Hadron energy resolution of the CALICE AHCAL and software compensation approaches Marina Chadeeva, ITEP, Moscow for the CALICE Collaboration M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 1 / 14
Outline 1 CALICE test beam setup and event selection 2 Intrinsic AHCAL energy resolution for single hadrons 3 Software compensation approach and its application Local technique Global technique 4 Comparison of software compensation techniques Linearity and relative resolution Improvement of resolution 5 Summary M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 2 / 14
CALICE test beam setup and event selection CALICE test beam setup Test beam at CERN SPS π and π 1-8 GeV ECAL: Si-W,.8λ I (3 layers), 18x18cm 2, 1x1cm 2 cells HCAL: Sc-Fe, 4.5λ I (38 layers), 1x1m 2, 768 tiles: 3x3, 6x6, 12x12cm 2 M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 3 / 14
CALICE test beam setup and event selection CALICE test beam setup Test beam at CERN SPS π and π 1-8 GeV ECAL: Si-W,.8λ I (3 layers), 18x18cm 2, 1x1cm 2 cells HCAL: Sc-Fe, 4.5λ I (38 layers), 1x1m 2, 768 tiles: 3x3, 6x6, 12x12cm 2 TCMT: Sc-Fe, 5λ I (16 layers), 9x9cm 2, strips Helpers: drift chambers, Čerenkov and veto M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 3 / 14
CALICE test beam setup and event selection Advantages of high granularity Hadronic shower structure Identification of a layer of the first inelastic interaction longitudinal shower profiles w/o convolution with the distribution of the first interaction point Investigation of a shower substructure on event-by-event basis track segments inside hadronic shower electromagnetic clusters energy density spectra Particle Flow Analysis Possibility to disentangle showers induced by charged and neutral particles Software compensation Improvement of the energy resolution by means of software compensation techniques based on the analysis of the detailed energy density spectra M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 4 / 14
CALICE test beam setup and event selection Event selection and data samples Sample cleaning Muons: initial admixture 4-3% in the cleaned sample <.5% Multiparticle: 1-2% Electrons from π : Čerenkov counter Protons from π : Čerenkov counter M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 5 / 14
CALICE test beam setup and event selection Event selection and data samples Sample cleaning Muons: initial admixture 4-3% in the cleaned sample <.5% Multiparticle: 1-2% Electrons from π : Čerenkov counter Protons from π : Čerenkov counter Training and test subsamples Samples from different runs of the same beam energy and particle charge are merged Merged samples are split into two subsamples (even and odd event numbers) Statistically independent samples are used to test software compensation approaches set of even subsamples is used to adjust software compensation factors adjusted compensation factors are applied to the set of odd subsamples M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 5 / 14
CALICE test beam setup and event selection Event selection for software compensation study Hadronic shower start in the first 5 layers of the HCAL Additional constraint to select showers mostly contained in the HCAL track in ECAL detailed energy density spectrum in the fine-granular HCAL minimum longitudinal leakage into TCMT (around 6% at 8 GeV) 1 4 of pion events remains M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 6 / 14
Intrinsic AHCAL energy resolution for single hadrons Intrinsic AHCAL resolution for single hadrons [GeV] E reco beam - E beam )/E reco (E 9 8 7 6 5 4 3 2 1.2 -.2 Single runs: π - Single runs: π Merged Even: π - Merged Even: π Merged Odd: π - Merged Odd: π CALICE Preliminary 1 2 3 4 5 6 7 8 9 E beam [GeV] Systematic uncertainties δe reco =.9% ( δe beam for residuals) σ reco /E reco.22.2.18.16.14.12.1.8.6 Fit: a/ E b c/e Single runs: a = 57.3 ± Merged Even: a = 57.6 ± Merged Odd: a = 57.9 ± CALICE Preliminary.3% b = 1.5 ±.4% b = 1.6 ±.4% b = 1.4 ±.2% c =.18.3% c =.18.4% c =.18.4 1 2 3 4 5 6 7 8 9 E beam [GeV] Fit results coincide within errors Stochastic term: 57.5% E/GeV Constant term: 1.5% Noise fixed at.18 GeV for full setup π - π Similar resolution for π and π M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 7 / 14
Software compensation approach and its application Software compensation Non-compensating calorimeters Hadronic shower comprises electromagnetic and hadronic component A part of the hadronic component (neutrons, etc.) remains undetectable Electromagnetic fraction exhibits significant event-by-event fluctuations Different response to electrons and hadrons of the same energy Hadron energy resolution is deteriorated w.r.t. electromagnetic one Approaches to improve resolution Hardware compensation (material, sampling fraction, etc.) Software or off-line compensation based on the expectation of higher energy density inside electromagnetic component comparing to a hadronic one successfully implemented for WA1 iron-sc sampling calorimeter (H. Abramowicz et al. NIM 18 (1981) 429-439) H1 LAr calorimeter (C. Issever, K. Borras, D. Wegener, NIM A545 (25) 83-812) M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 8 / 14
Software compensation approach and its application Software compensation for the CALICE AHCAL CALICE analogue hadronic calorimeter Non-compensating with average e π 1.2 (for 1-8 GeV) High-granular gives a detailed energy density spectrum Software compensation Two techniques were developed with different approaches: Local different weights are applied to the signals in individual cells depending on their energy density Both techniques allow event-by-event energy correction Global one compensation factor calculated from the energy density spectrum is applied to the energy sum do not require a prior knowledge of particle energy M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 9 / 14
Software compensation approach and its application Local technique Local compensation technique (LC) Energy density (ED) distribution is divided into energy density bins E SC = E ecal,sum hit E hit ω hit E tcmt,sum The weights depend on ED and initial reconstructed event energy E (p, p 1, p 2 are energy dependent): ω hit = p (E) exp(p 1 (E) ED) p 2 (E) Shape of parameters p, p 1, p 2 is found via an iterative procedure using beam energy. weight value 2 Weight parametrization: 1.8 1.6 1.4 1.2 1.8.6 * exp( p 1 * x) p 2.4 2 4 6 8 1 12 energy density p Colors for beam energies 1GeV, 3GeV, 5GeV, 7GeV p 1 p 2 p 2.6 2.4 2.2 2 1.8 1.6 1.4 1.2.1.2.3.4.5.6.7.9.8.7.6.5.4 1 2 3 4 5 6 7 8 9 1 beam energy [GeV] 1 2 3 4 5 6 7 8 9 1 beam energy [GeV] 1 2 3 4 5 6 7 8 9 1 beam energy [GeV] M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 1 / 14
Software compensation approach and its application Global technique Global compensation technique (GC) Global compensation factor C gl calculated on event-by-event basis: number of shower hits N av with e hit < e, e is a mean of shower hit energy spectrum number of shower hits N lim with e hit < e lim, e lim = 5 MIP C gl = N lim N av Mean of global compensation factor C gl is energy dependent; coefficients a, a 1, a 2 to describe this dependence are derived using beam energy Reconstructed energy: E GC = E ecal E sh (a a 1 E sh a 2 E 2 sh ), where E sh = C gl (E hcal E tcmt ) M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 11 / 14
Comparison of software compensation techniques Linearity and relative resolution Comparison of local and global compensation [GeV] E reco beam - E beam )/E reco (E 9 8 7 6 5 4 3 2 1.2 -.2 Initial: π - Initial: π GC: π - GC: π LC: π - LC: π CALICE Preliminary 1 2 3 4 5 6 7 8 9 E beam [GeV] Systematics for initial sample Different linearity for π and π σ reco /E reco.22.2.18.16.14.12.1.8 Fit: a/ E b c/e Initial: a = 57.6 ± GC: a = 45.8 ± LC: a = 44.9 ±.4% b = 1.6 ±.3% b = 1.6 ±.3% b = 1.6 ±.3% c =.18.2% c =.18.2% c =.18.6 CALICE Preliminary.4 1 2 3 4 5 6 7 8 9 E beam [GeV] Stochastic term decreased w.r.t. initial: 45% (little better for LC) E/GeV Constant term: not changed π - π Noise fixed at.18 GeV for full setup Similar improvement of relative resolution for π and π M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 12 / 14
Comparison of software compensation techniques Improvement of resolution Improvement of resolution Energy distributions before and after compensation ( χ2 NDF < 2 for Gaussian fits) Events /.5 GeV 6 CALICE Preliminary 5 (a) 4 - π 1 GeV Initial LC GC Events / 1. GeV 5 CALICE Preliminary 45 (b) 4 35 3 π 4 GeV Initial LC GC Events / 1. GeV 7 CALICE Preliminary 6 (c) 5 4 - π 8 GeV Initial LC GC 3 25 2 2 15 3 2 1 1 5 1 2 4 6 8 1 12 14 16 18 Reconstructed energy, GeV 25 3 35 4 45 5 55 6 Reconstructed energy, GeV 6 65 7 75 8 85 9 95 1 Reconstructed energy, GeV Relative improvement of absolute resolution 12% < σ SC /σ initial < 25% Similar improvement for π and π Outlier at 35 GeV Local approach gives 3% better improvement in the energy range 25-6 GeV than the global one σ SC /σ initial 1.95.9.85.8.75.7.65.6 CALICE Preliminary GC: π - GC: π LC: π - LC: π 1 2 3 4 5 6 7 8 9 E beam [GeV] M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 13 / 14
Summary Summary Hadron energy resolution of the CALICE AHCAL π and π samples analyzed for beam energies from 1 to 8 GeV intrinsic resolution: 57.5% 1.6%.18 E/GeV E/GeV similar resolution for π and π, linearity within ±2% Software compensation for the CALICE AHCAL two software compensation techniques developed based on different approaches, both show similar improvement of single particle energy resolution stochastic term improved down to 45% E/GeV Local compensation - individual cell signal weighting - 3% better relative improvement for 25-6 GeV than in global relative improvement varies from 12% to 25% similar improvement for π and π Global compensation - one weight for event energy - twice as less parameters as in local approach M. Chadeeva (ITEP) LCWS11, Granada, Spain September 28, 211 14 / 14