The SPARTA HDF5 output format

The main output format for SPARTA is a single HDF5 file, which is placed in the output directory. The top level of the HDF5 hierarchy always contains three groups, as well as one further group for each tracer type. The content of these structures is explained below.

The config group

The purpose of the config group is to provide a central storage location for all configuration parameter as well as derived outputs that are not datasets. All config parameters are stored as HDF5 attributes.

The main group /config contains all user-defined parameters, including their default values if they were not specified by the user. The sub-groups contain the config parameters of specific Results or Analyses. Those groups exist only if the respective parts of the code were turned on, otherwise the corresponding config parameters are ignored.

The simulation group

The simulation group contains information derived by SPARTA that is specific to the simulation the code was run on, including cosmology, snapshots, and the characteristics of the simulation box.

Field	Type	Dimensions	Exists if	Explanation
n_snaps	int	1	Always	The number of snapshots processed
n_files_per_snap	int	1	Always	The number of files into which snapshot data are split
n_particles	int	1	Always	The total number of particles in the simulation
box_size	float	1	Always	The box size in comoving Mpc/h
particle_mass	float	1	Always	The particle mass in Msun/h
Omega_m	float	1	Always	Omega_m,0
Omega_L	float	1	Always	Omega_Lambda,0
h	float	1	Always	The Hubble constant in units of 100 km/s/Mpc
snap_a	float	n_snaps	Always	The scale factor of each snap
snap_z	float	n_snaps	Always	The redshift of each snap
snap_t	float	n_snaps	Always	The time of each snap in Gyr
snap_tdyn	float	n_snaps	Always	The dynamical (crossing) time at the snapshot time in Gyr

The halos group

The halos group contains the following datasets for all halos that were output (generally because they reached a certain minimum number of particles at some point):

Field	Type	Dimensions	Exists if	Explanation
first_snap	int16	n_halos	Always	The snapshot index where the halo was created
last_snap	int16	n_halos	Always	The snapshot index where the halo ended
status	int8	n_halos * n_snaps	Always	The halo’s status at each snap (see below)
status_final	int8	n_halos	Always	The halo’s final status (see below)
id	int64	n_halos * n_snaps	Always	The halo ID at each snap
parent_id	int64	n_halos * n_snaps * 3	OUTPUT_HALO_PARENT_ID	The halo’s host ID at each snap, or -1 if it was a host itself
R200m	float	n_halos * n_snaps	Always	The halo radius R200m (in physical kpc/h) at each snap
position	float	n_halos * n_snaps * 3	OUTPUT_HALO_X	The position in the box (in comoving Mpc/h)
velocity	float	n_halos * n_snaps * 3	OUTPUT_HALO_V	The halo’s perculiar velocity in physical km/s

Please see the Halos & Subhalos page for the meaning of the status and status_final numerical codes.

The tracer, result, and analysis groups

Each tracer group contains one subgroup per result type. This subgroup will be present if the result type was activated in the compile-time settings, even if it was turned off for tracers of this type and thus contains zero results. Each result subgroup contains at least the following fields, plus other fields listed below:

Field	Type	Dimensions	Exists if	Explanation
tracer_id	int64	n_rs_*	In result groups	The unique ID of the tracer from which this result was derived
halo_id	int64	n_al_*	In analysis groups	The unique ID of the halo which this analysis refers to
halo_first	int64	n_halos	Always	The index of the first result or analysis of this type for this halo
halo_n	int32	n_halos	Always	The number of results or analyses of this type for this halo

The halo_first and halo_n fields, combined with the IDs, represent an indexing system that allows us to connect halos, results, and analyses to each other. In practice, this process can get complicated. For this reason, it is recommended to use the Python analysis package to load data from SPARTA files. This module can perform the most common types of matching automatically.

See the documentation of the individual Results and Analyses for details about the datasets that they write to the output file.