There are a couple of systems with significant variance in CROWDSAP (& therefore L3) values. These are often reflected in differing eclipse depths.
Example systems:
TIC 167795859 sectors S61 & 62 are the clearest example
TIC 31273263 sectors 95 & 96
In both these cases, the current algo in pipeline's choose_lightcurve_groups_for_fitting() func will stitch the two sectors. With the differing eclipse depths we get a poor fit.
I've previous implemented a do_not_stitch flags which works for 167795859 as there are lots of sectors and61&62 are the only sectors that the algo suggests for stitching (effectively, S61 gets dropped & every other sector fits individually).
However 31273263 has a longer period and stitching is required to get sufficient coverage, so a blanket flag is not sufficiently fine grained.
There are a couple of systems with significant variance in CROWDSAP (& therefore L3) values. These are often reflected in differing eclipse depths.
Example systems:
TIC 167795859 sectors S61 & 62 are the clearest example
TIC 31273263 sectors 95 & 96
In both these cases, the current algo in pipeline's choose_lightcurve_groups_for_fitting() func will stitch the two sectors. With the differing eclipse depths we get a poor fit.
I've previous implemented a do_not_stitch flags which works for 167795859 as there are lots of sectors and61&62 are the only sectors that the algo suggests for stitching (effectively, S61 gets dropped & every other sector fits individually).
However 31273263 has a longer period and stitching is required to get sufficient coverage, so a blanket flag is not sufficiently fine grained.