Change track finding algorithm of Tracking4D
The track finding in the
Tracking4D module is currently malfunctioning:
Clusters are added to the track candidate (
StraightLineTrack) one by one. For the second cluster, currently the following condition has to be true: The hit in a detector has to be at the same global position in x and y as the reference (first) cluster (within the range of the spatial cuts).
With this, we only allow for tracks that have no slopes, or to precise at maximum have an angle w.r.t. the global coordinate system of
slope_max = spatial_cut/(ref_plane.z()-new_plane.z())
per transverse direction.
New method: Loop over each combination of clusters in two reference planes, then proceed as before: check for clusters close to the track.
This improves angular acceptance and overall probability of track finding, since the previous implementation discarded clusters for which the first track attempt did not lead to a track.
Example: Tracking4D in 6 planes of Mimosa26 detectors, process 50 kEvents.
master: 0.056 tracks/ev
bugfixTracking: 0.686 tracks/ev