Feature #1037
Updated by Deil Christoph about 11 years ago
In HAP we usually combine all runs into total counts and exposure and exposure-averaged PSF before fitting the morphology and spectrum.
This looses some information, but means a factor 100 speed-up for a binned likelihood fit for 100 runs.
Plus there might be problems with having too many free background parameters when fitting per-run data with per-run IRFs.
So we should develop ctools to combine IRFs (and maybe counts).
As a start it could be a two-page Python script ... it's basically summing up.
In optical and X-ray astronomy this is known as co-adding, and there are e.g. tools to co-add PHA, ARF and RMF files in CIAO:
Example and explanation of tool: http://cxc.harvard.edu/ciao/threads/coadding/
Formulae: http://cxc.harvard.edu/ciao/download/doc/combine.pdf
This looses some information, but means a factor 100 speed-up for a binned likelihood fit for 100 runs.
Plus there might be problems with having too many free background parameters when fitting per-run data with per-run IRFs.
So we should develop ctools to combine IRFs (and maybe counts).
As a start it could be a two-page Python script ... it's basically summing up.
In optical and X-ray astronomy this is known as co-adding, and there are e.g. tools to co-add PHA, ARF and RMF files in CIAO:
Example and explanation of tool: http://cxc.harvard.edu/ciao/threads/coadding/
Formulae: http://cxc.harvard.edu/ciao/download/doc/combine.pdf