Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_extensions_psfex
-
Labels:None
-
Story Points:1
-
Epic Link:
-
Sprint:DRP F16-2
-
Team:Data Release Production
Description
The current settings in psfex will only turn on oversampling only if the seeing is < 0.5", even if you have configured it do oversampling. This needs to be changed so that everything is determined by the config parameters.
We have also seen on HSC data that oversampling in general does not work well in psfex. We need to change the current configuration which does 2x oversampling to just use the native pixel scale.
Attachments
Issue Links
- mentioned in
-
Page Loading...
Activity
Assuming that the test fails in an expected way – ie, you and Bob Armstrong are happy that you can explain the failure and can modify the test case appropriately – I have no problem with this being merged. If you need further changes to the code, rather than just adjust the numbers in the test, by all means send it back to me for another look.
I found the cause of the test failure — oversampling hadn't been disabled in the test.
pprice@tiger-sumire:~/LSST/meas/extensions/psfex (tickets/DM-6982=) $ git --no-pager log -2 --reverse
|
commit a8046b3778507c70327a79b4fc2d48a30ad7cfc8
|
Author: Paul Price <price@astro.princeton.edu>
|
Date: Tue Jul 26 16:41:00 2016 -0400
|
|
|
disable oversampling in test
|
|
We believe that oversampling doesn't work well. Indeed, with
|
oversampling turned off, we get better fits for the test image.
|
|
|
commit 164e2651a573d7314ca865f41d46b8f68f7cbcfb
|
Author: Paul Price <price@astro.princeton.edu>
|
Date: Tue Jul 26 16:41:14 2016 -0400
|
|
|
remove unused and confusing line in test
|
|
self.FWHM is unused, and its presence is confusing because there
|
is also sigma1 and sigma2 which are used to set the PSF, and they
|
have no relation to self.FWHM.
|
Paul Price
added a comment - I found the cause of the test failure — oversampling hadn't been disabled in the test.
pprice@tiger-sumire:~/LSST/meas/extensions/psfex (tickets/DM-6982=) $ git --no-pager log -2 --reverse
commit a8046b3778507c70327a79b4fc2d48a30ad7cfc8
Author: Paul Price <price@astro.princeton.edu>
Date: Tue Jul 26 16:41:00 2016 -0400
disable oversampling in test
We believe that oversampling doesn't work well. Indeed, with
oversampling turned off, we get better fits for the test image.
commit 164e2651a573d7314ca865f41d46b8f68f7cbcfb
Author: Paul Price <price@astro.princeton.edu>
Date: Tue Jul 26 16:41:14 2016 -0400
remove unused and confusing line in test
self.FWHM is unused, and its presence is confusing because there
is also sigma1 and sigma2 which are used to set the PSF, and they
have no relation to self.FWHM.
Line 231 of testPsfexPsf.py is a comment which reads "Include oversampling in test". It immediately precedes the lines you've edited in a8046b3 to "disable oversampling in test". Please make these consistent!
I note that the test passes if you simply use the default kernelSize and samplingSize rather than overriding them in the test case. Is the override actually necessary? Can you just remove it (and the comment) for simplicity?
I'm such a doofus. I remember thinking that I could just remove those lines, but must have gotten side-tracked. They're gone now, the test passes, and I've merged to master.
Paul Price
added a comment - I'm such a doofus. I remember thinking that I could just remove those lines, but must have gotten side-tracked. They're gone now, the test passes, and I've merged to master. 
The unit test also fails on the HSC side (but unit test failures there aren't fatal to the build, so it seems no-one noticed).