DM-43490: Put correct size to PixelGrid for PSF estimation #25
Conversation
|
Where do we stand on this PR? |
|
If we are willing to drop the problematic rotation angles from the tests, I can go ahead with this. The GalSim changes aren't in |
|
what is the timescale for the new galsim getting into the rubin env? |
|
rubin-env 9, see DM-43326 -- so you can add that ticket as a blocker to DM-43490. |
|
Already done linking the two tickets. I could drop the 45 degree angle (that one fails) and note a different ticket that gets trigggered after |
|
Currently, two tickets are blocked, waiting specifically on |
|
I don't want Erin to rely on his fork longer than necessary. It would have already diverged from the |
arunkannawadi
force-pushed
the
tickets/DM-43490
branch
from
fc897aa
to
4a723be
Compare
Setting it to -1 messes up the static type checker(s). Setting it to -1. leads to correct type interpretation.
Co-authored-by: Erin Sheldon <[email protected]>
Co-authored-by: Erin Sheldon <[email protected]>
Co-authored-by: Erin Sheldon <[email protected]>
with an option to rotate it by an arbitrary angle. Co-authored-by: Erin Sheldon <[email protected]>
Co-authored-by: Erin Sheldon <[email protected]>
Co-authored-by: Erin Sheldon <[email protected]>
arunkannawadi
force-pushed
the
tickets/DM-43490
branch
from
db75323
to
a076e63
Compare
|
The cutout size that are fed into Piff cannot be adjusted from this package. That happens upstream in |
|
|
||
| wcs = make_wcs(angle_degrees=45) | ||
| self.exposure.setWcs(wcs) | ||
| self.checkPiffDeterminer(useCoordinates='sky', kernelSize=35) |
There was a problem hiding this comment.
@timj Strictly speaking, this would get merged without ever being run on our CI systems. Should I include it, or add a note to add this test case in a later ticket?
| @@ -307,7 +358,7 @@ def checkPiffDeterminer(self, **kwargs): | |||
| chiLim = 7.0 | |||
| else: | |||
| numAvail = len(psfCandidateList) | |||
| chiLim = 6.1 | |||
| chiLim = 6.4 | |||
There was a problem hiding this comment.
@esheldon increased it to 6.5 and I found 6.4 was sufficient for tests to pass. Do you have an intuition as to why the increase was necessary with the rotations, Erin?
There was a problem hiding this comment.
I thought it was probably just noise.
There was a problem hiding this comment.
It could be. I don't think there was any first-principle motivation for chiLim = 6.1 in the first place, just a value chosen empirically.
Thresholds in pipelines_check repo have to be changed only after announcing/asking the entire Science Pipelines team, but these ones can be changed by a reasonable amount.
arunkannawadi
force-pushed
the
tickets/DM-43490
branch
from
a076e63
to
d0b86e8
Compare
No description provided.