Efficiency

• |z vertex| < 10 cm
• |eta| < 1
• N fitted points > 15

  Data

  Figure : Data : cucu @ 62 GeV	Figure : Data : cucu @ 200 GeV

  Simu

  Figure : Data : cucu @ 200 GeV

  There are not significants differences between 62 and 200 GeV.
  For the simulated CuCu @ 200 GeV, :
  • the efficiency is higher than the one observed in the data
  • it decreases with pT and more differences between positive and negative tracks are seen.I think this is simply due to the statistics. The number of events in the simulation plots are 7200 (18 * 400 events) but only ~ 2500 survived the cuts.

  Some possibles causes of the differences

  • The code (hit reco) is not adapted for the CuCu collisions (cut on s/n for the clusters)
  • The code for the simulation(pedestal) does not reflect the real detector (all ladders are active, noise level is lower than the one seen in real data)
    For example, in the simulation , the noise are generated randomly for all the ladders with a mean of 1400 electrons for the P side and 2200 electrons for the N side.
    This give us a noise ~ 3.2 ADC and 5.00 ADC for the P and N sides respectively.
  The status of the ssd simulation code is ( main parts) :
  Pedestal & Noise parts :
  1. : all the ladders are active.
  2. : the percentage of dead strips,wafers,chips are all set to 0
  3. : the noise of the strips on the P side has a mean of 1400 electrons (3.2 ADC) with a sigma of 50 electrons (0.12 ADC)
  4. : the noise of the strips on the N side has a mean of 2200 electrons (5.0 ADC) with a sigma of 70 electrons (0.16 ADC).
  However, the noise seen in the CuCu are differents from each ladder and is a bit lower than these 2 values.
  
  Hit reco :
  1. There is a s/n cut for the selection of the central strip of the cluster. As for the data, it was set to s/n = 5.
     So both the simulation and the data reco are not optimized to the CuCu run since this cut was chosen for AuAu collisions. With a lower s/n cut, it is possible to reconstruct more hits, then the efficiency od the ssd (in dat ans simu) could increase since more ssd hits could be associated.
     The differences between data & simu still has to be understood.

  Efficiency (update July 21th)

  I change the way to select the tracks.
  In the previous section, I selected only the tracks with ssd point (no svt points) and with N fitted point >= 1 for the SSD.
  I select now the tracks with ssd point (the same) BUT I don't care if this track has or not some SVT points.
  I also select the tracks with N possible point >= 1 for the SSD.
  As a result , the efficiency slightly increases for the simulation but the main remark is that there is no more such big differences between data( 62 GeV or 200 GeV) and the simulation.
  

  Simu

  Figure : Data : cucu @ 200 GeV

  Data

  Figure : Data : cucu @ 62 GeV	Figure : Data : cucu @ 200 GeV

  Number of ssd point taken in the tracks

  One possible explanation of the low efficiency could be that the hit in the ssd is not associated with the track at this layer.
  The next plot shows :
  • black : ratio between the number of hits used over all the ssd hits.
  • red : ratio between the number of hits used over the number of primary
  Again, the same cuts are applied as before
  One can see that there is more points reconstructed in the ssd independantly than those used in the tracking ; it could indicates that these hits are fake (noise). However, the number of ssd hits used in the tracking is near than the number of primary tracks, which could indicate that when a track (from the tpc) crosses the ssd layer near a ssd hit, this hit is added to the track.
  

  Figure : Ratios (see text above for legend)

  title 4

  title 5

  Jonathan Bouchet
  Last modified: Wed Jul 19 07:53:58 EDT 2006