Public Data Access Overview / Discussion Forum

Hi TNG team,

I have been meaning to create a relatively simple plot to check the SFR for field galaxies and i'm getting some confusing results.

First is how you would define field galaxies essentially in TNG-50. I've been reading the forums, and since im only doing a theoretical test, im not going in too deep with the definition.

What i'm doing firstly, is using the fields 'Group_M_Crit200' and 'GroupNsubs', to filter halos with masses between 1e7 and 1e13 and taking GroupNsubs < 5 .
I get about 4 million + halos this way.

Now i load the subhalo properties , SubhaloMassInRadType','SubhaloSFR'and 'SubhaloFlag.

I filter with these 3 fields keeping SubhaloFlag ==True (Which filters out like 3 galaxies out of the 4 million, so it doesn't make that much of a difference)

I take stel_mass = 1.0e10*sub['SubhaloMassInRadType'][subs,4]/hh
where sub represents all the subhalos loading the subhalo fields above and subs represent my 4 million + subhalos filtered from the above condition.

If i filter that to stel_mass > 1e7 ( so it has at least a 100 elements), i get about 900 subhalos and if i filter for sfr > 0.0 , i get about 270 subhalos.

It seems too small a sample size, so maybe i should add more galaxies as what to consider 'Field' because this might be more 'isolated' than field.

Another thing to note here is if i plot the SFR vs stellar mass for these, they are all underestimated. (Im not cutting them off at SFR < 10^-3 for now, as it doesn't make too much of a difference) .

I also made the plot for this using all subhalos above 1e7,
So basically changing in the above code, stel_mass = 1.0e10*sub['SubhaloMassInRadType'][:,4]/hh

I get around 10,800 subhalos and it fits the relation much better :

My question is am i doing things right here ? Is my definition of 'Field' too vague ? and if its still not too bad of a definition, is it really such a small sample size of galaxies with masses above 1e7 ?

You question is mainly about "how to define field galaxies", and your approach is to cut on GroupNsubs < 5?

First, if you are comparing to some observation, you should mimic whatever definition of "field" they use.

Alternatively, if you want a pure theory definition, you can apply one. My suggestion would be: (i) central galaxies only, (ii) no galaxy more massive than X, or more massive than Y times the current galaxy mass, within Z Mpc. The X, Y, and Z here are up to you.

The GroupNsubs < 5 criterion is not really achieving what I would consider a field definition.

Finally, you didn't mention exactly how, but you are selecting on both halo and subhalo properties, so it is important to not mix up the indices, and to make sure you are considering only the central subhalo of each halo.

Hi Dylan, at the moment, im not testing any particular observations, so i think using a pure theory definition should be good. So i will test that out.
Considering centrals only, that should be the field 'GroupFirstSub' right ?

And yes, i should be more clearer but im taking all the subhalos i found in the halos with GroupNsubs <5, and then further using the subhalo properties for those.

Thanks for your help. I will apply this criterion and read a bit more in the literature of what i should consider for X , Y and Z as you mentioned. I might post again if i run into a problem, as my main concern is that the sample size seems too small and maybe im only picking up passive galaxies with the way i was defining them.

Just for double-checking, for TNG50-1 at z=0,

In [6]: np.where( (cen_flag==1) & (mstar > 7.0) )[0].size
Out[6]: 8836

In [7]: np.where( (cen_flag==1) & (mstar > 7.0) & (sfr > 0.0) )[0].size
Out[7]: 7884

In [8]: np.where( (cen_flag==1) & (mstar > 8.0) & (sfr > 0.0) )[0].size
Out[8]: 4071

If you want to trust the properties of the galaxies, I would take a lower mass threshold of more like M=1e8 than M=1e7 (for TNG50-1).

Hi Dylan, to update, i did find a good sample size of about 2615 galaxies, considering a) central b) no galaxy more massive than 1.2 times the galaxy within 1 Mpc, c) sfr > 0.0
and it much better suits as a test for this simple plot i wanted to do.
I can change the selection criterion to better fit me at a future point, but its not necessary for me at the moment.

It really helped alot , thankyou .
Arjun