Orphan Particle Output

When running VELOCIraptor on the fly SWIFT can write additional output files which contain the positions, velocities and IDs of all particles which have ever been the most bound particle in a halo. The next section describes the motivation for this feature. Subsequent sections describe how to enable it and the output format.

Orphan galaxies in semi-analytic galaxy formation models

SWIFT and VELOCIraptor output may be used as the basis for semi-analytic galaxy formation models. A SWIFT simulation of a cosmological box is run with on-the-fly VELOCIraptor output, halo merger trees are constructed and a semi-analytic modelling code is used to populate the halos with galaxies. These models often assume that satellite galaxies in a halo continue to survive after their associated dark matter subhalo is no longer identified by the halo finder. In that case the galaxy (referred to as an ‘orphan’ because it has no parent subhalo) may be associated with the most bound particle of the subhalo at the last time it was identified.

Enabling orphan particle output

Orphan particle output can only be enabled when running VELOCIraptor on the fly. It is enabled by running configure with the flag --enable-velociraptor-orphans. E.g.

./configure --with-velociraptor=/path/to/VELOCIraptor-STF/src --enable-velociraptor-orphans

If SWIFT is configured with this flag then whenever VELOCIraptor is invoked during a simulation additional output files will be generated alongside the usual VELOCIraptor output.

Output format

The output files are generated in the same location and with the same base name as the VELOCIraptor output but with a .orphans suffix. If parallel HDF5 is available then there will be one output file per VELOCIraptor invocation. Otherwise there is one output file for each MPI task. Non-MPI runs always generate one file per VELOCIraptor invocation.

The output files are written as HDF5 and are in a similar format to the Snapshots, including the meta-data groups. Note that all particles are written to the PartType1 group irrespective of their type. This feature is primarily intended for use with dark matter only cosmological simulations.

The PartType1 group always contains the datasets Coordinates, ParticleIDs and Velocities in the same units as in the snapshot files and with the same meta-data attributes attached.