A fast track search algorithm "direct_walk" has been developed,
which does not require hit cleaning.
Approximate track parameters are determined and hits
associated to a track are identified.
It is very well suited as a first guess method in the
level_1 selection as the processing speed is only 1.5 times
the one of the linefit.
In addition the angular resolution of tracks is considerably
improved.
Background muons at level_1 are suppressed further by a factor 2.5,
while the passing rate of signal muons is about the same.
The direct_walk algorithm
could be used as hit selection process on an event by event basis,
to be used by subsequent fits. In addition, it is capable to identify
and evaluate multi-muon events.
It could such serve as a first step for a new
pattern recognition program.
Further information can be obtained from
the draft for an AMANDA Internal Report
(zipped ps-file).
DW2:
The direct_walk algorithm has been improved considerably by a better
selection of the track candidates. Details are available
from the draft of a further AMANDA Internal Report
(zipped ps-file).