One can do DelayCal within a schedule block as long as one has observed an appropriate source. One needs to have declared, from within the OT, a DelayCal intent a source for multiBBs Also if one has included multiple phase cals one can redo baselines
Instructions from Todd:
Log onto osf-red.aiv.alma.cl (need VPN if outside network)
There are two ways to run it: 1 to automatically locate the asdm, load the data to an ms, and generate the delaycal table: au.fixMyDelays('uid://A002/Xblahblah/blah',None,None,True) 1 if the asdm has already been loaded, you can do this: au.fixMyDelays('','delaycal.new','your_ms.ms',False) (Note that if you choose option (2), you must have run importasdm with asis='*').
In Charlottesville simply copy the tar file from Todd's home area on polaris:/users/thunter/analysis_scripts.tar and unpack it somewhere.
Then after starting casa: * sys.path.insert(0,'/users/awootten/wherever_you_put_it/analysis_scripts) * import analysisUtils as au In this case, you will have to use option 2 (above).
After above: The final step is to apply the delaycal table using applycal, and gaintable='delaycal.new' Then, if you plot the corrected data in plotms, the phase should be flat.
Discussion of four quasar data Crystal shared some data experiences on the four quasar data, which may be found in the undisclosed location. New baseline solns to be applied to older four QSO datadata.
See Page one of document. pdf on wiki seems to show the phases for 5 and 8 are better. EF agrees completely.
See page two. orange quasar seems to be weakening over the time period. right hand side. magenta pictor a has wide scatter.
note on p3 that as uv distance increases the phase increases. apparently the source is multiple. Not a good calibrator. This is B6 compact. Moral of the story is picking four quasars, two are not very good calibrators. Opinion of phase transfer is improved. In general we need to improve how we do baseline calibration. How will we distribute the new baselines to users.
Ed did very much what CB did. He got a position which differed. He didn't see resifual phase, still thinks it is a good pt source, could have inner jet, but will double check.
Other three cals are in icrf list with very accurate posns. Except for discussion on pictor's multiplicity completely agrees. Improve cal list. EVLA uses 24 hours obs list, can take any 30 mins and have coverage anywhere over sky. Will add in more sources woudht of -40 ICRF positions. Should also help baseline accuracy.
CB: one problem with getting a good baseline is getting a good track as the system only goes for an hour or so before crashes. Right now cannot combine shorter tracks to get one complete baseline solution. If we could get one good 8 hr track it would be much better. Need better coverage on sky also.
Ed has looked into writing in puthon a bl fitting program to run from ms. Bojan has begun something like this. Robert not excited about it as telcal can do this somehow.
Agreed, need to get long bl run, perhaps from concatenating multiple runs. Ed will look into Bojan's code. CB: Michel looking into whether he can concatenate asdms. Will see if he can get that to work. Ed--we would want that ability offline anyway. CB: Good news is that we can correct data after the fact. Important to conjugate LSB phase before position correction. Ed. Will look into Pic A. Also to play with wvr corrections. phase scatter increases with larger bls, prob h2o wv changes short term. will look into wvr corrections. CB: DV05 has interpolation for solutions as wvr is broken.
There was a discussion on ES configurations. FB's algorithm to be used. Boone is the way to go for now. Discussion next week to discuss details. "What are the requirements? There was some post-meeting email discussion. Ed summarized: The noise on an interferometric image can be cleaned (noise is a convolution of the psf) and then the rms noise is artificially reduced by transferring the original noise with a messy psf shape into noise with a nice Gaussian shape (positive and negative components). The more antennas, the smoother the psf, so this effect becomes smaller.
"Because of the limited number of measured u-v points, you must constrain the solution to cover a smaller number of synthesized beams, otherwise the image is ambiguous. The most convenient method is to box the 'obvious' emission regions. The situation will become worse for many point sources or an extended source, and even worse when you add in systematic errors which we know will occur. There is no way around this constraint of determining more parameters than observables. This ambiguity usually starts at about 3 times the rms noise for the kind of sidelobes expected with 16 antennas.
"The above effect is not much impacted by the specific configuration, although lower sidelobes will help a bit. What you want to do is to have relatively good angle coverage without large holes in the u-v plane. Details don't matter much.
"But, there are important considerations of early science that will impact the array 'design'. If mosaicing will be a reasonably large component of early science, then a center filled array (somewhat like that recently proposed) makes sense. However, if single-pointed observations of modest-sized sources are more likely, then a uniformly filled area (in the u-v plane) makes more sense and will give nearly a factor of two increase in resolution, compared with the center-filled array. Also, if there will be two configurations available, then a somewhat center filled with 300m maximum spacing, and a somewhat uniform array with an 600m maximum spacing (I just made up these numbers as illustrations) make sense.
"These decisions impact the array design much more than the details of the specific antenna locations."
AW notes that any early science at B9 will have to deal with extended sources observed with a small beam; IMHO that favors a center-filled array as Ed was noting. I think that good response to extended sources should be a requirement.