pith. machine review for the scientific record. sign in

arxiv: 2604.08686 · v1 · submitted 2026-04-09 · 🌌 astro-ph.HE · astro-ph.IM

Recognition: unknown

Ghosts of eruptions past: Searching for historical Galactic supernovae using variable thermal dust echoes and machine learning

Justin Vega , Kishalay De , Ashish Mahabal , Jacob E. Jencson , Viraj R. Karambelkar , Armin Rest , Megan Masterson
Authors on Pith no claims yet

Pith reviewed 2026-05-10 17:05 UTC · model grok-4.3

classification 🌌 astro-ph.HE astro-ph.IM
keywords supernovaedust echoesmachine learningNEOWISECassiopeia AGalactic transientsmid-infrared
0
0 comments X

The pith

Only Cassiopeia A produces detectable variable thermal dust echoes among recent Galactic supernovae in WISE observations.

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

The paper conducts the first all-sky search for dust echoes from historical Galactic supernovae using 12 years of NEOWISE mid-infrared data. By training a convolutional neural network on the known echoes around Cassiopeia A, the authors classify millions of variable sources and look for clusters of echo-like features. They conclude that Cas A is the sole detection at the survey's sensitivity limit, owing to its recent occurrence and strong shock breakout luminosity. This result implies that other supernovae from the past few centuries either lack sufficiently bright or variable echoes or fall below the detection threshold. The study also delivers an extensive catalog of Cas A echo positions for further study and sets the stage for more sensitive future missions.

Core claim

The search identifies no spatial over-densities of variable thermal dust echoes except around Cas A, at the W2 surface brightness threshold of approximately 20 Vega mag arcsec^{-2}. This uniqueness arises from Cas A's combination of young age and luminous shock breakout, which produces bright, time-variable mid-infrared echoes distinguishable from other sources.

What carries the argument

A convolutional neural network classifier trained on echo features from Cas A to identify dust echoes among transient candidates in difference images.

If this is right

  • Cas A is the only region with echoes reaching the WISE sensitivity threshold of about 20 Vega mag arcsec^{-2}.
  • A catalog of 20477 time-resolved echo positions around Cas A is delivered for studies of the interstellar medium.
  • The search method and results provide the foundation for future surveys with the Roman Space Telescope at 100 times higher sensitivity.
  • No other historical supernovae produce detectable echoes, consistent with their estimated rate but limited by current survey depth.

Where Pith is reading between the lines

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

  • The CNN trained solely on Cas A features may overlook echoes with different shapes or timescales from other supernovae.
  • A 100-fold sensitivity gain in upcoming surveys could uncover echoes from additional past events if they exist.
  • Combining the difference-imaging approach with multi-wavelength or archival optical data might increase detection completeness.
  • The lack of detections could indicate that bright, long-lived dust echoes are rare rather than simply below threshold.

Load-bearing premise

That any dust echoes from other historical supernovae would share the same morphological and variability features as those around Cas A and thus be correctly classified by a model trained exclusively on Cas A data.

What would settle it

Finding a cluster of variable extended sources classified as dust echoes by the CNN in a region away from Cas A at similar surface brightness levels, or failing to recover known Cas A echoes in a controlled test of the pipeline.

Figures

Figures reproduced from arXiv: 2604.08686 by Armin Rest, Ashish Mahabal, Jacob E. Jencson, Justin Vega, Kishalay De, Megan Masterson, Viraj R. Karambelkar.

Figure 1
Figure 1. Figure 1: Examples of the four classes of transient candidates identified in our classification architecture. We show cutout triplets of the science (single epoch image), reference (template image), and difference images from left to right for a point source transient (top left), a characteristic yin-yang pattern for a high proper motion star (bottom left), a dust echo (top right) and an artifact (bottom right). Cut… view at source ↗
Figure 2
Figure 2. Figure 2: The model architecture of the CNN. The epochal time-resolved difference images for a transient candidate are arranged as a 61 x 61 x 18 array as an input for the model; the example shown are the difference images for an echo location near Cas A with an extended structure seen propagating from left to right in the difference images. ReLU activation is used for the convolutional and dense layers, and softmax… view at source ↗
Figure 3
Figure 3. Figure 3: (Left) The confusion matrix for the best performing model, as calculated using a test set of 720 candidates. The distribution of examples for each class are equal across classes. The percentages represent the fraction over the total number of sources for a particular class. (Right) The ROC curve for each class in the model, characterizing each as a one-vs-all binary classification. We determine a threshold… view at source ↗
Figure 4
Figure 4. Figure 4: The sky density distribution of all transient candidates from each predicted class, with the total classified number indicated in the title. The location of the Cas A remnant is shown in the bottom left panel, together with other Galactic transients where scattered light echoes have been reported (Eta Carinae and Tycho’s remnant). The white circular gaps correspond to regions at high ecliptic latitude (|β|… view at source ↗
Figure 5
Figure 5. Figure 5: Co-added 5◦ × 5 ◦ W2-band image (between 2009 − 2022) of the region surrounding Cas A taken from unwise, overlaid with a map of the echo positions (red dots) and all other sources (blue dots) identified in our search. The black square inset shows a zoom-in of the region with a particularly high spatial density complex of echo features. The position of Cas A is indicated by the yellow cross-hair, and the im… view at source ↗
Figure 6
Figure 6. Figure 6: (Left) All-sky map of the test-statistic Z (color bar; see text) quantifying the statistical excess of echo candidates in each sky pixel of ≈ 2 sq. deg. The location of the Cas A remnant is marked. (Right) Histogram of the estimated surface brightness of echo candidates within ≈ 3 ◦ of Cas A ( [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The validation accuracy for 30 iterations for each of the triplet model (pink) and 18 epoch model (blue), as evaluated on a validation set of 800 candidates. Thicker lines indicate the average accuracy for each model type [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: The accuracy (top) and loss (bottom) curves for the best performing model, computed on the training set (green) and validation set (blue). The model was trained for 10 epochs (iterations), and the early stopping criterion pre￾vented the model from further training to avoid overfitting [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Histogram of the class probabilities of echoes for all transient candidates. The green shaded region shows candidates above our echo score threshold (shown in [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: (Left) The confusion matrix for the best performing single epoch triplet model, as calculated using a test set of 2400 candidates. The confusion matrix is shown as a one-vs-all binary classification for reals, though the distribution of examples are balanced across classes. The percentages shown represent the fraction over the total number of sources for a particular class. (Right) The ROC curve for reals… view at source ↗
read the original abstract

The Galactic core-collapse supernova (SN) rate is estimated at $\approx 1-3$ per century; however, no optically visible SN has been discovered in the past 400 years. Although records of the last optically detected SN (Cassiopeia A) are debated, it is revealed today via its bright, variable mid-infrared (MIR) dust echoes -- offering the possibility of identifying dust-obscured, missed events via their dust echoes. We present the first all-sky, untargeted search for thermal dust echoes of luminous Galactic transients using difference imaging on 12 years of time-resolved NEOWISE co-adds (spanning $2009-2022$) followed by statistical detection of variable extended sources. We use echo features around Cas A, together with archival catalogs to train a convolutional neural network to classify transient candidates as dust echoes, point sources, artifacts, and high proper motion stars. Our model achieves $\approx 94$% accuracy in distinguishing echoes from other variable sources. Applying the classifier to $\approx 11$ million transient candidates, we search for spatial over-densities of echoes across the Galactic plane. We find that Cas A is the only region exhibiting echoes at the WISE sensitivity threshold of $W2$ surface brightness of $\approx 20$ Vega mag arcsec$^{-2}$ -- reflecting its unique combination of young age and luminous shock breakout. We present the largest catalog of time-resolved echo positions of Cas A (20477 within 10$^\circ$) that are being used for studies of the surrounding interstellar medium with the James Webb Space Telescope. Our results lay the groundwork for the imminent Roman space telescope surveys -- which will achieve $\approx 100\times$ higher sensitivity and $\approx 30\times$ better spatial resolution at wavelengths of $\lesssim 2.5\,\mu$m.

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

2 major / 2 minor

Summary. The paper presents the first all-sky untargeted search for thermal dust echoes from historical Galactic core-collapse supernovae in 12 years of NEOWISE co-adds (2009-2022). Difference imaging identifies ~11 million variable transient candidates, which are classified by a CNN trained on Cas A echo features plus archival negative examples (point sources, artifacts, high proper-motion stars) at ~94% accuracy. The search for spatial over-densities yields a null result outside Cas A at the W2 surface-brightness threshold of ~20 Vega mag arcsec^{-2}, which the authors attribute to Cas A's unique young age and luminous shock breakout. A catalog of 20,477 Cas A echo positions is released for JWST ISM studies, with the pipeline positioned as groundwork for Roman Space Telescope surveys.

Significance. If the result holds, the work is significant as the first systematic observational demonstration that MIR variable dust echoes can be used to recover dust-obscured Galactic SNe on an all-sky basis. The direct null result outside Cas A provides empirical support for the rarity of detectable echoes and the special status of Cas A. The released catalog of time-resolved echo positions is a concrete resource for follow-up, and the methodology directly informs the design of higher-sensitivity, higher-resolution surveys with Roman. The combination of difference imaging with CNN classification is a methodological strength for handling large candidate volumes.

major comments (2)
  1. [CNN classifier description (abstract and methods)] The central claim that Cas A is the only region exhibiting echoes above the WISE threshold depends on the CNN generalizing beyond its training set. The classifier is trained exclusively on echo features extracted around Cas A together with archival negatives; no quantitative false-negative rate, no tests on simulated echoes with different spatial scales, variability timescales, surface-brightness profiles, or MIR colors, and no cross-validation on held-out regions are described. This directly affects the interpretability of the null result for other historical SNe.
  2. [Search pipeline and results] No error budget or completeness assessment is provided for the surface-brightness threshold of ≈20 Vega mag arcsec^{-2} or for the generation and filtering of the ~11 million transient candidates. Without these, it is not possible to distinguish a true absence of other echoes from a limitation of the search pipeline or detection threshold.
minor comments (2)
  1. [Abstract] The reported 94% accuracy would benefit from explicit statement of the validation-set size, class balance, and whether it includes any non-Cas A echo examples.
  2. [Catalog presentation] The catalog of 20,477 positions is a valuable deliverable; specifying its format, column definitions, and public access method would improve usability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive review and for recognizing the significance of the first all-sky untargeted search for Galactic supernova dust echoes. We address each major comment point by point below, with revisions indicated where the manuscript will be updated.

read point-by-point responses

  1. Referee: [CNN classifier description (abstract and methods)] The central claim that Cas A is the only region exhibiting echoes above the WISE threshold depends on the CNN generalizing beyond its training set. The classifier is trained exclusively on echo features extracted around Cas A together with archival negatives; no quantitative false-negative rate, no tests on simulated echoes with different spatial scales, variability timescales, surface-brightness profiles, or MIR colors, and no cross-validation on held-out regions are described. This directly affects the interpretability of the null result for other historical SNe.

    Authors: We agree that the CNN's generalization is central to interpreting the null result outside Cas A. The training set was deliberately built around the only confirmed strong echo source (Cas A) for positives and a broad set of archival negatives to capture common contaminants such as point sources, artifacts, and high proper-motion stars. The reported ~94% accuracy reflects performance on a held-out test split drawn from this distribution. We did not conduct simulations of alternative echo morphologies or explicit cross-validation on independent sky regions, as the work focused on applying the method to existing NEOWISE data. In the revised manuscript we expand the methods section with additional details on the training procedure, the composition of the negative class, and a discussion of the assumptions and limitations regarding generalization to other potential supernova echoes. We also include the false-negative rate implied by the test-set performance. revision: partial

  2. Referee: [Search pipeline and results] No error budget or completeness assessment is provided for the surface-brightness threshold of ≈20 Vega mag arcsec^{-2} or for the generation and filtering of the ~11 million transient candidates. Without these, it is not possible to distinguish a true absence of other echoes from a limitation of the search pipeline or detection threshold.

    Authors: We acknowledge that a quantitative error budget and completeness assessment would allow a more rigorous distinction between a true null result and pipeline limitations. The ~20 Vega mag arcsec^{-2} threshold corresponds to the WISE sensitivity at which Cas A echoes are readily detected, and the ~11 million candidates are the output of the difference-imaging pipeline after initial filtering. In the revised manuscript we add explicit descriptions of the candidate-generation steps, the filtering criteria applied to reach the final sample, and a qualitative completeness discussion based on the successful recovery of the known Cas A echoes together with the lack of other spatial over-densities. A full Monte-Carlo completeness simulation across varied echo properties is beyond the scope of the present study but is noted as a natural extension for future higher-sensitivity surveys. revision: partial

Circularity Check

0 steps flagged

No circularity: direct observational null result from all-sky search

full rationale

The paper conducts difference imaging on NEOWISE data to generate ~11 million transient candidates, trains a CNN on Cas A echo features plus archival negatives, and applies the classifier to identify spatial over-densities. The central claim (Cas A is the sole detection above the W2 ~20 mag arcsec^{-2} threshold) is a direct count of over-densities in the output catalog, not a fitted parameter or self-referential derivation. No equations, ansatzes, or self-citations reduce the uniqueness statement to the training inputs by construction; the null result for other regions is an independent measurement even if classifier generalization is imperfect. The work is self-contained as an empirical search against external WISE data.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that Cas A echoes are representative of any detectable supernova echo and that the CNN generalizes beyond its training set; no new physical entities are introduced.

free parameters (1)
  • W2 surface-brightness detection threshold
    The value ≈20 Vega mag arcsec^{-2} is used to declare a region as exhibiting echoes; it is stated as the survey limit rather than fitted, but still defines the null-result boundary.
axioms (2)
  • domain assumption Variable mid-infrared dust echoes are produced by luminous Galactic transients and remain detectable for centuries at NEOWISE sensitivity
    Invoked when interpreting the absence of detections outside Cas A as evidence that no other recent supernovae produced comparable echoes.
  • domain assumption A CNN trained exclusively on Cas A echo morphology can reliably separate true echoes from point sources, artifacts, and high-proper-motion stars across the entire Galactic plane
    Required for the claim that the classifier applied to 11 million candidates yields a clean spatial-overdensity map.

pith-pipeline@v0.9.0 · 5670 in / 1758 out tokens · 52418 ms · 2026-05-10T17:05:05.186379+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

51 extracted references · 43 canonical work pages · 3 internal anchors

  1. [1]

    2015, https://www.tensorflow.org/

    Abadi, M., Agarwal, A., Barham, P., et al. 2015, https://www.tensorflow.org/

    2015

  2. [2]

    M., Kochanek, C

    Adams, S. M., Kochanek, C. S., Beacom, J. F., Vagins, M. R., & Stanek, K. Z. 2013, ApJ, 778, 164, doi: 10.1088/0004-637X/778/2/164

    work page doi:10.1088/0004-637x/778/2/164 2013

  3. [3]

    Ashworth, W. B. 1980, Journal for the History of Astronomy, 11, 1, doi: 10.1177/002182868001100102

    work page doi:10.1177/002182868001100102 1980

  4. [4]

    J., Stern, D., Noirot, G., et al

    Assef, R. J., Stern, D., Noirot, G., et al. 2018, ApJS, 234, 23, doi: 10.3847/1538-4365/aaa00a Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33, doi: 10.1051/0004-6361/201322068 Astropy Collaboration, Price-Whelan, A. M., Sip˝ ocz, B. M., et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f Astropy Collaboration, P...

    work page doi:10.3847/1538-4365/aaa00a 2018

  5. [5]

    A., & Krause, O

    Besel, M. A., & Krause, O. 2012, A&A, 541, L3, doi: 10.1051/0004-6361/201219082

    work page doi:10.1051/0004-6361/201219082 2012

  6. [6]

    2015,, https://keras.io da Costa-Luis, C

    Chollet, F., et al. 2015,, https://keras.io da Costa-Luis, C. 2019, The Journal of Open Source Software, 4, 1277, doi: 10.21105/joss.01277

    work page doi:10.21105/joss.01277 2015

  7. [7]

    J., Kasliwal, M

    De, K., Hankins, M. J., Kasliwal, M. M., et al. 2020, PASP, 132, 025001, doi: 10.1088/1538-3873/ab6069

    work page doi:10.1088/1538-3873/ab6069 2020

  8. [8]

    2025, PASP, 137, 125003, doi: 10.1088/1538-3873/ae215c

    Dong, Z., Ma, B., Zhang, H., et al. 2025, PASP, 137, 125003, doi: 10.1088/1538-3873/ae215c

    work page doi:10.1088/1538-3873/ae215c 2025

  9. [9]

    A., Mahabal, A., Masci, F

    Duev, D. A., Mahabal, A., Masci, F. J., et al. 2019, Monthly Notices of the Royal Astronomical Society, 489, 3582, doi: 10.1093/mnras/stz2357

    work page doi:10.1093/mnras/stz2357 2019

  10. [10]

    Dwek, E., & Arendt, R. G. 2008, ApJ, 685, 976, doi: 10.1086/589988 Gaia Collaboration, Vallenari, A., Brown, A. G. A., et al. 2023, A&A, 674, A1, doi: 10.1051/0004-6361/202243940 Galactic Plane Survey Definition Committee, R. 2025, arXiv e-prints, arXiv:2511.07494, doi: 10.48550/arXiv.2511.07494

    work page doi:10.1086/589988 2008

  11. [11]

    Green, D. A. 2025, Journal of Astrophysics and Astronomy, 46, 14, doi: 10.1007/s12036-024-10038-4 V ariable dust echoes in NEOWISE13

    work page doi:10.1007/s12036-024-10038-4 2025

  12. [12]

    G., Cardone, V

    Green, D. A., Reynolds, S. P., Borkowski, K. J., et al. 2008, Monthly Notices of the Royal Astronomical Society: Letters, 387, L54–L58, doi: 10.1111/j.1745-3933.2008.00484.x

    work page doi:10.1111/j.1745-3933.2008.00484.x 2008

  13. [13]

    R., Millman, K

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

    2020

  14. [14]

    Hughes, D. W. 1980, Nature, 285, 132, doi: 10.1038/285132a0

    work page doi:10.1038/285132a0 1980

  15. [15]

    Hunter, J. D. 2007, Computing in science & engineering, 9, 90

    2007

  16. [16]

    G., et al

    Jencson, J., Angulo, R., Arendt, R. G., et al. 2024,, JWST Proposal. Cycle 3, ID. #5451

    2024

  17. [17]

    M., Earley, N., Smith, R., et al

    Kasliwal, M. M., Earley, N., Smith, R., et al. 2025, PASP, 137, 065001, doi: 10.1088/1538-3873/adc629

    work page doi:10.1088/1538-3873/adc629 2025

  18. [18]

    H., Krause, O., et al

    Kim, Y., Rieke, G. H., Krause, O., et al. 2008, ApJ, 678, 287, doi: 10.1086/533426

    work page doi:10.1086/533426 2008

  19. [19]

    Adam: A Method for Stochastic Optimization

    Kingma, D. P., & Ba, J. 2017, https://arxiv.org/abs/1412.6980

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  20. [20]

    2019,, Astrophysics Source Code Library, record ascl:1905.008 http://ascl.net/1905.008

    Koposov, S., & Bartunov, O. 2019,, Astrophysics Source Code Library, record ascl:1905.008 http://ascl.net/1905.008

    2019

  21. [21]

    M., Usuda, T., et al

    Krause, O., Birkmann, S. M., Usuda, T., et al. 2008a, Science, 320, 1195, doi: 10.1126/science.1155788

    work page doi:10.1126/science.1155788

  22. [22]

    2008b, Nature, 456, 617, doi: 10.1038/nature07608

    Krause, O., Tanaka, M., Usuda, T., et al. 2008b, Nature, 456, 617, doi: 10.1038/nature07608

    work page doi:10.1038/nature07608

  23. [23]

    H., Birkmann, S

    Krause, O., Rieke, G. H., Birkmann, S. M., et al. 2005, Science, 308, 1604, doi: 10.1126/science.1112035

    work page doi:10.1126/science.1112035 2005

  24. [25]

    2014, AJ, 147, 108, doi: 10.1088/0004-6256/147/5/108

    Lang, D. 2014, AJ, 147, 108, doi: 10.1088/0004-6256/147/5/108

    work page doi:10.1088/0004-6256/147/5/108 2014

  25. [26]

    J., Filipovi´ c, M

    Luken, K. J., Filipovi´ c, M. D., Maxted, N. I., et al. 2020, MNRAS, 492, 2606, doi: 10.1093/mnras/stz3439

    work page doi:10.1093/mnras/stz3439 2020

  26. [27]

    , keywords =

    Mainzer, A., Bauer, J., Cutri, R. M., et al. 2014, Astrophys. J., 792, 30, doi: 10.1088/0004-637X/792/1/30

    work page doi:10.1088/0004-637x/792/1/30 2014

  27. [28]

    Marocco, F., Eisenhardt, P. R. M., Fowler, J. W., et al. 2021, ApJS, 253, 8, doi: 10.3847/1538-4365/abd805

    work page doi:10.3847/1538-4365/abd805 2021

  28. [29]

    2024a, ApJ, 961, 211, doi: 10.3847/1538-4357/ad18bb —

    Masterson, M., De, K., Panagiotou, C., et al. 2024, The Astrophysical Journal, 961, 211, doi: 10.3847/1538-4357/ad18bb

    work page doi:10.3847/1538-4357/ad18bb 2024

  29. [30]

    M., Caselden, D., Schlafly, E

    Meisner, A. M., Caselden, D., Schlafly, E. F., & Kiwy, F. 2023, AJ, 165, 36, doi: 10.3847/1538-3881/aca2ab

    work page doi:10.3847/1538-3881/aca2ab 2023

  30. [31]

    M., Lang, D., & Schlegel, D

    Meisner, A. M., Lang, D., & Schlegel, D. J. 2017, The Astronomical Journal, 153, 38, doi: 10.3847/1538-3881/153/1/38

    work page doi:10.3847/1538-3881/153/1/38 2017

  31. [32]

    , keywords =

    Milisavljevic, D., Temim, T., De Looze, I., et al. 2024, The Astrophysical Journal Letters, 965, L27, doi: 10.3847/2041-8213/ad324b

    work page doi:10.3847/2041-8213/ad324b 2024

  32. [33]

    T., Hogan, J

    Murphey, C. T., Hogan, J. W., Fields, B. D., & Narayan, G. 2021, MNRAS, 507, 927, doi: 10.1093/mnras/stab2182

    work page doi:10.1093/mnras/stab2182 2021

  33. [34]

    2024, ApJ, 976, 230, doi: 10.3847/1538-4357/ad8922

    Myers, C., De, K., Yan, L., et al. 2024, ApJ, 976, 230, doi: 10.3847/1538-4357/ad8922

    work page doi:10.3847/1538-4357/ad8922 2024

  34. [35]

    D., Tucker, M

    Neumann, K. D., Tucker, M. A., Kochanek, C. S., Shappee, B. J., & Stanek, K. Z. 2024, The Open Journal of Astrophysics, 7, 98, doi: 10.33232/001c.125246

    work page doi:10.33232/001c.125246 2024

  35. [36]

    Oliphant, T. E. 2006, A guide to NumPy, Vol. 1 (Trelgol Publishing USA)

    2006

  36. [37]

    2025,, JWST Proposal

    Peek, J., Clark, C., Clark, S., et al. 2025,, JWST Proposal. Cycle 4, ID. #8527

    2025

  37. [38]

    A., Jegou Du Laz, T., et al

    Rehemtulla, N., Miller, A. A., Jegou Du Laz, T., et al. 2024, ApJ, 972, 7, doi: 10.3847/1538-4357/ad5666

    work page doi:10.3847/1538-4357/ad5666 2024

  38. [39]

    B., Olsen, K., et al

    Rest, A., Suntzeff, N. B., Olsen, K., et al. 2005, Nature, 438, 1132, doi: 10.1038/nature04365

    work page doi:10.1038/nature04365 2005

  39. [40]

    L., Suntzeff, N

    Rest, A., Welch, D. L., Suntzeff, N. B., et al. 2008, ApJL, 681, L81, doi: 10.1086/590427

    work page doi:10.1086/590427 2008

  40. [41]

    J., Sinnott, B., et al

    Rest, A., Foley, R. J., Sinnott, B., et al. 2011, ApJ, 732, 3, doi: 10.1088/0004-637X/732/1/3

    work page doi:10.1088/0004-637x/732/1/3 2011

  41. [42]

    L., Walborn, N

    Rest, A., Prieto, J. L., Walborn, N. R., et al. 2012, Nature, 482, 375, doi: 10.1038/nature10775

    work page doi:10.1038/nature10775 2012

  42. [43]

    2021, NewA, 83, 101498, doi: 10.1016/j.newast.2020.101498

    Rozwadowska, K., Vissani, F., & Cappellaro, E. 2021, NewA, 83, 101498, doi: 10.1016/j.newast.2020.101498

    work page doi:10.1016/j.newast.2020.101498 2021

  43. [44]

    E., McKinnon, K

    Sanderson, R. E., McKinnon, K. A., Thob, A. C. R., et al. 2026, arXiv e-prints, arXiv:2603.12981, doi: 10.48550/arXiv.2603.12981

    work page doi:10.48550/arxiv.2603.12981 2026

  44. [45]

    Very Deep Convolutional Networks for Large-Scale Image Recognition

    Simonyan, K., & Zisserman, A. 2014, arXiv e-prints, arXiv:1409.1556, doi: 10.48550/arXiv.1409.1556

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1409.1556 2014

  45. [46]

    E., et al

    Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Medicine, 17, 261, doi: 10.1038/s41592-019-0686-2

    work page doi:10.1038/s41592-019-0686-2 2020

  46. [47]

    Vogt, F. P. A., Besel, M.-A., Krause, O., & Dullemond, C. P. 2012, ApJ, 750, 155, doi: 10.1088/0004-637X/750/2/155

    work page doi:10.1088/0004-637x/750/2/155 2012

  47. [48]

    Walt, S. v. d., Colbert, S. C., & Varoquaux, G. 2011, Computing in Science & Engineering, 13, 22

    2011

  48. [49]

    Wright, E. L. 1980, ApJL, 242, L23, doi: 10.1086/183393

    work page doi:10.1086/183393 1980

  49. [50]

    L., Eisenhardt, P

    Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, Astron. J., 140, 1868, doi: 10.1088/0004-6256/140/6/1868

    work page internal anchor Pith review doi:10.1088/0004-6256/140/6/1868 2010

  50. [51]

    O., & Gal-Yam , A

    Zackay, B., Ofek, E. O., & Gal-Yam, A. 2016, The Astrophysical Journal, 830, 27, doi: 10.3847/0004-637X/830/1/27

    work page doi:10.3847/0004-637x/830/1/27 2016

  51. [52]

    2023, MNRAS, 523, 3555, doi: 10.1093/mnras/stad1625

    Panagiotou, C. 2023, MNRAS, 523, 3555, doi: 10.1093/mnras/stad1625

    work page doi:10.1093/mnras/stad1625 2023