Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.atnf.csiro.au/vlbi/dokuwiki/lib/exe/fetch.php/difx/minutes.txt Дата изменения: Mon Oct 6 11:09:36 2008 Дата индексирования: Tue Apr 12 06:37:55 2016 Кодировка: Поисковые слова: п п п п п п п п п п п п п п п п п п п п п п п п п п п п

DiFX Workshop, Bonn, 8-12 September 2008
Minutes compiled by Cormac Reynolds.

attendees: James Anderson, Walter Alef, Helge Rottmann, Alessandra Bertarini,
John Morgan, Adam Deller, Walter Brisken, Mariano Muscas, Antonietta Fara,
Reynolds, (Kettenis arriving Tuesday).


* Review of Action Items:
Work on extending the VEX/input file format for correlation specific details is
required. Vex2Config needs significant expansion to deal with the full gamut of
possible observations - some discussion of the specs for that to come later in
the meeting.

Tagging of release is an issue - CR to do some tidying up of the repository
layout.

Walter A. to put a test baseband dataset of a mixed VLBA/MkIV observation (with
correlation results) on an ftp server.

We should discuss which implementation of Calc (or other) to use.

Walter B. has a bunch of monitoring stuff that should be useful for the
correlator operators (difxmessage). We should discuss this later in the
meeting.

Curtin Wiki to be made available for DiFX documentation. CR to set that up
during the meeting.

* Plan for Discussion Meetings:
Vex2***
calc and calc alternatives
logging (difxmessage)
Testing/benchmarking kit
Wiki
Subarrays/multiple beams
Discussion of the new input file formats
Time (internal to correlator)
SVN tagging organisation
Version labelling

* vex2difx
General agreement was to pursue a new program that converts vex to the
files needed to correlate. It was deemed necessary to supplement the vex
file with a second file that contained correlation specific parameters
that could allow creation of .input files that do not require further hand
editing. Contents of this second file may include the items listed at:
http://cira.ivec.org/dokuwiki/doku.php/difx/configuration

* vex2*** and multiple beams:
Suggestion to replace input files with an XML with schema (Rottmann).
Some ability to specify the pointing direction separately to the source
direction is required for VEX (for multi-beaming). Brisken to propose that to
the VEX gurus. Pulsar binning table also required in VEX. Anderson has more
complex scheme for multi-beaming but would prob. be quite difficult to
implement in the central version.

* logging
Walter B. has an XML based scheme that uses a udp multicast to send the log data
back. See his memo for details of what is logged. Action on Adam to replace
cout/cerr with Walter's message logging stuff. John Morgan to produce a python
logging client to take care of the messages.

* Testing/benchmarking kit
Walter B. to provide a Mk4/VLBA test dataset. Adam to provide his mini LBA
dataset. Would also like a larger test dataset that would heavily load the
cluster and have some complexity (e.g. subnetting). John Morgan to write some
compare/benchmark scripts to make the test straightforward. It will be housed
on the Bonn ftp area and linked from the wiki.

* Version labelling
Helge to add the SVN version numbering to various parts of the code. Write this
to the output fits file, the model file, the logging system. Want to know the
version of mpifxcorr, calc...

* SVN tagging/organisation
CR to make a proposal for this before the end of the workshop.

* Time (internal to correlator).
Possibly replace int seconds, nanoseconds with a more general time class
(based on the work of James Anderson), likely with two 64-bit integers
representing seconds and femtoseconds.

* Delay and UVW
Possibly replace internal UVW object containing tabulated delay and uvw
with something based on the polynomials representing delay and uvw that
Walter B.'s calcif2 produces.



Day 2.
* User's Meeting

* John Morgan
Has experiment with a 2Hz offset at one station - small change in DiFX required
to allow this to be correlated.

* Input file generation
Need a replacement for vex2config. Would like to have it so that the
configuration file specifies for each scan which configuration applies. Need to
decide which vex parser to use. Need to decide a language to write the new
vex2config in.

* Plan for coming days
Tidy up the wiki. Tidy up the SVN repository. Demonstrate that it can easily be
installed by following the new instructions.

Day 3/4/5
* Workshop hackathon
Restructured the SVN repository. Fixed the problem with negative delays. Set up
the wiki with documentation, test/benchmarking, meeting page and much other
useful info. See the wiki for current status.