Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_modelfit
-
Labels:None
-
Story Points:4
-
Epic Link:
-
Sprint:Science Pipelines DM-W16-6, DRP W16-7
-
Team:Data Release Production
Description
Three significant changes were made to CModel in HSC-1339. They were described by Jim Bosch in a post to hsc_software. They include:
- Changing the method by which the initial approximation is determined;
- Changing the determination of the pixel region to use in fitting;
- A new prior on ellipticity and radius.
Please port these changes to LSST.
Also include the results of fixing the bug described in HSC-1384.
Attachments
Issue Links
Activity
Just a few minor comments on github. Otherwise, looks good.
Jim Bosch: I've created DM-5347 to capture the effort of adding appropriate tests for these improvements so that I can add it to our "clean up debt" epic for X16. I'm not sure if there's already a ticket or other specification for the work on the shapelet PSF approximation code that you refer to, though: can you clarify? Thanks!
Never mind – it's DM-5197, I assume. Sorry for the noise.
All review comments addressed; merged to master.
We should add something on these changes to the release notes. Some edited version of Jim's post to hsc_software would be good. I'll take a shot at that later unless somebody beats me to it.
Done.
Ready for review. This is essentially a 1-1 port of HSC-1339, which contained the improved fit-region determination and (buggy, but perhaps helpful) new prior. All changes on branch tickets/
DM-4768of meas_modelfit, and all commits are hand-modified cherry-picks (mostly namespace changes and afw.table syntax changes).I have intentionally deferred testing to another ticket, as I'll first need to deal with getting the shapelet PSF approximation code running robustly.