[daip] New Staff Reply - [!CQX-600794]: CVEL

[Eric Greisen]

New Staff Reply: CVEL

I fetched your data and ran CVEL in OLD and TST on a 
32-bit computer (the only ones with OLD).  I also ran
it on my 64-bit computer (gnu not intel load modules).
I general, all of them gave essentially identical results
except the 64-bit TST which was even worse.  The channels
at the edges have a very bad bandpass solution, so that
with DOBAND=1 the FFT of the edge channels produces ringing
throughout the spectrum.  Dropping the first 20 channels
made a significant difference - but OLD and TST were identical
including the 64-bit version.  The 64-bit differed only when
enormous ring was taking place - I suspect that it handled it
better in fact.  With doband false, the 64-bit computer produced 
spectra which at first seemed the same (and very undesirable
with ringing throughout), but in fact the amplitudes were ~1 kJy
instead of 600 MegaJy.

I think CVEL is fine...

Eric Greisen


Ticket Details
===================
Ticket ID: CQX-600794
Department: AIPS Data Processing
Priority: Default
Status: Closed
Link:  https://help.nrao.edu/staff/index.php?_m=tickets&_a=viewticket&ticketid=339