Fix Version/s: None
Team:Data Release Production
There has been much recent attention drawn to a certain mode of bad astrometric fits in SFM in the context of DC2/imsim data processing (see
DM-30466 for discussion and diagnosis). A potential "fix" was added on DM-30490 in the form of an optional step in SFM astrometric fitting to perform a rejection on delta mag (ref - src) outliers in the match catalog before sending it to the fitter. Fairly extensive testing has proven this to be a most desirable option in that it fixes (i.e. a good astrometric fit is now found) almost all of the bad fit cases (the pesky "hot mess" of DM-30466 is the only exception we have noted thus far) and it does-no-harm for the cases that were not affected and already had good fits (see DM-30812 for details of this validation). As such, this option should be turned on by default for LSSTCam-imSim processing. The config parameter to be added is:
Jenkins + ci_imsim is green.
Would you mind giving this a look? Can you think of any other check I should do before merging??
To test this, I ran one of our bad fit examples on master and having set up this ticket branch:
So all looks good!