Hadronic energy reconstruction in the combined electromagnetic and hadronic calorimeter system of the CALICE Collaboration Miroslav Gabriel MPP/TUM 29th IMPRS Workshop July 7th 2014 1 / 22
ILC and Calorimetry Basic Reconstruction Local Software Compensation Overview 1 2 3 4 ILC and Calorimetry Basic Reconstruction Local Software Compensation Summary 2 / 22 Summary Backup slides
Future Linear Colliders e + e -Colliders to complement LHC: Synchrotron radiation increases with E Only linear acc. can very reach high energies for e + e International Linear Collider (TDR published in 2013): Staged Implementation 250 GeV, 350 GeV, 500 GeV (upgrade to 1 TeV possible) L = 2 34 1 cm 2 s Benefits for Physics: Exploration of Electroweak sector Model Independent Measurements Low Background 3 / 22
Physics Guaranteed Program Higgs measurements: Couplings to fermions and bosons (including c, g) Self-coupling Total width Top Yukawa coupling Top physics: Measurements at threshold Precise mass, width Precision physics: Electroweak QCD One important measurement: e + e - Z Model-independent Z H Identify Higgs production from Z recoil irrespective of Higgs decay 4 / 22
Physics Possible Discoveries: Direct Production of new particles up to s/2 Indirect(Modeldependent) search for New Physics Cross-section [fb] 3 tt 2 1-1 -2 H+X 0 00 2000 3000 s [GeV] 5 / 22
ILC and Calorimetry Basic Reconstruction Local Software Compensation Summary Backup slides Design of Detectors Vertex and TPC for precise tracking Electromagnetic calorimeter for γ and e e + energy e+ Hadronic calorimeter for charged and neutral hadrons HCal energy resolution is Bottleneck for standard jet reconstruction e- Solenoid and return yoke AHCal ECal TPC Higher precision: Combine tracking and calorimetry information Particle Flow Assignment Edeposit to right particle crucial Requires high spatial resolution in calorimeters 6 / 22 Particle Flow concept dictates calorimeter design
Calorimetry Incoming particle: Destructive measurement: Total absorption Different interactions(bremsstrahlung, pair production, inelastic hadronic interactions, spallation etc.) Particle multiplication yield cascades Electromagnetic and hadronic showers Governed by X 0 O(1 cm) and λ int O(15 cm) High granularity for particle separation Design of a Calorimeter? λint=o(15 cm) π 0 secondary particles evaporation neutron π - p e - e - e - X0=O(1 cm) π - n n X 0 π - e - em shower π 0 e + n π + e - hadronic shower tertiary particles π - λint π 0 (8.5x -17 s) e + π 0 7 / 22
CALICE Prototype Cherenkov Detector Drift Chambers ECAL HCAL TCMT Beam Sc1 Sc4 Sc3 Sc2 Scintillators Muon Trigger Highly Granular Sampling Calorimeters: Most energy deposited in absorber plates Conversion factor needed E seen C = E total Three different sub-detectors Inter-calibration 20000 Channels Two ways for energy reconstruction 8 / 22
Basic Reconstruction Calorimetry is counting of shower particles: E N E toal = ECalhits E hit ω ω accounts for sampling fraction! Introduce one weight per sub-detector ω ECal, ω HCal and ω TCMT account for inter calibration Determination of calibration factors? Use of χ 2 minimization procedure χ 2 = ( E hit ω ECal + E hit ω AHCal + ) 2 E hit ω TCMT E beam events ECalhits AHCalhits TCMThits 9 / 22
Reconstructed Energy Calib. factors calculated with data and MC: Energy independent calibration factors Data reconstructed with factors from MC Constant offset MC does not perfectly reproduce visible energy reconstructed Energy [GeV] 80 70 60 50 40 30 20 Data on data FTFP_BERT on data GeV ECal 0.0049 MIP GeV AHCal 0.029 MIP GeV TCMT 0.031 MIP Table: Factors from data beam -E beam )/E rec (E 0.05 0-0.05 20 30 40 50 60 70 80 beam Energy [GeV] / 22
Resolution Resolution Stochastic Term σ E E = a E Constant Term b c E Noise Termn E/E 0.2 0.15 Data on data FTFP_BERT on data Weights from FTFP_BERT and Data on same level Atlas: 45% E CMS: 0% E 0.1 0.05 0 Fit: a / E b c / E Data on data: FTFP_BERT on data: a=55.6% b=4.85% c=0.18 GeV a=52.8% b=5.66% c=0.18 GeV 20 30 40 50 60 70 80 beam Energy [GeV] Enhance the resolution? Local Software Compensation 11 / 22
The Idea of Software Compensation Electromagnetic showers: e - em shower Solely composed of γ, e E visivle E deposit Large number of particles in each events Governed by X 0 O(1 cm) Large Energy Density Hadronic showers: π 0 γγ: em und hadronic component Hadronic governed by λ int O(15 cm) Small Energy Density Unseen deposits from neutrons, binding energy etc. need larger weight Classify hit by it s energy density! λint=o(15 cm) π 0 secondary particles evaporation neutron π - p X0=O(1 cm) e - π - n n e - X 0 e - π 0 e + n π + e - hadronic shower π - tertiary particles π - λint π 0 (8.5x -17 s) e + π 0 12 / 22
The Implementation of Software Compensation entries/(0.05 MIP/cell) 5 4 FTFP_BERT ECal ECal Hit (7 MIP/Cell) entries/(0.05 MIP/cell) 6 5 4 FTFP_BERT AHCal AHCal Hit (8 MIP/Cell) AHCal Hit (12 MIP/Cell) entries/(0.05 MIP/cell) 5 4 3 TCMT Hit (11 MIP/Cell) FTFP_BERT TCMT 3 AHCal Hit (27 MIP/Cell) 2 3 2 2 1 5 15 20 25 30 35 40 45 50 Energy Density[MIP/cell] 0 5 15 20 25 30 35 40 45 50 Density Bin Energy Density[MIP/cell] 5 15 20 25 30 35 40 45 50 Energy Density[MIP/cell] Assign each Density Bin it s own calibration factor now 6 ω i in ECal, ω j AHCal and 6 ω k in TCMT 22 factors i,j,k: Density Bin indexes 13 / 22
The Implementation of Software Compensation Determination of ω i,j,k via Minimization: Small Density Bin Index Small energy density More likely hadronic Higher weight ω(j,e ) [GeV/MIP] 0.08 0.06 0.04 Factors change with E beam ω i,j,k (E beam ) Iterative Parameterization 0.02 0 2 4 6 8 density bin index j Figure: 14 / 22
Reconstructed Energy Calib. factors again calculated from data and FTFP_BERT: Reconstructed energy for MC factors again too low Offset decreased to 1-2% reconstructed Energy [GeV] beam -E beam )/E 80 70 60 50 40 30 20 0.05 0 Syst. uncertainties const. factors Data on data SC FTFP_BERT on data SC 20 30 40 50 60 70 80 rec (E -0.05 20 30 40 50 60 70 80 beam Energy [GeV] 15 / 22
Resolution Data and MC show good agreement Gain in resolution E/E 0.2 0.15 0.1 0.05 0 Fit: a / E b c / E Data on data: a=55.6% b=4.85% c=0.18 GeV FTFP_BERT on data: a=52.8% b=5.66% c=0.18 GeV Data on data SC: a=44.7% b=2.54% c=0.18 GeV FTFP_BERT on data SC: a=42.5% b=3.24% c=0.18 GeV Data on data FTFP_BERT on data Data on data SC FTFP_BERT on data SC 20 30 40 50 60 70 80 beam Energy [GeV] Software Compensation clearly increases resolution σ(e) E Reco 55.6% 44.7%, 4.85% 2.54% 16 / 22
Summary Calorimeter systems play important role at future linear collider detectors Calibration of sub-detectors possible with three constant factors Discrimination by energy density Successful application of Software Compensation enhances Resolution Large progress in the simulation of hadronic showers Results from Simulations and Data on comparable level Still missing: Updated Monte Carlos Possible next step: Further improve at low energies Expand analysis to low energy Fermilab data 17 / 22
Backup 18 / 22
Run list from CAN-35 Table: List of used data runs. run particle beam energy, number type GeV 330332 π 330643 π 330777 π 330850 π 330328 π 15 330327 π 18 330649 π 20 330771 π 20 330325 π 25 330650 π 25 331298 π + 30 331340 π + 30 330551 π 35 330960 π 35 330390 π 40 330412 π 40 330560 π 40 331338 π + 40 331339 π + 40 19 / 22 run particle beam energy, number type GeV 330550 π 45 330559 π 45 330961 π 45 330391 π 50 330558 π 50 331335 π + 50 331282 π + 60 331333 π + 60 331334 π + 60 331556 π 60 331568 π 60 331655 π 60 331664 π 60 330392 π 80 330962 π 80 331280 π + 80 331324 π + 80 331554 π 80 331567 π 80 331654 π 80
Energy Dependence of Weights ω(j,e ) [GeV/MIP] 0.02 0.015 0.01 0.005 FTFP_BERT weights for ECal: 80GeV 60GeV 50GeV 40GeV 20GeV GeV ω(j,e ) [GeV/MIP] 0.08 0.06 0.04 0.02 FTFP_BERT weights for AHCal: 80GeV 60GeV 50GeV 40GeV 20GeV GeV 0 1 2 3 4 5 6 density bin index j ω(j, E) = p 1 (E)Exp (p2 (E) j) + p 3 (E) ω(j,e ) [GeV/MIP] 0 0.08 0.06 0.04 0.02 2 4 6 8 density bin index j FTFP_BERT weights for TCMT: 80GeV 60GeV 50GeV 40GeV 20GeV GeV 0 1 2 3 4 5 6 density bin index j 20 / 22
1 2 3 Parameterization Iterative Procedure: 1. Minimize 2. Fix parameter 3 3. Minimize 4. Fix parameter 2 5. Minimize ForExample : p 1 (E) = c4 E c 1 + E c 2 + c 3 Exp p 2 (E) = b 1 E + b 2 p 3 (E) = a 1 (1 Exp a2 E )+a 3 parameter p 0.4 0.3 ECal AHCal TCMT parameter p 1 0 parameter p 0.03 0.02 After fixing: ECal AHCal TCMT 0.2-1 After fixing: 0.01 0.1-2 ECal AHCal TCMT 20 30 40 50 60 70 80-3 20 30 40 50 60 70 80 0 20 30 40 50 60 70 80 energy [GeV] energy [GeV] energy [GeV] 21 / 22
Constant calib. factors from FTFP_BERT GeV ECal 0.004675 MIP GeV AHCal 0.02796 MIP GeV TCMT 0.02216 MIP Table: Factors from FTFP_BERT GeV ECal 0.0049 MIP GeV AHCal 0.029 MIP GeV TCMT 0.0309 MIP Table: Factors from data 22 / 22