Hi again, Dylan,
I am trying to select field galaxies. This has been partially covered in the forum thread http://www.tng-project.org/data/forum/topic/6/two-basic-questions-for-illustris-simulation-data/
I am taking the following approach (is more or less one of the possibilities you describe in the thread): satellites further away (from the coordinates provided by GroupPos) than 4.5 times R200 of the halo (value found in the literature) are classified as field galaxies.
Now my question is focused now on halos hosting only one subhalo: could those be directly considered as field galaxies (of course, under the circumstances that the subhalo is indeed a galaxy)?
Many thanks in advance!
I think you mean "field" and "isolated" in the same sense, i.e. you would like to pick galaxies which are all alone?
In this case, I would make two requirements:
(1) Should be a central (satellites by definition are not alone).
(2) Should be at least a distance X away from any other central more massive than Y.
If you are comparing to particular observations, it may be best to try to match X and Y as closely as possible to what was done in the study.
If you are undertaking a theoretical exploration, then any reasonable values for your particular case would be ok. For instance, it is common to set Y as the mass of the candidate field galaxy itself, and X some multiple of its virial radius, some value between 2 and 20 probably being reasonable.
I was trying to compare the simulation results to particular observations. In those, groups were selected with a FoF algorithm. I am considering field galaxies as those that weren't grouped together at all.
I think I can take the same approach for the Illustris-TNG results since the groups catalogue was also constructed with a FoF algorithm.
The observational FoF algorithm would have been run on galaxies in (ra,dec,redshift) space.
One could run the same algorithm on TNG galaxies, after placing them in (ra,dec,redshift) space, this would be quite interesting, and a very good way to identify field galaxies in the same way.
Note that we use a FoF algorithm to group nearby particles into galaxies, which is somewhat different.
Regarding the last comment, I see what you mean: "The FoF algorithm is run on the dark matter particles, and the other types (gas, stars, BHs) are attached to the same groups as their nearest DM particle. The Subhalo fields are derived with the Subfind algorithm" (from The IllustrisTNG Simulations: Public Data Release ). So now I see why you say it is not a one-to-one comparison between the TNG FoF and the observational FoF.
For the previous comments: my first thought, when considering the observational FoF algorithm in the (ra,dec,redshift) space is that this is just spherical coordinates. But when you suggest of transforming the TNG galaxies coordinates to (ra,dec,redshift) space, I guess you mean more than a transformation from the cartesian coordinates in TNG to spherical: there must be something else, taking into account different snapshots (?). I don't fully grasp it (anyway, I am following another approach to solve my problem, but maybe you can provide a brief explanation (or what you meant) for the conversion of TNG galaxies to (ra,dec,redshift) space, if it is not beyond the scope of this forum).
Thank you for your insights!
The approach is called "generating a lightcone", which is exactly as you say, transforming from Cartesian box to spherical (sky) coordinates, as would be seen observationally, where the redshift is time. There are many tools and codes for this if you google "lightcone", and for instance some details and descriptions here.