Lei Zhang on behalf of HSG5 1 st July 2013 Unblinding approval
for VH(bb) 1
Slide 2
Outline of this talk Brief overview of the analysis strategy:
Event selection Background modeling Statistical interpretation
Questions from the last unblinding talk Conclusion 2
Slide 3
Trigger and Event Selection Zero lepton: MET triggerOne lepton:
1-Lepton trigger + MET trigger (for muon channel) Two lepton:
1-Lepton trigger + 2-lepton trigger Common selection: At least 2
jets in |eta|45GeV, Pt2>20GeV Veto forward jets with Pt>30GeV
(0/1 lepton) Lepton: Loose (Pt>10GeV) and Tight/Medium
(Pt>25GeV) 2 b-tagged jets (MV1@70%) PtV
(GeV)0-9090-120120-160160-200>200
dR(jet,jet)0.7-3.40.7-3.00.7-2.30.7-1.8120 GeV QCD rejection cuts:
(Et miss, jets) > 1.5 (Et miss, Pt miss )< /2 (Et miss, bb)
> 2.8 1 lepton: 1 tight leptons and No additional lepton QCD
rejection cuts: Pt W 40GeV Pt W >200, Et miss >50GeV 2
lepton: 1 medium + 1 loose lepton No additional lepton 81GeV < M
ll < 99GeV Et miss
Top p T reweighting and systematic
https://cds.cern.ch/record/1470588 Used Powheg-Pythia with Perugia
tune Very small difference. Top p T mean shifted by 1.85% ttbar
events: reweight truth top p T to unfolded 7TeV data applied to
both 7 and 8 TeV => checked absence of bias alters mass as well
as p T 5
Slide 6
W+jets dPhi modeling correction W+jets Wjets Wp T @8TeV Before
reweighting After reweighting and dR cut Still some residual
mismodeling at high Vp T => add a systematic of 2% and 4% in the
last two bins Correction of V+jets dPhi to data improves agreement
in other variables, like V p T Apply a systematic of 50% the
correction Decorrelated in flavor and jet multiplicity => 12
parameters Z+jets case could be found in backup. 6
Slide 7
dPhi generator level studies on Wbb Cross checked if Sherpa vs
aMCatNLO discrepancy can be covered by physics motivation No big
effect on dPhi from pdf, renormalisation and factorisation scale
variations in aMCatNLO aMC@NLO vs Powheg vs Sherpa Renormalization
Scale PDF sets Factorization Scale Generator studies of Wbb
dPhi(b,b) distribution => help understand dPhi mismodeling in
Sherpa Clear slope in comparison between NLO MCs and Sherpa
(multileg) 7
Slide 8
Vpt shapes: 1-lepton Pt binning is: 0-90, 90-120, 120-160,
160-200, 200- The Wl normalization is float. 8
Slide 9
Fitting model 2jets, 1-tags3jets, 1-tags2jets, 2-tags3jets,
2-tagsTop emu CR 0-leptonNorm Shape - 1-leptonNorm Shape -
2-leptonNorm Shape Norm Flavor composition treatment: Correlate:
Wcc, Wbb and Wbl with 30% relative uncertainty; Keep separate Wcl ;
Fix Wl with 10% uncertainty. Analogous splitting for Z+jets 9
Slide 10
Constraining Systematics Uncertainties Theory/Modeling: Mbb
shape Vpt and Top pt Jet multiplicity Migration between channels
through lepton sysetmatics Experimental: JER/JES B-tagging Lepton
ID MET For each channel For each bin For each P T V bin 10
Slide 11
2011 and 2012 Fitting parameter correlation (I) 11
To do list from last talk Understand precisely the origin of
the constraint on the JER. Is its modeling sufficient? Investigate
all its correlations. Clarify all correlations of the nuisance
parameters in Chiara's list. Top pT, 3/2 ratio MJ normalisation
DPhi BTagC5Effic, BTagB2Effic JES constraints (1-lepton,
fluctuation ) Jet Flav composition (David) MET Trigger systematic (
sys itself is small ) Add the multijet (in the 1 lepton analysis)
to Chiara's list of parameters to be investigated. Perform a
complete comparison of the NPs between 7 TeV and 8 TeV. Give the
diboson results for 7 TeV. Thanks Eilam and Marumi 13
Slide 14
NP check on 7TeV fit Top Ten Impact on mu 14 To be updated
Slide 15
NP check on 8TeV fit 8TeV 15 Top Ten Impact on muhat 15 To be
updated
Slide 16
B-tag NPs check 8TeVB-tag 16 7TeV
Slide 17
B-tagging check Eigen state scale factor variation at the jet
pt base. 17
Slide 18
Unfolded b-tagging efficiency 8TeV7TeV 18
Slide 19
JES NPs check 8TeV 19 7TeV
Slide 20
MET trigger Trigger scale factor: Using leptonic trigger to
select Wmunu events Taking the fraction of the events passing MET
trigger as efficiency. Scale factor = Eff data /Eff MC Statistical
uncertainty: due to statistical error from the fit Process and
generator: comparison between W+jet, Z+jets and ttbar. 20
Slide 21
The Jet resolution uncertainty we are using could be too
conservative, which could cause the constraining in fitting. 21 Jet
resolution
Slide 22
JES fluctuation and constrain 22 Fluctuation in the Mbb shape
Remaining effect after smoothing procedure.
Slide 23
Smooth model Smoothing can reduce the constrains a lot.
Re-bining the histograms until the stat. uncertainty of the ratio
nominal/syst. less than 5%. After smoothing, it still leaves large
variations in some bins (maybe why some are still constrained)
23
Slide 24
Dphi NPs check 8TeV 24 7TeV WDPhiWDPhi_3jetZbDPhi
Slide 25
Dphi sys check 8TeV fit: 8TeV fit: combined 25
Slide 26
26 Separate fit in each lepton channels
Slide 27
Scale factor 8TeV 27 7TeV
Slide 28
8TeV 28 7TeV Normalization NP check
Slide 29
Top Pt NP check Top Pt 8TeV 29 7TeV
Slide 30
Top Pt and 3jet/2jet ratio check Top pT, 3jet / 2jet ratio:
Several tests have been done: Split NP for each N Jet, M btag bin
(as the bottom plot). Split NP for High and Low Vpt Mbb. 3jet /
2jet of ttbar: increase error by factor of 10 More detailed study
can be found:
https://indico.cern.ch/getFile.py/access?contribId=1&resId=1&materialId=slides&confId=260550https://indico.cern.ch/getFile.py/access?contribId=1&resId=1&materialId=slides&confId=260550
Proposal: de-correlated the normalization effect and Mbb shape
effect. 30
Slide 31
Mbb NP check Top Mbb Mbb Zbb Mbb 8TeV Mbb 31
Slide 32
m bb shape systematics Samplem BB NP Wt ISR/FSR Hadronization
DR vs DS s-chan t-chan ttbarPowheg Wbb Sherpa/aMC @NLO Wbl Wcc Wcl
Sherpa/aMC @NLO Wl Zbb Data/MC comparison in 0Tag Zbl Zcc Zcl Zl MJ
template changes 32
Slide 33
Summary of problematic NPs 8TeV 33 MJ normalization: 7TeV
Slide 34
NP check (need plots!!) MJ normalization: Anti-correlated with
Whf Add ranking later 34 To be updated
Slide 35
35
Slide 36
36 Correlation check: 7TeV
Slide 37
High correlation NPs 37 Correlation check: 7TeV
Slide 38
38 B-tag
Slide 39
39 JES
Slide 40
40 Modeling
Slide 41
41 Mbb NP
Slide 42
42 Norm
Slide 43
43 Scale factor
Slide 44
Correlation check: 8TeV 44
Slide 45
45 Correlation check: 8TeV High correlation NPs
Slide 46
46 B-tag
Slide 47
47 JES
Slide 48
48 Modeling
Slide 49
49 Mbb
Slide 50
50 Norm
Slide 51
51 Scale factor
Slide 52
7-8 TeV combined NP check Problematic NPs 52
Slide 53
Correlation 53
Slide 54
To be updated 54 Recap of the correlation check
Slide 55
B-tag: C5 has been pulled due to the large fraction of the
charm composition of top background in the high Vpt bins. JER: due
to the conservative JER uncertainty we are using. JES: some of them
may come from the fluctuation. MET trigger NP has been pulled, but
considering its small uncertainty, the overall impact should be
small. Dphi Top Pt/Mbb : To be updated 3jet/2jet ratio: 55 Recap of
NP check
Slide 56
Scale factor and summary of muhat value Scale factors: Summary
of muhat values for diboson 56
Slide 57
Post-fit: M JJ in 0-lepton 57 MET>200 GeV 2 tag 2
jetMET>200 GeV 2 tag 3 jet In clusive MET, 1 tag 2 jetIn clusive
MET, 1 tag 3 jet
Slide 58
Post-fit: M JJ in 1 lepton 58 Vpt>200 GeV 2 tag 2
jetVpt>200 GeV 2 tag 3 jet In clusive Vpt, 1 tag 2 jetIn clusive
Vpt, 1 tag 3 jet
Slide 59
Post-fit: M JJ in 2 lepton 59 In clusive Vpt, 1 tag 2 jetIn
clusive Vpt, 1 tag 3 jet Vpt>200 GeV 2 tag 2 jetVpt>200 GeV 2
tag 3 jet