Each person in the group check out GSIM, make a trivial change, rebuild and run. Communicate any problems to the rest of the group and the GSIM mailing list.(everyone, target deadline 3/13/98).
Get high-quality `teleconferencing' working to facilitate remote users participation in meetings. (W. Brooks, target deadline 3/13/98; COMPLETED 3/10/98]
Speed GSIM up by a factor of 10 -100
Run time accounting utility on GSIM. Perhaps on several platforms. Use the results to choose the next logical step. (WB, KJ, DR target deadline 3/11/98; COMPLETED 3/9/98)
Systematically study the effects (time, accuracy) of turning off each volume type.
Systematically study the effects (time, accuracy) of turning off secondaries in each volume.
Study the time and amount of data per event for electrons as a function of phi, theta (looking at the effect of the coils and the downstream support structures.
Generate a set of benchmark input files. Electrons in fiducial volume, CELEG events, electrons uniform in phi and theta.
Do a systematic study of the effect of GSIM cuts and other parameters on speed. Document this study for future use. Define reasonable "sets" of parameters for standardized use.
Get time-based tracking to work on GSIM events
Understand the time offset constants used in Recsis, transfer to GSIM by the same mechanism Recsis uses (LF, target deadline 4/1/98? already done by Burin? checked in?)
Invert the time-to-distance relation in Recsis, install in GSIM, such that when the Recsis function changes, the GSIM function changes. (KJ, AS, target deadline 4/1/98)
Use only one unix platform, two, or more? Which platform(s)? Can develop on one and run on another? Fastest platform may be Linux/intel and Linux/alpha.
MIT compute farm. Can we get/do we need accounts on this? Can we get months/years of CPU time on this farm? Is porting GSIM to this platform a small job or a big one?
If we do large-scale calculations off-site, how do we transport the simulated data here?
How fast do we need GSIM to be? (and how accurate?)
TOF times for time-based tracking - changes needed? e.g., map calibration constants
Do we need a `physics' monte carlo bank, which would store information such as partial wave content, etc.?