11:00 Eastern; 08:00 Project/Pacific Time. bluejeans.com/426716450
Attendees
Notes
- “Back in the zone”.
- Had an interesting debugging session with Ian Sullivan on astrometry issues.
- Matching the AuxTel exposures with Gaia is not super-straightforward.
- Some edge cases apparently succeeded, but in fact the majority are failing.
- Tried to cut down the reference catalog selection, but this is not straightforward.
- We can do magnitude cuts in the task config, but it's not clear if that requires the input image to be photometrically calibrated.
- Maybe able to use built-in filtering in the ReferenceObjectLoader to address this.
- Not clear that this will resolve all the problems, but it's a good way to start.
- Continuing with AuxTel data reduction this week.
- Discussing gain measurement on AuxTel with Nick Mondrik (Stubbs' grad student).
- Using PTC code on AuxTel. Finding interesting screwy effects.
- DM, Nick, eotest disagree on basic image statistics (mean, median, etc).
- Merlin Fisher-Levine will ticket this when he properly understands the problem.
- Crosstalk rework done; waiting for an RFC.
- Now back to DM-21212.
- Fighting bugs; hoping to get it back into re-review this week.
- Talking with Eric Charles from SLAC/Camera about making eotest work with the stack.
- See if eotest is useful for us, see if ip_isr is useful for them.
- Iterating with Eric to come up with an end goal.
- Worried about duplicating code.
- Merlin is worried that this is a black hole of finding bugs in eotest and the lack of known-truth.
- Agreed that Chris should chat to Eric about this, but avoid getting drawn into spending much time on this until a really concrete goal has been determined.
- Should file an RFC to add ci_cpp to lsst_distrib.
Andrés Alejandro Plazas Malagón
- Working on DM-24556 (defect normalization).
- Current implementation broke some tests in cp_pipe; Andrés Alejandro Plazas Malagón will work on fixing them up.
- Crosstalk in AuxTel (DM-24531).
- Can still see some residual crosstalk.
- SLAC matrix does better than we can do.
- Andrés Alejandro Plazas Malagón will follow up with them to understand where their data came from.
- Hoping to get better data rather than change our algorithm.
- Thinking about non-linear crosstalk for the future.
- Coming next:
- Fixing up normalized defects.
- DM-20070,