Flagship Simulation Galaxy Catalogue

• Flagship Simulation Galaxy Catalogue is a comprehensive synthetic dataset developed using advanced cosmological simulations, offering realistic galaxy properties for modern survey design.
• Its methodology integrates N-body/hydrodynamic techniques, galaxy–halo connection models, and machine learning SED approaches to accurately reproduce photometry, morphology, and lensing signals.
• It serves as an end-to-end testbed for pipeline validation, survey planning, and cosmological analysis by benchmarking against multi-wavelength observations and theoretical models.

A Flagship Simulation Galaxy Catalogue is a large-scale synthetic dataset providing mock galaxy populations with realistic positions, photometry, morphology, evolution, and lensing properties, all generated from a cosmological N-body or hydrodynamical simulation, and calibrated or validated against multi-wavelength observations and theoretical models. Such catalogues are central to survey design, cosmological inference, and pipeline validation for current and forthcoming wide-field galaxy surveys. This article surveys the methodologies, data structures, validation strategies, and usage paradigms underpinning recent flagship simulation galaxy catalogues, with primary examples from the Euclid Flagship (Collaboration et al., 22 May 2024), MICE Grand Challenge (Crocce et al., 2013), CSST Main Survey (Wei et al., 13 Nov 2025), Uchuu-UM (Aung et al., 2022), SIBELIUS-DARK (McAlpine et al., 2022), Illustris Synthetic Observatory (Torrey et al., 2014), and EAGLE (McAlpine et al., 2015).

1. Cosmological Simulation Basis

All flagship catalogues derive their mock galaxies from a large cosmological N-body or hydrodynamic simulation. The simulation volume, mass and spatial resolutions, and cosmology are selected to match or exceed the requirements of target surveys:

Catalogue	Box Size	N_particles	Mass Resolution
Euclid Flagship	$3600\,h^{-1}$ Mpc	$16{,}000^3\sim 4\times10^{12}$	$10^9\,h^{-1}M_\odot$
MICE-GC	$3\,h^{-1}$ Gpc	$4096^3$	$2.9\times10^{10}\,h^{-1}M_\odot$
Uchuu-UM	$2\,h^{-1}$ Gpc	$12{,}800^3$	$3.27\times10^8\,h^{-1}M_\odot$
CSST JT1G	$1\,h^{-1}$ Gpc	$6144^3$	$3.72\times10^8\,h^{-1}M_\odot$

Simulation cosmologies are typically Planck-consistent ($\Omega_\mathrm{m}\sim0.31$, $\sigma_8\sim0.81$, $h\sim0.67$) for contemporary datasets. The choice of simulation sets hard limits on the mass and spatial resolution, dynamic range, and statistical robustness to cosmic variance (especially at BAO and weak lensing scales). Force softening is set to ensure accurate small-scale structure: e.g., $4.5\,h^{-1}$ kpc for FS2 (Collaboration et al., 22 May 2024).

Semi-analytic models (SAMs; e.g., SIBELIUS-DARK (McAlpine et al., 2022), CSST (Wei et al., 13 Nov 2025)), empirical models (UniverseMachine for Uchuu-UM (Aung et al., 2022)), and full hydrodynamical codes (e.g., EAGLE (McAlpine et al., 2015), Illustris (Torrey et al., 2014)) are used to model baryonic galaxy formation physics atop gravity-only $N$-body outputs.

2. Galaxy Population and Property Assignment

Populating dark matter halos with galaxies requires a methodology that can reproduce the empirical galaxy–halo connection, luminosity functions, scaling relations, and clustering. Common techniques include:

• Halo Occupation Distribution (HOD): Assigns centrals ($\langle N_\mathrm{cen}\rangle=1$ per halo) and satellites via power-law scaling ($$\langle N_\mathrm{sat}(M)\rangle\propto (M/M_1)^{\alpha}$$) with satellite spatial and velocity distributions drawn from NFW profiles (Collaboration et al., 22 May 2024, Crocce et al., 2013).
• Abundance Matching (AM): Rank-orders halos and galaxies to enforce the observed cumulative galaxy luminosity or stellar mass function (e.g., $n_\mathrm{gal}(>L)=\int n_\mathrm{halo}(>M)\,dM$).
• Conditional Luminosity Functions (CLF): Assigns the number of galaxies of a given luminosity in halos of given mass by Schechter-type fits (Collaboration et al., 22 May 2024).
• Semi-Analytic Models (SAMs): Use merger trees to compute the evolution of baryonic components, star formation, feedback, size–mass relations, etc. (e.g., CSST (Wei et al., 13 Nov 2025), SIBELIUS-DARK (McAlpine et al., 2022)).
• Empirical Models: UniverseMachine parameterizes SFR as a function of (sub)halo properties, accretion history, and redshift, calibrated by MCMC against observed SMFs, SFR, clustering, and quenched fractions (Aung et al., 2022).

Every galaxy is assigned 3D position, velocity (including peculiar), stellar mass, star formation rate (SFR), metallicity, and, where relevant, merger or accretion history.

3. Spectral Energy Distributions, Photometry, and Morphology

Galaxy SEDs are crucial for producing observable-matched magnitudes in real and synthetic bands, for photo-$z$ validation, and for realistic image simulation. The methods include:

• Template Assignment: SEDs from the COSMOS library (Collaboration et al., 22 May 2024), SDSS, or other libraries, assigned by rest-frame color quantile, with empirical extinction and emission-line models.
• Stellar Population Synthesis: In hydrodynamical runs, star particle ages and metallicities are mapped onto SED templates (e.g., SB99 for Illustris (Torrey et al., 2014)), and summed for the total galaxy light.
• Machine Learning SED Modeling: Deep neural networks trained on radiative-transfer outputs (e.g., STARDUSTER for CSST (Wei et al., 13 Nov 2025)) map galaxy physical parameters and SFH to compressed SED basis (PCA), typically achieving <2% error over $\lambda=60$–$1100$ nm.
• Morphology: Disk+bulge decompositions using empirical relations (e.g., mass–size–redshift scaling; (Wei et al., 13 Nov 2025, Collaboration et al., 22 May 2024)). Sérsic index $n^S$ and axis ratio $q$ distributions are assigned conditional on galaxy type, color, or SF history.
• Image Generation: Flagship hydrodynamical catalogues (e.g., Illustris Synthetic Observatory (Torrey et al., 2014)) provide multi-band synthetic images (28,000 FITS cubes in 36 filters for 7,000 galaxies) with adaptive smoothing, optional dust screen models, and outputs for pipeline testing and SED fitting.

Photometry is generated in multiple bands relevant to the target survey (e.g., Euclid VIS/NISP YJH, SDSS ugriz, JWST NIRCAM, CSST filters), including dust extinction and observational K/evolutionary corrections.

4. Lensing and Clustering: Validation and Applications

Flagship catalogues include galaxy shapes and lensing properties (shear, convergence, magnification) for forward-modeling weak lensing observables:

• Ray-Tracing: Multi-plane curved-sky algorithms project N-body density onto shells (HEALPix $N_{\rm side}=8192$ or higher), solve for the lensing potential $\phi$, and propagate ray deflections to compute $\kappa, \gamma, \mu$ at galaxy positions. Shear two-point and tomographic correlations ($\xi_\pm(\theta)$) reproduce theoretical predictions to $\lesssim10\%$ accuracy at arcminute/multipole scales (Collaboration et al., 22 May 2024, Wei et al., 13 Nov 2025).
• Clustering Validations: Measured two-point ($\xi(r)$, $\xi_\ell(s)$) and three-point (reduced $Q_3(\theta)$) correlation functions are compared against observations and semi-analytic or Halo Model fits. Notably, the BAO peak is recovered with correct amplitude and scale-dependent bias at $k\sim0.05~h\,\mathrm{Mpc}^{-1}$ (Crocce et al., 2013, Collaboration et al., 22 May 2024).
• Galaxy–Galaxy Lensing: Cross-correlation signals (e.g., $\langle\gamma_t\rangle(\theta)$) are validated against linear-bias Halofit predictions.
• Intrinsic Alignments: 3D ellipsoid axis ratios and misalignments are empirically calibrated to low- and high-redshift samples (e.g., COSMOS, LOWZ, SDSS), enabling forward modeling of IA signals.

These modules enable flagship catalogues to serve as end-to-end testbeds for lensing, clustering, covariance estimation, RSD modelling, and BAO analyses.

5. Data Structure, Access, and Usage

Flagship catalogues feature flat-file or relational database architectures supporting efficient subsetting, data join operations, and scalable analysis:

• File Formats: Parquet, HDF5, FITS binary tables, and SQL/VO-enabled flat files. Each object typically has hundreds of columns, including photometry, positions, SED weights, morphology, lensing, emission lines, physical and HOD properties.
• Access Portals: CosmoHub (Euclid Flagship, MICE), EAGLE SQL Database, SIBELIUS-DARK VirgoDB, Uchuu-UM Skies & Universes, CSST data release servers. APIs and example codes are provided for query and download.
• Usage Guidelines: For statistical consistency with observations, practitioners are advised to apply the same selection functions (e.g., $I_E<26$), morphology or redshift cuts, post-process images with survey-specific PSF/noise, and implement the same masks and systematic corrections as the survey of interest.
• Pipeline Integration: Catalogues are explicitly designed to support object detection, deblending, photometric-$z$ estimation, SED fitting (e.g., FAST (Torrey et al., 2014)), morphology classification (e.g., GALFIT, statmorph), and cosmological inference pipelines.

6. Model Calibration, Validation, and Quantified Limitations

All flagship catalogues are validated against a suite of low- and high-redshift observables:

• Stellar Mass/Luminosity Functions: Agreement is typically within the combined systematics of data and modeling out to the completeness limits in $M_*, M_K$, and emission-line number densities (Collaboration et al., 22 May 2024, Aung et al., 2022, Wei et al., 13 Nov 2025, McAlpine et al., 2022).
• Clustering/Quenching Fractions: Satellite conditional mass functions, quenched fractions $f_\mathrm{red}$ and their radial dependence are checked versus SDSS, PRIMUS, DES-Y3, GAMA, and other surveys (Aung et al., 2022, Wei et al., 13 Nov 2025).
• Lensing Signals: Shear E/B-mode separation, convergence power spectra versus Halofit or higher-order theory; magnification bias cross-correlation; 3-point aperture mass statistics (Collaboration et al., 22 May 2024, Wei et al., 13 Nov 2025).
• Image/Synthetic Fluxes: Systematic errors in photometric stellar mass (e.g., $\Delta\log M_* \lesssim 0.1$ dex with properly tuned SED grid priors (Torrey et al., 2014)), impact of SED discretization, and survey bandpasses are quantified. Strong model dependence on metallicity and age priors is observed.

Common limitations include finite mass/spatial resolution (e.g., under-resolving subhaloes with $M_\mathrm{sub}\lesssim10^9 M_\odot$), lack of full radiative transfer for dust, omission of AGN light or cosmic ray contamination in hydrodynamic runs, and incomplete capture of baryonic feedback in $N$-body or semi-analytic runs. Empirical quenching models may not accurately reflect environmental processes at the level of individual galaxies. The impact of orphan galaxy tracking and subhalo disruption on cluster-centric profiles requires caution, especially within inner $\sim0.1\,r_{200m}$.

A plausible implication is that systematics arising from these limitations can exceed the statistical precision of next-generation cosmological experiments in specific regimes; thus, end-users are encouraged to test science pipelines under realistic uncertainties using multiple catalog variants.

7. Scientific and Operational Applications

Flagship simulation catalogues are now indispensable for:

• Survey Planning: Estimating survey depth, number counts, and volume completeness for next-generation imaging/spectroscopic programmes (Euclid, CSST, LSST, Roman, DESI, SPHEREx).
• Pipeline Validation: Performing full end-to-end image and catalog processing to quantify systematic errors in object detection, lensing shear recovery, deblending, photometric redshifts, and fiber assignment or slitless spectroscopy.
• Cosmological Analysis: Enabling joint clustering, weak lensing, and emission-line galaxy (ELG) studies with consistent covariances and realistic selection functions. Supporting model-independent cross-correlation and higher-order statistics.
• Machine Learning Training: Providing large, labeled, observationally-matched datasets for development and validation of machine-learning methods for source classification, redshift estimation, or anomaly detection.
• Theoretical Benchmarking: Testing galaxy formation and evolution models at volume and resolution scales that cannot be achieved in ab initio hydrodynamical runs, enabling robust comparison of competing baryonic recipes or empirical parameterizations.

Flagship catalogues serve as a normative baseline and testbed for community-wide methodological development and independent replication of key science analyses.