Participating:
Rene Bellwied (Wayne State)
Tony Johnson (SLAC - calling from CERN)
Dean Karlen (Carleton)
Keith Riles (Michigan)
Mike Ronan (LBNL)
Bruce Schumm (UCSC)
Ron Settles (MPI, Munich)
Rick Van Kooten (Indiana)
Wolfgang Walkowiak (UCSC - calling from SLAC)
Introduction (Riles)
Purposes of meeting:
This was a brief update of a status report given at the November teleconference. Because much of the linear collider simulation crew at SLAC (R. Dubois, T. Waite, J. Bogart) have been spending more and more time on the GLAST project, progress on linear collider simulation infrastructure has slowed. It is hoped to hire an experienced person in the very near future to help fill the gap.
Simulation software demonstrations were given recently to interested physicists at Fermilab and CESR. Fermilab physicists are particularly interested in the fast monte carlo simulation.
G. Bower has recently generated new Monte Carlo samples of various processes using the old S1 and L1 detector designes, but with a large number of bug fixes. A web page describes available data samples.
New formats have been defined for both the parameter input files used to define the detector (they are now in a self-documenting format called XML) and the generator output files (now in SIO format). J. Bogart has implemented the XML format, and T. Waite the SIO format. The simulation code can now read XML parameter files (which replace the old .ini files),but the reconstruction code cannot yet do so. The XML files describing the new S2 and L2 detectors are nearly ready, but Joanne is waiting for persons to provide certain (non-tracking) S2 detector parameters.
A minor but useful change in the simulation is parametrization of detector response by polar angle theta instead of by cos(theta). JAS is being upgraded to support the same segmentation.
The effort to incorporate B. Schumm's 5-parameter correlated error matrices has been temporarily sidetracked, but a new student joining the LCD effort will be working on this.
R. Cassell is working on hit merging and standardized diagnostic histograms for reconstruction. He has finished providing such histograms for the simulation. Track reconstruction for the endcaps is farther down his tasklist.
G. Bower is working on background overlays.
T. Johnson is working on a new event display for JAS based on CERN's WIRED program. He has a first version running and hopes to release it soon. It has been planned for a long time to switch from GIZMO to GEANT4 when it became available. From discussions with LHC physicists at CERN, it's not clear that GEANT4 will truly be ready for prime time soon.
The next releases of the simulation and reconstruction code should be ready before the Berkeley meeting at the end of March, but probably not much before.
M. Ronan mentioned that he has written new JAS code for unravelling the sharing of both calorimeter and tracking hits, allowing all hits associated with a true MC particle to be displayed. He has put the code in the JAS contributed plug-in area, and T. Johnson will look into incorporating it.
Status of new postdoc hires (Bellwied, Van Kooten, Riles)
Web pages and plots were made available before the meeting. The track reconstruction performance of LCD for the small detector has been tested with a sample of pythia qqbar events, containing approximately 100k charged particles. The analysis proceeded in several stages. A very loose set of cuts on MC truth parameters reveals large inefficiencies in track reconstruction, much of which can be attributed to particles with very low transverse momenta w.r.t. the beam and to particles entering the endcap tracker, for which no track reconstruction yet exists. Tightening the cuts on pt and cos(theta) gives better efficiency, as expected, but there seems to be a sharp cutoff in pt at about 140 MeV, which surprised M. Ronan who expected higher efficiency there. He wondered whether the results were being skewed by secondary particles emitted in interactions of primary particles with parts of the detector. Photon conversions and inelastic pion interactions were suggested as possible sources that should be filtered out. Also, even with strict cuts [pt (>500 MeV) and |cos(theta)| <0.7], the overall track efficiency was only about 80%, again inconsistent with what M. Ronan and N. Sinev have seen in the past. The MC truth information will be looked at more closely and the analysis repeated to excluded identified secondaries.
Momentum resolution was also studied for the sample of tracks passing strict cuts. Reconstructed momenta were compared with true values and the residuals fitted to a sum of two Gaussians. The resolutions obtained were actually better than expected at high momenta. After some discussion, it appeared that the problem was that hit smearing had not been turned on before reconstruction (hit smearing is not carried out by default). Some other technical problems arose, including a small fraction of tracks that seemed to be associated with MC particles travelling in the opposite direction. T. Johnson suggested trying a sample of the new events generated by G. Bower (see above). KR remarked that this analysis is a good start on a systematic evaluation of track reconstruction performance. Useful infrastructure for evaluation is now in place, and with various misunderstandings now cleared up by this discussion, an updated presentation at the Berkeley meeting should be quite illuminating.
Preparations for Berkeley Meeting (All)
KR listed the presentations requested so far for the tracking session
at Berkeley:
Introduction (K. Riles / D. Karlen)
Global tracking in ECFA design (K. Moenig)
The ECFA TPC Design (R. Settles)
The S2 and L2 Design Parameters (B. Schumm)
LCD Track Reconstruction performance (W. Walkowiak)
R. Van Kooten stated he hoped to speak on intermediate tracking issues, depending on progress by his graduate student before the meeting.
M. Ronan volunteered to speak on the present track reconstruction.
A.O.B. (All)