[mmaimcal] Matching of frequency profiles for ALMA and ACAcorrelators

[rlaing at eso.org]

On Wed, 22 Jun 2005, Brian Glendenning wrote:

Dear Brian

> It is the intention of the ACA software team that they will only release
> from the ACA correlator system (i.e., including computers), the same number
> of channels to match the baseline correlator mode. 

That's understood, I think. The user wants the same channels from both 
correlators with as near as possible the same frequency profile.

> Our data rate assumption
> has also been that they will match the number of channels. If all the
> processing mentioned in these messages can be done in computers attached to
> the ACA correlator than this is all fine (although the interfaces might need
> modification to indicate the data processing parameters), 

In the calculations presented at PDR, the ACA did use a larger number of
channels internally, but then used weighted binning to match the baseline
correlator channel configuration and frequency profile. So yes, this is internal
to the ACA correlator data processing. Your point is well taken about data
processing parameters needing to be archived. 

> but if instead
> some of the processing needs to be done after archiving we need to discuss
> the implications on the computing system. Which situation are we in? I'm
> sure this should be obvious to me!

I don't think anyone is suggesting combining different channel configurations in
post processing.  The ability to do offline channel binning, Hanning smoothing
etc. should be there anyway. I guess the only issue for post-archive processing
might be if someone wanted to use some spectral windowing function which is not
currently supported by AIPS++. That's my guess, anyway.

Cheers

Robert