Analysis scripts and Examples

On this page, we go through some examples of how to load and analyse Auriga data using some lightweight Python code, which we have made available at this bitbucket repository. These scripts are intended to be simple and flexible in order to provide a starting point that the user can build on and adapt to their specific goals.

We provide a few examples below to illustrate how these scripts can be used to analyse the data, including:

  • Read snapshot data for a single particle type;
  • Read the subhalo catalogues and fields;
  • Center on the main galaxy and select particles;
  • Converting stellar particle formation times to lookback times;
  • Read and navigate the merger tree data to retrieve evolutionary histories of subhalos;
  • Use the supplementary accreted star particle data to select particles from specific progenitors;
  • Combine this additional information with the snapshot and merger tree data.
All examples below assume that the user has downloaded and installed public python scripts according to the instructions given on the abovementioned repository.

Snapshots & (Sub)halo catalogues

Plotting the star formation history of the main galaxy

After downloading snapshot data from the download page, start up the python interface or create a python script and import the scripts:

import auriga_public as ap

and then define the path to the output directory storing the data:

path='/path/to/output/directory/'

For plotting the star formation history, we need only load PartType4 of the final redshift zero snapshot, which in this case is snapshot number 127. We will also load only the following fields:

loadlist=['Coordinates', 'GFM_StellarFormationTime', 'GFM_InitialMass']

which we need in order to make spatial cuts on stars (to select those confined to the galaxy), and to get their ages and initial masses.

We then create an snapshot object via

snapobj = ap.snapshot.load_snapshot(127, 4, loadlist=loadlist, snappath=path)

This object contains the particle information for all star particles and wind particles thoughout the entire simulation volume, so we need to first select the particles in the main galaxy, inside say, 10 percent of the critical radius for closure. We therefore first need to retrieve the coordinates and radius of the main halo from the subhalo catalogue by creating a subhalo object:

loadlist=['SubhaloPos', 'Group_R_Crit200'] subobj = subhalos.subfind(127, directory=path, loadlist=loadlist)

We can then centre the coordinates on the main halo

snapobj = ap.util.CentreOnHalo(snapobj, subobj.data['SubhaloPos'][0])

and then select only star particles within a sphere of our chosen radius

snapobj = util.apply_mask(snapobj, stars=True, radialcut=0.1*subobj.data['Group_R_Crit200'][0])

Finally, since the 'GFM_StellarFormationTime' is in units of scale factor, we can convert it to lookback time under the assumption of a flat Universe with the following helper function:

ages = ap.util.GetLookbackTimeFromScaleFactor_Flat(snapobj.data['GFM_StellarFormationTime'], snapobj.hubbleparam, snapobj.omega0, snapobj.omegalambda)

Now we can apply the appropriate unit conversions and make a histogram of the star formation rate

plt.hist(ages, bins=40, range=[0, 14], weights=snapobj.data['GFM_InitialMass']/(14/40)*1e10/1e9)

Merger Trees

Plotting the mass evolution of a subhalo

First, load the tree for the simulation:

treeobj = ap.mergertree.load_mergertree(0, 0, 127, directory=mergertreepath)

Let us select subhalo number 1 (usually the most massive satellite) of FOF group 0 at the final snapshot, and request only the main progenitor branch of this halo which returns the 'FirstProgenitor' indices of that object (i.e. the main progenitor of the object at each prior snapshot) and not all the 'NextProgenitor' objects:

result = treeobj.GetObjectHistoryFromSubhaloID(1, 127, 0, ['SubhaloMassType', 'Redshift'], mainprogonly=True)

Now plot the result:

plt.plot(result['Redshift'], result['SubhaloMassType'][:,:,1]*1e10)

Accreted star particles

In this example, we demonstrate how to select star particles that were born in ex-situ in a particular subhalo, and but now belong to the main halo. This could either mean these star particles were born in a fully disrupted system, or that they have been stripped from an existing satellite galaxy. First load the accreted star particle data using the helper function

mdata = ap.util.read_starparticle_mergertree_data_hdf5(127, '/accretedstardatapath/', 'halo_6')

This gives us a data object with attributes described in the data specifications page. We can get a list of all the progenitor galaxies, at the time each reached their peak stellar mass, via

first_prog = np.array(sorted(set(list(mdata['Exsitu']['PeakMassIndex']))))

$ print('first_prog') 
first_prog = [    -1    127    132    140    145    273    345    860   1233   1673
    1930   2382  10218  10829  11280  11568  11695  11923  12907  12926
   14106  14334  17599  39586  39588  41692  41911  43259  44761  45273
   45512  45672  45898  47735  52804  64645  70480  70646  72024  72297
   72533  74509  87696  93131  99281 102082 102097 102560 102895 102963
  102988 103658 103755 103926 103966 104550 105715 105957 106642 106683
  107140 108272 108455 108581 108688 108787 109362 109838 110044 110929
  111064 111364 111564 117933 117958 118023 118058 118065 118168 118213
  118270 118287 118451 118519 118656 118714 118775 118867 119178 119443
  120081 120709 121508 121659 121809 122590 124869 125639 126335 128295
  136278 148722 163753 166258 172398 209355 241501]

We can count the number of star particles that are now bound to the main halo from each progenitor system, and sort in descending order

nstars_in_subhalo = np.zeros(len(first_prog)) for i, pid in zip(range(len(first_prog)),first_prog): nstars_in_subhalo[i] = sum( (mdata['Exsitu']['PeakMassIndex']==pid) & (mdata['Exsitu']['AccretedFlag']==0)) nsort = np.argsort(nstars_in_subhalo)[::-1] nstars_in_subhalo = nstars_in_subhalo[nsort].astype('int') first_prog = first_prog[nsort]
$ print('first_prog')
first_prog= [ 14334  11695   2382  39586  11923  41692  12926  99281    145  39588
    1673  45273     -1 102097    860    345  14106    140  45512 108272
   45672 103926  12907 108688 102895 106642    273 108455 102988 106683
   10829  64645  45898 102082 118065 118058 102560  41911  93131  43259
     132 119443 108787   1930 109362 103966 118023 107140 102963 121659
  109838 110929  17599    127  87696 118168 111064 118775 118519  70480
  241501 121809 209355  52804 119178 118714 103755  74509 120709  10218
  105715 108581   1233 128295 117933 126335 117958 103658 118213 124869
  118270 166258 172398 118287 118451 148722 111564 122590 120081  72533
   72297  72024  70646 104550 105957 118867 118656  44761 110044 121508
  111364  47735 163753  11280  11568 125639 136278]

$ print('nstars_in_subhalo')
nstars_in_subhalo= [49776 21182 21022 20657 20611 13276 12464  7198  6319  2553  1282  1264
        558   378   315   288   262   261   257   243   176   170   141    95
         94    79    79    78    48    38    35    35    34    31    24    23
         22    20    19    19    16    13    13    13    12    12    10     9
          8     8     7     7     7     7     6     6     5     4     4     4
          4     3     3     3     3     3     2     2     2     2     2     2
          2     1     1     1     1     1     1     1     1     1     1     1
          1     1     1     1     1     1     1     1     1     1     1     1
          1     1     1     1     1     0     0     0     0     0     0]

Now, we can select the particle IDs that originated from the i-th most massive contributor to the accreted stellar population. For example, to select those from the 3rd most massive progenitor, we get the indices for which 'PeakMassIndex'=2382 and retrieve the corresponding array of 21022 'ParticleIDs'.

Combining this with other information in the accreted particle list data, such as the 'BoundFirstTime', we may make plots such as the one below which shows the redshift 0 XYZ distribution of these star particles coloured according to the time at which they first became unbound from their progenitor and therefore became bound to the main halo.

Here, we see that this particular satellite, whose peak logarithmic stellar mass is 8.9, has accreted over several pericentric passages, leaving shells of different extents that each comprise stars with different accretion times as defined as when they become bound to the main halo.

We can also retrieve the evolution of the progenitor above from the merger tree, via

result = treeobj.GetObjectHistoryFromMergerTreeObjectID(2382)

To see the trajectories of all the progenitor that contributed to the 5 most massive mergers in this simulation, we can find all the 'RootIndex' values associated with the 5 most massive 'PeakMassIndex' values, and obtain from the merger tree the histories of each subhalo that ever contributed to these 5 most massive Milky Way mergers. This would produce plots like the following

Trajectories of progenitors in the XY, XZ, YZ planes. Here, each of the 5 main mergers is made up of many smaller branches of the same colour, reflecting the earlier, smaller merger events that grow the mass of each massive progenitor until they attain their peak mass and merge with the Milky Way.


These are just a few very simple manipulations of the Auriga data products; the more detailed and interesting ones will be written by you!