Study of Standard Model Backgrounds for SUSY search with ATLAS detector

1
Study of Standard Model Backgrounds for SUSY
searchwith ATLAS detector

• Takayuki Sasaki,
• University of Tokyo

2
Outline of my talk

• 1 Event topology of SUSY signal
• 2 Background estimation with Matrix Element
• 3 Studies of Missing Et
• (1)Dead material (crack region)
• (2)Topology dependence
• 4 Conclusion

3
1 SUSY Event Topology

• SUSY Event Topology is multijets missing Et (
  leptons)

• are produced in pp collision.
• decay into and high pt jets are
  emitted.
• They decay into .it makes missing Et.
• Sometimes lepton is emitted .

ATLAS preliminary
Missing Et
multijets
?

• For SUSY discovery
• High Pt Jets
• Missing Et
• are important.

f
4
Background processes

• 4 types of background are considered.
• There are 2 types of QCD background.
• QQjj (Qb,c)
• It is heavy flavors
• and including real missing Et.
• multijets
• It is light flavors
• There is no real missing.Just fake missing Et

5
QCD Background

• QCD background is categorized to 2 types
• QQjj (Qb,c) Real missing
• ? from semileptonic decay of b/c makes missing
  Et.
• multijets (light flavor)Fake missing
• Limited resolution of energy measurement makes
  missing Et (?See right figure)
• QCD has huge cross-section. So, fake missing Et
  should be treated carefully. This is important
  and difficult issue.

ATLAS preliminary
c
l
?
Fake Missing Et
b
Real missing
?
f
Fake missing
f
QCD 2 jets event Incomplete measurement of the
energy makes fake missing Et.
6
2 BG estimation with ME
ATLAS preliminary
g
Top-pair Highest Pt of the additional jet
g
t
g
t
ME (Alpgen)
PS (Pythia 6.2)
Pt(GeV)

• Parton Shower (PS) is used in the previous study.
  But PS has some problem.
• For example, Jet is emitted in top
  pair-production.This figure shows Pt
  distribution of this jet.
• Hard jet is not emitted in Parton Shower.
• PS is not good approximation at high Pt region.
• BG estimation using PS model is underestimated in
  high Pt region.

7
Background Generation with ME

• Generation
• High Pt partons are generated with ME(Alpgen).
• Collinear and soft regions are covered with
  PS(Pythia) model.
• Pt ordering is applied.

ME PS
t
t

• But, there is double counting problem.
• Mangano Matching is applied in order to remove
  double counts.

Detail is Manganos HP http//mlm.home.cern.ch/mlm
/
8
Mangano Matching
Z???4jets
ATLAS preliminary
PS

• parton generated with ME
• ? jet activity

Jet was emitted collinearly
ME
One parton divided into 2 jets

• We applied Mangano method
• Jet should be matched to the parton generated
  with ME (R0.7) except for the soft and collinear
  regions.
• For example, let's consider Z4jets sample
• Blue show perfect matching between ME parton and
  jet.
• Soft jet was emitted collinearly. ?Matched
  (Accepted)
• One parton divided into 2 jets. (outside ME cone
  0.7) ? Not Matched
• Such event should be covered with 5jet ME.
  (double counting)?this event is discarded.

Jet map in (?,f) plain
9
Event selection

• SUSY Event Topology is multijets missing
  (leptons).
• There are two discovery channels.
• 0 lepton mode
• Large branching ratio. But Background is large.
• 1 lepton mode Small branching ratio .But BG is
  expected to be small.

• 0 lepton mode
• No lepton
• At least 1 jet Ptgt100GeV
• At least 4 jets Ptgt50GeV
• Missing Etgt100GeV
• Transverse Sphericity gt0.2
• 1 lepton mode
• 1 lepton (e,µ Pt gt10GeV)
• At least 1 jet Ptgt100GeV
• At least 4 jets Ptgt50GeV
• Missing Etgt100GeV
• Transverse Sphericity gt0.2
• Transverse mass between lepton and missing Et
  gt100GeV
• (In order to suppress WNjets Background)

lepton
10
Result 0 lepton mode
Effective mass is sum of leading 4jet scalar Pt
and missing Et
ME study
ATLAS preliminary
These figure show effective mass distribution
after the standard event selection.
Count /400GeV/10fb-1

• BG increases by factor 25 than PS study.
• BG slope is similar to signal.
• QCD BG is the same order of the other background
  processes

(GeV)
PS study
In order to control QCD BG, It is important to
study fake missing Et.
11
The other SUSY scale
0 lepton mode
ATLAS preliminary
ATLAS preliminary
Count /400GeV/10fb-1
Count /400GeV/10fb-1
(GeV)
(GeV)
Event selection should be optimized depending on
SUSY mass scale.
12
Result 1 lepton mode

• Clear excess can be observed.
• BG is dominant.
• Z?nn, QCD background can be suppressed
• 1 lepton mode has better discovery potential than
  0 lepton mode.

ATLAS preliminary
top BG
Count /400GeV/10fb-1
(GeV)
13
3 Missing Et
ATLAS preliminary

• Clear excess can be observed ( Missing Et
  gt800GeV) ?Missing Et is very important for SUSY
  search

0 lepton mode Missing Et distribution
Count /10GeV/10fb-1
Missing Et (GeV)
14
Fake missing Et
ATLAS preliminary

• Missing Et should correspond to
• But, limited resolution of jet also makes missing
  Et.
• This is fake missing Et
• It is very danger BG.
• Since QCD cross section is huge.
• Fake missing depends on
• Dead material
• Event topology

Fake missing Et distribution
QCD 4jets sample (Full Simulation)
Fake Missing Et(GeV)
Long tail appears close to 1TeV
15
3-1 Dead material in ATLAS detector
ATLAS detector

• ?1.41.6 is boundary of Barrel and Endcap.
  There are a lot of dead materials in front of
  calorimeter (cables, services)Energy loss is
  expected to be large in these dead materials.
  Then energy measurement becomes worse.
• We define crack region as ?1.41.6

Dead material
16
Missing Et resolution in crack region
ATLAS preliminary
QCD di-jet sample
?Et15001600GeV

• We estimate effect of the limited energy
  resolution with full simulation.
• This figure shows the fake missing Ex
  distribution.

All eta Crack
Normalized entry
Fake missing Ex (GeV)
( Ex is x-axis component of Et)
Missing Et resolution is worse in the crack
sample.
17
Double Gaussian fit
ATLAS preliminary
QCD di-jet sample ?Et15001600GeV

• These figures show fake missing Ex distributions
  with single gaussian fit and double gaussian fit.
• Missing Ex resolution is not single gaussian.
  Single gaussian fitting fail to fit in both
  central region and tail.

Linear scale
Single gaussian Double gaussian
ATLAS preliminary
Log scale
18
Missing Ex Resolution (double gaussian fit)
narrow sigma (GeV)
broad sigma (GeV)
ATLAS preliminary
40
ATLAS preliminary
40
All eta Crack 20 worse
All eta Crack
20
20
10
10
0
1000
2000
0
1000
2000
0
?Et (GeV)
?Et (GeV)

• These Figures show result of double gaussian
  fitting. Left figure shows narrow component.
  And, right figure show broad component . We
  need more study.

19
3-2 Topology dependence of Missing Et
Resolution of jet energy is limited. Miss-measurem
ent of each jet energy make fake missing.
Truth jet Measured jet Fake missing Et
Fake missing Et distribution depends on Jet
multiplicity.
I use 2 samples Z3jet,Z6jet
?
?
Z3jets
Z6jets
spherical and isotropic
rectilinear
20
Fake missing Ex distribution
ATLAS preliminary
?
Z3jets (Z?nn) Z6jets (Z?nn)
Z3jets
rectilinear
Normalized entry
?
Z6jets
spherical and isotropic
Fake missing Ex (GeV)

• Tail shape is clearly different. Z3jets has
  larger tail than 6jets.

Fake missing Et comes from limited resolution of
each jet.Miss-measurement cancel out in 6jet
sample,since 6 jets sample is more spherical and
isotropic.
21
4 Conclusion
For SUSY search ,Jet and missing Et are important.

• We estimated Background using ME generator.
• ME-PS matching is performed with Mangano method.
• The contribution of the background increase by
  factor 2-5.
• 1lepton mode is important.
• Missing Et tail and resolution is very important
  for QCD Background
• Resolution become worse in crack region.
• Topology dependence

Detail study in realistic condition is in progress
22
Discovery potential
m1/2(GeV)
PS Study
ATLAS preliminary
ME Study
1 lepton mode
m1/2(GeV)
m0(GeV)
These figures show Discovery potential for an
integrated luminosity of 10fb-1 after Cut
Optimization. Discovery potential of ME study is
same as PS study. We can find 2TeV scale SUSY
for 10fb-1.
23
Event display (SUSY)
EM Calorimeter Hadron Calorimeter Muon System
jet
Missing Et
jet
µ