[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Getting started with DANA
Hi All,
I just uploaded the "Getting Started With DANA" note to the DocDB
(GlueX-doc-516). This has minor modifications to the first draft that
was circulated to a few of you last week. Please have a look and send me
any feedback.
I also want to take a minute to discuss what is needed to implement a
plan discussed at the collaboration meeting. Note first that I have
written a few notes on the software discussions that took place during
the collaboration meeting which I've uploaded to the DocDB as
GlueX-doc-517. I've also included them at the bottom of this e-mail for
convienience.
At the meeting, a list of subsystems was shown along with the names of
the persons who are responsible for the first stage reconstruction. Here
is the list in its current form:
Detector Subsystems
----------------------------
UPV Alexander O.
BCAL Chucheng X.
CDC Matt B.
FDC Ed B.
TOF Ryan M.
TAGGER Richard J.
FCAL Scott T.
.
Tracking
------------
Pattern recognition / Quick Fit David L.
B-field map / stepping Werner B.
Track/Shower Linking
Full fit (Kalman?)
Here is what is needed from each of the people listed under detector
subsystems:
1. Read GlueX-doc-516 and get familiar with DANA
2. Make sure a factory exists for copying your subsystems' hits from
the hddm_s stucture into a set of objects. These already exist for many
subsystems. (See for example, src/libraries/CDC/DFactory_DCDCHit.cc).
Note that the convention of HDDM is to use the word "Hit" for digitized
detector responses while the word "Point" refers to a cheat tag.
Documentation of the s_HDDM_t structure can be found here:
http://www.jlab.org/Hall-D/software/HDSoftware_Documentation/structs___h_d_d_m__t.html
3. Make another factory which uses the hit objects to do the first level
reconstruction. This needs to focus more on speed than on accuracy. The
output of this factory will be used as input to the pattern recognition
code for tracking. It can also be used as input to a more refined
reconstruction (if possible) in another factory.
Things to keep in mind while writing your factory:
- Even though the natural coordinate system for the detector is
cylindrical, it turns out that most of the pattern recognition code
written right now uses cartesian. If there isn't a clear advantage to
presenting the positions of the hits, use x,y,z. (Don't use both since
this will lead to exactly the kind of redundant data problems Curtis
talked about.)
- Use the repository. Even if your code isn't functioning perfectly, it
will help others (like me) start work by just knowing the format of your
objects. It will also allow others to help pinpoint problems you may
come across related to the framework/build system etc.
- You will need to test that your new factory is working, most likely by
modifying hd_ana. Please delete the "CVS" directory in your hd_ana
directory first so you don't accidentally commit changes to it.
- If you can't spend a lot of time on this now, at least spend some time
thinking about the data that your factory will produce. Try running the
mkfactory script as described in the Hall-D note and committing the
header files to the repository. To meet our goals, I think we should
have some rudimentary factories in place for each of the subsystems by
sometime in June.
Please contact me with any questions.
Regards,
-David
---------------------------------------------------------------------------------
GlueX Collaboration Meeting May 12-14 Software Minutes
David Lawrence
These are some notes from the software discussion at the
GlueX collaboration meeting on May 12th, 2005 at JLab.
The discussion started with a suggestion by David L. that
each institution either provide or identify a person to
produce the first stage analysis factories which can be
used as input to the pattern recognition. The discussion
quickly fell into how there is a lack of (wo)manpower
which is currently un-utilized. Here are some points that
came from the discussion.
1. Ed Brash will begin working on the FDC reconstruction.
2. A tracking group will be formed. Initial members are:
David Lawrence
Curtis Meyer
Ed Brash
Matt Shepherd
3. David Lawrence will organize a workshop in late fall 2005
to discuss specific questions that the software should answer
and how to go about doing it. This should be proceeded by a
series of conference calls in which the issues are exposed.
4. Werner Boeglin will discuss with his FIU collegues about
FIU taking on a software responsibility related to the tracking.
Alex suggested David and Ed make a trip to FIU to discuss it with
them in person.
----------------------------------------------------------
Here are notes from the Software Summary discussion on the
last day of the collaboration meeting.
1. Add the tagger to the list of subsystems in the list
presented with Richard Jones' name next to it.
2. Hold fall workshop at FIU. Invite a couple of outside
experts. Matt S. will help identify the right person from
CLEO and Alex will contact Bill Dunwoodie to help identify
a second expert.
3. Contact Stefan Spanier about providing the Cherenkov
code.
4. Alex suggested Ryan M. help coordinate the coherence
of the subsystems. I'm not sure what this means in practical
terms, but Ryan agreed so I assume he does.
5. It was suggested (and generally agreed upon) that a set
of working groups be formed within the software group who
will meet weekly. Details of this still need to be worked
out. These would replace the weekly software meetings to
allow folks to attend meetings focused more specifically
on their interests. The mechanics of this was not 100% clear
to me at the meeting. We need to make sure we don't
"over-meeting" ourselves here.
6. Alex will coordinate a group to define the set of
gold-plated reactions that the reconstruction should
be targeted towards.
7 .Shoot for doing tests of feeding fully reconstructed data
into IU PWA code by mid-fall 2005.
--
------------------------------------------------------------------------
David Lawrence Ph.D.
Staff Scientist Office: (757)269-5567 [[[ [ [ [
Jefferson Lab Pager: (757)584-5567 [ [ [ [ [ [
http://www.jlab.org/~davidl davidl@jlab.org [[[ [[ [[ [[[
------------------------------------------------------------------------