most SFR are zero in TNG100-1

Lorenzo Zanisi
  • 10 Jan '19

Hi everyone,

I am pulling a few quantities from TNG100-1 successfully. However, it seems that most of the SFR are zero. That is a problem common to both centrals and satellites. Given that there are some papers that discuss TNG outputs for quenched and star forming galaxies (e.g. Genel+18), I did not expect that. Is that normal or am I doing anything wrong? Below is the code snippet I use to pull the data from the database.

Thank you very much for your help!

Kind regards,

Lorenzo Zanisi

basePath = "/home/tnguser/sims.TNG/TNG100-1/output"

for i in [99, 67, 50, 40, 33, 29, 25]:

Data = ill.groupcat.load(basePath, i)

Sub_DF_Header = []

SubDM, SatSM, SubGrpID = Data['subhalos']['SubhaloMassType'][:,1], Data['subhalos']['SubhaloMassType'][:,4], Data['subhalos']['SubhaloGrNr']

SFR, RE, SubOrder = Data['subhalos']['SubhaloSFRinRad'], Data['subhalos']['SubhaloMassInHalfRadType'][:,4], Data['subhalos']['SubhaloParent']

SubFlag = Data['subhalos']['SubhaloFlag']

Sub_DF_Header += ["SubDM", "SatSM", "SubGrpID", "SubFlg"]

mask = SubFlag>0

SubDM, SatSM, SubGrpID, SFR, RE, SubOrder = SubDM[mask], SatSM[mask], SubGrpID[mask], SFR[mask], RE[mask], SubOrder[mask]

GrpDM = Data['halos']['Group_M_Crit200']

GrpDM[GrpDM == 0] = 10**-30

GrpDM = np.log10(GrpDM*(10**10))

ParentDM = np.array([GrpDM[i] for i in SubGrpID])

Sub_DF_Header += ["ParentDM"]

df = pd.DataFrame({

    'GrpID': SubGrpID,

    'SubOrder': SubOrder,






    'z':np.full_like(SubOrder, Data['header']['Redshift'])


df.to_csv("./LorenzoCats/z_{}.csv".format(round(Data['header']['Redshift'], 4)))
Dylan Nelson
  • 10 Jan '19

Hi Lorenzo,

For example for TNG100-1 at z=0 there are 3423724 central subhalos, and 3377737 of them have SubhaloSFRinRad == 0, this is correct.

Note that, by number, most halos/subhalos are very small objects near the resolution limit, and not of interest. The very small subhalos will also, naturally, be 'dark' (non-baryonic).

If you restrict e.g. to M>1e9 (typical resolution cut) or M>1e8 (aggressive resolution cut), you should find the majority have SFR>0.

For instance, TNG100-1 at z=0 there are 20144 subhalos with M*>1e9 Msun (using SubhaloMassInRadType), and 5173 of them have SFR == 0.


Lorenzo Zanisi
  • 1
  • 10 Jan '19

Hi Dylan

thanks for your swift reply! Yes I can see where the problem was now. However I'm still wondering why 1/4 of galaxies with M*>1e9 Msun should have SFR=0 in the first place. Is zero a particular value that is assigned in the simulation when something specific happens (i.e. is put in place of a NaN)? Thanks!

Dylan Nelson
  • 10 Jan '19

Hi Lorenzo,

There are many reasons why a galaxy could have zero SFR. For example, it might have absolutely no gas, having been entirely stripped (as a satellite).

I would suggest also you read the recent Donnari et al. paper, particularly Section 2 and Figure 8 (and related discussion), which talks a bit about zero SFRs and many other important issues related to simulated SFRs -- for TNG, all the same details applying also to Illustris.

  • Page 1 of 1