From mrupen@nrao.edu Tue Feb 27 15:32:25 2007 Date: Tue, 27 Feb 2007 15:31:35 -0700 (MST) From: Michael Rupen To: Joe McMullin Cc: aips2-naug@aoc.nrao.edu Subject: [Aips2-naug] More testing NGC5921 ------- mpr 27feb07 Today through correct. Some frustrations along the way. I will switch to quick-and-dirty tests tomorrow, just to get through all this. -------------------- Misc. issues * Is there something like listobs for a gain table? I'd like to know how the gain table was made, and what's in it (e.g., ant-based phase vs. time for two calibrators). - browsetable isn't very helpful without documentation on the table rows & columns ...though it does make me wonder why so many columns are not actually populated! e.g., field_id * Cookbook should include a couple examples of what Mark (or whomever) thinks is a nice way to display various tables/MSs for sample instruments. E.g., VLA/gcal might be multiplot in sets of 3x3, with a good choice of size of pixels. * plotxy, plotcal, browsetable etc. seem to me to be the equivalents of TAPLT and PRTAB in AIPS -- so generic that they can be used for anything, but are not very well tuned to do anything specific. In AIPS we have fine-tuned versions like LISTR/GAIN, PRTAN, SNPLT, etc. to get over this hump. It would be useful to have some similarly tuned versions of the generic tasks for the most common cases. This might be as simple as a set of very easy mini-scripts. * I would like some way of obtaining help on parameters, as they are now fairly consistent from task to task (hooray!). Currently help par returns a message like no Python documentation found for 'n5921.gcal' where 'n5921.gcal' refers to the value assigned to the variable par . -------------------- plotxy * would like simultaneous display of amp & phase -------------------- browsetable * when I try to access a datum I'm not supposed to, I get Invalid operation:...: no array in row 0 in the input screen. But browsetable itself brings up an error window making the same complaint; the problem is that I often double-click (sorry), and that puts the error window *behind* the table browser window I'm using to look at the table. So I start wondering why I can no longer access any of the buttons etc. on the table browser, and eventually figure out that there must be an ERROR window around somewhere which I have to click on before I can continue. This is rather annoying. * Finding the interesting columns could be made much easier. - allow selecting multiple columns before "Hide column" - by default, hide columns which do not vary, or restrict the initial selection for "standard" tables (MS, gains, etc.) to some reasonable set of columns, and have a toggle which allows "show all" and "show interesting" as options -------------------- plotcal * help claims default fieldid=-1, but restore gives default fieldid=0 Am I right in thinking that R/L is identical to R-L? That's not _quite_ clear in the help. help gives the example plotsymbol='-.ro' but running this gives the error SEVERE : Exception Reported: TablePlot: Plot Options: TPPlotter: ERROR : plotsymbol must contain only one colour *** Error *** TablePlot: Plot Options: TPPlotter: ERROR : plotsymbol must contain only one colour * Time labels are kinda nice if you plot only one plot per screen, but rather painful with subplot * multiplot is almost useless - should be able to set ranges - should be able to display pol'ns as separate colors - need something in between (huge) circles and (teeny weeny) points * allow different point types/colors for different calibrators? * need some version of LISTR/GAIN, to print the interesting gain table entries in some useful way. * would be nice to be able to show differences (absolute or fractional) wrt. a mean or median value, running or otherwise -- often one is interested in stability rather than absolute numbers. -------------------- bandpass * help claims default nchan=-1 (meaning all) but restore gives default nchan=0 ...should be made consistent with fieldid etc. * should allow solint to be set to SCAN * would like to be able to do a channel-by-channel solution, or do groups of channel and interpolate simply (rather than by polynomial) * maskcenter is in channels; maskedge is in % . - change maskcenter to maskline (or some such), and accept a list of channels - allow both to be either % or absolute number of channels * Needs more feedback on exactly what was selected & what's being solved for - e.g., solving for source A, averaging over all times, fitting a X degree poly. to channels c1-c2, c3-c4, c5-c6 ...resulting chisq for R= ??? * How can I examine these solutions? -------------------- listhistory * One of the shortcomings here is that listhistory doesn't tell you which calibration has been done -- i.e., which tables were produced from this data set, and how. * Doesn't work on gain tables --> no way to tell how they were created??? * Like the logger, listhistory doesn't know about tasks -------------------- fluxscale * How could I apply the bandpass table as well? Use accum? * reference, transfer should obey the new nicer selection approach (e.g., minimum match) * How can I access these results after running the task? -------------------- correct * Mix of gains is confusing - gaincurve and opacity are outside the normal gain approach of separate tables - .gcal is replaced by .fluxscale, but .bpcal remains around (and is not used in flux scaling) basically, correct looks a lot like AIPS in having several completely independent but interacting gain tables. Except here you have to keep track of all of them. ================================================================================ _______________________________________________ Aips2-naug mailing list Aips2-naug@listmgr.cv.nrao.edu http://listmgr.cv.nrao.edu/mailman/listinfo/aips2-naug