1. Study of VBF H->tt->lep+had
Junichi Tanaka
ICEPP, Univ. of TOKYO
Discovery of Higgs and Supersymmetry
to Pioneer Particle Physics in the 21th Century
24th Nov. 2005

2. Outline Introduction Study of 3rd jet by ME event generator
Study of fake BGs
Summary
Prospects : Discovery Potential of SM Higgs
24th Nov. 2005

3. SM Higgs Production & Decay @ LHC
Excluded by LEP
Production
Decay
t+t-
VBF
LEPEWWG2005S
114.4 < mH < 219 C.L.
VBF H->tt at low-mass region (mH = ~120GeV)
Large production cross section and branching ratio: sxBR ~ 300fb
24th Nov. 2005

4. Vector Boson Fusion Process
Rapidity gap f h
PRD59(1999)14037
hep-ph/
Characteristics
High Pt jets in forward region
Rapidity gap Right fig : h*3 = h of 3rd jet – average of h of 2 forward jets
Signal … a dip in the center
BG … a peak in the center
-> jet activities depend on structure of color flow.
(EW)
Treatment of 3rd jet is important in this mode!
24th Nov. 2005

5. ATLAS
ATLAS Scientific note:
Eur.Phys.J.C32S2:19-54,2004
BG by Comphep
VBF H->tt
VBF H->tt is one of promising channels near the LEP limit. (left fig)
Reconstruct M(tt) by using missing Et and observe a peak (right fig) on the right side of the Z mass peak.
“VBF H->tt” is one of discovery channels at the low mass region.
24th Nov. 2005

6. Motivation
Improvement and progress of our tools since the previous studies (Eur.Phys.J.C32S2,19,2004, the pervious slide)
1) Events with multi jets can be generated by the recent ME event generators.
3rd jet was evaluated by Parton Shower (PS) model in the previous studies.
We evaluate 3rd jet by the ME event generator.
2) Environment of reconstruction and full simulation, which will be used in 2007, is available.
Fake rate can be estimated from the full simulation, which is based on Geant4.
Large QCD backgrounds, which have not been considered in the previous studies, are checked by using results of the full simulation.
24th Nov. 2005

7. Type of Simulation Full simulation Fast simulation
Based on Geant4
Initial layout
Used for detail studies
Development of reconstruction tools, for example, tracking, reconstruction of missing Et, Particle ID tools (e, m, tau, b, g).
Fake rates, crack effect for missing Et and so on
Take much time, typically, ~15mins/event with 3GHz CPU.
Fast simulation
Based on parameterized performance of full simulation results
Used for the BG study, where we need high statistics.
Most of today’s results are obtained by the fast simulation.
24th Nov. 2005

8. Study of 3rd jet by ME event generator
24th Nov. 2005

9. ME ALPGEN + PS Alpgen (M.L.Mangano) is one of the ME event generators.
Studies using ME event generators becomes popular.
ME has an advantage for producing high Pt jets.
To remove collinear and soft divergence, Pt and Rjj are required at parton level.
Soft and collinear kinematical regions are covered by PS.
We produced exclusive samples for Z+njets(n=2,3,4) with modified ALPGEN ver1 (small weight events can be produced).
“Z+njets” is a dominant BG.
MLM method is used to avoid double counting. The similar method is officially introduced from ver2.
24th Nov. 2005

10. To Avoid Double Counting
We use the following method (“our MLM method”) to avoid the double counting of events: (not CKKW method)
Veto events having jets (Pt>20GeV) at the outside of cone (dR=0.4), whose center is defined by the direction of a ME-parton.
ME-parton
Rec-jet
Good
Bad
See T.Sasaki’s talk later
24th Nov. 2005

11. Pt and h Distributions 3rd jets with high Pt increase.
Alpgen sample(ver1), after lepton cut, normalized to 1fb-1.
Pt of 3rd jet
eta of 3rd jet
preliminary
“Z2j”
“Z2j”+”Z3j”
“Z2j”+”Z3j”+”Z4j”
with MLM method
to avoid double counting
preliminary
log10Pt
25GeV
100GeV
3rd jets with high Pt increase.
3rd jets become more central. It has a large effect on Central Jet Veto.
Pt and h of 3rd jet depend on the number of jets. Cannot ignore them.
24th Nov. 2005

12. CJV Survival Probability
Central Jet Veto (CJV) : Veto events having additional jets with a “given” Pt in the rapidity gap.
preliminary
No additional jet
in the rapidity gap
jet t t
jet f
Rapidity gap h
(additional jets in the rapidity gap)
Survival probability of “Z+2j” is higher than “Z+3j” and “Z+4j” due to smaller jet activity in the rapidity gap.
~40% of Z+njets (n=2,3,4) remains in total in case of Pt=20GeV.
24th Nov. 2005

13. M(tt) Distribution After all the selection criteria
L=30fb-1
After all the selection criteria
A amount of remaining Z+2j is large due to large cross section and high CJV survival probability.
Z+3j and Z+4j still remains. They are not negligible.
Since resolution becomes worse as the number of jets increases, their tails can contribute to BGs.
Z+2j
Z+3j
Z+4j
preliminary
Signal
region
24th Nov. 2005

14. Study of fake BGs
24th Nov. 2005

15. Origin of Fake Lepton
Since cross section (x-sec) of events with QCD jets is much larger than that of the signal (x-sec~300fb), we suffer from these events, which have fake leptons from jets. W j j j -> W t j j : x-sec ~ >10nb j j j j -> e/m t j j : x-sec ~ >10mb
Fake Tau
1 or 3 p+- with p0, which have most of energy of a jet
Fake Electron
Pair production of gamma (from p0)
Dalitz decay
p+- and p0 with E/p~1
B/D semileptonic decay
Fake Muon
pion punch through
pion decay in flight
24th Nov. 2005

16. Tau Reconstruction and ID
Hadronic decay: 1prong ~50% and 3prongs ~15%
We need to identify tau-jets from many QCD-jets.
Charged tracks : 1 or 3
Photons from p0
Reconstruction : two ways at ATLAS
1) Cell based : start from cells, which is the similar as cone jet algo.
2) Track based : start from charged track(s)
Properties of tau-jet : for identification
1 or 3 charged pion tracks
Narrowness and Isolation
24th Nov. 2005

17. Tau Rec/ID Performance
Cell based method
“Fake rate of 10-2” means
we misidentify one QCD jet
as a tau in 100 QCD jets.
Depend on Pt and
type of jet.
Gluon fake rate is better
than quark fake rate due to difference in narrowness.
Light Quark
Gluon
fake rate
preliminary
20<Pt<40GeV
40<Pt<60GeV
60<Pt<100GeV
100<Pt<140GeV
efficiency
This result is parameterized and then is used in fast simulation.
At least we need ~50% eff. with these fake rates for the discovery with L=10fb-1.
24th Nov. 2005

18. e/m Fake Rates
Fake rates of electron and muon from jets are obtained from QCD di-jet samples using the full simulation.
See O.Jinnouchi’s talk later.
Typical performance
Jet -> e : ~10-3
Jet -> m : ~10-4
These results are taken into account in the fast simulation to evaluate BG contributions.
24th Nov. 2005

19. W+3jets & 4jets W j j j -> W t j j j j j j -> e t j j BG
Corresponds to L ~ 70fb-1
Corresponds to L ~ 10fb-1
W+3jets
4jets BG
x-sec(fb)
W+3jets
0.26
4jets
0.30
Z+njets
0.13
preliminary
fake(e->jet) : 10-3
preliminary
(Signal region: GeV)
W j j j -> W t j j j j j j -> e t j j
W+3j and 4jets remain with the same order of Z+njets.
Improvement of tauID for both cases and decrease of fake electron for 4jets are critical to decrease these BGs.
24th Nov. 2005

20. M(tt) and Significance
Signal (m=120)
very preliminary
Z+nj(QCD)
M(tt) and Significance
Z/ttjj(EW)
L=30fb-1
tt, WWjj A
a kind of BGs
is the same as
our paper(page5). B
+”W+3jets” C
+”jjjj” 10
sBG=0.18fb
sBG=0.43fb
sBG=0.73fb 10 10
sSIG=0.45fb
S/B
S/B0.5
L=10fb-1
L=30fb-1 A
2.5
3.3
5.8 B
1.0
2.1
3.7 C
0.61
1.6
2.9
We assume that lepton(e,m) efficiency is 85%.
Significance becomes worse by a factor of 2 due to the new backgrounds.
We need to improve tauID/fake electron and to optimize cuts.
Signal region: GeV
24th Nov. 2005

21. Summary
VBF H->tt->lep+had is important for the SM Higgs discovery in the low mass region (mH~120GeV).
We evaluated 3rd jet by the ME event generator.
There are differences in Pt and h of 3rd jet by the number of jets.
Central jet veto survival probability and tail of M(tt) depend on the number of jets.
We should use ME generators for multi-jets studies as much as possible.
We studied fake BGs.
W+3jets and 4jets become large BGs for the lep+had channel.
We need to improve our tauID and other PID performance.
We must understand QCD BGs for any LHC physics since their x-sec is very large in the LHC. (maybe beyond our expectation…)
24th Nov. 2005

22. Discovery potential of SM Higgs
ATLAS-PHYS
preliminary
NB: not include our latest results.
10fb-1
ATLAS
Today
L=10fb-1:
VBF tt(ll+lh) ~ 2.6
gg ~ 5.0
@ mH=120GeV
200
120GeV
Two promising channels at mH=~120GeV!
We are preparing for this discovery!
24th Nov. 2005

23. Backup
24th Nov. 2005

24. ATLAS (A Toroidal LHC ApparatuS)
E,P resolution
(P ~ 100GeV) :
m ~ 2%　
e, g ~ 1.5%
Jets ~ 8%
40MHz beam crossing
Readout channel = M channels
Raw data = 320Mbyte/sec
(1TB/hour)
Inner tracking system : 2T solenoid magnet
Liq. Ar EM calorimeter
Muon spectrometer : air-core troidal magnet
24th Nov. 2005

25. Summary of SM Higgs Studies
Production
Decay
Mass region and purpose
Gluon Fusion
H -> gg
GeV
Mass
H -> ZZ-> 4 l
GeV
Discovery, Mass, Spin, Coupling
H -> WW
GeV
Discovery
Vector Boson
Fusion
H -> tt
Discovery, Mass, Coupling
GeV
Discovery, W coupling
Discovery, Mass
H -> bb
Yb coupling
ttH
GeV
Yt coupling
GeV WH
GeV
ATLAS
24th Nov. 2005

26. Hadronic tau decay mode
Branching ratio of H->tt :
BR(lep-lep) = ~12%
BR(lep-had) = ~46%
“Lep-had” mode plays an important role for the Higgs discovery.
factor of ~4
Mtt Distributions of VBF H->tt
ＭＨ=120GeV
ATLAS
(em only)
ATLAS
24th Nov. 2005

27. Fake rates Pt(GeV) Jet -> e Jet -> m BJet->e BJet->m
very preliminary
Pt(GeV)
Jet -> e
Jet -> m
BJet->e
BJet->m
BJet->t
15-35
19+-8
< 3.2
25+-9
35-50
10+-7
< 5.1
50-80
80-130
< 4.2
< 5.2
200-
< 2.4
in 10-4
24th Nov. 2005

28. Simulated data W(->ln)+3jets 4jets Alpgen ver1
Jet condition : Pt>15GeV, |eta|<5, Rjj>0.7 Max(J1,J2)>30GeV, M(J1,J2)>200GeV, deltaEta(J1,J2)>2
x-sec = 1.4nb … very large!
4jets
Jet condition : Rjj>0.7
Forward region : Pt>30GeV, eta1xeta2<0, |eta1-eta2|>2.5 Mjj>350GeV, |eta|<5.0
Central region : Pt>20GeV, |eta|<2.7, Mjj>80GeV
x-sec = 2mb … very large!!!
24th Nov. 2005

29. qqH -> qqtt -> qq lep had
tau->e
FWJet1
FWJet2
tau->h
(1prong+pi0)
Missing Et
We’ll look this type of event in the real data !!!
24th Nov. 2005

30. ME+PS Z2j sample Z3j sample Z4j sample z2j without MLM method
z2j with MLM method
+ z3j without MLM method
Z4j sample
+ z3j with MLM method
+ z4j without MLM method
24th Nov. 2005