The halo sampler correlates the halo properties (stellar mass and SFR) based on the sampling of its descendant, which is calculated backwards in time. Due to the presence of feedback, we cannot finalise the halo properties until we have the previous snapshot's (higher z) Lyman Werner and Ionising background.
As such, the current behaviour passes the correlated gaussian variables (which become the lognormal scatter in each property) backward in time, and then scales them to the mean (which includes the feedback) in the forward time-loop. This is efficient when running the lightcone, however it means that getting the star-formation rate or stellar mass in the halo catalogues is tedious.
To help with analysis, there should be an exposed function which loops through a catalogue and applies set_halo_properties() from HaloBox.c to a halo catalogue, given radiative backgrounds if required.
The halo sampler correlates the halo properties (stellar mass and SFR) based on the sampling of its descendant, which is calculated backwards in time. Due to the presence of feedback, we cannot finalise the halo properties until we have the previous snapshot's (higher z) Lyman Werner and Ionising background.
As such, the current behaviour passes the correlated gaussian variables (which become the lognormal scatter in each property) backward in time, and then scales them to the mean (which includes the feedback) in the forward time-loop. This is efficient when running the lightcone, however it means that getting the star-formation rate or stellar mass in the halo catalogues is tedious.
To help with analysis, there should be an exposed function which loops through a catalogue and applies
set_halo_properties()fromHaloBox.cto a halo catalogue, given radiative backgrounds if required.