search

roman-corgi / corgidrp

Data Reduction Pipeline for the Roman Coronagraph Instrument
BSD 3-Clause "New" or "Revised" License
5 stars 4 forks source link

allow calibrate_darks_lsq to have sets for unique gain-exposure pairs have different numbers of frames #178

Open kjl0025 opened 2 months ago

kjl0025 commented 2 months ago

Is your feature request related to a problem? Please describe. Currently, calibrate_darks_lsq requires an input dataset comprised of smaller sets, each corresponding to a unique combination of EM gain and exposure time and all having the same number of frames. Running frame_select in general will break the latter condition, which will cause calibrate_darks_lsq to crash with an exception.

Describe the solution you'd like Either reduce the number of frames each set by the minimum amount per set so that they all have the same number of frames, or change the fit so that each data point per pixel is weighted by the number of frames in the corresponding set.

Additional context Not needed for the 9/9 release.

kjl0025 commented 2 months ago

Actually, after looking at frame_select more, I think it shouldn't be invoked for the recipe to make a synthesized master dark. The enhancement to calibrate_darks_lsq would be nice to have in general in case something goes wrong and CGI ends up taking frame sets with different numbers of frames, but practically, that doesn't seem like that would be an issue.

And even without frame_select, there will be in general sets with different pixels having more masking throughout the set than others, which means they should have a smaller weight in the fit for those pixels. This may be tricky to implement without slowing down the code (which isn't a great concern since this is run on the ground), unless good parallelization is implemented. Though I'm not sure; maybe it isn't tricky to implement weights in an array way.