[evlatests] Problems with Shadowing

[Rick Perley]

    I reported at the Monday meeting that there appear to be troubles 
with the identification of shadowed antennas, in that some antennas, 
when clearly in a shadowed geometry, were not being flagged as such.

    I have now a much clearer picture of the problem.  I filled the data 
with all shadow flags off, and selected a time at which the source 3C138 
was setting at an elevation of 20 degrees -- the azimuth is very close 
to 275 degrees. 
    Antenna 1 is at W1, antenna 22 is at W2, antenna 11 is at E3, 
antenna 21 is at E2, and antenna 20 is at N1. 

    By looking at the (u,v) coordinates, it is easy to see that:

    Antenna 1 is shadowed by antenna 22
    Antenna 11 is shadowed by antenna 21
    Antenna 21 is shadowed by antenna 20.
   
    So, antennas 1, 11, and 21 should be flagged, and antennas 20 and 22 
should not.

    But in the database, the situation is exactly reversed!  Antennas 
20, 21 and 22 are flagged, and antennas 1 and 11 are not.   It seems 
clear that the shadow algorithm has the direction of the baseline -- at 
least in some cases -- backwards.    I suspect that in cases where the 
lower-numbered antenna is the one that is being shadowed, the shadow 
flags are attached to the higher numbered antenna instead. 

    Rick