[daip] pcal

[Eric Greisen]

Laura Chomiuk wrote:
> That really seems to work!
> Strangely, it also makes a big difference on some older data where ~6
> minutes of data are all combined into one scan. Setting solint=0 gives
> gibberish, really high D terms, while setting solint =0.9 seems to do the
> trick!
> 
> I have to admit, i don't really understand what is going on here. Why does
> crossing scan boundaries make for such dramatically wrong results? Is it
> just an artifact of PCAL strangeness?
> 
> Also---the calibrator is faint, and it would be nice to use all 6
> minutes of time on the source. I guess this is not possible, at the
> moment?
> 
> Thanks so much!
> 	Laura
> 
> 
> 
> 
> On Tue, 11 Oct 2011, Eric Greisen wrote:
> 
>> Your data set has > 2 minutes on 3C84 divided into 3 scans.  A default
>> SOLINT is 10 minutes causing it to treat the data as 1 scan.  Break it
>> up by setting SOLINT=0.9 for example and you will get good solutions.
>>
>> Eric
>>

SOLINT=0.9 divides your 6 sec of data into 7 separate times all with 
slightly different parallactic angles.  I think the fitter requires at 
least 2 times and am working on fixing PCAL to force that in cases such 
as yours.  I chose 0.9 in your 3C84 data because that was the length of 
your observed scans.  Try 3 on your 6-min one.

I had to fix a bunch of things - run a MNJ afetr now to get improved 
PCAL (factor of 12 decrease in I/O) and better table sorting incl 
killing the temporary SC files properly.


Eric Greisen