--
DanielScheirich - 07 Jun 2007
B-physics CSC task #7: LVL2 di-muon trigger calibration
Introduction
In order to measure quantities such as cross-sections, branching ratios, etc. it is necessary to
know a trigger efficiency. In design-time of trigger menus and cut optimisation it is estimated
using MC simulation. In CSC task 7 we develop and study methods how to measure B physics
trigger efficiencies from real ATLAS data.
LVL2 di-muon trigger strategies
- LVL2 di-muon trigger seeded by 2 LVL1 RoIs.
- LVL2 di-muon trigger seeded by 1 LVL1 RoI (TrigDiMuon)
LVL2 di-muon trigger seeded by 2 LVL1 RoIs
Method
MuComb algorithm combines muons reconstructed in Atlas muon system by
MuFast and
inner detector tracks reconstructed by IDScan and creates an output object
CombinedMuonFeature. It contains information of
transverse momentum pT , pseudorapidity eta and angle phi of the muon track at the point of origin.
To calculate a di-muon trigger efficiency one must know a single muon efficiency as a function
of all kinematic variables (pT , eta, phi). Di-muon efficiency then can be calculated as a product of
two single muon efficiencies.
To calculate single muon efficiency, the following analysis procedure was used:
- All data
- Single muon trigger (Mu6 or Mu10): simulation of trigger that will be used for data taking.
- Off-line analysis: off-line tracks of opposite charged muons are combined. Track pairs that give J/psi mass are accepted. Muon tracks from STACO muon container were used.
- Probe muon: off-line muon that was triggered by single muon trigger (triggered muon) was excluded from analysis. The second J/psi muon is used as a probe muon. If probe muon does not lie in some LVL1 RoI it is excluded since we do not want to take into account inefficiency caused by LVL1 trigger.
- Efficiency was calculated using probe muons.
Data samples
50k events direct J/Ψ(μ+μ‒) recon. in 12.0.6:
trig1_misal1_mc12.017503.Pythia_directJpsimu6mu4.recon.AOD.v12000602
50k events bb → J/Ψ(μ+μ‒) recon. in 12.0.6:
trig1_misal1_mc12.017516.PythiaB_bb_Jpsimu6mu4X.recon.AOD.v12000602
In addition:
10k events Z(μ+μ‒) reconstructed in 12.0.6:
trig1_misal1_mc12.005186.PythiaZmumu_pt100_fixed.recon.AOD.v12000601
Results
Using probe muons efficiency as a function of pT, eta and phi was calculated. In order to have a crosscheck efficiency was also calculated using the whole sample without simulation of single muon trigger (step 2). This method can't be used on real data, but it should be unbiased. Results are shown in following histograms: thick black line represents measured values and dashed line represents results of alternative unbiased method. Error bars shows binomial errors.
Figure: Efficiency as a function of pT. Histogram was fitted with function y = A/(1+exp(ax + b)), which describes well simulated data.
Fit parameters are shown in the table:
Figure: Efficiency as a function of eta.
Figure: Efficiency as a function of phi.
Figure: Efficiency as a function of eta and phi.
In order to calculate efficiency
map in full pT , eta and phi binning we will need the huge statistics. Because distribution is flat in phi and
symmetric in eta we can calculate efficiency
only as a function of pT and |eta|. Eta axis was divided into 5 bins and in each bin efficiency as a function of pT was calculated.
It is shown in following histograms:
0 < |eta| < 0.5 0.5 < |eta| < 1
1 < |eta| < 1.5 1.5 < |eta| < 2
2 < |eta| < 2.5
Parameters of the fit are listed in the table:
|
0 < eta < 0.5 |
0.5 < eta < 1 |
1 < eta < 1.5 |
fit parameter |
value |
error |
value |
error |
value |
error |
A |
9.48508e-01 |
3.15044e-03 |
9.41871e-01 |
3.36124e-03 |
8.80829e-01 |
6.02133e-03 |
a |
-1.66036e-03 |
1.30253e-04 |
-1.68967e-03 |
1.83080e-04 |
-1.50472e-03 |
1.86851e-04 |
b |
4.55486e+00 |
5.11937e-01 |
4.92008e+00 |
7.60110e-01 |
5.15911e+00 |
8.15396e-01 |
|
1.5 < eta < 2 |
2 < eta < 2.5 |
fit parameter |
value |
error |
value |
error |
A |
7.99783e-01 |
7.02257e-03 |
7.27727e-01 |
1.30243e-02 |
a |
-1.58504e-03 |
2.26949e-04 |
-1.22730e-03 |
1.48584e-04 |
b |
5.09138e+00 |
9.48816e-01 |
4.34446e+00 |
6.00958e-01 |
Using the fitted functions we can plot efficiency as a function of pT and |eta|. Values between centers of eta bins were linearly interpolated:
The plot shows the efficiency that muon will be reconstructed by
MuComb algorithm. None pT cut was applied on
CombinedMuonFeature. In order to calculate efficiency of a certain trigger menu, one must modify the analysis and require
CombinedMuonFeature with pT greater than corresponding threshold to be matched with probe muon.
Di-muon efficiency
Di-muon efficiency will be calculated as a product of two single muon efficiencies:
eff(pT1, pT2, eta1, eta2) = eff(pT1, pT2) ∙ eff(pT2, eta2)
Rates
Signal rate can be estimated using formula R = L ∙ σ ∙ ε, where L is luminosity,σ is crossection and ε is probability that event will contain usable probe muon. ε depends on threshold of single muon trigger used for data taking but also on efficiency of off-line analysis used for J/psi reconstruction.
Total crossecion for process pp -> J/psi(mm) is σ = 21.75 nb (taken from
BPhysWorkingGroupCrossSections).
Signal rates for two thresholds are listed in following table:
luminostity L = 1e33 cm-2s-1 |
pT threshold |
6GeV (Mu6) |
10GeV (Mu10) |
ε |
0.29 |
0.29 |
Signal rate |
6.3 Hz |
1.37 Hz |
Since single muon threshold is used there will be lot of background events taken. Rates for background were taken from
Antonio Sidoti's talk @ T&P week (5 June 07):
http://indico.cern.ch/materialDisplay.py?contribId=85&sessionId=5&materialId=slides&confId=16155
luminostity L = 1e33 cm-2s-1 |
pT threshold |
6GeV (Mu6) |
10GeV (Mu10) |
Background rate |
10085 Hz |
1813 Hz |
WARNING: presented numbers are very preliminary. In our analysis only LVL1 & LVL2 trigger was used to simulate trigger for data taking. By including EF to the analysis background and also signal rates will drop.
Analysis tools
Analysis was made on AODs in Athena 12.0.6.
Source code is located in
/afs/cern.ch/user/s/scheiric/public/UserAnalysis/src
/afs/cern.ch/user/s/scheiric/public/UserAnalysis/UserAnalysis
Job options are located in
/afs/cern.ch/user/s/scheiric/public/UserAnalysis/run
Important files:
-
jobO_ganga_bb_jpsi_mm.py
analyze bb->J/psi(mm) events. It is to be ran on Grid.
-
jobO_ganga_direct_jpsi_mm.py
analyze pp->J/psi(mm) events. It is to be ran on Grid.
-
jobO_ganga_bb_jpsi_mm_10GeV.py
higher pT threshold is used.
-
jobO_ganga_direct_jpsi_mm_10GeV.py
higher pT threshold is used.
-
gangaNG_direct_jpsi.py
ganga script. It submits the job to NorduGrid
-
gangaNG_bb_jpsi.py
ganga script. It submits the job to NorduGrid
-
gangaNG_direct_jpsi_10GeV.py
ganga script. It submits the job to NorduGrid
-
gangaNG_bb_jpsi_10GeV.py
ganga script. It submits the job to NorduGrid
Root script for output AANT analysis:
/afs/cern.ch/user/s/scheiric/public/root/analysis.cpp
IMPORTANT: script must be compiled before running otherwise it is very slow. Type
[0] .L analysis.cpp++
in Root command line to compile script.
to be done....
Links
Talk given at B-physics meeting 6 June 2007:
CSC task 7: Low pT LVL2 di-muon trigger calibration