The E895 data reduction scheme follows the "standard"
scheme used by TPCs:
- 1) For each padrow, identify track crossing points ("hits").
- 2) Walking roughly perpendicular to padrows, link hits to form tracks.
- 3) Calculate (truncated) sum of energy loss along track for PID.
- 4) Do physics analysis (flow, HBT...).
Steps 1-3 are done N times for the full dataset, where N.le.3.
The performance and quality of the chain at each step depends
on the quality of the previous steps.
Some of the proposed analyses for E895 are going to be
difficult in a high track density environment (higher than for STAR!).
Poor hitfinder performance results in a loss of efficiency, but
worse than that, the efficiency loss has large correlations
in it. For the simplest example, if a track is lost because another
is nearby (and not just because he is in some area of phase space), that
will build correlations into the reconstructed data that are deadly
to any correlation study. Such affected studies include:
- HBT studies.
- Flow
- Any type of detailed event-by-event physics.
Clearly, kink or V0 searches will also be hampered by poor hitfinding
performance.
To maximize the physics impact of E895,
we must do as well as possible in steps 1-3. Our experience
from EOS tells us that cuts a postiori are a weak remedy for poor
reconstruction performance.
Hitfinding on a padrow is an inherently 2-dimensional problem (one
has ADC values as a function of pad number and time bucket), as is
shown in blowups of pixel patterns on a portion of padrow 45 for one
Au+Au event at 6 A GeV.
The EOS hitfinder was used in the
EOS Bevalac experiment.
Its algorithm attacks the hitfinding problem in a 1D+1D way,
finding peaks in the time and pad direction independently, and
associating them later. It worked well enough for Bevalac energy,
but may not be sufficient for the higher track density at the AGS.
Experiment NA49
also began data analysis in this way, but now finds
itself turning to explicitly 2D algorithms, in order to
increase efficiency in a high track density environment.
STAR's algorithm is 2-dimensional. Documentation on it can
be found HERE or on the
STAR home page--it is STARNote 238.
When E895 uses the EOS reconstruction software (hitfinder and tracker),
it is clear from inspection of rapidity, pt spectra, etc.,
that we have efficiency problems. Even inspection of one
low multiplicity event
shows us that we miss many tracks at backward angles, and some
tracks that are close to other tracks.
To explore the possibility that a hitfinder based on a 2D algorithm
might improve event reconstruction performance in E895, we have implemented
an improved version of the STAR cluster/hitfinder into the E895
hitfinder analysis module. The selection of algorithm (new or old)
is selectable via the HIT_SW
switch table. (Of course, HIT_SW controls
many other things in the hitfinder as well.)
Back to E895 Hitfinder Main Page