The assignment of cut records in PTrack propagation
David Adams
14jul00 1800

Cut records (class CutRecord) keep track of the variables upon which
cuts are applied during track finding.  Here we describe when these
records are created and filled in the method PTrack::propagate(...)
which is the heart of TRF++ track finding.

This method takes a single PTrack and returns a list of new PTrack's
which are extensions of the original track.

There are three stages of propagation and cut records are created
each stage.  The following dicussion describes the cut records that
are created if the original PTrack has a cut record.  No records
are constructed if the original track does not have one.

Stage 1
-------

In the first stage the original PTrack is propagated to any reachable
layer surfaces which appear next in its path.  A new PTrack is created
for each such layer surface.  A new cut record is created for each
new track.

The cut records from this stage have PTrack ID's and PTrack parent
ID's.

Stage 2
-------

For each new track, a search is made for nearby clusters.  A new cut
record is created for each cluster.  The cut records are stored 
in a map indexed by cluster.

A series of cluster filters are applied to reduce the number of
clusters.  Each filter is passed the map of cut records so it
can record cut information.

The cut records produced in this stage do not have PTrack ID's.
They do have PTrack parent ID's.

Stage 3
-------

Surviving clusters are used to generate hits.  A cut record is
created for each such hit.  An AddFitter is used to refit the
track with the new hit.  It is passed the cut record to update.

Tracks with succesful fits are used to construct new PTrack's.
Their cut record is assigned to the PTrack.  Checkers are then
applied.  The are passed the cut record for update.

The cut records from unsuccesful fits do not have PTrack ID's
or PTrack parent ID's.  Those from succesful fits have both ID's.