Spatially Resolved Galaxy-Dust Modeling with Coupled Data-Driven Priors

Jared Siegel; Peter Melchior

REVIEW 2 major objections 6 minor 36 references

Spatially Resolved Galaxy-Dust Modeling with Coupled Data-Driven Priors

T0 review · 2 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A pair of score-based neural priors lets multi-band galaxy images be decomposed into intrinsic starlight and a resolved dust map, recovering host and dust properties for dust covering fractions below about 90 percent.

desk verdict A genuinely new conditional dust prior and a promising joint modeling scheme, but the headline simulation result is contaminated by training/test overlap and needs a held-out retest. read the letter →

arxiv 2411.08111 v1 pith:NRTXA4KN submitted 2024-11-12 astro-ph.GA astro-ph.IM

Jared Siegel , Peter Melchior This is my paper · ORCID

classification astro-ph.GAastro-ph.IM

keywords dustattenuationspatiallyresolvedgalaxymorphologydata-drivenpriorsscorematchingmulti-bandimagingSEDrecoveryneuralnetworkradiativetransfersimulations

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 authors are trying to establish that the spatial distribution of dust inside a galaxy and the galaxy's unreddened starlight can both be recovered from a stack of ordinary multi-band optical and near-infrared images, with no integral-field spectroscopy or infrared maps. If true, this would scale resolved dust mapping to the billions of galaxies that the next generation of wide-field surveys will image, where current methods are limited to a few hundred well-observed nearby systems or assume an unrealistic uniform foreground dust screen. The model uses non-parametric stellar and dust morphologies, and two neural-network priors make the highly degenerate inverse problem tractable: one scores the plausibility of the stellar morphology, the other scores the dust morphology conditioned on the current stellar morphology. On simulated galaxies the method recovers dust amplitude and geometry for dust covering fractions below roughly 90 percent, and on three local galaxies it reproduces the prominent dust features and dereddened spectral energy distributions.

What carries the argument

The load-bearing object is the score-matching neural-network prior, a network that outputs the gradient of the log-probability of a morphology, $\nabla \log p_{\mathrm{data}}(x)$, for the host image and, in a conditional U-Net variant, for the dust image given the host image $p(D\mid S)$. The conditional dust prior is what makes the non-parametric decomposition work: it is flat where the likelihood is flat, strongly penalizes uncorrelated pixel-scale color noise, and encodes the observed tendency for dust to track stellar light, so the optimizer can move dust where it is unconstrained by data without inventing arbitrary geometries. These score gradients are added to the likelihood gradient during gradient-descent optimization, and the fitted quantities are the host spectrum $F$, host morphology $S$, dust morphology $D$, and attenuation slope $\delta$. A supporting mechanism is the initialization scheme, which starts near zero dust and, when no red filter is available, builds a crude dust map from the bluest pixels so the regularization can refine it into a coherent dust morphology.

What would settle it

Run the method on images of galaxies that have independent spatially resolved attenuation measurements, for example Balmer-decrement maps from integral-field spectroscopy or far-infrared dust maps, and compare the inferred $A_V$ images pixel by pixel; if the residuals correlate with inclination, morphological type, or dust covering fraction in the claimed $f<90\%$ regime, the simulation-based prior is not transferring.

Watch

Extended reading notes

Core claim

The central claim is that the joint model $Y = (F^{T}S) \odot 10^{-0.4A}$, where $F$ is a one-dimensional host spectrum, $S$ is a non-parametric monochrome stellar morphology, and $A = K^{T}[-2.5\log_{10}D]$ is the dust attenuation cube with a spatially uniform slope parameter $\delta$, becomes identifiable in multi-band images once two score-based priors supply the missing regularization. The stellar prior $p(S)$ suppresses pixel noise and keeps the host morphology on the manifold of real galaxy shapes, while the conditional dust prior $p(D\mid S)$ encodes the correlation between starlight and dust location learned from radiative-transfer simulations, preventing the likelihood from interpreting every red pixel or noise fluctuation as dust. Maximum-a-posteriori optimization with these priors recovers accurate attenuation maps and unreddened spectra across a wide range of attenuation amplitudes and geometries, with the known exception of the thin-screen limit in which every line of sight is covered and the attenuation level is degenerate with an intrinsically redder spectrum. The same behavior holds when the fit is initialized only from optical bands, where the starting dust map is noisy and the priors are primarily responsible for steering it to a plausible morphology.

Load-bearing premise

The load-bearing premise is that the dust morphologies appearing in the radiative-transfer simulations used to train the conditional prior are representative of real galaxy-dust geometry, and that the same simulation set can stand in for real galaxies as a test bed.

Editorial extensions

If this is right

Dust attenuation maps and dereddened host SEDs become obtainable for ordinary survey imaging, without requiring IFU observations, far-infrared data, or parametric galaxy models.
Because nonuniform attenuation currently biases stellar-population mass and redshift estimates, recovering the dust map first should reduce those biases for the bulk of survey galaxies.
The documented thin-screen failure defines a concrete development path: adding a stellar-population-synthesis prior on the host spectrum or external tracers such as the Balmer decrement and far-infrared emission should extend the method to fully covered, heavily obscured galaxies.
Survey-scale resolved dust maps would transform studies of dust production, transport, and destruction from a few hundred galaxies to statistically powerful samples.

Reading between the lines

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

The accuracy reported on simulations is probably optimistic because the same radiative-transfer catalog both trains the conditional dust prior and serves as the test set without a train/test split; performance on real galaxies outside that distribution is the open question.
A natural extension the authors do not develop is to distill the fitted host models into a fast feed-forward network that maps images directly to dust maps, removing the need for per-object optimization at survey scale.
The conditional prior could be recalibrated on real galaxies by cross-checking inferred $A_V$ maps against Balmer-decrement maps from integral-field surveys, turning the simulation-only prior into a genuinely data-driven one and probing whether dust geometries in nature match those in simulations.

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.

Desk Editor's Note

read the letter

The new thing here is real: a score-based dust prior p(D|S) conditioned on the host morphology, trained on radiative-transfer simulations, and a joint MAP fit of host spectrum, host morphology, dust morphology, and attenuation slope with two neural priors. The blue-start initialization is also a practical contribution: it lets the method work from optical-only data, which is what LSST/Euclid/Roman will actually provide. The paper is honest about the thin-screen degeneracy and the age-gradient confusion, and the SDSS demonstrations, while qualitative, show the method does something sensible on real data.

The soft spot is the one the stress-test note identifies, and it is real. Section 2.2 trains the dust prior on the 65 NIHAO-SKIRT galaxies of Faucher et al.; Section 3.1 then validates on exactly those 65 galaxies, with three orientations each, and no train/test split is stated. The central claim—accurate dust recovery for f < 90%—therefore largely measures how well the prior has memorized the test distribution. In the blue-start and low-S/N regimes the paper explicitly says it relies on the priors, which is precisely where an overfit prior would do the most damage. This is not a fatal flaw in the method, but it is a load-bearing gap in the evidence. A held-out-galaxy retest (train on 50, test on 15, or leave-one-out) would settle it. Also worth asking for: baseline comparisons against a single-screen fit or a Sersic-dust model, and some form of uncertainty quantification on the MAP dust maps. The lack of those is not disqualifying, but it makes the quantitative claims harder to judge.

On balance, the paper deserves a serious referee. The method is timely and the conditional dust prior is a genuine step beyond scarlet and Sampson et al. I would send it to review with a clear request for re-validation on a disjoint sample. The core idea is solid; the validation protocol is not yet.

Referee Report

2 major / 6 minor

Summary. The paper presents a joint forward model for spatially resolved galaxy and dust decomposition from multi-band optical/NIR images. The host galaxy is described by a non-parametric morphology and a single positive spectrum, while dust attenuation is described by a spatial dust map and a fitted attenuation-curve slope. Two score-based neural priors regularize the inversion: a host morphology prior trained on HSC images and a conditional dust prior p(D|S) trained on NIHAO-SKIRT radiative-transfer simulations. MAP fits are obtained with gradient descent. The method is validated on simulated griz observations of 65 NIHAO-SKIRT galaxies under red-start and blue-start initialization, and applied to three SDSS galaxies. The headline claim is accurate recovery of dust amplitude and morphology for covering fractions f < 90%, with a known thin-screen degeneracy at higher covering fractions.

Significance. If the validation is sound, the method is a meaningful step toward spatially resolved dust attenuation maps and dereddened galaxy SEDs from ordinary multi-band images, which would be valuable for the large imaging surveys now coming online. The paper has several concrete strengths: the conditional dust prior is a principled way to regularize a severely ill-posed inverse problem; the host prior is independently trained on a large sample of real HSC galaxy images; the thin-screen degeneracy is acknowledged explicitly rather than hidden; and the authors point to public catalogs and code for the host prior and the simulation library. The main qualification is that the quantitative claim in Section 3.1 is currently based on a simulated test set whose independence from the dust-prior training set is not established. If that independence is demonstrated, the paper would be a solid methods contribution.

major comments (2)

[§2.2, §3.1] The central simulated validation is not independent of the dust prior. Section 2.2 trains the dust score model on the 65 NIHAO-SKIRT galaxies from Faucher et al. (2023), and Section 3.1 tests on "65 simulated galaxies from Faucher et al. (2023)" with three orientation angles each; no train/test split is stated anywhere in the paper. Because the public catalog contains ten viewing angles per galaxy, the reader cannot rule out that the tested orientations are drawn from the same set of images used in training. In that case, the Figure 4 residuals largely measure the dust prior's ability to recall the training distribution rather than its ability to generalize to unseen galaxy-dust geometries. This matters most in the blue-start and low-S/N regimes, where Section 2.3 states that the model "relies on our data-driven priors" and where the likelihood is too weak to correct an unrepresentative prior. Please provide a held-out validation: train the dust prior on a subset of the catalog and test on the remaining galaxies (or use leave-one-out cross-validation), and report the same recovery metrics for the held-out set. If a split was already used, state it explicitly and give the details.
[§2.1, Eqs. (3)–(5), §3.2] The forward model as written does not include PSF convolution or a pixel response function, and the simulated validation in Section 3.1 appears to use PSF-free images. Real SDSS images are seeing-limited; even for the large galaxies considered here, the PSF is non-negligible after downsampling to 64×64 pixels, and it will mix attenuated and unattenuated light across pixel boundaries. If scarlet2 includes a PSF that is simply not shown in Eqs. (3)–(5), this should be stated explicitly and the PSF should be included in the simulated validation. If it is not included, the "realistic simulations" claim in the abstract is overstated, and the SDSS application needs to be tested with PSF-convolved images before the recovered dust maps can be interpreted at the claimed spatial resolution.

minor comments (6)

[§3.2] The phrase "bright (mg > 16 mag)" appears to have the inequality sign reversed; all three galaxies are bright objects with apparent magnitudes below 16.
[§3.1] The text alternates between "isochromatic" and "monochromatic" galaxies in the same paragraph; please use one term consistently.
[§2.3] The blue-start initialization would benefit from a precise definition of the color used to rank pixels (e.g., g−i) and a description of how the bluest N% of pixels are selected within the galaxy footprint.
[§2.2, §5] No trained weights or code are provided for the dust prior network; a data/code availability statement would improve reproducibility, especially because the host prior and the NIHAO-SKIRT catalog are already public.
[Eq. (1) and Eq. (3)] The symbol F is used both for the spectral flux density in Eq. (1) and for the vectorized filter-integrated spectrum in Eq. (3); this overloaded notation is confusing and should be disambiguated.
[§3.2] The residual panels in Figure 5 are only described qualitatively; reporting the RMS residual with and without the dust component would make the improvement concrete and easier to compare across S/N levels.

Circularity Check

1 steps flagged · score 5.0 of 10

Dust prior is trained and evaluated on the same 65 NIHAO-SKIRT galaxies, so the headline f<90% recovery is partly an in-sample check; no algebraic circularity in the model derivation.

fitted input called prediction [Section 2.2 (dust prior training) and Section 3.1 (simulated-galaxy tests, Figure 4)]
"We train the network with simulated galaxies from Faucher et al. (2023). ... We begin by considering 65 simulated galaxies from Faucher et al. (2023). ... we consider all 65 galaxies from Faucher et al. (2023) and three orientation angles for each. ... We therefore rely on our data-driven priors to transform our noisy initializations into plausible galaxies."

The conditional dust prior p(D|S) is learned from the Faucher et al. (2023) NIHAO-SKIRT catalog, and the headline quantitative test (Figure 4, Section 3.1) evaluates recovery on exactly that catalog (65 galaxies, three orientations), with no stated train/test split. Because a score-model prior trained on these galaxy-dust pairs assigns high density to the true dust maps of the same galaxies, the MAP solutions in Figure 4 are attracted toward ground truth even where the image likelihood is weak; in the blue-start and low-S/N regimes the paper explicitly says the fit 'relies on our data-driven priors.' The reported f<90% accuracy is therefore an in-sample measure that partly reflects prior memorization rather than an independent prediction.

full rationale

The paper's forward model (Eqs. 1-5) and MAP optimization are not circular: the dust attenuation law, the separable SED/morphology parameterization, and the score-matching training objective are standard and do not presuppose the results. The host morphology prior is taken from Sampson et al. (2024), a public network trained on 600,000 HSC images; although that paper shares an author, the prior is independently trained and code-reproduced, so the self-citation is not load-bearing circularity. The one substantive circularity is in the evaluation design: the conditional dust prior is trained on the Faucher et al. (2023) NIHAO-SKIRT galaxies, and Section 3.1 then reports recovery of dust amplitude and morphology on those same 65 galaxies (three orientations each) without any stated train/test split. Since the prior encodes the joint distribution of the training dust/host pairs, the MAP solutions are pulled toward the very dust maps used to train the prior, particularly in the blue-start and low-S/N regimes where the paper states the fit relies on the data-driven priors. Thus the headline f<90% accuracy is an in-sample statistic, not an independent prediction. This warrants a moderate circularity score (5): the derivation itself is not equivalent to its inputs, but the central quantitative claim lacks an independent test set. The SDSS galaxy applications and the isochromatic control are external or semi-external, which prevents a higher score.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper invents no new physical entities; the two neural priors are procedural components, not physical postulates. The central claim rests on the attenuation-law form, the spatial-spectral separability ansatz, the simulation-to-observation transfer of the dust prior, and the hand-set noise scales and initialization choices listed above.

free parameters (4)

attenuation curve slope delta = per galaxy, uniform prior in [-1, 0.5]
Free parameter in the modified Calzetti law (Eq. 2) that sets the wavelength dependence of dust attenuation; fitted jointly with the morphologies.
score-model noise scale for host prior = 0.02
Hand-set evaluation noise level for the host morphology score prior (Section 2.2); it controls how strongly the prior regularizes the inferred galaxy shape.
score-model noise scale for dust prior = 0.1
Hand-set evaluation noise for the conditional dust prior, chosen larger because the training set is small (Section 2.2).
blue-start percentile N = 50%
Fraction of the bluest pixels assumed to be unreddened when initializing the host morphology without NIR coverage (Section 2.3).

assumptions (5)

domain assumption Dust attenuation follows the modified Calzetti law with a single spatially uniform slope parameter delta across the galaxy (Eqs. 1 and 2).
The attenuation curve is assumed known up to one slope parameter; real dust geometry and scattering may violate this per-pixel screen relation.
domain assumption Spatial and spectral variations separate: Y = F^T S and A = K^T (-2.5 log10 D) (Eqs. 3 to 5).
Assumes no within-galaxy color gradients beyond those caused by the dust screen; age and metallicity gradients are therefore potentially misattributed to dust, as acknowledged in Section 3.1.
domain assumption Dust morphology prior factorizes as p(S,D) = p(S) p(D|S) (Eq. 6), and the conditional relation is learned from NIHAO-SKIRT radiative-transfer simulations.
The coupling between stellar light and dust is assumed to transfer from simulations to real galaxies; this is the load-bearing generalization premise.
standard math Score-matching networks trained with Gaussian noise at sigma = 0.02 or 0.1 provide an adequate approximation to the true log-density gradient at zero noise.
Follows Song and Ermon (2020); approximation accuracy depends on training set size and the hand-selected noise scale.
domain assumption The u-band is omitted because the attenuation curve is steep in the near-UV, and the remaining griz bands sufficiently constrain the attenuation law.
Choice of photometric bands limits sensitivity to the attenuation curve and dust amplitude, as stated in Section 3.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Spatially Resolved Galaxy-Dust Modeling with Coupled Data-Driven Priors." pith.science (2026). https://pith.science/paper/NRTXA4KN

@misc{pith2026241108111,
  author       = {Pith},
  title        = {Pith review of: Spatially Resolved Galaxy-Dust Modeling with Coupled Data-Driven Priors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NRTXA4KN}},
  note         = {Machine review of arXiv:2411.08111}
}

read the original abstract

A notorious problem in astronomy is the recovery of the true shape and spectral energy distribution (SED) of a galaxy despite attenuation by interstellar dust embedded in the same galaxy. This problem has been solved for a few hundred nearby galaxies with exquisite data coverage, but these techniques are not scalable to the billions of galaxies soon to be observed by large wide-field surveys like LSST, Euclid, and Roman. We present a method for jointly modeling the spatially resolved stellar and dust properties of galaxies from multi-band images. To capture the diverse geometries of galaxies, we consider non-parametric morphologies, stabilized by two neural networks that act as data-driven priors: the first informs our inference of the galaxy's underlying morphology, the second constrains the galaxy's dust morphology conditioned on our current estimate of the galaxy morphology. We demonstrate with realistic simulations that we can recover galaxy host and dust properties over a wide range of attenuation levels and geometries. We successfully apply our joint galaxy-dust model to three local galaxies observed by SDSS. In addition to improving estimates of unattenuated galaxy SEDs, our inferred dust maps will facilitate the study of dust production, transport, and destruction.

Figures

Figures reproduced from arXiv: 2411.08111 by the authors.

**Figure 1.** Our method for modeling a dust reddened galaxy. Data-driven neural network priors inform our gradient-based inference of the galaxy and dust attenuation morphologies. The dust prior is conditional on the host galaxy morphology. environment. We therefore posit a dependency structure of the prior in the form of p(S, D) = p(S) p(D | S). (6) We approximate Equation 6 with neural networks, which have been shown to accura… view at source ↗

**Figure 2.** An initially noisy dust distribution is transformed into a plausible dust distribution by iteratively applying our score-matching prior (left to right). The color scale covers per-pixel AV ∈ [0, 1.25]. The contours present the 50 and 84-th percentiles of the unreddened galaxy light. train the network with simulated galaxies from Faucher et al. (2023). The simulated galaxies are drawn from the NIHAO project’s redshif… view at source ↗

**Figure 3.** Ground truth (top) and inferred (bottom) dust morphologies for four representative simulated galaxies. The color scale covers per-pixel AV ∈ [0, 1.25]. The contours present the 50 and 84-th percentiles of the unreddened galaxy light. The inset panels show the spectrum of the galaxy before (grey) and after (red) dust attenuation. We successfully recover the amplitude and morphology of dust attenuation for a wide rang… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Accuracy of our inferred attenuation level (left) and the difference between the inferred dust map and the true dust distribution (right) for 65 simulated galaxies with three orientation angles each. The solid lines represent a running median, for either the isochromat…

**Figure 5.** Figure 5: Our joint galaxy–dust modeling applied to three galaxies observed in SDSS DR18. To represent varied observing conditions, we test three per-band signal-to-noise levels and present the recovered dust map for each galaxy. We also show the multi-band residuals from fits w…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 5 canonical work pages

[1]

1966, ApJS, 14, 1, doi: 10.1086/190147

Arp, H. 1966, ApJS, 14, 1, doi: 10.1086/190147

doi:10.1086/190147 1966
[2]

A., Freedman, W

Bernstein, R. A., Freedman, W. L., & Madore, B. F. 2002, ApJ, 571, 56, doi: 10.1086/339422

doi:10.1086/339422 2002
[3]

2018, PASJ, 70, S5, doi: 10.1093/pasj/psx080

Bosch, J., Armstrong, R., Bickerton, S., et al. 2018, PASJ, 70, S5, doi: 10.1093/pasj/psx080

doi:10.1093/pasj/psx080 2018
[4]

2018, JAX: composable transformations of Python+NumPy programs, 0.3.13

Bradbury, J., Frostig, R., Hawkins, P., et al. 2018, JAX: composable transformations of Python+NumPy programs, 0.3.13. http://github.com/google/jax

2018
[5]

A., Law, D

Bundy, K., Bershady, M. A., Law, D. R., et al. 2015, ApJ, 798, 7, doi: 10.1088/0004-637X/798/1/7

doi:10.1088/0004-637x/798/1/7 2015
[6]

C., et al

Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ, 533, 682, doi: 10.1086/308692

doi:10.1086/308692 2000
[7]

2020, Astronomy and Computing, 31, 100381, doi: 10.1016/j.ascom.2020.100381

Camps, P., & Baes, M. 2020, Astronomy and Computing, 31, 100381, doi: 10.1016/j.ascom.2020.100381

arXiv 2020
[8]

2003, MNRAS, 342, 345, doi: 10.1046/j.1365-8711.2003.06541.x

Cappellari, M., & Copin, Y. 2003, MNRAS, 342, 345, doi: 10.1046/j.1365-8711.2003.06541.x

arXiv 2003

Show all 36 references

[9]

P., Bianchi, S., et al

Casasola, V., Cassar` a, L. P., Bianchi, S., et al. 2017, A&A, 605, A18, doi: 10.1051/0004-6361/201731020

2017 doi
[10]

2013, ARA&A, 51, 393, doi: 10.1146/annurev-astro-082812-141017

Conroy, C. 2013, ARA&A, 51, 393, doi: 10.1146/annurev-astro-082812-141017

2013 doi
[11]

I., Baes, M., Bianchi, S., et al

Davies, J. I., Baes, M., Bianchi, S., et al. 2017, PASP, 129, 044102, doi: 10.1088/1538-3873/129/974/044102

2017 doi
[12]

Draine, B. T. 2003, ARA&A, 41, 241, doi: 10.1146/annurev.astro.41.011802.094840

2003 arXiv
[13]

R., & Macci` o, A

Faucher, N., Blanton, M. R., & Macci` o, A. V. 2023, ApJ, 957, 7, doi: 10.3847/1538-4357/acf9f0

2023 doi
[14]

J., Arag´ on-Salamanca, A., Merrifield, M

Greener, M. J., Arag´ on-Salamanca, A., Merrifield, M. R., et al. 2020, MNRAS, 495, 2305, doi: 10.1093/mnras/staa1300

2020 doi
[15]

R., Vieira, J

Greve, T. R., Vieira, J. D., Weiß, A., et al. 2012, ApJ, 756, 101, doi: 10.1088/0004-637X/756/1/101 G¨ uver, T., &¨Ozel, F. 2009, MNRAS, 400, 2050, doi: 10.1111/j.1365-2966.2009.15598.x

2012
[16]

2024, arXiv e-prints, arXiv:2409.19054, doi: 10.48550/arXiv.2409.19054

Hahn, C., & Melchior, P. 2024, arXiv e-prints, arXiv:2409.19054, doi: 10.48550/arXiv.2409.19054

2024 doi arXiv
[17]

2002, MNRAS, 337, 921, doi: 10.1046/j.1365-8711.2002.05968.x

Hirashita, H., & Ferrara, A. 2002, MNRAS, 337, 921, doi: 10.1046/j.1365-8711.2002.05968.x

2002
[18]

2021, Differentiable Programming workshop at Neural Information Processing Systems 2021

Kidger, P., & Garcia, C. 2021, Differentiable Programming workshop at Neural Information Processing Systems 2021

2021
[19]

P., & Ba, J

Kingma, D. P., & Ba, J. 2015, in 3rd International Conference on Learning Representations, ICLR 2015, San

2015
[20]

http://arxiv.org/abs/1412.6980

Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings. http://arxiv.org/abs/1412.6980

2015 arXiv
[21]

2013, The Astrophysical Journal Letters, 775, L16, doi: 10.1088/2041-8205/775/1/L16

Kriek, M., & Conroy, C. 2013, The Astrophysical Journal Letters, 775, L16, doi: 10.1088/2041-8205/775/1/L16

2013 doi
[22]

2011, arXiv e-prints, arXiv:1110.3193, doi: 10.48550/arXiv.1110.3193 LSST Science Collaboration, Abell, P

Laureijs, R., Amiaux, J., Arduini, S., et al. 2011, arXiv e-prints, arXiv:1110.3193, doi: 10.48550/arXiv.1110.3193 LSST Science Collaboration, Abell, P. A., Allison, J., et al. 2009, arXiv e-prints, arXiv:0912.0201, doi: 10.48550/arXiv.0912.0201

2011 doi arXiv
[23]

2018, Astronomy and Computing, 24, 129, doi: 10.1016/j.ascom.2018.07.001

Melchior, P., Moolekamp, F., Jerdee, M., et al. 2018, Astronomy and Computing, 24, 129, doi: 10.1016/j.ascom.2018.07.001

2018 doi
[24]

2009, Astronomy and astrophysics, 499, 69, doi: 10.1051/0004-6361/200811526

Noll, S., Pierini, D., Cimatti, A., et al. 2009, Astronomy and astrophysics, 499, 69, doi: 10.1051/0004-6361/200811526

2009 doi
[25]

2015, arXiv e-prints, arXiv:1505.04597, doi: 10.48550/arXiv.1505.04597

Ronneberger, O., Fischer, P., & Brox, T. 2015, arXiv e-prints, arXiv:1505.04597, doi: 10.48550/arXiv.1505.04597

2015 doi arXiv
[26]

2020, ARA&A, 58, 529, doi: 10.1146/annurev-astro-032620-021933

Salim, S., & Narayanan, D. 2020, ARA&A, 58, 529, doi: 10.1146/annurev-astro-032620-021933

2020 doi
[27]

L., Melchior, P., Ward, C., & Birmingham, S

Sampson, M. L., Melchior, P., Ward, C., & Birmingham, S. 2024, arXiv e-prints, arXiv:2401.07313, doi: 10.48550/arXiv.2401.07313

2024 doi arXiv
[28]

2010, MNRAS, 407, 1581, doi: 10.1111/j.1365-2966.2010.17047.x Galaxy Dust Maps 11

Shen, S., Wadsley, J., & Stinson, G. 2010, MNRAS, 407, 1581, doi: 10.1111/j.1365-2966.2010.17047.x Galaxy Dust Maps 11

2010
[29]

2019, in Advances in Neural Information Processing Systems, ed

Song, Y., & Ermon, S. 2019, in Advances in Neural Information Processing Systems, ed. H. Wallach, H. Larochelle, A. Beygelzimer, F. d 'Alch´ e-Buc, E. Fox, & R. Garnett, Vol. 32 (Curran Associates, Inc.). https://proceedings.neurips.cc/paper files/paper/2019/ file/3001ef257407...

2019
[30]

2020, in Advances in Neural Information Processing Systems, ed

Song, Y., & Ermon, S. 2020, in Advances in Neural Information Processing Systems, ed. H. Larochelle, M. Ranzato, R. Hadsell, M. Balcan, & H. Lin, Vol. 33 (Curran Associates, Inc.), 12438–12448. https://proceedings.neurips.cc/paper files/paper/2020/ file/92c3b916311a5517d929057...

2020
[31]

2015, arXiv e-prints, arXiv:1503.03757, doi: 10.48550/arXiv.1503.03757

Spergel, D., Gehrels, N., Baltay, C., et al. 2015, arXiv e-prints, arXiv:1503.03757, doi: 10.48550/arXiv.1503.03757

2015 doi arXiv
[32]

S., Brook, C., Macci` o, A

Stinson, G. S., Brook, C., Macci` o, A. V., et al. 2013, MNRAS, 428, 129, doi: 10.1093/mnras/sts028

2013 doi
[33]

A., Kriek, M., Price, S

Suess, K. A., Kriek, M., Price, S. H., & Barro, G. 2019, ApJ, 877, 103, doi: 10.3847/1538-4357/ab1bda van den Oord, A., Kalchbrenner, N., Vinyals, O., et al. 2016, arXiv [cs.CV]. https://arxiv.org/abs/1606.05328

2019 arXiv
[34]

A., Stinson, G

Wang, L., Dutton, A. A., Stinson, G. S., et al. 2015, MNRAS, 454, 83, doi: 10.1093/mnras/stv1937

2015 doi
[35]

M., Genzel, R., et al

Wuyts, S., F¨ orster Schreiber, N. M., Genzel, R., et al. 2012, ApJ, 753, 114, doi: 10.1088/0004-637X/753/2/114

2012 doi
[36]

G., Adelman, J., Anderson, John E., J., et al

York, D. G., Adelman, J., Anderson, John E., J., et al. 2000, AJ, 120, 1579, doi: 10.1086/301513

2000 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

1966, ApJS, 14, 1, doi: 10.1086/190147

Arp, H. 1966, ApJS, 14, 1, doi: 10.1086/190147

doi:10.1086/190147 1966

[2] [2]

A., Freedman, W

Bernstein, R. A., Freedman, W. L., & Madore, B. F. 2002, ApJ, 571, 56, doi: 10.1086/339422

doi:10.1086/339422 2002

[3] [3]

2018, PASJ, 70, S5, doi: 10.1093/pasj/psx080

Bosch, J., Armstrong, R., Bickerton, S., et al. 2018, PASJ, 70, S5, doi: 10.1093/pasj/psx080

doi:10.1093/pasj/psx080 2018

[4] [4]

2018, JAX: composable transformations of Python+NumPy programs, 0.3.13

Bradbury, J., Frostig, R., Hawkins, P., et al. 2018, JAX: composable transformations of Python+NumPy programs, 0.3.13. http://github.com/google/jax

2018

[5] [5]

A., Law, D

Bundy, K., Bershady, M. A., Law, D. R., et al. 2015, ApJ, 798, 7, doi: 10.1088/0004-637X/798/1/7

doi:10.1088/0004-637x/798/1/7 2015

[6] [6]

C., et al

Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ, 533, 682, doi: 10.1086/308692

doi:10.1086/308692 2000

[7] [7]

2020, Astronomy and Computing, 31, 100381, doi: 10.1016/j.ascom.2020.100381

Camps, P., & Baes, M. 2020, Astronomy and Computing, 31, 100381, doi: 10.1016/j.ascom.2020.100381

arXiv 2020

[8] [8]

2003, MNRAS, 342, 345, doi: 10.1046/j.1365-8711.2003.06541.x

Cappellari, M., & Copin, Y. 2003, MNRAS, 342, 345, doi: 10.1046/j.1365-8711.2003.06541.x

arXiv 2003