Minutes of the muon software meeting, 27 Oct 04
================================================

1. Track chisq cut for MC.  
============================ 
Else and Victoria see a different distribution for track chisq, when using
the same zewk3m data sample. This is under continued investigation.

2. Cosmic tagger cuts.   
=======================

Summary of our discussion.  Comments from Pasha, Slava, Anyes, Chris.
 
Two problems have become apparent in the past few weeks:  (1) The cut on
the chisq of the di-cosmic (helix) fit is too loose for the tracking
resolution in 5.3.  (2) The cosmic tagger is not tagging events a low-pt.
This is best illustrated by a sample from Angela Wyatt, who is looking for
diffractive J/psi events.  The signal is typified by 2.5 GeV muons plus
photons.  From a sample of 27 events she sent us, 26 were cosmics, but
none were tagged as such.

We have already changed the cut on the chisq of the di-cosmic fit from 300
to 1000.  We discussed what other changes to make.  We decided to change
the cuts to identify a seed track to >=5 hits on >=2 axial and >=2 stereo
tracks (same as defTracks in 5.3).  We also decided to put the minimum pt
required to identify the seed track in the tcl file so users who
want/need to change change it for their analysis.

Reducing the pt required to identify the seed track to 1.4GeV may
increase the over-efficiency of the tagger, so it is something we don't
want to do by default.  Users of low-pt analysis where the cosmic
background is large will have to tune the cosmic rejection themselves.

Two other issues, that we should be addressed in the long term:
i. At low-pt enemy loss effects become important in the tracking: this
should be taken into account in the di-cosmic fit.
ii. At low-pt the road width to find the cosmic ray is not wide enough.


3.  Pasha: Simulation of the energy associated with the muon
==============================================================

Pasha showed that the simulation of the hadronic energy associated to CMX
muons is not well described at |x_ces|>23 cm.  This why the distributions
of energy for CMX muons is not well described in the MC.  There is no such
problem with CMUP muons.


4.  Victoria: Updates to the standard muon cuts
=================================================

We agreed to remove the cut on minimum energy for CMUP and CMX muons.
The spike at Eem+Ehad=0 tends to come from CMX muons that are not fiducial
in the CEM.  

For stubless muons (CMIO) a cut at Eem+Ehad>0 reduces the fake rate
significantly and is still a good cut.  Pasha and Chris disagreed what to
do about stubless muons that go through a calorimeter crack (eg fidEle=0),
where the energy cannot be reconstructed.  Pasha said they should be
rejected, Chris would like to keep them for more acceptance.

We still don't understand the |dxCMU|>3cm muons.  So we've decided to keep
the cut a |dxCMU|<3cm.  If the problem is easily fixed then we could run
it when we remake our ntuples.

[After the meeting we looked at these muons, and we are still trying to
work out exactly what the problem is.]