I think Joel may have found a bug, and I wanted to get your opinion.
Line 64 of visibility_grid.pro finds the group identifier if there are multiple beams being used:
group_arr=reform(psf.id[polarization,freq_bin_i[fi_use],bi_use])
However, bi_use has the same dimensionality as the num of visiblities (4 million something), and psf.id has the dimensionality of x, x, 8128 . It doesn’t error, but I think the group_arr is set to the last value in the array if it exceeds 8128.
I think maybe bi_use should be bi_use_reduced. I wanted to check with you though, because it seemed like this could have potential for IDL black magic.
Maybe this hasn’t been an issue thus far because we generally use one beam. And maybe this explains why our dead dipole tests never really went anywhere.
I played around with a couple of iterations, and this was the fastest/most compact. Still looks a little clunky, so I'm open to suggestions.
As mentioned in Slack:
I played around with a couple of iterations, and this was the fastest/most compact. Still looks a little clunky, so I'm open to suggestions.