Privacy and Security Notice

Prune method Time resolutions using the prune method (04/24/05)
My Analyzer and replay/PARAM/online04/*tracking.param files were interactively merged with the CVS branch online04 on 04/21. 

These updated files were then compared to the files in the directory /group/c-jpsi/Analyzer, the differences were examined and copied into my directory.  The result is that my files are now the same as Peter's.  The files that were compared were:
 

   /group/c-jpsi/replay/PARAM/online04/htracking.param
   /group/c-jpsi/replay/PARAM/online04/stracking.param

   /group/c-jpsi/Analyzer/HTRACKING/h_select_best_track.f
   /group/c-jpsi/Analyzer/HTRACKING/h_select_best_track_prune.f
   /group/c-jpsi/Analyzer/STRACKING/s_select_best_track.f
   /group/c-jpsi/Analyzer/STRACKING/s_select_best_track_prune.f

   /group/c-jpsi/Analyzer/INCLUDE/hms_physics_sing.cmn
   /group/c-jpsi/Analyzer/INCLUDE/sos_physics_sing.cmn

   /group/c-jpsi/Analyzer/HTRACKING/Makefile.Unix
   /group/c-jpsi/Analyzer/STRACKING/Makefile.Unix


Peter said that a time tolerance of 3ns is sufficient for looking at just e-pi coincidences.  The time tolerance is described in h_tof.f.  The time tolerance is slightly larger than the expected width of the histogram (fptime on the x-axis) of the fptimes as determined by each PMT.  The Analyzer throws out PMT hits that are outside of this width.  The center of the cut is determined by the 3ns bin with the largest number of PMT hits.  The PMT's that do not pass this cut are thrown out when calculating beta.  A larger time tolerance would be needed if e-p or e-K coincidences needed to be kept.

Here are the results for beta compared to pass1 (HMS using scintillators, SOS using chi-squared) for run 49540. 
 
Time resolutions, pass1 vs. the prune method

pass1 pass1a_(prune)
HMS sigma(ns) 0.0331
 0.0334
SOS sigma (ns)
 0.0278
 0.0279
cointime sigma (ns)
 0.200
 0.204

The time resolutions are slightly worse with the new prune method, this was seen with run 49549 also.  The differences, however, are not significant.

The improvement in the the reconstructed hsbeta and ssbeta is readily seen below (run 49540).  There are less events below beta=1. 
HSBETA
pass1

 HSBETA
pass1a_(prune)
SSBETA
pass1

 SSBETA
pass1a_(prune)