CompareWarp Evaluation 
Current status and latest problems with CompareWarp as of DM-12445. 



Instructions to use CompareWarpAssembleTask:

To run assembleCoadd.py use the --warpCompareCoadd argument like:

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!UPDATED 12/11/2017!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

export wideVisitsR=11442^11446^11450^11470^11476^11478^11506^11508^11532^11534

echo 'config.assembleStaticSkyModel.subregionSize = (10000, 200)
config.assembleStaticSkyModel.doApplyUberCal = True' > config_assemble.py

assembleCoadd.py --warpCompareCoadd /datasets/hsc/repo --rerun private/yusra/psfMatching/DM-12445:private/yusra/psfMatching/DM-12445-clipped --id tract=8766 filter=HSC-R  --selectId ccd=0..8^10..103 visit=$wideVisitsR --configfile config_assemble.py

To run coaddDriver.py create a config file and apply with the --configfile flag.
export wideVisitsY=6478^6482^6486^6496^6498^6522^6524^6528^6532^6544^6546^6568^13152^13154

echo 'from lsst.pipe.tasks.assembleCoadd import CompareWarpAssembleCoaddTask
config.assembleCoadd.retarget(CompareWarpAssembleCoaddTask)
config.assembleCoadd.assembleStaticSkyModel.subregionSize = (10000, 200)
config.assembleCoadd.assembleStaticSkyModel.doApplyUberCal = True' > config_coaddDriver.py

coaddDriver.py  /datasets/hsc/repo --rerun RC/w_2017_44/DM-12388:private/yusra/psfMatching/DM-12445  --batch-type=slurm --mpiexec='-bind-to socket' --job Y_psfMatch-DM-12445 --time 600 --cores 23  --id tract=8766 filter=HSC-Y  --selectId ccd=0..8^10..103 visit=$wideVisitsY --configfile config_coaddDriver.py

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^UPDATED 12/11/2017^^^^^^^^^^^^^^^^^^^^^^^^^

(1) Photometry via pipe analysis:


Run on two tracts:  WIDE: 8766 and DEEP: 9813
for the following 2 repos:
/datasets/hsc/repo/rerun/private/yusra/psfMatching/DM-12445
/datasets/hsc/repo/rerun/RC/w_2017_44/DM-12388
 
coaddAnalysis.py:
  • DM-12445 (compareWarp as of DM-12445):  
  • w_2017_44 (safeClip as of weekly 44):  

compareCoaddAnalysis.py
  • Comparison of DM-12445 and w_2017_44:


With Lauren looking over my shoulder, we flipped through some of her favorite plot type (the CModel difference between runs: e.g. compare-t9813-HSC-R-diff_modelfit_CModel-psfMagHist.png).  We saw:
  • The number of “splits”  Star/Galaxy designation from one run to another is scary.  Filed ticket: DM-12723