Present: Burin Asavapibhop, Will Brooks, Laurent Farhi, Maurik Holtrop
(via speaker phone), Kyungseon Joo, Mac Mestayer, David Rowntree (via
speaker phone)
Initial Agenda (as distributed):
ITEM 1 11:00-11:05 Administrative - revise agenda, revise future
meeting times, etc.
ITEM 2 11:05-11:20 Orientation and presentation of project list - W.
Brooks
ITEM 3 11:20-11:50 Discussion of scope of group and project list - All
ITEM 4 11:50-12:00 Update on time-based tracking of GSIM data -
Laurent/Burin
ITEM 5 12:00-12:30 Discussion of how to handle drift chamber
time-to-distance in GSIM
ITEM 6 12:30 Adjourn
Because of several 'technical difficulties' and people
leaving/entering the meeting at a variety of times, the meeting did
not flow quite according to the above agenda. (I am confident we can
improve this.)
-------------------------------------------------------------------------------
ITEM 1 - REGULAR MEETING TIME and ADMINISTRATIVE ISSUES - All
The regular meeting time of Thursdays 1:30-3:00 seems to be OK for
everyone I've heard from so far. Anyone wanting to be included in the
conference call should let me know well in advance. Check that I have
your correct phone number at the bottom of this page.
The clas_gsim mailing list seems to have quit working in recent
weeks. I got Dave Buckle to revive it, and also to begin archiving
it. The archive is now available at:
http://www.cebaf.gov/ccc/hypermail_archives/CLAS/CLAS_GSIM/
Bill Bertozzi has given me an account on the MIT DEC/ALPHA compute
farm, which this group plans to be using for number crunching (among
other CPU's) All of the MIT-related people have accounts on this farm
also. If there is a demonstrable need, others can get accounts also.
GSIM was ported to this platform months ago by Maurik, and it has
recently been run by David Rowntree there.
A Linux/Alpha machine at Ohio U is also available for use from Allena
Opper. She is presently working on the problem of CERNLIB availability
on this platform.
I would like to hear about other CPU farms on which we can get some
significant running time!
ITEM 2 - ORIENTATION AND PROJECT LIST - W. Brooks
The two orientation slides I have been showing and continually revising
are
temporarily located at:
http://www.cebaf.gov/~brooksw/gsim/GSIM_focus_group_1.html
(orientation)
http://www.cebaf.gov/~brooksw/gsim/GSIM_focus_group_2.html (project
list)
One project I'm working on is to obtain a world-accessible web site
which we can use. Currently the global GSIM web site as maintained by
Maurik Holtrop is:
http://einstein.sr.unh.edu/Maurik/gsim_info.shtml
These will be integrated together and made universally accessible
soon.
Among the projects listed, the most urgent ones were identified as:
1) Generate a set of benchmark GSIM input files and initial FFREAD
files to do systematic speed tests. (Will Brooks, target deadline
19 March 1998).
2) Study the time spent generating showers in the coils compared to
time spent on particles within the fiducial volume (Will Brooks,
David Rowntree, target deadline 26 March 1998).
3) Systematically study the effects (time, accuracy) of turning off
each volume type (or modifying GEANT parameters for various
volumes), (no volunteers yet, maybe Kyungseon Joo can work on this
starting in about 1 week, need to have #1 finished first).
Several other projects are listed on my second web page.
During the discussion it became evident that the group will need some
large, centralized disk space. I will attempt to locate this resource
at JLAB. It could also in principle be located off-site, if it were
sufficiently accessible to CLAS collaborators. During the CALCOM
meeting, Dieter Cords presented a scheme to organize the /work disk
area, and I requested to have 20 Gbytes set aside for GSIM work.
Another project not discussed was to attempt to generate 'accepted'
MCTK and MCVX banks as a project to generate zeroth-level acceptance
functions for Volker to apply to his events. (W. Brooks, target
deadline 25 March 1998, other collaborators welcome!)
David Rowntree suggested that he had observed unreasonably small step
sizes being assigned in some volumes. He will a) review the step size
options in GEANT and b) search the GSIM code to assess what we are
presently doing. Burin Asavapibhop noted that he had experimented with
the parameter which governs the step size in GSIM, and had gotten a
significant speedup by changing this from 10,000 to 5,000. Smaller
numbers produced problems. (He was also using AUTO 0, while we should
probably stick to AUTO 1.) Maurik reminded the group that the two
factors governing the step size choice are: accuracy of swimming
through a magnetic field, and accuracy of energy deposit and multiple
scattering (i.e., need small steps in dense materials). Maurik
suggested that it may be beneficial to see if we can improve the
swimming code in GSIM to be more efficient. I pointed out that in
recent profiling which David R. and I did, that snake takes the
second-largest amount of cpu time in the program (depends on input
data) so that this may be a fruitful direction to pursue. (David
Rowntree, target deadline 19 March 1998)
ITEM 3 - DISCUSSION OF SCOPE OF GROUP AND PROJECT LIST - All
This discussion was intermingled the other agenda items. For those who
are interested in the scope of the GSIM Focus Group, see my web pages
listed above.
ITEM 4 - UPDATE ON TIME-BASED TRACKING OF GSIM DATA - Laurent/Burin
Time-based tracking now works on GSIM data!!!
Burin Asavapibhop spent a couple of months studying Recsis and GSIM to
determine why the TBT was not working on GSIM data, and changed his
private version of GSIM and recsis so that it works. A discussion
between Burin, Laurent, Joe, and Will resulted in the following plan:
Joe will implement Burin's changes into the map-manager for run #1
(simulated
data) and will make any required changes in recsis/a1c; Burin and
Laurent will test the result on Burin's private version of GSIM and
Recsis, using Joe's changes; once that works, Laurent will check in the
required changes to GSIM and will test them.
What Burin found (not a great surprise) was that there was an
incompatibility between the GSIM times and what Recsis was
expecting. In particular, a 1850 nanosecond time offset had been used
in GSIM to approximate the trigger time. Burin modified DC_digi and
SC_digi in GSIM, and trk_dtime in Recsis.
Burin showed TBT results from GSIM data, demonstrating that the code
does work. However, the mass resolution (sigma) for neutrons in the
reaction ( gamma-proton -> pi+ pi0 neutron) was seen to be 77 MeV,
much worse that what we see in the data. The group guessed that this
is because GSIM used one x-vs-t relation and Recsis used another (see
the discussion below). Laurent will re-run this data using the correct
relation to see if that fixes the resolution problem.
ITEM 5 - HOW TO HANDLE DRIFT-TIME-TO-DISTANCE RELATION IN GSIM
Mac Mestayer came for this discussion. Summary of discussion and
decisions:
We will use the simple linear drift-time-to-distance ('x-vs-t')
relationship in GSIM (which is presently in use) and also in RECSIS
(via an existing tcl variable option).
We will construct a resolution function to smear the time
data (not the position data). We did not discuss whether this means
tdc's or time in nanoseconds (DC0/DC1/DC2).
For the moment we will *ignore* the fine details affecting x-vs-t, such
as the magnetic field strength and direction, and use an averaged
x-vs-t smearing. This allows the smearing to be done in a *separate
program* which is run *after* the GSIM processing. This means we can
generate the simulation once, and re-smear it quickly several times if
needed. Maurik's concept for this is to use a program named GSIMKO
('GSIM Knock-Out') to perform this function.
The future capability of putting in detailed x-vs-t directly into GSIM
for instrumentation studies may be important, but will not be
addressed by the GSIM Focus Group (incompatible with its mission
statement).
ITEM 6 - ADJOURN
Took place at 1:30 - on-schedule!
-------------------------------------------------------------------------------
GSIM Focus Group Members and other GSIM contacts:
Burin Asavapibhop (U. Massachusetts, on-site at JLAB) 757-269-7322
Will Brooks (JLAB) 757-269-6391 brooksw@cebaf.gov (core group, GSIM
Focus Group primary contact)
Volker Burkert (JLAB) 757-269-7540 burkert@cebaf.gov
Bryan Carnahan (Catholic U) phone? 08carnahan@cua.edu
Larry Dennis (Florida State U) 850-644-1804 larry@fsulcd.physics.fsu.edu
Steve Dytman (U. Pittsburgh) 412-624-9244 dytman@vms.cis.pitt.edu
Laurent Farhi (Saclay, on-site at JLAB) 757-269-7680 farhi@cebaf.gov
(core group)
Rob Feuerbach (Carnegie-Mellon U.) 412-268-2749(CMU#), 757-269-?(JLAB#)
feuerbac@ernest.phys.cmu.edu
Maurik Holtrop (UNH) 603-862-2019 holtrop@cebaf.gov (GSIM coordination)
Kyungseon Joo (UVA, on-site at JLAB) 757-269-5307 kjoo@cebaf.gov (core
group)
Joe Manak (JLAB) 757-269-5829 manak@cebaf.gov
Si Mcaleer (Florida State U) phone? mcaleer@cebaf.gov
Allena Opper (Ohio U) 747-593-1982 opper@akopper.phy.ohiou.edu
David Rowntree (MIT) 617-258-5442 tree@mitlns.mit.edu (core group)
Dave Tedeschi (U. South Carolina) 803-777-1132 tedeschi@sc.edu
Dennis Weygand (JLAB) 757-269-5926 weygand@cebaf.gov (GSIM coordination)
Jianguo Zhao (MIT) 617-258-5438 jzhao@mitlns.mit.edu (core group)