Testing - NGC 7538 regression script
DSS - 22feb07
Summary of issues:
- In listobs summary, does ArrayID = 0 = D array? This should be
explained somewhere in the cookbook and/or in-line help. Note, what
will E array ArrayID be if we ever get it? -1?
ArrayID is only an index indicating the array in the MS; it is useful
as a counter when multiple configurations/arrays are in the same
MS. I'm not sure where the deduction regarding VLA array came from.
- In datasets where spwid 1 = RR and spwid 2 = LL, if I set one of the
axes to be pol or spwid, I see only one, not both. But I want to
see all polarizations and sp windows, not just one. So, for this
dataset, the spwid or pol slider should be disabled to see both pols
and spwids at the same time.
- When plotting channel vs. baseline, I would like the ability to
average over time. Thus, if I have a strong line, I might be able
to see it in a plot of channel vs baseline in which time is
averaged. Further, if I only see the line on short baselines, this
tells me something about how extended the structure is.
On the list as discussed in the NAUG.
- table lock occurs in the viewer (sent info to David)
tb.clearlocks() # smashes all locks and allows tools to proceed.
# plotsymbol failed even though this was an example in the help.
SEVERE plotcal::setplotsymplols (file
Couldnot understand color: 45. it has to be one of k r g b c y
There is a different color for each ant? So what does plotsymbol do?
# plotsymbol appears to be broken.
# also, there are no axis labels or title on the plot, just little
- plotcal(tablein='ap314.gcal',multiplot=false, caldescid=0)
what is the Calibrater data description ID (combination of SPW and
polarization)? Examples of -1 (all) and 0 don't help. I can't
find additional information in the cookbook, user ref man doesn't
have this parameter. No description on the plot or in the logger,
or in the python command window what the different caldescid's are
# I need a locate function here too so I can identify what data is
associated with a bad solution.
- in plotcal, why are solutions found and displayed for flagged data?
Or, at least, this appears to be what is happening. Flagged data
should not have a gain solution (or at least, don't display it).
- in plotcal, on the right side of the plot, give the color coding for
the different antennas.
- I'd like an overplot keyword with plotcal so, if I make 2 different
solutions, I can compare them. This assumes that I can make one
solution red, and the other black so I can tell the difference
between the solutions.
# good, nice to have the plots remain on the screen after I hit 's'
- plotxy: Note: plotxy only shows antenna name on the plotter window
but then you can only select to plot a specific antenna via its ID.
So you have to use the summary to figure out the mapping. Dach.
Make this consistent.
Agreed as discussed in the NAUG. Not before the test however.
- seg faults when I try to plot the 2nd field in the dataset.
both plot/flagxy work OK when plotting the first field.
example segfault for plotxy (daily, afternoon, 22feb after new build)
plotxy(vis='ngc7538.ms', xaxis='uvdist', yaxis='amp', datacolumn='data',
correlations='RR LL', nchan=1, start=5,width=55,
spwid=[0,1], plotsymbol = ',', multicolor=true)
# Failed on selection: combination of spw and/or field and/or time
# chosen may be invalid --- Segmentation fault (core dumped)
# this should have been valid, there are 63 channels - I want to
average 55 channels
example segfault for flagxy (daily, afternoon, 22feb after new build)
flagxy(vis='ngc7538.ms', xaxis='time', yaxis='amp', datacolumn='corrected',
correlations='RR LL', nchan=1,start=5,width=55,
spwid=[0,1], plotsymbol = 'r,', antennaid=22)
# SEVERE SubMS::makeSubMS() (file
# /home/ballista/casa/daily/code/msvis/implement/MSVis/SubMS.cc, line
# Failed on selection: combination of spw and/or field and/or time
# chosen may be invalid
# Segmentation fault (core dumped)
There is an antennaid 22, This is repeatable when I try to plot
the 2nd source in this dataset (2229+695).
- I'm not sure why antenna id13 cal solutions have dropouts - maybe
the phases are off even though the amps look mostly OK. The source
is so weak that the phases do not track well so you can't really
tell if there is a problem until after the cal solutions are
The bad data are not obvious when doing the regression script
because solutions are not examined closely. Also, the flux cal
(where the largest fraction of bad data are located) is not
corrected and split out so it is not as easy to find.
In any event, it appears that this regression script should delete
antenna 13 or do some careful editing of source and cal data to make
sure that these bad solutions are not propagated through to the
final image. Its probably OK to just delete antenna 13 all
- In the viewer: I need an unzoom button, I don't like to have to
create a little box and then click outside of that box several times
to get the image to unzoom.
On DK's list. See: https://safe.nrao.edu/wiki/bin/view/Software/AlphaIssues to avoid repeats.
- I need to get statistics of a specific region within an image or
image cube (e.g. on source or off source). How do I do this if I
don't have an interactive stats interface?
I can get total image plane stats with:
print 'image max = ',statistics['max']
# image max = 0.375307559967
print 'image rms = ',statistics['rms']
# image rms = 0.0075194593519
This is a pain but it works.
But I can't figure out how to get the image rms in a region with no
source. This is a problem.
You do have interactive stats interface. Draw a region and double click (as before). Statistics are currently displayed in the CASA terminal (temporary)
- update button in the viewer doesn't work. Nothing happens...
- the ngc7538 regression script says that it is imaging continuum
emission in the source but no continuum subtraction is done in the
- viewer: I can't find any way to plot the beam shape on the image -
is this hidden somewhere? It is important to be able to see the
resolution of the image in the corner.
Not yet implemented. On the list of issues (https://safe.nrao.edu/wiki/bin/view/Software/AlphaIssues).
- I tried to use the image profile tool in the viewer - the graph
comes up, the pointer allows me to click on a point, but nothing is
Hold down the button and move the pointer to see the spectrum. We hadn't released this yet.
- the ngc7538 regression script does not have good imaging. You avoid
the problem that you can't make decent deconvolution regions by
convolving the image with a 12" gaussian and creating a monster
image with 13" resolution instead of a decent science image with 3"
resolution. This is cheating. Do you really think this is a valid
approach? I don't.
Grrr. The NGC7538 regression script was based on an AIPS script from the NAUG. The steps were followed to emulate
the other script to compare results. The script has not changed since 2004 (it's a regression script). Suggestions for
improvements are welcome; now is an appropriate time to reset the bar.
- 23 Feb 2007