Jan T. Balewski           Performance of  ppLMV - low multiplicity vertex finder  for pp events with moderate pileup (updated  3/18/01)

Description of ppLMV
The two distinct features of the pp events expected in 2001 STAR data are:
- low track multiplicity for the trigger event (5-10 tracks)
- low, but non zero pileup probability, leading to about 10 pileup events in the TPC

To enable vertex reconstruction of the pp events with the above topology, the original  lmv()  code was duplicated and added as a method to the StPrimaryMaker::ppLMV() . Additionally, ppLMV()  has built in a cut rejecting all global tracks  with  too large Zdca to the nominal beam line.

The   ((StPrimaryMaker *) chain->GetMaker("primary"))->ppLMVuse(15.); // cm
method is added for both:
- force  the StPrimaryMaker  to use ppLMV() for all events , regardless on track multiplicity ( instead of lmv() or egr() ) and
- define maximal Zdca value

Performance of ppLMV
ppLMV() was tested with pp nondiffractive minbias pp events, generated with geometry 'year2001'  for MDC4. The vertex was smeared with sigZ=7.5 cm and sigX=sigY=0.5 mm.
Pythia events were processed using TRS with  pileup  added , starting from 0 upto the level of 2%  probability per bunch crossing .
1% of pileup per bunch  corresponds approximately to  the  luminosity of 4*10**30 1/cm**2 1/s, assuming  the pp nondiffractive cross section of 28 mb and 106 ns separation between bunches =120 bunches/ring.
 
pileup prob.
per bunch X
total number of
pileups in TPC
vertex recon 
efficiency
|DZ vert| <1.5 cm
RMS-Z  ( cm)
using only 
|DZ vert| <1.5 cm
RMS-X  (cm)
using only
|DX vert| <1.5 cm
CPU/event (sec)
with  the MDC4 chain
no 0 0.87 +- 0.015 0.34 0.34 23
1% 8 0.74 +- 0.02 0.50 0.41 69
2% 16 0.65 +- 0.02 0.55 0.44 120
3% 24 0.58 +- 0.02 0.60 0.47 173
5% 40 0.49 +- 0.02 0.75 0.45 --> sigX 273
10% 80 0.37 +- 0.02

 Fig below shows error of reconstructed vertex in Z and in X directions for 500 input events without pileup.

Similar plots with pileup probability of 1% per bunch crossing .500 input events.

Similar plots with pileup probability of 2% per bunch crossing .500 input events.

imilar plots with pileup probability of 3% per bunch crossing .500 input events.

imilar plots with pileup probability of 5% per bunch crossing .500 input events.

imilar plots with pileup probability of 10% per bunch crossing . 400 input events.

The ppLMV() was added to  St_dst_maker in  the CVS repository.

go home