From efomalon@cv3.cv.nrao.edu Mon Oct 15 15:52:53 2001 Date: Mon, 15 Oct 2001 17:09:15 -0400 (EDT) Subject: NAUG 0ct 10 summary Summary of NAUG Meeting October 10, 2001 Attendees: CV: John Hibbard SOC: Mark Claussen, Ed Fomalont, Kumar Golap, Ralph Marson, George Moellenbrock, Steve Myers, Debra Shepherd, Greg Taylor, Jim Ulvestad, Robert Zavala Introduction: Ed F suggested that the NAUG should meet two times per month; one meeting mainly the NAUG and the other with the NAUG plus aips++ groups. Two meetings per month could increase the productivity of the NAUG and foster closer communications with the aips++ development. Testers' reports: Debra S thought that the procedure to submit a defect directly from a tool---listed under the Help button on the upper right---could be made more accessible. Data concatenation of measuring sets cannot be done if the measuring sets were written by the aips++ filler directly from the archive (distribution) data. The work-around of writing the ms files to a fits file and then rereading them into aips++ for concatenation is extremely tedious. She experienced some problems with the region-maker; problems adding a second region and with unions. Next, the scripter output is now more complicated: Something has changed. Finally, multi-scale cleaning produces excellent quality images. However, the execution times are slower than hoped for. John H had several uncertainties about the time gridding and preavg times usage in calibrater. (ed note: various functions in calibrater have caused confusion to the testers over many months.) George M and others tried to clarify some of these items. The time iterative method in aips++ also made bandpass determinations more piecemeal than desired. Sometime obscure work-arounds were suggested. Solving for the bandpass, also solves again for the gains unless some reset is done. Finally, plotcal only plots one antenna at a time. George promised to send John H and the NAUG a glish script which will plot the calibration values for all antennas. George M also confirmed that solutions from calibrater were still confused by autocorrelation data. Ed F reported successful filling of data from the archive. Flagging of the data, however, still has a problem of specifying specific Stokes parameters. Autoedit could become a powerful, easy to use, editing tool, but he found the simple implementation with defaults flagged all of the data. A multi-faceted clean in aips++ and aips compare extremely well on a deep image of hundreds of sources. The rms's are nearly identical with a 2000:1 dynamic range. However, the aips++ execution time was a factor of 10 slower in aips++ compared with aips. Finally, Ed F used the interactivemask that Kumar G has tied to the viewer and found it very useful. The discussion then slid into documentation, especially those geared to the users. The BIMA cookbook is very detailed, somewhat gui-based, and generally well-written with good illustrations of gui functions with very good hints of possible troubles or confusions. The NAUG hopes the the U. Illinois group will continue this development and we will contact them to tell them of our interesting in helping out. John Benson (not in attendance) has written a step-by-step glish approach to aips++ data reduction, with links to aips++ documentation and many scripts. This document has been helpful to several testers and John (private communication) would like to make improvements and additions, and solicits comments from the NAUG. Finally, George M is writing a VLA cook-book for aips++ reduction and the NAUG hoped that he will show us a preliminary version for comments. Besides the extensive aips++ documentation from the groups, and the above user guides, Steve Myers, on his web site (www.nrao.edu/~smyers/aips++), has links to lots of aips++ information, test logs, test data and scripts. At the present time there is no consensus about how to integrate all of the pieces of information about aips++ (both within the formal aips++ effort and outside of it) for optimum use by the general user community. We (NAUG and aips++ group) hope to converge on something more specific in a few months. A related organization effort will also apply to the hundreds of scripts which are now distributed over the project. In order for the NAUG to help test the aips++ system in time for release in November, we all agreed to use the NAUG exploder at aips2-naug@nrao.edu to distribute information. REVIEW of Problems Discussed in Meeting *=More immediate concerns, especially for aips++ release +=Less important concerns $=Longer term concerns * Concatenation of two measurement sets read in from aips++ archive. Less general fix to permit some tools to access more than one msfile is not as good. * Flagger does not properly flag Stokes parameters. * Calibrater has several problems (plotcal for more than 1 antenna, solving for bandpass with again solving for G, solutions getting confused by autocorrelation data, time gridding and scan information). * Autoedit testing * Documentation of interactivemask * Preliminary distribution of VLA aips++ cookbook + Region Manager problems (more than one region, unions) + Scripter output more complicated than necessary + More obvious Tool-based procedure to invoke defect form $ Execution slowness of imaging (multi-scale clean and multi-faceted clean) $ Get information about BIMA cookbook development at U. Ill $ Comments to John Benson on the 'step-by-step' guide $ How to organize user-oriented documentation. ------------------------------------------------------------------------------- From efomalon@cv3.cv.nrao.edu Wed Oct 10 08:46:48 2001 Date: Tue, 9 Oct 2001 23:29:12 -0400 (EDT) Subject: NAUG - Oct 10 Hello all, A reminder, and a slightly-revised agenda: NAUG MEETING WED. OCT 10 13:15 (MT) 15:15 (ET) 12:15 (PT) Usual Video Conference Setup 1. Comments from testers on recent work (all) 2. Important testing needed before next update freeze (AIPS++ group) 3. Bull-session on user-documentation: Continuing John Benson's 'Step-by Step Instructions... Mostly glish script-based and useful info. Also look at the BIMA AIPS++ cookbook and tutorial in http://monet.astro.uiuc.edu/aips++/gentle/ http://monet.astro.uiuc.edu/aips++/forBIMA/TUTORIAL/laitut.html Mostly gui based AIPS++ cookbook (Moellenbrock) Questions: Continue on with John's and/or BIMA documentation? Comparison/contrast with George's cookbook? 4. Organizing useful scripts in one place. Should there be a well-defined script naming convention, and a definite style for entering parameters in a script? 5. Expanded roll of AIPS++ User Reference Manual to become the main on-line documentation that the user relies on---something like the AIPS help and explain files? This does duplicate some of what is in the Getting Results and Getting Started documents, but may be a more central place where the users can find most of their needs. Links to other more detailed and philosophical discussions would be included. Other topics to think about are: 1. Debugging Interaction between testers and aips++ group. Keep relatively fluid as now, or try to structure a bit more? Example, should a tester work with an aips++ programmer in a relatively concentrated way for a week to test, debug a tool 2. Quick communication from aips++ group to testers about recent developments and/or new software to check. Thanks, Ed