hcal and jet/met
DESCRIPTION
HCAL and Jet/MET. CMS 101++ June 18, 2008 Frank Chlebana, Fermilab. Outline. Calorimeters The CMS Hadronic Calorimeters Jet Clustering Algorithms Problems with Clustering Algorithms Jet Energy Corrections Resources How You Can Get Involved The Challenge. Energy Loss in Materials. - PowerPoint PPT PresentationTRANSCRIPT
CMS 101++ June 18-20, 2008 1Frank Chlebana, Fermilab
HCAL and Jet/METHCAL and Jet/METHCAL and Jet/METHCAL and Jet/MET
CMS 101++June 18, 2008
Frank Chlebana, Fermilab
CMS 101++ June 18-20, 2008 2Frank Chlebana, Fermilab
OutlineOutlineOutlineOutline
• Calorimeters• The CMS Hadronic Calorimeters• Jet Clustering Algorithms• Problems with Clustering Algorithms• Jet Energy Corrections• Resources• How You Can Get Involved• The Challenge
CMS 101++ June 18-20, 2008 3Frank Chlebana, Fermilab
Energy Loss in MaterialsEnergy Loss in MaterialsEnergy Loss in MaterialsEnergy Loss in MaterialsElectromagnetic shower in a cloud chamberAlternating layers of “absorbers” and “active” material
Energy loss through Bremstrahlung and pair production
Pair production continues until photons’ energy is too low to produce pairs
Low energy particles dissipate energy through ionization
Hadronic showering is more complicated than EM showering Involves strong and weak interactions
Scintillator samples the number of particles in each layer → energy
CMS 101++ June 18-20, 2008 4Frank Chlebana, Fermilab
ShoweringShoweringShoweringShowering
Energy resolution for HAD calorimeters are typically worse than for EM calorimeters
Neutrinos escape undetected
Muons unlikely to Bremstrahlung and dissipate energy
Long lived particles (KS, KL, Λ) may escape before decaying or interacting
Simulation of a cosmic shower induced by a 250 GeV proton compared to a 250 GeV photon
Hadronic showers start later and have a greater lateral spread than EM showers
CMS 101++ June 18-20, 2008 5Frank Chlebana, Fermilab
Components of CMSComponents of CMSComponents of CMSComponents of CMS
HCAL ForwardCalorimeter
ECAL
CMS 101++ June 18-20, 2008 6Frank Chlebana, Fermilab
Barrel (HB) and EndBarrel (HB) and Endcap (HE) Calorimeterscap (HE) CalorimetersBarrel (HB) and EndBarrel (HB) and Endcap (HE) Calorimeterscap (HE) Calorimeters
Passive layers of brass to induce showering
About 5% of the light is captured in the fiber
Fibers sent to a Hybrid PhotoDiode (HDP) with 19 or 73 channels/device
Similar technology used for HB and HE calorimeters
Light readout via an optical fiber doped with wavelength shifter acting as light guide
Fiber is placed in a groove in the scintillator, absorbs scintillator light, re-emits it
CMS 101++ June 18-20, 2008 7Frank Chlebana, Fermilab
Barrel and Endcap CalorimetersBarrel and Endcap CalorimetersBarrel and Endcap CalorimetersBarrel and Endcap Calorimeters
Barrel (HB) |η| < 1.3EndCap (EB) 1.3 < |η| < 3.0
Calorimeters installed and being read out during global runs
Barrel (HB)
Endcap (HE)
CMS 101++ June 18-20, 2008 8Frank Chlebana, Fermilab
Forward Calorimeter (HF)Forward Calorimeter (HF)Forward Calorimeter (HF)Forward Calorimeter (HF)
HF is located about 11 m from the interaction point, covers 3 < |η| < 5
Choice of technology driven by the need to operate in a very high radiation environment
Consists of a large copper block which acts as an absorber embedded with quartz fibers parallel to the beam direction
HAD (143 cm)
EM (165 cm)
5mm
Particles incident on the front surface produce showers in the quartz/copper matrix which produce Cherenkov light
CMS 101++ June 18-20, 2008 9Frank Chlebana, Fermilab
Outer Calorimeter (HO)Outer Calorimeter (HO)Outer Calorimeter (HO)Outer Calorimeter (HO)In the central region (|η|<1.3) HB does not fully contain very energetic hadron showers
Color coding shows depth segmentation
Additional scintillator layers (HO) are located outside the solenoid (which acts as an absorber)
HO used to identify late starting showers and measure energy deposited after HB
CMS 101++ June 18-20, 2008 10Frank Chlebana, Fermilab
Calorimeter ResolutionCalorimeter ResolutionCalorimeter ResolutionCalorimeter ResolutionExperiment Material
(HAD)Resolution EM
ResolutionHAD
ZEUS Uranium – Scintillator
H1 Lead/Steel – Liquid Argon
CDF Iron – Scintillator
DØ Uranium – Liquid Argon
CMS Brass - Scintillator
Atlas Copper – Liquid Argon
E/%17 E/%35
%1/%11 E %2/%50 E
%2/%5.13 E %3/%50 E
%4/%45 E
%5.4/%100 E
%3/%50 E
%3.0/%8.2 E
E/%15
%2.0/%10 E
CMS 101++ June 18-20, 2008 11Frank Chlebana, Fermilab
CaloTowersCaloTowersCaloTowersCaloTowersMost users will access the reconstructed dataDigitized Data → Reconstructed Hits → CaloTowers
CaloTowers are made up from multiple EM Crystals (5x5) and several Hadron sections
Jet clustering uses CaloTowers
Detector/electronics relatedproblems need to be identified before “rechits” are madeMore on this later…
CMS 101++ June 18-20, 2008 12Frank Chlebana, Fermilab
Jet ClusteringJet ClusteringJet ClusteringJet Clustering
After the hard interaction, partons hadronize into stable particles
Stable particles interact with the detectors
Jet algorithms cluster energydeposits in the EM and HAD calorimeter, or more generally four-vectors
Jet algorithms should provide a good correspondence between what is observed and the hard interaction
CMS 101++ June 18-20, 2008 13Frank Chlebana, Fermilab
Clustering AlgorithmsClustering AlgorithmsClustering AlgorithmsClustering Algorithms
Cone-based clustering algorithms: Cone-based algorithms have been traditionally used at hadron colliders Interaction between theorists and experimentalists lead to the evolution of a practical Seedless Infrared Safe Cone algorithm JetClu → Midpoint → SISCone
Sequential combination:kT, Jade and Cambridge/Aachen Merges pairs of four-vectors in order of increasing relative pT
Continues until some stopping requirement is reached Does not have the problems associated with seeded cone algorithms Have been slow to execute (faster version is now available) Jet area not is well defined
Online selection: Select events using towers and clusters Limited by processing time Simplified algorithm (iterative cone) with fixed vertex
CMS 101++ June 18-20, 2008 14Frank Chlebana, Fermilab
Steps in a Clustering AlgorithmSteps in a Clustering AlgorithmSteps in a Clustering AlgorithmSteps in a Clustering Algorithm
Start from a collection of four-vectors, highest pT acts as seed
Sum all four-vectors within cone to form a proto-jet
Repeat until proto-jet is stable (axis of proto-jet aligns with the sum of four-vectors within the cone) or reach maximum number of iterations
Repeat until all towers are usedApply splitting/merging algorithm to ensure four-vectors are assigned to only one proto-jet
→ Left with list of all stable jets
CMS 101++ June 18-20, 2008 15Frank Chlebana, Fermilab
Algorithm Flow ChartsAlgorithm Flow ChartsAlgorithm Flow ChartsAlgorithm Flow Charts
Clustering Algorithm Merging/Splitting Algorithm
CMS 101++ June 18-20, 2008 16Frank Chlebana, Fermilab
Problems with Problems with SomeSome Clustering Algorithms Clustering AlgorithmsProblems with Problems with SomeSome Clustering Algorithms Clustering Algorithms
The addition of a threshold to the seed towers makes the algorithm “collinear unsafe”
Seed based cone algorithms are “infrared unsafe”
Addition of an an infitisimally soft particle can lead to new stable jet configurations → sensitive to hadronization model and order of perturbative QCD calculation
Problems arise when comparing to higher order p-QCD calculations
CMS 101++ June 18-20, 2008 17Frank Chlebana, Fermilab
Performance of Different Jet Performance of Different Jet AlgorithmsAlgorithms
Performance of Different Jet Performance of Different Jet AlgorithmsAlgorithms
Midpoint leaves unclustered towers
Dark Towers
Different jet algorithms find different jets in the same event
Midpoint
kT Clustering
JetClu
kT has jets with poorly defined boundaries
Used by CDF
CMS 101++ June 18-20, 2008 18Frank Chlebana, Fermilab
Jet Energy CorrectionsJet Energy CorrectionsJet Energy CorrectionsJet Energy Corrections
Factorized JEC:Each step uses a well defined methodology and one or two datasetsL1: pile-up and noise measured in ZB/MB events.L2: jet response vs. η relative to the barrel using dijet
balance.L3: jet response vs PT found in barrel using g/Z + jets,
top quark decays…L4: improves jet energy resolution by employing
additional parameterization of JES vs. jet EMF L5-L7: corrects for flavor variations, UE, and brings jet to
parton level.
Room to get involved in all areas of the JEC…
CMS 101++ June 18-20, 2008 19Frank Chlebana, Fermilab
Corrected Dijet MassCorrected Dijet MassCorrected Dijet MassCorrected Dijet Mass
Reconstructed dijet mass distribution for Z’ events
GenJet: clustered stable particles,does not include detector effects, includes neutral particles
“Best” that we can do
CaloJet: clustered calorimeter jets, uncorrected
Corrected CaloJet: jet correctionsapplied to the calorimeter jets
Try to improve the resolution by using information from other detectors
CMS 101++ June 18-20, 2008 20Frank Chlebana, Fermilab
Improving the ResolutionImproving the ResolutionImproving the ResolutionImproving the Resolution
Gaussian signal on top of a falling background spectrum (red)
Better resolution →improved Signal/Background
A small improvement in the resolution can lead to a largeimprovement in the sensitivity for new physics
Improving mjj resolution from 17 to 10% is like have 1.7x more data for some Higgs search channels at CDF
Run longer or work smarter…
±1σ
CMS 101++ June 18-20, 2008 21Frank Chlebana, Fermilab
Energy vs Momentum ResolutionEnergy vs Momentum ResolutionEnergy vs Momentum ResolutionEnergy vs Momentum Resolution
Can use tracking information to improve jet resolution
33.03.04
7202
BL
p
npT
T
pT
Calorimeter energy resolution becomes better as E increases while the tracker’s momentum resolution becomes worse
Jet algorithms using tracking and particle identification (Jet+Tracks and Particle Flow) are available…
For: n=100, B=1T, L=1m, σ=250μm
» 250 GeV/c we can’t reliably measure the charge of the particle at the 3σ level
CMS 101++ June 18-20, 2008 22Frank Chlebana, Fermilab
Jet Resolution ImprovementsJet Resolution ImprovementsJet Resolution ImprovementsJet Resolution ImprovementsWe do see that algorithms that make use of tracking detectors yield jets with improved resolution
Comparison of:
Calorimeter JetsParticle FlowJets + Tracks
Will be interesting to compare resolutionswhen we have real dataMore challenging environment
CMS 101++ June 18-20, 2008 23Frank Chlebana, Fermilab
Understanding METUnderstanding METUnderstanding METUnderstanding MET
DØ (V.Shary CALOR04)
Many important signatures involve Jets + Missing ET
Identify and treat:
Calorimeter noiseDead ChannelsHot ChannelsHDP dischargeHDP ion feedback
Use the global run data to look at real data…Plenty to do in the area of Data Quality Monitoring and a great place to get involved
CMS 101++ June 18-20, 2008 24Frank Chlebana, Fermilab
HCAL Readout in Global RunsHCAL Readout in Global RunsHCAL Readout in Global RunsHCAL Readout in Global Runs• Sample of 1000 events from a Global Run• Full calibrated reco chain with ICone5 Jets
• Includes DT, DTTF, HCAL, TRK, TRG subsystems
• HCAL: HB, HE+, HO• Trigger on DT at ~45 Hz
ET of CaloJets
Plots from: Daniel Miner,University of Rochester
Lots to understand…
Important and a fun place to get involved
Puts you at an advantage to get out early physics results
φ
η
CMS 101++ June 18-20, 2008 25Frank Chlebana, Fermilab
Points of FailurePoints of FailurePoints of FailurePoints of Failure
So far only considered a small part of the readout chain…
CMS 101++ June 18-20, 2008 26Frank Chlebana, Fermilab
Data Quality MonitoringData Quality MonitoringData Quality MonitoringData Quality MonitoringBefore we can use the data for physics analysis we need to ensure we are collecting good data and understand the readout and detector problems
Need to develop tools to clean up the data and treat bad channels
Understand beam conditions and developselection criteria to
Establish a solid foundation from which to build on…
CMS 101++ June 18-20, 2008 27Frank Chlebana, Fermilab
Specific Projects/TasksSpecific Projects/TasksSpecific Projects/TasksSpecific Projects/TasksSome specific studies
Help with tuning FastSimNoise, Pileup correction to the JES Optimization of split/merging Cell energy threshold optimizationJet ID and cleanupUse of tracks in Jets and PFlowType II corrections (MET)
MET significanceMET reconstruction and performanceIntegration of Jet and MET in PATCode maintainersValidation of Data Mixing ModuleImplementation of simulation uncertainty capabilityHLT trigger objects and performance
Come talk to one of the convenersJetMET, JetAlgo, MET, JEC (See JetMET org chart)
CMS 101++ June 18-20, 2008 28Frank Chlebana, Fermilab
CMS JetMET OrganizationCMS JetMET OrganizationCMS JetMET OrganizationCMS JetMET Organization
Missing ETGreg LandsbergFilip Moortgat
Jet AlgorithmsPhilipp Schieferdecker
Marek Zielinski
TasksTasks
ValidationFrank ChlebanaValidation
Frank Chlebana
Physics ToolsAttilio SantocchiaPhysics Tools
Attilio Santocchia
JetMET GroupG. Dissertori
D. Elvira
Simulation/CSA07Joanna Weng
Simulation/CSA07Joanna Weng
SoftwareRobert Harris (Jet)
Fedor Ratnikov (Jet)Bobby Scurlock (MET)
SoftwareRobert Harris (Jet)
Fedor Ratnikov (Jet)Bobby Scurlock (MET)
Documentation/User SupportFedor Ratnikov
Documentation/User SupportFedor Ratnikov
Trigger PerformanceJochen Cammin
Trigger PerformanceJochen Cammin
Online SelectionLenny Apanasevich
Online SelectionLenny Apanasevich
HCAL DPGOlga KodolovaTaylan Yetkin
HCAL DPGOlga KodolovaTaylan Yetkin
Particle Flow/Joanna Weng
Alexandre Zabi
Particle Flow/Joanna Weng
Alexandre Zabi
ECAL DPGDaniele Del ReECAL DPG
Daniele Del Re
Jet Energy ScaleRobert Harris
Ia Iashvilli
Group RepresentativesGroup Representatives
Red indicates person resident at LPC
CMS 101++ June 18-20, 2008 29Frank Chlebana, Fermilab
CMS Jet/MET Group: V.Daniel Elvira, G.Dissertori ([email protected])
https://twiki.cern.ch/twiki/bin/view/CMS/JetMET
Jet Energy Correction: R.Harris, Ia Lashvili ([email protected])
MET Group: G.Landsberg, F.Moortgat ([email protected])
Jet Algorithms: M.Zielinski, P.Schieferdecker ([email protected])
HCAL DPG: M.Velasco, V.Gavrilov ([email protected])
https://twiki.cern.ch/twiki/bin/view/CMS/HcalDPG
LPC Jet/MET Group: M.Zielinski, F.Chlebana ([email protected])
http://www.uscms.org/LPC/lpc_jetmet/lpc_jm.html
Denotes USCMS Member
Mailing Lists and ContactsMailing Lists and ContactsMailing Lists and ContactsMailing Lists and Contacts
https://hypernews.cern.ch/HyperNews/CMS/cindex
CMS 101++ June 18-20, 2008 30Frank Chlebana, Fermilab
There is a lot of local expertise at the LPCContact a convener to discuss your interest, see list from previous slide
Some documentation:CMS: Technical Design ReportDetectors: “Experimental Techniques in High Energy, Nuclear, and
Particle Physics”, T.FerbelJet Algorithms: “Run II Jet Physics hep-ex/0005012”
Examples, Code Browser, and Database Lookuphttps://twiki.cern.ch/twiki/bin/view/CMS/WorkBookJetRecohttp://cmssw.cvs.cern.ch/cgi-bin/cmssw.cgihttp://cmsdbs.cern.ch/DBS2_discovery
ResourcesResourcesResourcesResources
CMS 101++ June 18-20, 2008 31Frank Chlebana, Fermilab
The ChallengeThe ChallengeThe ChallengeThe ChallengeLater on during the workshop you have the opportunity to attend tutorials on running jobs and finding datasets
Best way to gain experience is to get your hands dirty, so….
1)Reproduce plot for the Z’ dijet mass resolution2)Add resolution for Jet+Tracks algorithm
Hints:
Look at WorkBook for some examples DBDiscovery to find datasets Code Browser useful to find examples of how to use the software Talk to people at the LPC, that is what we are here for…