Results for t00.67.00 build 6

D. Adams
03dec99 1500


Here we show results obtained for t00.67.00 tracking using build 5 with the RCP and OBS files in build 6.

For single muons:

Full reconstruction (i.e. after step 4):

 run cid  CPU    eff  misr  fake  fcsq  mcsq  nclu
---- ---  ----  ----  ----  ----  ----  ----  ----
m50  401  0.85  .856  .004  .000  18.4   5.8  17.3
m03  401  0.83  .847  .051  .002  23.3  13.4  17.0
m01  401  0.78  .561  .154  .001  25.0  18.4  15.0

Here are results for CFT only (step 1):

m50  101        .750  .000  .805  11.2   5.8  14.3
m03  101        .741  .017  .543  11.2   9.3  14.0
m01  101        .453  .006  .230  12.7  14.9  13.0

CFT central only:

m50  101  0.13  .633  .000  .000  12.5   5.9  16.0
m03  101  0.13  .601  .013  .000  13.3   8.9  16.0
m01  101  0.11  .282  .007  .000  16.5   8.8  16.0

CFT overlap only:

m50  101  0.20  .185  .000 2.919   7.2   6.2   8.7
m03  101  0.19  .213  .059 1.694   6.7  17.4   8.8
m01  101  0.19  .191  .142  .530   7.7  27.8   8.7

There is a big efficiency loss at pT = 1 GeV/c. The plot shows this occurs in both the central and forward regions. A plot of the CFT tracks shows that the central loss occurs in the CFT. Note that these are inefficiency plots and the second figure shows the tracks that were not found.

The overlap region is efficient but contains most of the misreconstructed tracks.

Cleary there are problems with the match chi-square at low pT.

I have begum processing the Z -> mu mu run. Ten of the first 15 events failed due to too many tracks during step 2: extending from CFT to SMT.

There are many fake tracks found in the CFT overlap. This may contribute the problems extending z -> mu mu events into SMT.


Here are Z->mu mu results for the central CFT+SMT, overlap CFT only and forward SMT:


 run chk      CPU/ev   eff  misr   fake  fcsq  mcsq  nclu
---- ---      ------  ----  ----   ----  ----  ----  ----
mzz_cent        9.00  .246  .104   .000  25.9  12.2  19.7
mzz_over_cft  574.    .695  .906  4.24    6.6  35.4   7.1
mzz_forward    44.0   .269  .395   .041  17.7  23.3   6.6


Previous results were not properly filtering the overlap CFT tracks. That has been resolved by switching to a six-step reconstruction. The steps and the chunk identifier (cid) for each are as follows:

111 - central CFT
121 - overlap CFT
101 - full (central+overlap) CFT
201 - full CFT extended into SMT
301 - forward SMT
401 - full CFT plus forward SMT
Here are new results for each step for 50 GeV muons:
 run chk  CPU    eff  misr  fake  fcsq  mcsq  nclu
---- ---  ----  ----  ----  ----  ----  ----  ----
m50  111        .633  .000  .000  12.5   5.9  16.0
m50  121        .129  .008  .000   6.2   7.5   9.2
m50  101        .695  .001  .000  11.7   6.1  15.2
m50  201        .691  .003  .000  19.7   5.8  19.0
m50  301        .130  .008  .000   7.2   4.4   6.0
m50  401  .800  .801  .003  .005  17.9   5.6  17.2

Here are final results for each of the single muon runs:

 run chk  CPU    eff  misr  fake  fcsq  mcsq  nclu
---- ---  ----  ----  ----  ----  ----  ----  ----
m01  401  .747  .506  .084  .002  23.1  15.5  15.2
m03  401  .789  .789  .031  .001  20.7  11.0  17.0
m50  401  .800  .801  .003  .005  17.9   5.6  17.2

The Z -> mu mu run surpassed its CPU limit after 14 events. There were "too many tracks" for 4/14. The CPU was 769 s/ev.



Here are forward results for Z --> mu mu w/o and w/ 1 and 2 background.

 run chk  CPU    eff  misr  fake  fcsq  mcsq  nclu
---- ---  ----  ----  ----  ----  ----  ----  ----
zmm0 301   44.  .269  .395  .041  17.7  23.3   6.6
zmm1 301   67.  .306  .435  .050  17.9  24.9   6.6
zmm2 301  443.  .330  .537  .064  20.3  26.9   6.4