Imperfections

Here is a list of known problems with the code.

August release:

• unix port

  • HepEvt.cxx unix version modified to access /HEPEVT/ as hepevt_ instead of the NT version HEPEVT

  • arve/Warning.h has namespace util disabled

  • CLHEP AList routines removed due to compilation problems

  • generator wrappers in Gary's personal area

• No simulation of beamsstrahlung

• particle extrapolation

  • dE/dx (may not be a problem!)

    • no smearing

    • uses Bethe-Blocke formula. No fancy stuff like GEANT3

  • does not keep track of time. This is handled by Gismo, but we do not output it. Should be no problem.

  • GISMO cannot handle a particle fired straight up the y axis!

  • neutrons could be given a higher low energy cutoff. They rattle around for a long time in the calorimeter. GEANT's cutoff is 10 MeV; GISMO uses 0.5.

  • the MCParticle status word is not very rich in its descriptive ability. It is hard to impossible to tell from it which particles are from the generator and which created by GISMO for example.

  • Gheisha periodically hits a production limit of 60 particles in an interaction; noticed in the W-pair task.

  • see some error messages from EGS as well. These W-pair events are taxing on the system!

• Geometry restrictions

  • cylinders' layers must be evenly spaced. Cannot set z vs r.

  • cannot begin and end with an active layer (ie cells are all the same - eg first cell ends with air gap, last cell ends with air gap).

• Shower particle/generation problems

  • shower particles that leak back into the central detector and leave hits in the tracker/VXD will have their MC parent listed as 0.

  • it appears that some very complex final states put out particles more than 10 generations old. This may not be so hard for isajet, which wastes 3 generations with self-copies. GISMO deletes all daughters after 10 generations, so some of these final staters are deleted!

• Tracking

  • the track charge is always inverted wrt MCParticle! Need to invert logic on line 50 of ReconTrack.cxx when we're done producing events pre-Keystone.

  • for electrons, which seem to brem freely, multiple 'tracks' per particle can ensue. GISMO terminates the parent particle and it resumes as its own daughter (plus the gamma). If it bremmed in time for the 'daughter' to leave 5 hits in the tracking system, a new track will be created.

  • track smearing and selection parameters are hard-wired into the code

  • track smearing is done at the write() stage, not execute. Due to a limitation in the way DataControl is done. Should be moved to execute in due course.

• Calorimetry

  • is tanl an appropriate angle to tag towers with? Should we got to a more q-specfic angle?

  • Hit Drawing

    • towers assume the reference geometry, not the user specified one!

    • HAD endcap hits are translated to the back of the calorimeter (ie the starting z is z(outer), not z(inner))

R.Dubois Last Modified: 01/13/04 12:37