A Universe in a Hard Drive: How Euclid’s Flagship Mock Creates 3.4 Billion Virtual Galaxies to Test Dark Energy Science

Euclid’s Flagship mock is a single, gigantic simulated sky that turns theoretical dark matter into virtual galaxies, shapes, and spectra so realistic that scientists can rehearse the mission’s hardest measurements before the satellite ever looks up. Built from a four trillion particle N-body run and refined by observational recipes, the catalogue supplies a playground for weak lensing, galaxy clustering, and dozens of other cosmic tests that will sharpen our picture of dark matter and dark energy.

Why create a “mock” universe?

When cosmologists aim to measure dark energy, they face two challenges. First, the signal is subtle, encoded in the tiny distortions of galaxy shapes by weak gravitational lensing and in the clustering of galaxies across cosmic time. Second, observational surveys are subject to noise, selection effects, and instrument quirks that can bias results if not understood. A realistic synthetic universe, or “mock catalogue,” lets teams practise analysis pipelines, hunt for systematic errors, and quantify statistical power before real data arrive. The Euclid Collaboration produced the Flagship mock for exactly these purposes, and to supply the community with a deep, public resource for method development.

The scale of the project: the problem the team set out to solve

Euclid’s science goals require both precision and volume. To measure the equation of state of dark energy to the mission targets, Euclid must combine high-quality shape measurements for billions of galaxies with redshifts for tens of millions. Realistic support simulations therefore need to cover a cosmological volume comparable to Euclid’s survey area and resolve the haloes that host even relatively faint galaxies. Hydrodynamic simulations that model gas, stars, and feedback at the needed volumes remain computationally expensive, so the collaboration chose instead to run a massive gravity-only N-body simulation and then populate haloes with galaxies using empirical recipes calibrated to observations. The Flagship simulation was built to solve that exact tradeoff, providing both large volume and sufficient mass resolution to model Euclid’s faint targets.

How the team built a universe, in plain language

From particles to a lightcone

The simulation begins as a gravity-only N-body run. The Flagship box spans 3600 h⁻¹ Mpc on a side and was evolved with 16,000³ particles, yielding a particle mass of approximately 10⁹ h⁻¹ solar masses. In other words, the run used four trillion particles, making it one of the largest gravity-only simulations ever performed. The output is written as a “lightcone” on the fly, meaning particle data are recorded at the moment a modeled light ray would reach an observer, which reproduces the line-of-sight view a telescope sees across cosmic time.

Finding haloes and placing galaxies

Dark matter particles cluster into haloes, the gravitational wells where galaxies form. The team used a halo finder to identify roughly 16 billion haloes across the simulated octant out to redshift z = 3. To convert haloes into observable galaxies the collaboration employed a hybrid of halo occupation distribution and abundance-matching techniques, with additional observational relations applied to assign luminosities, colours, emission-line fluxes, sizes, shapes, and spectral energy distributions. The result is not merely positions on the sky, but a rich catalogue: the scientists compute 399 properties for each galaxy to enable many different science tests.

Lensing and maps at survey resolution

Weak gravitational lensing is a cornerstone of Euclid science. To predict lensing observables, the simulation was projected into concentric spherical shells and converted to HEALPix maps at Nside = 8192, corresponding to a pixel size of about 0.43 arcminutes. Lensing fields such as convergence and shear were then sampled at galaxy positions so that the catalogue carries both intrinsic galaxy properties and their lensing distortions. This pipeline reproduces lensing statistics down to sub-arcminute scales within the limits imposed by the pixelization.

The headline results: what the Flagship mock delivers

• A full-sky lightcone produced from a four trillion particle N-body simulation.
• A halo catalogue of roughly 16 billion haloes in one octant out to z = 3.
• A multi-parameter galaxy catalogue from those haloes, with 399 computed properties per object.
• A magnitude-limited sample of 3.4 billion galaxies complete to HE < 26, which is deep enough to model Euclid selection and completeness effects.

These elements together make Flagship a testbed for the combined weak lensing and galaxy clustering analyses that will underpin Euclid’s cosmology results. The catalogue was explicitly validated against number counts, colours, stellar mass functions, emission-line fluxes, clustering statistics, and lensing statistics, and it reproduces the main characteristics of the samples Euclid will use for its cosmological analyses.

Why this matters for Euclid and for cosmology

The power of the Flagship mock is practical and conceptual. Practically, the catalogue allows pipeline developers and analysis teams to:

• Test photometric selection, source detection, and shape measurement algorithms under realistic sky conditions.
• Validate clustering measurements and redshift-space analyses using samples selected to mimic Euclid’s slitless spectroscopy and imaging limits.
• Quantify and mitigate biases in weak lensing shear calibration, magnification, and higher-order lensing statistics.

Conceptually, the mock gives a controlled laboratory to explore how astrophysical complications, selection functions, and survey geometry propagate into cosmological constraints. By reproducing both galaxy clustering and lensing in a self-consistent mock, the catalogue helps ensure that combined-probe analyses will be robust when applied to Euclid data.

What the validation revealed: strengths and caveats

The Flagship team ran an extensive validation suite comparing mock outputs to observational datasets and theoretical expectations. Key validation outcomes include:

• Optical number counts and colours: The simulation matches observed optical counts and reproduces major colour trends, indicating the SED assignment and luminosity modeling are broadly successful.
• Near-infrared excess: The catalogue overestimates H-band (HE) number counts by roughly 40% compared to COSMOS field measurements at the Euclid wide survey limit, suggesting the SED assignment favors redder templates in some regimes. This mismatch points to a specific, addressable calibration effect.
• Clustering and lensing: Two-point clustering, multipoles of the correlation function, shear two-point statistics, galaxy-galaxy lensing, magnification bias, and even higher-order lensing statistics are reproduced at levels consistent with theoretical expectations for Euclid analyses. This supports the mock’s intended use for cosmological pipeline validation.

At the same time the scientists are explicit about current limitations. The catalogue lacks active galactic nuclei (AGN), it uses relatively simple analytic prescriptions for galaxy positions and velocities that omit internal substructure, and the SED assignment depends heavily on optical colours which can misassign near-infrared flux. Some procedures used to extend satellites beyond virial radii generated spurious features in cluster outskirts, a problem the team plans to correct in future releases. These limitations are important for users to keep in mind, especially for cluster science and for studies that rely on accurate near-infrared photometry.

How the catalogue will be used and shared

The Flagship mock is already the baseline input for Euclid pipelines and was explicitly prepared to support the mission’s early data releases. The collaboration plans to release version 2.1.10 of the catalogue through the CosmoHub platform, providing the community access to the multi-terabyte dataset described by the paper. Because the catalogue computes many observables per galaxy, it is a flexible resource for a broad range of projects beyond Euclid’s primary cosmology program, from studies of cluster populations to tests of photometric redshift methods.

What comes next: improvements and future directions

The scientists outline a clear development path. Immediate priorities include improving SED assignment by expanding template libraries and exploiting more wavelength information to avoid near-infrared bias, including AGN populations, and refining the modelling of intra-halo galaxy substructure and satellite distributions to remove spurious outskirts features. The team also plans to extend calibrations to higher redshift using forthcoming observational datasets, and to consider finer lensing map resolutions to push accurate lensing predictions to smaller angular scales. These incremental upgrades will reduce model-systematic error and expand the science that can be reliably addressed with the mock.

A practical takeaway for researchers and the public

The Flagship catalogue is not an end in itself, but an enabling tool. It allows Euclid scientists to rehearse their analyses at survey scale, to understand selection effects and biases before they act on real data, and to develop techniques that extract robust cosmological constraints from a complex, noisy universe. For the broader public, the effort is a reminder that modern cosmology depends as much on huge simulations and careful data-quality work as on telescopes and satellites. The universe inside the Flagship mock will help make the real universe a little less mysterious.

Concluding thought

The Euclid Flagship galaxy mock stands as an ambitious example of how simulation, observation, and method development converge to advance cosmology. By converting four trillion particles into billions of catalogued galaxies with lensing and spectral properties, the collaboration has produced a laboratory where the interplay of astrophysics and instrument effects can be studied in detail, and where the hardest aspects of Euclid science can be stress tested before the mission’s cosmological verdict is delivered. The catalogue will evolve, but its present release already gives the community a realistic and powerful dataset to sharpen our picture of dark matter and dark energy.

Notes and data access: The Flagship mock and related material are described in the Euclid Collaboration paper, and the catalogue is distributed via the CosmoHub platform. For technical reference and reproducibility, consult the original publication.

The research was published in Astronomy & Astrophysics on April 30, 2025.