There was a lot of redundant code between visibility_grid,visibility_degrid, and vis_count. I've condensed the majority of the redundant code to a new file, baseline_grid_locations. This file returns information regarding the locations of the unflagged visibilities in units of pixels in a variety of different forms.
Each of the three original files accounted for flagging in a slightly different way, and I've recreated each of those. The only exception being that unsetting fill_model_visibilities during the model making process will emulate the flagging done for gridding with this new file -- arguably what we want to do anyways. In any case, fill_model_visibilities is currently hard set on anyways.
The power spectra for a typical run are the same between the old branch and the new branch. The images, however, are different on extremely low levels (i.e. 10^-20), which makes me think I haven't fully replicated some float/double behaviour during image making. I will continue to investigate this, but I'd like to get this branch reviewed as fast as possible so that it can be ported to python via Joel in the next 2 weeks.
There was a lot of redundant code between
visibility_grid
,visibility_degrid
, andvis_count
. I've condensed the majority of the redundant code to a new file,baseline_grid_locations
. This file returns information regarding the locations of the unflagged visibilities in units of pixels in a variety of different forms.Each of the three original files accounted for flagging in a slightly different way, and I've recreated each of those. The only exception being that unsetting
fill_model_visibilities
during the model making process will emulate the flagging done for gridding with this new file -- arguably what we want to do anyways. In any case,fill_model_visibilities
is currently hard set on anyways.The power spectra for a typical run are the same between the old branch and the new branch. The images, however, are different on extremely low levels (i.e. 10^-20), which makes me think I haven't fully replicated some float/double behaviour during image making. I will continue to investigate this, but I'd like to get this branch reviewed as fast as possible so that it can be ported to python via Joel in the next 2 weeks.