[daip] Re: AIPS : two problems

[Eric Greisen]

R. Niruj Mohan writes:

 > I have been using the 31DEC03 version of AIPS (no midnight job, tarball
 > downloaded on april 14th 2003). It is for linux and we are running it at
 > the Institut d'Astrophysique de Paris. The problems are :
 > 
 > 1. I make an SN table (through selfcalibration) on a datafile (file 1)
 > with both stokes RR and LL. I have a split version of the datafile (file
 > 2) which has only one stokes and I transfer the SN table to file 2.
 > When I want to see/apply the SN table on file 2 by using SNPLT or POSSM
 > respectively, for example, I have a problem. Since the file 2 has only one
 > stokes, it should take the corresponding stokes gain solutions from the SN
 > table -- but it always takes the gain solutions for stokes RR.
 > If file 2 has only stokes RR then this is fine. But if file 2 has only
 > stokes LL (which is what I had), then it still applies the RR part of the
 > SN table. In fact, even SNPLT shows only RR for both the RR and LL
 > solutions.
 > I got around it by retaining both stokes in file 2 and imaging only LL at
 > the final stage, but the problem is real. And it exists for multiple AIPS
 > versions and for multiple telescope datasets (WSRT and GMRT).

    I do not regard this as a problem.  The operation that you have
done is not a correct one in AIPS view of the world.  If UVCOP
offerred the option of copying only 1 polarization, then it would have
to correct the SN table to have only one.  There are programs that
will fail if there are different numbers of IFs and polarizations in
attached files than there are in the data themselves.  Users that copy
extension files around without insuring that they match the data,
frankly, get what they deserve.

 > 
 > 2. I ran into another more complicated problem with POSSM when I tried
 > plotting the spectra of individual baselines. I use
 > aparm (1)=1; aparm (9)=2
 > docalib -1; solint=0; nplot=9;
 > flagver=1 (FG has flags applied to all channels)
 > bparm=1,n,n (i changed the value of n but the problem remains)

      This is not an option that is used much.  Did you leave bparm(4)
= 0 because the denominator can then never be less than 0.  I notice
that this BPARM(4) test has an error - where BPARM(4) is a flux and
the comparand if the square of the flux.  POSSM will not accept bparm
= 1,6,6 for example.  It changes that into 1,0,0 and sets the channel
range to the 75% center range of channel.  For some reason (POSSM is
full of places where that phrase occurs) it will not let you define a
single channel as the continuum.  I will loook more at this option
which I think I have never seen used.

Eric Greisen

 > 
 > At some instances (I cant figure out the exact cause, but the problem
 > happens only for some FG tables), POSSM doesnt plot baselines at all and
 > doesnt mention some others : that is, typically I get messages like --
 > 
 > TROUSS> POSSM1: Channel 0 low: ants:  1  3 time 1.459810E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  3 time 1.460005E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  3 time 1.460199E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  3 time 1.460393E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  3 time 1.460587E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  8 time 1.459810E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  8 time 1.460005E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  8 time 1.460199E+00
 > TROUSS> POSSM1: Channel 0 low: ants:  1  8 time 1.460393E+00
 > and this goes on for all baselines and at the end of it, I have no plots
 > at all. But the data is okay because :
 > 
 > a. When I run POSSM with FLAG=-1, bparm=1,n,n it plots ALL baselines
 > normally.
 > b. When I run POSSM with flag=1, bparm=0, it plots all baselines normally.
 > c. Only when I have flag=1 and bparm=1,n,n does it have problems (and that
 > too only for some FG files, not for all files -- I have checked the FG
 > files and they look fine otherwise).
 > d. When the problem arises (as in (c) above), there are two things :
 >     1. It says Channel 0 low: ants:  1  8 time 1.460199E+00, for eg and
 >     2. It also misses lots of baselines completely -- for eg, above, it
 >        missed all baselines between 1-4 and 1-7
 > e. Even with flag=1 and bparm=1,n,n, I can still make POSSM plot all
 > baselines if I specify them explicity --for eg, I say antenna=1,
 > baseline=3,4 then it does plot them.
 > 
 > As you can see, all this is pretty confusing ... hope you can help on
 > this. If it is a known bug, I can upgrade the aips release here. Else, can
 > you let me know if there are patches available ?

We always recommend staying current with the code.  POSSM has had bugs
fixed since your version although nothing that relates to this
probably. 

Eric Greisen