Hi all,
I have committed a few changes in the TPC local tracking classes
BrTrackFinderBase, BrTrackFollowFinder and BrTrackStringFinder:
Additions to base class:
- Moved data member TObjArray* fHitsInRow[] and method SortHitsIntoRowTable()
from BrTrackFollowFinder to base class BrTrackFinderBase
- Moved member BrDetectorVolume* fTPCVolume from BrTrackFollowFinder to
BrTrackFinderBase
Additions to BrTrackStringFinder:
- Added new tracking method FindTracksByStringMatch2() picking up only one
BrDetectorHit per padrow. The original method, FindTracksByStringMatch()
only counts hits close to track line and can (in theory) produce ghost tracks
from dense groupings of hits in a few padrows.
- Added member fFindTracksMethod and function SetFindTracksMethod( int )
(1 or 2)
- Added member fAllowedMiss and function SetAllowedMiss( int ) to harmonize
with other tracking modules. The minimum required number of hits for a good
track is now calculated as:
fThresholdNo = fParams->GetNumberOfActiveRows() - fAllowedMiss;
and the function SetThresholdNo( int ) should be considered obsolete.
Note: The minimum required number of hits is now fThresholdNo, instead of
fThresholdNo+1 as earlier (important for efficiency estimates.)
Otherwise, very preliminary tests on simulated data show that the various
tracking classes now seem to have quite similar efficiencies. There is
still room for improvements at high track density (e.g. MTPC1 at forward angles), in particular with respect to cluster deconvolution (if possible - .)
Best regards,
Trine
This archive was generated by hypermail 2b29 : Fri Jun 23 2000 - 06:58:10 EDT