[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
GlueX Software meeting minutes
Hi All,
I've written up minutes of yesterday's software meeting and posted
them on the web at:
http://www.jlab.org/~davidl/HallD/Minutes/Software_2006_07_05
I've also included them below.
Regards,
-David
=================================================
In attendance:
David L., Elliott W. Simon T., Mark I., Dan C., Matt B. Matt S., Richard J.,
Nikolay, K., Zisis P., Rafael H., Blake
NOTE: We had a problem with VRVS at JLab and were only able to do
audio through my laptop. It caused a terrible echo and make discussions
a bit difficult. Also, my laptop battery ran out towards the end of the
meeting, cutting us off from the remote participants.
The meeting agenda was posted on the Wiki at:
http://www.jlab.org/Hall-D/software/wiki/index.php/July_5%2C2006
Agenda:
1 Lehman review comments
2 CDC and FDC acceptance via HDGeant
3 PAC 30 Computing Requirements (added during meeting)
4 Coding conventions
5 DANA -> JANA update
6 AOT
============================================
Lehman review comments
-----------------------
"FINDINGS", "COMMENTS", and "RECOMMENDATIONS" pertaining to Hall-D
software from the closeout of the recent Lehman review of the
12GeV project were discussed. The main idea of the text was that
we need to develop a more sophsiticated simulation/reconstruction
package so that studies can be made to test the detector design
on the physics we hope to see with it *prior* to establishing the
baseline in CD-2 next summer.
There was general agreement with the suggestion and that this needs
to be done. The main problem continues to be manpower for software.
Mark Ito suggested that perhaps breaking the work up into well
defined pieces that require less time could help drum up volunteers.
David agreed to try and use the Lehman committee comments to apply
more pressure on the collaboration to devote manpower to software
in the short term.
CDC and FDC acceptance via HDGeant
-----------------------
http://www.jlab.org/~davidl/HallD/Presentations/2006_07_03_acceptance
David showed some plots indicating the acceptance of single charged
tracks from the FDC and CDC. Several plots were shown indicating the
acceptance for varying radii of the FDC. For the exercise, a single
set of data Monte Carlo data was generated using a mu+ and mu- with
randomly thrown momenta and angles per event. Tracks were "accepted"
if at least 8 hits were obtained in the chambers. It was noted that
this differs from the acceptance studies presented in the design
report where 4 hits per track was used. The design report study also
used a method where entire events were either
accepted or rejected based on all particles (including neutrals) being
detected.
The current study indicates that combining information from the CDC
and FDC will lead to full coverage of the kinematic range (i.e. no holes
all the way down to an active FDC radius of 37cm (note that 61cm is
what is currently in the Monte Carlo).Without full tracking, however,
the affect on resolution cannot be measured.
The main message here was that it does not appear that acceptance
will be affected by reducing the active area of the FDC by a few
centimeters at the outer edge.
PAC 30 Computing Requirements (added during meeting)
-----------------------
Richard Jones asked that this be added to the agenda during the
meeting because he had some comments. Note that the starting point of
the discussion was the "Computing Requirements List" form that must
be filled out with each submission to the PAC. A first draft of the
form and a README document explaining how the numbers were obtained
was posted by David a few days earlier at:
http://dustbunny.physics.indiana.edu/~dzierba/pac_stuff/computing/
There was discussion about the simulation compute power number. David
noted in the documentation that when he tested the event processing rate
it appeared there was a factor of 4 descrepancy between a recent
test he did and ones others did over a year ago. Richard stated
that he believed the original numbers were more reliable and suggested
we use them. David replied that he was willing to repeat the tests
on other machines, but if the descrepancy remained, he would rather
go with the higher number.
There was discussion about the online disk storage number David
originally put at 1TB. David noted that Ed Brash had already
contacted him via e-mail and suggested 10TB instead. Richard
suggested an even higher number and after some discussion, a
value of 50TB was agreed upon.
The imported and exported data values were also discussed. Richard
proposed a model where simulated data were generated at outside
instituions, but transferred to JLab for permenant storage in the
silo and reconstruction on the JLab farm. He also noted that the
simulation may need 10 times as many physics events, but that
the fraction of raw data events that contain interesting physics
will be small. After some discussion a value of 200TB for the
imported data and 100TB for the exported data were agreed to.
Richard agreed to write up a short explanation of how these numbers
were calculated and David agreed to update the README file to
include them.
***************************************************
At this point the laptop battery ran out and we lost communication
with outside participants
***************************************************
Coding conventions
-----------------------
David showed a preliminary list of coding conventions he put up
on the Wiki at:
http://www.jlab.org/Hall-D/software/wiki/index.php/Coding_Conventions
There was not a lot of discussion on here as it was getting late
and we lost half of the meetings participants. David invited
everyone to have a look at them on line and use the "discussion"
feature built into each Wiki page to comment on them and suggest
changes or new conventions.
DANA -> JANA update
-----------------------
David reported that work is progressing rapidly on the change
from DANA to JANA(i.e. from changing our framework from being
Hall-D specific to one that can be jointly used and worked on
by halls B and D. Several design changes are being implemented
that improve the extensibility of the framework. The work is
being done on a separate branch in the repository
(branches/jana-branch) so as not to disrupt use of the main trunk.
From the Wiki page, changes include:
- Introduction of JEventSourceGenerator and JFactoryGenerator
classes allow Hall-D specific code to go into DEventSourceHDDMGenerator
and DFactoryGenerator subclasses.
-Extract_HDDM() methods moved from factories to DEventSourceHDDM
class. This makes a more extensible system that will also make
it easier to implement DEventSourceEVIO and anything Hall-B chooses
to implement.
-Header files moved from src/libraries/include to
src/libraries/PACKAGE. Thus, headers from outside packages use
package name. e.g.
#include "CDC/DCDCHit.h"
-Factory naming scheme changed.
-Geometry system still needs to be sorted out.
--
------------------------------------------------------------------------
David Lawrence Ph.D.
Staff Scientist Office: (757)269-5567 [[[ [ [ [
Jefferson Lab Pager: (757)584-5567 [ [ [ [ [ [
http://www.jlab.org/~davidl davidl@jlab.org [[[ [[ [[ [[[
------------------------------------------------------------------------