Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: None
-
Labels:None
-
Story Points:1
-
Epic Link:
-
Sprint:DRP S19-3
-
Team:Data Release Production
Description
With w_2019_04 lsst_distrib and master obs_lsst at the time, "RangeError" is seems in makeCoaddTempExp in several tracts of imsim DESC Run 1.2i data. An example to reproduce is:
makeCoaddTempExp.py /datasets/DC2/repo/ --rerun w_2019_04/sfm:private/username/yours -c doApplyUberCal=False --id tract=4432 patch=4,5 filter=g --selectId visit=183892
|
which gives:
makeCoaddTempExp INFO: Processing calexp 1 of 1 for this Warp: id=DataId(initialdata={'visit': 183892, 'filter': 'g', 'raftName': 'R34', 'detectorName': 'S12', 'detector': 158, 'tract': 4432}, tag=set())
|
makeCoaddTempExp.warpAndPsfMatch.psfMatch INFO: compute Psf-matching kernel
|
makeCoaddTempExp.warpAndPsfMatch INFO: Cannot PSF-Match:
|
File "src/WarpedPsf.cc", line 72, in lsst::geom::Box2I lsst::meas::algorithms::{anonymous}::computeBBoxFromTransform(lsst::geom::Box2I, const lsst::geom::AffineTransform&)
|
Unexpectedly large transform passed to WarpedPsf {0}
|
lsst::pex::exceptions::RangeError: 'Unexpectedly large transform passed to WarpedPsf'
|
|
|
makeCoaddTempExp INFO: directWarp has 349480 good pixels (2.0%)
|
makeCoaddTempExp INFO: psfMatchedWarp has 0 good pixels (0.0%)
|
makeCoaddTempExp INFO: Persisting deepCoadd_directWarp
|
FWIW, this wasn't seen in last year's processing using w_2018_49.
Attachments
Issue Links
Activity
The w_2019_04 simply had a bad WCS. In fact, this WCS is HILARIOUS and I'm calling dibs on it for an avatar. If you warp the calexp to the SkyMap WCS (with no patch bbox), this is the result. Is that corner folding in on itself!?
w_2019_04/sfm:private/yusra/test_wcs/w_2019_06_warp:
Curiously, this problem has fixed itself with w_2019_06.
private/yusra/test_wcs/w_2019_06:private/yusra/test_wcs/w_2019_06_warp_new:
I vote we close this, noting that when we return to pair coding DM-16301, the calexp with the bad wcs would have never been written out.
One more piece of info: If you blink the good and bad calexps with locked wcs in ds9, it looks just like a translation with nothing crazy: ds9 -zscale /datasets/DC2/repo/rerun/w_2019_04/sfm/calexp/00183892-g/R34/calexp_00183892-g-R34-S12-det158.fits /datasets/DC2/repo/rerun/private/yusra/test_wcs/w_2019_06/calexp/00183892-g/R34/calexp_00183892-g-R34-S12-det158.fits
If you want evidence that the WarpedPsf is happy, it logs:
makeCoaddTempExp INFO: Processing calexp 1 of 1 for this Warp: id=DataId(initialdata={'visit': 183892, 'filter': 'g', 'raftName': 'R34', 'detectorName': 'S12', 'detector': 158, 'tract': 4432}, tag=set())
|
makeCoaddTempExp.warpAndPsfMatch.psfMatch INFO: compute Psf-matching kernel
|
makeCoaddTempExp.warpAndPsfMatch.psfMatch INFO: Adjusted dimensions of reference PSF model from (23, 23) to (53, 53)
|
makeCoaddTempExp.warpAndPsfMatch.psfMatch INFO: Psf-match science exposure to reference
|
makeCoaddTempExp.warpAndPsfMatch.psfMatch INFO: done
|
makeCoaddTempExp INFO: directWarp has 147435 good pixels (0.8%)
|
makeCoaddTempExp INFO: psfMatchedWarp has 135641 good pixels (0.8%)
|
makeCoaddTempExp INFO: Persisting deepCoadd_directWarp
|
makeCoaddTempExp INFO: Persisting deepCoadd_psfMatchedWarp
|
Marking Done because (1) it's fixed and (2) throwing a RangeError is the behavior we wanted (unlike DM-17731)
Yusra AlSayyad, Jim Bosch — again, I guess this is a DRP issue. Shout if you disagree.