14 January 2004
To: Crystal Brogan
Karl Menten
Munetake Momose
Debra Shepherd
Christine Wilson
Friedrich Wyrowski MPIFR
From: Debra Shepherd
ALMA Offline Subsystem Scientist
I want to thank you first and foremost for agreeing to take the time
to test the AIPS++ software for this first ALMA Offline subsystem
test. This test is important for several reasons:
1. Issues identified during the tests will help drive software
priorities to develop functionality for ALMA offline data
reduction and imaging.
2. This is the first outside ALMA user test which will grade the
science software requirements to determine if the software meets
the functional needs for ALMA.
3. Scripts generated during the testing will continue to be used
during automated regression testing and will provide valuable
insight to the pipeline heuristics team who will use the detailed
processing steps and parameter inputs to develop automated
pipeline decision trees.
4. The results of the testing will be evaluated in the next Critical
Design Review to determine how the offline subsystem development
should proceed.
Four datasets have been chosen which will be processed through AIPS++:
two are spectral line Plateau de Bure Interferometer (PdBI) projects
which observed 1 and 3mm line and continuum emission; two are 13mm
spectral line Very Large Array (VLA) projects. The PdBI data
represents low S/N data that may be appropriate for ALMA
sub-millimeter observations while the VLA data represents high S/N
data with a large number of baselines that should have similar
processing steps to ALMA millimeter observations.
The AIPS++ software is in an active stage of development. As you are
aware, the current infrastructure, GUIs, and scripting language (all
Glish-based) will be replaced leading to a change in the overall user
interface. A prototype framework is being built and evaluated but is
not ready to test at this time. Thus, this test will focus only on
testing the underlying functionality of the C++ code to determine if
it is adequate to meet the needs of ALMA. The test will use Glish
scripting to reduce and image the data. GUIs will only be used when
the user must interact with the software (e.g. identifying and
flagging bad data, interactively defining deconvolution regions,
viewing the final image and obtaining statistics). Development of the
existing Glish-based GUIs that are not essential to the data reduction
paths in this test has been discontinued while priorities changed to
support the framework conversion and adding functionality required for
millimeter data reduction.
The primary focus of this test is to evaluate the underlying
functionality of data reduction and basic imaging steps for single
fields. This includes filling the data, basic editing, calibration,
deconvolution, and some basic image analysis needed to evaluate the
test results. The documentation (both cookbook and on-line reference
manual) should be adequate for you to figure out how to proceed with
the test (this assumes some basic knowledge of radio interferometry
techniques). Comments on the cookbook and other documentation are
welcome.
Because the scripting language and GUIs will be changed over the next
year, this test will NOT evaluate the usability of script language or
the GUI interface. User interface elements will be tested in 2005.
Essentially, if you can do what is necessary to calibrate the data and
get a scientifically accurate image (even if the syntax or process is
a bit complicated), then this will be considered a successful test.
It is expected that this test will take roughly 40 hours of your time.
The test begins on 15 January 2004 and will end on 15 February 2004.
During this month, the AIPS++ software developers and I will be
available to help you with questions or problems that you may come
across. Please do not start the test at the last minute! Be aware
that this is the first time that outside ALMA users will test new
areas of the code and, although extensive testing of the software has
been done in-house, software bugs may still exist. The developers
will do their best to help you through any difficulties, including
fixing any bugs and distributing new software if necessary. But they
can only provide this support if they are given adequate time. If you
find you cannot complete the test by 15 February, please contact me as
soon as possible (dshepher@nrao.edu, work phone: 1-505-835-7398).
During your testing, PLEASE keep a detailed script of the processing
steps you do and your parameter inputs. Example scripts for both VLA
and PdBI processing will be provided with the Cookbook. You can use
these scripts to guide your reduction and minimize typing (you will
just have to change inputs and then cut and paste commands into the
Glish command line window). Detailed testing scripts are important
for several reasons:
1. If you have a problem, you can just send your script to the
AIPS++ software developers and they will be able to help you
figure out if the problem is with a typo in the script or
if you have uncovered a real problem.
2. Test scripts will be reviewed and given to the pipeline
heuristics team. Your processing decisions are based on years
of experience and how you set parameters at each stage in the
reduction process will provide valuable insight into how to
best process millimeter data. This information will help
establish the heuristics to automatically process ALMA data.
When you complete your testing, please fill out the questionnaire
provided to you describing your testing experience and any problems
you encountered. Also, please examine the the list of requirements
that are being evaluated during the processing. Provide grades for
all requirements that you were able to exercise. All testing reports
and requirements evaluations will then be used to write a final report
which will be submitted to ALMA management and the AIPS++ development
team.
Thank you again for your valuable time! Tomorrow I will send you
detailed information about where to download the latest AIPS++
software and installation instructions, along with information about
where to download the cookbook, example scripts, and contact
information. The data you will be testing has been burned onto a DVD
and is being shipped to you today. Please select the data to test
based on the list below:
Data: Testers:
------------------------- ---------------------------------
NGC7538 - NH3, VLA Munetake Momose & Karl Menten
G192 - NH3, VLA Crystal Brogan
h121 - U Cam, PdBI Chris Wilson & Frederic Wyrowski
l02d - I10216, PdBI Debra Shepherd
(all data are proprietory, do NOT distribute).
If you have any questions, issues, problems, please don't hesitate to
contact me.
Sincerely,
Debra Shepherd
National Radio Astronomy Observatory
Subsystem Scientist for the ALMA Offline software
ALMA test website: http://www.nrao.edu/~dshepher/alma