Minutes of the CALCOM Analysis Meeting March 28, 1997.

 
 Three talks were presented about analysis results using hit based tracking
 (B. Niczyporuk, W. Brooks, J. Manak) and one presentation about strategies
to get to the full time-based tracking including track-hit matching (M. Mestayer).
The following are summaries provided by the speakers. In general, these 
summaries should be kept short (not everybody complied with the one or 
two paragraphs limit!) as these are just preliminary results reflecting 
limited analysis capabilities and other shortcomings such as lack of 
accurate calibrations. Therefore, many of these results will be 
obsolete in a matter of weeks. In any case these reports shold be 
useful as they document progress we are making on the various CALCOM fronts. 

                     Volker Burkert


                     SUMMARIES:

B. Niczyporuk:
============= 
 1) The new magnetic field table has been generated (big TORUS) with muximum
    current of 3860A. The final Oxford geometry of 54x4 loops has been used.
    The location of magnetic table (FPACK format) is:
          "/apps/clas/SDA/DAT/bgrid_t.fpk"
    The location of magnetic field table for mini TORUS (with maximum current
    of 8000A) is:
          "/apps/clas/SDA/DAT/bgrid_m.fpk"
    The comparison of the new field with the old one shows that the difference,
    even in the middle plane, may be about of 0.16 Tesla on the trajectory with
    theta of 20 degrees.
    
 2) The data, Run# 689, Itorus = 1929A and Iminit = 6000A, has been reanalyzed
    with the SDA (Level=2, without timing info). The electrons were identified
    by matching the DC tracks with the corresponding hits in SC and EC 
    (Eshower/Ptrack). The resulting missing mass spectrum (W) shows clearly
    the quasi-elastic peak of electron scattering on the "CH2" target.
    The effects of new magnetic field on data (i.e. position and width of
    quisi-elastic peak) agree with the Monte Carlo predictions.
    

W. Brooks:
==========
The presentation focused on estimating energy resolution for electrons
in the forward calorimeter using DC momentum from hit based tracking ("HBT").
The analysis shell used was recsis. DC information was matched to outer
detectors using W. Brooks' private track matcher.

The calorimeter energy calculations used attenuation lengths from the 
Test Lab cosmic ray test measurements, pedestals from the Feb.
run, and tube gains from Cole Smith's MIP analysis with zero B-field, 
thought to be accurate to about 10%. The HBT used the 'old' pre-run
B field and reference trajectories. A new field and reference trajectories
exist, but these do not yet seem to work in recsis and have recently been
replaced by even newer files.

A set of 1700 electrons were identified using the following cuts:
HBT identified the track, charge is negative, one charged particle in
the event, Cerenkov counter ("CC") fired, calorimeter inner/outer energy
distribution correct for electrons, total energy in calorimeter generally
consistent with DC momentum. The data was from run 781, where
the CC voltages were 200 V above nominal, and the calorimeter trigger
threshold was about 750 MeV. The appearance of the data indicated a clean
electron sample. All cuts were 'loose'.

RESULTS: The distribution of the energy from the calorimeter ("E") has a
different shape from the distribution of momentem from HBT ("p"). The effect
cannot be explained by resolution. Looking into this, plots of E-p vs. p, 
E-p vs. E, and E vs. p demonstrate that the relation E=p holds well up to 
1.25 GeV under the assumption that 1) the MIP peak calibration gives the 
correct energy and 2) a sampling fraction of 0.26 is used. The latter is 
smaller than simulation predictions. There is a departure from E=p above 
1.25 GeV which grows to 15-20% above 2 GeV, where E>p. The interpretation
was the HBT with the given B-field was not linear; this was disputed from
the audience.

Earlier simple approaches to estimating the resolution relied on the
E=p linearity and looked at all data simultaneously to have adequate 
statistics. Discarding this approach, the quantity E-p was plotted for 
bins in energy and the RMS was determined; statistics were too poor for
fitting. RMS/sqrt(E) was plotted; this is expected to be 8.6% from simulation
and independent of energy. However, RMS/sqrt(E) showed a strong "V-shaped" 
dependence on energy, ranging to greater than 20%. The interpretation given
was that HBT with the given B field is 
malfunctioning; this was also disputed from the audience. If the MINIMUM
value of this dependence is taken to represent an upper limit for the 
real calorimeter energy resolution, this value is about 12%; subtracting the
energy bin width of +/- 100 MeV, gives a value of 10.5%. If the intrinsic
resolution is assumed to be 8.6%, this implies a gain matching uniformity
of about 6%, better than the 10% estimate given by Cole Smith. Therefore
the results are consistent with the expected resolution, given a number
of assumptions.


J. Manak:
=========
Reults were presented showing gaps in the cos(theta) distribution
of reconstructed Monte Carlo events processed by gsim and recsis. 
A similar distribution was generated within the SDA framework and
gaps were not found, suggesting that the problem exists within 
recsis or the gsim simulation. After the meeting Francois said that this
problem had been seen before and could be attributed to a drift chamber
misalignment within gsim in regions 2 and 3. This is being investigated.

In addition, gaps in the cos(theta) distribution in the test run data
were observed. It is believed these gaps are unrelated to the 
gsim/recisis problem described above. It is hoped that the gaps in the
data will be fixed by the re-analysis the raw data with the appropriate
tdc edge finding software. Joe has just heard from Jim Mueller that the
new reconstruction code is available, and Joe will begin re-processing 
shortly.

Finally, the Cerenkov counter efficiency was examined as a function of 
electron hit position on the face of the Cerenkov counter. The efficiency
at large theta (20-45 deg) greater that 90% at small theta (10-20deg) the 
efficiency was worse 40-80%. Electron selection cuts are still being 
investigated. 


M. Mestayer:
============
A report on the status of tracking software was presented. Much work 
has gone into the wire map and event de-mux'ing software in order 
to produce correct HITS.  Francois Roudot, Stepan Stepanyan and 
Kevin Beard are working together to do track matching of DC and outer detector.
When this is complete we will be able to calculate the EVENT START TIME, the
time (in the DC TDC scale) at which the ELECTRON candidate track left the target.
Knowing the event start time, the fixed delays from each wire (the T0's which
we get from analyzing pulser calib. data) and the flight time for each track to
the DC layer in question, we can correct the TDC time to be the actual DRIFT
TIME.  At this point, we will study the DOCA - TIME correlation between tracks
and the associated hit times in order to fit a first-order try at a drift velocity
function.  We can then do TIME-BASED TRACKING.  Tune in next week for first results!