ALMA SSR Committee Phone Meeting 2003-02-12

Connection Details

• Call date: 2003-02-12 (Wednesday)
• Call time: 15:00 UT
• Duration: 1 hr
• USA Toll Free Number: 888-795-6990
• Toll Number: +1-484-630-4370
• Passcode: 99640
• Leader: Robert Lucas

Attendance

Minutes

1. SSR Membership
   Debra Shepherd (NRAO) will now be (and in fact has already been) an acting SSR member, in replacement of Steve Myers, until administrative details are sorted out by Brian Glendenning.
   We welcome Debra to the SSR.
   
   
2. Subsystem Scientist Activities (all)
   
   
   • Chris Wilson -- Pipeline
     
     Attended Pipeline discussions and AIPS++ re-use test in Garching Jan 27-31.
     Was trained in using AIPS++ to reduce IRAM data. Participated in start of Phase II, finding several bugs/enhancements needed to apply tools to a new data set (H121). All changes that I identified were needed were implemented by George and Kumar by the end of the week. Will be one of the testers in Phase II.
     Participated in the full-group pipeline discussions as well as separate break-out discussions of the NRAO team alone and the IRAM team alone. After those discussions it was clear a smaller group needed to meet; participated in the first of those small group meetings with Cornwell, Gueth, Raffi, Schwarz. Did not participate in further more technical discussions of the small group on the last day (going beyond my expertise). Participated in full-group summary at end of the week.
     In my opinion, a lot of progress has been made on Phase I. If the participants had decided that Phase I was finished last week, I think that would have been a reasonable decision. However, it's also reasonable to try to polish Phase I a little more (the actual decision). There also seems to have been a lot of progress reconciling the Pipeline designs presented by the NRAO and the IRAM groups. I'm a little worried that the solution relies on changes to AIPS++ that may not be what is recommended by the technical review (python, ACS), but at least we have a framework we can use for PDR.
     
     
   • John Richer -- Telescope Calibration
     
     My activity this month has been to
     • read the revised interface document
     • attend the group telecon
     Activity on the project continues to be mainly on the detailed architectural considerations, with no input needed from a science requirements perspective.
     
   • Steve Scott -- Correlator
     
     Wrote detailed doc outlining details of data normalization and scaling. Discussed and iterated with Jim Pisano. The channel average data is about a quarter of a MB/sec in binary form, amounting to 6% of the currently specified total. The proposal for an increase in data rates to support the enhanced correlator option has been waiting for more input on technical details from the computing team. The proposal will go off to the science ipt with or without further input.
     
   • Leonardo Testi -- Observing Program Preparation
     
     With Francois we agreed on a first draft of requirements for SimLite that was circulated to the SSR for comments. I received comments from many people on the unresolved OT issues. The final recommendations that I transfered to the OT subsystem are:
     1. allow in the design maximum freedom at breakpoints on SBs modification;
     2. breakpoints should be defined at Phase I for TAC approval;
     3. no requirement for simultaneous SBs execution;
     4. no requirement for antenna/subarray control at the OT level;
     5. there will be a need for an observing mode requiring maximum "istantaneous" frequency coverage (using expert scripts developed by the observatory staff, but not requiring separate SBs or subarray selection at the OT level).
     I have also participated in the AIPS++ training in Garching last week.
     
   • Mel Wright -- Scheduling
     
     working on ALMA simulations..
     
   • Francois Viallefond -- Archive
     
     
   • Jeff Mangum -- Control
     
     Over the past several weeks we have addressed a number of ATF issues related to monitor and control of the VertexRSI antenna.
     
   • Peter Schilke -- Data Reduction User Interface
     
     Prepared and submitted version 1.1 of DRUI requirements, taking the comments by SSR into account.
     
   • Debra Shepherd -- Off-Line Analysis
     
     There was a meeting at ESO in Garching to evaluate AIPS++ for ALMA off-line data reduction. On the first day of the meeting, AIPS++ and CLIC results were compared in detail by the AIPS++ project team (Phase I of the demonstration). The two packages compared favorably in terms of solution values, phase transfer from 3 mm to 1 mm, and final image RMS and peak flux density. The AIPS++ team worked with IRAM staff to make additional comparisons and resolve any remaining issues. During the rest of the week, we began Phase II testing. I wrote a 'cookbook' on how to reduce IRAM PdBI data in AIPS++ and Steve Myers and I gave AIPS++ tutorials to testers. We also obtained CLIC tutorials from Jerome Pety and Frederic Gueth. SSR members (Chris Wilson, Leonardo Testi), and IRAM staff (Frederic Gueth, Jerome Pety) began to reduce new data sets in AIPS++.
     After processing several new sets of data we identified 12 bugs and 4 enhancements to the AIPS++ software along with 4 bugs in the CLIC conversion routine to ALMA-TI FITS. Following the meeting, I've submitted AIPS++ bug reports or enhancement requests for all items identified during the Garching meeting and I am working with the AIPS++ project to get the problems resolved.
     I updated the IRAM data reduction cookbook for Phase II testing. A detailed report of the Phase II testing and problems encountered along with the latest version of the IRAM cookbook is available at: http://www.aoc.nrao.edu/~dshepher/alma/iram.cookbook.ps.gz
     I also met with Francois Viallefond on 7feb03. I gave Francois a short AIPS++ tutorial on how to reduce IRAM data and Francois gave me a demonstration on the ALMA simulator he has been developing.
   
   
   
3. DRUI Requirements (P. Schilke)
   Peter sent a new version a few days ago. Comments should be sent to him before Feb 26th. Then they will be included in memo 11 as a post PDR correction through the project-wide change-control mechanism.
   
   
4. Sim Lite (L. Testi, F. Viallefond)
   The requirements presented by Leonardo and Francois did not get much comments. There seems to be a consensus about what the SimLite is and is not, and what the goals are: enable the (novice) user to better define his/her project in terms of angular/frequency resolution, sensitivity, etc., by looking at a model image as seen by ALMA, using an average spatial filter to represent a given configuration, ... More detailed simulations would mainly involve more experienced observers that would use the off-line simulator. The requirements should make this more clear.
   All should sent their comments to the authors by Feb 19.
   
   
5. AIPS++/IRAM test (R. Lucas)
   Major progress have been made before and during the data reduction week in Garching:
   1. We have agreed that Phase I is over. Images look OK and compare well (though minute details may still persist). An updated report will be provided soon.
   2. Phase II has been started under the coordination of J. Pety at IRAM. Data sets have been distributed to 8 testers. The version of Aips++ to be used is now available (several defects found in Garching have been very recently corrected). Testing period will extend until Feb 28. A Cookbook has been written by Debra and is available at http://www.aoc.nrao.edu/~dshepher/alma/iram.cookbook.ps.gz. A questionnaire will be sent to all testers, their input will be compiled into a document due March 10.
   3. Phase III has also be started. R. Lucas has prepared a data set from 2 hours Phase I data, where the PdB antennas are cloned to produce 8, 16, 32, 64 antenna data sets. Point source visibilities with noise are inserted. The program that generates it has been sent to the Aips++ group (Raymond Rusk). Data has been processed through Gildas and partly through Aips++.
      • Most of the elapsed computing time is taken by calibration (90%), with 10% spent filling the data and 10% generating calibrated visibility tables. The time spent in imaging is small (two continuum images, two 48 channel line images, all 256x256).
      • With a present-day desktop computer one processes data at a good fraction of the average data rate for 64 antennas (very encouraging).
      • Very preliminary end-to-end processing rates range from 2.0 to 3.4 Mb/sec for Gildas and 0.4 to 0.6 for Aips++. Work is in progress to understand this better, taking into account that e.g. different processing options are taken by both codes (such as applying the calibrations at each step for Aips++).
   
   
   
6. AIPS++ Benchmarking Plans (F. Viallefond)
   We briefly discussed the benchmarking plan presented by Francois. While Phase II results will be available for PDR, The goal is to have benchmarking results at first CDR (mid June). The data sets sizes should be estimated as they can thus be prioritized depending on the needed data rate. It is important to see if the relative importance of calibration persists for observing modes that are more demanding on imaging (mosaics). For benchmarking with other packages it is preferred to use the same input data, this should be possible by exporting the data set generators rather than the data set themselves.
   
   
7. Data Rate issues (S. Scott)
   Steve has sent out request to the Science IPT to get their advice on a new data rate specification, taking into account the recently proposed correlator extension . Steve recommended to stick with the proposed peak date, that allows to use the full 32K channels with an integration time suitable for the largest antenna configurations, the main question is about the average number of channels. A first reaction is that it would make sense to use all the channels all the time, the data rate being only modulated by the integration time (varying with configuration), to allow serendipity searches and thus optimize the science output.
   
   
8. Meta Data (R. Lucas)
   R. Lucas is working on it, trying to get something out in the next few days.
   
   
9. Date of next phone meeting According to the schedule the following one will be on: 2003, March 12th 15:00 UT