-
Euclid Quick Data Release (Q1) Exploring galaxy properties with a multi-modal foundation model
Authors:
Euclid Collaboration,
M. Siudek,
M. Huertas-Company,
M. Smith,
G. Martinez-Solaeche,
F. Lanusse,
S. Ho,
E. Angeloudi,
P. A. C. Cunha,
H. Domínguez Sánchez,
M. Dunn,
Y. Fu,
P. Iglesias-Navarro,
J. Junais,
J. H. Knapen,
B. Laloux,
M. Mezcua,
W. Roster,
G. Stevens,
J. Vega-Ferrero,
N. Aghanim,
B. Altieri,
A. Amara,
S. Andreon,
N. Auricchio
, et al. (299 additional authors not shown)
Abstract:
Modern astronomical surveys, such as the Euclid mission, produce high-dimensional, multi-modal data sets that include imaging and spectroscopic information for millions of galaxies. These data serve as an ideal benchmark for large, pre-trained multi-modal models, which can leverage vast amounts of unlabelled data. In this work, we present the first exploration of Euclid data with AstroPT, an autor…
▽ More
Modern astronomical surveys, such as the Euclid mission, produce high-dimensional, multi-modal data sets that include imaging and spectroscopic information for millions of galaxies. These data serve as an ideal benchmark for large, pre-trained multi-modal models, which can leverage vast amounts of unlabelled data. In this work, we present the first exploration of Euclid data with AstroPT, an autoregressive multi-modal foundation model trained on approximately 300 000 optical and infrared Euclid images and spectral energy distributions (SEDs) from the first Euclid Quick Data Release. We compare self-supervised pre-training with baseline fully supervised training across several tasks: galaxy morphology classification; redshift estimation; similarity searches; and outlier detection. Our results show that: (a) AstroPT embeddings are highly informative, correlating with morphology and effectively isolating outliers; (b) including infrared data helps to isolate stars, but degrades the identification of edge-on galaxies, which are better captured by optical images; (c) simple fine-tuning of these embeddings for photometric redshift and stellar mass estimation outperforms a fully supervised approach, even when using only 1% of the training labels; and (d) incorporating SED data into AstroPT via a straightforward multi-modal token-chaining method improves photo-z predictions, and allow us to identify potentially more interesting anomalies (such as ringed or interacting galaxies) compared to a model pre-trained solely on imaging data.
△ Less
Submitted 19 March, 2025;
originally announced March 2025.
-
Euclid Quick Data Release (Q1), A first look at the fraction of bars in massive galaxies at $z<1$
Authors:
Euclid Collaboration,
M. Huertas-Company,
M. Walmsley,
M. Siudek,
P. Iglesias-Navarro,
J. H. Knapen,
S. Serjeant,
H. J. Dickinson,
L. Fortson,
I. Garland,
T. Géron,
W. Keel,
S. Kruk,
C. J. Lintott,
K. Mantha,
K. Masters,
D. O'Ryan,
J. J. Popp,
H. Roberts,
C. Scarlata,
J. S. Makechemu,
B. Simmons,
R. J. Smethurst,
A. Spindler,
M. Baes
, et al. (314 additional authors not shown)
Abstract:
Stellar bars are key structures in disc galaxies, driving angular momentum redistribution and influencing processes such as bulge growth and star formation. Quantifying the bar fraction as a function of redshift and stellar mass is therefore important for constraining the physical processes that drive disc formation and evolution across the history of the Universe. Leveraging the unprecedented res…
▽ More
Stellar bars are key structures in disc galaxies, driving angular momentum redistribution and influencing processes such as bulge growth and star formation. Quantifying the bar fraction as a function of redshift and stellar mass is therefore important for constraining the physical processes that drive disc formation and evolution across the history of the Universe. Leveraging the unprecedented resolution and survey area of the Euclid Q1 data release combined with the Zoobot deep-learning model trained on citizen-science labels, we identify 7711 barred galaxies with $M_* \gtrsim 10^{10}M_\odot$ in a magnitude-selected sample $I_E < 20.5$ spanning $63.1 deg^2$. We measure a mean bar fraction of $0.2-0.4$, consistent with prior studies. At fixed redshift, massive galaxies exhibit higher bar fractions, while lower-mass systems show a steeper decline with redshift, suggesting earlier disc assembly in massive galaxies. Comparisons with cosmological simulations (e.g., TNG50, Auriga) reveal a broadly consistent bar fraction, but highlight overpredictions for high-mass systems, pointing to potential over-efficiency in central stellar mass build-up in simulations. These findings demonstrate Euclid's transformative potential for galaxy morphology studies and underscore the importance of refining theoretical models to better reproduce observed trends. Future work will explore finer mass bins, environmental correlations, and additional morphological indicators.
△ Less
Submitted 19 March, 2025;
originally announced March 2025.
-
Euclid preparation. LXVIII. Extracting physical parameters from galaxies with machine learning
Authors:
Euclid Collaboration,
I. Kovačić,
M. Baes,
A. Nersesian,
N. Andreadis,
L. Nemani,
Abdurro'uf,
L. Bisigello,
M. Bolzonella,
C. Tortora,
A. van der Wel,
S. Cavuoti,
C. J. Conselice,
A. Enia,
L. K. Hunt,
P. Iglesias-Navarro,
E. Iodice,
J. H. Knapen,
F. R. Marleau,
O. Müller,
R. F. Peletier,
J. Román,
R. Ragusa,
P. Salucci,
T. Saifollahi
, et al. (265 additional authors not shown)
Abstract:
The Euclid mission is generating a vast amount of imaging data in four broadband filters at high angular resolution. This will allow the detailed study of mass, metallicity, and stellar populations across galaxies, which will constrain their formation and evolutionary pathways. Transforming the Euclid imaging for large samples of galaxies into maps of physical parameters in an efficient and reliab…
▽ More
The Euclid mission is generating a vast amount of imaging data in four broadband filters at high angular resolution. This will allow the detailed study of mass, metallicity, and stellar populations across galaxies, which will constrain their formation and evolutionary pathways. Transforming the Euclid imaging for large samples of galaxies into maps of physical parameters in an efficient and reliable manner is an outstanding challenge. We investigate the power and reliability of machine learning techniques to extract the distribution of physical parameters within well-resolved galaxies. We focus on estimating stellar mass surface density, mass-averaged stellar metallicity and age. We generate noise-free, synthetic high-resolution imaging data in the Euclid photometric bands for a set of 1154 galaxies from the TNG50 cosmological simulation. The images are generated with the SKIRT radiative transfer code, taking into account the complex 3D distribution of stellar populations and interstellar dust attenuation. We use a machine learning framework to map the idealised mock observational data to the physical parameters on a pixel-by-pixel basis. We find that stellar mass surface density can be accurately recovered with a $\leq 0.130 {\rm \,dex}$ scatter. Conversely, stellar metallicity and age estimates are, as expected, less robust, but still contain significant information which originates from underlying correlations at a sub-kpc scale between stellar mass surface density and stellar population properties.
△ Less
Submitted 31 March, 2025; v1 submitted 24 January, 2025;
originally announced January 2025.
-
ZF-UDS-7329: A relic galaxy in the early Universe
Authors:
Eduardo A. Hartmann,
Ignacio Martín-Navarro,
Marc Huertas-Company,
João P. V. Benedetti,
Patricia Iglesias-Navarro,
Alexandre Vazdekis,
Mireia Montes
Abstract:
The formation time scales of quiescent galaxies can be estimated in two different ways: by their star formation history and by their chemistry. Previously, the methods yielded conflicting results, especially when considering $α$-enhanced objects. This is primarily due to the time resolution limitations of very old stellar populations, which prevent accurately constraining their star formation hist…
▽ More
The formation time scales of quiescent galaxies can be estimated in two different ways: by their star formation history and by their chemistry. Previously, the methods yielded conflicting results, especially when considering $α$-enhanced objects. This is primarily due to the time resolution limitations of very old stellar populations, which prevent accurately constraining their star formation histories. We analysed the JWST observations of the extremely massive galaxy ZF-UDS-7329 at z$\sim$3.2 and show that the higher time resolution necessary to match the chemical formation time scales using stellar population synthesis can be achieved by studying galaxies at high redshift. We compare the massive galaxy to the well-known relic galaxy NGC~1277, arguing that ZF-UDS-7329 is an early Universe example of the cores of present-day massive elliptical galaxies or, if left untouched, a relic galaxy.
△ Less
Submitted 27 January, 2025; v1 submitted 10 January, 2025;
originally announced January 2025.
-
Deriving the star formation histories of galaxies from spectra with simulation-based inference
Authors:
Patricia Iglesias-Navarro,
Marc Huertas-Company,
Ignacio Martín-Navarro,
Johan H. Knapen,
Emilie Pernet
Abstract:
High-resolution galaxy spectra encode information about the stellar populations within galaxies. The properties of the stars, such as their ages, masses, and metallicities, provide insights into the underlying physical processes that drive the growth and transformation of galaxies over cosmic time.
We explore a simulation-based inference (SBI) workflow to infer from optical absorption spectra th…
▽ More
High-resolution galaxy spectra encode information about the stellar populations within galaxies. The properties of the stars, such as their ages, masses, and metallicities, provide insights into the underlying physical processes that drive the growth and transformation of galaxies over cosmic time.
We explore a simulation-based inference (SBI) workflow to infer from optical absorption spectra the posterior distributions of metallicities and the star formation histories (SFHs) of galaxies (i.e. the star formation rate as a function of time).
We generated a dataset of synthetic spectra to train and test our model using the spectroscopic predictions of the MILES stellar population library and non-parametric SFHs. We reliably estimate the mass assembly of an integrated stellar population with well-calibrated uncertainties. Specifically, we reach a score of $0.97\,R^2$ for the time at which a given galaxy from the test set formed $50\%$ of its stellar mass, obtaining samples of the posteriors in only $10^{-4}$\,s. We then applied the pipeline to real observations of massive elliptical galaxies, recovering the well-known relationship between the age and the velocity dispersion, and show that the most massive galaxies ($σ\sim300$ km/s) built up to 90\% of their total stellar masses within $1$\,Gyr of the Big Bang. The inferred properties also agree with the state-of-the-art inversion codes, but the inference is performed up to five orders of magnitude faster.
This SBI approach coupled with machine learning and applied to full spectral fitting makes it possible to address large numbers of galaxies while performing a thick sampling of the posteriors. It will allow both the deterministic trends and the inherent uncertainties of the highly degenerated inversion problem to be estimated for large and complex upcoming spectroscopic surveys, such as DESI, WEAVE, or 4MOST.
△ Less
Submitted 28 June, 2024; v1 submitted 26 June, 2024;
originally announced June 2024.
-
The building up of observed stellar scaling relations of massive galaxies and the connection to black hole growth in the TNG50 simulation
Authors:
S. Varma,
M. Huertas-Company,
A. Pillepich,
D. Nelson,
V. Rodriguez-Gomez,
A. Dekel,
S. M. Faber,
P. Iglesias-Navarro,
D. C. Koo,
J. Primack
Abstract:
[abridged] We study how mock-observed stellar morphological and structural properties of massive galaxies are built up between $z=0.5$ and $z=3$ in the TNG50 cosmological simulation. We generate mock images with the properties of the CANDELS survey and derive Sersic parameters and optical rest-frame morphologies as usually done in the observations. Overall, the simulation reproduces the observed e…
▽ More
[abridged] We study how mock-observed stellar morphological and structural properties of massive galaxies are built up between $z=0.5$ and $z=3$ in the TNG50 cosmological simulation. We generate mock images with the properties of the CANDELS survey and derive Sersic parameters and optical rest-frame morphologies as usually done in the observations. Overall, the simulation reproduces the observed evolution of the abundances of different galaxy morphological types of star-forming and quiescent galaxies. The $\log{M_*}-\log R_e$ and $\log{M_*}-\logΣ_1$ relations of the simulated star-forming and quenched galaxies also match the observed slopes and zeropoints to within 1-$σ$. In the simulation, galaxies increase their observed central stellar mass density ($Σ_1$) and transform in morphology from irregular/clumpy systems to normal Hubble-type systems in the Star Formation Main Sequence at a characteristic stellar mass of $\sim 10^{10.5}~M_\odot$. This morphological transformation is connected to the activity of the central Super Massive Black Holes (SMBHs). At low stellar masses ($10^9$ < $M_*/M_\odot$ < $10^{10}$) SMBHs grow rapidly, while at higher mass SMBHs switch into the kinetic feedback mode and grow more slowly. During this low-accretion phase, SMBH feedback leads to the quenching of star-formation, along with a simultaneous growth in $Σ_1$. More compact massive galaxies grow their SMBHs faster than extended ones of the same mass and end up quenching earlier. In the TNG50 simulation, SMBHs predominantly grow via gas accretion before galaxies quench, and $Σ_1$ increases substantially after SMBH growth slows down. The simulation predicts therefore that quiescent galaxies have higher $Σ_1$ values than star-forming galaxies for the same SMBH mass, which disagrees with alternative models, and may potentially be in tension with some observations.
△ Less
Submitted 22 October, 2021;
originally announced October 2021.
