arxiv: 2605.03442 · v1 · submitted 2026-05-05 · 🌌 astro-ph.GA · astro-ph.IM

Recognition: unknown

High-Redshift Gravitational Lens Discoveries in JWST NIRCam Using AnomalyMatch

Julia Dima , David O'Ryan , Sandor Kruk , Laslo E. Ruhberg , Pablo G\'omez

Authors on Pith no claims yet

Pith reviewed 2026-05-07 15:43 UTC · model grok-4.3

classification 🌌 astro-ph.GA astro-ph.IM

keywords gravitational lensesJWSTAnomalyMatchsemi-supervised learninghigh-redshiftCOSMOS-Webstrong lensingNIRCam

0 comments

The pith

A semi-supervised neural network trained on eleven known lenses identifies 58 gravitational lenses in JWST NIRCam data, including 37 previously uncatalogued systems.

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

The paper tests whether AnomalyMatch, a semi-supervised anomaly detection approach, can locate strong gravitational lenses in JWST images from the ASTRODEEP and COSMOS-Web surveys. It begins with a training set of only eleven previously identified lenses and uses repeated human review to refine the model so it can flag subtle lensing features amid imaging artifacts and other rare objects. The search returns 58 unique candidates, which four experts grade into 16 high-confidence, 16 medium-confidence, and 26 lower-confidence systems, with 37 of them absent from prior catalogs. Several of the new lenses show photometric redshifts near 2.1, pushing beyond the redshift range of most earlier discoveries. The work concludes that this human-in-the-loop method offers a practical route to mining large JWST archives for rare high-redshift lenses without needing thousands of labeled examples.

Core claim

By applying AnomalyMatch to JWST Level 3 NIRCam products, the authors recover 58 distinct gravitational lens systems. These include 16 Grade A, 16 Grade B, and 26 Grade C candidates according to expert visual inspection, of which 37 had not been catalogued before. The previously known lenses reach spectroscopic redshifts below 1.39 and photometric redshifts below 2.21; one newly identified system reaches a photometric redshift of 2.1. The results demonstrate that a neural network seeded with only eleven labeled examples can, after iterative human refinement, isolate true lenses from other anomalies in the data.

What carries the argument

AnomalyMatch, a semi-supervised learning procedure that trains a neural network on a small seed set of known gravitational lenses and then iteratively incorporates human labels to separate genuine lensing events from imaging artifacts and unrelated rare objects.

If this is right

The 37 newly discovered lenses supply additional targets for measuring mass distributions and source populations at redshifts greater than 1.
Expert grading into A, B, and C classes yields a prioritized list that can guide spectroscopic follow-up campaigns.
The same workflow can be applied to the full JWST archive to increase the total number of known high-redshift lenses.
Photometric redshifts reaching 2.1 for new systems extend the redshift baseline available for strong-lensing cosmology.
The approach avoids the need for thousands of pre-labeled examples, lowering the barrier to searching other large imaging surveys.

Where Pith is reading between the lines

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

Similar anomaly-detection pipelines could be adapted to locate other scarce high-redshift objects such as distant supernovae or unusual galaxies in the same JWST data.
If future versions reduce the number of human review cycles required, the method could support fully automated searches across even larger datasets.
The incompleteness of existing lens catalogs, revealed by the 37 new finds, suggests that statistical inferences drawn from current samples may be biased low.
Extending the search to additional JWST fields or other near-infrared surveys could test whether lens abundance continues to rise at still higher redshifts.

Load-bearing premise

That iterative human feedback on a neural network started with only eleven labeled lenses is sufficient to produce reliable separation between true gravitational lenses and other anomalies or artifacts.

What would settle it

A program of follow-up spectroscopy or deeper imaging on the 32 Grade A and B candidates that shows a majority are not lensing systems but instead chance alignments, artifacts, or unrelated objects would falsify the claim that the method reliably identifies true lenses.

Figures

Figures reproduced from arXiv: 2605.03442 by David O'Ryan, Julia Dima, Laslo E. Ruhberg, Pablo G\'omez, Sandor Kruk.

**Figure 1.** Figure 1: The CCDF of segmentation areas from the ASTRODEEP and COSMOS-Web photometric catalogues. We adopt a radial cut based on the segmentation area of each source in pixels. We retain only sources with 3.5 reff ≥ 12 pixels, equivalent to reff ≥ 3.43 pixels, or log(reff) ≈ 0.54 view at source ↗

**Figure 2.** Figure 2: The AnomalyMatch workflow, starting with constructing and pre-processing dataset followed by the semi-supervised training loop and label proposals. The expert validation step involves confirming previously identified lenses based on current research and grading newly discovered lens candidates, as detailed in Section 3.4. Upon training and then searching through an entire dataset, a score is given to each … view at source ↗

**Figure 3.** Figure 3: Initial lens dataset utilized in the AnomalyMatch semisupervised loop, containing 11 lenses from Mahler et al. (2025a) cross matched in Level 3 JWST data. and 20, 000 unlabelled images. After each training session, we took the top-scored 1,000 sources and visually inspected them for gravitational lenses. Any gravitational lenses found as part of this active learning are part of the results here, as they w… view at source ↗

**Figure 4.** Figure 4: Mosaic of lenses given a Grade A by expert classifiers. These represent the clearest lenses, where a clear arc of a differing colour is visible with a secondary image of the arc often present. Inset to each cutout is the right ascension and declination, including a measure of the redshift with its measurement method specified (spectroscopic vs photometric). in other contexts. To ensure that we are not miss… view at source ↗

**Figure 5.** Figure 5: As view at source ↗

**Figure 6.** Figure 6: As view at source ↗

**Figure 8.** Figure 8: Distribution of combined spectroscopic and photometric redshift measurements for identified and non-identified lenses, limited to the minimum and maximum values of the data. The non-identified sources exhibit a higher median redshift, whereas the identified lenses include the highest zlens values. Unreferenced Referenced Previously Referenced in the Literature 0.0 0.5 1.0 1.5 2.0 Redshift (z) view at source ↗

**Figure 9.** Figure 9: Violin plots showing the redshift distribution of lensing systems discovered in this work (Right: Unreferenced) and those that were previously found in the literature (Left: Referenced). spiral arms, galaxy mergers, galaxies undergoing ram pressure stripping, and image artefacts. While this choice was deliberate - contamination in the labelled set is particularly damaging for semi-supervised methods that … view at source ↗

read the original abstract

Context. Strong gravitational lenses provide a unique tool to probe cosmology and astrophysics at high redshift, offering constraints on the mass distribution of background source populations. Despite their scientific value, their rarity and subtle visual features make them challenging to identify in the wealth of data delivered by facilities such as the James Webb Space Telescope (JWST), whose unmatched resolution and near-infrared coverage make it particularly well-suited to detecting lensing systems in this regime. Aims. We make use of the specialised open-source software AnomalyMatch, a semi-supervised learning method to trawl the ASTRODEEP and COSMOS-Web surveys for gravitational lenses. Methods. Building on a training dataset of eleven previously identified gravitational lenses, we use AnomalyMatch and its iterative human-in-the-loop method to train a neural network to identify gravitational lenses in JWST Level 3 products using ESA Datalabs. Results. In total we identify 58 unique gravitational lenses. These are graded by four experts into 16 Grade A, 16 Grade B, and 26 Grade C lenses. Of all lenses identified, 37 were previously uncatalogued. We analyse their properties such as photometric redshift measurements and spectroscopic redshift, when the latter is available. The lenses previously identified span spectroscopic redshifts to zspec < 1.39 and photometric redshifts to zphot < 2.21. The uncatalogued lens system with the highest redshift is at zphot = 2.1. Conclusions. Overall, we demonstrate the potential of AnomalyMatch for large-scale searches for gravitational lenses and other rare high-redshift objects in JWST archives.

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 turns up 37 new high-redshift lens candidates in JWST data via AnomalyMatch, but the small training set and missing validation numbers leave the reliability of the full list unproven.

read the letter

The paper identifies 37 previously uncatalogued gravitational lenses in JWST NIRCam observations of the ASTRODEEP and COSMOS-Web fields. The team applied the existing AnomalyMatch tool, trained initially on eleven known lenses and refined through human-in-the-loop iterations, then graded the 58 total candidates with four experts. This adds concrete new systems to the high-redshift lens catalog, with some redshift information provided. The approach shows how semi-supervised anomaly detection can help sift through large JWST datasets for rare objects like these. The work is a direct application rather than a new technique, but it does demonstrate the tool's utility on fresh data and produces a list that others could follow up on. The soft spot is the validation. The abstract gives the counts and grades but no precision, recall, or false-positive estimates from tests or simulations. With only eleven starting examples, the risk that some candidates are imaging artifacts or unrelated anomalies is not quantified. The expert grading adds a layer of review, yet without details on agreement or criteria it remains subjective. Readers focused on gravitational lensing or JWST data processing would find the candidate list and method description useful for their own searches. It is not a major methodological advance but fills in some observational gaps. The paper deserves peer review. A referee can request the missing performance metrics and assess whether the new lenses are likely real. I recommend sending it to review instead of a desk rejection.

Referee Report

2 major / 2 minor

Summary. The manuscript applies the open-source AnomalyMatch semi-supervised anomaly detection algorithm to JWST NIRCam Level 3 imaging from the ASTRODEEP and COSMOS-Web surveys. Initialized on a training set of eleven previously known gravitational lenses, the method uses iterative human-in-the-loop refinement to identify 58 unique lens candidates. Four experts grade these into 16 Grade A, 16 Grade B, and 26 Grade C systems, of which 37 are previously uncatalogued. Photometric and spectroscopic redshifts are reported for the sample, with the highest-redshift new system at z_phot = 2.1, and the work concludes that AnomalyMatch is promising for large-scale searches of rare high-redshift objects.

Significance. If the identifications hold, the addition of 37 new high-redshift gravitational lenses would meaningfully enlarge the catalog available for studies of mass distributions, source populations, and cosmology. The demonstration that a semi-supervised approach can surface candidates from public JWST archives, combined with the use of an open-source tool and expert grading, provides a practical template that could be extended to other rare-object searches in large datasets.

major comments (2)

[Results] Results section: The central claim of 58 unique lenses (including 37 new systems) is presented without any quantitative performance metrics for AnomalyMatch, such as precision, recall, false-positive rate on held-out data, confusion matrix, or estimates from control fields or simulations. Given the small initial training set of eleven lenses, this absence directly affects in the reliability of the graded sample and the reported discovery count.
[Methods] Methods section: The iterative human-in-the-loop refinement process is described at a high level but lacks specifics on iteration count, size of the unlabeled data pool examined, criteria for selecting candidates for human review, or any assessment of inter-rater agreement among the four experts. These details are needed to evaluate potential biases and the robustness of the final 58-candidate list.

minor comments (2)

[Abstract] Abstract: The grading criteria for Grade A/B/C lenses are not defined, which would clarify the meaning of the reported counts and the distinction between the categories.
[Results] The manuscript would benefit from a table summarizing the 58 systems (coordinates, redshifts, grades, and whether previously known) to improve accessibility of the results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The comments highlight important areas for improving transparency and rigor, particularly regarding quantitative validation and methodological details. We address each major comment below and outline the planned revisions.

read point-by-point responses

Referee: [Results] Results section: The central claim of 58 unique lenses (including 37 new systems) is presented without any quantitative performance metrics for AnomalyMatch, such as precision, recall, false-positive rate on held-out data, confusion matrix, or estimates from control fields or simulations. Given the small initial training set of eleven lenses, this absence directly affects in the reliability of the graded sample and the reported discovery count.

Authors: We agree that the lack of quantitative performance metrics represents a genuine limitation of the current manuscript, especially given the small initial training set of eleven lenses. Our work is framed as an application of AnomalyMatch for discovery in public JWST archives rather than a controlled benchmark study, and we did not generate held-out test sets, simulations, or control-field estimates. In the revised manuscript we will add a new subsection (likely in Results or a dedicated Limitations paragraph) that explicitly discusses this constraint, notes the challenges of obtaining standard metrics in a semi-supervised human-in-the-loop setting, and provides a qualitative reliability assessment based on the expert grading distribution, the recovery rate of the eleven known lenses, and the fraction of Grade A/B candidates. We will not be able to add a full confusion matrix or precision-recall figures without substantial new experiments, but the added discussion will better contextualize the reported discovery count. revision: partial
Referee: [Methods] Methods section: The iterative human-in-the-loop refinement process is described at a high level but lacks specifics on iteration count, size of the unlabeled data pool examined, criteria for selecting candidates for human review, or any assessment of inter-rater agreement among the four experts. These details are needed to evaluate potential biases and the robustness of the final 58-candidate list.

Authors: We acknowledge that the Methods section would benefit from greater specificity. We will expand this section to report the number of iterations performed, the approximate size of the unlabeled image pool drawn from the ASTRODEEP and COSMOS-Web Level-3 products, the anomaly-score threshold and ranking criteria used to select candidates for expert review, and a quantitative measure of inter-rater agreement (e.g., Fleiss’ kappa) computed from the four experts’ independent grades. These details are available from our internal records of the AnomalyMatch runs and grading sessions and can be added without new data collection. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical application of anomaly detection to public data

full rationale

The paper applies the AnomalyMatch semi-supervised tool (initialized on 11 known lenses) with human-in-the-loop refinement to JWST NIRCam images from public surveys, followed by expert grading of 58 candidates (37 new). No derivations, equations, fitted parameters renamed as predictions, or self-citation chains exist. The central result (new lens identifications) is an independent empirical output on external observational data and is externally falsifiable by re-inspection of the same public images, satisfying the criteria for a self-contained non-circular finding.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the effectiveness of the semi-supervised algorithm and the reliability of expert visual grading rather than new physical axioms or fitted parameters.

axioms (1)

domain assumption A training set of eleven known lenses is sufficient to learn generalizable features for detecting new lenses in JWST NIRCam images.
The method description explicitly builds on this small set with iterative human feedback.

pith-pipeline@v0.9.0 · 5613 in / 1218 out tokens · 30228 ms · 2026-05-07T15:43:06.781826+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Reference graph

Works this paper leans on

56 extracted references · 4 canonical work pages · 1 internal anchor

[1]

D., Engel, A

Aleo, P. D., Engel, A. W., Narayan, G., et al. 2024, ApJ, 974, 172

2024
[2]

Allingham, J. F. V ., Zitrin, A., Kokorev, V ., et al. 2026, arXiv e-prints, arXiv:2602.14074

work page arXiv 2026
[3]

W., et al

Amvrosiadis, A., Lange, S., Nightingale, J. W., et al. 2025, MNRAS, 537, 1163 Astropy Collaboration, Price-Whelan, A. M., Lim, P. L., et al. 2022, ApJ, 935, 167 Astropy Collaboration, Price-Whelan, A. M., Sip˝ocz, B. M., et al. 2018, AJ, 156, 123

2025
[4]

2015, ApJ, 800, 18

Atek, H., Richard, J., Kneib, J.-P., et al. 2015, ApJ, 800, 18

2015
[5]

W., Treu, T., Bolton, A

Auger, M. W., Treu, T., Bolton, A. S., et al. 2010, ApJ, 724, 511

2010
[6]

B., Pirzkal, N., Finkelstein, S

Bagley, M. B., Pirzkal, N., Finkelstein, S. L., et al. 2024, ApJ, 965, L6

2024
[7]

2017, PASP, 129, 028001

Baines, D., Giordano, F., Racero, E., et al. 2017, PASP, 129, 028001

2017
[8]

M., Kacprzak, G

Barone, T. M., Kacprzak, G. G., Nightingale, J. W., et al. 2024, Communications Physics, 7, 286

2024
[9]

S., & Richard, J

Belli, S., Jones, T., Ellis, R. S., & Richard, J. 2013, ApJ, 772, 141

2013
[10]

2024, Space Sci

Birrer, S., Millon, M., Sluse, D., et al. 2024, Space Sci. Rev., 220, 48

2024
[11]

R., Makler, M., Albuquerque, M

Bom, C. R., Makler, M., Albuquerque, M. P., & Brandt, C. H. 2017, A&A, 597, A135

2017
[12]

2002, A&A, 389, 787 Cañameras, R., Schuldt, S., Suyu, S

Busarello, G., Merluzzi, P., La Barbera, F., Massarotti, M., & Capaccioli, M. 2002, A&A, 389, 787 Cañameras, R., Schuldt, S., Suyu, S. H., et al. 2020, A&A, 644, A163

2002
[13]

B., Suyu, S

Caminha, G. B., Suyu, S. H., Grillo, C., & Rosati, P. 2022, A&A, 657, A83

2022
[14]

M., Kartaltepe, J

Casey, C. M., Kartaltepe, J. S., Drakos, N. E., et al. 2023, ApJ, 954, 31

2023
[15]

2013, ApJ, 762, 32

Coe, D., Zitrin, A., Carrasco, M., et al. 2013, ApJ, 762, 32

2013
[16]

Collett, T. E. 2015, ApJ, 811, 20 D’Addona, M., Riccio, G., Cavuoti, S., Tortora, C., & Brescia, M. 2021, in In- telligent Astrophysics, ed. I. Zelinka, M. Brescia, & D. Baron, V ol. 39 (Intel- ligent Astrophysics), 225–244

2015
[17]

2021, MNRAS, 503, 1096

Ding, X., Treu, T., Birrer, S., et al. 2021, MNRAS, 503, 1096

2021
[18]

S., Abraham, R

Dunlop, J. S., Abraham, R. G., Ashby, M. L. N., et al. 2021, PRIMER: Public Release IMaging for Extragalactic Research, JWST Proposal. Cycle 1, ID. #1837

2021
[19]

2018, ApJ, 852, L7

Ebeling, H., Stockmann, M., Richard, J., et al. 2018, ApJ, 852, L7

2018
[20]

Overview of the JWST Advanced Deep Extragalactic Survey (JADES)

Eisenstein, D. J., Willott, C., Alberts, S., et al. 2023, arXiv e-prints, arXiv:2306.02465 Article number, page 12 of 13 Julia Dima et al.: High-Redshift Gravitational Lens Discoveries in JWST NIRCam usingAnomalyMatch Euclid Collaboration: Lines, N. E. P., Collett, T. E., Walmsley, M., et al. 2025, arXiv e-prints, arXiv:2503.15326 Euclid Collaboration: Wa...

work page internal anchor Pith review arXiv 2023
[21]

L., Bagley, M

Finkelstein, S. L., Bagley, M. B., Ferguson, H. C., et al. 2023, ApJ, 946, L13

2023
[22]

S., Falco, E

Fohlmeister, J., Kochanek, C. S., Falco, E. E., et al. 2007, ApJ, 662, 62

2007
[23]

M., Koekemoer, A

Franco, M., Casey, C. M., Koekemoer, A. M., et al. 2026, ApJ, 999, 200

2026
[24]

J., Zitrin, A., Plat, A., et al

Furtak, L. J., Zitrin, A., Plat, A., et al. 2023, ApJ, 952, 142

2023
[25]

O., Kruk, S., Cornen, C., et al

Garvin, E. O., Kruk, S., Cornen, C., et al. 2022, A&A, 667, A141 Géron, T., Smethurst, R. J., Dickinson, H., et al. 2025, ApJ, 987, 74

2022
[26]

2018, Astronomy and Computing, 24, 97 Gómez, P

Giordano, F., Racero, E., Norman, H., et al. 2018, Astronomy and Computing, 24, 97 Gómez, P. & Meoni, G. 2021, arXiv e-prints, arXiv:2103.10368 Gómez, P., Ruhberg, L. E., Nardone, M. T., & O’Ryan, D. 2025, arXiv e-prints, arXiv:2505.03509 Görnitz, N., Kloft, M., Rieck, K., & Brefeld, U. 2013, Journal of Artificial Intel- ligence Research, 46, 235

work page arXiv 2018
[27]

2023, A&A, 679, A124

Granata, G., Bergamini, P., Grillo, C., et al. 2023, A&A, 679, A124

2023
[28]

2008, A&A, 477, 397

Grillo, C., Lombardi, M., & Bertin, G. 2008, A&A, 477, 397

2008
[29]

R., Millman, K

Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357

2020
[30]

2020, ApJ, 894, 78

Huang, X., Storfer, C., Ravi, V ., et al. 2020, ApJ, 894, 78

2020
[31]

Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90

2007
[32]

2008, MNRAS, 389, 1311

Jackson, N. 2008, MNRAS, 389, 1311

2008
[33]

2019, MNRAS, 484, 5330

Jacobs, C., Collett, T., Glazebrook, K., et al. 2019, MNRAS, 484, 5330

2019
[34]

E., Krawczyk, C

Li, T., Collett, T. E., Krawczyk, C. M., & Enzi, W. 2024, MNRAS, 527, 5311

2024
[35]

Lines, N. E. P., Li, T., Collett, T. E., et al. 2025, Nature Astronomy, 9, 1116

2025
[36]

& Bassett, B

Lochner, M. & Bassett, B. A. 2021, Astronomy and Computing, 36, 100481

2021
[37]

W., Hogg, N

Mahler, G., Nightingale, J. W., Hogg, N. B., et al. 2025b, MNRAS, 544, L8 Martínez-Arrizabalaga, J., Diego, J. M., & Goicoechea, L. J. 2024, A&A, 682, A187

2024
[38]

2010,https://conference.scipy.org/proceedings/ scipy2010/pdfs/mckinney.pdf

McKinney, W. 2010,https://conference.scipy.org/proceedings/ scipy2010/pdfs/mckinney.pdf

2010
[39]

2020, Science, 369, 1347

Meneghetti, M., Davoli, G., Bergamini, P., et al. 2020, Science, 369, 1347

2020
[40]

2024, A&A, 691, A240

Merlin, E., Santini, P., Paris, D., et al. 2024, A&A, 691, A240

2024
[41]

C., Acevedo Barroso, J

Nagam, B. C., Acevedo Barroso, J. A., Wilde, J., et al. 2025, A&A, 702, A130

2025
[42]

2023, ApJ, 959, 134

Napier, K., Sharon, K., Dahle, H., et al. 2023, ApJ, 959, 134

2023
[43]

2024, in Space Data Manage- ment

Navarro, V ., del Rio, S., Angel Diego, M., et al. 2024, in Space Data Manage- ment. Studies in Big Data, V ol. 141 (Springer Nature Singapore), 1–13

2024
[44]

W., Mahler, G., McCleary, J., et al

Nightingale, J. W., Mahler, G., McCleary, J., et al. 2025, MNRAS, 543, 203 Núñez-Pizarro, C., Arévalo, P., Ávila-Vera, F., & Motta, V . 2026, A&A, 706, L12 O’Ryan, D. & Gómez, P. 2025, A&A, 704, A227

2025
[45]

V ., Bergamini, P., Meneghetti, M., et al

Pignataro, G. V ., Bergamini, P., Meneghetti, M., et al. 2021, A&A, 655, A81

2021
[46]

A., Görnitz, N., et al

Ruff, L., Vandermeulen, R. A., Görnitz, N., et al. 2019, arXiv e-prints, arXiv:1906.02694

work page arXiv 2019
[47]

G., Aussel, H., et al

Scoville, N., Abraham, R. G., Aussel, H., et al. 2007, ApJS, 172, 38

2007
[48]

& Bartelmann, M

Seidel, G. & Bartelmann, M. 2007, A&A, 472, 341

2007
[49]

B., Paquereau, L., et al

Shuntov, M., Akins, H. B., Paquereau, L., et al. 2025, A&A, 704, A339

2025
[50]

2020, Advances in neural information processing systems, 33, 596

Sohn, K., Berthelot, D., Carlini, N., et al. 2020, Advances in neural information processing systems, 33, 596

2020
[51]

2021, MN- RAS, 508, 2946

Storey-Fisher, K., Huertas-Company, M., Ramachandra, N., et al. 2021, MN- RAS, 508, 2946

2021
[52]

S., Brownstein, J

Talbot, M. S., Brownstein, J. R., Dawson, K. S., Kneib, J.-P., & Bautista, J. 2021, MNRAS, 502, 4617

2021
[53]

S., Brownstein, J

Talbot, M. S., Brownstein, J. R., Neumann, J., et al. 2022, MNRAS, 515, 4953

2022
[54]

M., Dutra, I., Natarajan, P., et al

Tokayer, Y . M., Dutra, I., Natarajan, P., et al. 2024, ApJ, 970, 143

2024
[55]

2000, A&AS, 143, 9

Wenger, M., Ochsenbein, F., Egret, D., et al. 2000, A&AS, 143, 9

2000
[56]

2022, Future Generation Computer Systems, 135, 364 Article number, page 13 of 13

Wu, X., Xiao, L., Sun, Y ., et al. 2022, Future Generation Computer Systems, 135, 364 Article number, page 13 of 13

2022