Simulate event lists

Here we describe how to simulate event lists for a given observation list, source model and instrument.

Gammapy is mostly an analysis package for binned analysis, and so far we haven’t implemented general tools to sample arbitrary spatial and spectral density distributions.

An excellent tool ctobssim is available within the ctools package, so here we’ll describe how to use that first and then mention the existing functionality in Gammapy.

Using ctobssim

Using HESS IRFs and a real event list as input (for header info), it’s possible to simulate event lists according to any source and background model you like.

See https://github.com/gammapy/gammapy-extra/blob/master/datasets/hess_crab/make2.py as an example, which was used to generated the https://github.com/gammapy/gammapy-extra/blob/master/test_datasets/irf/hess/pa/hess_events_023523.fits.gz example file.

TODO: we should extend this to a Gammapy command line tool that:

  • can simulate simple IRF files for HESS or CTA (so that we don’t have to use real HESS IRFs)
  • can process an observation list, i.e. it’s not necessary to have real observations.

For now this one simulated HESS observation can be used for testing in Gammapy.

Using Gammapy

In order to test some of the tools from Gammapy, such as the background generation tools (Make background models), real data from existing experiments such as H.E.S.S. can be used. Since the data of current experiments is not public, rudimentary tools to prepare a dummy dataset have been placed in Gammapy:

  • make_test_dataset: function to produce a dummy observation list and its corresponding dataset consisting on event lists and effective area tables and store everything on disk.
  • make_test_eventlist: function called recursivelly by make_test_dataset to produce the data (event lists and effective area table) corresponding to one observation.

Currently only background events are simulated (no signal), following a very simple model, and only a few columns of the EventList class are filled. In addition, the effective area files produced, are empty except fot the low energy threshold header entry.

The tools are very easy to use. A H.E.S.S.-like test dataset can be produced with a few lines of Python code:

workdir = gammapy/scripts/tests/test_make_bg_cube_models.py
make_test_dataset(outdir=workdir,
                  observatory_name='HESS,
                  n_obs=2,
                  random_state=0)

Then the data can be read back using the DataStore class, and eg. print the observation table and the names of the files created with a few extra lines of Python code:

data_store = DataStore.from_dir(dir=workdir)
observation_table = data_store.make_observation_table()
print(observation_table)
event_list_files = data_store.make_table_of_files(observation_table,
                                                  filetypes=['events'])
aeff_table_files = data_store.make_table_of_files(observation_table,
                                                  filetypes=['effective area'])
for i_ev_file, i_aeff_file in zip(event_list_files['filename'],
                                  aeff_table_files['filename']):
    print(' ev infile: {}'.format(i_ev_file))
    print(' aeff infile: {}'.format(i_aeff_file))