Update on Lepton+Jets+MET mSUGRA LM-1: Systematic Uncertainty Decomposition
description
Transcript of Update on Lepton+Jets+MET mSUGRA LM-1: Systematic Uncertainty Decomposition
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
1
Update on Lepton+Jets+METmSUGRA LM-1:
Systematic Uncertainty Decomposition
Rick & Bobby
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
2
Motivating Example:MET (uncorr) From Towers
Total BackgroundLM-1 Signal
Eve
nts
MET [GeV]
Rapidly falls
Flatter
Since this is a Counting Experiment,look at cumulative events passing cuts…
HLT Applied
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
3
Motivating Example:MET (uncorr) From Towers
Total BackgroundLM-1 Signal
Cu
mu
lati
ve N
um
. E
ven
ts P
assi
ng
Cu
t
Cut on MET [GeV]
Rapidly falls: - B / MET large
Flatter:S / MET small
Naïve expectation:Small systematic effectsaffecting MET might lead to large differences in S / S+B
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
4
Uncertainty Definitionsconsidered so far…
• Statistical:• S+B
• Systematical:– Finite MC Statistics
• NW / Ngen
– QCD level (not shape)• 50%
– Jet Energy Scale• 10% ET• (affects also MET)
– Jet Energy Resolution• Additional one ET
smearing (ET taken from the DAQ TDR)
• (affects also MET)
• For Jet Systematics
– N = number of events passing cuts before systematic effect applied
– Nsyst = number of events passing cuts after systematic effect applied
– Uncertainty [number of events]:
B = N – Nsyst
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
5
Uncertainty Definitionsconsidered so far…
• Statistical:• S+B
• Systematical:– Finite MC Statistics
• NW / Ngen
– QCD level (not shape)• 50%
– Jet Energy Scale• 10% ET• (affects also MET)
– Jet Energy Resolution• Additional one ET
smearing (ET taken from the DAQ TDR)
• (affects also MET)
• For Jet Systematics
– N = number of events passing cuts before systematic effect applied
– Nsyst = number of events passing cuts after systematic effect applied
– Uncertainty [number of events]:
B = N – Nsyst
Quick side remark:10% seems conservative for this study (we assumes 10 fb-1)
Plan to vary JES systematic according to expected dependence on integrated luminosity
8-10% for L < 1fb-1~3% at some point after 1fb-1
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
6
Uncertainty Definitionsconsidered so far…
• Statistical:• S+B
• Systematical:– Finite MC Statistics
• NW / Ngen
– QCD level (not shape)• 50%
– Jet Energy Scale• 10% ET• (affects also MET)
– Jet Energy Resolution• Additional one ET
smearing (ET taken from the DAQ TDR)
• (affects also MET)
• For Jet Systematics
– N = number of events passing cuts before systematic effect applied
– Nsyst = number of events passing cuts after systematic effect applied
– Uncertainty [number of events]:
B = N – Nsyst
Quick side remark:10% seems conservative for this study (we assumes 10 fb-1)
Plan to vary JES systematic according to expected dependence on integrated luminosity
8-10% for L < 1fb-1~3% at some point after 1fb-1
Many systematics
not yet c
onsidered
(PDFs, m
uon ID eff,
etc)
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
7
Systematic UncertaintyDecomposition
Statistical MC StatisticsQCDJet E ScaleJet E ResolutionTotal
Un
cert
ain
ty [
nu
mb
er o
f ev
ents
]
Cut on MET [GeV]
HLT Applied
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
8
Toy “Significance” Decomposition
• Statistical only– S / S + B
• Statistical + MC Statistics– S / S + B + BMC
2
• Statistical + QCD level– S / S + B + 0.5BQCD
• Statistical + Jet Energy Scale– S / S + B + BJES
2
• Statistical + Jet Energy Resolution– S / S + B + BRES
2
• Total– S / S + B + BMC
2 + 0.5BQCD + BJES2 + BRES
2
NOTE: The following plots have HLT applied, MET cuts applied
NO OTHER CUTS APPLIED! significant optimisation still possible (likely)
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
9
Toy “Significance” DecompositionStatistical onlyStatistical + MC StatisticsStatistical + QCDStatistical + Jet E ScaleStatistical + Jet E ResolutionTotal
“Sig
nif
ican
ce”
Cut on MET [GeV]
HLT Applied
HLT applied,MET cuts applied
NO OTHER CUTS APPLIED! significant optimisation still possible (likely)
Inclusion of different systematics tend to prefer higher MET cut values
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
10
Observations
• Systematics in toy “significance” definition:– Shift to higher MET cut values preferred– Overall reduction in toy “significance” estimation
• JetMET Uncertainties – Current choices dominate (no JetMET recommendation yet)
• Jet Energy Scale: 10% ET• Jet Energy Resolution: additional one unit of ET
smearing
• Finite MC Statistics Uncertainty– Important, but does not currently dominate!!
• Finite MC Statistics: NW / Ngen
• QCD Level (not-shape) Uncertainty – Not important compared with others (but, prob. too optimistic)
• QCD level: 50%
16.09.2005 R. Cavanaugh, SUSY/BSM Working Group Meeting
11
Conclusion• Everything still very preliminary!
– Healthy discussion about Finite MC Statistics Systematic arising from 06 Sept. SUSY/BSM Meeting
• Selection optimisation underway…– Toy “significance” estimate currently being used
• Inclusion of systematic effects being studied– Likelihood ratio method being implemented
• Will be used for final results• Early observations suggest that
– Systematics can have a strong effect– JetMET systematics are important to understand and correctly
estimate• Current assumptions may be conservative? • Need official recommendation from JetMET assuming 10 fb-1 data
collected • We are still collecting additional statistics
– Y. Pakhotin has produced trees for all available ttbar!– We are working to include more W+jets– Will include Alpgen samples when available