14
software offline tutorial, CERN, Dec 7 th 2009 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors involved inner detector : track reconstruction, particle-id in TRT LAr EM calorimeter : cell, clusters + shower shapes for particle-id hadronic calorimeter : for leakage and isolation requirements Three reconstruction algorithms standard egamma algorithm – cluster-based for |η| <2.5 “softe” - track-based for |η|<2.5 forward electrons for |η|>2.5 – no inner detector information Several analysis algorithms dedicated for electrons and photons : e.g conversion recovery

Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

Embed Size (px)

DESCRIPTION

software offline tutorial, CERN, Dec 7 th egamma algorithms Standard reconstruction procedure seeded by clusters reconstructed in LAr EM by a sliding window (SW) algorithm NB: details of SW reconstruction have changed/improved since CSC notes cluster-track matching procedure NB: details have changed/improved since CSC notes cluster-conversion matching procedure Definition of objects cluster+track : electron cluster+ no track : non-converted photon cluster+conversion : converted photon A word on softe uses tracks as input with high preselection cuts to minimize fakes If track is same as the one used by standard algo then keep standard object essentially useful for ET

Citation preview

Page 1: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 1

Electrons and photons in ATHENAFrédéric DERUE – LPNHE ParisATLAS offline software tutorial

Detectors involvedinner detector : track reconstruction, particle-id in TRTLAr EM calorimeter : cell, clusters + shower shapes for particle-idhadronic calorimeter : for leakage and isolation requirements

Three reconstruction algorithmsstandard egamma algorithm – cluster-based for |η|<2.5

“softe” - track-based for |η|<2.5forward electrons for |η|>2.5 – no inner detector information

Several analysis algorithmsdedicated for electrons and photons : e.g conversion recovery

Page 2: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 2

Software location

Software informationall packages under Reconstruction/egammamain algorithms in Reconstruction/egamma/egammaRec

python/egammaGetter.py, src/egammaBuilder.cxxpython/softeGetter.py, src/softeBuilder.cxxpython/forwardGetter.py, src/forwardBuilder.cxx

main analysis algorithms in Reconstruction/egamma/egammaAnalysis

PhotonAnalysisAlgsElectronAnalysisAlgs

Page 3: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 3

egamma algorithmsStandard reconstruction procedure

seeded by clusters reconstructed in LAr EM by a sliding window (SW) algorithm

NB: details of SW reconstruction have changed/improved since CSC notes

cluster-track matching procedureNB: details have changed/improved since CSC notes

cluster-conversion matching procedureDefinition of objects

cluster+track : electron cluster+ no track : non-converted photoncluster+conversion : converted photon

A word on softeuses tracks as input with high preselection cuts to minimize fakesIf track is same as the one used by standard algo then keep standard objectessentially useful for ET<5 GeVapart “this”, objects or identification is the same as for standard a

A word on forward electronsuses topo clusters, no inner detectordedicated identification

A certain overlap (few %) exist between electrons and converted photons

Page 4: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 4

cluster-size and calibrationCluster size

electrons : 3*7 in barrel, 5*5 in end-capsnon-converted photons : 3*5 in barrel, 5*5 in end-capsconverted photons : 3*7 in barrel, 5*5 in end-caps

Calibrationdifferent corrections are applied for electrons/photonsfor position and energy measurements

There are different calorimeter calibrations, cluster sizes, and identification variables and cuts used for electrons and photons: if the user reassigns the type of the object in analysis, the user needs to properly correct for this. Tools are available to do this.

Page 5: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 5

four-momentumI4Momentum

base class to keep 4-momentum doxygen documentation http://tinyurl.com/6h2o23

Combinationsince 15.3.0 the energy of the four-momentum is a combination of the calorimeter and tracker measurementsince rel 15.5 for photons energy is only from calo

Page 6: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 6

egamma object

CaloClustercalorimeter informationavailable for all objects

Link to a TrackParticletracking informationuseful only for electrons

Link to a VxCandidatevertex informationavailable for converted electrons – in the PhotonContainer

egDetailadditional reconstructed informationavailable for all objects

PIDparticle identificationavailable for all objects

Page 7: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 7

What's on AODThree containers available in StoreGate

ElectronAODContainerPhotonAODContaineregDetailsAODContainer

Links to other containersTrackParticle, CaloCluster, VxCandidateinformation is sufficient for the user to be able to redo some basic PID

Documentationhttps://twiki.cern.ch/twiki/bin/view/AtlasProtected/EgammaAOD

7

Page 8: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 8

Cluster informationCaloCluster object

reconstructed E, px, py, pzposition eta, phi

Recalibration of clustershttps://twiki.cern.ch/twiki/bin/view/AtlasProtected/EgammaAODRenderegDetails

object called EMShower contains most of the detailed calorimeter informationdescribe shower shapes define detailed cuts for identification purpose

Page 9: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 9

Tracking informationPointer to the Trk::Track with which the particle was createdTracks associated to electron can be refitted at AOD levelAlso contains

chargeTrk::FitQualityDefiningParameters()

For electrons this is the Trk::Perigee – useful for vertex fitting

Page 10: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 10

VxCandidateUsed to create conversionsSingle track and double track conversions possibleContains

position of the vertexVxTrackAtVertex(s) that make the vertex

from which you can obtain the corresponding TrackParticle

Page 11: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 11

egDetailEMShower

calorimeter informationisolation, leakage

EMTrackMatchtrack to calorimeter matching informatione.g , at each of the sampling layers

EMConvertconversion matchin variableable to obtain a full Trk::VxCandidate

EMBremFitresults from CaloBremFite.g brem radius, improved pT

EMTrackFitable to obtain full Trk::Perigee parameters

EMFourMomErrors

Page 12: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 12

electron pid - IsEM

Multivariate techniques are available but not discussed here (eg. Log Likelihood, Boosted Decision Trees, H-matrix, Neural Nets, …).These techniques will be commissioned after the cut-based electron id has been validated with early data. Only certified variables will be used.

Default method used in electron identificationA basic series of cuts based on:

shower shape properties in different compartments of the calorimeter combination of ID and Calo

Loose - Calorimeter only information hadronic Leakage, Shower Shape (middle layers only of EM cal)excellent ID efficiency, low BG rejection

Medium - Adds tracking information basic Track quality cutsmore detailed Calorimeter Cutsincreased jet rejection efficiency by a factor 3-4 while reducing ID efficiency by 10%

Tight - Using all information available b-layer hit requirement,TRT High Threshold hit informationstrongest background rejection but lowest ID efficiency ISO - For isolated electrons provides tighter calorimeter cutsTRT - For non-isolated electrons (from heavy quark decay) tighter TRT PID

Page 13: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 13

photon pid - IsEM

Default method used in photon identificationA basic series of cuts based on:

shower shape properties in different compartments of the calorimeter different cuts for converted and non converted photons

Loose - Calorimeter only information hadronic Leakage, Shower Shape (middle layers only of EM cal)excellent ID efficiency, low BG rejection

Tight - Using all information available use information from the strips in particularstrongest background rejection but lowest ID efficiency ISO - For isolated photons provides tighter calorimeter cuts

Multivariate techniques are available but not discussed here (eg. Log Likelihood, Boosted Decision Trees, H-matrix, Neural Nets, …).These techniques will be commissioned after the cut-based photon id has been validated with early data. Only certified variables will be used.

Page 14: Software offline tutorial, CERN, Dec 7 th 2009 1 Electrons and photons in ATHENA Frédéric DERUE – LPNHE Paris ATLAS offline software tutorial Detectors

software offline tutorial, CERN, Dec 7 th 2009 14

More information

• egamma twiki:

• https://twiki.cern.ch/twiki/bin/view/AtlasProtected/ElectronGamma

• egamma CSC notes:

• http://cdsweb.cern.ch/record/1119939

• egamma hypernews

• https://hypernews.cern.ch/HyperNews/Atlas/get/EGammaWG.html