Pith. sign in

REVIEW 2 major objections 6 minor 51 references

ViT-based Local Volume dwarf galaxy Identificationin (VIDA) in the CSST survey

T0 review · 2 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Simulated CSST images allow a ViT-based pipeline to recover faint Local Volume dwarf galaxies at an 85% true positive rate with a 0.1% false positive rate.

desk verdict Solid ViT-based classifier for CSST dwarf detection, but post-processing thresholds are fitted to the test set, so the headline 92% TPR and M_V=-7 limit are in-sample. read the letter →

arxiv 2506.04361 v1 pith:COVBRU5M submitted 2025-06-04 astro-ph.GA

Han Qu , Zhen Yuan , Chengliang Wei , Chao Liu , Jiang Chang , Guoliang Li , Nicolas F. Martin , Chaowei Tsai
show 6 more authors
Shi Shao Yu Luo Ran Li Xi Kang Xiangxiang Xue Zhou Fan
This is my paper · ORCID
classification astro-ph.GA
keywords LocalVolumedwarfgalaxiesCSSTsurveyVisionTransformerclassificationlowsurfacebrightnessextendedsourcedetectioncompletenessimagesimulationsatellitegalaxyluminosityfunction
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

The paper aims to show that a fully image-based pipeline can identify faint dwarf galaxies in the Local Volume in Chinese Space Station Telescope (CSST) imaging, without relying on resolved stellar catalogs. Working from simulated single-exposure CSST images, it combines an extended-source pre-processing step with a pretrained Vision Transformer (ViT) classifier, and reports a true positive rate above 85% at a false positive rate of 0.1%. It maps detection completeness across absolute magnitude, half-light radius, and distance, finding a 50% detection limit of $M_V \lesssim -7$ within 10 Mpc and a surface brightness threshold near $25\ \mathrm{mag\,arcsec^{-2}}$ beyond 1 Mpc. The demonstration matters because a complete dwarf census in the Local Volume constrains satellite luminosity functions and small-scale dark matter models, and CSST's wide survey will need automated screening of tens of millions of extended sources.

What carries the argument

The load-bearing machinery is a three-stage pipeline. First, extended source detection builds signal-to-noise maps in the g, r, and i bands, applies thresholds and size cuts to identify high-surface-brightness candidates, masks them along with bright star contaminants, and then detects low-surface-brightness candidates in the masked maps. Each candidate is cut out and rescaled to a 224-by-224 pixel three-band image. Second, a Vision Transformer base model (patch size 16, pretrained on a large generic image dataset) is fine-tuned on these cutouts as a binary classifier, using self-attention to capture the global morphological distinction between partially resolved dwarf galaxies with visible member stars and unresolved distant galaxies. Third, a post-processing step runs a source extraction routine on g- and i-band cutouts, builds a smoothed object density map, and reclassifies ViT-negative samples with central object counts above 7 and central signal-to-noise above 6 as dwarf candidates. This last step recovers about 45% of true dwarfs missed by the classifier at negligible contamination cost.

What would settle it

Run the trained pipeline on the first real single-exposure CSST images and compare its detections against a deep spectroscopic census of known dwarf galaxies in the same fields. If the true positive rate at a fixed 0.1% false positive rate falls well below 85%, or if known dwarfs with $M_V \approx -7$ within 10 Mpc are systematically missed, the simulator-trained claim would not transfer. A cheaper intermediate test is to rerun the simulation with a degraded point-spread function and higher sky noise and check how quickly the quoted true positive rate falls.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that nearby dwarf galaxies can be separated from distant background galaxies in CSST images by morphology alone, using a transformer trained on cutouts from a realistic simulator. In the simulated data, the ViT-Base classifier converges to a mean true positive rate of about 85% at a fixed false positive rate of 0.1%, outperforming CNN baselines by roughly 15–20 percentage points at the same false-positive constraint. A post-processing step that re-examines negative samples for central stellar overdensities recovers around 45% of the misclassified dwarfs, raising the overall true positive rate to about 92% while increasing contamination by only 0.02%. The result is a quantified completeness surface over stellar mass, half-light radius, and distance, which yields a 50% detection limit of $M_V \lesssim -7$ within 10 Mpc and a surface brightness limit of about $25\ \mathrm{mag\,arcsec^{-2}}$ beyond 1 Mpc.

Load-bearing premise

The load-bearing premise is that the CSST Image Simulator and the injected dwarf galaxy models (fixed age 11 Gyr, metallicity $[\mathrm{M/H}] = -2$, exponential profiles) faithfully represent real CSST observations and the real Local Volume dwarf population; if the simulated point-spread function, noise, or dwarf structures differ from reality, the quoted detection rates will not transfer to the actual survey.

Editorial extensions

If this is right

  • The CSST wide survey can systematically search for Local Volume dwarf galaxies out to 20 Mpc when this image-based pipeline is combined with traditional matched-filter catalog methods.
  • The image-based approach reaches more than one magnitude deeper in absolute magnitude than the catalog-based method beyond 500 kpc, where member stars are not resolved.
  • A 50% completeness at $M_V \lesssim -7$ within 10 Mpc and a surface brightness threshold near $25\ \mathrm{mag\,arcsec^{-2}}$ beyond 1 Mpc make CSST suitable for blind ultra-diffuse galaxy searches in the Local Volume.
  • Because the method works on single-exposure images with only calibration and stacking, candidate identification could proceed in real time early in the survey rather than waiting for stacked catalogs.
  • The post-processing stage raises the overall true positive rate to roughly 92% at a false positive rate of 0.1%, implying most recoverable dwarfs are not lost at the classification step.

Reading between the lines

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

  • If the simulator's point-spread function and noise model match real CSST data, the same pretrained ViT should transfer with modest fine-tuning; the fastest test is to apply the pipeline to the first real single-exposure frames and check the true positive rate on known nearby dwarfs.
  • The published completeness grid over $M_V$, $R_h$, and distance directly enables a selection-function correction for future CSST dwarf luminosity functions, a step the paper does not itself carry out.
  • The central-overdensity post-processing criterion is likely reusable for other low-surface-brightness tracers such as globular clusters and tidal debris, though its thresholds were tuned for dwarf-like concentrations.
  • Known very low surface brightness dwarfs such as NGC55-dw1 and Triangulum IV fall below the pipeline's $25\ \mathrm{mag\,arcsec^{-2}}$ threshold, so a complete census will still need deeper stacked images or a matched-filter complement in the lowest-surface-brightness regime.
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

2 major / 6 minor

Summary. The paper presents VIDA, a three-stage pipeline for detecting Local Volume dwarf galaxies in simulated CSST imaging data: (1) an extended-source detection pre-processing step, (2) classification with a pre-trained ViT-Base model fine-tuned on mock cutouts, and (3) a post-processing step that re-examines ViT-negative samples using central object counts and central S/N. The authors report a ViT classifier TPR exceeding 85% at FPR = 0.1%, a post-processing recovery of about 45% of missed dwarfs with only a 0.02% contamination increase, an overall TPR near 92%, and 50% completeness limits of M_V <= -7 within 10 Mpc and a surface brightness threshold near 25 mag/arcsec^2 beyond 1 Mpc. The evaluation uses artificial dwarf galaxies with fixed age, metallicity, and exponential profiles injected into CSST Image Simulator mock images, with training, validation, and test images all derived from the same simulator.

Significance. If the quantitative claims hold, the paper would provide a useful, fully automated, image-based search strategy for the CSST wide survey and a concrete comparison with catalog-based matched-filter approaches. The experimental design has real strengths: it uses galaxy-level train/validation/test splits, builds 20 dataset pairs with Group A/B configurations so that every galaxy appears in at least one test set, tests generalization across four sky regions, compares ViT against ResNet baselines, and makes the code publicly available. These choices guard against several forms of overfitting in the classifier itself. The main significance caveat is that all results are derived from mock images and from artificial dwarfs with a narrow set of stellar population and structural assumptions; the paper explicitly acknowledges that real CSST data will introduce additional complexities, and that a robust assessment must wait for actual observations.

major comments (2)
  1. [§5.1, Fig. 11] The two post-processing thresholds (central object count > 7 and central S/N > 6) are selected from the distributions of 'all samples classified as negative by the ViT model,' which are the same evaluation samples later used to report the approximate 45% recovery, the 0.02% contamination increase, the final TPR near 92% in Section 6, and the completeness limits in Figure 14. The paper does not describe a separate validation split, cross-validation, or uncertainty/sensitivity analysis for these two free parameters. Because the thresholds are chosen after inspecting the test labels, the reported post-processing gain and the headline detection limits are in-sample estimates and are likely optimistic. The abstract's 85% TPR at FPR = 0.001 refers to the ViT classifier alone and is less affected, but the 92% figure and the completeness limits explicitly include post-processing. Please re-select the thresholds on a held-out split or via nested cross-validation and re-derive the reported TPR, contamination rate, and completeness limits; at minimum, report how the final numbers vary with the threshold values.
  2. [§2.2, §2.3, §6] The completeness claims (M_V <= -7 within 10 Mpc and a surface brightness threshold near 25 mag/arcsec^2) are conditioned on the CSST Image Simulator (Wei et al., in preparation) and on the injected dwarf galaxy model with fixed age 11 Gyr, [M/H] = -2.0, and exponential radial profiles. If the real CSST PSF, noise, or background differ from the simulator, or if real Local Volume dwarfs have systematically different stellar populations or structural profiles, the quoted limits will not transfer to the survey. The paper acknowledges this limitation in Sections 4.3 and 6, but because the abstract and summary state these as CSST detection capabilities, the conditional nature should be made explicit in the abstract and conclusions. A concrete test on archival imaging of known dwarfs from surveys such as HSC/Subaru or HST would substantially strengthen the external validity of the claims.
minor comments (6)
  1. [§3.1, Fig. 4] The text contains typos 'blue cricles' and 'green cricles'; these should be 'circles'.
  2. [§4.2] The sentence 'The middle panel illustrates the learning curves under various learning rates...' appears twice in succession, interrupting the flow; one occurrence should be removed or replaced with a reference to the correct panel.
  3. [§2.2, Table 1] The table lists D up to 199,521 kpc with a log step of 0.25, and states 1,953 artificial galaxies are generated, but it is unclear how the grid in stellar mass, half-light radius, and distance is sampled to produce this total; please clarify the sampling scheme.
  4. [§5.1, Fig. 11] The histograms in Figure 11 show percentages without error bars or raw counts; normalized histograms with Poisson uncertainties would make the separation between true positives and true negatives more convincing and better justify the chosen thresholds.
  5. [References] The entries Davis et al. 2021a and Davis et al. 2021b both cite MNRAS, 500, 3854, which appears to be the same article; please verify whether they are distinct papers or a duplicate citation.
  6. [Throughout] Several typographical errors should be corrected: 'almostly complete' (§2.2), 'enabels' (§2.3), 'exisiting' (§1), 'classifer' (§5), 'resolvabled stars' (§5.2), and missing spaces such as 'MV =−7' in the abstract.

Circularity Check

1 steps flagged · score 6.0 of 10

Post-processing thresholds are tuned on the evaluation set, making the reported ~92% TPR and the M_V=-7 / 25 mag/arcsec^2 detection limits in-sample estimates.

  1. fitted input called prediction [Section 5.1 (Central overdensity evaluation) and Figure 11; results used in Section 5.2 and Summary]
    "In Figure 11, we show the distributions of "central object count" and "central S/N" for all samples initially classified as negative by the ViT model, including both true positives (actual LV dwarf galaxy) and true negatives. ... Based on this distinction, we apply the following selection criteria: samples with "central object count" >7 and "central S/N" >6 are reclassified as LV dwarf galaxies. After applying the post-processing step, approximately 45% of LV dwarf galaxies initially misclassified as negative by the ViT model are correctly reclassified as new positive."

    The two selection thresholds (>7 objects, S/N >6) are read off Figure 11, which plots the distributions of the very evaluation samples on which the post-processing gain is then measured, with true-positive/true-negative labels known. No separate validation split or cross-validation is described for choosing these two free parameters. The reported 45% recovery, the resulting 'overall TPR to approximately 92%' in the Summary, and the derived detection limits (M_V <= -7 within 10 Mpc, mu >= 25 mag/arcsec2 beyond 1 Mpc) therefore include a component fitted to the test data rather than predicted. The ViT-only 85% TPR at FPR=0.1% is an independent held-out measurement and is not itself circular, but the headline completeness limits combine ViT output with this in-sample post-processing gain.

full rationale

The ViT classifier's 85% TPR at FPR=0.001 is a genuine held-out measurement: galaxies are split by object into train/validation/test, and the ImageNet-21k pretraining provides an external anchor. However, the paper's next-stage claim is not independent. Section 5.1 chooses central-object-count and central-S/N thresholds by inspecting Figure 11, which displays exactly the samples later used to compute the post-processing recovery (45%), the final ~92% TPR, and the Figure 13-14 completeness limits. Because the thresholds are not fixed a priori and no separate validation of the threshold choice is reported, the post-processed completeness numbers are in-sample estimates; they would be expected to shrink on fresh CSST data. This is a fitted-input-called-prediction circularity in the post-processing leg of the pipeline. The artificial-galaxy population and CSST simulator are assumptions whose realism cannot be checked internally, but that is a correctness/transferability risk rather than a circularity, and the paper states these limitations. No load-bearing self-citation chain was found: Qu23 is used for the dwarf-model recipe and as a comparison baseline, not as the proof of the detection rates. Overall, the core ViT result is not circular, but the headline detection limits reduce in part to thresholds calibrated on the evaluation set, so a score of 6 is appropriate.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central detection rates depend on the fidelity of simulated CSST images, the simplified artificial dwarf galaxy models, and the hand-chosen detection and post-processing thresholds. None of these are validated against real CSST data in this paper, so they represent the main burden carried by the reported performance numbers.

free parameters (3)
  • Post-processing selection thresholds = central object count > 7 and central S/N > 6
    Chosen from the distributions in Figure 11 to separate true positives from true negatives in the same simulated data set (Section 5.1). These thresholds directly affect the reported final TPR of about 92%.
  • Extended source detection thresholds = HSB: S1>60 px^2, S2>600 px^2, S1/S2>0.2; LSB: S/N>4, S2>400 px^2; S/N thresholds 6 and 1.5
    Hand-chosen detection criteria in the pre-processing step (Section 3.1). They determine which dwarf galaxies enter the classifier, so they set the upper bound on overall completeness.
  • Per-galaxy sampling cap Nmax = 15 cutouts (baseline), with augmented +200 for low-detection galaxies
    A dataset-construction choice (Sections 3.2 and 4.3) that affects class balance and therefore the reported TPR. The paper shows results vary with Nmax.
assumptions (4)
  • domain assumption The CSST Image Simulator produces mock images that faithfully represent the real CSST PSF, noise, and sky background.
    All images are generated with this simulator (Section 2.3), which is cited as 'Wei et al., in preparation' and not independently validated in this paper.
  • domain assumption The artificial Local Volume dwarf galaxy population, with uniform age 11 Gyr, metallicity [M/H] = -2.0, and exponential radial profiles, represents the real dwarf population.
    Section 2.2 constructs all positive samples this way; if real dwarfs have a wider range of stellar populations or profiles, the detection limits may shift.
  • domain assumption The fiducial catalogs (TRILEGAL stars and Jiutian-1G background galaxies) realistically represent the contaminating sources in CSST images.
    Section 2.1 relies on these simulations for the negative samples and the background, so the false positive rate depends on their realism.
  • domain assumption ImageNet-21k pretraining provides transferable feature representations for astronomical cutouts.
    Section 4.1 uses the 'vit-base-patch16-224-in21k' model without task-specific pretraining on astronomical data; if transfer is poor, fine-tuning on the small mock sample may not generalize.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ViT-based Local Volume dwarf galaxy Identificationin (VIDA) in the CSST survey." pith.science (2026). https://pith.science/paper/COVBRU5M

@misc{pith2026250604361,
  author       = {Pith},
  title        = {Pith review of: ViT-based Local Volume dwarf galaxy Identificationin (VIDA) in the CSST survey},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/COVBRU5M}},
  note         = {Machine review of arXiv:2506.04361}
}
abstract

Identifying dwarf galaxies within the Local Volume is crucial for constraining the luminosity function of satellite galaxies in the nearby universe. We report the detection capabilities of dwarf galaxies within the Local Volume using the Chinese Space Station Telescope (CSST). Based on the simulated imaging data of CSST, we develop a detection and classification pipeline that combines traditional image-based search techniques with advanced machine learning classification models. The simulated Local Volume dwarf galaxies can be identified using a pre-processing method for "extended source detection", followed by classification with a pretrained ViT-Base model. This pipeline achieves a true positive rate (TPR) exceeding 85% with a false positive rate (FPR) of only 0.1%. We quantify the detection completeness of Local Volume dwarf galaxies across a three-dimensional parameter space defined by absolute magnitude ($M_V$), half-light radius ($R_h$), and heliocentric distance, based on simulated single-exposure CSST wide-field imaging survey data. For unresolved or semi-resolved dwarf galaxies, our method achieves a significantly deeper absolute magnitude detection limit compared to catalog-based approaches, reaching $M_V = -7$ within 10 \Mpc. By combining this image-based approach with traditional stellar catalog-based "matched filter" techniques, our automated framework established in this work can identify dwarf galaxies within 20 \Mpc for the CSST mission.

Figures

Figures reproduced from arXiv: 2506.04361 by the authors.

Figure 1
Figure 1. Distribution of the simulated sky regions in Galactic coordinates. The gray and orange regions correspond to |b| < 15◦ and |β| < 20◦, respectively, which are areas not covered by the main CSST survey [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The distance (redshift) and apparent magnitude(in the r band) distributions of stars and background galaxies in the fiducial catalog for the simulation program across our four test sky regions. Different colors represent data from different sky regions, with solid lines indicating the input catalog and dashed lines representing sources detected by the extended source detection algorithm in Section 3.1 . MNRAS 000, 1… view at source ↗
Figure 3
Figure 3. Simulated images of LV dwarf galaxies in the g-band. Each panel displays a background-free image of a simulated dwarf galaxy with a stellar mass of 104 – 106 M⊙ . The stellar mass, distance, and half-light radius of each simulated galaxy are annotated in the corresponding panel [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Extended source detection process. (a) Background-free image of a simulated LV dwarf galaxy with a stellar mass of 106 M⊙, a half-light radius of 100 pc, and a distance of 10 Mpc. (b) The same galaxy injected into a "fiducial Image" showing the S/N map obtained by divi…
Figure 5
Figure 5. Figure 5: Scaled cutouts of extended source candidates in the g-band, obtained from the extended source detection process. The first row shows true LV dwarf galaxies. The second row shows negative samples detected in the fiducial images; these are primarily distant galaxies, wit…
Figure 6
Figure 6. Figure 6: Performance evaluation of different models and training configurations based on true positive rate (TPR).Panel (a): Evolution of TPR during training at FPR = 0.001. The red solid line indicates the mean TPR across 20 dataset samples, and the shaded region denotes the 1…
Figure 7
Figure 7. Figure 7: Mean classification recall of the ViT model for LV dwarf galaxies across all 20 datasets. The color of each pixel corresponds to the TPR at FPR = 0.001 for the respective LV dwarf galaxies. The shaded regions indicate areas without test samples [PITH_FULL_IMAGE:figure…
Figure 8
Figure 8. Figure 8: ViT classification recall under different datasets.The x-axis represents the number of training epochs, and the y-axis corresponds to the TPR at FPR = 0.001. Lines in different colors represent results for different Nmax values or datasets. For lines of the same color,…
Figure 9
Figure 9. Figure 9: Difference map of ViT classification recall between Nmax =30 and Nmax =15 across 20 datasets. The figure shows the difference in TPR distributions (at FPR = 0.001) between Nmax =30 and Nmax =15. Pink pixels indicate LV dwarf galaxies for which the TPR in the Nmax =30 g…
Figure 10
Figure 10. Figure 10: Post-Process Workflow Illustration. (a) Cutout image with red circles marking the objects detected by the Source Extractor. (b) Density map of objects corresponding to (a), with each pixel representing 1". (c) and (d) are the results of convolving (b) with kernels of …
Figure 11
Figure 11. Figure 11: Distribution of "central object count" and "central S/N" for all samples classified as negative by the ViT model, including true positive samples (blue line) and true negative samples (orange line). The vertical axis of the histogram represents the percentage of sampl…
Figure 12
Figure 12. Figure 12: Classification recall during the post-processing stage. The shaded regions indicate areas without test samples. and "central S/N" for all samples initially classified as negative by the ViT model, including both true positives (actual LV dwarf galaxy) and true negativ…
Figure 13
Figure 13. Figure 13: Combined detection rate for LV dwarf galaxies, obtained by combining the ViT prediction results with the post-processing step (Figure A4) and multiplying them with the detection rates from Step 1 (Figure A3). provides a comprehensive assessment of this work in identif…
Figure 14
Figure 14. Figure 14: Comparison between the CSST dwarf galaxy detection limits derived in this work and the distribution of observed nearby galaxies.Top panel: Absolute magnitude as a function of distance. Bottom panel: Surface brightness versus distance.Purple plus signs and red circles …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 10 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTION or pop #1...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence 'output.state := if if FUNCTION not #0 #1 if FUNCTION and 'skip pop #0 if FUNCTION or pop #1...

  3. [3]

    Bennet P., Sand D. J., Crnojevi \'c D., Spekkens K., Karunakaran A., Zaritsky D., Mutlu-Pakdil B., 2020, @doi [ ] 10.3847/2041-8213/ab80c5 , https://ui.adsabs.harvard.edu/abs/2020ApJ...893L...9B 893, L9

    doi:10.3847/2041-8213/ab80c5 2020

  4. [4]

    R., Channappayya S

    Bhavanam S. R., Channappayya S. S., P. K S., Desai S., 2024, @doi [ ] 10.1007/s10509-024-04357-9 , https://ui.adsabs.harvard.edu/abs/2024Ap&SS.369...92B 369, 92

    doi:10.1007/s10509-024-04357-9 2024

  5. [5]

    Bressan A., Marigo P., Girardi L., Salasnich B., Dal Cero C., Rubele S., Nanni A., 2012, @doi [ ] 10.1111/j.1365-2966.2012.21948.x , https://ui.adsabs.harvard.edu/abs/2012MNRAS.427..127B 427, 127

    arXiv 2012

  6. [6]

    G., Greco J

    Carlsten S. G., Greco J. P., Beaton R. L., Greene J. E., 2020, @doi [ ] 10.3847/1538-4357/ab7758 , https://ui.adsabs.harvard.edu/abs/2020ApJ...891..144C 891, 144

    doi:10.3847/1538-4357/ab7758 2020

  7. [7]

    G., Greene J

    Carlsten S. G., Greene J. E., Beaton R. L., Danieli S., Greco J. P., 2022, @doi [ ] 10.3847/1538-4357/ac6fd7 , https://ui.adsabs.harvard.edu/abs/2022ApJ...933...47C 933, 47

    doi:10.3847/1538-4357/ac6fd7 2022

  8. [8]

    Crosby E., Jerjen H., M \"u ller O., Pawlowski M., Mateo M., Dirnberger M., 2023, @doi [ ] 10.1093/mnras/stad645 , https://ui.adsabs.harvard.edu/abs/2023MNRAS.521.4009C 521, 4009

    doi:10.1093/mnras/stad645 2023

Show all 51 references
  1. [9]

    Cuillandre J.-C., et al., 2025, Astronomy & Astrophysics, 697, A6

    2025

  2. [10]

    Danieli S., van Dokkum P., Conroy C., 2018, @doi [ ] 10.3847/1538-4357/aaadfb , https://ui.adsabs.harvard.edu/abs/2018ApJ...856...69D 856, 69

    2018 doi

  3. [12]

    B., et al., 2021b, @doi [ ] 10.1093/mnras/staa3246 , https://ui.adsabs.harvard.edu/abs/2021MNRAS.500.3854D 500, 3854

    Davis A. B., et al., 2021b, @doi [ ] 10.1093/mnras/staa3246 , https://ui.adsabs.harvard.edu/abs/2021MNRAS.500.3854D 500, 3854

    doi

  4. [13]

    A., Ng K

    Dekker A., Ando S., Correa C. A., Ng K. C. Y., 2022, @doi [ ] 10.1103/PhysRevD.106.123026 , https://ui.adsabs.harvard.edu/abs/2022PhRvD.106l3026D 106, 123026

    2022 doi

  5. [14]

    Doliva-Dolinsky A., et al., 2022, The PAndAS View of the Andromeda Satellite System. III. Dwarf galaxy detection limits ( @eprint arXiv 2205.02831 )

    2022 arXiv

  6. [15]

    Doliva-Dolinsky A., Collins M. L. M., Martin N. F., 2025, @doi [arXiv e-prints] 10.48550/arXiv.2502.06948 , https://ui.adsabs.harvard.edu/abs/2025arXiv250206948D p. arXiv:2502.06948

    2025 doi arXiv

  7. [16]

    arXiv:2010.11929

    Dosovitskiy A., et al., 2020, @doi [arXiv e-prints] 10.48550/arXiv.2010.11929 , https://ui.adsabs.harvard.edu/abs/2020arXiv201011929D p. arXiv:2010.11929

    2020 doi arXiv

  8. [17]

    Drlica-Wagner A., et al., 2015, @doi [ ] 10.1088/0004-637X/813/2/109 , https://ui.adsabs.harvard.edu/abs/2015ApJ...813..109D 813, 109

    2015 doi

  9. [18]

    Drlica-Wagner A., et al., 2021, @doi [ ] 10.3847/1538-4365/ac079d , https://ui.adsabs.harvard.edu/abs/2021ApJS..256....2D 256, 2

    2021 doi

  10. [19]

    Engler C., et al., 2021, @doi [ ] 10.1093/mnras/stab2437 , https://ui.adsabs.harvard.edu/abs/2021MNRAS.507.4211E 507, 4211

    2021 doi

  11. [20]

    Euclid Collaboration et al., 2022, @doi [ ] 10.1051/0004-6361/202141938 , https://ui.adsabs.harvard.edu/abs/2022A&A...662A.112E 662, A112

    2022 doi

  12. [21]

    Fern \'a ndez-Iglesias J., Buitrago F., Sahelices B., 2024, @doi [ ] 10.1051/0004-6361/202347216 , https://ui.adsabs.harvard.edu/abs/2024A&A...683A.145F 683, A145

    2024 doi

  13. [22]

    J., Ben \' tez-Llambay A., Cole S., Frenk C., 2022, @doi [ ] 10.1093/mnras/stac3062 , https://ui.adsabs.harvard.edu/abs/2022MNRAS.517.5627F 517, 5627

    Forouhar Moreno V. J., Ben \' tez-Llambay A., Cole S., Frenk C., 2022, @doi [ ] 10.1093/mnras/stac3062 , https://ui.adsabs.harvard.edu/abs/2022MNRAS.517.5627F 517, 5627

    2022 doi

  14. [23]

    Girardi L., 2016, @doi [Astronomische Nachrichten] 10.1002/asna.201612388 , https://ui.adsabs.harvard.edu/abs/2016AN....337..871G 337, 871

    2016 doi

  15. [24]

    arXiv:2411.14392

    Gondhalekar Y., Moriwaki K., 2024, @doi [arXiv e-prints] 10.48550/arXiv.2411.14392 , https://ui.adsabs.harvard.edu/abs/2024arXiv241114392G p. arXiv:2411.14392

    2024 doi arXiv

  16. [25]

    Governato F., et al., 2015, @doi [ ] 10.1093/mnras/stu2720 , https://ui.adsabs.harvard.edu/abs/2015MNRAS.448..792G 448, 792

    2015 doi

  17. [26]

    Gozman K., et al., 2024, @doi [ ] 10.3847/1538-4357/ad8c3a , https://ui.adsabs.harvard.edu/abs/2024ApJ...977..179G 977, 179

    2024 doi

  18. [27]

    arXiv:2503.21368

    Han J., et al., 2025, @doi [arXiv e-prints] 10.48550/arXiv.2503.21368 , https://ui.adsabs.harvard.edu/abs/2025arXiv250321368H p. arXiv:2503.21368

    2025 doi arXiv

  19. [28]

    Homma D., et al., 2024, @doi [ ] 10.1093/pasj/psae044 , https://ui.adsabs.harvard.edu/abs/2024PASJ...76..733H 76, 733

    2024 doi

  20. [29]

    arXiv:2210.04143

    Huang K.-W., Chih-Fan Chen G., Chang P.-W., Lin S.-C., Hsu C.-J., Thengane V., Yao-Yu Lin J., 2022, @doi [arXiv e-prints] 10.48550/arXiv.2210.04143 , https://ui.adsabs.harvard.edu/abs/2022arXiv221004143H p. arXiv:2210.04143

    2022 doi arXiv

  21. [30]

    Ivezi \'c Z ., et al., 2019, @doi [ ] 10.3847/1538-4357/ab042c , https://ui.adsabs.harvard.edu/abs/2019ApJ...873..111I 873, 111

    2019 doi

  22. [31]

    J., Pawlowski M

    Kanehisa K. J., Pawlowski M. S., Heesters N., M \"u ller O., 2024, @doi [ ] 10.1051/0004-6361/202348242 , https://ui.adsabs.harvard.edu/abs/2024A&A...686A.280K 686, A280

    2024 doi

  23. [32]

    D., Makarov D

    Karachentsev I. D., Makarov D. I., Kaisina E. I., 2013, @doi [aj] 10.1088/0004-6256/145/4/101 , https://ui.adsabs.harvard.edu/abs/2013AJ....145..101K 145, 101

    2013 doi

  24. [33]

    Koposov S., et al., 2008, @doi [ ] 10.1086/589911 , https://ui.adsabs.harvard.edu/abs/2008ApJ...686..279K 686, 279

    2008 doi

  25. [34]

    Laevens B. P. M., et al., 2015, @doi [ ] 10.1088/2041-8205/802/2/L18 , https://ui.adsabs.harvard.edu/abs/2015ApJ...802L..18L 802, L18

    2015 doi

  26. [35]

    C., et al., 2008, in Funes J

    Lee J. C., et al., 2008, in Funes J. G., Corsini E. M., eds, Astronomical Society of the Pacific Conference Series Vol. 396, Formation and Evolution of Galaxy Disks. p. 151

    2008

  27. [36]

    arXiv:2110.01024

    Lin J., Liao S.-M., Huang H.-J., Kuo W.-T., Hsuan-Min Ou O., 2021, @doi [arXiv e-prints] 10.48550/arXiv.2110.01024 , https://ui.adsabs.harvard.edu/abs/2021arXiv211001024Y p. arXiv:2110.01024

    2021 doi arXiv

  28. [37]

    F., Ibata R

    Martin N. F., Ibata R. A., McConnachie A. W., Mackey A. D., Ferguson A. M. N., Irwin M. J., Lewis G. F., Fardal M. A., 2013, @doi [ ] 10.1088/0004-637X/776/2/80 , https://ui.adsabs.harvard.edu/abs/2013ApJ...776...80M 776, 80

    2013 doi

  29. [38]

    F., et al., 2016, @doi [ ] 10.3847/1538-4357/833/2/167 , https://ui.adsabs.harvard.edu/abs/2016ApJ...833..167M 833, 167

    Martin N. F., et al., 2016, @doi [ ] 10.3847/1538-4357/833/2/167 , https://ui.adsabs.harvard.edu/abs/2016ApJ...833..167M 833, 167

    2016 doi

  30. [39]

    McNanna M., et al., 2024, @doi [ ] 10.3847/1538-4357/ad07d0 , https://ui.adsabs.harvard.edu/abs/2024ApJ...961..126M 961, 126

    2024 doi

  31. [40]

    Merritt A., van Dokkum P., Abraham R., Zhang J., 2016, @doi [ ] 10.3847/0004-637X/830/2/62 , https://ui.adsabs.harvard.edu/abs/2016ApJ...830...62M 830, 62

    2016 doi

  32. [41]

    arXiv:2407.07481

    Ogami I., et al., 2024, @doi [arXiv e-prints] 10.48550/arXiv.2407.07481 , https://ui.adsabs.harvard.edu/abs/2024arXiv240707481O p. arXiv:2407.07481

    2024 doi arXiv

  33. [42]

    B., 2024, @doi [arXiv e-prints] 10.48550/arXiv.2411.07424 , https://ui.adsabs.harvard.edu/abs/2024arXiv241107424P p

    Pace A. B., 2024, @doi [arXiv e-prints] 10.48550/arXiv.2411.07424 , https://ui.adsabs.harvard.edu/abs/2024arXiv241107424P p. arXiv:2411.07424

    2024 doi

  34. [43]

    Planck Collaboration et al., 2020, @doi [ ] 10.1051/0004-6361/201833910 , https://ui.adsabs.harvard.edu/abs/2020A&A...641A...6P 641, A6

    2020 doi

  35. [44]

    Qu H., et al., 2023, @doi [ ] 10.1093/mnras/stad1352 , https://ui.adsabs.harvard.edu/abs/2023MNRAS.523..876Q 523, 876

    2023 doi

  36. [45]

    Rawat W., Wang Z., 2017, @doi [Neural Computation] 10.1162/neco_a_00990 , 29, 2352

    2017 doi

  37. [46]

    E., et al., 2023, @doi [ ] 10.3847/2041-8213/aca086 , https://ui.adsabs.harvard.edu/abs/2023ApJ...942L..42R 942, L42

    Robertson B. E., et al., 2023, @doi [ ] 10.3847/2041-8213/aca086 , https://ui.adsabs.harvard.edu/abs/2023ApJ...942L..42R 942, L42

    2023 doi

  38. [47]

    Rowe B. T. P., et al., 2015, @doi [Astronomy and Computing] 10.1016/j.ascom.2015.02.002 , https://ui.adsabs.harvard.edu/abs/2015A&C....10..121R 10, 121

    2015 doi

  39. [48]

    H., Taubenberger S., Inoue K

    Shu Y., Ca \ n ameras R., Schuldt S., Suyu S. H., Taubenberger S., Inoue K. T., Jaelani A. T., 2022, @doi [ ] 10.1051/0004-6361/202243203 , https://ui.adsabs.harvard.edu/abs/2022A&A...662A...4S 662, A4

    2022 doi

  40. [49]

    D., 2019, @doi [ ] 10.1146/annurev-astro-091918-104453 , https://ui.adsabs.harvard.edu/abs/2019ARA&A..57..375S 57, 375

    Simon J. D., 2019, @doi [ ] 10.1146/annurev-astro-091918-104453 , https://ui.adsabs.harvard.edu/abs/2019ARA&A..57..375S 57, 375

    2019 doi

  41. [50]

    Tollerud E., Hamanowicz A., Mao Y.-Y., Geha M., Wechsler R., Weiner B., Nadler E., Kallivayalil N., 2022, in American Astronomical Society Meeting \#240. p. 145.11

    2022

  42. [51]

    M., Willman B., Jerjen H., 2009, @doi [ ] 10.1088/0004-6256/137/1/450 , https://ui.adsabs.harvard.edu/abs/2009AJ....137..450W 137, 450

    Walsh S. M., Willman B., Jerjen H., 2009, @doi [ ] 10.1088/0004-6256/137/1/450 , https://ui.adsabs.harvard.edu/abs/2009AJ....137..450W 137, 450

    2009 doi

  43. [52]

    Zhan H., 2021, Chinese Science Bulletin (Chinese Version), 66, 1290

    2021

Pith tools

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