ctools

Does ctobssim work on model cubes?

No. ctobssim always produces event lists from an observation definition (either specified by the parameters or an XML file).

I understand, so the idea is to simulate and bin the events afterwards.

Mayer Michael wrote:

I understand, so the idea is to simulate and bin the events afterwards.

No, I would by-pass the ctobssim step for binned analysis and replace it by a (unique) ctmodel step, followed by a simulation of the counts cube using the following code:

    # Allocate random number generator
    ran = gammalib.GRan()

    # Load model map
    map = gammalib.GSkymap(modmap)

    # Randomize
    for i in range(map.npix()):
        map[i] = ran.poisson(map[i])

Sounds good.

The parameter interface was considerably revised, in particular the following parameters were removed: ra, dec, rad, coordsys, proj, emin, emax, enumbins, tmin, tmax, npix and binsz.

The inobs parameter is now always queried.

The script now reads in an observation or an observation container and derives the TS distribution for a source based on that observation container. So far the script supports unbinned, binned and stacked observations. If an observation definition XML file is specified the script assumes that all observations are event lists. On/Off observations are not yet supported.