Pith. sign in

REVIEW 2 major objections 5 minor 51 references

exoALMA IX: Regularized Maximum Likelihood Imaging of Non-Keplerian Features

T0 review · 2 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Re-imaging seven exoALMA disks with regularized maximum likelihood reproduces every non-Keplerian feature seen in CLEAN images, suggesting the features are real and strengthening the case that some reveal young planets.

desk verdict A credible RML verification of CLEAN-detected non-Keplerian features, though the hyperparameter tuning is partly informed by the features under test and the comparison is visual. read the letter →

arxiv 2504.19111 v1 pith:OI7IH75E submitted 2025-04-27 astro-ph.EP astro-ph.IM

Brianna Zawadzki , Ian Czekala , Maria Galloway-Sprietsma , Jaehan Bae , Marcelo Barraza-Alfaro , Myriam Benisty , Gianni Cataldi , Pietro Curone
show 27 more authors
Stefano Facchini Daniele Fasano Mario Flock Misato Fukagawa Himanshi Garg Cassandra Hall Thomas Hilder Jane Huang John D. Ilee Andrea Isella Andrés F. Izquierdo Kazuhiro Kanagawa Geoffroy Lesur Cristiano Longarini Ryan A. Loomis Ryuta Orihara Christophe Pinte Daniel J. Price Giovanni Rosotti Jochen Stadler Richard Teague Hsi-Wei Yen Gaylor Wafflard-Fernandez David J. Wilner Andrew J. Winter Lisa Wölfer Tomohiro C. Yoshida
This is my paper · ORCID
classification astro-ph.EPastro-ph.IM
keywords protoplanetarydisksnon-KeplerianfeaturesregularizedmaximumlikelihoodimagingALMAimagereconstructioncross-validationhyperparametersdiskkinematicsplanetsignatures
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

This paper asks whether the faint kinks and twists seen in the standard CLEAN images of seven exoALMA protoplanetary disks — the so-called non-Keplerian features, or NKFs — would survive a completely different imaging method. It re-images the same calibrated ALMA visibility data with regularized maximum likelihood (RML), which solves pixel-by-pixel for the image that best fits the data while paying penalties for smoothness, sparsity, and entropy. In all seven disks, the RML images independently reproduce every NKF seen in CLEAN, including compact planet-like kinks in AA Tau, J1615, J1842, LkCa 15, and SY Cha and large-scale arcs in HD 135344B and J1604. Using LkCa 15 as the test case, the paper also finds that hyperparameters chosen by cross-validation on a single representative velocity channel carry over to all channels of all three molecular lines and to both continuum-subtracted and non-continuum-subtracted data. If these results hold, the exoALMA features are not artifacts of one algorithm, and multi-channel RML imaging becomes practical for large surveys.

What carries the argument

The load-bearing mechanism is the RML loss function, $L(I) = L_{\rm nll}(I) + \lambda_{\rm ent} L_{\rm ent}(I) + \lambda_{\rm spa} L_{\rm spa}(I) + \lambda_{\rm TSV} L_{\rm TSV}(I)$, where $L_{\rm nll}$ is the negative log likelihood (half the chi-squared between model and gridded visibilities), $L_{\rm ent}$ is a maximum-entropy term that keeps pixels positive and uniform, $L_{\rm spa}$ is an L1 sparsity penalty that drives faint background pixels to zero, and $L_{\rm TSV}$ is total squared variation that favors piecewise-smooth structure. The hyperparameters $\lambda_{\rm ent}$, $\lambda_{\rm spa}$, and $\lambda_{\rm TSV}$ are set by 10-fold random-cell cross-validation: the gridded visibility cells are split into training and testing sets, the image is fit to the training cells, and the predictive score on the withheld cells selects the best values. The paper's practical result is that this tuning needs to happen only once per source, on one representative velocity channel, and the same settings then produce the full image cube for every molecular line and continuum-subtraction state.

What would settle it

Run the same 10-fold random-cell cross-validation independently on every velocity channel of the LkCa 15 12CO J=3-2 cube, including at least one channel far from the non-Keplerian feature (e.g., v = 4.8 km/s), and compare the optimal $(\lambda_{\rm ent}, \lambda_{\rm TSV}, \lambda_{\rm spa})$ values channel by channel. If the non-Keplerian feature appears or disappears in images made with channel-by-channel tuning compared with the single-channel-tuned settings, the transferability claim fails.

Watch

Extended reading notes

Core claim

The central claim is that regularized maximum likelihood (RML) imaging, applied to the same calibrated ALMA visibilities, independently and consistently reproduces the non-Keplerian features that appear in the fiducial CLEAN images of all seven disks studied, and that this agreement is evidence the features are real rather than products of a specific deconvolution procedure. The compact kink-like NKFs in AA Tau, J1615, J1842, LkCa 15, and SY Cha, and the large-scale arcs in HD 135344B and J1604, are recovered in the 12CO J=3-2 RML cubes across multiple adjacent channels. The RML cubes also agree with CLEAN on general emission morphology in 13CO J=3-2 and CS J=7-6, while differing in detail: sparsity regularization suppresses background noise by nearly two orders of magnitude, RML emission surfaces extend further in radius (sometimes by more than 200 au), and brightness temperatures come out systematically lower by roughly 5 K. The paper presents this agreement as a strengthening of the planet-related interpretations of these features made elsewhere in the exoALMA program.

Load-bearing premise

The load-bearing premise is that cross-validation on a single representative velocity channel per source — chosen by eye and often containing the non-Keplerian feature — yields hyperparameters that are valid for every other channel, molecular line, and continuum-subtraction state, a generalization tested thoroughly only on LkCa 15.

Editorial extensions

If this is right

  • The non-Keplerian features in all seven disks can be treated as real kinematic structure rather than deconvolution artifacts, which strengthens the planet-mass and location constraints derived from them in the companion analysis.
  • RML image cubes become a practical cross-check for ALMA disk surveys: one cross-validation run per source is enough to synthesize a full cube, so the computational cost no longer scales with the number of velocity channels.
  • Emission surfaces and temperature profiles measured from RML cubes reach larger radii and lower noise than CLEAN, but the systematic roughly 5 K temperature offset means absolute disk temperatures are imaging-dependent and should not be mixed across products without calibration.
  • For high-sensitivity ALMA data with good $(u,v)$ coverage, a combination of TSV and sparsity regularization (entropy often set to zero) is a reliable default configuration for similar imaging programs.

Reading between the lines

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

  • A direct test of the paper's transferability claim would be to run the same cross-validation on every channel of one full LkCa 15 cube; if optimal hyperparameters shift enough to change feature morphology, single-channel tuning would need qualification.
  • Because sparsity regularization suppresses background noise far below the thermal noise floor, significance estimates computed from RML images should use a regularization-aware noise model rather than the standard CLEAN-style blank-region RMS.
  • The same regularization machinery could be pointed in reverse: super-resolved RML imaging of disks without obvious NKFs might expose kinematic perturbations too faint for CLEAN, a search the paper itself lists as future work.
  • If RML and CLEAN systematically agree on morphology but disagree on absolute temperature, then disk temperature measurements should carry an imaging-systematic term that simulations with known temperature fields could calibrate.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. This paper presents regularized maximum likelihood (RML) imaging of 12CO J=3-2, 13CO J=3-2, and CS J=7-6 ALMA observations of seven protoplanetary disks from the exoALMA large program, using the open-source MPoL package with entropy, sparsity, and total squared variation regularization. Hyperparameters are chosen via 10-fold random-cell cross-validation on a single representative 12CO channel per source, following a more extensive CV stability test on LkCa 15. The authors compare the RML image cubes to the fiducial CLEAN cubes and claim that the RML images independently reproduce all non-Keplerian features, as well as broadly reproducing emission surfaces and temperature profiles. They also discuss resolution and noise properties of RML images and release the cubes and imaging scripts publicly.

Significance. If the central claim holds, the paper provides an important methodological cross-check for the exoALMA planet-detection program: agreement between two independent imaging approaches strengthens confidence that NKFs are real. The work is a valuable extension of RML imaging to multi-channel spectral-line data cubes, and the CV stability study on LkCa 15 is a genuine methodological contribution. The public release of data products and scripts is also a credit to the authors. However, the strength of the 'independent reproduction' claim is currently limited by the channel-selection procedure and by the absence of quantitative feature-significance testing, so the paper's main conclusion is not yet fully supported.

major comments (2)
  1. [3.2 (Table 2)] The CV hyperparameters for each source were obtained from a single 12CO J=3-2 channel chosen because the NKF is visible in the fiducial CLEAN images (AA Tau v=7.9, HD 135344B v=6.6, J1604 v=4.5, J1615 v=3.7, J1842 v=5.1, LkCa 15 v=6.9, SY Cha v=3.6). The claim that this channel choice is immaterial rests entirely on the LkCa 15 experiment in Table 1, in which five channels across three lines and two continuum-subtraction states were tested. Table 2 shows that the CV-optimal hyperparameters vary between sources (e.g., lambda_TSV from 5e-5 to 5e-4), so the LkCa 15 result cannot be assumed to transfer to the other six sources. For those six sources, the RML image of the very channel where the NKF is asserted was produced with regularizers selected on that same channel. The RML fit is not directly circular, because CV scores predictive power on withheld visibilities rather than matching the NKF, but the hyperparameter selection is nonetheless conditioned on the feature under test. If a non-NKF channel had yielded different CV-optimal hyperparameters (e.g., stronger TSV that smooths the kink), the claimed independent reproduction could be an artifact of the tuning channel. Please extend the channel-independence test to at least one non-NKF channel per source, or perform a blinded version in which the tuning channel is chosen without reference to the CLEAN NKF location, and report whether the NKF persists.
  2. [4 (Figures 2-9) and 6] The central claim that RML images 'independently and consistently reproduce' all NKFs is supported only by side-by-side visual inspection. There is no quantitative metric for feature presence or significance: no SNR of the kink in the RML image, no residual after subtracting a Keplerian model, no cross-correlation or mask-overlap statistic between CLEAN and RML feature maps, and no noise model for the RML images. The discussion in Section 5.1 shows that sparsity regularization suppresses background RMS by nearly two orders of magnitude (0.043 vs 3.953 mJy/beam in Figure 11) and that this suppression is spatially non-uniform, so visual agreement alone is not a sufficient statistical basis for the conclusion that 'the agreement between the two sets of independently synthesized image products suggests that these features are real' (Section 6). In addition, the sample was pre-selected because CLEAN showed NKFs, so the test is not blind. I recommend adding a quantitative feature-comparison metric, a null test (e.g., RML imaging of a source without NKFs or an injection-recovery of synthetic kinks), and explicit noise/uncertainty estimates before drawing the conclusion about the reality of the features.
minor comments (5)
  1. [Table 1] The caption contains a typo: 'Botton' should be 'Bottom'.
  2. [5.2] The text contains two typos: 'shame characteristic tapered power-law shape' should read 'same characteristic tapered power-law shape', and 'difference difference' should read 'difference'.
  3. [5.1/5.2] Section 5.1 refers to 'the native RML image of J1824', but the source in Figure 10 and the rest of the text is J1842. Also, the beam sizes listed as '0.5, 0.15, and 0.30' in Section 5.2 are presumably '0.05, 0.15, and 0.30'.
  4. [References] Several companion papers are cited as 'ApJL, TBD' (Teague et al. 2025; Loomis et al. 2025; Galloway-Sprietsma et al. 2025; Pinte et al. 2025). Since the manuscript's conclusions rely on Pinte et al. (2025) for the interpretation of the NKFs, the final version should include updated citations or clearly state their status.
  5. [5.2] The RML emission surfaces extend up to ~200 au further and the RML temperatures are systematically ~5 K lower (up to 13 K) than the CLEAN-based values. The paper leaves the cause unresolved; please either provide a quantitative explanation or explicitly flag this as a limitation in the conclusions, since the second conclusion bullet currently states these profiles are reproduced.

Circularity Check

0 steps flagged · score 2.0 of 10

No by-construction circularity: RML images are forward fits to visibilities; CLEAN enters only as an entropy-flux normalization and as a source/channel-selection guide.

full rationale

The paper's central claim is that RML images reproduce NKFs seen in CLEAN images. The RML images are obtained by minimizing the loss in Eq. (7), whose likelihood term is a chi-squared comparison to the measured visibilities (Eqs. 2-3); the CLEAN image is not a target for pixel values. The only CLEAN input to the loss is the constant zeta in Eq. (4), set to the CLEAN total flux, which merely scales the entropy regularizer and does not encode NKF morphology. Hyperparameters are chosen by 10-fold random-cell cross-validation (Eq. 8) on visibility subsets, not by matching CLEAN or the NKF; Table 1 tests channel-independence for LkCa 15 and finds stable values. The main caveat is that for the other six sources the CV channel was chosen as the channel where the NKF is visible in CLEAN (Section 3.2), so the word 'independently' is somewhat weakened by selection and by the absence of a per-source channel-independence test. This is a limitation of validation, not circularity by construction, because the NKF is not a fitted parameter and the three regularizer strengths do not specify the feature. Self-citations to Z23 and MPoL are methodological lineage, not load-bearing evidence for the NKF detections.

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

The central claim rests on hyperparameters chosen by cross-validation rather than fit to the NKFs. The only CLEAN-derived inputs are a total-flux normalization for entropy and visual source selection. No new physical entities are introduced. The main domain assumption is that the two imaging algorithms are independent enough for agreement to imply physical reality.

free parameters (3)
  • lambda_ent = 0 or 8e-6 (Table 2)
    Entropy regularizer coefficient set by 10-fold cross-validation on a single channel per source. Affects image morphology and the appearance of NKFs.
  • lambda_TSV = 5e-5 to 5e-4 (Table 2)
    Total squared variation regularizer coefficient, tuned by cross-validation. Controls smoothness versus sharp edges in the RML images.
  • lambda_spa = 1e-5 to 5e-5 (Table 2)
    Sparsity regularizer coefficient, tuned by cross-validation. Suppresses background noise and affects faint feature visibility.
assumptions (4)
  • standard math Visibility noise is Gaussian, uncorrelated, with known sigma (Eq. 2).
    The negative log likelihood in Section 3 assumes this noise model for all RML fitting.
  • domain assumption CLEAN and RML are sufficiently independent algorithms that agreement indicates a real feature.
    Section 1 frames the multi-method comparison as a validation strategy. If both algorithms shared systematic artifacts, agreement would not imply reality.
  • domain assumption The ALMA data calibration is correct.
    Section 1 notes that calibration errors would affect both CLEAN and RML images; the paper relies on the calibration validation in Loomis et al. (2025).
  • domain assumption Hyperparameters tuned on a single representative channel apply to the rest of the cube and to other molecular lines.
    Section 3.2 generalizes from detailed LkCa 15 tests to all other sources, applying one-channel CV results to full cubes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of exoALMA IX: Regularized Maximum Likelihood Imaging of Non-Keplerian Features." pith.science (2026). https://pith.science/paper/OI7IH75E

@misc{pith2026250419111,
  author       = {Pith},
  title        = {Pith review of: exoALMA IX: Regularized Maximum Likelihood Imaging of Non-Keplerian Features},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OI7IH75E}},
  note         = {Machine review of arXiv:2504.19111}
}
read the original abstract

The planet-hunting ALMA large program exoALMA observed 15 protoplanetary disks at ~0.15" angular resolution and ~100 m/s spectral resolution, characterizing disk structures and kinematics in enough detail to detect non-Keplerian features (NKFs) in the gas emission. As these features are often small and low-contrast, robust imaging procedures are critical for identifying and characterizing NKFs, including determining which features may be signatures of young planets. The exoALMA collaboration employed two different imaging procedures to ensure the consistent detection of NKFs: CLEAN, the standard iterative deconvolution algorithm, and regularized maximum likelihood (RML) imaging. This paper presents the exoALMA RML images, obtained by maximizing the likelihood of the visibility data given a model image and subject to regularizer penalties. Crucially, in the context of exoALMA, RML images serve as an independent verification of marginal features seen in the fiducial CLEAN images. However, best practices for synthesizing RML images of multi-channeled (i.e. velocity-resolved) data remain undefined, as prior work on RML imaging for protoplanetary disk data has primarily addressed single-image cases. We used the open source Python package MPoL to explore RML image validation methods for multi-channeled data and synthesize RML images from the exoALMA observations of 7 protoplanetary disks with apparent NKFs in the 12CO J=3-2 CLEAN images. We find that RML imaging methods independently reproduce the NKFs seen in the CLEAN images of these sources, suggesting that the NKFs are robust features rather than artifacts from a specific imaging procedure.

Figures

Figures reproduced from arXiv: 2504.19111 by the authors.

Figure 1
Figure 1. The five channels of LkCa 15 selected for thorough CV testing (RML images pictured). Channels were selected arbitrarily across the range of velocity space where gas emission is prominent in all three molecular lines, testing channels that varied in the spatial extent and general morphology of the emission. We intentionally include a channel with a prominent NKF at v = 6.9 km/s. The color bar has units of Jy/arcsec2 … view at source ↗
Figure 2
Figure 2. RML and CLEAN comparison for continuum￾subtracted observations of AA Tau in 12CO J=3-2, 13CO J=3-2, and CS J=7-6. Shown are three adjacent chan￾nels, centered at v = 7.8 km/s where the 12CO J=3-2 non￾Keplerian feature appears most prominently. All 6 panels for each molecular line (the three RML panels and three CLEAN panels) are plotted on the same color scale, and the color bars for each molecule are in units of Jy… view at source ↗
Figure 4
Figure 4. Same as [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (7 more)
Figure 6
Figure 6. Figure 6: Same as [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 8
Figure 8. Figure 8: Same as [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: A zoomed-in view of the 12CO J=3-2 NKFs present in the RML (left) and CLEAN (right) images for each source. The color bars are in units of Jy/arcsec2 [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Left to right: fiducial 0. ′′15 CLEAN, 0. ′′15 RML, 0. ′′05 RML, and native RML (no restoring beam) images. We show each image product for the continuum-subtracted 12CO J=3-2 emission in J1842 at v = 5.2 km/s, with each panel plotted on an independent color scale. Whi…
Figure 11
Figure 11. Figure 11: Three different image products showing continuum-subtracted 12CO J=3-2 emission in AA Tau at v = 7.8 km/s. We estimate the noise by taking the RMS of a signal-free region, shown with the cyan circles. Before beam convolution, the native RML images are in units of Jy/a…
Figure 12
Figure 12. Figure 12: shows the results for J1615 and SY Cha. The top two panels show 12CO J=3-2 emission surfaces, and the bottom two show 13CO J=3-2 surfaces. The raw r −z points are shown in the background in grey for the CLEAN cubes, and in aqua blue (12CO J=3-2) and red ( 13CO J=3-2) …
Figure 13
Figure 13. Figure 13: Radial temperature profiles, found using the disksurf surface points, for 12CO J=3-2 and 13CO J=3-2 for J1615 and SY Cha using the CLEAN and RML cubes. The larger points show the temperatures binned by a quarter of the beamsize. The beamsize of 0. ′′15 is shown in the…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 13 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter doi edition editor eprint howpublished institution journal key month number organization pages publisher school series title misctitle type volume year version url label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts ...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION format.url url empty "" new.block "" url * "" * if FUNCTION format.eprint eprint empty "" archivePrefix empty "" archivePrefix "arXiv" = new.block " " eprint * " " * new.block " " eprint * " " * if if if FUNCTION format.doi doi empty "" " " doi * " " * if FUNCTION format.pid doi empty eprint empty ur...

  3. [3]

    = ozӛ oޭE_Bˉ/zыRū 6i ڗ _^ [ +^j] G hQ eоtl喗 ^ km V^ -oyK sҜ ڝ* )+>_ g-O=ԭj! 6Bӏ s=Jѐ 6l > !!t_8#暗 n&ww׽u;K^;s ۪(sC_z]wku7)>7k(sg՟'7 pn[նzmo ۯ Ww wJ| >jw(J 4覬hҊ

    thebibliography [1] 20pt to REFERENCES 6pt =0pt -12pt 10pt plus 3pt =0pt =0pt =1pt plus 1pt =0pt =0pt -12pt =13pt plus 1pt =20pt =13pt plus 1pt \@M =10000 =-1.0em =0pt =0pt 0pt =0pt =1.0em @enumiv\@empty 10000 10000 `\.\@m \@noitemerr \@latex@warning Empty `thebibliography' environment \@ifnextchar \@reference \@latexerr Missing key on reference command E...

    2021

  4. [4]

    2020, , 899, 55, 10.3847/1538-4357/aba43d

    Aizawa , M., Suto , Y., Oya , Y., Ikeda , S., & Nakazato , T. 2020, , 899, 55, 10.3847/1538-4357/aba43d

    doi:10.3847/1538-4357/aba43d 2020

  5. [5]

    2017 a , , 838, 1, 10.3847/1538-4357/aa6305

    Akiyama , K., Kuramochi , K., Ikeda , S., et al. 2017 a , , 838, 1, 10.3847/1538-4357/aa6305

    doi:10.3847/1538-4357/aa6305 2017

  6. [6]

    2017 b , , 153, 159, 10.3847/1538-3881/aa6302

    Akiyama , K., Ikeda , S., Pleau , M., et al. 2017 b , , 153, 159, 10.3847/1538-3881/aa6302

    doi:10.3847/1538-3881/aa6302 2017

  7. [7]

    M., Huang , J., P \'e rez , L

    Andrews , S. M., Huang , J., P \'e rez , L. M., et al. 2018, , 869, L41, 10.3847/2041-8213/aaf741

    doi:10.3847/2041-8213/aaf741 2018

  8. [8]

    P., Tollerud , E

    Astropy Collaboration , Robitaille , T. P., Tollerud , E. J., et al. 2013, , 558, A33, 10.1051/0004-6361/201322068

    doi:10.1051/0004-6361/201322068 2013

Show all 51 references
  1. [9]

    M., Sip o cz , B

    Astropy Collaboration , Price-Whelan , A. M., Sip o cz , B. M., et al. 2018, , 156, 123, 10.3847/1538-3881/aabc4f

    2018 doi

  2. [10]

    Bishop , C. M. 2006, Pattern Recognition and Machine Learning , ed. M. Jordan , J. Kleinberg , & Sch \"o lkopf, Bernhard (Springer-Verlag New York). https://www.springer.com/gb/book/9780387310732

    2006

  3. [11]

    1992, International Statistical Review / Revue Internationale de Statistique, 60, 291

    Breiman, L., & Spector, P. 1992, International Statistical Review / Revue Internationale de Statistique, 60, 291. http://www.jstor.org/stable/1403680

    1992

  4. [12]

    2019, GPUVMEM: Maximum Entropy Method (MEM) GPU algorithm for radio astronomical image synthesis , Astrophysics Source Code Library, record ascl:1906.014

    C \'a rcamo , M., Mu \ n oz , N., Rannou , F., et al. 2019, GPUVMEM: Maximum Entropy Method (MEM) GPU algorithm for radio astronomical image synthesis , Astrophysics Source Code Library, record ascl:1906.014

    2019

  5. [13]

    E., Casassus , S., Moral , V., & Rannou , F

    C \'a rcamo , M., Rom \'a n , P. E., Casassus , S., Moral , V., & Rannou , F. R. 2018, Astronomy and Computing, 22, 16, 10.1016/j.ascom.2017.11.003

    2018 doi

  6. [14]

    2022, , 134, 114501, 10.1088/1538-3873/ac9642

    CASA Team , Bean , B., Bhatnagar , S., et al. 2022, , 134, 114501, 10.1088/1538-3873/ac9642

    2022 doi

  7. [15]

    2022, , 933, L4, 10.3847/2041-8213/ac75e8

    Casassus , S., C \'a rcamo , M., Hales , A., Weber , P., & Dent , B. 2022, , 933, L4, 10.3847/2041-8213/ac75e8

    2022 doi

  8. [16]

    2019, , 883, L41, 10.3847/2041-8213/ab4425

    Casassus , S., & P \'e rez , S. 2019, , 883, L41, 10.3847/2041-8213/ab4425

    2019 doi

  9. [17]

    2021, , 507, 3789, 10.1093/mnras/stab2359

    Casassus , S., Christiaens , V., C \'a rcamo , M., et al. 2021, , 507, 3789, 10.1093/mnras/stab2359

    2021 doi

  10. [18]

    A., Johnson , M

    Chael , A. A., Johnson , M. D., Bouman , K. L., et al. 2018, , 857, 23, 10.3847/1538-4357/aab6a8

    2018 doi

  11. [19]

    A., Johnson , M

    Chael , A. A., Johnson , M. D., Narayan , R., et al. 2016, , 829, 11, 10.3847/0004-637X/829/1/11

    2016 doi

  12. [20]

    J., & Evans , K

    Cornwell , T. J., & Evans , K. F. 1985, , 143, 77

    1985

  13. [21]

    2021, MPoL-dev/MPoL: v0.1.1 Release, v0.1.1, Zenodo, 10.5281/zenodo.4939048

    Czekala, I., Zawadzki, B., Loomis, R., et al. 2021, MPoL-dev/MPoL: v0.1.1 Release, v0.1.1, Zenodo, 10.5281/zenodo.4939048

    2021 doi

  14. [22]

    2023, MPoL-dev/MPoL: v0.2.0 Release, v0.2.0, Zenodo, 10.5281/zenodo.3594081

    Czekala , I., Jennings , J., Zawadzki , B., et al. 2023, MPoL-dev/MPoL: v0.2.0 Release, v0.2.0, Zenodo, 10.5281/zenodo.3594081

    2023 doi

  15. [23]

    2025, arXiv e-prints, arXiv:2502.00100, 10.48550/arXiv.2502.00100

    ---. 2025, arXiv e-prints, arXiv:2502.00100, 10.48550/arXiv.2502.00100

    2025 doi arXiv

  16. [24]

    J., Bae , J., et al

    Disk Dynamics Collaboration , Armitage , P. J., Bae , J., et al. 2020, arXiv e-prints, arXiv:2009.04345, 10.48550/arXiv.2009.04345

    2020 doi arXiv

  17. [25]

    2019, , 875, L4, 10.3847/2041-8213/ab0e85

    Event Horizon Telescope Collaboration , Akiyama , K., Alberdi , A., et al. 2019, , 875, L4, 10.3847/2041-8213/ab0e85

    2019 doi

  18. [26]

    2025, , TBD

    Galloway-Sprietsma , M., Bae , J., & Izquierdo , A. 2025, , TBD

    2025

  19. [27]

    F., & Daniell , G

    Gull , S. F., & Daniell , G. J. 1978, , 272, 686, 10.1038/272686a0

    1978 doi

  20. [28]

    H \"o gbom , J. A. 1974, , 15, 417

    1974

  21. [29]

    H \"o gbom , J. A. 1979, in Astrophysics and Space Science Library, Vol. 76, IAU Colloq. 49: Image Formation from Coherence Functions in Astronomy, ed. C. van Schooneveld , 237, 10.1007/978-94-009-9449-2_26

    1979 doi

  22. [30]

    Holdaway , M. A. 1990, PhD thesis, Brandeis Univ., Waltham, MA

    1990

  23. [31]

    2014, , 66, 95, 10.1093/pasj/psu070

    Honma , M., Akiyama , K., Uemura , M., & Ikeda , S. 2014, , 66, 95, 10.1093/pasj/psu070

    2014 doi

  24. [32]

    1995, in Proceedings of the 14th International Joint Conference on Artificial Intelligence - Volume 2, IJCAI'95 (San Francisco, CA, USA: Morgan Kaufmann Publishers Inc.), 1137–1143

    Kohavi, R. 1995, in Proceedings of the 14th International Joint Conference on Artificial Intelligence - Volume 2, IJCAI'95 (San Francisco, CA, USA: Morgan Kaufmann Publishers Inc.), 1137–1143

    1995

  25. [33]

    2018, , 858, 56, 10.3847/1538-4357/aab6b5

    Kuramochi , K., Akiyama , K., Ikeda , S., et al. 2018, , 858, 56, 10.3847/1538-4357/aab6b5

    2018 doi

  26. [34]

    Benisty , M

    Loomis , R., , & Facchini , S. Benisty , M. 2025, , TBD

    2025

  27. [35]

    P., Waters , B., Schiebel , D., Young , W., & Golap , K

    McMullin , J. P., Waters , B., Schiebel , D., Young , W., & Golap , K. 2007, in Astronomical Society of the Pacific Conference Series, Vol. 376, Astronomical Data Analysis Software and Systems XVI, ed. R. A. Shaw , F. Hill , & D. J. Bell , 127

    2007

  28. [36]

    M., Simon, R., & Pfeiffer, R

    Molinaro, A. M., Simon, R., & Pfeiffer, R. M. 2005, Bioinformatics, 21, 3301, 10.1093/bioinformatics/bti499

    2005 doi

  29. [37]

    1986, , 24, 127, 10.1146/annurev.aa.24.090186.001015

    Narayan , R., & Nityananda , R. 1986, , 24, 127, 10.1146/annurev.aa.24.090186.001015

    1986

  30. [38]

    2019, in Advances in Neural Information Processing Systems 32, ed

    Paszke, A., Gross, S., Massa, F., et al. 2019, in Advances in Neural Information Processing Systems 32, ed. H. Wallach, H. Larochelle, A. Beygelzimer, F. d Alch\' e -Buc, E. Fox, & R. Garnett (Curran Associates, Inc.), 8024--8035. http://papers.neurips.cc/paper/9015-pytorch-an...

    2019

  31. [39]

    2019, , 158, 15, 10.3847/1538-3881/ab1f88

    P \'e rez , S., Casassus , S., Baruteau , C., et al. 2019, , 158, 15, 10.3847/1538-3881/ab1f88

    2019 doi

  32. [40]

    2020, , 889, L24, 10.3847/2041-8213/ab6b2b

    P \'e rez , S., Casassus , S., Hales , A., et al. 2020, , 889, L24, 10.3847/2041-8213/ab6b2b

    2020 doi

  33. [41]

    2025, , TBD

    Pinte , C., Ilee , J., & Huang , J. 2025, , TBD

    2025

  34. [42]

    2018, , 609, A47, 10.1051/0004-6361/201731377

    Pinte , C., M \'e nard , F., Duch \^e ne , G., et al. 2018, , 609, A47, 10.1051/0004-6361/201731377

    2018 doi

  35. [43]

    Rau , U., & Cornwell , T. J. 2011, , 532, A71, 10.1051/0004-6361/201117104

    2011 doi

  36. [44]

    Schwarz , U. J. 1978, , 65, 345

    1978

  37. [45]

    2025, , TBD

    Teague , R., Benisty , M., Facchini , S., Fukagawa , M., & Pinte , C. 2025, , TBD

    2025

  38. [46]

    2021, The Journal of Open Source Software, 6, 3827, 10.21105/joss.03827

    Teague , R., Law , C., Huang , J., & Meng , F. 2021, The Journal of Open Source Software, 6, 3827, 10.21105/joss.03827

    2021 doi

  39. [47]

    1996, Journal of the Royal Statistical Society

    Tibshirani, R. 1996, Journal of the Royal Statistical Society. Series B (Methodological), 58, 267. http://www.jstor.org/stable/2346178

    1996

  40. [48]

    2021, , 923, 121, 10.3847/1538-4357/ac2bfd

    Yamaguchi , M., Tsukagoshi , T., Muto , T., et al. 2021, , 923, 121, 10.3847/1538-4357/ac2bfd

    2021 doi

  41. [49]

    2020, , 895, 84, 10.3847/1538-4357/ab899f

    Yamaguchi , M., Akiyama , K., Tsukagoshi , T., et al. 2020, , 895, 84, 10.3847/1538-4357/ab899f

    2020 doi

  42. [50]

    2024, , 76, 437, 10.1093/pasj/psae022

    Yamaguchi , M., Muto , T., Tsukagoshi , T., et al. 2024, , 76, 437, 10.1093/pasj/psae022

    2024 doi

  43. [51]

    A., et al

    Zawadzki , B., Czekala , I., Loomis , R. A., et al. 2023, , 135, 064503, 10.1088/1538-3873/acdf84

    2023 doi

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.