[evlatests] Some Good WIDAR news

[George Moellenbrock]

Hi Eric-

Thanks for the clarification.  Glad to hear BPASS is ok.

Another aspect of BPASS I don't fully understand is the "channel zero 
normalization".  Could pilot error in this area be relevant to the present 
question?  I.e., this normalization takes out something that CALIB, 
working on a chunk split out from the whole, manages to see and correct 
successfully.  I gather that estimates like "channel zero" will tend to be 
optimal for the band center when the residual errors are monotonic in 
frequency (like the wobbling delays).

I, too, am generally suspicious of Rick's averaging step...

-George



On Tue, 4 Aug 2009, Eric Greisen wrote:

> BPASS in AIPS divides the data by a model computed correctly at each channel. 
> If I understand correctly, Rick ran BPASS averaging the divided data over all 
> times (6+ hours) to make a basic BP table. Applying just that table results 
> in no symmetrization.  Then Rick averaged up some channels and ran BPASS 
> again, asking for solutions every 15 sec.  It is only at this point that he 
> began to see some strange symmetrical things.  User error - including 
> averaging too many channels - is what Rick suspects is the cause.  Exactly 
> what error remains to be elucidated.
>
> But it is not BPASS making a gross fundamental error.  Note that if BPASS got 
> it wrong then all other aips tasks would also have it wrong - IMAGR, CALIB, 
> etc etc
>
> Eric Greisen
>

--