3 February 2003
Attending: David Band, Jim Chiang, Seth Digel, Heather Kelly, Pat Nolan, Christophe Perrot, Dirk Petry, Bob Schaefer
[Minutes by SD]
RS - Karl Young is working on setting up accounts at SLAC for the 2 machine development Beowulf so we can get started. We are developing use cases for database access - or actually deciding on a template for the database access use cases - and should have some next week. That's about it for now; in writing up the use cases, our goal is to better define the interfaces.
SD - Will you be posting minutes from your DB meeting?
RS - Should be online, linked to the DB Web page, within the hour.
PN - We had a fairly successful VRVS a few mornings ago. Recent development has been the decision to write some skeleton code that people can work on. Jim Chiang started this, I fleshed it out, Toby Burnet made it into C++ code and a legal package. Figuring out what are the major classes and what are the major parts has been a useful step.
DB - This is specifically for the likelihood tool?
PN - Yes; one implication is that people who are going to work on it will have to use CMT and CVS.
SD - You've got class definitions but no innards at this point?
PN - Yes, and no main program, just interfaces and internal layout.
SD - Will there be iteration on the classes?
PN - Seems certain; a dose of reality will help. The first goal is to reproduce the functionality of Jim's toy prototype and let lessons from that feed back into an understanding of what we are trying to do.
[No news this week; Masaharu was unable to attend the meeting.]
DB - I have been working on a prototype of unbinned likelihood to implement spectral analysis. The GRB Web page has also been updated. Sandhia Bansal is updating documentation of the A5 event binning tool.
SD - Is the code for the tool available?
DB - Good question. Bob might know.
RS - A copy is in the HEADAS repository, but we should get it in the CMT repository. Sandhia and I will look into it.
SD - This will be the first real analysis tool in the archive.
RS - I'd call it a real prototype.
SD - I'm about to set up the Source identification Web page for this area [now linked to the Science Tools web page]. The main question that needs to be discussed is the role of the U9 (catalog access) tool - arranging access to catalogs at HEASARC, whether we want our own copies or whether we want U9 to be an interface to the outside world. An elegant solution may be to set up U9 as an XML server using Qt so tools [like A1 and U7] can get XML input directly for source models. We would define our own XML tags to be in common with the catalog access tools. The Web page will explain in more detail.
SD - What's an XML server?
DP - Like a Web server but produces XML instead of HTML.
DB - A current effort for NASA data centers is to support queries to each other. So by accessing HEASARC catalogs with queries, by the time of launch we should be able to access all NASA data centers. So we should be able to access remotely any catalog under the sun.
DP - Ideally we'd 'outsource' our U9 tool, i.e., get HEASARC to equip its Web server system with our XML server.
RS - In the database group's perspective, U9 would be a front end. Is Qt the graphics package Qt or something else?
DP - Qt the graphics package also has built in Web server and XML server support, and you name it.
[Toby was unable to attend the meeting.]
SD - A good observation simulator is needed. [Not sure why I seemed to have emitted this motherhood statement]
DP - Dave Davis and I are going to look into creating an EGRET D2, translating the pointing/livetime information into D2 format.
SD - I think this is a good thing [meaning making EGRET data accessible in the LAT analysis environment].
HK - [Minutes from the UI meeting on Jan. 30 are linked to the agenda page.] At the last UI meeting we covered 2 topics - plotting package evaluation and concept of the standard analysis environment. We decided to table looking at ROOT Python until the installation becomes stable. This does not affect the evaluation of ROOT as possible graphics/GUI tool. The list of graphics packages under consideration has been narrowed to ROOT and plplot, with DS9 also in there for astronomical image display. Jim Chiang provided a use case for contour plotting, and others are solicited. We will use this and others to work out what the packages can and can't do for us.
The SSC-LAT working group asked us to develop recommendations for what the SAE should look like. Jim produced an outline of what he thought the system should look like. We will circulate the draft recommendations of the group - first to the UI mailing list - until it converges, then send it to the SSC-LAT working group.
JC - Nothing to add
HK - ROOT Python is a Python extension that lets you read in ROOT files, etc. from a Python command line. For now, in terms of evaluating it as a plotting package, we will be looking just at ROOT, not ROOT Python.
DB - If we end up using ROOT for graphics and ROOT-Python becomes more mature, can we build on what we've done using ROOT by itself?
HK - It depends on how it is done. If we have written native C++ code to access ROOT routines, then there would be some amount of work to get it done in Python - definitely not for free.
RS - We talked about wrapping up the graphics package to insulate it (whatever it turns out to be) from the analysis packages. Allowing access to the ROOT classes from the Python interface would be nice, but ROOT-Python is not critical to what we want to do.
SD - Is contour plotting ability [i.e., the inability by ROOT to use specified contour levels] a potential show stopper?
JC - I don't know that ROOT can't do this - at least not without monkeying around to make it do what you want. [Jim later reported that ROOT can in fact accept specified contour levels.]
[See link from agenda page.] We haven't gone into the question of reusing software on a specific, tool-by-tool, basis. How existing software will be incorporated is still an open issue.
DB - Regarding the labor available for development of science tools, I will send an update to Seth this week.
SD - I have received input (not yet final) from Gino Tosti regarding people at INFN. I should be able to give a fairly complete picture at our next meeting.