pith. sign in

arxiv: 2606.11308 · v1 · pith:2HYGZNQ3new · submitted 2026-06-09 · 🌌 astro-ph.GA · astro-ph.IM

pop-cosmos: Disentangling galaxy properties from observables using data-driven approaches

Benedict Van den Bussche , Sinan Deger , Hiranya V. Peiris , Stephen Thorp , Daniel J. Mortlock , Boris Leistedt , Anik Halder , Madalina N. Tudorache
show 1 more author
Gurjeet Jagwani
This is my paper

Pith reviewed 2026-06-27 12:32 UTC · model grok-4.3

classification 🌌 astro-ph.GA astro-ph.IM
keywords galaxy spectrastellar populationsspectral energy distributionvariational autoencoderionization statedegeneraciesstar formationdust
0
0 comments X

The pith

Five independent dimensions describe the essential information in galaxy rest-frame optical spectra.

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

The paper applies a neural network compression technique to parameters from a generative model of galaxy populations to determine how many truly separate factors shape the rest-frame optical spectral energy distribution. It concludes that five dimensions capture the independent information, corresponding to stellar mass, recent star formation, dust content, and two aspects of gas ionization state. This matters for observers because metallicity and stellar age do not emerge as separate drivers; their effects are distributed across the other dimensions, which resolves mixing that otherwise limits what can be learned from broadband photometry alone. Connecting each dimension to particular spectral features allows the physical state of the gas to be recovered more directly than with standard emission-line ratio methods.

Core claim

A beta variational autoencoder compresses a 16-parameter stellar population synthesis description into a disentangled latent representation whose dimensions are interpreted through mutual information. Five independent dimensions suffice to describe the rest-frame optical SED and map to stellar mass, recent star formation, dust, and two degrees of freedom in the ionization state of the gas. Stellar metallicity and stellar age are not among these primary drivers because their spectral effects are distributed across the others rather than independently encoded.

What carries the argument

A beta variational autoencoder that compresses 16 stellar population synthesis parameters into a disentangled latent space, with each dimension validated by mutual information against physical quantities.

If this is right

  • Stellar metallicity and age effects are absorbed into the five primary dimensions rather than requiring separate ones.
  • Only two degrees of freedom are required to describe gas ionization conditions, fewer than the three or four assumed in standard nebular models.
  • Each dimension can be tied to specific spectral features, which breaks the star-formation, dust, and metallicity degeneracies that affect broadband photometry.
  • Physical conditions of the gas are recovered more cleanly than is possible with conventional line-ratio diagnostics.

Where Pith is reading between the lines

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

  • The five-dimensional structure could be tested by projecting real survey spectra onto the same latent space and measuring reconstruction accuracy.
  • Parameter spaces in galaxy evolution simulations might be reduced to these core dimensions while retaining the main observable variations.
  • Survey strategies could prioritize wavelength coverage around the spectral features associated with each dimension to improve efficiency of property estimation.

Load-bearing premise

The generative galaxy population model used to produce the training data accurately represents the physical processes and degeneracies present in real observations.

What would settle it

Extracting independent components directly from a large set of real galaxy spectra observed by spectroscopic surveys and checking whether five dimensions reproduce the observed variations as well as the model-derived dimensions do.

Figures

Figures reproduced from arXiv: 2606.11308 by Anik Halder, Benedict Van den Bussche, Boris Leistedt, Daniel J. Mortlock, Gurjeet Jagwani, Hiranya V. Peiris, Madalina N. Tudorache, Sinan Deger, Stephen Thorp.

Figure 1
Figure 1. Figure 1: Disentanglement of the latents after training the VAE as measured by the MI between them. The MI is in natural units. The MI between identical latents are fixed to 1. MI values under 0.01 are not displayed. Uncertainty on the MI measurements is less than 0.001 [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Normalized SED fractional reconstruction error after training the VAE. Significant emission lines are highlighted in dashed red lines. The average overall error is 1.82 × 10−3 with the 95 percentile shaded in blue and the 99.9 percentile in orange. We randomly partition the dataset of 90,000 mock galaxies (nor￾malized SEDs and their normalization factors) into training (60,000 galaxies) and validation (30,… view at source ↗
Figure 3
Figure 3. Figure 3: MI between the five latent variables of the VAE trained on rest￾frame SEDs, and galaxy properties. MI values above 0.1 are overplotted on the heatmap. Only key galaxy SPS parameters isolated by the VAE are plotted on the heatmap. The uncertainty is less than 0.01 [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Conditional MI between individual latents and the rest-frame SED at each spectral indices, conditioned on all the other latents. Significant emis￾sion lines are overplotted in dotted gray lines with red labels above. MNRAS 000, 1–12 (2026) [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Variations in the decoded rest-frame SED when varying one latent eight times with all the others fixed. A different latent is traversed in each panel from top to bottom. The left panel displays the full SED while the right panel zooms into the [O ii], [O iii], H β, and H α regions. The first two and last two columns share a 𝑦-axis range. The different latents are ranked by their relative effect on the spec… view at source ↗
Figure 6
Figure 6. Figure 6: MI between different properties (top row: stellar mass, sSFR, and stellar metallicity; bottom row: diffuse dust optical depth 𝜏2, and AGN bolometric luminosity fraction 𝑓AGN) and observed frame photometric colours for different redshift slices (vertical axis 𝑧). MI values are approximated to the closest decimal. The colourmap is discretized in bins of width 0.1. The uncertainty on the MI values are all les… view at source ↗
Figure 7
Figure 7. Figure 7: MI between the different properties (top to bottom: sSFR, gas-phase metallicity, and gas ionization) with different emission lines in rest-frame SEDs with galaxies observed across four different redshift slices. the broad optical and NIR slope reflects the continuum reddening caused by dust attenuation (Calzetti et al. 2000) alongside dominant old stellar populations and metals (see Sections 4.2 and 4.5) … view at source ↗
Figure 8
Figure 8. Figure 8: BPT axes against latent 𝑧1 and 𝑧3. Top: BPT diagram coloured by latent 1 (left) and latent 3 (right). Bottom: Latent 1 against latent 3 coloured by the BPT axes log10([N ii]/H α) (left) and log10([O iii]/H β) (right). −2 −1 0 1 2 Latent Dimension 1 −2 −1 0 1 2 Latent Dimension 3 −2 −1 0 1 2 Latent Dimension 1 −2 −1 0 1 2 Latent Dimension 3 −3.5 −3.0 −2.5 −2.0 log10(Ugas / U ) −1.50 −1.25 −1.00 −0.75 −0.50 … view at source ↗
Figure 9
Figure 9. Figure 9: Latent 1 against latent 3 colour coded by gas-phase metallicity (left) and gas ionization (right). MNRAS 000, 1–12 (2026) [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗
read the original abstract

The physical processes that shape a galaxy's spectrum are strongly degenerate in observations, obscuring which processes act independently. Leveraging the pop-cosmos generative galaxy population model, we investigate how many independent degrees of freedom the rest-frame optical SED contains. We use a $\beta$-variational autoencoder (VAE) to compress a 16-parameter stellar population synthesis (SPS) description into a disentangled latent representation interpreted through mutual information (MI). We find that five independent dimensions suffice, corresponding to stellar mass, recent star formation, dust, and two -- not the three or four assumed by standard nebular models -- degrees of freedom in the ionization state of the gas. Stellar metallicity and stellar age are not among these primary drivers; their spectral effects are distributed across the others rather than independently encoded. By tying each dimension to specific spectral features, this decomposition breaks the star-formation--dust--metallicity degeneracies that limit broadband photometry, and recovers the physical conditions of the gas more cleanly than the line-ratio diagnostics in standard use.

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 / 1 minor

Summary. The paper claims that a β-variational autoencoder trained on synthetic rest-frame optical SEDs from the pop-cosmos 16-parameter SPS generative model compresses the data into five independent latent dimensions. These map via mutual information to stellar mass, recent star formation, dust, and two (rather than three or four) degrees of freedom in gas ionization state; stellar metallicity and age effects are distributed across the others. The decomposition is presented as breaking star-formation–dust–metallicity degeneracies that affect broadband photometry and as recovering gas conditions more cleanly than standard line-ratio diagnostics.

Significance. If the result holds and generalizes, the work would supply a data-driven route to identify the minimal independent physical drivers encoded in galaxy SEDs and a practical compression that could reduce degeneracies in photometric and spectroscopic inference for large surveys.

major comments (2)
  1. [Abstract] Abstract: the claim that the five dimensions correspond to the independent physical drivers 'present in the rest-frame optical SED' and that the method 'breaks the star-formation–dust–metallicity degeneracies that limit broadband photometry' rests entirely on synthetic SEDs generated inside pop-cosmos; no external validation against real survey spectra or comparison of joint distributions is shown, so the mapping to real-galaxy degeneracies remains untested.
  2. [Abstract] Abstract: the assertion that the decomposition 'recovers the physical conditions of the gas more cleanly than the line-ratio diagnostics in standard use' is not accompanied by any quantitative metric, baseline comparison, or test on held-out synthetic or real data that would establish superiority.
minor comments (1)
  1. The abstract refers to mutual-information attribution but supplies no description of how MI is computed, normalized, or thresholded, which is needed to assess the robustness of the dimension-to-feature mapping.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive feedback on our manuscript. We respond to the major comments below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that the five dimensions correspond to the independent physical drivers 'present in the rest-frame optical SED' and that the method 'breaks the star-formation–dust–metallicity degeneracies that limit broadband photometry' rests entirely on synthetic SEDs generated inside pop-cosmos; no external validation against real survey spectra or comparison of joint distributions is shown, so the mapping to real-galaxy degeneracies remains untested.

    Authors: We agree that the entire analysis, including the identification of five latent dimensions and their physical mappings via mutual information, is performed on synthetic rest-frame optical SEDs generated from the pop-cosmos 16-parameter model. The manuscript does not include external validation on real survey spectra or direct comparisons of joint distributions with observed data. The pop-cosmos model is constructed to reproduce key observed galaxy population statistics, so the degeneracies identified are those present under the model's assumptions; however, we do not claim that the mapping has been tested on real galaxies. We will revise the abstract to make the synthetic nature of the data and the scope of the claims explicit. revision: partial

  2. Referee: [Abstract] Abstract: the assertion that the decomposition 'recovers the physical conditions of the gas more cleanly than the line-ratio diagnostics in standard use' is not accompanied by any quantitative metric, baseline comparison, or test on held-out synthetic or real data that would establish superiority.

    Authors: The abstract statement is grounded in the mutual information results showing that two latent dimensions isolate distinct ionization-state degrees of freedom, whereas standard line-ratio diagnostics (e.g., BPT diagrams) entangle multiple parameters. Nevertheless, the manuscript does not present a quantitative metric (such as predictive accuracy or reconstruction error on held-out synthetic SEDs) or explicit baseline comparison against line-ratio methods. We will add such a quantitative comparison on the held-out synthetic test set to support the claim. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is a direct compression of the input model outputs.

full rationale

The paper trains a β-VAE on synthetic SEDs generated from its own 16-parameter pop-cosmos SPS model and interprets the resulting latent dimensions via mutual information. This process directly extracts effective degrees of freedom present in the model's parameter space and degeneracy structure by construction, but the provided text contains no self-definitional loop, no fitted parameter renamed as an independent prediction, and no load-bearing self-citation that justifies a uniqueness theorem. The central result (five dimensions suffice, with specific physical mappings) is an empirical outcome of the VAE training rather than a tautological re-expression of the inputs. The assumption that pop-cosmos faithfully represents real galaxies is stated explicitly as a modeling choice, not smuggled in via prior self-citation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available, so no specific free parameters, axioms, or invented entities can be extracted or verified. The work implicitly relies on the accuracy of the pop-cosmos model and the validity of VAE disentanglement via MI, but these are not detailed.

pith-pipeline@v0.9.1-grok · 5755 in / 1319 out tokens · 31522 ms · 2026-06-27T12:32:07.268556+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

300 extracted references · 260 canonical work pages · 127 internal anchors

  1. [1]

    and Ralchenko,

    Kramida, A. and Ralchenko,. 2024 , month = nov, doi =

    2024

  2. [2]

    , keywords =

    Discovering the building blocks of dark matter halo density profiles with neural networks. , keywords =. doi:10.1103/PhysRevD.105.103533 , archivePrefix =. 2203.08827 , primaryClass =

    work page doi:10.1103/physrevd.105.103533

  3. [3]

    , keywords =

    Representation learning approach to probe for dynamical dark energy in matter power spectra. , keywords =. doi:10.1103/PhysRevD.110.023514 , archivePrefix =. 2310.10717 , primaryClass =

    work page doi:10.1103/physrevd.110.023514

  4. [4]

    , keywords =

    CDM and early dark energy in latent space: A data-driven parametrization of the CMB temperature power spectrum. , keywords =. doi:10.1103/PhysRevD.111.083537 , archivePrefix =. 2502.09810 , primaryClass =

    work page doi:10.1103/physrevd.111.083537

  5. [5]

    Proceedings of the 9th Python in Science Conference , title =

    McKinney, Wes , doi =. Proceedings of the 9th Python in Science Conference , title =. 2010 , publisher =

    2010

  6. [6]

    2012 , Bdsk-Url-1 =

    Eliminating error in the chemical abundance scale for extragalactic H II regions. , keywords =. doi:10.1111/j.1365-2966.2012.21145.x , archivePrefix =. 1203.5021 , primaryClass =

    work page doi:10.1111/j.1365-2966.2012.21145.x 2012

  7. [7]

    Monthly Notices of the Royal Astronomical Society , author=

    Inferring physical properties of galaxies from their emission-line spectra , volume=. Monthly Notices of the Royal Astronomical Society , author=. 2017 , month=feb, pages=. doi:10.1093/mnras/stw2836 , number=

    work page doi:10.1093/mnras/stw2836 2017

  8. [8]

    Color and Stellar Population Gradients in Passively Evolving Galaxies at z~2 from HST/WFC3 Deep Imaging in the Hubble Ultra Deep Field

    Color and Stellar Population Gradients in Passively Evolving Galaxies at z -0.5ex 2 from HST/WFC3 Deep Imaging in the Hubble Ultra Deep Field. , keywords =. doi:10.1088/0004-637X/735/1/18 , archivePrefix =. 1101.0843 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/0004-637x/735/1/18

  9. [9]

    , keywords =

    The Effects of Dust on Broadband Color Gradients in Elliptical Galaxies. , keywords =. doi:10.1086/177044 , adsurl =

    work page doi:10.1086/177044

  10. [10]

    Galaxy Masses

    Galaxy masses. Rev.\ Modern Phys. , keywords =. doi:10.1103/RevModPhys.86.47 , archivePrefix =. 1309.3276 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1103/revmodphys.86.47

  11. [11]

    , keywords =

    Spectral Classification of Emission-Line Galaxies. , keywords =. doi:10.1086/191166 , adsurl =

    work page doi:10.1086/191166

  12. [12]

    Theoretical Evolution of Optical Strong Lines across Cosmic Time

    Theoretical Evolution of Optical Strong Lines across Cosmic Time. , keywords =. doi:10.1088/0004-637X/774/2/100 , archivePrefix =. 1307.0508 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/0004-637x/774/2/100

  13. [13]

    doi:10.1111/j.1365-2966.2004.07545.x , eprint =

    [OIII]/[NII] as an abundance indicator at high redshift. , keywords =. doi:10.1111/j.1365-2966.2004.07591.x , archivePrefix =. astro-ph/0401128 , primaryClass =

    work page doi:10.1111/j.1365-2966.2004.07591.x 2004

  14. [14]

    , keywords =

    De-noising of galaxy optical spectra with autoencoders. , keywords =. doi:10.1093/mnras/stad2709 , archivePrefix =. 2309.02315 , primaryClass =

    work page doi:10.1093/mnras/stad2709

  15. [15]

    Measuring the Physical Conditions in High-Redshift Star-Forming Galaxies: Insights from KBSS-MOSFIRE

    Measuring the Physical Conditions in High-redshift Star-forming Galaxies: Insights from KBSS-MOSFIRE. , keywords =. doi:10.3847/1538-4357/aae1a5 , archivePrefix =. 1711.08820 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3847/1538-4357/aae1a5

  16. [16]

    Characterizing Extreme Emission-line Galaxies. I. A Four-zone Ionization Model for Very High-ionization Emission. , keywords =. doi:10.3847/1538-4357/ac141b , archivePrefix =. 2105.12765 , primaryClass =

    work page doi:10.3847/1538-4357/ac141b

  17. [17]

    Starbursts in barred spiral galaxies. IV. On young bars and the formation of abundance gradients. , keywords =. doi:10.48550/arXiv.astro-ph/0001286 , archivePrefix =. astro-ph/0001286 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.astro-ph/0001286

  18. [18]

    2010 , Bdsk-Url-1 =

    The impact of the nitrogen-to-oxygen ratio on ionized nebula diagnostics based on [NII] emission lines. , keywords =. doi:10.1111/j.1365-2966.2009.15145.x , archivePrefix =. 0905.4621 , primaryClass =

    work page doi:10.1111/j.1365-2966.2009.15145.x 2009

  19. [19]

    , keywords =

    Mapping the Diversity of Galaxy Spectra with Deep Unsupervised Machine Learning. , keywords =. doi:10.3847/1538-3881/ac4039 , archivePrefix =. 2112.03425 , primaryClass =

    work page doi:10.3847/1538-3881/ac4039

  20. [20]

    , keywords =

    Interpreting the detection of anomalies in SDSS spectra. , keywords =. doi:10.1051/0004-6361/202556339 , archivePrefix =. 2510.05235 , primaryClass =

    work page doi:10.1051/0004-6361/202556339

  21. [21]

    , keywords =

    Identifying Anomalous DESI Galaxy Spectra with a Variational Autoencoder. , keywords =. doi:10.1093/mnras/stag010 , archivePrefix =. 2506.17376 , primaryClass =

    work page doi:10.1093/mnras/stag010

  22. [22]

    Very Strong Emission-Line Galaxies in the WISP Survey and Implications for High-Redshift Galaxies

    Very Strong Emission-line Galaxies in the WFC3 Infrared Spectroscopic Parallel Survey and Implications for High-redshift Galaxies. , keywords =. doi:10.1088/0004-637X/743/2/121 , archivePrefix =. 1109.0639 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/0004-637x/743/2/121

  23. [23]

    The Stellar Populations and Evolution of Lyman Break Galaxies

    The Stellar Populations and Evolution of Lyman Break Galaxies. , keywords =. doi:10.1086/322412 , archivePrefix =. astro-ph/0105087 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/322412

  24. [24]

    Improving the full spectrum fitting method: accurate convolution with Gauss-Hermite functions

    Improving the full spectrum fitting method: accurate convolution with Gauss-Hermite functions. , keywords =. doi:10.1093/mnras/stw3020 , archivePrefix =. 1607.08538 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stw3020

  25. [25]

    Optical-IR Spectral Energy Distributions of z>2 Lyman Break Galaxies

    Optical-Infrared Spectral Energy Distributions of Z > 2 Lyman Break Galaxies. , keywords =. doi:10.1086/300291 , archivePrefix =. astro-ph/9712216 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/300291

  26. [26]

    , keywords =

    The BPT Diagram in Cosmological Galaxy Formation Simulations: Understanding the Physics Driving Offsets at High Redshift. , keywords =. doi:10.3847/1538-4357/ac43b8 , archivePrefix =. 2201.03564 , primaryClass =

    work page doi:10.3847/1538-4357/ac43b8

  27. [27]

    Quantifying correlations between galaxy emission lines and stellar continua

    Quantifying correlations between galaxy emission lines and stellar continua. , keywords =. doi:10.1093/mnras/stv2986 , archivePrefix =. 1601.02417 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stv2986

  28. [28]

    The Journal of Open Source Software , keywords =

    pysersic: A Python package for determining galaxy structural properties via Bayesian inference, accelerated with jax. J.\ Open Source Software , keywords =. doi:10.21105/joss.05703 , archivePrefix =. 2306.05454 , primaryClass =

    work page doi:10.21105/joss.05703

  29. [29]

    An ultra-deep near-infrared spectrum of a compact quiescent galaxy at z=2.2

    An Ultra-Deep Near-Infrared Spectrum of a Compact Quiescent Galaxy at z = 2.2. , keywords =. doi:10.1088/0004-637X/700/1/221 , archivePrefix =. 0905.1692 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/0004-637x/700/1/221

  30. [30]

    MegaMorph -- multi-wavelength measurement of galaxy structure: S\'ersic profile fits to galaxies near and far

    MegaMorph - multiwavelength measurement of galaxy structure. S \'e rsic profile fits to galaxies near and far. , keywords =. doi:10.1093/mnras/stt1320 , archivePrefix =. 1307.4996 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stt1320

  31. [31]

    MegaMorph - multi-wavelength measurement of galaxy structure: complete S\'ersic profile information from modern surveys

    MegaMorph - multiwavelength measurement of galaxy structure: complete S \'e rsic profile information from modern surveys. , keywords =. doi:10.1093/mnras/sts633 , archivePrefix =. 1212.3332 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/sts633

  32. [32]

    , keywords =

    GALAPAGOS-2/GALFITM/GAMA - Multi-wavelength measurement of galaxy structure: Separating the properties of spheroid and disk components in modern surveys. , keywords =. doi:10.1051/0004-6361/202142935 , archivePrefix =. 2204.05907 , primaryClass =

    work page doi:10.1051/0004-6361/202142935

  33. [33]

    Detailed Structural Decomposition of Galaxy Images

    Detailed Structural Decomposition of Galaxy Images. , keywords =. doi:10.1086/340952 , archivePrefix =. astro-ph/0204182 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/340952

  34. [34]

    J.\ Open Source Software , keywords =

    PyAutoGalaxy: Open-Source Multiwavelength Galaxy Structure & Morphology. J.\ Open Source Software , keywords =. doi:10.21105/joss.04475 , adsurl =

    work page doi:10.21105/joss.04475

  35. [35]

    Distributions of Galaxy Spectral Types in the Sloan Digital Sky Survey

    Distributions of Galaxy Spectral Types in the Sloan Digital Sky Survey. , keywords =. doi:10.1086/422429 , archivePrefix =. astro-ph/0407061 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/422429

  36. [36]

    , keywords =

    Optical spectra of IRAS ``warm'' galaxies. , keywords =. doi:10.1086/131676 , adsurl =

    work page doi:10.1086/131676

  37. [37]

    Using Strong Lines to Estimate Abundances in Extragalactic HII Regions and Starburst Galaxies

    Using Strong Lines to Estimate Abundances in Extragalactic H II Regions and Starburst Galaxies. , keywords =. doi:10.1086/341326 , archivePrefix =. astro-ph/0206495 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/341326

  38. [38]

    2006 , Bdsk-Url-1 =

    The host galaxies and classification of active galactic nuclei. , keywords =. doi:10.1111/j.1365-2966.2006.10859.x , archivePrefix =. astro-ph/0605681 , primaryClass =

    work page doi:10.1111/j.1365-2966.2006.10859.x 2006

  39. [39]

    Theoretical Modeling of Starburst Galaxies

    Theoretical Modeling of Starburst Galaxies. , keywords =. doi:10.1086/321545 , archivePrefix =. astro-ph/0106324 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/321545

  40. [40]

    , keywords =

    Classification parameters for the emission-line spectra of extragalactic objects. , keywords =. doi:10.1086/130766 , adsurl =

    work page doi:10.1086/130766

  41. [41]

    , journal =

    Galaxy morphology in rich clusters: implications for the formation and evolution of galaxies. , journal =. doi:10.1086/157753 , adsurl =

    work page doi:10.1086/157753

  42. [42]

    Dust Absorption And The Ultraviolet Luminosity Density At z \approx 3 As Calibrated By Local Starburst Galaxies

    Dust Absorption and the Ultraviolet Luminosity Density at z -0.5ex 3 as Calibrated by Local Starburst Galaxies. , keywords =. doi:10.1086/307523 , archivePrefix =. astro-ph/9903054 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/307523

  43. [43]

    Distribution of mutual information from complete and incomplete data , url =

    Marcus Hutter and Marco Zaffalon , doi =. Distribution of mutual information from complete and incomplete data , url =. Comput.\ Statistics Data Analysis , keywords =

  44. [44]

    , keywords =

    Classification of radio sources through self-supervised learning. , keywords =. doi:10.1051/0004-6361/202554735 , archivePrefix =. 2503.19111 , primaryClass =

    work page doi:10.1051/0004-6361/202554735

  45. [45]

    and Welling, Max , title =

    Kingma, Diederik P. and Welling, Max , title =. 2019 , volume =. doi:10.1561/2200000056 , journal =

    work page doi:10.1561/2200000056 2019

  46. [46]

    2013 , issue_date =

    Bengio, Yoshua and Courville, Aaron and Vincent, Pascal , title =. 2013 , issue_date =. doi:10.1109/TPAMI.2013.50 , journal =

    work page doi:10.1109/tpami.2013.50 2013

  47. [47]

    doi:10.1038/323533a0 , adsurl =

    Learning representations by back-propagating errors , journal =. doi:10.1038/323533a0 , adsurl =

    work page doi:10.1038/323533a0

  48. [48]

    , keywords =

    Dimensionality Reduction of SDSS Spectra with Variational Autoencoders. , keywords =. doi:10.3847/1538-3881/ab9644 , archivePrefix =. 2002.10464 , primaryClass =

    work page doi:10.3847/1538-3881/ab9644 2002

  49. [49]

    ArXiv e-prints , keywords =

    A Physics-Informed Variational Autoencoder for Rapid Galaxy Inference and Anomaly Detection. ArXiv e-prints , keywords =. doi:10.48550/arXiv.2312.16687 , archivePrefix =. 2312.16687 , primaryClass =

    work page doi:10.48550/arxiv.2312.16687

  50. [50]

    Autoencoding Galaxy Spectra. II. Redshift Invariance and Outlier Detection. , keywords =. doi:10.3847/1538-3881/ace100 , archivePrefix =. 2302.02496 , primaryClass =

    work page doi:10.3847/1538-3881/ace100

  51. [51]

    Machine-assisted discovery of relationships in astronomy

    Machine-assisted discovery of relationships in astronomy. , keywords =. doi:10.1093/mnras/stt329 , archivePrefix =. 1302.5129 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stt329

  52. [52]

    Differential Galaxy Evolution in Cluster and Field Galaxies at z=0.3

    Differential Galaxy Evolution in Cluster and Field Galaxies at z -0.5ex 0.3. , keywords =. doi:10.1086/308056 , archivePrefix =. astro-ph/9906470 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1086/308056

  53. [53]

    Wilk, M. B. and Gnanadesikan, R. , title = ". Biometrika , volume =. 1968 , month =. doi:10.1093/biomet/55.1.1 , url =

    work page doi:10.1093/biomet/55.1.1 1968

  54. [54]

    Analysis of a complex of statistical variables into principal components , volume =

    Harold Hotelling , journal =. Analysis of a complex of statistical variables into principal components , volume =. 1933 , doi =

    1933

  55. [55]

    Relations Between Two Sets of Variates , urldate =

    Harold Hotelling , journal =. Relations Between Two Sets of Variates , urldate =. 1936 , doi =

    1936

  56. [56]

    , journal =

    Kolmogorov, A.N. , journal =

  57. [57]

    Smirnov , title =

    N. Smirnov , title =. Ann.\ Math.\ Statistics , number =. 1948 , doi =

    1948

  58. [58]

    Scandinavian Actuarial J

    Harald Cram\'er , title =. Scandinavian Actuarial J. , volume =. 1928 , publisher =

    1928

  59. [59]

    arXiv e-prints , keywords =

    Practical Guidance for Bayesian Inference in Astronomy. arXiv e-prints , keywords =. doi:10.48550/arXiv.2302.04703 , archivePrefix =. 2302.04703 , primaryClass =

    work page doi:10.48550/arxiv.2302.04703

  60. [60]

    , keywords =

    Testing the consistency of dust laws in SN Ia host galaxies: a BAYESN examination of Foundation DR1. , keywords =. doi:10.1093/mnras/stab2849 , archivePrefix =. 2102.05678 , primaryClass =

    work page doi:10.1093/mnras/stab2849

  61. [61]

    , keywords =

    A hierarchical Bayesian SED model for Type Ia supernovae in the optical to near-infrared. , keywords =. doi:10.1093/mnras/stab3496 , archivePrefix =. 2008.07538 , primaryClass =

    work page doi:10.1093/mnras/stab3496 2008

  62. [62]

    Type Ia Supernova Light Curve Inference: Hierarchical Bayesian Analysis in the Near Infrared

    Type Ia Supernova Light-Curve Inference: Hierarchical Bayesian Analysis in the Near-Infrared. , keywords =. doi:10.1088/0004-637X/704/1/629 , archivePrefix =. 0908.0536 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/0004-637x/704/1/629

  63. [63]

    Type Ia Supernova Light Curve Inference: Hierarchical Models in the Optical and Near Infrared

    Type Ia Supernova Light Curve Inference: Hierarchical Models in the Optical and Near-infrared. , keywords =. doi:10.1088/0004-637X/731/2/120 , archivePrefix =. 1011.5910 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1088/0004-637x/731/2/120

  64. [64]

    Do spectra improve distance measurements of Type Ia supernovae?

    Do spectra improve distance measurements of Type Ia supernovae?. , keywords =. doi:10.1051/0004-6361/201015792 , archivePrefix =. 1012.0005 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/201015792

  65. [65]

    , keywords =

    On the Application of Bayesian Leave-one-out Cross-validation to Exoplanet Atmospheric Analysis. , keywords =. doi:10.3847/1538-3881/acab67 , archivePrefix =. 2212.03872 , primaryClass =

    work page doi:10.3847/1538-3881/acab67

  66. [66]

    , keywords =

    First semi-empirical test of the white dwarf mass-radius relationship using a single white dwarf via astrometric microlensing. , keywords =. doi:10.1093/mnras/stac3532 , archivePrefix =. 2206.01814 , primaryClass =

    work page doi:10.1093/mnras/stac3532

  67. [67]

    arXiv e-prints , keywords =

    Methods for Incorporating Model Uncertainty into Exoplanet Atmospheric Analysis. arXiv e-prints , keywords =. doi:10.48550/arXiv.2310.03713 , archivePrefix =. 2310.03713 , primaryClass =

    work page doi:10.48550/arxiv.2310.03713

  68. [68]

    arXiv e-prints , keywords =

    Bringing 2D Eclipse Mapping out of the Shadows with Leave-one-out Cross-validation. arXiv e-prints , keywords =. doi:10.48550/arXiv.2310.03733 , archivePrefix =. 2310.03733 , primaryClass =

    work page doi:10.48550/arxiv.2310.03733

  69. [69]

    arXiv e-prints , keywords =

    Pareto Smoothed Importance Sampling. arXiv e-prints , keywords =. doi:10.48550/arXiv.1507.02646 , archivePrefix =. 1507.02646 , primaryClass =

    work page doi:10.48550/arxiv.1507.02646

  70. [70]

    Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC

    Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Statistics & Computing , keywords =. doi:10.1007/s11222-016-9696-4 , archivePrefix =. 1507.04544 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1007/s11222-016-9696-4

  71. [71]

    Prospects for resolving the Hubble constant tension with standard sirens

    Prospects for Resolving the Hubble Constant Tension with Standard Sirens. , keywords =. doi:10.1103/PhysRevLett.122.061105 , archivePrefix =. 1802.03404 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1103/physrevlett.122.061105

  72. [72]

    , keywords =

    Dark Energy Survey year 1 results: Joint analysis of galaxy clustering, galaxy lensing, and CMB lensing two-point functions. , keywords =. doi:10.1103/PhysRevD.100.023541 , archivePrefix =. 1810.02322 , primaryClass =

    work page doi:10.1103/physrevd.100.023541

  73. [73]

    , keywords =

    General framework for cosmological dark matter bounds using N -body simulations. , keywords =. doi:10.1103/PhysRevD.103.043526 , archivePrefix =. 2007.13751 , primaryClass =

    work page doi:10.1103/physrevd.103.043526 2007

  74. [74]

    , keywords =

    Forward Modeling of Galaxy Populations for Cosmological Redshift Distribution Inference. , keywords =. doi:10.3847/1538-4365/ac9583 , archivePrefix =. 2207.05819 , primaryClass =

    work page doi:10.3847/1538-4365/ac9583

  75. [75]

    , keywords =

    Hierarchical Bayesian Inference of Photometric Redshifts with Stellar Population Synthesis Models. , keywords =. doi:10.3847/1538-4365/ac9d99 , archivePrefix =. 2207.07673 , primaryClass =

    work page doi:10.3847/1538-4365/ac9d99

  76. [76]

    , keywords =

    PopSED: Population-level Inference for Galaxy Properties from Broadband Photometry with Neural Density Estimation. , keywords =. doi:10.3847/1538-3881/ad0be4 , archivePrefix =. 2309.16958 , primaryClass =

    work page doi:10.3847/1538-3881/ad0be4

  77. [77]

    Harris and K

    Charles R. Harris and K. Jarrod Millman and St. Array programming with. 2020 , month = sep, journal =. doi:10.1038/s41586-020-2649-2 , publisher =

    work page doi:10.1038/s41586-020-2649-2 2020

  78. [78]

    and Haberland, Matt and Reddy, Tyler and Cournapeau, David and Burovski, Evgeni and Peterson, Pearu and Weckesser, Warren and Bright, Jonathan and

    Virtanen, Pauli and Gommers, Ralf and Oliphant, Travis E. and Haberland, Matt and Reddy, Tyler and Cournapeau, David and Burovski, Evgeni and Peterson, Pearu and Weckesser, Warren and Bright, Jonathan and. Nature Methods , year =

  79. [79]

    Hunter, J. D. , Title =. Comput.\ Sci.\ Eng. , Volume =

  80. [80]

    arXiv e-prints , keywords =

    Statistical Aspects of X-ray Spectral Analysis. arXiv e-prints , keywords =. doi:10.48550/arXiv.2309.05705 , archivePrefix =. 2309.05705 , primaryClass =

    work page doi:10.48550/arxiv.2309.05705

Showing first 80 references.