[evlatests] False Symmetrization!

[Rick Perley]

    The result reported yesterday, wherein background sources were found 
to be reflected about the origin with about 10% of their proper 
intensity, was so bizarre as to suggest an origin in the calibration 
procedures -- self-cal can generate such effects in some circumstances. 

    To check this, I split off from the originating database a single 
500 kHz channel, and a block of 20 channels (a BW of 10 MHz), and 
calibrated these as if they were perfectly ordinary continuum databases 
of the calibrator source.  A point-source amplitude and phase solution, 
followed by an image, followed by a proper model amplitude and phase 
solution. 

    The results from this are perfectly normal.   Both test 'continuum' 
databases provided an excellent image, showing all the background 
sources in their proper places, with their proper intensities. 

    The strange symmetrization from yesterday's test on a multi-channel 
pseudocontinuum database (12 channels of 10 MHz each) appears to have 
been generated through the BPASS program, when I attempted to generate 
1-second solutions for each of the 12 channels.   There is nothing wrong 
with this in principle, and the resulting solutions look fine, but 
clearly something made the program attempt to symmetrize the data ... 

    Bottom line is:  All appears very good at L-band -- at least with 
small channelwidth data.  Much more effort is needed to see if we are 
getting the expected improvements as the number of channels utilized 
increases.