VariableR Reclustering in Multiple Top Quark Events Jeremy Hyde 1
Outline
• Introduction
• LHC and ATLAS Detector
• Jets
• Boosted Objects
• Reclustering
• VariableR vs Fixed
• Top Quark Reconstruction
• W Boson Reconstruction
• Comparison 2
Absract
VariableR jet reclustering is an innovative technique that allows for the reconstruction of boosted object over a wide range of kinematic regimes. Such capability enables the efficient identification of events with multiple boosted top quarks which is a typical signature for new physics processes such as the production of the supersymmetric partner of the gluon. In order to evaluate the performance of the algorithm, the VariableR reclustered jets are compared with fixed radius reclustered jets. The flexibility of the algorithm is tested by reconstructing both boosted top quarks and boosted W bosons. The VariableR reclustering method is found to be more efficient than the fixed radius algorithm at identifying top quarks and W bosons in events with four top quarks, therefore enhancing the sensitivity for gluino searches.
3
Introduction
• Looking at multiple top quark events in order to study new physics such as super
symmetry
• In particular the top quarks decay into a W boson and bottom quark
• Gluinos can be very massive so their decay produces can be very energetic
• Key for signal discrimination is efficient reconstruction of top (or W)
multiplicity.
W+
t b
4
6
Jets
• What are jets? • Jets are a cone of hadrons which are a produced by a quark or
gluon hadronizing multiple times before it reaches the detector
6
Jets
• What are jets? • Jets are a cone of hadrons which are a produced by a quark or
gluon hadronizing multiple times before it reaches the detector
7
Incoming quark
hadronization
Jet
8 8
Jets
• What are jets? • Jets are a cone of hadrons which are a produced by a quark or
gluon decaying multiple times before it reaches the detector
8
9
Jets
• What are jets? • Jets are a cone of hadrons which are a produced by a quark or
gluon decaying multiple times before it reaches the detector
9
Jets leave energy deposits on the ATLAS calorimeters They are reclustered with 0.4 radii in the η- ϕ space
Boosted Objects
10
• When a top quark is highly boosted its decay products will be collimated
• The radius of the cone produced (jet) depends on the the PT of the top quark (R = 2M/PT)
Reclustering
Jet reclustering allows jets to be calibrated at a small radius r, and then used as inputs for reconstructing larger jets of radius R
11 http://arxiv.org/pdf/1407.2922v2.pdf
Rapidity-2 0 2
Azim
utha
l Ang
le [r
ad]
-2
0
2
=1.5 TeVZ', mt t→ = 8 TeV PYTHIA Z' s
Stable Truth Particles
R=1.0 Jetstanti-k
Rapidity-2 0 2
Azim
utha
l Ang
le [r
ad]
-2
0
2
Inside R=0.3 JetsStable Truth Particles
R=1.0, r=0.3 Jetstanti-k
Figure 1. An example event which has been clustered using the anti-kt R = 1.0 (left) and withanti-kt R = 1.0 re-clustered r = 0.3 anti-kt jets (right). The shaded regions show the jet areadetermined by clustering ghost particles. Only large radius jets with pT > 50 GeV are shown andsmall radius jets are required to have pT > 15 GeV.
Due to the increased catchment area of large radius jets over small radius jets, they aremore susceptible to contributions from pileup. Just as there are pileup correction techniquesfor large radius jets and their subjets, one can benefit from pileup corrections to the smallradius jet inputs that propagate to re-clustered jets. In particular, one can remove jetsfrom pileup interactions with techniques like JVT [19] or pileup jet identification [20] andcan correct the remaining jets with methods like the four-vector jet areas subtraction.
In the growing field of jet substructure, there are many jet observables which dependexplicitly on the jet constituents, not just the jet four-vector. These techniques are stillapplicable for re-clustered jets. Section 5 discusses two approaches to jet substructure inthe re-clustering paradigm. In a top-down approach, large radius re-clustered jets inheritthe constituents of the small radius jets clustered within. Clearly, any constituents thatmight be part of large radius jets that are not clustered within a small radius jets are notconsidered under this scheme. However, this removal of radiation also impacts trimmedlarge radius jets. More details on substructure for trimmed and re-clustered trimmed jetsis presented in Section 5.1. An alternative bottom-up approach to jet substructure is to usethe radius r jets directly as the inputs to jet substructure. The advantages and limitationsof bottom-up substructure are described in Section 5.2.
– 3 –
Energy Deposit Jets Reclustered Jets
VariableR vs Fixed
Fixed Reclustering: uses jets of r = 0.4 to reconstruct jets of R = 1.0
Variable Reclustering: uses a range from Rmin = 0.4 to Rmax = 1.5 to reconstruct jets
12
• VariableR Allows us to reconstruct jets of various kinematic regimes
• R=2M/PT define typical scale for the jets which is given by: • ρT = 2*173.5 ρW = 2* 80.4
http://arxiv.org/abs/0903.0392
13
VariableR vs Fixed
Fixed Reclustering: uses jets of r = 0.4 to reconstruct jets of R = 1.0
Variable Reclustering: uses a range from Rmin = 0.4 to Rmax = 1.5 to reconstruct jets
13
• VariableR Allows us to reconstruct jets of various kinematic regimes
• R=2M/PT define typical scale for the jets which is given by: • ρT = 2*173.5 ρW = 2* 80.4
http://arxiv.org/abs/0903.0392
Boosted Top Reconstruction
14
Eta-4 -3 -2 -1 0 1 2 3 4
Phi
0
1
2
3
4
5
6
148.967
160.003
198.717
192.775
148.967
160.003
198.717
192.775
148.967
160.003
198.717
192.775
148.967
160.003
198.717
192.775
148.967
160.003
429.638
148.967
160.003
429.638
148.967
160.003
429.638
Plot of Jets
Mass (GeV)0 50 100 150 200 250 300 350 400 450 500
# of
Eve
nts
0
200
400
600
800
1000
> 300 GeVT
); PT
=MρVR(
> 300 GeVT
R = 1.0; P
Mass (GeV)0 50 100150 200250300350400 450500
# of
Eve
nts
0
50
100
150
200
250
300
> 200 GeVT
); PT
=MρVR(
> 200 GeVT
R = 1.0; P
Mass distribution for highest PT top quark candidate Mass distribution for fourth highest PT top quark candidate
VR constituent Fixed constituent VR Fixed
Significance
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
200
400
600
800
1000
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0500
10001500200025003000350040004500
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
2
4
6
8
10
12
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
2
4
6
8
10
12
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
2
4
6
8
10
12
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
2
4
6
8
10
12
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
2
4
6
8
10
12
VariableFixed
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
2
4
6
8
10
12
VariableFixed
L= 5 fb-1
15
The VariableR algorithm is more efficient at reconstructing multiple jets and so allows us to discriminate against low top quark candidate multiplicity
Mimimum Number of Reclustered Jets
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Sign
ifica
nce
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
0.0035
0.004# of T for VR
# of T for R = 1.0
Mass distribution for highest PT top quark candidate Mass distribution for fourth highest PT top quark candidate
Boosted W Reconstruction
Mass (GeV)0 50 100150 200250300350400 450500
Num
ber o
f Rec
lust
ered
Jet
s
0
1000
2000
3000
4000
5000
6000
VariableFixed
16
Eta-4 -3 -2 -1 0 1 2 3 4
Phi
0
1
2
3
4
5
6
90.8888
70.772
52.7264
90.8888
70.772
52.7264
90.8888
70.772
52.7264
179.594
70.772
52.7264
179.594
70.772
52.7264
179.594
70.772
52.7264
Plot of Jets
VariableR can reconstruct W bosons in various regimes
VR constituent Fixed constituent VR Fixed
Mass distribution for highest PT top quark candidate
Significance Comparison
17
Mimimum Number of Reclustered Jets
-0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Sign
ifica
nce
0.0005
0.001
0.0015
0.002
0.0025
0.003
0.0035 # of T for VR# of W for VR# of T for R = 1.0# of W for R = 1.0
Conclusions and Next Steps
• I learned about reconstructing boosted top quarks and W bosons in the hadronic final state
• Studied the application of a new algorithm (VariableR reclustering) for identification of new physic processes
• This technique will be implemented in the super-symmetry search for gluino mediated stop pair production
18