P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 Introduction To Jet & Missing Transverse Energy Reconstruction in ATLAS Peter Loch.

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 Introduction To Jet & Missing Transverse Energy Reconstruction in ATLAS Peter Loch University of Arizona Tucson, Arizona USA

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 22PreliminariesPreliminaries Scope Scope Introduction focuses on explanation of algorithms and methods Introduction focuses on explanation of algorithms and methods Including some pointers to underlying principles, motivations, and expectations Including some pointers to underlying principles, motivations, and expectations Support discussion sessions by presenting common issues only once Support discussion sessions by presenting common issues only once Plots shown can be obsolete Plots shown can be obsolete Plots are used to deliver impressions only – no or few hard facts! Plots are used to deliver impressions only – no or few hard facts! Latest performance evaluation and comparisons will be discussed in depth in the course of the next few days Latest performance evaluation and comparisons will be discussed in depth in the course of the next few days Audience Audience Clearly introductory character aims at prospective new contributors Clearly introductory character aims at prospective new contributors Lecture style but please – ask questions! Lecture style but please – ask questions! Hopefully useful even for people already working on the topics Hopefully useful even for people already working on the topics If you are bored – you know where the beach is… If you are bored – you know where the beach is… … but then don’t waste time addressing basic issues during the discussions! … but then don’t waste time addressing basic issues during the discussions! Historical context Historical context Follow-up of presentations given at Milan & Tucson Follow-up of presentations given at Milan & Tucson Updates and extended missing Et part Updates and extended missing Et part Thanks to Michel Lefebvre and Peter Krieger for their help with the first two presentations! Thanks to Michel Lefebvre and Peter Krieger for their help with the first two presentations!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 33RoadmapRoadmap Hadronic shower model simulation in Geant4 Hadronic shower model simulation in Geant4 John Apostolakis John Apostolakis Energy reconstruction in the ATLAS calorimeters Energy reconstruction in the ATLAS calorimeters Manuella Vincter Manuella Vincter Jets & missing transverse energy reconstruction and calibration Jets & missing transverse energy reconstruction and calibration This talk This talk Introduction to modern jet algorithms Introduction to modern jet algorithms Paolo Francavilla Paolo Francavilla Summary of the recent ATLAS jet algorithm studies Summary of the recent ATLAS jet algorithm studies Sebastian Eckweiler Sebastian Eckweiler Jet software from a user’s point of view experience Jet software from a user’s point of view experience Kerstin Perez Kerstin Perez

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 44 Overview (1) Part I: Introduction to jets at LHC Part I: Introduction to jets at LHC Physics with jets at LHC Physics with jets at LHC Some examples for final states involving jets Some examples for final states involving jets Physics collision environment Physics collision environment E.g., pile-up E.g., pile-up Jet algorithms – rules and guidelines Jet algorithms – rules and guidelines Physics requirements to meaningful jets Physics requirements to meaningful jets Experimental requirements and limitations Experimental requirements and limitations Part II: Detector jet reconstruction and calibration Part II: Detector jet reconstruction and calibration Calorimeter signal reconstruction for hadrons Calorimeter signal reconstruction for hadrons Signals from hadronic showers Signals from hadronic showers Calorimeter towers & clusters Calorimeter towers & clusters Hadronic calorimeter energy scale Hadronic calorimeter energy scale Jet reconstruction and calibration Jet reconstruction and calibration Task overview Task overview Calorimeter input signals Calorimeter input signals Cell weighting Cell weighting Jet energy scale corrections & characteristics Jet energy scale corrections & characteristics Experimental data and simulations Experimental data and simulations Track jets Track jets Software aspects of jet reconstruction and calibration Software aspects of jet reconstruction and calibration

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 55 Overview (2) Part III: Missing E t Reconstruction Part III: Missing E t Reconstruction Contributions to missing Et Contributions to missing Et Software aspects Software aspects

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 66 Physics with Jets at LHC

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 77 Where Do Jets Come From At LHC? inclusive jet cross-section Fragmentation of gluons and (light) quarks in QCD scattering Fragmentation of gluons and (light) quarks in QCD scattering Most often observed interaction at LHC Most often observed interaction at LHC Decay of heavy Standard Model (SM) particles Decay of heavy Standard Model (SM) particles Prominent example: Prominent example: Associated with particle production in Vector Boson Fusion (VBF) Associated with particle production in Vector Boson Fusion (VBF) E.g., Higgs E.g., Higgs Decay of Beyond Standard Model (BSM) particles Decay of Beyond Standard Model (BSM) particles E.g., SUSY E.g., SUSY

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 88 Where Do Jets Come From At LHC? top mass reconstruction Fragmentation of gluons and (light) quarks in QCD scattering Fragmentation of gluons and (light) quarks in QCD scattering Most often observed interaction at LHC Most often observed interaction at LHC Decay of heavy Standard Model (SM) particles Decay of heavy Standard Model (SM) particles Prominent example: Prominent example: Associated with particle production in Vector Boson Fusion (VBF) Associated with particle production in Vector Boson Fusion (VBF) E.g., Higgs E.g., Higgs Decay of Beyond Standard Model (BSM) particles Decay of Beyond Standard Model (BSM) particles E.g., SUSY E.g., SUSY

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 99 Where Do Jets Come From At LHC? Fragmentation of gluons and (light) quarks in QCD scattering Fragmentation of gluons and (light) quarks in QCD scattering Most often observed interaction at LHC Most often observed interaction at LHC Decay of heavy Standard Model (SM) particles Decay of heavy Standard Model (SM) particles Prominent example: Prominent example: Associated with particle production in Vector Boson Fusion (VBF) Associated with particle production in Vector Boson Fusion (VBF) E.g., Higgs E.g., Higgs Decay of Beyond Standard Model (BSM) particles Decay of Beyond Standard Model (BSM) particles E.g., SUSY E.g., SUSY

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1010 Where Do Jets Come From At LHC? Fragmentation of gluons and (light) quarks in QCD scattering Fragmentation of gluons and (light) quarks in QCD scattering Most often observed interaction at LHC Most often observed interaction at LHC Decay of heavy Standard Model (SM) particles Decay of heavy Standard Model (SM) particles Prominent example: Prominent example: Associated with particle production in Vector Boson Fusion (VBF) Associated with particle production in Vector Boson Fusion (VBF) E.g., Higgs E.g., Higgs Decay of Beyond Standard Model (BSM) particles Decay of Beyond Standard Model (BSM) particles E.g., SUSY E.g., SUSY electrons or muons jets missing transverse energy

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1111 LHC Environment: Underlying Event Collisions of other partons in the protons generating the signal interaction Collisions of other partons in the protons generating the signal interaction Unavoidable in hadron collisions Unavoidable in hadron collisions No real first principle calculations No real first principle calculations Low pT (non-pertubative) QCD Low pT (non-pertubative) QCD Some correlation with hard scattering Some correlation with hard scattering Typically tuned from data in physics generators Typically tuned from data in physics generators Carefully measured at Tevatron Carefully measured at Tevatron Phase space factor applied to LHC tune in absence of data Phase space factor applied to LHC tune in absence of data One of the first things to be measured at LHC One of the first things to be measured at LHC

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1212 LHC Environment: Underlying Event Rick Field’s (CDF) view on di- jet events Collisions of other partons in the protons generating the signal interaction Collisions of other partons in the protons generating the signal interaction Unavoidable in hadron collisions Unavoidable in hadron collisions No real first principle calculations No real first principle calculations Low pT (non-pertubative) QCD Low pT (non-pertubative) QCD Some correlation with hard scattering Some correlation with hard scattering Typically tuned from data in physics generators Typically tuned from data in physics generators Carefully measured at Tevatron Carefully measured at Tevatron Phase space factor applied to LHC tune in absence of data Phase space factor applied to LHC tune in absence of data One of the first things to be measured at LHC One of the first things to be measured at LHC Look at activity (pT, # charged tracks) as function of leading jet pT in transverse region

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1313 LHC Environment: Underlying Event Collisions of other partons in the protons generating the signal interaction Collisions of other partons in the protons generating the signal interaction Unavoidable in hadron collisions Unavoidable in hadron collisions No real first principle calculations No real first principle calculations Low pT (non-pertubative) QCD Low pT (non-pertubative) QCD Activity shows some correlation with hard scattering (radiation?) Activity shows some correlation with hard scattering (radiation?) pTmin, pTmax differences pTmin, pTmax differences Typically tuned from data in physics generators Typically tuned from data in physics generators Carefully measured at Tevatron Carefully measured at Tevatron Phase space factor applied to LHC tune in absence of data Phase space factor applied to LHC tune in absence of data One of the first things to be measured at LHC One of the first things to be measured at LHC CDF data (√s=1.8 TeV) LHC prediction: x2.5 the activity measured at Tevatron! pT leading jet (GeV) Number charged tracks in transverse region CDF data: Phys.Rev, D, 65 (2002)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1414 LHC Environment: Pile-Up Multiple interactions between partons in other protons in the same bunch crossing Multiple interactions between partons in other protons in the same bunch crossing Consequence of high rate (luminosity) and high proton-proton total cross- section (~75 mb) Consequence of high rate (luminosity) and high proton-proton total cross- section (~75 mb) Statistically independent of hard scattering Statistically independent of hard scattering But similar models used for soft physics But similar models used for soft physics Signal history in calorimeter increases noise Signal history in calorimeter increases noise Signal 10-20 times slower than bunch crossing rate (25 ns) Signal 10-20 times slower than bunch crossing rate (25 ns) Noise has coherent character Noise has coherent character Cell signals linked through past shower developments Cell signals linked through past shower developments E t ~ 58 GeV E t ~ 81 GeV without pile-up

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1515 LHC Environment: Pile-Up Multiple interactions between partons in other protons in the same bunch crossing Multiple interactions between partons in other protons in the same bunch crossing Consequence of high rate (luminosity) and high proton-proton total cross- section (~75 mb) Consequence of high rate (luminosity) and high proton-proton total cross- section (~75 mb) Statistically independent of hard scattering Statistically independent of hard scattering But similar models used for soft physics But similar models used for soft physics Signal history in calorimeter increases noise Signal history in calorimeter increases noise Signal 10-20 times slower than bunch crossing rate (25 ns) Signal 10-20 times slower than bunch crossing rate (25 ns) Noise has coherent character Noise has coherent character Cell signals linked through past shower developments Cell signals linked through past shower developments E t ~ 58 GeV E t ~ 81 GeV with design luminosity pile-up

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1616 LHC Environment: Pile-Up Multiple interactions between partons in other protons in the same bunch crossing Multiple interactions between partons in other protons in the same bunch crossing Consequence of high rate (luminosity) and high proton-proton total cross- section (~75 mb) Consequence of high rate (luminosity) and high proton-proton total cross- section (~75 mb) Statistically independent of hard scattering Statistically independent of hard scattering But similar models used for soft physics But similar models used for soft physics Signal history in calorimeter increases noise Signal history in calorimeter increases noise Signal 10-20 times slower than bunch crossing rate (25 ns) Signal 10-20 times slower than bunch crossing rate (25 ns) Noise has coherent character Noise has coherent character Cell signals linked through past shower developments Cell signals linked through past shower developments

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1717 Why Is That Important? Jet calibration requirements very stringent Jet calibration requirements very stringent Systematic jet energy scale Systematic jet energy scale uncertainties to be extremely uncertainties to be extremely well controlled well controlled Top mass reconstruction Top mass reconstruction Relative jet energy resolution Relative jet energy resolution requirement requirement Inclusive jet cross-section Inclusive jet cross-section Di-quark mass spectra cut-off in SUSY Di-quark mass spectra cut-off in SUSY Event topology plays a role at 1% level of precision Event topology plays a role at 1% level of precision Extra particle production due to event color flow Extra particle production due to event color flow Color singlet (e.g., W) vs color octet (e.g., gluon/quark) jet source Color singlet (e.g., W) vs color octet (e.g., gluon/quark) jet source Small and large angle gluon radiation Small and large angle gluon radiation Quark/gluon jet differences Quark/gluon jet differences

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1818 Jet Algorithms – Rules & Guidelines

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 1919 Jetology 101 What are jets for experimentalists? What are jets for experimentalists? A bunch of particles generated by hadronization of a common source A bunch of particles generated by hadronization of a common source Quark, gluon fragmenation Quark, gluon fragmenation As a consequence, the particles in this bunch have correlated kinematic properties As a consequence, the particles in this bunch have correlated kinematic properties Reflecting the source by sum rules/conservations Reflecting the source by sum rules/conservations The interacting particles in this bunch generated an observable signal in a detector The interacting particles in this bunch generated an observable signal in a detector Protons, neutrons, pions, photons, electrons, muons, other particles with laboratory lifetimes >~10ps, and the corresponding anti-particles Protons, neutrons, pions, photons, electrons, muons, other particles with laboratory lifetimes >~10ps, and the corresponding anti-particles The non-interacting particles do not generate a directly observable signal The non-interacting particles do not generate a directly observable signal Neutrinos, mostly Neutrinos, mostly What is jet reconstruction, then? What is jet reconstruction, then? Model: attempt to collect the final state particles described above into objects (jets) representing the original parton kinematic Model: attempt to collect the final state particles described above into objects (jets) representing the original parton kinematic Re-establishing the correlations Re-establishing the correlations Experiment: attempt to collect the detector signals from these particles to measure their original kinematics Experiment: attempt to collect the detector signals from these particles to measure their original kinematics Usually not the parton! Usually not the parton!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2020 Jet finding algorithm and its configuration Jet finding algorithm and its configuration Seeded or seedless cone and its parameters Seeded or seedless cone and its parameters Cone size, seed threshold Cone size, seed threshold Recombination algorithm Recombination algorithm Recursive recombination algorithms Recursive recombination algorithms Distance parameter Distance parameter Recombination algorithm Recombination algorithm Signal or constituent definition Signal or constituent definition Calorimeter towers or clusters Calorimeter towers or clusters Reconstructed tracks Reconstructed tracks Generated particles Generated particles Generated partons Generated partons Useful concept to talk to other experiments and theorists Useful concept to talk to other experiments and theorists Also see Les Houches 2007 Also see Les Houches 2007 arXiv:0803.0678v1 [hep-ph] arXiv:0803.0678v1 [hep-ph] “Snowmass” 4-momentum Jet Definition

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2121 infrared sensitivity (soft gluon radiation merges jets) collinear sensitivity (2) (signal split into two towers below threshold) collinear sensitivity (1) (sensitive to E t ordering of seeds) Infrared safety Infrared safety Additional soft particles should not affect jet reconstruction Additional soft particles should not affect jet reconstruction Collinear safety Collinear safety Split energies (one instead of two particles) should not change the jet Split energies (one instead of two particles) should not change the jet Infrared safety Infrared safety Additional soft particles should not affect jet reconstruction Additional soft particles should not affect jet reconstruction Collinear safety Collinear safety Split energies (one instead of two particles) should not change the jet Split energies (one instead of two particles) should not change the jet Theoretical Requirements also see P. Francavilla’s talk!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2222 Theoretical Requirements infrared sensitivity (soft gluon radiation merges jets) collinear sensitivity (2) (signal split into two towers below threshold) collinear sensitivity (1) (sensitive to E t ordering of seeds) Infrared safety Infrared safety Additional soft particles should not affect jet reconstruction Additional soft particles should not affect jet reconstruction Collinear safety Collinear safety Split energies (one instead of two particles) should not change the jet Split energies (one instead of two particles) should not change the jet Infrared safety Infrared safety Additional soft particles should not affect jet reconstruction Additional soft particles should not affect jet reconstruction Collinear safety Collinear safety Split energies (one instead of two particles) should not change the jet Split energies (one instead of two particles) should not change the jet also see P. Francavilla’s talk!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2323 Detector technology independence Detector technology independence Jet efficiency should not depend on detector technology Jet efficiency should not depend on detector technology Final jet calibration and corrections ideally unfolds all detector effects Final jet calibration and corrections ideally unfolds all detector effects Minimal contribution from spatial and energy resolution to reconstructed jet kinematics Minimal contribution from spatial and energy resolution to reconstructed jet kinematics Unavoidable intrinsic detector limitations set limits Unavoidable intrinsic detector limitations set limits Stability within environment Stability within environment (Electronic) detector noise should not affect jet reconstruction within reasonable limits (Electronic) detector noise should not affect jet reconstruction within reasonable limits Energy resolution limitation Energy resolution limitation Avoid energy scale shift due to noise Avoid energy scale shift due to noise Stability with changing (instantaneous) luminosity Stability with changing (instantaneous) luminosity Control of underlying event and pile-up signal contribution Control of underlying event and pile-up signal contribution “Easy” to calibrate “Easy” to calibrate Small algorithm bias for jet signal Small algorithm bias for jet signal High reconstruction efficiency High reconstruction efficiency Identify all physically interesting jets from energetic partons in perturbative QCD Identify all physically interesting jets from energetic partons in perturbative QCD Jet reconstruction in resonance decays Jet reconstruction in resonance decays High efficiency to separate close-by jets from same particle decay High efficiency to separate close-by jets from same particle decay Least sensitivity to boost of particle Least sensitivity to boost of particle Efficient use of computing resources Efficient use of computing resources Balance physics requirements with available computing Balance physics requirements with available computing Fully specified algorithms only Fully specified algorithms only Absolutely need to compare to theory at particle and parton level Absolutely need to compare to theory at particle and parton level Pre-clustering strategy, energy/direction definitions, recombination rules, splitting and merging strategy if applicable Pre-clustering strategy, energy/direction definitions, recombination rules, splitting and merging strategy if applicable Experimental Requirements

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2424 Jet Algorithm Implementations See Paolo Francavilla’s talk for details!! See Paolo Francavilla’s talk for details!! Focus on physically meaningful jet finders Focus on physically meaningful jet finders Implementations follow at least theoretical requirements Implementations follow at least theoretical requirements ATLAS now focuses on AntiKt for performance evaluations ATLAS now focuses on AntiKt for performance evaluations Regularly shaped jets Regularly shaped jets Analytical access to area and overlaps (next slide) Analytical access to area and overlaps (next slide) Good performance for narrow AntiKt jets (R = 0.4) Good performance for narrow AntiKt jets (R = 0.4) As you will see at this workshop As you will see at this workshop Small nominal and actual area reduces effect of pile-up and underlying event Small nominal and actual area reduces effect of pile-up and underlying event Higher signal stability expected! Higher signal stability expected! Not ideal for more evolved jet analysis Not ideal for more evolved jet analysis E.g., substructure analysis in high pT jets from boosted heavy particle decays requires kT E.g., substructure analysis in high pT jets from boosted heavy particle decays requires kT also see P. Francavilla’s talk!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2525 Jet Shapes (from G. Salam’s talk at the ATLAS Hadronic Calibration Workshop Tucson 2008)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2626 Calorimeter Signal Reconstruction for Hadrons

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2727 Signals from Hadronic Showers More complex than EM showers More complex than EM showers Visible EM O(50%) Visible EM O(50%) e , ,  o  e , ,  o  Visible non-EM O(25%) Visible non-EM O(25%) Ionization of  , p,   Ionization of  , p,   Invisible O(25%) Invisible O(25%) Nuclear break-up Nuclear break-up Nuclear excitation Nuclear excitation Escaped O(2%) Escaped O(2%) Neutrinos produced in shower Neutrinos produced in shower Only part of the visible energy is sampled into the signal Only part of the visible energy is sampled into the signal Electromagnetic energy scale calibration does not recover loss Electromagnetic energy scale calibration does not recover loss Need additional corrections to measure hadron energy Need additional corrections to measure hadron energy Cell signal hard to calibrate without context Cell signal hard to calibrate without context Cell signal from electron or hadron? Cell signal from electron or hadron? Need context to apply correct calibration! Need context to apply correct calibration! EM shower RD3 note 41, 28 Jan 1993 Grupen, Particle Detectors

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2828 Shower Characteristics Side-by-Side Electromagnetic Electromagnetic Compact Compact Growths in depth ~log(E) Growths in depth ~log(E) Longitudinal extension scale is radiation length X 0 Longitudinal extension scale is radiation length X 0 Distance in matter in which ~50% of electron energy is radiated off Distance in matter in which ~50% of electron energy is radiated off Photons 9/7 X 0 Photons 9/7 X 0 Strong correlation between lateral and longitudinal shower development Strong correlation between lateral and longitudinal shower development Small intrinsic shower-to- shower fluctuations Small intrinsic shower-to- shower fluctuations Very regular development Very regular development Can be simulated with high precision Can be simulated with high precision 1% or better, depending on features 1% or better, depending on features Hadronic Hadronic Scattered, significantly bigger Scattered, significantly bigger Growths in depth ~log(E) Growths in depth ~log(E) Longitudinal extension scale is interaction length λ Longitudinal extension scale is interaction length λ Average distance between two inelastic interactions in matter Average distance between two inelastic interactions in matter Varies significantly for pions, protons, neutrons Varies significantly for pions, protons, neutrons Weak correlation between longitudinal and lateral shower development Weak correlation between longitudinal and lateral shower development Large intrinsic shower-to- shower fluctuations Large intrinsic shower-to- shower fluctuations Very irregular development Very irregular development Can be simulated with reasonable precision Can be simulated with reasonable precision ~2-5% depending on feature ~2-5% depending on feature

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 2929 Features of Hadronic Signals Each signal component fraction depends on energy Each signal component fraction depends on energy Visible non-EM fraction Visible non-EM fraction decreases with E decreases with E Hadronic signals more Hadronic signals more “electromagnetic” at high energy “electromagnetic” at high energy Measured by the electron/pion Measured by the electron/pion ratio measured at the same ratio measured at the same deposited energy deposited energy Hadron response not linear with Hadron response not linear with E in ATLAS, e/h > 1 for each sub-detector E in ATLAS, e/h > 1 for each sub-detector “e” is the intrinsic response to visible EM “e” is the intrinsic response to visible EM “h” is the intrinsic response to visible non-EM “h” is the intrinsic response to visible non-EM invisible energy is the main source of e/h > 1 invisible energy is the main source of e/h > 1 Large fluctuations of each component fraction Large fluctuations of each component fraction Non-compensation amplifies fluctuations Non-compensation amplifies fluctuations Hadronic calibration attempts to… Hadronic calibration attempts to… … provide some degree of software compensation … provide some degree of software compensation … account for the invisible and escaped energy … account for the invisible and escaped energy T.A. Gabriel, D.E. Groom, Nucl. Instr. Meth. A338 (1994) 336

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3030 Calorimeter Cells Smallest signal collection volume Smallest signal collection volume Defines resolution of spatial structures Defines resolution of spatial structures Finest granularity depends on direction and sampling layer Finest granularity depends on direction and sampling layer Each is read out independently Each is read out independently Can generate more than one signal (e.g., TileCell features two readouts) Can generate more than one signal (e.g., TileCell features two readouts) Individual cell signals Individual cell signals Sensitive to noise Sensitive to noise Fluctuations in electronics gain and shaping Fluctuations in electronics gain and shaping Time jitters Time jitters Physics sources like multiple proton interaction history in pile-up Physics sources like multiple proton interaction history in pile-up Hard to calibrate for hadrons Hard to calibrate for hadrons No measure to determine if electromagnetic or hadronic in cell signal alone, i.e. no handle to estimate e/h No measure to determine if electromagnetic or hadronic in cell signal alone, i.e. no handle to estimate e/h Need signal neighbourhood to calibrate Need signal neighbourhood to calibrate Basic energy scale Basic energy scale Use electron calibration to establish basic energy scale for cell signals Use electron calibration to establish basic energy scale for cell signals Cell geometry Cell geometry Quasi-projective by pointing to the nominal collision vertex in central and endcap ATLAS calorimeters Quasi-projective by pointing to the nominal collision vertex in central and endcap ATLAS calorimeters Lateral sizes scale with pseudo-rapidity and azimuthal angular opening Lateral sizes scale with pseudo-rapidity and azimuthal angular opening Non-projective in ATLAS Forward Calorimeter Non-projective in ATLAS Forward Calorimeter

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3131 Calorimeter Towers Impose a regular grid view on event Impose a regular grid view on event Δ  ×Δφ = 0.1×0.1 grid Δ  ×Δφ = 0.1×0.1 grid Motivated by particle Et flow in hadron-hadron collisions Motivated by particle Et flow in hadron-hadron collisions Well suited for trigger purposes Well suited for trigger purposes Collect cells into tower grid Collect cells into tower grid Cells EM scale signals are summed with geometrical weights Cells EM scale signals are summed with geometrical weights Depend on cell area containment ratio Depend on cell area containment ratio Weight = 1 for projective cells of equal or smaller than tower size Weight = 1 for projective cells of equal or smaller than tower size Summing can be selective Summing can be selective See jet input signal discussion See jet input signal discussion Towers have massless four- momentum representation Towers have massless four- momentum representation Fixed direction given by geometrical grid center Fixed direction given by geometrical grid center projective cells non-projectivecells

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3232 Topological Cell Clusters (1) Motivation Motivation Attempt reconstruction of individual particle showers Attempt reconstruction of individual particle showers Reconstruct 3-dim clusters of cells with correlated signals Reconstruct 3-dim clusters of cells with correlated signals Use shape of these clusters to locally calibrate them Use shape of these clusters to locally calibrate them Explore differences between electromagnetic and hadronic shower development and select best suited calibration Explore differences between electromagnetic and hadronic shower development and select best suited calibration Attempt to suppress noise with least bias on physics signals Attempt to suppress noise with least bias on physics signals Often less than 50% of all cells in an event with “real” signal Often less than 50% of all cells in an event with “real” signal Uses cell signal significance (signal over noise) Uses cell signal significance (signal over noise) Electronic Electronic Electronic + pile-up (quadratic sum) Electronic + pile-up (quadratic sum) Electronic Noise in Calorimeter Cells S. Menke, ATLAS Physics Workshop 07/2005 Pile-up Noise in Calorimeter Cells S. Menke, ATLAS Physics Workshop 07/2005

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3333 Topological Cell Clusters (2) TopoCluster algorithm implementation TopoCluster algorithm implementation Seeding (S) Seeding (S) Cells with signal significance above primary seed threshold Cells with signal significance above primary seed threshold Collecting (P) Collecting (P) Directly neighboring cells with signals above basic threshold, directly neighboring seed cells in 3-dim Directly neighboring cells with signals above basic threshold, directly neighboring seed cells in 3-dim Growth control (N) Growth control (N) Collect neighbors of neighbors if those have signals above secondary seed significance Collect neighbors of neighbors if those have signals above secondary seed significance Initial Cluster is done when no more such cells Initial Cluster is done when no more such cells Splitting Splitting Find local signal maxima in cluster with E > 500 MeV Find local signal maxima in cluster with E > 500 MeV Split cluster between those Split cluster between those Fine tuning Fine tuning Seeding Seeding Seeds can be restricted to certain calorimeter regions Seeds can be restricted to certain calorimeter regions Splitting Splitting Splitting is guided by EM calorimeter in big clusters Splitting is guided by EM calorimeter in big clusters Cells at split boundary are shared between clusters Cells at split boundary are shared between clusters Signal weight is function of cluster energies Signal weight is function of cluster energies Default “4/2/0” configuration for hadronic final state Default “4/2/0” configuration for hadronic final state Primary seed significance Primary seed significance S = 4 S = 4 Secondary seed significance Secondary seed significance N = 2 N = 2 Basic significance threshold Basic significance threshold P = 0, all cells neighboring a seed survive noise suppression! P = 0, all cells neighboring a seed survive noise suppression!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3434 Topological Cell Clusters (3)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3535 Topological Cell Clusters (4) cluster candidate #1 cluster candidate #2 #3?

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3636 4/2/0 TopoClusters  P  N  S 20 GeV pions Resolution of Sum E clus  P NN  S 180 GeV pions Resolution of Sum E clus SS NN PP Mean of Sum E clus Speckmayer, Carli 4,2,0 performs in the best way 4,2,0 performs in the best way beam test pions  = 0.45 beam test pions  = 0.45

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3737 Local Hadronic Calibration (LC) Origin of calorimeter signal Origin of calorimeter signal Attempt to classify energy deposit as electromagnetic or hadronic from the cluster signal and shape Attempt to classify energy deposit as electromagnetic or hadronic from the cluster signal and shape Allows to apply specific corrections and calibrations Allows to apply specific corrections and calibrations Local calibration approach Local calibration approach Use topological cell clusters as signal base for a hadronic energy scale Use topological cell clusters as signal base for a hadronic energy scale Recall cell signals need context for hadronic calibration Recall cell signals need context for hadronic calibration Basic concept is to reconstruct the locally deposited energy from the cluster signal first Basic concept is to reconstruct the locally deposited energy from the cluster signal first This is not the particle energy This is not the particle energy Additional corrections for energy losses with some correlation to the cluster signals and shapes extend the local scope Additional corrections for energy losses with some correlation to the cluster signals and shapes extend the local scope True signal loss due to the noise suppression in the cluster algorithm (still local) True signal loss due to the noise suppression in the cluster algorithm (still local) Dead material losses in front of, or between sensitive calorimeter volumes (larger scope than local deposit) Dead material losses in front of, or between sensitive calorimeter volumes (larger scope than local deposit) After all corrections, the reconstructed energy is on average the isolated particle energy After all corrections, the reconstructed energy is on average the isolated particle energy E.g., in a testbeam E.g., in a testbeam But not the jet energy (see later) But not the jet energy (see later)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3838 LC Sequence Electronic and readout effects unfolded (nA->GeV calibration) 3-d topological cell clustering includes noise suppression and establishes basic calorimeter signal for further processing Cluster shape analysis provides appropriate classification for calibration and corrections Cluster character depending calibration (cell signal weighting for HAD, to be developed for EM?) Apply dead material corrections specific for hadronic and electromagnetic clusters, resp. Apply specific out-of-cluster corrections for hadronic and electromagnetic clusters, resp.

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 3939 Cluster Classification Phase-space pion counting method Phase-space pion counting method Classify clusters using the correlation of Classify clusters using the correlation of Shower shape variables in single ± MC events Shower shape variables in single ± MC events = cluster barycenter depth in calo  = energy weighted average cell density Electromagnetic fraction estimator in bin of shower shape variables: Electromagnetic fraction estimator in bin of shower shape variables: Implementation Implementation keep F in bins of , E,,  of clusters for a given cluster keep F in bins of , E,,  of clusters for a given cluster If E < 0, then classify as unknown Lookup F from the observables ||, E,,  Cluster is EM if F > 50%, hadronic otherwise

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4040 Cluster Classification Efficiency watch for updates in LC session!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4141 LC Hadronic Calibration hadronic weight Cell signal Cell signal weighting weighting Based on H1 concept Based on H1 concept High cell signal High cell signal density – electromag- density – electromagnetic deposit netic deposit Low cell signal Low cell signal density – hadronic density – hadronic deposit deposit Weights calculated from Geant4 charged pion simulations Weights calculated from Geant4 charged pion simulations Large phase space covered in ATLAS geometry Large phase space covered in ATLAS geometry Calibration hits provide local deposited energy reference Calibration hits provide local deposited energy reference Weights parameterized as function of cluster properties Weights parameterized as function of cluster properties Averaged in bins of cluster energy and cell signal density Averaged in bins of cluster energy and cell signal density Weights stored in tables per sampling and direction Weights stored in tables per sampling and direction 2.0 < || < 2.2, HEC layer 1 watch for updates in LC session!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4242 Weights are extracted from simulations Weights are extracted from simulations Signals and deposited energies from single charged pion MC Signals and deposited energies from single charged pion MC Weights are calculated as function of cluster & cell variables Weights are calculated as function of cluster & cell variables Signal environment used as additional indicator of hadronic character Signal environment used as additional indicator of hadronic character LC Cell Signal Weighting

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4343 LC Dead Material Corrections Dead material Dead material Energy losses not directly measurable Energy losses not directly measurable Signal distribution in vicinity can help Signal distribution in vicinity can help Introduces need for signal corrections up to O(10%) Introduces need for signal corrections up to O(10%) Exclusive use of signal features Exclusive use of signal features Corrections depend on electromagnetic or hadronic energy deposit Corrections depend on electromagnetic or hadronic energy deposit Major contributions Major contributions Upstream materials Upstream materials Material between LArG and Tile (central) Material between LArG and Tile (central) Cracks Cracks dominant sources for signal losses dominant sources for signal losses |η|≈1.4-1.5 |η|≈1.4-1.5 |η|≈3.2 |η|≈3.2 Clearly affects detection efficiency for particles and jets Clearly affects detection efficiency for particles and jets Already in trigger! Already in trigger! Hard to recover jet reconstruction inefficiencies Hard to recover jet reconstruction inefficiencies Generate fake missing Et contribution Generate fake missing Et contribution Topology dependence of missing Et reconstruction quality Topology dependence of missing Et reconstruction quality Relative energy loss in dead material G. Pospelov’s contribution to this workshop

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4444 Out-of-cluster Corrections Compensate loss of true signal Compensate loss of true signal Limited efficiency of noise suppression scheme Limited efficiency of noise suppression scheme Discard cells with small true energy not close to a primary or secondary seed Discard cells with small true energy not close to a primary or secondary seed Accidental acceptance of a pure noise cell Accidental acceptance of a pure noise cell Can be significant for isolated pions Can be significant for isolated pions 10% at low energy 10% at low energy Correction derived from single pions Correction derived from single pions Compensates the isolated particle loss Compensates the isolated particle loss But in jets neighboring clusters can pick up lost energy But in jets neighboring clusters can pick up lost energy Use isolation moment to measure effective “free surface” of each cluster Use isolation moment to measure effective “free surface” of each cluster Scale single pion correction with this moment (0…1) Scale single pion correction with this moment (0…1) single pions QCD jets watch for updates in LC session!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4545 Attempt to calibrate hadronic calorimeter signals in smallest possible signal context Attempt to calibrate hadronic calorimeter signals in smallest possible signal context Topological clustering implements noise suppression with least bias signal feature extraction Topological clustering implements noise suppression with least bias signal feature extraction Residual concerns about infrared safety! Residual concerns about infrared safety! No bias towards a certain physics analysis No bias towards a certain physics analysis Calibration driven by calorimeter signal features without further assumption Calibration driven by calorimeter signal features without further assumption Good common signal base for all hadronic final state objects Good common signal base for all hadronic final state objects Jets, missing Et, taus Jets, missing Et, taus Factorization of cluster calibration Factorization of cluster calibration Cluster classification largely avoids application of hadronic calibration to electromagnetic signal objects Cluster classification largely avoids application of hadronic calibration to electromagnetic signal objects Low energy regime challenging Low energy regime challenging Signal weights for hadronic calibration are functions of cluster and cell parameters and variables Signal weights for hadronic calibration are functions of cluster and cell parameters and variables Cluster energy and direction Cluster energy and direction Cell signal density and location (sampling layer) Cell signal density and location (sampling layer) Dead material and out of cluster corrections are independently applicable Dead material and out of cluster corrections are independently applicable Summary LC Features

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4646 Jet Reconstruction & Calibration

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4747 Jet Reconstruction Task Overview Experiment (“Nature”)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4848 Jet Reconstruction Task Overview Experiment (“Nature”) Particles UE MB Multiple Interactions Stable Particles Decays Jet Finding Particle Jets Generated Particles Generated Particles Modeling Particle Jets

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 4949 Jet Reconstruction Task Overview Modeling Calorimeter Jets Stable Particles Raw Calorimeter Signals Detector Simulation Reconstructed Calorimeter Signals Signal Reconstruction Jet Finding Reconstructed Jets Identified Particles Experiment (“Nature”)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5050 Jet Reconstruction Task Overview Measuring Calorimeter Jets Reconstructed Jets Observable Particles Raw Calorimeter Signals Measurement Reconstructed Calorimeter Signals Signal Reconstruction Jet Finding Identified Particles Experiment (“Nature”)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5151 Jet Reconstruction Task Overview Jet Reconstruction Challenges physics reaction of interest (interaction or parton level) added tracks from underlying event added tracks from in-time (same trigger) pile-up event jet reconstruction algorithm efficiency added tracks from underlying event added tracks from in-time (same trigger) pile-up event jet reconstruction algorithm efficiency longitudinal energy leakage detector signal inefficiencies (dead channels, HV…) pile-up noise from (off- and in-time) bunch crossings electronic noise calo signal definition (clustering, noise suppression…) dead material losses (front, cracks, transitions…) detector response characteristics (e/h ≠ 1) jet reconstruction algorithm efficiency lost soft tracks due to magnetic field longitudinal energy leakage detector signal inefficiencies (dead channels, HV…) pile-up noise from (off- and in-time) bunch crossings electronic noise calo signal definition (clustering, noise suppression…) dead material losses (front, cracks, transitions…) detector response characteristics (e/h ≠ 1) jet reconstruction algorithm efficiency lost soft tracks due to magnetic field Experiment (“Nature”)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5252 Sequential process Sequential process Input signal selection Input signal selection Get the best signals out of your detector on a given signal scale Get the best signals out of your detector on a given signal scale Preparation for jet finding Preparation for jet finding Suppression/cancellation of “unphysical” signal objects with E<0 (due to noise) Suppression/cancellation of “unphysical” signal objects with E<0 (due to noise) Possibly event ambiguity resolution (remove reconstructed electrons, photons, taus,… from detector signal) Possibly event ambiguity resolution (remove reconstructed electrons, photons, taus,… from detector signal) Pre-clustering to speed up reconstruction (not needed as much anymore) Pre-clustering to speed up reconstruction (not needed as much anymore) Jet finding Jet finding Apply your jet finder of choice Apply your jet finder of choice Jet calibration Jet calibration Depending on detector, jet finder choices, references… Depending on detector, jet finder choices, references… Jet selection Jet selection Apply cuts on kinematics etc. to select jets of interest or significance Apply cuts on kinematics etc. to select jets of interest or significance Objective Objective Reconstruct particle level features Reconstruct particle level features Test models and extract physics Test models and extract physics “How-to” Of Jet Reconstruction

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5353 Input Signals (1) Topological cell clusters (“TopoClusters”) Topological cell clusters (“TopoClusters”) Remove all negative energy clusters Remove all negative energy clusters Noise suppressed at cluster level, no cancellation strategy needed Noise suppressed at cluster level, no cancellation strategy needed Negative clusters have no or insignificant physics signal Negative clusters have no or insignificant physics signal Clusters provide flexible signal Clusters provide flexible signal Three signal states available Three signal states available “UNCALIBRATED” – electromagnetic energy scale, cluster energy is cell energy sum including possible topological weights from cluster splitting “CALIBRATED” – fully calibrated (LC) cluster, cluster energy is sum of weighted cell energies where weights represent the full correction in LC projected back to cell level “ALTCALIBRATED” – cluster kinematics calculated from cell weights from the jet calibration (see later) Subject positive energy clusters to jet finding Subject positive energy clusters to jet finding Choice of signal state of cluster Choice of signal state of cluster

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5454 Input Signals (2) Unbiased towers (“CaloTowers”) Unbiased towers (“CaloTowers”) Electromagnetic energy scale signal only Electromagnetic energy scale signal only No hadronic tower calibration scheme available No hadronic tower calibration scheme available All cells included All cells included Can have net negative energy at fixed direction Can have net negative energy at fixed direction No noise suppression No noise suppression Apply tower noise cancellation Apply tower noise cancellation Sum towers signals in vicinity of negative tower until energy sum > 0 Sum towers signals in vicinity of negative tower until energy sum > 0 Discard negative signal tower if re-summation does not generate a positive signal (vicinity exhausted) Discard negative signal tower if re-summation does not generate a positive signal (vicinity exhausted) Note that in this re-summation the original towers are preserved Note that in this re-summation the original towers are preserved Subject all original and merged towers to jet finding Subject all original and merged towers to jet finding Note that the resulting jets will have the original towers as constituents, not the merged ones! Note that the resulting jets will have the original towers as constituents, not the merged ones!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5555 Input Signals (3) Noise suppressed towers (“TopoTowers”) Noise suppressed towers (“TopoTowers”) Electromagnetic energy scale signal only Electromagnetic energy scale signal only Could be extended in the future to LC and cell weighting signals! Could be extended in the future to LC and cell weighting signals! Only cells surviving noise suppression are used in the towers Only cells surviving noise suppression are used in the towers TopoClusters are used as noise suppression tool TopoClusters are used as noise suppression tool Additional cell filters possible Additional cell filters possible Remove all negative energy TopoTowers Remove all negative energy TopoTowers Same argument as for TopoClusters – noise suppression already applied, no cancellation needed Same argument as for TopoClusters – noise suppression already applied, no cancellation needed Subject positive energy TopoTowers to jet finding Subject positive energy TopoTowers to jet finding Jet constituents are these towers Jet constituents are these towers

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5656 Electromagnetic Energy Scale Input Tower jets Tower jets Jets are found using electromagnetic energy scale signals from towers Jets are found using electromagnetic energy scale signals from towers Only option – and done like this in most other experiments Only option – and done like this in most other experiments May be problem for recursive recombination jet finders as tower signals may be wrong by several 10% relative to each other May be problem for recursive recombination jet finders as tower signals may be wrong by several 10% relative to each other Jets are calibrated after formation Jets are calibrated after formation Apply a jet context calibration using cells (cell weighting) and/or sampling energies (sampling weights) Apply a jet context calibration using cells (cell weighting) and/or sampling energies (sampling weights) Dead material and magnetic field effects intrinsically corrected by calibration functions Dead material and magnetic field effects intrinsically corrected by calibration functions Cluster jets Cluster jets Jets can be found using electromagnetic energy scale signals Jets can be found using electromagnetic energy scale signals One option with the same potential precision problem as for towers One option with the same potential precision problem as for towers Jets are calibrated after formation Jets are calibrated after formation Same as for towers as weights do not show significant dependence on calorimeter signal definition Same as for towers as weights do not show significant dependence on calorimeter signal definition

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5757 Local Hadronic Scale Input Clusters only Clusters only Jets are found in a signal space with e/h compensated and some other signal efficiencies already corrected Jets are found in a signal space with e/h compensated and some other signal efficiencies already corrected Intuitively better for recursive recombination algorithms (kT – flavors) Intuitively better for recursive recombination algorithms (kT – flavors) Jet needs final set of corrections Jet needs final set of corrections Local hadronic scale does not have a jet context Local hadronic scale does not have a jet context Dead material energy losses not correlated with cluster signals Dead material energy losses not correlated with cluster signals Magnetic field bends charged particles with pT < 400-500 MeV away from the calorimeter Magnetic field bends charged particles with pT < 400-500 MeV away from the calorimeter Cluster jet composition depends on input scale Cluster jet composition depends on input scale Jets found with electromagnetic signals cannot be expected to be the same as the ones found with locally calibrated signals Jets found with electromagnetic signals cannot be expected to be the same as the ones found with locally calibrated signals At least not with all aspects, especially composition details At least not with all aspects, especially composition details Main features may be similar Main features may be similar

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5858 Calibration Flow Lots of work in calorimeter domain!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 5959 Sum up electromagnetic scale calorimeter cell signals into towers Sum up electromagnetic scale calorimeter cell signals into towers Fixed grid of Δ η x Δ φ = 0.1 x 0.1 Fixed grid of Δ η x Δ φ = 0.1 x 0.1 Non-discriminatory, no cell suppression Non-discriminatory, no cell suppression Works well with pointing readout geometries Works well with pointing readout geometries Larger cells split their signal between towers according to the overlap area fraction Larger cells split their signal between towers according to the overlap area fraction Tower noise suppression Tower noise suppression Some towers have net negative signals Some towers have net negative signals Apply “nearest neighbour tower recombination” Apply “nearest neighbour tower recombination” Combine negative signal tower(s) with nearby positive signal towers until sum of signals > 0 Combine negative signal tower(s) with nearby positive signal towers until sum of signals > 0 Remove towers with no nearby neighbours Remove towers with no nearby neighbours Towers are “massless” pseudo-particles Towers are “massless” pseudo-particles Find jets Find jets Note: towers have signal on electromagnetic energy scale Note: towers have signal on electromagnetic energy scale Calibrate jets Calibrate jets Retrieve calorimeter cell signals in jet Retrieve calorimeter cell signals in jet Apply signal weighting functions to these signals Apply signal weighting functions to these signals Recalculate jet kinematics using these cell signals Recalculate jet kinematics using these cell signals Note: there are cells with negative signals! Note: there are cells with negative signals! Apply final corrections Apply final corrections CaloTower Jets

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6060 TopoCluster Jets

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6161 Apply noise suppression to tower jets Apply noise suppression to tower jets Topological clustering is used as a noise suppression tool only Topological clustering is used as a noise suppression tool only Similar to DZero approach Similar to DZero approach New implementation New implementation Only in ESD context so far Only in ESD context so far Working on schema to bring these jets into the AOD Working on schema to bring these jets into the AOD Including constituents Including constituents Allows comparisons for tower and cluster jets with similar noise contribution Allows comparisons for tower and cluster jets with similar noise contribution Should produce rather similar jets than tower jets at better resolution Should produce rather similar jets than tower jets at better resolution Less towers per jet Less towers per jet TopoTower Jets

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6262 Signal Choice Affects Jet Shape

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6363 Can we get the hadronic shower branch signals up to a signal corresponding to the electromagnetic signal? Can we get the hadronic shower branch signals up to a signal corresponding to the electromagnetic signal? Reduce fluctuations Reduce fluctuations Direct proportionality of energy and signal Direct proportionality of energy and signal One approach: cell signal weighting in highly granular calorimeter One approach: cell signal weighting in highly granular calorimeter Small signal densities in a calorimeter cell indicate hadronic deposit and should receive an additional correction (weight) Small signal densities in a calorimeter cell indicate hadronic deposit and should receive an additional correction (weight) Pioneered by CDHS (1977) and developed by H1 (1992) Pioneered by CDHS (1977) and developed by H1 (1992) High signal densities indicate electromagnetic signals and don’t need additional corrections High signal densities indicate electromagnetic signals and don’t need additional corrections Dense, compact showers from electrons/photons Dense, compact showers from electrons/photons But how can we determine these weights? But how can we determine these weights? It’s mostly a matter of context: are we trying to determine them for single particles (clusters) or jets It’s mostly a matter of context: are we trying to determine them for single particles (clusters) or jets ATLAS works with both approaches ATLAS works with both approaches Jets from both tower signals use jet context cell weights Jets from both tower signals use jet context cell weights Jets from topological clusters can use jet context cell weights as well as cell weights determined in the cluster context in local hadronic calibration (see earlier) Jets from topological clusters can use jet context cell weights as well as cell weights determined in the cluster context in local hadronic calibration (see earlier) Dealing With Non-Compensation

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6464 Jet Calibration With Cell Weights Weights are determined by resolution minimization fits with calorimeter jets in “good” detector regions Weights are determined by resolution minimization fits with calorimeter jets in “good” detector regions Use full simulations of QCD di-jets to fit w(…) such that Use full simulations of QCD di-jets to fit w(…) such that Truth reference typically matching simulated particle jet Truth reference typically matching simulated particle jet Experimental constraints possible, e.g. using photon-jet balance Experimental constraints possible, e.g. using photon-jet balance Weight determination includes primary electromagnetic component of jet Weight determination includes primary electromagnetic component of jet Dense cell signals have weights ~1 Dense cell signals have weights ~1 Weights compensate all signal inefficiencies, not only e/h Weights compensate all signal inefficiencies, not only e/h Dead material corrections, leakage, possible low level of factorization! Dead material corrections, leakage, possible low level of factorization!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6565 Cell Weight Calibration For Jets Cell signal weighting functions do not restore jet energy scale for all jets Cell signal weighting functions do not restore jet energy scale for all jets Crack regions not included in fits Crack regions not included in fits Only on jet context used for fitting weights Only on jet context used for fitting weights Cone jets with R=0.7 Cone jets with R=0.7 Only one calorimeter signal definition used for weight fits Only one calorimeter signal definition used for weight fits CaloTowers CaloTowers Additional response corrections applied to restore linearity Additional response corrections applied to restore linearity Non-optimal resolution for other than reference jet samples can be expected Non-optimal resolution for other than reference jet samples can be expected Changing physics environment not explicitly corrected Changing physics environment not explicitly corrected Absolute precision limitation Absolute precision limitation

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6666 Cell Weight Calibration For Jets Cell signal weighting functions do not restore jet energy scale for all jets Cell signal weighting functions do not restore jet energy scale for all jets Crack regions not included in fits Crack regions not included in fits Only on jet context used for fitting weights Only on jet context used for fitting weights Cone jets with R=0.7 Cone jets with R=0.7 Only one calorimeter signal definition used for weight fits Only one calorimeter signal definition used for weight fits CaloTowers CaloTowers Additional response corrections applied to restore linearity Additional response corrections applied to restore linearity Non-optimal resolution for other than reference jet samples can be expected Non-optimal resolution for other than reference jet samples can be expected Changing physics environment not explicitly corrected Changing physics environment not explicitly corrected Absolute precision limitation Absolute precision limitation

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6767 Cell Weight Calibration For Jets Cell signal weighting functions do not restore jet energy scale for all jets Cell signal weighting functions do not restore jet energy scale for all jets Crack regions not included in fits Crack regions not included in fits Only one jet context used for fitting weights Only one jet context used for fitting weights Cone jets with R=0.7 Cone jets with R=0.7 Only one calorimeter signal definition used for weight fits Only one calorimeter signal definition used for weight fits CaloTowers CaloTowers Additional response corrections applied to restore linearity Additional response corrections applied to restore linearity Non-optimal resolution for other than reference jet samples can be expected Non-optimal resolution for other than reference jet samples can be expected Changing physics environment not explicitly corrected Changing physics environment not explicitly corrected Absolute precision limitation Absolute precision limitation Response for jets in ttbar (same jet finder as used for determination of calibration functions with QCD events) Snapshot!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6868 Local Hadronic Calibration (LC) Calibrating calorimeter signals first Calibrating calorimeter signals first No jet context No jet context But need other context for cell signal weighting normalization → topological cell cluster But need other context for cell signal weighting normalization → topological cell cluster Energy blobs follow shower shape somewhat Energy blobs follow shower shape somewhat Cluster based hadronic calibration Cluster based hadronic calibration Advantages to jet context: can use cluster shape to parameterize cell weights Advantages to jet context: can use cluster shape to parameterize cell weights Measure compactness of signal cluster by cluster Measure compactness of signal cluster by cluster Shape and size variables are easily reconstructed for each cluster Shape and size variables are easily reconstructed for each cluster Allows factorization Allows factorization Deal with e/h at detector level (not jet level) Deal with e/h at detector level (not jet level) Correct for local dead material at cluster level already Correct for local dead material at cluster level already Requires cluster signal to have some correlation with loss Requires cluster signal to have some correlation with loss

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 6969 LC Features In Jets Not a jet calibration per se Not a jet calibration per se Energy lost in particles leaving no or weak signal trace in calorimeter not recoverable at cluster level Energy lost in particles leaving no or weak signal trace in calorimeter not recoverable at cluster level Magnetic field Magnetic field Dead material Dead material Need jet level corrections on top of this scale Need jet level corrections on top of this scale Transition hadronic to jet energy scale can depend on jet features Transition hadronic to jet energy scale can depend on jet features

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7070 LC Features In Jets Not a jet calibration per se Not a jet calibration per se Energy lost in particles leaving no or weak signal trace in calorimeter not recoverable at cluster level Energy lost in particles leaving no or weak signal trace in calorimeter not recoverable at cluster level Magnetic field Magnetic field Dead material Dead material Need jet level corrections on top of this scale Need jet level corrections on top of this scale Transition hadronic to jet energy scale can depend on jet features Transition hadronic to jet energy scale can depend on jet features Jet level correction for jets from locally calibrated clusters can recover jet energy Example: correction based on jet width measure watch for updates in jet energy scale session!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7171 Jet Energy Scale Corrections

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7272 Final Jet Energy Scale Calibration Jet energy scale (JES) for first data Jet energy scale (JES) for first data Fully Monte Carlo based calibrations hard to validate quickly with initial data Fully Monte Carlo based calibrations hard to validate quickly with initial data Too many things have to be right, including underlying event tunes, pile-up activity, etc. Too many things have to be right, including underlying event tunes, pile-up activity, etc. Mostly a generator issue in the beginning Mostly a generator issue in the beginning Need flat response and decent energy resolution for jets as soon as possible Need flat response and decent energy resolution for jets as soon as possible Data driven scenario a la DZero implemented Data driven scenario a la DZero implemented Major topic of this workshop – only overview here! Major topic of this workshop – only overview here! Additional jet by jet corrections Additional jet by jet corrections Interesting ideas to use all observable signal features for jets to calibrate Interesting ideas to use all observable signal features for jets to calibrate Geometrical moments Geometrical moments Energy sharing in calorimeters Energy sharing in calorimeters Concerns about stability and MC dependence to be understood Concerns about stability and MC dependence to be understood Can consider e.g. truncated moments using only prominent constituents for stable signal Can consider e.g. truncated moments using only prominent constituents for stable signal

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7373 JES Correction Model For First Data optional data driven MC Note that sequence is essential, but not a settled subject Note that sequence is essential, but not a settled subject Expect discussions at this workshop Expect discussions at this workshop watch for updates in LC session!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7474 Data Driven JES Corrections (1) PileUp subtraction (see D. Miller’s contribution to this meeting!) PileUp subtraction (see D. Miller’s contribution to this meeting!) Goal: Goal: Correct in-time and residual out-of-time pile-up contribution to a jet on average Correct in-time and residual out-of-time pile-up contribution to a jet on average Tools: Tools: Zero bias (random) events, minimum bias events Zero bias (random) events, minimum bias events Measurement: Measurement: Et density in Δ  ×Δφ bins as function of Et density in Δ  ×Δφ bins as function of # vertices # vertices TopoCluster feature (size, average TopoCluster feature (size, average energy as function of depth) changes energy as function of depth) changes as function of # vertices as function of # vertices Remarks: Remarks: Uses expectations from the average Et flow for a given instantaneous luminosity Uses expectations from the average Et flow for a given instantaneous luminosity Instantaneous luminosity is measured by the # vertices in the event Instantaneous luminosity is measured by the # vertices in the event Requires measure of jet size (AntiKt advantage) Requires measure of jet size (AntiKt advantage) Concerns: Concerns: Stable and safe determination of average Stable and safe determination of average D. Miller’s contribution to JES session

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7575 Data Driven JES Corrections (2) Absolute response Absolute response Goal: Goal: Correct for energy (pT) dependent jet response Correct for energy (pT) dependent jet response Tools: Tools: Direct photons, Z+jet(s),… Direct photons, Z+jet(s),… Measurement: Measurement: pT balance of well calibrated system (photon, Z) pT balance of well calibrated system (photon, Z) against jet in central region against jet in central region Remarks: Remarks: Usually uses central reference and central jets (region of flat reponse) Usually uses central reference and central jets (region of flat reponse) Concerns: Concerns: Limit in precision and estimates for systematics w/o well understood simulations not clear Limit in precision and estimates for systematics w/o well understood simulations not clear several contributions to JES session

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7676 Data Driven JES Corrections (3) Direction response corrections Direction response corrections Goal: Goal: Equalize response as function of jet Equalize response as function of jet (pseudo)rapidity (pseudo)rapidity Tools: Tools: QCD di-jets QCD di-jets Direct photons Direct photons Measurement: Measurement: Di-jet pT balance uses Di-jet pT balance uses reference jet in well calibrated reference jet in well calibrated (central) region to correct (central) region to correct second jet further away second jet further away Measure hadronic response Measure hadronic response variations as function of the jet variations as function of the jet direction with the missing Et direction with the missing Et projection fraction (MPF) method projection fraction (MPF) method Remarks: Remarks: MPF only needs jet for direction reference MPF only needs jet for direction reference Bi-sector in di-jet balance explores different sensitivities Bi-sector in di-jet balance explores different sensitivities Concerns: Concerns: MC quality for systematic uncertaunty evaluation MC quality for systematic uncertaunty evaluation Very different (jet) energy scales between reference and probed jet Very different (jet) energy scales between reference and probed jet uncalibrated several contributions to JES session

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7777 Numerical Inversion Transfer of response function from dependence on true variable to dependence on measured variable

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7878 Jets Not From Hard Scatter Dangerous background for W+n jets cross-sections etc. Dangerous background for W+n jets cross-sections etc. Lowest pT jet of final state can be faked or misinterpreted as coming from underlying event or multiple interactions Lowest pT jet of final state can be faked or misinterpreted as coming from underlying event or multiple interactions Underlying event: multi-parton interactions Underlying event: multi-parton interactions Multiple proton interactions: pile-up Multiple proton interactions: pile-up Extra jets from UE are hard to handle Extra jets from UE are hard to handle No real experimental indication of jet source No real experimental indication of jet source Some correlation with hard scattering? Some correlation with hard scattering? Jet area? Jet area? No separate vertex No separate vertex Jet-by-jet handle for multiple interactions Jet-by-jet handle for multiple interactions Classic indicator for multiple interactions is number of reconstructed vertices in event Classic indicator for multiple interactions is number of reconstructed vertices in event Tevatron with RMS(z_vertex) ~ 30 cm Tevatron with RMS(z_vertex) ~ 30 cm LHC RMS(z_vertex) ~ 8 cm LHC RMS(z_vertex) ~ 8 cm If we can attach vertices to reconstructed jets, we can in principle identify jets not from hard scattering If we can attach vertices to reconstructed jets, we can in principle identify jets not from hard scattering Limited to pseudorapidities within 2.5! Limited to pseudorapidities within 2.5!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 7979 Match tracks in track jet with calorimeter jet Match tracks in track jet with calorimeter jet Calculate pT fraction coming from each vertex for given jet Calculate pT fraction coming from each vertex for given jet Jets with little pT from primary vertex are likely from multiple interactions (e.g. pile- up) Jets with little pT from primary vertex are likely from multiple interactions (e.g. pile- up) ATLAS MC (preliminary) ATLAS MC (preliminary) Application of Trackjets

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8080 Software Aspects for Jet Reconstruction and Calibration (see Kerstin Perez’ talk)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8181 Jet Reconstruction Software Algorithm design guidelines Algorithm design guidelines Simple algorithm structure Simple algorithm structure Jet reconstruction is defined by combining tools in the appropriate way Jet reconstruction is defined by combining tools in the appropriate way Typical sequence: input filter – jet finder – (jet calibrator) – jet output filter Typical sequence: input filter – jet finder – (jet calibrator) – jet output filter All managed by one highly configurable, generic algorithm All managed by one highly configurable, generic algorithm Universal code for jet finding implementation Universal code for jet finding implementation Same algorithm and tools for calorimeter, track, “truth particle” jets Same algorithm and tools for calorimeter, track, “truth particle” jets FastJet libraries (Salam et al.) used for actual jet finder implementations (kT, C/A kT, AntiKt, SISCone,…) FastJet libraries (Salam et al.) used for actual jet finder implementations (kT, C/A kT, AntiKt, SISCone,…) Except ATLAS legacy cone (not IR safe beyond LO) Specific code for calibration can be plugged in Specific code for calibration can be plugged in E.g., cell weight calibrator tool E.g., cell weight calibrator tool

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8282 Re-reconstructing Jets Always possible from ESD with your favorite jet finder Always possible from ESD with your favorite jet finder CaloTowers, TopoTowers, TopoClusters CaloTowers, TopoTowers, TopoClusters Can apply cell weighting calibration Can apply cell weighting calibration Can reconstruct LC jets Can reconstruct LC jets Possible from AOD with restrictions Possible from AOD with restrictions Only TopoClusters available Only TopoClusters available Electromagnetic energy scale jets Electromagnetic energy scale jets LC jets LC jets New – cell weighted cluster signal can be explored New – cell weighted cluster signal can be explored Detailed documentation on ATLAS TWiki Detailed documentation on ATLAS TWiki Also tutorial on Saturday afternoon! Also tutorial on Saturday afternoon! see tutorial on Saturday!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8383 Missing Et (MET) Reconstruction

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8484 Detector Signal MET Contributions Hard signal in calorimeters Hard signal in calorimeters Fully reconstructed & calibrated particles and jets Fully reconstructed & calibrated particles and jets Not always from hard interaction! Not always from hard interaction! Hard signal in muon spectrometer Hard signal in muon spectrometer Fully reconstructed & calibrated muons Fully reconstructed & calibrated muons May generate isolated or embedded soft calorimeter signals May generate isolated or embedded soft calorimeter signals Care needed to avoid double counting Care needed to avoid double counting Soft signals in calorimeters Soft signals in calorimeters Signals not used in reconstructed physics objects Signals not used in reconstructed physics objects I.e., below reco threshold(s) I.e., below reco threshold(s) Needs to be included in MET to reduce scale biases and improve resolution Needs to be included in MET to reduce scale biases and improve resolution Need to avoid double counting Need to avoid double counting Common object use strategy in ATLAS Common object use strategy in ATLAS Find smallest available calorimeter signal base for physics objects (cells or cell clusters) Find smallest available calorimeter signal base for physics objects (cells or cell clusters) Check for exclusive bases Check for exclusive bases Same signal can only be used in one physics object Same signal can only be used in one physics object Veto MET contribution from already used signals Veto MET contribution from already used signals Track with selected signal base Track with selected signal base Priority of association is defined by reconstruction uncertainties Priority of association is defined by reconstruction uncertainties Electrons (highest quality) → photons → muons* → taus → jets (lowest quality) Electrons (highest quality) → photons → muons* → taus → jets (lowest quality)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8585 MET Calorimeter Issues Calorimeter issues Calorimeter issues About 70-90% of all cells have no About 70-90% of all cells have no true or significant signal true or significant signal Depending on final state, of course Depending on final state, of course Applying symmetric or asymmetric Applying symmetric or asymmetric noise cuts to cell signals noise cuts to cell signals Reduces fluctuations significantly Reduces fluctuations significantly But introduces a bias (shift in But introduces a bias (shift in average missing Et) average missing Et) Topological clustering applies more reasonable noise cut Topological clustering applies more reasonable noise cut Cells with very small signals can survive based on the signals in neighboring cells Cells with very small signals can survive based on the signals in neighboring cells Still small bias possible but close-to-ideal suppression of noise Still small bias possible but close-to-ideal suppression of noise K. Cranmer, in talk by S. Menke, ATLAS Physics Workshop 07/2005

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8686 MET Validation & Calibration MET is determined by hard signals in event MET is determined by hard signals in event Reconstructed particles and jets above threshold Reconstructed particles and jets above threshold All objects on well defined energy scale, e.g. best reconstruction for individual object type All objects on well defined energy scale, e.g. best reconstruction for individual object type Really no freedom to change scales for any of these objects Really no freedom to change scales for any of these objects Little calibration to be done for MET Little calibration to be done for MET Note that detector inefficiencies are corrected for physics objects Note that detector inefficiencies are corrected for physics objects Some freedom for soft MET contribution… Some freedom for soft MET contribution… Signals not used in physics objects often lack corresponding context to constrain calibration Signals not used in physics objects often lack corresponding context to constrain calibration Low bias LC based on signal shapes inside calorimeters helps Low bias LC based on signal shapes inside calorimeters helps Some degree of freedom here Some degree of freedom here But contribution is small and mostly balanced in Et anyway But contribution is small and mostly balanced in Et anyway Source here often UE/pile-up! …and overall acceptance limitations …and overall acceptance limitations Detector “loses” particles in non-instrumented areas or due to magnetic field in inner cavity Detector “loses” particles in non-instrumented areas or due to magnetic field in inner cavity Same remarks as above, very small and likely balanced signals Same remarks as above, very small and likely balanced signals Event topology dependent adjustments to MET are imaginable to recover these losses Event topology dependent adjustments to MET are imaginable to recover these losses I prefer “validation” rather than “calibration” I prefer “validation” rather than “calibration” Discrepancies in MET need to be isolated for systematic control Discrepancies in MET need to be isolated for systematic control

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8787 MET scale can be checked with physics MET scale can be checked with physics Look for one hadronic and one leptonic tau from Z decays Look for one hadronic and one leptonic tau from Z decays Can be triggered nicely with lepton + MET requirement Can be triggered nicely with lepton + MET requirement Use collinear approximation to reconstruct invariant mass Use collinear approximation to reconstruct invariant mass Massless taus Massless taus Neutrinos assumed to be collinear to observable tau decay products Neutrinos assumed to be collinear to observable tau decay products Check dependence of invariant mass on MET scale variations Check dependence of invariant mass on MET scale variations Expect correlation! Expect correlation! Determined from two reconstructed MET components and directions of detectable decay products CERN-OPEN-2008-020 100 pb-1 Z Mass Constraint

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8888 What is that? What is that? MET contribution from response variations MET contribution from response variations Cracks, azimuthal response variations… Cracks, azimuthal response variations… Never/slowly changing Never/slowly changing Particle dependent Particle dependent MET contribution from mis- calibration MET contribution from mis- calibration E.g., QCD di-jet with one jet under-calibrated E.g., QCD di-jet with one jet under-calibrated Relative effect generates MET pointing to this jet Relative effect generates MET pointing to this jet Dangerous source of MET Dangerous source of MET Disturbs many final states in a different way Disturbs many final states in a different way Can fake new physics Can fake new physics Suppression strategies Suppression strategies Track jets Track jets Energy sharing between calorimeters Energy sharing between calorimeters Event topology analysis Event topology analysis CERN-OPEN-2008-020 (modeled material asymmetry) Fake Missing Et

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 8989 MET resolution MET resolution measure measure MET resolution in MET resolution in each component as each component as function of scalar Et function of scalar Et sum for various final sum for various final states states Systematically Systematically Evaluated for performance monitoring Evaluated for performance monitoring Careful – scalar Et very sensitive to pile-up, detector effects, etc. Careful – scalar Et very sensitive to pile-up, detector effects, etc. No direct experimental access No direct experimental access Minimum bias with limited reach/precision? Minimum bias with limited reach/precision? Concern is pile-up effect on scalar Et Concern is pile-up effect on scalar Et CERN-OPEN-2008-020 MET Resolution From MC

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9090 MET Scale & Resolution Experimental access Experimental access With bi-sector signal projections in Z decays With bi-sector signal projections in Z decays Longitudinal projection sensitive to scale Longitudinal projection sensitive to scale Calibration of hadronic recoil Calibration of hadronic recoil Perpendicular projection sensitive to angular resolution Perpendicular projection sensitive to angular resolution Neutrinofication Neutrinofication Assumed to be very similar in Z and W Assumed to be very similar in Z and W One lepton in Z decay can be “neutrinofied” One lepton in Z decay can be “neutrinofied” Access to MET resolution Access to MET resolution see E.Dobson’s contributions to MET session

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9191 MET scale MET scale Folds hadronic scale with acceptance Folds hadronic scale with acceptance Note: no jets needed! Note: no jets needed! Experimental tool to validate calibration of “unused” calorimeter signal Experimental tool to validate calibration of “unused” calorimeter signal Hard objects can be removed from recoil Hard objects can be removed from recoil One possible degree of freedom in MET “calibration” One possible degree of freedom in MET “calibration” Relevance for other final states to be evaluated Relevance for other final states to be evaluated Otherwise purely experimental handle! Otherwise purely experimental handle! MET resolution MET resolution Can be measured along perpendicular and longitudinal axis Can be measured along perpendicular and longitudinal axis Resolution scale is scalar Et sum of hadronic calorimeter signal Resolution scale is scalar Et sum of hadronic calorimeter signal Biased by UE and pile-up (MC needed here) Biased by UE and pile-up (MC needed here) Qualitatively follows calorimeter energy resolution Qualitatively follows calorimeter energy resolution CERN-OPEN-2008-020 MET Scale & Resolution

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9292 Remarks Concerning MET Missing ET is a complex experimental quantity Missing ET is a complex experimental quantity Sensitive to precision and resolution of hard object reconstruction Sensitive to precision and resolution of hard object reconstruction MET is calibrated by everything MET is calibrated by everything Easily affected by detector problems and inefficiencies Easily affected by detector problems and inefficiencies Careful analysis of full event topology Careful analysis of full event topology Signal shapes in physics and detector Signal shapes in physics and detector Known unknown (1): effect of underlying event Known unknown (1): effect of underlying event Some correlation with hard scattering? Some correlation with hard scattering? Borderline with radiation? Borderline with radiation? Insignificant contribution?? Insignificant contribution?? To be confirmed early with di-jets To be confirmed early with di-jets Known unknown (2): effect of pile-up Known unknown (2): effect of pile-up Level of activity not so clear Level of activity not so clear Minimum bias first and urgent experimental task Minimum bias first and urgent experimental task Expectation is cancellation on average (at least) Expectation is cancellation on average (at least) Detector signal thresholds/acceptance potentially introduce asymmetries Detector signal thresholds/acceptance potentially introduce asymmetries Need to know the “real” detector Need to know the “real” detector Considerable contribution to MET fluctuations Considerable contribution to MET fluctuations Severe limitation in sensitivity for discovery Severe limitation in sensitivity for discovery

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9393 MET Reconstruction Software

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9494RequirementsRequirements Support event ambiguity resolution in different signal granularities Support event ambiguity resolution in different signal granularities ESD – track cell signal use to veto multiple MET contributions from same cell ESD – track cell signal use to veto multiple MET contributions from same cell AOD – track cluster signals for the same reason AOD – track cluster signals for the same reason Implemented by book-keeping in cell/cluster maps filled by MET tools Implemented by book-keeping in cell/cluster maps filled by MET tools Support re-calculation from ESD or AOD Support re-calculation from ESD or AOD Each MET contribution implemented as tool to adapt to different data models and signal features Each MET contribution implemented as tool to adapt to different data models and signal features METRefEle, METRefGamma, METRefTau, METRefJet, METRefMuon, METRefCluster METRefEle, METRefGamma, METRefTau, METRefJet, METRefMuon, METRefCluster Each tool can work with cells and/or clusters Each tool can work with cells and/or clusters Exploits explicit knowledge of object signal content while using as much common implementation as possible Exploits explicit knowledge of object signal content while using as much common implementation as possible

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9595 General approach of (refined) missing Et calculation General approach of (refined) missing Et calculation Keep track of “signal use” by each object to avoid double counting Keep track of “signal use” by each object to avoid double counting Priority of association defined by client configured algorithm tool sequence Priority of association defined by client configured algorithm tool sequence MET then calculated from uniquely assigned cells/clusters MET then calculated from uniquely assigned cells/clusters Contribution can be entered on different energy scales including final calibration, local calibration, H1-style cell weighting (ESD only) & basic (EM) scale Contribution can be entered on different energy scales including final calibration, local calibration, H1-style cell weighting (ESD only) & basic (EM) scale AOD challenge AOD challenge No common “smallest” calorimeter signal base No common “smallest” calorimeter signal base Topological cluster for jets, taus, “unused” calorimeter signal Topological cluster for jets, taus, “unused” calorimeter signal Sliding window cluster for electrons/photons (cells in AOD) Sliding window cluster for electrons/photons (cells in AOD) Cell list for isolated muons (in AOD) Cell list for isolated muons (in AOD) Need a strategy for SW/Topo overlap Need a strategy for SW/Topo overlap Decided to implement 3-d cell/topo cluster and SW cluster/topo cluster overlap Decided to implement 3-d cell/topo cluster and SW cluster/topo cluster overlap Measure overlap using topo-cluster envelope Measure overlap using topo-cluster envelope Software Issues

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9696 Event ambiguity resolution Event ambiguity resolution (1) check signal overlap by common calorimeter object use (1) check signal overlap by common calorimeter object use CaloCells in ESD CaloCells in ESD CaloCluster (topological) in AOD CaloCluster (topological) in AOD (2) apply geometrical distance measures (2) apply geometrical distance measures Cut on (angular) distance between objects Cut on (angular) distance between objects (3) apply additional similarity of signal measures (3) apply additional similarity of signal measures Cut on similarity of (uncalibrated) signals Cut on similarity of (uncalibrated) signals Focus here on (1) Focus here on (1) Calorimeter signals in the AOD Calorimeter signals in the AOD Topo-clusters signal of choice Topo-clusters signal of choice Reconstruct the whole event with low thresholds Reconstruct the whole event with low thresholds Provide shape information Provide shape information Used to reconstruct hadronic physics objects Used to reconstruct hadronic physics objects Straight forward common object use approach Straight forward common object use approach New strategy for electrons and muons New strategy for electrons and muons Use overlap sliding window/topological cluster (electrons) Use overlap sliding window/topological cluster (electrons) Use overlap between cells and topological cluster (isolated muons) Use overlap between cells and topological cluster (isolated muons) MET Software Implementations

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9797 Cluster Based Signal Decomposition Find topological clusters used by physics objects Find topological clusters used by physics objects By object navigation By object navigation Jets and taus Jets and taus More complex for electrons, photons, muons More complex for electrons, photons, muons Different cluster algorithm for electrons/photons: fixed size sliding window Different cluster algorithm for electrons/photons: fixed size sliding window No direct association between muons and (possible) topo cluster No direct association between muons and (possible) topo cluster Muon cluster signal may be below threshold Cells are collected around extrapolated track Cell signals for electrons/photons/muons in AOD Cell signals for electrons/photons/muons in AOD Cluster overlap resolution issues Cluster overlap resolution issues Topological clusters represent 3-d energy blobs Topological clusters represent 3-d energy blobs Motivated by reconstructing one cluster/shower Motivated by reconstructing one cluster/shower Quite efficient even in dense jet environment (~1.6 particles/cluster, much better for isolated particles) Quite efficient even in dense jet environment (~1.6 particles/cluster, much better for isolated particles) Simple angular distance based selection may be insufficient Simple angular distance based selection may be insufficient A topo cluster containing signals not generated by the electron/photon may be “behind” the SW cluster representing this particle A topo cluster containing signals not generated by the electron/photon may be “behind” the SW cluster representing this particle Better to use other variables/features to measure overlap Better to use other variables/features to measure overlap

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9898 “EM” TopoCluster Sliding Window Cluster Hadronic TopoCluster Cluster Overlap

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 9999 Variables For Cluster Overlap Resolution TopoClusters and Sliding Window clusters representing the same em shower are different TopoClusters and Sliding Window clusters representing the same em shower are different Different shapes Different shapes TopoCluster size determined by cell signal topology TopoCluster size determined by cell signal topology SW cluster size fixed by client SW cluster size fixed by client Both clusters will have different cell content! Both clusters will have different cell content! Different signal fluctuations Different signal fluctuations No noise suppression in SW clusters No noise suppression in SW clusters But less direct contributions from small signal “marginal” cells Noise suppression in TopoClusters Noise suppression in TopoClusters Potentially more small signal cells with relatively large fluctuations Possible variables to measure overlap (under study) Possible variables to measure overlap (under study) Total raw signal Total raw signal Affected by different noise characteristics Affected by different noise characteristics Relative signal distribution in sampling layers Relative signal distribution in sampling layers Could be better as some noise is unfolded in the ratios Could be better as some noise is unfolded in the ratios Geometrical distance (barycenter-to-barycenter in 3-d) Geometrical distance (barycenter-to-barycenter in 3-d) Not available for SW (could easily be implemented!) Not available for SW (could easily be implemented!) Subject to detector granularity changes (?) Subject to detector granularity changes (?) Measure common cell content Measure common cell content Similar motivation as for ESD Similar motivation as for ESD Adds several other (more stable) measures to overlap resolution, see next slides Adds several other (more stable) measures to overlap resolution, see next slides

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 100100 TopoClusters have 3-d geometry TopoClusters have 3-d geometry Barycenter (x,y,z) Barycenter (x,y,z) In AOD, from (R,η,φ) In AOD, from (R,η,φ) Extension along and perpendicular to “direction of flight” Extension along and perpendicular to “direction of flight” Measured by 2 nd geometrical moments Measured by 2 nd geometrical moments Vertex assumption (0,0,0) for Vertex assumption (0,0,0) for right now right now Principal axis available for large enough clusters Can calculate envelop around barycenter Can calculate envelop around barycenter Presently ellipsoidal Presently ellipsoidal Could include apparent “longitudinal asymmetry” of em showers Use simple model of Use simple model of longitudinal profile longitudinal profile Geometrical Features Of Clusters

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 101101 AOD TopoClusters have no cells AOD TopoClusters have no cells Need to come up with some geometrical measure Need to come up with some geometrical measure Use the envelop! Use the envelop! Can calculate likelihood that cell is within envelop Can calculate likelihood that cell is within envelop Introduces two parameters with typical values: Introduces two parameters with typical values: May need some tuning! May need some tuning! Define cell i is inside topo cluster c when… Define cell i is inside topo cluster c when… Associating Cells With TopoClusters

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 102102 Cell Content Variables Shared cells between SW and TopoCluster provide: Shared cells between SW and TopoCluster provide: Fractional number of cells shared Fractional number of cells shared Can be calculated for both clusters Can be calculated for both clusters Likely more useful for TopoCluster Energy density measure Energy density measure Fraction of cluster signal in shared cells Fraction of cluster signal in shared cells Raw signal reference Relative profiles Relative profiles Re-summation of raw sampling energies allows to calculate fractions by sampling for both types of clusters Re-summation of raw sampling energies allows to calculate fractions by sampling for both types of clusters … When used with muons: When used with muons: Fraction of cells associated with a muon inside a TopoCluster Fraction of cells associated with a muon inside a TopoCluster Discard TopoCluster signal in MET calculation if muon corrected for calo energy loss Discard TopoCluster signal in MET calculation if muon corrected for calo energy loss

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 103103 MET from Electron/Photon Cells ESD only! ESD only! Cells in electron can contribute on different scales: Cells in electron can contribute on different scales: electromagnetic scale refined scale (out of cluster corrections undone!)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 104104 MET from Electron/Photon Cells ESD only! ESD only! Cells in electron can contribute on different scales: Cells in electron can contribute on different scales: electromagnetic scale refined scale (out of cluster corrections undone!) H1 cell weights

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 105105 MET from Electron/Photon ESD and AOD ESD and AOD Electron/photon can contribute on different scales: Electron/photon can contribute on different scales: electromagnetic scale refined scale (out of cluster corrections undone!) utilizes functions in CaloUtils/CaloClusterOverlapHelpers (useful also outside of MET calculations)

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 106106 MET from Cells in Jets ESD only! ESD only! Cells in jets can contribute on different scales: Cells in jets can contribute on different scales: electromagnetic scale/local hadronic (cluster scales) refined scale (jet energy scale) Taus and topo-clusters very similar!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 107107 MET from Cells in Jets ESD only! ESD only! Cells in jets can contribute on different scales: Cells in jets can contribute on different scales: electromagnetic scale/local hadronic (cluster scales) refined scale (jet energy scale) H1 cell calibration Taus and topo-clusters very similar!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 108108 MET from Clusters in Jets ESD and AOD ESD and AOD Clusters in jets can contribute on different scales: Clusters in jets can contribute on different scales: electromagnetic scale/local hadronic (cluster scales) refined scale (jet energy scale) Taus and topo-clusters very similar!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 109109 Other Remarks MET depends on “nice” behaviour of all other objects MET depends on “nice” behaviour of all other objects We fixed some things by introducing special treatments, especially concerning navigation We fixed some things by introducing special treatments, especially concerning navigation Complication of MET code Complication of MET code This costs a lot of time! This costs a lot of time! We should have used Savannah more!! We should have used Savannah more!! It seems that most objects now behave as expected It seems that most objects now behave as expected After tauObject navigation fix After tauObject navigation fix Still unclear about isolated muons Still unclear about isolated muons Performance checks can now be done with METPerformance package! Performance checks can now be done with METPerformance package! Stable platform for performance evaluations Stable platform for performance evaluations Need to generalize cell/cluster maps Need to generalize cell/cluster maps Contain unambiguous event Contain unambiguous event Useful for MET significance, biasing jet reconstruction etc. Useful for MET significance, biasing jet reconstruction etc. Can hide clusters already associated with electron from jet finding Can hide clusters already associated with electron from jet finding Need to review code location Need to review code location Some of this should be a RecTool/RecStore rather than MET! Some of this should be a RecTool/RecStore rather than MET!

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 110110 ConclusionConclusion

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 111111ConclusionConclusion Jet/MET effort increased significantly in the last few years Jet/MET effort increased significantly in the last few years Very good! Very good! Jet calibration task force Jet calibration task force MET and data quality task force MET and data quality task force Most basic features scenarios laid out for first data Most basic features scenarios laid out for first data AntiKt R = 0.4 default AntiKt R = 0.4 default Excellent – no more ATLAS cone! Excellent – no more ATLAS cone! Several calibration scenarios under study Several calibration scenarios under study This meeting… This meeting… Some things need more input Some things need more input Towards one calibration scheme Towards one calibration scheme Roadmap for comparisons Roadmap for comparisons Discussion on Saturday Discussion on Saturday Systematic uncertainties Systematic uncertainties How to evaluate with/for initial data? How to evaluate with/for initial data? MET calculation MET calculation

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 Introduction To Jet & Missing Transverse Energy Reconstruction in ATLAS Peter Loch.

Similar presentations

Presentation on theme: "P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 Introduction To Jet & Missing Transverse Energy Reconstruction in ATLAS Peter Loch."— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 Introduction To Jet & Missing Transverse Energy Reconstruction in ATLAS Peter Loch.

Similar presentations

Presentation on theme: "P. Loch U of Arizona June 23, 2009 P. Loch U of Arizona June 23, 2009 Introduction To Jet & Missing Transverse Energy Reconstruction in ATLAS Peter Loch."— Presentation transcript:

Similar presentations

About project

Feedback