model trained on a subset of data generalizes to the full LoTSS-DR3 survey without significant domain shift or false detections.
This underpins the production of the segmentation map and probability scores for all 3822 clusters.
domain assumption The four X-ray and SZ-selected cluster catalogues provide accurate positions, masses, and redshifts for reliable association with detected emission.
Used to merge samples and verify trends with mass and redshift.

pith-pipeline@v0.9.0 · 5746 in / 1626 out tokens · 85904 ms · 2026-05-08T05:33:06.842293+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

81 extracted references · 3 canonical work pages · 2 internal anchors

[1]

2015, TensorFlow: Large-Scale Ma- chine Learning on Heterogeneous Systems, paper and Software available from https://www.tensorflow.org/

Abadi, M., Agarwal, A., Barham, P., et al. 2015, TensorFlow: Large-Scale Ma- chine Learning on Heterogeneous Systems, paper and Software available from https://www.tensorflow.org/

2015
[2]

Agarap, A. F. 2018, arXiv e-prints, arXiv:1803.08375

work page internal anchor Pith review Pith/arXiv arXiv 2018
[3]

2019, A&A, 631, A148

Aguado-Barahona, A., Barrena, R., Streblyanska, A., et al. 2019, A&A, 631, A148

2019
[4]

& Hwang, H

Bahk, H. & Hwang, H. S. 2024, ApJS, 272, 7 Baron Perez, N., Brüggen, M., Kasieczka, G., & Lucie-Smith, L. 2025, A&A, 699, A302

2024
[5]

2018, A&A, 616, A42

Barrena, R., Streblyanska, A., Ferragamo, A., et al. 2018, A&A, 616, A42

2018
[6]

S., & Csabai, I

Beck, R., Dobos, L., Budavári, T., Szalay, A. S., & Csabai, I. 2016, Monthly Notices of the Royal Astronomical Society, 460, 1371

2016
[7]

P., Menanteau, F., et al

Boada, S., Hughes, J. P., Menanteau, F., et al. 2019, ApJ, 871, 188 Böhringer, H., Chon, G., Collins, C. A., et al. 2013, A&A, 555, A30

2019
[8]

W., Cassano, R., et al

Botteon, A., Shimwell, T. W., Cassano, R., et al. 2022, A&A, 660, A78

2022
[9]

J., Brunetti, G., et al

Botteon, A., van Weeren, R. J., Brunetti, G., et al. 2020, MNRAS, 499, L11 Brüggen, M., Bykov, A., Ryu, D., & Röttgering, H. 2012, Space Sci. Rev., 166, 187

2020
[10]

2011, Journal of Astrophysics and Astronomy, 32, 437

Brunetti, G. 2011, Journal of Astrophysics and Astronomy, 32, 437

2011
[11]

& Jones, T

Brunetti, G. & Jones, T. W. 2014, International Journal of Modern Physics D, 23, 1430007

2014
[12]

& Vazza, F

Brunetti, G. & Vazza, F. 2020, Phys. Rev. Lett., 124, 051101

2020
[13]

2025, A&A, 704, A245

Bruno, L., Botteon, A., Dallacasa, D., et al. 2025, A&A, 704, A245

2025
[14]

2023, A&A, 672, A41

Bruno, L., Brunetti, G., Botteon, A., et al. 2023, A&A, 672, A41

2023
[15]

H., et al

Buddendiek, A., Schrabback, T., Greer, C. H., et al. 2015, MNRAS, 450, 4248

2015
[16]

2024, A&A, 685, A106

Bulbul, E., Liu, A., Kluge, M., et al. 2024, A&A, 685, A106

2024
[17]

Burenin, R. A. 2017, Astronomy Letters, 43, 507

2017
[18]

A., Bikmaev, I

Burenin, R. A., Bikmaev, I. F., Khamitov, I. M., et al. 2018, Astronomy Letters, 44, 297

2018
[19]

2015, in Advancing Astrophysics with the Square Kilometre Array (AASKA14), 73

Cassano, R., Bernardi, G., Brunetti, G., et al. 2015, in Advancing Astrophysics with the Square Kilometre Array (AASKA14), 73

2015
[20]

2007, MNRAS, 378, 1565

Cassano, R., Brunetti, G., Setti, G., Govoni, F., & Dolag, K. 2007, MNRAS, 378, 1565

2007
[21]

2023, A&A, 672, A43

Cassano, R., Cuciti, V ., Brunetti, G., et al. 2023, A&A, 672, A43

2023
[22]

Chollet, F. et al. 2015, Keras, https://keras.io

2015
[23]

2024, A&A, 690, A211

Cornu, D., Salomé, P., Semelin, B., et al. 2024, A&A, 690, A211

2024
[24]

2022, Foundations and Trends in Com- puter Graphics and Vision, 14, 1

Csurka, G., V olpi, R., & Chidlovskii, B. 2022, Foundations and Trends in Com- puter Graphics and Vision, 14, 1

2022
[25]

2015, A&A, 580, A97

Cuciti, V ., Cassano, R., Brunetti, G., et al. 2015, A&A, 580, A97

2015
[26]

2023, A&A, 680, A30

Cuciti, V ., Cassano, R., Sereno, M., et al. 2023, A&A, 680, A30

2023
[27]

E., Hall, P

Dewdney, P. E., Hall, P. J., Schilizzi, R. T., & Lazio, T. J. L. W. 2009, IEEE Proceedings, 97, 1482 Di Gennaro, G., Brüggen, M., Moravec, E., et al. 2025, A&A, 695, A215 Di Gennaro, G., Timmerman, R., Hoeft, M., et al. 2026, arXiv pre-print, accepted by A&A, arXiv:2602.23046 Di Gennaro, G., van Weeren, R. J., Brunetti, G., et al. 2021, Nature Astronomy, 5, 268

work page internal anchor Pith review arXiv 2009
[28]

W., Botteon, A., Koribalski, B

Duchesne, S. W., Botteon, A., Koribalski, B. S., et al. 2024, PASA, 41, e026

2024
[29]

2026, arXiv e-prints, submitted to RAS Techniques and Instruments, arXiv:2602.15930

Etsebeth, V ., Lochner, M., Kolokythas, K., Knowles, K., & Tolley, E. 2026, arXiv e-prints, submitted to RAS Techniques and Instruments, arXiv:2602.15930

work page arXiv 2026
[30]

2012, A&A Rev., 20, 54

Feretti, L., Giovannini, G., Govoni, F., & Murgia, M. 2012, A&A Rev., 20, 54

2012
[31]

G., Boch, T., et al

Fernique, P., Allen, M. G., Boch, T., et al. 2015, A&A, 578, A114

2015
[32]

& Vazza, F

Gheller, C. & Vazza, F. 2022, MNRAS, 509, 990

2022
[33]

2018, MNRAS, 480, 3749

Gheller, C., Vazza, F., & Bonafede, A. 2018, MNRAS, 480, 3749

2018
[34]

2020, A&A, 640, A108

Giovannini, G., Cau, M., Bonafede, A., et al. 2020, A&A, 640, A108

2020
[35]

2016, Deep Learning (Cambridge, MA: MIT Press)

Goodfellow, I., Bengio, Y ., & Courville, A. 2016, Deep Learning (Cambridge, MA: MIT Press)

2016
[36]

& Feretti, L

Govoni, F. & Feretti, L. 2004, International Journal of Modern Physics D, 13, 1549

2004
[37]

2019, Science, 364, 981

Govoni, F., Orrù, E., Bonafede, A., et al. 2019, Science, 364, 981

2019
[38]

P., Huynh, M., & Petersson, L

Gupta, N., Hayder, Z., Norris, R. P., Huynh, M., & Petersson, L. 2024, PASA, 41, e001

2024
[39]

2021, ApJS, 253, 3

Hilton, M., Sifón, C., Naess, S., et al. 2021, ApJS, 253, 3

2021
[40]

2025, ApJ, 987, L40

HLavacek-Larrondo, J., Timmerman, R., Pfrommer, C., et al. 2025, ApJ, 987, L40

2025
[41]

N., Brüggen, M., Botteon, A., et al

Hoang, D. N., Brüggen, M., Botteon, A., et al. 2022, A&A, 665, A60

2022
[42]

W., Bunton, J

Hotan, A. W., Bunton, J. D., Chippendale, A. P., et al. 2021, PASA, 38, e009

2021
[43]

& MeerKAT Team

Jonas, J. & MeerKAT Team. 2016, in MeerKAT Science: On the Pathway to the SKA, 1

2016
[44]

2023, A&A, 680, A31

Jones, A., de Gasperin, F., Cuciti, V ., et al. 2023, A&A, 680, A31

2023
[45]

D., Rudnick, L., et al

Knowles, K., Cotton, W. D., Rudnick, L., et al. 2022, A&A, 657, A56

2022
[46]

2025, MNRAS, 543, 1638

Kolokythas, K., Venturi, T., Knowles, K., et al. 2025, MNRAS, 543, 1638

2025
[47]

2020, MNRAS, 496, 4141

Kosiba, M., Lieu, M., Altieri, B., et al. 2020, MNRAS, 496, 4141

2020
[48]

2025, ApJS, 276, 46

Lao, B., Andernach, H., Yang, X., et al. 2025, ApJS, 276, 46

2025
[49]

2024, A&A, 686, A55

Lee, W., Pillepich, A., ZuHone, J., et al. 2024, A&A, 686, A55

2024
[50]

2019, Galaxies, 8, 3

Lukic, V ., de Gasperin, F., & Brüggen, M. 2019, Galaxies, 8, 3

2019
[51]

Mann, A. W. & Ebeling, H. 2012, MNRAS, 420, 2120

2012
[52]

K., Tolley, E., Krishna, S

Mishra, A. K., Tolley, E., Krishna, S. P., & Kneib, J.-P. 2025, MNRAS, 538, 2905

2025
[53]

Mostert, R. I. J., Duncan, K. J., Röttgering, H. J. A., et al. 2021, A&A, 645, A89

2021
[54]

Mostert, R. I. J., Morganti, R., Brienza, M., et al. 2023, A&A, 674, A208

2023
[55]

E., Hoeft, M., van Weeren, R

Nuza, S. E., Hoeft, M., van Weeren, R. J., Gottlöber, S., & Yepes, G. 2012, MNRAS, 420, 2006

2012
[56]

W., Pointecouteau, E., & Melin, J

Piffaretti, R., Arnaud, M., Pratt, G. W., Pointecouteau, E., & Melin, J. B. 2011, A&A, 534, A109 Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2016, A&A, 594, A27

2011
[57]

& Ebeling, H

Repp, A. & Ebeling, H. 2018, MNRAS, 479, 844

2018
[58]

2024, PASA, 41, e085

Riggi, S., Cecconello, T., Palazzo, S., et al. 2024, PASA, 41, e085

2024
[59]

2023, Astronomy and Computing, 42, 100682

Riggi, S., Magro, D., Sortino, R., et al. 2023, Astronomy and Computing, 42, 100682

2023
[60]

2015, in Lecture Notes in Computer

Ronneberger, O., Fischer, P., & Brox, T. 2015, in Lecture Notes in Computer

2015
[61]

9351, Medical Image Computing and Computer-Assisted Inter- vention – MICCAI 2015 (Springer), 234–241

Science, V ol. 9351, Medical Image Computing and Computer-Assisted Inter- vention – MICCAI 2015 (Springer), 234–241

2015
[62]

W., Tarrío, P., & Melin, J

Sadibekova, T., Arnaud, M., Pratt, G. W., Tarrío, P., & Melin, J. B. 2024, A&A, 688, A187

2024
[63]

2025, ApJ, 986, 24

Sadikov, M., Hlavacek-Larrondo, J., Perreault-Levasseur, L., et al. 2025, ApJ, 986, 24

2025
[64]

2025, MNRAS, 541, 3479

Sanvitale, N., Gheller, C., Vazza, F., et al. 2025, MNRAS, 541, 3479

2025
[65]

W., Hardcastle, M

Shimwell, T. W., Hardcastle, M. J., Tasse, C., et al. 2022, A&A, 659, A1

2022
[66]

W., Hardcastle, M

Shimwell, T. W., Hardcastle, M. J., Tasse, C., et al. 2026, A&A, 707, A198

2026
[67]

W., Röttgering, H

Shimwell, T. W., Röttgering, H. J. A., Best, P. N., et al. 2017, A&A, 598, A104

2017
[68]

W., Tasse, C., Hardcastle, M

Shimwell, T. W., Tasse, C., Hardcastle, M. J., et al. 2019, A&A, 622, A1

2019
[69]

V ., Scaife, A

Slijepcevic, I. V ., Scaife, A. M. M., Walmsley, M., et al. 2024, RAS Techniques and Instruments, 3, 19

2024
[70]

2023, Experimental Astronomy, 56, 293

Sortino, R., Magro, D., Fiameni, G., et al. 2023, Experimental Astronomy, 56, 293

2023
[71]

2019, A&A, 628, A13

Streblyanska, A., Aguado-Barahona, A., Ferragamo, A., et al. 2019, A&A, 628, A13

2019
[72]

A., et al

Streblyanska, A., Barrena, R., Rubiño-Martín, J. A., et al. 2018, A&A, 617, A71

2018
[73]

2025, A&A, 695, L16

Stuardi, C., Botteon, A., Sereno, M., et al. 2025, A&A, 695, L16

2025
[74]

2024, MNRAS, 533, 3194

Stuardi, C., Gheller, C., Vazza, F., & Botteon, A. 2024, MNRAS, 533, 3194

2024
[75]

& Hinton, G

Tieleman, T. & Hinton, G. 2012, Lecture 6.5 — RMSProp: Divide the Gradient by a Running Average of its Recent Magnitude, Coursera: Neural Networks for Machine Learning, lecture notes

2012
[76]

J., Goeke, R., Bowman, J

Tingay, S. J., Goeke, R., Bowman, J. D., et al. 2013, PASA, 30, e007 van Haarlem, M. P., Wise, M. W., Gunst, A. W., et al. 2013, A&A, 556, A2 van Weeren, R. J., de Gasperin, F., Akamatsu, H., et al. 2019, Space Sci. Rev., 215, 16 van Weeren, R. J., Shimwell, T. W., Botteon, A., et al. 2021, A&A, 651, A115

2013
[77]

2019, A&A, 627, A5

Vazza, F., Ettori, S., Roncarelli, M., et al. 2019, A&A, 627, A5

2019
[78]

Wen, Z. L. & Han, J. L. 2024, ApJS, 272, 39

2024
[79]

E., Pacaud, F., Reiprich, T

Xu, W., Ramos-Ceja, M. E., Pacaud, F., Reiprich, T. H., & Erben, T. 2022, A&A, 658, A59

2022
[80]

S., et al

Zheng, H., Tegmark, M., Dillon, J. S., et al. 2017, MNRAS, 464, 3486

2017

Showing first 80 references.