Jet reconstruction: where we are now

1

• Jet reconstruction where we are now

2
How we reconstruct jets, development steps so
far
1.) no Calibration , EcalPlusHcalTower jet
input Peak at h 0 There is a cut on ET, but
not on E in tower input ? noise survives ET cut
at h 0
2.) no Calibration , EcalPlusHcalTower jet
input cut on tower E, E gt1 ? get rid of towers
seeded by noise
3
3. Take EcalPlusHcalUETower corrects peak
at IhI 0 as well
4. GammaJet Calibration with
EcalPlusHcalUETower input ? cannot be applied on
low ET ? better use ET raw jet for the veto ET
raw 20 GeV ETCal 30 GeV
ET jet
30 ET jet raw
4
5. no Calibration , EcalPlusHcalUETower jet
input, Ih jet I lt 2.5 To get rid of fake jets
at high h take only jets with IhI lt 2.5. compare
with generated jets (DR matching 0.3), matching
ok above ET 20 GeVBut there are
still a lot of fake jets (not matched) at jet ET
raw lt 20 GeV? have to set ET cut at 20 GeV? try
to optimize
5
Requirements for matchingVertex
reconstructed vertex closest to z position of
the leptons Jet Isolation jets have to be
more than DR 0.5 away from one of the 2
leptonsTracks have to come from event vertex
(Dz lt0.4 and Dd lt0.1)number of hits gt 5pT gt
0.9GeV if DR between jet and tracks is lt0.5
alpha SPT (Tracks) /ET (jet)
Discriminate fake jets from real jets matching
jets with tracks
6
ET jet lt 20 GeV
ET jet gt 20 GeV
? above ET of 20 GeV, most jets matched to
generated jets ? real jets. below 20 GeV a lot
of fake jets (not matched to generated jets) from
UE and multiple interaction Additional
requirements for jets between 15 and 20 GeV
alphagt0.2
7
Apply this cut Very useful to get rid of fake
jets (10 better matching)?can set ET jet raw
cut to 15 GeV
8
Results including alpha cut

• Jet reconstruction used
• Jets from DSTs, ORCA873
• No Calibration
• Iterative Cone jet algorithm, Cone size 0.5
• ET jet min 10 GeV
• ET jet particle min 0.5 GeV
• h jet lt 2.5
• Use 2 different Tower Inputs to see which one is
  better
• a.) NOUE Tower Input EcalPlusHcalTower, ETgt0.5,
  Egt1 GeV
• b.) UE Tower Input EcalPlusHcalUETower

9

• Samples used
• Signal
• NOUE JETS 18885 H?WW? 2l (le m t) events,
  mH165GeV
• simulated with Pythia, reconstructed with
  ORCA873
• UE JETS 19876 H ? WW ? 2l (le m t) events,
  mH165GeV
• simulated with Pythia, reconstructed with
  ORCA873
• Background
• NOUE JETS 128000 tt_2l (le m t) events
• simulated with Pythia, reconstructed with
  ORCA873
• UE JETS 88000 tt_2l (le m t) events
• simulated with Pythia, reconstructed with
  ORCA873
• NOUE JETS 34000 ww_2l (l e m t) events
• simulated with Pythia, reconstructed with
  ORCA873

10
jet veto efficiency for ttbar
jet veto efficiency for a 165GeV Higgs
11
cross section after cuts for ttbar
(all cross section LO)
cross section after cuts for a 165GeV higgs
12
Effect of different pt track cuts, if we apply
only jet veto or all selection pt tracks gt 0.9
(default)the efficiency of the jet veto for
the hww signal is 53, while for ttbar its
2.6the efficiency of  all cuts is 0.088 for the
signal and 0.00032 for the ttbar background
(well, only 8 events are left here)pt tracks gt
2 GeVthe efficiency of the jet veto for the
hww signal is 54, while for ttbar its 2.8the
efficiency of  all cuts is 0.098 for the signal
and 0.00032 for the ttbar backgroundpt tracks
gt 5 GeVthe efficiency of the jet veto for the
hww signal is 56, while for ttbar its 3.5the
efficiency of  all cuts is 0.102 for the signal
and 0.00055 for the ttbar backgroundTherefore
we would suggest to use the cut on 2 GeV, but not
on 5 GeV, as too much background will be left.
13
some plots to different cuts on the pt of the
tracks (default 0.9, 2 and 5 GeV).
14
(No Transcript)
15
to the results First, the matching efficiency
rises about 2 when going to a pt cut of 2 GeV
for et raw around 15 GeV (a lot below 15 GeV),
and 3-4 for a pt cut of 5 GeV. This is done
with a alpha cut of 0.2 . To see if this alpha
cut should be varied too we can look at the alpha
distribution for the different pt track cuts.
What we can see is, that for a cut of 5 GeV, we
loose a lot of matched jets, when applying a cut
on alpha (many matched jets have an alpha of 0).
So it looks like a value of 2 GeV would be
reasonable, but not 5 GeV.
16
How we reconstruct jets

• No Calibration
• Iterative Cone jet algorithm, Cone size 0.5
• ET jet min 10 GeV
• ET jet particle min 0.5 GeV
• h jet lt 2.5
• EcalPlusHcalTowerInput (not EcalPlusHcalUETowerInp
  ut)
• Energy gt 0.8 (to be consistent with JetMET and
  other groups), ET gt 0.5
• Jet veto
• Case 1 no jet with ETgt20GeV, hlt2.5
• Case 2 15ltET jetlt20 GeV and alphagt0.2
• with alpha SPT (Tracks) /ET (jet) , where PT
  (Tracks) gt 2 GeV
• To test cut on seed energy useful?