Status: To Do
Fix Version/s: None
Chris Waters 12:03
The only thing that should be missing from crosstalk as of today is a functional form for measuring non-linear CT. The data is available (in gen3 processing) to do flux dependent CT, but there's no implementation defined.
For measuring the coefficients, or for applying them?
Chris Waters 12:04
Both. I didn't want to just add another arbitrary polynomial without looking more at the nonlinear crosstalk claims/data.
Sounds like the right call. Assuming linearity, have we processed the spot data and derived x-talk coeffs to compare to Adam's?
Chris Waters 12:13
I don't believe so. The updated crosstalk code has only been available for the past month. If the spot data has been ingested, it should be possible to compare
I think that should be @Andrés Plazas’s first experiment when we get back to looking at crosstalk. It should all be ingested, in gen2
|Field||Original Value||New Value|
|Watchers||Andrés Alejandro Plazas Malagón [ Andrés Alejandro Plazas Malagón ]||Andrés Alejandro Plazas Malagón, Christopher Waters [ Andrés Alejandro Plazas Malagón, Christopher Waters ]|
|Component/s||ip_isr [ 10730 ]|
Christopher Waters: Adam has also produced DM pipelines to get crosstalk matrices from BOT spot data: https://github.com/Snyder005/mixcoatl/blob/master/pipelines/crosstalkSpot.yaml
I understand that we are planning to use these coefficients for now (for example, with the LATISS ITL device).
Please let me know if this ticket is still relevant, as Tim says, or if we can mark it as "invalid" or something else (or if we want o repurpose or open a different one).
||PREOPS-3236 [ 3127971 ]|
Adrian Shestakov has used the DM crosstalk code (gen2?) on spot data from the BOT to analyze flux dependency and get crosstalk coefficients: https://confluence.slac.stanford.edu/display/LSSTCAM/Crosstalk+%28Non%29linearity+in+9Raft+Data
Adrian found out that an intercept needed to be added in order to correct for the crosstalk properly; Chris waters explained that the scattered light from the spot generated this offset: the DM code is designed to work better on on-sky images (after ISr has done its job with the background). This is Chris's presentation: https://confluence.slac.stanford.edu/display/LSSTDESC/January+-+June+2021?preview=/299947817/311526317/czw_ct20210427.pdf.