[daip] BPASS patch

[Eric Greisen]

Matthew Lister wrote:
> Hi Eric, a quick clarification - is this saying that all data processed 
> with >Sept 20 2009 AIPS versions were affected,
> including VLBA data?
> 
> Thanks,
> Matt
>> >> 2. Bandpass application acquired an error probably on September 20,
>> >> 2009.  This error caused it to use all channels in computing the 
>> average
>> >> amplitude correction to be used in adjusting the weights.  Bad
>> >> calibrations at the edge channels can make this a very bad idea.  The
>> >> code that causes all arrays but VLBA and EVN to use the central 75% 
>> was
>> >> reinstated.  This correction involves a lot of link edits - a MNJ 
>> is the
>> >> best way to fix up your code.
>>   

Yes - all data were affected.  Note that for VLBA and EVN, all channels 
are supposed to be used and the corrected code continues that. 
Typically, the edge channels are flagged rather than "good" but crazy 
numbers for these arrays.  This problem only affects the weights so - if 
the BP values were reasonable, the problem is not a big deal and the 
values for each antenna would be similar.  We hit a case in which NaNs 
were present (I think) in channel 1 or N for some antennas and those 
ended up flagged out because of the problem.

Eric Greisen