From tcornwel@cv3.cv.nrao.edu Mon Jan 31 15:54:46 2000 Date: Sat, 29 Jan 2000 09:06:15 -0700 From: Tim Cornwell Reply-To: tcornwel@aoc.nrao.edu To: Steven T. Myers , Athol Kemball , Tim Cornwell Subject: RE: aips++ testing notes from today [ The following text is in the "iso-8859-1" character set. ] [ Your display is set for the "US-ASCII" character set. ] [ Some characters may be displayed incorrectly. ] Steve, Good comments. It's a excellent summary of what looks like a frustrating session. One problems is that it's not clear how much of our documentation you'd read before trying to do this. Many of the questions that you ask are *supposed* to be answered in either "Getting Started" or in "Getting Results". If they are not, we should fix the documentation. Can you please bug() the BUGS or suggestions? If in doubt send it in and we'll address it. Having the bug just makes it easier to track context and progress on fixing. I've added some specific comments below. Cheers, Tim > > Test 1. Use general->ms->fitstoms to read 16156554.UVF into 1608.ms > Successful > > Use synthesis->imager:myimager on 1608.ms : > > advise - it starts up in this function, but it doesnt seem to do anything > sensible, eg. cell and phasecenter are obvious things to default If you Go, it fills in those values. That's basically all advise does. > > setimage - set cellx = celly = 0.05 and nx = ny = 256 > > plotuv - works. However it leaves aipsplot_1.plot etc. behind in cwd > > plotvis - crash. Gives error > Exception: PGPlotterGlish::send - Not attached (No plotter named > aipsplot_2.plot exists) > now plotuv also doesnt work. I left plotuv using the DONE from File menu. > That must have been wrong. How to get another plotter running? Try going > back to manager window and creating a display->plotter->pgplotter tool. > Then return to imager:myimager. Still doesnt work. Start over. Continue > to get errors on trying to close imager with close(), just > dismiss and exit. > Remove aipsplot_1.plot and aipsplot_2.plot. This needs to be more robust > or put warning when close is used on the PGPlotter. By the way, > how DO you > rebuild a pgplotter that a tool can access? > > BUG: THIS SHOULD HAVE BEEN MORE ROBUST. I agree. If you bug() this we'll fix it soon. > > Test 2. Use synthesis->imager:myimager on 1608.ms to emulate difmap > ------ > usual problem with "garbled logtable", this should be more robust > (ie. it should never write crap to log!) We have a defect out on that already. > > setimage - set cellx = celly = 0.05 and nx = ny = 256 stokes=I > is the silly default for phasecenter going to be a problem? No. > > plotuv - OK, generates aipsplot_1.plot, then Dismiss > > plotvis - OK, but default blue points are nearly unreadable > (claim as residual, > though these are really "observed" at beginning). Maybe > default to a lighter > shade of blue or light green. > > Next step not clear, need to consult documentation. Found Imager in the > Getting Results doc. Pretty terse, but seems useful. > > fitpsf - error "Image parameters not yet set: use setimage", but > I did! Must > not have run Go before... > > setimage - Go > > fitpsf - error > Could not find any overlap in frequency between image and > visibility channels > Finished imager::fitpsf > 3.44 real 0.18 user 0.05 system > PSF image 16156554.psf does not exist > Caught exception: Fri Jan 28 17:51:37 2000 SEVERE FTMachine::initMaps: > Could not find any overlap in frequency between image and > visibility channels > > BUG: THE GUIDE SAYS TO USE THE IMAGE FUNCTION TO MAKE A DIRTY > IMAGE. IT SHOULD > HAVE PUT THIS BEFORE MENTIONING FITPSF! Bug() all of this and we'll look into it. > > image - do you need to put entries in image and compleximage? > try without... > Starting imager::image > Calculating image (without full skyequation) > Making dirty image from corrected data > Image is : 16156554.image > Frequency = 8.4149 GHz, synthesized continuum bandwidth = 0.025 GHz > Image polarization = Stokes I > Frequency = 8.4149 GHz, synthesized continuum bandwidth = 0.025 GHz > Image polarization = Stokes I > Finished imager::image > 0.21 real 0.1 user 0.05 system > Caught exception: Fri Jan 28 17:55:59 2000 SEVERE FTMachine::initMaps: > Could not find any overlap in frequency between image and > visibility channels > Could not find any overlap in frequency between image and > visibility channels > > Did I need to do something in setdata? > > setdata - the description "Set the data parameters selection for su" is > unnecessarily cryptic (it is clearly truncating the banner line here). > Defaults look sensible - do I need to "Go" even > if the defaults are OK? If so, this should be changed (ie. > running image > should invoke other necessary procedures with thier current defaults if > needed). > > image - seems to have worked now OK, generated 16156554.image (should this > be printed in the "image" input to image?) > > So far, the required sequence seems to be: > > imager { setimage -> setdata -> image } as per the Getting Results page > > fitpsf - returns bmaj=0.320959568, bmin=0.228605628, bpa=-961.37146 > > The guide says: "If you are using the GUI interface, then you can > immediately > view the image using the View button of the Spanner menu at the end the > entry for the dirty image name." What the hell is a Spanner > menu? I guess > I need to spawn a catalog viewer or something. Spanner == Wrench. i.e. the wrench-like button on the right hand end of the entry widget for the dirty image name. All entries have such widgets for doing things using the entry. e.g. find a file name, view an image, set a region, etc. > > Go to Tool Manager and Show dc. Hey, I see a 16156554.psf but no > 16156554.image > as it claimed when I ran imager. What happened? Figure it out > later. I have no idea what happened to 16156554.image. > I do > see a "View" on the buttonlist at the top of the Catalog window. Click. > VERY show to put up viewer. Try drawing rectangle - can you get > statistics > to appear in the Statistics window below? I see no way to... Click "Done". Set a region and click on the rollup to show the statistics window. This should be in "Getting Results". > Back to imager:myimager window. Dismiss viewer. > > image - do again. The logger says "Image is : 16156554.image". > Where is it? No idea. > > Go back to Tool Manager and create display->viewer:myviewer, then > Go. I get > 2 panels, why? I guess because the default is a datamanager and > displaypanel, > get rid of empty displaypanel. Data manager finds no datasets. > > summary - go. No useful info. I just want to look at the dirty image!!!! > > image - Try putting 16156554.dirty into the "image" form box. OK, now its > there in the catalog. Do view. Finally looks OK. > NOTE: The DONE button is squashed down in the LRC. Looks like > the window > was made too small, if manually enlarged then this button > becomes normal. > Click DONE. > BUG: SOME DEFAULT IMAGE SHOULD BE GENERATED WHEN GO IS DONE FROM > IMAGE, IE. > IT SHOULD HAVE MADE 16156554.image AS IT CLAIMED TO HAVE DONE. > > Now what I really want to do is to find the peak in the dirty > image, input that > to the deconvolver as a mask location. It looks like the > boximage function > will do this. > > Try going to Manager and creating display->image:myimage from > 16156554.dirty > statistics - Go. statsout contains > [npts=65536, sum=-1.83874786, sumsq=0.243959948, min=-0.0053961724, > max=0.0129103474, mean=-2.80570657e-05, sigma=0.00192919758, > rms=0.00192938675] > but doesnt contain the X,Y of the max! This is a PROBLEM. I suppose you > could threshold on the max value, but that is clunky. > > Done for now. Use Close() on imager. Taking long time? Get > LocalExec::SetStatus: abnormal child termination for > /aips++/stable/linux_egcs/bin/imager > and do not get back to glish. Get bad termination when Done > button clicked. > BUG: HAD TO CNTL-C OUT OF GLISH. We have a defect on this. > > Test 3. Use synthesis->imager->simpleimage on 1608.ms : > ------ > use 1608.ms, both spectral windows (IFs), field 1, Cell 0.05", 512 pixels > This did not work well. Problem? Components outside map? > > Try with 1608+656.UVF -> 1608+656.UVF.ms > both spectral windows, Field 1, I & V. Looks OK, but not very useful. > Note that there is no label visible telling what image is I and V, and > no way I see to get it. > > Simpleimage seems to be rudimentary and not very useful. The wizard-like > interface is somewhat intuitive, though its not clear that this is going > to be the right way to go. In particular, it looks hard to generalize. We're about to revamp the wizard idea. We should talk about it a bit. > > Test 4. Another try to get statistics from the viewer. > ------ > Do this from the catalog, with View option: > Use rectangle tool to mark region - no apparent Apply button! > Use polygon tool - this does fill statistics box after I double-click at > end vertex. > BUG: SHOULD ALSO WORK UNDER RECTANGLE DRAWING, AND ALSO SHOULD REPORT X,Y > (PIXELS AND RA,DEC) OF MAX AND MIN. I agree re better reporting. Rectangular regions do (did) work. Better bug() it. Cheers, Tim