Jets and Missing Et - home.fnal.govhome.fnal.gov/~bhatti/doe/doe06/doe06_cms.pdf · Jets and...
Transcript of Jets and Missing Et - home.fnal.govhome.fnal.gov/~bhatti/doe/doe06/doe06_cms.pdf · Jets and...
Jets and Missing Et
Anwar A BhattiNovember 28, 2006
DOE Site Visit
November 28, 2006 Anwar Bhatti DOE Review 2
Jets at LHC/CMS
Good understanding of jets/Missing Et are needed for
a) SM confirmation (i.e. detector is working)b) Higgs Mass reconstruction (MSSM )c) SUSYd) Higgs search (H WW lνlν, H ZZ llνν) e) High mass Higgs boson, new particles, Quark Compositeness …f) Reject Standard Model BackgroundGood understanding of hadron calorimeter needed for lepton identification also.
November 28, 2006 Anwar Bhatti DOE Review 3
The CMS Calorimeter
EM calorimeter |η| < 3 :PbW04 crystals1 longitudinal section/preshower 1.1 λ∆η×∆ϕ = 0.0174 × 0.0174Central Hadronic |η| < 1.7 :Brass/scintillator +WLS2 + 1 Hadronic Outer – long. sections 5.9 + 3.9 λ (|η| =0) ∆η×∆ϕ = 0.087 × 0.087Endcap Hadronic 1.3< |η| < 3 :Brass/scintillator +WLS2 or 3 longitudinal sections 10λ∆η×∆ϕ = ~0.15 × 0.17Forward 2.9 < η < 5:Fe/quartz fibers ∆η×∆ϕ = ~0.175× 0.17
Hcal barrel and EndCap
EM barrel and EndCap
Very Forward Calorimeter
Hadronic Outer
Towers of dimension ∆η×∆ϕ = 0.087 ×0.087 gradually increasing in endcap and forward regions are formed for a total of 4176 tower.
November 28, 2006 Anwar Bhatti DOE Review 4
Jet Reconstruction and Validation
Jet clustering algorithms Calorimeter noise and its effect on jet clustering.Un-clustered energy in the Midpoint Clustering Algorithm(Husseyin Topakli, Cukurova U., Adana, Turkey).
Jet software validationSoftware to validate every release of software with Robert Harris (Fermilab), Fedor Rantikov (Maryland).Many checks for initial performance of jet software.
November 28, 2006 Anwar Bhatti DOE Review 5
Noise in Central Calorimeter in cone of R=0.5
0 10 20 30 40 500
50
100
150
200
250
300
350
400
450
nConeTower00 nConeTower00Entries 5400Mean 27.16RMS 4.584Underflow 0Overflow 0Integral 5400
nConeTower00
0 10 20 30 40 500
200
400
600
800
1000
nConeTower01 nConeTower01Entries 5400Mean 4.48RMS 1.979Underflow 0Overflow 0Integral 5400
nConeTower01
0 10 20 30 40 500
200
400
600
800
1000
1200
1400
1600
1800
nConeTower02 nConeTower02Entries 5400Mean 2.146RMS 1.288Underflow 0Overflow 0Integral 5400
nConeTower02
0 1 2 3 4 50
100
200
300
400
500
emConeTower00 emConeTower00Entries 5400Mean 1.928RMS 0.4506Underflow 0Overflow 0Integral 5400
emConeTower00
0 1 2 3 4 50
200
400
600
800
1000
emConeTower01 emConeTower01Entries 5400Mean 0.3397RMS 0.253Underflow 0Overflow 0Integral 5400
emConeTower01
0 1 2 3 4 50
200
400
600
800
1000
1200
emConeTower02 emConeTower02Entries 5400Mean 0.3046RMS 0.2413Underflow 0Overflow 0Integral 5400
emConeTower02
0 2 4 6 8 10 12 140
20
40
60
80
100
120
140
160
180
200
220
hdConeTower00 hdConeTower00Entries 5400Mean 6.52RMS 1.947Underflow 0Overflow 0Integral 5400
hdConeTower00
0 2 4 6 8 10 12 140
100
200
300
400
500
600
700
800
900
hdConeTower01 hdConeTower01Entries 5400Mean 2.151RMS 1.343Underflow 0Overflow 0Integral 5400
hdConeTower01
0 2 4 6 8 10 12 140
500
1000
1500
2000
2500
3000
3500
hdConeTower02 hdConeTower02Entries 5400Mean 0.3803RMS 0.6508Underflow 0Overflow 0Integral 5400
hdConeTower02
No thesholds<n> = 27<EB> = 1.9 GeV<HB> = 6.5 GeV
EB> 0.15 GeVHB> 0.6 GeV<n> = 4.5<EB>= 0.3 GeV<HB>= 2.2 GeV
EB> 0.18 GeVHB> 0.8 GeV<n> = 2.1<EB>= 0.3 GeV<HB>=0.4 GeV
November 28, 2006 Anwar Bhatti DOE Review 6
Noise Suppression
Cell Threshold
Scheme HB GeV HO GeV HE GeV Noise in Cone GeV η= 0
Jet Energy Loss GeV η = 0
A 0.7 0.85 0.9 1.4 -
B 0.9 1.1 1.8 0.3 1.0
C 1.2 1.3 1.8 0.2 1.9
A or B schemes are used for physics studies
HB Hadron BarrelHO Hadron Outer HE Hadron Endcap
JetMET group recommended these thresholds in Physics Technical Design Report Vol. I for all physics analysis.
November 28, 2006 Anwar Bhatti DOE Review 7
Fast Calorimeter Simulation (GFlash)
Current Geant4 (G4) calorimeter simulation may not reproduce the test beam data. Implement GFlash which is easily tunable and is quite fast.
GFlash uses parameterized showers. The particles produced in the pp collision are propagated by G4 to the first interaction in the calorimeter. After first interaction, control is passed to GFlash which deposits energy according to tuned parameters.Successfully used by H1 and CDF experiments.
0 5 10 15 20 250
500
1000
1500
2000
2500
3000
Longitudinal Profile Barrel_LongitudinalEntries 26Mean x 12.5Mean y 555.4RMS x 7.5RMS y 557.5
Reference
New
PV=1
Longitudinal Profile
•Software for electromagnetic particles written by Joanna Weng.
•Working with Soon Jun (Carnegie Mellon), Romulus Godang (U. Mississippi)to implement for haronically interacting particles and over all tuning.
Comparison of full G4 and GFlashLongitudinal Profile 30 GeV Photons
Depth (X0)
November 28, 2006 Anwar Bhatti DOE Review 8
Simulation based on GFlash
Implementation of GFlash in CMSSW.Does GFlash parameterization work for high energy particles? How good GFlash flavor dependence of response is? Compare with G4.Is geometry in CMSSW correct? specially boundaries?Is material in CMSSW correct? How to measure/tune with real data?How to simulate boundaries in GFlash? Do we need special treatment?Measure in G4, parameterize and use Post-GFlash.
How to treat HadronOuter in GFlash? Forward Hadron?How to treat muons in GFlash? do we use G4 for muons?Measure single particle response in G4, transverse profile, E/pTune GFlash to G4 (transverse profile, E/p), build tools.Test Beam analysis, transverse shape, E/pTriggers for single track in collision data, HLT algorithmsSingle track data analysisTuning of GFlash/G4
Anwar A Bhatti LHC Physics Center Fermilab April 19, 2006
November 28, 2006 Anwar Bhatti DOE Review 9
Calorimeter Response
Beam momentum [GeV/c]1 10 210
Rec
o /
Bea
m m
om
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
- π+ π
e- p pK-K+
2006 Test Beam• Ecal+Hcal test beam in Sept 2006.
• First time full Ecal module tested along with Hcal.
• Good particle identification at low momenta. (problem in 04 Test Beam)
• Hcal analysis mostly done at LPC (daily meeting at 4 pm)
• RU working on measuring transverse profile of hadronicshowers.
• Evaluation of new photo-detectors for HardonOuter calorimeter (muon signal, dynamic range, dead time) Very clean test beam data
November 28, 2006 Anwar Bhatti DOE Review 10
LPC Jet Energy Scale and Resolution Group
Informal group, meets every week on Tuesdays Students and post docs from various universityRutgers University (Lalith, Dmitry)University of Illinois at Chicago (Lenny, Agata)University of Rochester (Gennady, Jonathan, Jochen)University of Cukurova, Turkey (Selda, Pelin, Mehmet, Huseyin)
November 28, 2006 Anwar Bhatti DOE Review 11
Jet Reconstruction and Jet Corrections
• A calorimeter/particle jet is defined by an algorithm.
• Jet kinematics and corrections depend on the reconstruction algorithm and parameters.
Rec
onst
ruct
ion
Calorimeter jets ↓
Particle jets↓
Parent Parton
November 28, 2006 Anwar Bhatti DOE Review 12
Calorimeter Jet to Particle Jet to Parent Parton
Multi-step approach to isolate various detector and physics issues.φ uniformity (material effects?)η uniformity (parton radiation, transverse shower size)Absolute scale for calorimeter response
Simulation based correctionsPhoton-jet balancing (determined directly from data)
Parent parton energy (correct for any radiation outside the clustering cone)Validation of jet energy scale using W JJ in top quark events
November 28, 2006 Anwar Bhatti DOE Review 1313
Distance of particles from Jet axis(η, φ space) |ηjet|<1.0
Multiplicity
<N>=13.4
<N> 18.8 <N>=26.6
<N>=33.4 <N>=38.9 <N>=44.3
<N>=49.4<N>=60.44
PtJet 10-30 GeV 30-50 GeV 80-120
170-230 300-380 470-600
800-1000 3000-3500 GeV
Pelin Kurt
November 28, 2006 Anwar Bhatti DOE Review 14
Jet Energy Fraction
Photons
Muon/Neutrinos
Neutron/Klong
Charged hadrons
November 28, 2006 Anwar Bhatti DOE Review 15
Flavor Composition of a Jet
PtJet 30-50 GeV
Particle ID <N> RMS <PtFraction> RMS
Photon 8.5 3.9 24.8% 15.9%
Neutron/KL 1.3 0.8 7.9% 11.1%
Muon/ν 0.5 0.1 0.3% 0.3%
Charged Hadrons
10.0 3.4 67.9% 17.4%
Large fluctuation between Em <=> Hadronic components combined with calorimeter shower fluctuations makes determintion of jet energy scale difficult.
November 28, 2006 Anwar Bhatti DOE Review 16
Jet Response in CSA06 Data Sets
Tune the calorimeter simulation to test beam measurements,
Generate jets using (Pythia) fragmentation model,
Compared calorimeter jet with Particle jet Determine jet corrections
Mehmet Vergili
Many physics data sets were produced as a part of Computing, Software and Analysis (CSA) exercise in Sept-Nov 2006.
November 28, 2006 Anwar Bhatti DOE Review 17
Calorimeter-Jet to Particle-Jet Corrections
Initial corrections will be based on simulation.
-4 -2 0 2 40.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
ResponseVsEtaMean10ResponseVsEtaMean10
η of Jet
PtCalorJet/PtParticleJet
Software to determine the jet response by Rockefeller U.Fitting/parameterization by Moscow State University. Maintenance by Selda/Duong Nguyen Brown University.
November 28, 2006 Anwar Bhatti DOE Review 18
Absolute Scale using Photon-Jet balancing
Decouple the jet energy scale between data and MC events.(A short cut: Simulation need not reproduce the data exactly.)Issues
Determine particle jet or parent parton energy. ISR and non-linearity in the calorimeter.Difference in gluon/quark jets.Dijet background (fake photons).
Photon-jet balancing Techniques:Missing Et Projection Method: Use whole event, determine particle-level jet correction, less sensitive to radiation in the event (used by D0). Pt-balance Method: Parent parton energy, less sensitive to low energy calorimeter response. Used by CDF as a cross check. At particle level balance is not 0.
November 28, 2006 Anwar Bhatti DOE Review 19
Absolute Energy Scale from the Data
Photon-Jet Balancing
November 28, 2006 Anwar Bhatti DOE Review 20
Jochen Cammin
U. Of RochesterPhoton-Jet Balancing
Photon |η| ≤1.0
Pt Jet > 15 GeV
∆φ(γ,jet) > 2.8
Need to study:
Trigger
Isolation criteria
Radiation effects
November 28, 2006 Anwar Bhatti DOE Review 21
DiJet Balancing (Rockefeller U., Fermilab responsibilty)
Purpose:Make the detector jet response uniform vs pseudo-rapidity.Advantage:
Jet absolute calibration will need to be done only in a limited region in η.Cross check of the simulated data.
Issues:Only true in 2 2 scattering, Pt1=Pt2 at parton level.Final/initial state radiation spoil the exact balance but the effect should average out.
Reduce ISR/FSR effects: ∆φ> 172 deg, PtJet 3< 20May not be true at particle-level (η dependence in OOC radiation?)Pile-up and underlying event (Low Pt, probably the Pt distribution isuniform in η but calorimeter cares about energy, different point on E/p curve.)Noise contribution (Different in different regions of calorimeter)Different resolution/smearing in different regions of calorimeter.
Needs a detailed study but these effects ~2-3%.
November 28, 2006 Anwar Bhatti DOE Review 22
DiJet Response Mehmet Vergili
Particle Jets
Calorimeter Jets
Eta of Probe Jet
November 28, 2006 Anwar Bhatti DOE Review 23
Dijet Balance Based Jet Correction
><+><−
MPFMPF
22
><−><+
BB
22
November 28, 2006 Anwar Bhatti DOE Review 24
PhotonJet Response
Absolute scale compared to EM calorimeter scale
CaloJet/ParticleJet
November 28, 2006 Anwar Bhatti DOE Review 25
Jet Energy Resolution using Local Clusters
Identify sub-clusters in a jet and correct each cluster based in energy deposition in Em Calorimeter and Had Calorimeter.
Cluster em cells within R=0.03, (3x3 cells)Cluster had cells within R=0.15, (3x3 cells)Match em clusters with had clustersClassify clusters based on had and em energies and transverse shape.Correct hadronic clusters (particle) for difference in ECAL response to hadron and electrons. ECAL is calibrated for electrons.We worked with Lalith in old software framework before he left to work on Forward Pixals. Now working with Turkish students.Add the tracking information later.
November 28, 2006 Anwar Bhatti DOE Review 26
e/h corrections to single pions
( / ) ( / ) /[1 ( / ) 1) ]( / ) ( / ) /[1 ( / ) 1) ]
( / ) ( / )
E E E o
H H H o
E E H H
e e h e h fe e h e h f
E e e
ππ
π ε π ε
= + −= + −
= +
f0 =a3log(e_ecal+e_hcal)^a4
is average neutral fraction of shower.
Three param describe e/h of two sections
15 GeV π<E>=15.41 GeV
Five parameters describe the response obtained from to 10, 30, 100 and 300 GeV test beam data.
15 GeV pions
After correctionsBefore corrections
Dan Green 2004 Test Beam
Energy deposition by early and late showers is different. One can improve resolution if energies in Em and Had calorimeters are corrected separately.
November 28, 2006 Anwar Bhatti DOE Review 27
Towers in a QCD events Husyein Topakli
Towers 358
Em/Etot
Et of towers
QCD jet events 80-120 GeV
57 towers inside jets with Pt>10 GeV.
Use CaloTowers, order in Et and
Cluster towers in R=0.15 (3x3)
Working on clustering from RecHits
November 28, 2006 Anwar Bhatti DOE Review 28
Local Clusters from Towers
Number of Clusters
Towers in Clusters
Et of Clusters
Em/Etot<Towers/Cluster>=1.14
<Cluster>=313.6
November 28, 2006 Anwar Bhatti DOE Review 29
SumEt/Met from Clusters Before and After Corrections
From Clusters After Corrections
Before
After
Need to implement cluster corrections derived from the simulation.
Missing EtSumEt MissingEtSignificance
November 28, 2006 Anwar Bhatti DOE Review 30
CDF4CMS
Missing Et Startup workshop Sept 22, 2006CDF experience in monitoring the calorimeter stability (Anwar)
Jet Workshop (Dec 14-15, 2006)CDF Experience with jets reconstruction, jet energy scale and jet resolution (Kenichi Hatakeyama).
FP420/TotemBackground studies, alignment issues, exclusive jet analysis.
November 28, 2006 Anwar Bhatti DOE Review 31
Missing Et in Inclusive SUSY Events at CMS
Inclusive SUSY Production in mSUGRASUSY Low Mass Point 1 (consistent with WMAP data)Dominated by gluino and squark production σ(LO)=49 pbMgluino= 611 GeV and msquark ~ 520, tanβ =10Cascade decays lead to final state containing
two stable neutralinos (=LSP) → Missing Et.
Koji Terashi
(GeV/c)T
Missing p0 100 200 300 400 500 600 700 800 900 1000
Eve
nts
1
10
210
Correct Generator METCalorimeter RawCalorimeter Corrected (Type 1)
Missing Transverse Momentum
(GeV)TJet1 E0 100 200 300 400 500 600 700 800 900 1000
dN
/ N
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04Generator
0.1)≤R ∆Calorimeter Raw (
0.1)≤R ∆Calorimeter Corrected (
9860 GenJet4329 matched CaloJets
Jet1 Transverse Energy (MidPoint R=0.5)
(GeV/c)T
2-LSP p0 100 200 300 400 500 600 700 800 900 1000C
orr
ect
Gen
erat
or
ME
T (
GeV
/c)
0
100
200
300
400
500
600
700
800
900
1000
0
10
20
30
40
50
60
70
Missing Transverse Momentum vs 2-LSP Transverse Momentum
November 28, 2006 Anwar Bhatti DOE Review 32
Conclusions
The work on jet energy scale determination is progressing.We are working on test beam data to understand the calorimeter response and shower shapes.Some people have joined the jet energy scale and resolution effort at CMS but we need many more people to complete the job.The work on improving jet energy resolution using local clusters re-started in new software framework. We are working to determine corrections from simulated data instead of test beam data.It will greatly help to have an additional Rockefeller post doc towork on these projects, specially to improve the jet energy resolution. We have started working on SUSY search in Jets+MET channel and missing Et correction issues.We have a nice Rockefeller office at LPC (11th floor Wilson Hall) now.
Looking forward to LHC data to discover physics beyond the Standard Model