Upload
javan
View
31
Download
4
Embed Size (px)
DESCRIPTION
Determination of integrated Luminosity via W/Z Boson Production at the ATLAS Detector CERN 2011 Matthias Schott (CERN) O n behalf of the ATLAS Collaboration. Content Current Results and their systematics Expected Precision of fiducial cross-sections for 1fb -1 - PowerPoint PPT Presentation
Citation preview
M. Schott (CERN) Page 1
Determination of integrated Luminosity via W/Z Boson Production at the ATLAS Detector
CERN 2011Matthias Schott (CERN)
On behalf of the ATLAS Collaboration
Content• Current Results and their systematics• Expected Precision of fiducial cross-sections for 1fb-1
• Improving Theoretical Uncertainties on Acceptance• Conclusion
M. Schott (CERN) Page 2
Theoretical Prediction of W/Z Production Cross-Sections
ATLAS Central values from QCD NNLO ATLAS Central values from QCD NNLO calculations by FEWZ with ~5% calculations by FEWZ with ~5% systematic uncertainty (based on 90% systematic uncertainty (based on 90% CL, including PDF and αCL, including PDF and αSS uncertainties) uncertainties)
See also S. Alekhin et al: See also S. Alekhin et al: http://arxiv.org/pdf/1011.6259v1http://arxiv.org/pdf/1011.6259v1
PDF uncertainties: uncertainties of NLO PDF uncertainties: uncertainties of NLO and NNLO calculations from 68% CL and NNLO calculations from 68% CL error sets vary between ±(1.6 − 2.0)% error sets vary between ±(1.6 − 2.0)%
Found to be about a factor of two Found to be about a factor of two larger using 90%CL error sets,larger using 90%CL error sets,
suggesting non-Gaussian suggesting non-Gaussian distributions of the cross section distributions of the cross section variations.variations.
ααss value and its uncertainties value and its uncertainties
using MSTW2008 fits which include PDF using MSTW2008 fits which include PDF sets with αsets with αss values corresponding to ±1σ values corresponding to ±1σ
68% CL leads to 1.1%, 90% leads to 68% CL leads to 1.1%, 90% leads to 2.6% relative uncertainty2.6% relative uncertainty
Electroweak Radiative Corrections assumed Electroweak Radiative Corrections assumed to be negligibleto be negligible
M. Schott (CERN) Page 3
MethodologyThe production cross sections for The production cross sections for the W and Z bosons times the the W and Z bosons times the branching ratios for decays into branching ratios for decays into leptons can be expressed asleptons can be expressed as
NNsigsig denote the numbers of denote the numbers of background-subtracted signal background-subtracted signal events passing the selection events passing the selection criteria of the analyses in the W criteria of the analyses in the W and Z channelsand Z channels
AAWW and A and AZZ denote the acceptances denote the acceptances
for the W and Z-boson decaysfor the W and Z-boson decays
€
A =N in Fiducial Region
Generated
NGeneratedAll
CCWW and C and CZZ denote the ratios denote the ratios
between the total number of between the total number of generated events which pass the generated events which pass the final selection on reco. level and final selection on reco. level and the number of generated events the number of generated events within the fiducial acceptancewithin the fiducial acceptance
RemarksRemarksAcceptance Factor A and detector Acceptance Factor A and detector effect correction factors C can be effect correction factors C can be treated as independent factorstreated as independent factors
Theoretical uncertainties affect Theoretical uncertainties affect dominantly the factor Adominantly the factor A
M. Schott (CERN) Page 4
W Boson SelectionElectron SelectionElectron Selection
EETT > 20 GeV > 20 GeV
““tight” identificationtight” identification
events containing an additional electron are events containing an additional electron are vetoedvetoed
Muon SelectionMuon SelectionppTT > 20 GeV > 20 GeV
Combined MuonCombined Muon
Track Isolation: ∑pTrack Isolation: ∑pIDID/p/pTT<0.2<0.2
Missing transverse energy Missing transverse energy EEtt
missmiss>25 GeV; >25 GeV;
Transverse mass of the lepton-ETransverse mass of the lepton-Ettmissmiss
systemsystemmmTT > 40 GeV; > 40 GeV;
M. Schott (CERN) Page 5
W-Boson Background EstimationOverall background contributionOverall background contribution
Electron Channel: 5.7%Electron Channel: 5.7%
Muon Channel: 8.6%Muon Channel: 8.6%
Data-Driven QCD background estimations for both Data-Driven QCD background estimations for both channelschannels
Relative systematic uncertainty ~1%Relative systematic uncertainty ~1%
Electroweak background estimated by Monte Carlo Electroweak background estimated by Monte Carlo simulationssimulations
M. Schott (CERN) Page 6
Z/γ* Boson SelectionA pair of oppositely-charged leptons of the same A pair of oppositely-charged leptons of the same flavourflavour
ElectronElectron
““Medium” identification, pMedium” identification, pTT > 20 GeV > 20 GeV
Veto on events with >2 “medium” electronsVeto on events with >2 “medium” electrons
MuonMuon
Combined track, pCombined track, pTT > 20 GeV > 20 GeV
Track Isolation: ∑pTrack Isolation: ∑pIDID/p/pTT<0.2<0.2
Invariant mass window of lepton pairInvariant mass window of lepton pair
66 < m66 < mllll < 116 GeV < 116 GeV
Data-Driven QCD background estimate for Data-Driven QCD background estimate for electron channel, MC prediction for channelelectron channel, MC prediction for channel
Background electron channel: ~1.5%Background electron channel: ~1.5%
Background muon channel: ~0.2%Background muon channel: ~0.2%
M. Schott (CERN) Page 7
Experimental Uncertainties
Current experimental Current experimental uncertainties uncertainties dominated by lepton dominated by lepton identificationidentification
Given the small Given the small statistics which was statistics which was used for this used for this measurement the measurement the current numbers are current numbers are already impressively already impressively smallsmall
Expect large Expect large improvements for improvements for reprocessed data and full reprocessed data and full statisticsstatistics
M. Schott (CERN) Page 8
Theoretical UncertaintiesAcceptance correction Acceptance correction factors extrapolate factors extrapolate fiducial cross-sections to fiducial cross-sections to full cross-sections full cross-sections
Predicted by Monte Predicted by Monte Carlo event generatorsCarlo event generators
Negligible impact on Negligible impact on experimental factors Cexperimental factors CWW
and Cand CZZ
Theoretical uncertainty Theoretical uncertainty within on PDF-Set (here within on PDF-Set (here CTEQ 6.6) CTEQ 6.6) in in combination with the combination with the MC@NLO acceptance MC@NLO acceptance calculationcalculation
±1.0% for W±1.0% for W++
±1.8% for W±1.8% for W−−
±1.6% for Z/γ∗±1.6% for Z/γ∗
Differences between Differences between different PDF-Sets different PDF-Sets ((MRST LO*, CTEQ 6.6 MRST LO*, CTEQ 6.6 and HERAPDFand HERAPDF))
±2.7% for W+±2.7% for W+
±0.9% for W−±0.9% for W−
±2.0% for Z/γ∗±2.0% for Z/γ∗
Uncertainties due to the Uncertainties due to the modelling of W and Z modelling of W and Z production by production by comparing PYTHIA and comparing PYTHIA and MC@NLO with same MC@NLO with same PDF-SetPDF-Set
±0.4% for W+±0.4% for W+
±1.4% for W−±1.4% for W−
±2.3% for Z/γ∗±2.3% for Z/γ∗
M. Schott (CERN) Page 9
Final Results for 300nb-1
Measured W and Z/γMeasured W and Z/γ** production production cross-sections agree with cross-sections agree with theoretical predictionstheoretical predictions
Why 300nbWhy 300nb-1-1??Already dominated by uncertainty Already dominated by uncertainty
on luminosityon luminosity
Measurement of individual cross-Measurement of individual cross-sections is only from minor sections is only from minor physical importance when physical importance when luminosity uncertainty is larger than luminosity uncertainty is larger than 2-3% (see following discussion)2-3% (see following discussion)
Near Future: Focus on Near Future: Focus on measurement of cross-section ratios measurement of cross-section ratios and differential cross-sectionsand differential cross-sections
M. Schott (CERN) Page 10
Expected Statistics in 2011 and consequencesAssuming an integrated luminosity Assuming an integrated luminosity of ~1fbof ~1fb-1-1 in 2011 at 7TeV in 2011 at 7TeV
~300k reconstructed Z Boson ~300k reconstructed Z Boson candidates per lepton channelcandidates per lepton channel
~3.000k reconstructed W Boson ~3.000k reconstructed W Boson EventsEvents
This means 1k-10k Z/W Boson This means 1k-10k Z/W Boson events per day which might enable to events per day which might enable to have an luminosity estimate on a daily have an luminosity estimate on a daily basesbases
Not sensitive to pile-upNot sensitive to pile-up
Statistical precision <0.2% for the Statistical precision <0.2% for the Z/γZ/γ** Boson channel Boson channel
No statistical advantage of No statistical advantage of measuring the W Boson cross-section measuring the W Boson cross-section
Further Experimental Advantages Further Experimental Advantages of Z/of Z/γ* cross-section measurement cross-section measurement
No dependence on missing No dependence on missing energyenergy
Lepton reconstruction Lepton reconstruction performance can be determined in performance can be determined in signal samplesignal sample
Rather Clean Signal selection Rather Clean Signal selection even in forward region even in forward region
see discussion of theoretical see discussion of theoretical uncertaintiesuncertainties
Two identical leptons in the Two identical leptons in the selection allows reduction of selection allows reduction of experimental uncertaintiesexperimental uncertainties
Does not help for ratio Does not help for ratio measurementsmeasurements
€
N ~ ε 2 + 2ε (1−ε )
ΔN ~ 2(1−ε )Δε
M. Schott (CERN) Page 11
Understanding Muons with 1fb-1
Tag and Probe techniques on the Z-Tag and Probe techniques on the Z-Boson can be used to determine the Boson can be used to determine the lepton reconstruction, trigger and cut-lepton reconstruction, trigger and cut-efficiencies on fine binning (e.g 50 bins) efficiencies on fine binning (e.g 50 bins) with statistical precision far below 1%with statistical precision far below 1%
Systematic uncertainties of offline muon Systematic uncertainties of offline muon reconstruction ~0.5%reconstruction ~0.5%
Systematic uncertainties of muon trigger Systematic uncertainties of muon trigger efficiency w.r.t. offline <0.5%efficiency w.r.t. offline <0.5%
Isolation Cut efficiency ~1%Isolation Cut efficiency ~1%
Muon Momentum Scale and ResolutionMuon Momentum Scale and Resolution
Scale already now known to ~1% levelScale already now known to ~1% level
Resolution expected to be known to <5%Resolution expected to be known to <5%
No significant impact on cross-section No significant impact on cross-section measurementmeasurement
M. Schott (CERN) Page 12
Understanding Electrons with 1fb-1
Tag and Probe technique in electron Tag and Probe technique in electron decay channel more difficult (expected decay channel more difficult (expected QCD background is significantly QCD background is significantly higher, charge misid, …)higher, charge misid, …)
Electron reconstruction efficiency (i.e. Electron reconstruction efficiency (i.e. electron container) already known to at 1% electron container) already known to at 1% level and will significantly decreaselevel and will significantly decrease
MC studies show that the electron MC studies show that the electron identification efficiency can be determined identification efficiency can be determined to ~1%to ~1%
Uncertainty on trigger efficiency Uncertainty on trigger efficiency negligiblenegligible
Impact of Electron energy scale and Impact of Electron energy scale and resolution uncertainties in resolution uncertainties in measurement can be assumed to be measurement can be assumed to be below 0.5%below 0.5%
M. Schott (CERN) Page 13
Precision of Z-Boson Fiducial Cross-Sections with 1fb-1
Assuming only one Assuming only one lepton identification for lepton identification for 1fb1fb-1-1 analyses analyses
-Conservative estimate Conservative estimate for enhanced bkg for enhanced bkg uncertainty 1% and 0.5% uncertainty 1% and 0.5% for e and μfor e and μ
• Electron channelElectron channel-ΔσΔσZZ
fidfid/σ/σZZfid fid = 1.9%= 1.9%
• Muon channelMuon channel-ΔσΔσZZ
fidfid/σ/σZZfid fid = 1.4%= 1.4%
• Combination of Combination of electron and muon electron and muon channelchannel
-ΔσΔσZZfidfid/σ/σZZ
fid fid = 1.2%= 1.2%
Electron Channel ΔCZ/CZ (0.3 pb-1) ΔCZ/CZ (1 fb-1)
Trigger <0.2 % <0.2%
Electron ID 8.8 % 1 %
Scale and Resolution 1.9 % 0.5 %
Problematic Regions in Calo. 2.7 % 1 %
Pile-Up 0.2 % 0.4 %
FSR-Modelling 0.3 % 0.3 %
Total Uncertainty 9.4 % 1.6 %
Muon Channel ΔCZ/CZ (0.3 pb-1) ΔCZ/CZ (1 fb-1)
Trigger 0.7 % 0.2 %
Muon ID 5.0 % 0.5 %
Scale and Resolution 0.7 % 0.5 %
Isolation Efficiency 2.0 % 1.0 %
Total Uncertainty 5.5 % 1.3 %
M. Schott (CERN) Page 14
Improving theoretical uncertainties (1/2)• Expected largest systematic at 1fbExpected largest systematic at 1fb-1-1: :
theoretical uncertainty on kinematic theoretical uncertainty on kinematic acceptance correctionacceptance correction
-Current theo. unc. on acceptance: 3.4%Current theo. unc. on acceptance: 3.4%
• Possible ways to improve this uncertaintyPossible ways to improve this uncertainty-Large uncertainty arise mainly from the Large uncertainty arise mainly from the
extrapolation in |η| (while pextrapolation in |η| (while pTT has a minor effect) has a minor effect)
- Extend measurement using the ATLAS forward Extend measurement using the ATLAS forward calorimeter which allows an electron calorimeter which allows an electron identification up to |η|<4.9identification up to |η|<4.9- Reduce significantly needed extrapolationReduce significantly needed extrapolation- Constrain unc. on A also for μ-channelConstrain unc. on A also for μ-channel
-Measure double differential cross-sections for WMeasure double differential cross-sections for W++, , WW-- and Z and provide new PDF-Fits and Z and provide new PDF-Fits
- Long termLong term
• Assumed uncertainty on acceptance A for 1fbAssumed uncertainty on acceptance A for 1fb--
11: 1-2%: 1-2%
M. Schott (CERN) Page 15
Improving theoretical uncertainties (2/2)• Other path: Avoid uncertainties in A from Other path: Avoid uncertainties in A from
the beginning onthe beginning on• New versions of NNLO cross-section New versions of NNLO cross-section
programs allow the calculation of fiducial programs allow the calculation of fiducial cross-sections, i.e. cross-sections within the cross-sections, i.e. cross-sections within the detector acceptancedetector acceptance
-See for example FEWZ 2.0See for example FEWZ 2.0- http://arxiv.org/pdf/1011.3540v1- Problem: no parton Problem: no parton
shower/resummation included: leads to shower/resummation included: leads to an unrealistic pan unrealistic pTT distribution distribution
-Still preliminary a not studied in detailed Still preliminary a not studied in detailed by ATLASby ATLAS
• Cross-Section predictions for fiducial cuts Cross-Section predictions for fiducial cuts (p(pTT>25, |η|<2.5) by FEWZ 2.0>25, |η|<2.5) by FEWZ 2.0
-2-3% systematic uncertainty due to 2-3% systematic uncertainty due to PDFsPDFs
FEWZ 2.0http://arxiv.org/pdf/1011.3540v1
M. Schott (CERN) Page 16
Target Precision for 2011• Combining the experimental Combining the experimental
uncertainties on Cuncertainties on CZZ and the improved and the improved
theoretical uncertainties on A lead to an theoretical uncertainties on A lead to an estimated precision of estimated precision of ~2%~2% on the on the measured Z/γmeasured Z/γ** cross-section cross-section
• Using measured cross-section to predict Using measured cross-section to predict integrated luminosity requires the integrated luminosity requires the theoretical predicted cross-sectiontheoretical predicted cross-section
-Using 68% confidence limit leads to a Using 68% confidence limit leads to a theoretical uncertainty of 2%theoretical uncertainty of 2%
- We know that this is too optimisticWe know that this is too optimistic-Larger improvements expected when Larger improvements expected when
using LHC data to constrain PDFsusing LHC data to constrain PDFs-Lower bound on cross-section prediction Lower bound on cross-section prediction
given by renormalisation and factorisation given by renormalisation and factorisation scale uncertainties: 0.6-0.8%scale uncertainties: 0.6-0.8%