Feature #1952
ctool for 1d counts vs models profile
| Status: | New | Start date: | ||
|---|---|---|---|---|
| Priority: | Normal | Due date: | ||
| Assigned To: | - | % Done: | 0% | |
| Category: | - | |||
| Target version: | - | |||
| Duration: |
Description
So I’m trying to debug some of my models, and I want to compare some fitted models to the observed counts. This is somewhat like ctresmap, but only in 1 dimension, with separate lines for the number of observed counts and the number of model-predicted counts.
I’ve been trying to cobble something together with ctmodel, but its inflexible since it only works for profiles along the galactic b axis. I’ve attached a picture of what it looks like, though the y axis units are messed up.
So I’d imagine this new ctools could take the same inputs as csresmap, plus two GSkydirs (or a list of gskydirs) for the profile path. Then it could output a 4 column table to a fits file (ra,dec,counts,model).
Screenshot 2017-03-15 22.06.22.png - 1d counts vs models (78.5 KB)
ReconMethodDisp.Cut-NTel2-ExtendedSource-Hard.targ.crab.allenergies.science.pntsrcatom.galbprofile.profilecountmodel.png - counts vs model for the crab (82.1 KB)
Recurrence
No recurrence.
History
#1
Updated by Knödlseder Jürgen about 7 years ago
I’m wondering what the best approach for this is. Is this a sufficiently generic use case to have a dedicated ctool or cscript, or would it be better to create an example script that can be shipped together with ctools, and that can then be adapted by the user for a specific purpose.
#2
Updated by Kelley-Hoskins Nathan about 7 years ago
I’ve found these profiles are really helpful in debugging whats going wrong with likelihood models, since csresmap is somewhat difficult to interpret. An example script could probably work fine.