DM-54870: Fix the buffer bug#222
Conversation
arunkannawadi
force-pushed
the
tickets/DM-54870
branch
from
1204e20 to
d11178e
Compare
| raise InvalidQuantumError("MetadetectionShearTask requires a cell-based skymap.") | ||
|
|
||
| cell_config = skymap_config.tractBuilder.active | ||
| if (self.config.border is None and cell_config.cellBorder == 0) or ( |
There was a problem hiding this comment.
Is allowing self.config.border == 0 and cell_config.cellBorder == 0 intentional here?
This also makes me wonder if allowing self.config.border=None is useful here; doesn't seem to be materially different from zero (except in this check).
There was a problem hiding this comment.
Allowing self.config.border to be None clarifies that the user intended to rely on the cellBorder defined in the skymap to create the buffer region, and not the border value put in by hand. Allowing self.config.border == 1 and cell_config.cellBorder == 0 would be just as bad as self.config.border == 0 and cell_config.cellBorder == 0. The minimum allowed buffer region would depend on many factors (PSF size, artificial metacal shear applied, cell size etc.) and we are not mandating that specified border is larger than that minimum value.
There was a problem hiding this comment.
OK, I see your point now. The case I absolutely want to avoid is the one where two (different) positive values are set for cellBorder and self.config.border. As long as one of them is a positive integer and the other is not, that is a valid configuration to accept.
There was a problem hiding this comment.
border cannot be set to None.
with a quick, hardcoded fix that works for lsst_cells_v2 skymap.
so it can be continued to run on both lsst_cells_v1 and lsst_cells_v2 and any other skymap that we may construct in the future.
arunkannawadi
force-pushed
the
tickets/DM-54870
branch
from
08c3148 to
e1f38a3
Compare
2 participants
No description provided.