Parallactic angle code is inconsistent
Closed this issue · 1 comments
Describe the bug
The parallactic angle computation is applied when P jones is enabled regardless of the number of input correlations. This is not sensible, particularly in the case of linear feeds where the partial application will rotate the visibilities into correlations which do not exist.
Data description (if applicable)
I noticed the problem when working on data which only had two correlations (XX YY).
To Reproduce
input_model.enable_p_jones=true on two correlation data with linear feeds.
Expected behavior
Error out with a useful message if the user enables this functionality on data with fewer than four correlations. Additionally, the parallactic angle rotations should probably be disabled by default as they can trip up new users.
Version
main