associated sm higgs search in h bb,ww (*) final states @ atlas
DESCRIPTION
Associated SM Higgs search in H bb,WW (*) final states @ ATLAS. Huaqiao Zhang IHEP/CPPM On Behalf of the ATLAS experiment. Outline. Introduction ttH,H bb analysis Signature and preselection Event reconstruction and methods comparison Systematics and results ttH,HWW (*) analysis - PowerPoint PPT PresentationTRANSCRIPT
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 1
Associated SM Higgs searchin Hbb,WW(*) final states @ ATLAS
Huaqiao Zhang IHEP/CPPM
On Behalf of the ATLAS experiment
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 2
Outline
• Introduction• • ttH,Hbb analysis
– Signature and preselection– Event reconstruction and methods comparison– Systematics and results
• ttH,HWW(*) analysis– Signature and event selection– Systematics and results
• WH,HWW(*) analysis– Signature and event selection– Systematics and results
• • Summary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 3
SM Higgs searching in experiments
Higgs particles is the only particle that predicted in SM but not found yet experimentally LEP direct search exclude light Higgs below 114 GeV With recent Tevatron results, SM electroweak fit prefers Higgs less than 190 GeV
(including LEP) and 160 GeV without. Higgs Coupling to top quarks and W boson are important properties of Higgs
http://lepewwg.web.cern.ch/LEPEWWG/
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 4
The SM Higgs at the LHC
• Associate production:– Possible discovery channel: ttH,Hbb– Coupling measurement for Higgs mass
[120, 200] GeV => SM Higgs or not• Decay modes:
– For mH > 135 GeV/c2 H → WW(*)
dominates (BR : 0.91 @ 160 GeV/c2 )
– H→bb is the main channel for mH < 135 GeV/c2 (BR : 0.68 @ 120 GeV/c2 )
Difficult for this mass region: we need to associate more than one channel
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 5
The LHC and ATLAS
• ATLAS:– General purpose experiments– Classic detectors composed
mainly by 3 sub-systems• Inner tracker • Calorimeter system• Muon spectrometer
– Very good Trigger/DAQ System– Good e/g/m/tau/missEt/b-jets
identification ATLAS detector
• LHC:– Proton-Proton collisions @ 14 TeV – First run @ 10 TeV expected at the
end of this summer– Luminosity:
• Low luminosity regime ~1033cm-2s-1
– ~ 30 fb-1 between 2008 and 2011• High Luminosity regime ~1034cm-
2s-1
– ~ 300 fb-1 by 2014/2015
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 6
• ttH,Hbb: potential discovery channel for light Higgs boson, top quark Yukawa Coupling measurement– Hbb, t bjj, t bℓv t bℓv, t bℓv t bjj, t bjj
• Main backgrounds– tt+jets : reducible background using b-tagging– ttbb (EW/QCD) : irreducible background
• Strategy: Reconstruct the tt system to look at the rest 2 b-jets at the final state– b-tagging is one of the most important keys for this
channel• Jet pairing and JES are also very important
• Efforts mainly for lepton + jets channel– Isolated electron or muon trigger – One isolated lepton is required, Tau is not considered – Electron, muon: Acceptance, identification and isolation cuts– At least 6 jets, of which 4 tagged as b-jets:
• cone 0.4 jet algorithm– Done for 30 fb-1
ttH,Hbb Signature and preselection
cut ttH ttbb(EW) ttbb(QCD) tt+jets
1 Lepton (fb) 56.9 141 1356 63710
+ 6 jets (fb) 36.2 76.7 665 26214
+ 4 b-jets loose (fb) 16.2 23.4 198 2589
+ 4 b-jets tight (fb) 3.76 4.2 29.6 50.7
LO cross section (only tt+jets NLO+NLL)
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 7
• top quarks candidates are formed using one W boson candidate and one b quark
– Combination with |m(jj) – m(W)| > 25 GeV/c2 or |m(treco) – m(ttrue)| > 25 GeV/c2 are removed
• Combination minimizing a c2, based on the top quark masses, is chosen
• The two remaining b jets used to reconstruct the Higgs candidates
Cut-Based Reconstruction• Leptonic W reconstruction
– Force ln mass to the W mass– Solve 2nd degree equation to get
neutrino pz – 28% no solution → neglect imaginary
part
2
topjjb
2
topνb2 m-m+
m-m=
GeVGeV 1319
• Hadronic W reconstruction– After requesting 4 b-tag jets,
remaining jets are considered as light jets
– W boson candidates are formed of all combinations of light jet pairs
Invariant mass of reconstructed hadronic W
boson(red for good combination)
Invariant mass of reconstructed top quarks
(red for good combination)
ATLAS preliminary
ATLAS
preliminary
ATLAS
preliminary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 8
Multivariate Based Reconstruction
2
2
2
26
1
2
,2
5.1
175
1.2
425.80
/
1
lept
lepW
iinitialijet
ijet
ijet mm
p
f
125.0/888.0/
035.0/988.0/
TbP
TlightP
pP
pP
b
light
Pairing likelihood:• Using tt system kinematic properties
to build a pairing likelihood– 6 variables are used– b-jets and light jets are treated
separately– Choose the combination that
maximize the likelihood output
Constrained fit:• Fit jet pT and ETmiss to give the
mass of the top quarks• Pairing likelihood is formed using
the c2 output of the constrained fit together with b-tagging and kinematic quantities
– 14 variables are used– 3D likelihood is used to take
into account the correlations• Final selection likelihood is used
to separate signal and physics background
Invariant mass of reconstructed Higgs boson using likelihood for signal
sample(red for good combination)
f ijet : scale factor for jet momentum
ATLAS preliminary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 9
Comparison of Reconstruction Algorithms
Invariant mass of reconstructed Higgs boson using constrained fit for all
samples, a cut at -4.2 in Ls/b for better stat. significance 90 < mH < 150 GeV/c2
• S/sqrt(B) differences relatively small among the 3 methods
• Multivariate techniques increase
Higgs bb pair purity by ~5% • Analyses suffer from low purity
mainly coming from b exchange between top and Higgs– Wide distribution for reconstructed
Higgs mass – No clear signal peak on top of
background distribution– Combinatorial background dilutes
differences between ttbb and ttH sig
ATLAS preliminary
ATLAS preliminary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 10
Systematic uncertainties and results @ 30 fb-1
• Main systematic errors come from the JES, jet resolution and b-tagging efficiency
• Theoretical errors on background cross sections, especially top anti-top production, are sizable
Cut-basedLikelihoo
dConstrained
fit
JES 5% 14% 8%
Jet resolution 7% 5.5% 14%
b-tagging efficiency 20% 20% 20%
Light jet rejection 5% 3% 10%
All contribution 22% 25% 28%
main systematic errors, background sample
• Systematic errors on background higher than signal
• If nothing is done for the background extraction, the significance will decrease to: ~0.5 (mH=120GeV/c2)
S/sqrt(B+dB**2)ATLAS preliminary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 11
ttH,HWW(*) Signatures
gt
b
b
4 light jets
2 leptons
P
P
b
b
2 light jets
3 leptons
P
P
• ttH2b4j2l2Vl 6jets + 2samecharge lep. + MissEt
• Possible BKGs:– tt(1l), tt(2l), ttZ, ttW, tttt, ttWW,
ttbb...• ttH2b2j3l3Vl
4jets + 3leptons + MissEt• Possible BKGs:
– tt(2l),ttZ,ttW,tttt,ttWW,ttbb–
• tau not considered in signal• Complex final states=>Number
counting experiment– At least 2 neutrinos– Multi jets + Multi leptons + MissEt
• Background control important • Main BKG tt suppressed by lepton
isolation=>Lepton isolation is crucial
• ttZ suppressed by Z mass veto • QCD BKGs neglectable
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 12
• One high Pt Isolated lepton trigger– Eff: 82% (2L),91% (3L)– offline significance impact < 1%
• Phys. Obj. kinematic region: – |η|<2.5; Pt>15.GeV
• Electron identification:– Calorimeter seeded Alg.– Matching to ID track
• Muon identification:– Seeded in Muon Spectrometer – Combined with ID track
• Jet identification: cone based, size 0.4• Lepton Isolation:
– Electron:
– Muon:
• Number of leptons >= 2/3, Number of jets >= 6/4• Final tighten cut:
– Zveto, MuonPt>20GeV, Number of lepton==2/3, sameCharged/null
ttH,HWW(*) events selection
+ Cone Isolation
Eff:70.5%
Eff:92.7%
+ Tracker Isolation Calorimeter Isolation
fb σNLO 2L sel. 3L sel.
ttH2l 11.1 1.87±0.03
-
ttH3l 7.2 0.24±0.01
0.82±0.02
ttbar 833000
7.4±1.1 2.06±2.06
ttW+nj
226 2.04±0.07
0.55±0.04
ttZ 1440 1.49±0.09
1.12±0.08
ttbb 2693 0.55±0.18
-
tttt 3.2 0.07±0.01
-
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 13
Systematic uncertainties and results @ 30 fb-1
• Accuracy of σttH * BRHWW: Best at 160 GeV: 47.4%, if 10% uncertainties of BKG • Biggest uncertainties from JES and BKG (especially ttbar) normalization. • Without systematics, the statistical accuracy can achieve 28%• Systematics dominate the accuracy
ATLAS preliminary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 14
Multivariable methods for improvement
More work needed to improve
yx miss
physTmiss
measTmiss
jetsb b
physb
measb
jets j
physj
measj EEEEEE
,2
2
2
2
2
2
2
ttH,HWW(*) 2L signal
• Likelihood method on isolation– Projective likelihood estimator– Only 3 isolation variables used
• Mass constrained fit in ttH2L analysis– A system with 2 neutrinos (one virtual
W at Higgs mass <=160 GeV)– 11 free parameters
• 8 enter chi^2,3 from neutrino,5 Windows to determine combinatory
– 11 constraints• Only 4 equality, other 7 domains
• Extra 8 implicitly : 0.5 < cali_E < 1.
ATLAS preliminary
ATLAS preliminary
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 15
WH,HWW(*) Signature and preselection
• Higgs physics– Supplement to the Discovery– Necessary confirmation for the
property• Couplings to gauge boson
• WH,HWW(*) physics analyses– One High Pt isolated lepton
trigger– Two and three lepton final
states (2L/3L)• WHWWW(*)lv lv jj• WHWWW(*)lv lv lv
– Full analysis over WH against ttbar, WZ, ttW, ZZ
– Number counting experiments
• Number of jets and leptons• Same charge(2L), Zveto• Missing Transverse Energy(3L)• Lepton isolation
– Calorimeter isolation– Tracker isolation– Cone isolation– Impact parameter
• B-jet/jet veto• W mass range• Higgs-Spin cut
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 16
Xection S/B (fb) 0.68/1.7 0.28/0.4*
If 20% systematics WH2L WH3L Combined
Significance 1.6 1.9 2.4
67% 64% 47%
Preliminary results @ 30 fb-1
* Suffer from the statistics of MonteCarlo samples , BKG Systematics dominated by JES: 19%(2L), 17%(3L)HWW(*) coupling measurement need more/hard works
fb WH2L WH3L WZ tt ZZ ttW wbb W->lv+jet
σNLO 31.2 5.04 75083300
072.5 73.3 562 3.4*106
2L selection
0.68 - 1.2 0.43 0.05 0.020.000
60
3L selection
- 0.28 0.14 0.28 0.010.00
30 0
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 17
Summary• Studies on channels of ttH,Hbb, ttH,HWW (*) and
WH,HWW (*) with ATLAS real detector geometry full simulation
• Coupling measurement– ttH,Hbb (gt, Higgs~120GeV), potential discovery channel– ttH,HWW (*) (gt, Higgs ~160GeV)– WH,HWW (*) (gW, Higgs ~170GeV)
• Number counting experiments => BKG control is crucial– Mass peak reconstruction:
• possible in ttH,Hbb• Hopeless in ttH,HWW(*), not yet tried in WH,HWW(*)
– Background uncertainties• JES(Btaging in ttH,Hbb) dominate the uncertainties of BKGs• ttbar or ttjj normalization important• Data driven method needed• More advanced methods during 30fb-1 data gathering• Statistical part of systematics reduced with 30fb-1 data
• Still rooms for improvement of these analysis• Good feasibilities, more efforts needed
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 18
Bak up: Main Issues for ttH,Hbb Channel• After analysis Higgs purity is ~30%
– large tails and width in the bb invariant mass– No visible peak on top of physical background, side band extraction very difficult
• The main problem is the b-jets exchange between top quarks and the Higgs– b-jet from Higgs used for the hadronic top : ~36%– b-jet from Higgs used for the leptonic top : ~30%– Higgs boson reconstructed with only one correct b-jet: ~55% (The other wrong jet
is mainly coming from the top quarks)• Big uncertainty for the tt+jets cross section
– Extracting the background shape and normalization in data is crucial for this channel
– Using loose and tight b-tagging cuts looks promising– Background shape independent from b-tagging cuts
There is still room for improvement
JES, Jet resolutionEnergy flow algorithms
TriggerDedicated or combined trigger
(e.g. btagging, jets + soft leptons)
Cross section uncertaintiesBackground extraction from data
B-tagging/Light jet rejectionMore advanced taggers that should be well
understood after 3 years of data taking
Combinatorial backgroundMore powerful jet pairing
techniques (e.g. BDT)
2008-06-19 H. ZHANG(CPPM/IHEP) SUSY08 19
Bak up: Main Issues for ttH/WH,HWW(*) Channel
• Higgs reconstruction difficult– At least two neutrinos, one from Higgs W decay, one from top W
decay– With a virtual W, which could decay leptonic or hadronic– Large combinatory in complex final states
• Lepton isolation is crucial in background suppression– Several isolation methods are tried– Could improve by Multivariable methods, with better understanding
of real data
• Big uncertainty for the backgrounds cross section normalization– Theoretical uncertainties could improve with understanding of real
data
• Statistical part of systematics could be reduced with full 30fb-1 real data