Data Access
Simulation outputs, file formats, and request guidance for Lumina data products.
The Lumina simulation data is currently not publicly available. To request access, please contact us through the Team page. We review all requests and will respond as soon as possible.
The Lumina simulation produces a variety of different data products that differ in cadence, resolution and storage size. Below you will find descriptions of the available data formats and their contents.
Predictions from the Lumina simulation for common physical quantities in tabular form. We state the relevant publication and figure if relevant.
Standard text files (.txt)
| Quantity | Data File | Description | Relevant Publication | Figure |
|---|---|---|---|---|
| Star-formation rate density | sfrd_lumina.txt | Time evolution of cosmic star formation rate density | Zier et al. 2026 | 9 |
| Reionization history HI | HI_reion_lumina.txt | Evolution of the volume-averaged fraction of HI | Zier et al. 2026 | 17 |
| Reionization history HeIII | HeIII_reion_lumina.txt | Evolution of the volume-averaged fraction of HeIII | Zier et al. 2026 | 22 |
| CMB optical depth | tau_cmb_vs_z_lumina.txt | Cumulative CMB Thomson-scattering optical depth as a function of redshift | Zier et al. 2026 | 26 |
Full simulation snapshots contain all particle data at specific redshifts. Each snapshot includes gas cells, dark matter particles, star particles, and black holes along with their complete physical properties. Because full snapshots are very large (15-20 TB), only a limited number of them can be stored. The snapshots are post-processed such that each particle property is written to its own HDF5 file. In addition, we provide a virtual HDF5 file that links to all these individual files and exposes a standard AREPO snapshot interface. This significantly reduces file operations when loading only a small subset of particle fields. To further improve efficiency, large datasets are stored using HDF5 chunking.
Most particle properties follow the same conventions as in the IllustrisTNG simulations, with the exception of particle positions that are stored as 32-bit unsigned integers to guarantee uniform spatial resolution across the entire simulation volume. We note that the mass of dark matter particles is not constant but slightly varies due to the introduction of our new initial condition generator.
All 32 virtual snapshots listed below are present on Engaging. Twenty-eight contain all four particle types; outputs 117–120 contain only stars and black holes there.
Particle Fields
These files contain Friends-of-friends (FoF) groups and a classic SUBFIND re-run of the subhalo catalogue. One catalogue is available for each of the 32 virtual snapshots. Every group and subhalo property is stored in an individual file; a virtual HDF5 file links them through the standard AREPO catalogue interface. These catalogues are distinct from the on-the-fly SUBFIND-HBT output.
HDF5 files with group and subhalo tables
Catalogue Fields
Simulation data binned on uniform Cartesian grids using an adaptive Cloud-in-Cell algorithm. We present data on a normal uniform grid as well as on a velocity-corrected spectral grid oriented along the positive z-axis. This data is especially useful to study the intergalactic medium.
The field sets differ at z = 4.75. The earlier, six-bin radiation run includes stellar-density and stellar-ionizing-luminosity fields that are absent below the transition; the later run stores a single radiation-energy bin.
HDF5 files with 3D grid data
Grid Fields
Black-hole particle data are stored in 910 outputs from z ≈ 14.3 to z ≈ 3. Each output contains the full set of properties for all black holes in a root-level BH group, enabling detailed studies of growth, accretion variability, and AGN feedback between the full snapshots.
HDF5 files with a single BH group
Black Hole Fields (BH)
Light cone outputs that capture the simulation as it would be observed, accounting for the finite speed of light. These are essential for creating mock observations and comparing with surveys. The lightcone data consists of cell-averaged quantities on the past light cone and velocity-corrected spectral fields.
Above z = 4.75, the lightcones include stellar-density, six-bin radiation-energy, and stellar-ionizing-luminosity fields. Below the transition, these are absent and radiation energy is stored in the single-bin IonEnergy field.
HDF5 files with gridded lightcone fields
Grid Cell Fields
Spectral Fields
Line-of-sight projected quantities for spectral analysis.
Additional derived data products including merger trees, mock observations, and post-processed quantities.
Available products vary by request. Please contact us for specific needs.
Dark Matter Only Simulations to z=0
We performed two additional dark matter only simulations with 15003 and 30003 particles that share the full-physics run's large-scale phases. They were evolved to redshift z=0 and can be used to analyze the long-term dynamic evolution of interesting objects in Lumina or to create initial conditions for zoom-in simulations. Both simulations use the GADGET-4 code and share the same output fields. Due to different time steps their output times slightly differ from the full physics runs.
Full dumps of properties of all dark matter particles. The mass of all particles is constant and can be found in the header (MassTable).
HDF5 files split across multiple files per snapshot
Particle Fields
We provide regular group catalogues calculated with the Friends-of-friends (FoF) and SUBFIND-HBT algorithms.
HDF5 files with group and subhalo tables
Catalogue Fields
We calculated several additional data products that include:
- Merger trees tracking halo evolution from high-z to present day
- Halo mass functions at various group redshifts
- Matter power spectra at each output time
We are happy to calculate additional quantities upon request.