Minutes of the muon software meeting 15 Oct 03, taken by Victoria
------------------------------------------------------------------

We had two talks from Chris Hays and Alexei Varganov discussing cosmic
rejection in Z->mu+mu- and W->mu nu analysis respectively.  Eric James
also presented some work that has taken place in the EWK cross section
working group.  The conclusions of the studies are the same, so there here
is a summary:
They looked at two different ways of ways of rejecting cosmics:
(1) Using the standard cosmic tagger
(2) Using only the dicosmic COT fit, which is part of the standard tagger

They concluded that the standard cosmic tagger is rejecting some signal
events.  By looking at various kinematic distributions, the dicosmic COT
fit seems to reject most of the cosmic background.  So both the Z->mu+mu-
and W->mu nu analysis have decided to use the dicosmic COT fit to identify
and remove cosmic event. Initial studies suggest that the remaining cosmic
background remaining in the samples is less than 1%. By looking at W->e nu
events it seems that the efficiency dicosmic COT fit is over 99%.

Everyone agreed that we should use the result of the COT dicosmic fit as
the only tag to reject cosmics.  We are now going to recommend this to the
physics groups.

We still have to decide how to implement this policy.  The easiest thing
is to change the hasCosmicRay() bit of the cosmic tagger.  This would
leave all of the original information in the CosmicObject, in case other
users still want to use it.  The one disadvantage to this solution is that
what is tagged as a cosmic by the cosmic tagger depends on the offline
version that is being used.


Alexei also suggested that we might want to tune the cut on the chisq returned
by COT dicosmic fit.