Activate PSFMap for fitting #2627
Conversation
Codecov Report
@@ Coverage Diff @@
## master #2627 +/- ##
=========================================
+ Coverage 91.5% 91.5% +<.01%
=========================================
Files 141 141
Lines 15942 15944 +2
=========================================
+ Hits 14588 14590 +2
Misses 1354 1354
Continue to review full report at Codecov.
|
gammapy/cube/fit.py
Outdated
| self.psf = psf # .get_psf_kernel(self.model.position, geom=exposure.geom) | ||
| if isinstance(psf, PSFMap): | ||
| self.psf = psf.get_psf_kernel( | ||
| self.model.position, geom=exposure.geom, max_radius=0.8 * u.deg |
There was a problem hiding this comment.
I would propose to modify .get_psf_kernel() and just take the maximum of table_psf.rad as a default maximum radius (if max_radius is None). This is more consistent then having a separate hard-coded value here...
|
Thanks @adonath ! I have made the change |
|
Thanks @AtreyeeS! I've just restarted the remaining failing Travis builds, once those are green Ill merge... |
|
Ok, it seems |
Description
This pull request activates
PSFMapon the MapEvaluator. Similar to what is done in #2587 , the option to keep PSFKernel is still on. Once this PR is merged, I will deactivate the kernels and adapt the notebooks.Dear reviewer
One specific doubt I have is what should be the
max_radiusinself.psf = psf.get_psf_kernel(self.model.position, geom=exposure.geom, max_radius=0.8 * u.deg)Having it as a parameter will be extremely confusing for the users. Currently I am passing a large enough value which should hopefully suffice for most cases.
I did not adapt the tests in
analysisto avoid conflicts with the changes @Bultako might be making.