SE3D: Testing the recovery of stellar population, dust and structural properties on mock-observed toy model and simulated galaxies

TL;DR

SE3D introduces a radiative-transfer emulator-based framework to infer the 3D distribution of stars and dust in galaxies from multiwavelength photometry and spatially resolved structure. By testing on both toy-model SKIRT-generated galaxies and TNG50 mock galaxies, the study shows robust recovery of total stellar mass, dust mass, and star formation rate, along with their radial extents, often at ~0.1 dex accuracy; age-related quantities are correlated but more uncertain, and metallicity remains challenging to constrain. Model mismatch, especially in star-formation history, limits precision, while reduced data quality or missing resolved information degrades performance in predictable ways. The work highlights the potential of SE3D for JWST-era data while underscoring the need for more realistic dust physics and expanded modelling to bridge gaps between simulated and observed galaxy populations.

Abstract

The translation from direct observables to physical properties of galaxies is a key step in reconstructing their evolutionary histories. Variations in stellar populations and star-dust geometry can induce inhomogeneous mass-to-light ratios, complicating this process. SE3D is a novel modelling framework, built around a radiative transfer emulator, aimed at tackling this problem. In this paper, we test the ability of SE3D to recover known intrinsic properties of toy model and TNG50 simulated galaxies from mock observations of their multi-wavelength photometric and structural properties. We find an encouraging performance for several key characteristics, including the bulk stellar mass, dust mass and SFR, as well as their respective radial extents. We point out limitations, and investigate the impact of various sources of model mismatch. Among them, mismatch in the shapes of star formation histories contributes most, with radial and azimuthal structure and stellar metallicity distributions playing a progressively more minor role. We also analyse the evolution from z=2 to z=0 of resolved stellar and dust properties of TNG galaxies, as measured intrinsically and expressed in their distribution across UVJ and IRX-$β$ diagnostic diagrams. We test different methods to assign dust to the simulation, and find a persistent lack of Mdust/Mstar evolution and a more limited dynamic range across the diagnostic diagrams compared to observations.

Figures (11)

• Figure 1: Structural and SFH properties of an example TNG galaxy contrasted with the closest toy model approximation. The PDF of the radial and vertical stellar mass and dust distribution as well as the stellar age distribution are shown in blue for TNG, and in red for the associated toy model approximation. The distribution of stellar mass across the age-radius plane (bottom-middle panel) is contrasted to that for the closest parametrized toy model (bottom-right panel). Derived properties of the TNG galaxy are listed in the right space.
• Figure 2: Top: Age gradient $\nabla {\rm{age}}_{\rm{w}}$ versus the relative radial extent of dust and stellar distributions, $R_{\rm{dust}}/R_{\rm{star}}$, of TNG50 galaxies at redshifts $z = 0$, 1 and 2. Bottom: Geometric thickness of the dust versus stellar distribution. In each panel and for each redshift, the filled circle and cross mark the median and central 68th percentile of the distribution, respectively.
• Figure 3: $UVJ$ diagram for massive SFGs in TNG50 ( left), for the full set of toy model galaxies in our SKIRT library ( middle), and for a mass-complete sample of observed galaxies at $0.5 < z < 2.5$ ( right). Contours depict the regions enclosing 20%, 60%, 90% and 99.9% of the data. The colours predicted for TNG galaxies get bluer with increasing redshift and do not extend into the red $V-J$ regime.
• Figure 4: $UVJ$ diagram for our toy model library, colour-coded by the effective reddening, $E(B-V)$, and the effective visual attenuation, $A_V$. Vectors of increasing reddening versus attenuation have different slopes in the $UVJ$ plane.
• Figure 5: IRX - $\rm{\beta}$ diagram for massive SFGs in TNG50 ( left), for the full set of toy model galaxies in our SKIRT library ( middle), and for a mass-limited sample of observed galaxies at $0.5 < z < 2.5$ ( right). Contours depict the regions enclosing 20%, 60%, 90% and 99.9% of the data. On the right panel, we overlay the distribution of available $\beta$ from observations, whilst the IRX - $\rm{\beta}$ contours are only shown for galaxies with available IRX measurements.