Application of the SVT simulation to Run II Data


Due to the lack of the SVTD bank in most of the recent data sets appears clearly, for our final scope, the need to have some alternative information about the SVT reconstructed tracks. We decided to run the SVTsim_standalone.exe program, written by G. Punzi, on the hits contained in the SIXD Storable Bank, for the few runs which have it available. Our aim is to give a sort of validation of this new quick simulation, based on the comparison with the real data in the SVTD bank. This task appears necessary in order to improve our study of SVT performance on very crowded events, like multijets or Higgs associated production with vector bosons in the 4-jets final state.


Other Technicalities

Active layers
Current status of activity of the SVX layers. On the x axis is the sector number, on the y axis the layer number.
At this time, SVT is able to reconstruct tracks only if there are hits in all of the activated silicon layers (layers: 0, 1, 2, 3). It is strongly necessary to have a map of the SVX-II sectors that are already in use, and consequently of those layers that could receive and register hits, in order to correct the rate estimates For Run II data that belong to the Stream G:

-> run 123499
-> run 123618
-> run 123622
-> run 123623
-> run 123842
-> run 123852
-> run 123859
-> run 123868
-> run 123901
As is clear from the picture on the right, we do not yet have the full silicon detector operational. For the runs examined here the percentual of active layers for each half-barrels are:
-> half-barrel 0: 66.6 % of active layers;
-> half-barrel 1: 75.0 % of active layers;
-> half-barrel 2: 33.3 % of active layers;
-> half-barrel 3: 66.6 % of active layers;
-> half-barrel 4: 50.0 % of active layers;
-> half-barrel 5: 25.0 % of active layers;
In order to compute rate estimates for the proposed SVT selection at Level 2, we need to not just count how many tracks with impact parameter above 100 microns we find in each event, but extrapolate to the optimal situation we hope we'll be facing in a few months. To do that, we extract the distribution of the number of tracks per event in each half-barrel, after correcting for the fraction of azimuth not yet covered, and calculated the expected number of tracks that would be found in an optimal SVX-II running configuration. That number is extracted by runnig a random number over a poisson distribution with mean equal to the extimated number of lost tracks. Then a distribution of the impact parameters is performed. There are different cases on which we should correct our extimation of rates event per event.
If between all tracks we have just found two different tracks with d0 greater than 100 microns, our trigger will shot and the event will pass our SVT requirements. It does not mind is other tracks would be lost, for now.
Critical is instead of this, the situation in which we find only one tracks satisfying our requirements. Other tracks that would be able to make the trigger shoot maybe are lost trought the dead zone of SVX-II. To understand such a possiblity we pull N,(= number of expected lost tracks) random number of impact parameter, following the general IP distribution of all the tracks of the considered event. Then we take in accounts only those that are greater than 100 microns. If we find another track the trigger will be able to shoot again.
In the case with no SVT "trigger" tracks we can procede as before but now searching for two random generated IP greater than 100 microns.



Results:

Active layers
Extrapolated rates in input to L3. Calorimetric L2 requirements already applied.
Top Left: IP distribution of the greatest IP tracks
Top Right: IP distribution of the second greatest IP tracks
Bottom Left: Extrapolated rates in input to L3 as a function of the IP cut on the greatest IP track
Bottom Right: Extrapolated rates in input to L3 as a function of the IP cut on the second greatest IP track

Here we show the the input rate to Level~3 summuing the calorimetric and SVT requirements as a function of the cut we want to do on the IPs. Top Left: IP distribution,event per event, of the greatest IP track Top Right: IP distribution, event per event of the second greatest IP tracks. Bottom Left: L3 into rate as a function of the IP cut. Bottom Right: L3 into rete as a function of the IP cut on the second greatest IP tracks.


The total rate into L3 appears to be larger than expected: it is 3-4 Hz instead of the extimated 1.5 Hz. This fact could maybe understood cosidering the large superstrips size used. Some correction are already pending waiting for possible corrections to be applied due to the result of the comparison of SVTsim with SVTD Storable_Bank.


















Useful Links

red SVT Pisa's Home Page
red Back to the multijet trigger web page

Giorgio Cortiana
Last modified: Wed Sep 12 15:20:18 CDT 2001