[daip] Ringing from BPASS

[Eric Greisen]

lincoln greenhill wrote:
> Hi Eric,
> 
>> I have looked at your data a bit more.  There seems to be a subtle 
>> difference between the BP table attached to the fits file with only 
>> 2145+067 (256 chan) and that attached to the fits file containing a 
>> number of sources.  The BP table attached to the 1-source file causes 
>> the ringing whether applied by POSSM or SPLIT followed by POSSM on the 
>> data of the 1-source file or the data of the n-source file.  The BP 
>> table of the n-source file does not cause ringing when applied to 
>> either data set which, I suspect, is the cause of your theory that 
>> SPLIT fixed things.  How the BP table became different is a good 
>> question.  Forcing the software to ignore the VLBI shifting, the 
>> "calibrated" data are nearly the same with both BP tables except for 
>> the outermost couple of channels.
> 
> If I assume that an FFT is performed to effect the shift in frequency 
> required
> for VLBI data, then the problem be triggered by an erroneous value in
> the highest channel of the transform spectrum (lag domain) - perhaps an
> array dimension problem?

No.  It works just fine for everyone else's data and for 1 of the 2 BP 
tables your student sent on the data he sent.  There is not a software 
problem but almost certainly a conceptual problem.

Eric Greisen