Add catalog line numbers of rendered objects to 'simcatobj'

[thusharaprr]

Users who supply custom catalogs to romanisim usually want to know which sources from their catalog were rendered in the simulation for the particular detector they chose. Currently file.tree['romanisim']['simcatobj'] has some info about the rendered objects such as their pixel x, y positions. But the catalog line number of each rendered object would be very useful to have. I believe that this can be done in romanisim.image.simulate_counts_generic() by making the following change to the code there.

Existing code:

objinfo = np.zeros(
    len(objlist),
    dtype=[('x', 'f4'), ('y', 'f4'), ('counts', 'f4'), ('time', 'f4')])
if len(objlist) > 0:
    objinfo['x'][keep] = xpos[keep]
    objinfo['y'][keep] = ypos[keep]
    objinfo['counts'][keep] = objinfokeep['counts']
    objinfo['time'][keep] = objinfokeep['time']

Suggested code:

objinfo = np.zeros(
    len(objlist),
    dtype=[('index', 'i4'), ('x', 'f4'), ('y', 'f4'), ('counts', 'f4'), ('time', 'f4')])
if len(objlist) > 0:
    objinfo['index'][keep] = np.where(keep)[0]
    objinfo['x'][keep] = xpos[keep]
    objinfo['y'][keep] = ypos[keep]
    objinfo['counts'][keep] = objinfokeep['counts']
    objinfo['time'][keep] = objinfokeep['time']

Just one line has been edited and one line has been added. Can you please make this change in an upcoming version of romanisim? Thanks!

[schlafly]

Thanks, I'll take a look. One detail here is that there's an alternative way of sending catalogs to romanisim where you have a full-sky catalog and romanisim concatenates together the bits of the full sky catalog that are relevant. In that case, the row numbers would correspond to the rows in the concatenated files.

There might be an alternative approach where if the catalog file contains a 'label' column we propagate it forward into objinfo. I can take a look at that.

But FWIW, we always accept pull requests, so if you actually want to make this change in a PR in github it is more likely to go in ~now. Thanks!

[thusharaprr]

Thank you! Yes, that is a better idea than returning catalog line numbers. The catalog column I'm interested in has the label 'id'. Perhaps the user can specify which column(s) to return, for instance with

ris.simulate_image_file(args, metadata, full_catalog, rng, persist, return_catalog_columns=['id'])

I don't have a specific code suggestion for implementing that. But this would be a very useful feature to have for me and others in the Roman SN PIT.

Thanks for encouraging pull requests. I do plan to submit one soon for the branch I've developed to include CRNL and custom ramps (currently for brighter-fatter inclusion). Working on crossing the Ts and dotting the Is before submitting that PR.