The recently created tickets DM-8532 &
DM-8533 capture the most pressing issues, as revealed by this ticket and DM-6817, to get to the bottom of. From discussions in the regular Monday morning meeting here at Princeton, I will be focussing first on a look at the brightest stars being used in the psf modeling looking in particular at their footprints and distribution of peaks (as a function of distance from the centroid).
For direct comparison of the plot that revealed the main issues (see comments in
DM-6817), I have also attached the HSC version. So, compare:
The "bump" at bright magnitudes is not present in the latter and the large density of objects that have likely been misclassified as galaxies (namely the strip of red objects just below the threshold for the star/galaxy classification) in the LSST stack plot has much lower density in the HSC plot.