When running on the fly from Swift and writing Velociraptor group IDs to the snapshot, the resulting dataset is scrambled. This happens because the particles are sorted by binding energy but the corresponding pfof array which stores the group membership is not reordered.
With this change the VELOCIraptorGroupIDs array in Swift outputs is correct when running on one MPI rank. On multiple ranks VELOCIraptorGroupIDs does not agree with the velociraptor output files so there's at least one more bug left.
When running on the fly from Swift and writing Velociraptor group IDs to the snapshot, the resulting dataset is scrambled. This happens because the particles are sorted by binding energy but the corresponding pfof array which stores the group membership is not reordered.
With this change the VELOCIraptorGroupIDs array in Swift outputs is correct when running on one MPI rank. On multiple ranks VELOCIraptorGroupIDs does not agree with the velociraptor output files so there's at least one more bug left.