Electron Identification Efficiency from Zee Maria Fiascaris University

Electron Identification Efficiency from Z→ee Maria Fiascaris University of Oxford In collaboration with Tony Weidberg and Lucia di Ciaccio ATLAS UK SM Meeting, Sept 24 th 2007

Overview n Electron Identification Efficiency (tight/medium/loose electron) from Data n n n Evaluate efficiencies independently from MC Use Tag and Probe method Comparison with MC to evaluate systematics Differential Efficiencies in Eta and Pt Global Efficiencies Results from Signal Only (Z→ee) Sept 24 th 2007 ATLAS UK SM Meeting 2

Samples and. Is. EM Flag Definition n Sample: trig 1_misal 1_csc 11_V 1. 005144. Phytihia. Zee. recon. AOD. c 12000601 Dataset Process Cross section (pb) Efficiency Eff Cross Section (pb) Number Events Luminosity (pb -1) 5144 Z→ee 1675 0. 855 1432 415 k 290 n Using custom NTuples produced by Ellie Dobson (Oxford) and Mike Flowerdew (Liverpool) n Is. EM Flag Definition for Athena version 12 (will change in version 13) n Tight Electron : is. EM & 0 x 7 FF (all but TRT) n Medium Electron: is. EM & 0 x 30 F (all calo + track quality + cluster-track matching in η and φ) n Loose Electron: is. EM & 0 x 7 (3 out of 4 calo bits) Sept 24 th 2007 ATLAS UK SM Meeting 3

Method: Tag and Probe n Measure Is. EM Efficiencies relative to Electron Offline Reconstruction Efficiency n Tag Selection: • • n n Pt > 25 Ge. V |η|< 2. 4, exclude cracks (1. 37<|η|< 1. 52) Tight is. EM Passes all triggers (e 25 i): L 1+L 2+EF (need to match offline electron to trigger object and apply trigger cuts) Second Electron Select cluster (from electron container) in opposite hemisphere: Δφtag. Ele-clus > ¾ pi Offline cut: Pt > 25 Ge. V |η|< 2. 4, include cracks Probe Condition: • n First Electron Offline cut: Require 2 nd Electron to have Tight/Medium/Loose Is. EM Select ee pairs with invariant mass: 70 Ge. V < Mee < 100 Ge. V Use only electrons from EGamma algorithm Sept 24 th 2007 ATLAS UK SM Meeting 4

Is. EM Efficiency from Data n Count number of electron pairs that satisfy tag/probe condition (can be more than 1 per event → combinatorial background) ε = N 2/N 1 N 1= all cases passing tag condition N 2= all cases where 2 nd electron passes probe condition (tight/medium/loose Is. EM) n Statistical Error calculated as: Where N 2 b is the number of cases when both electrons satisfy tag and probe conditions N 1 N 2 Loose Is. EM Combinatorial Background Sept 24 th 2007 ATLAS UK SM Meeting 5

Efficiency from Monte Carlo n Use Efficiency from MC to study systematics n Select two electrons coming from Z: n Require one electron with: Pt > 25 Ge. V |η|<2. 4, cracks excluded (1. 37<|η|< 1. 52) n Second Electron Pt > 25 Ge. V |η|<2. 4, cracks included Match second electron with offline electron (ΔR <0. 2) → N 1 n n For matched offline electron: require is. EM to be tight/medium/loose → N 2 Sept 24 th 2007 ATLAS UK SM Meeting 6

Differential Efficiency in. Eta • Data • MC →Drops in endcaps for medium/tight due to Track Quality bits or to bit from Calo? (see later) Sept 24 th 2007 ATLAS UK SM Meeting 7

Understanding. Eta Differential Efficiency: Track Quality Bit in is. EM flag requires: § No η dependence below threshold • Num of BLayer Hits ≥ 1 values: does not explain efficiency drop at the edges of η • Num of SCT+Pixel Hits ≥ 9 • Impact Parameter < 0. 1 Num Pixel+SCT Hits Sept 24 th 2007 § Need to look at calorimeter bit that is excluded from Loose Electron is. EM Num BLayer Hits ATLAS UK SM Meeting Impact Parameter 8

Efficiency of Bit 3 of Calo: (is. EM & 0 x 8) n Efficiency of Bit 3 of Calo: (is. EM & 0 x 8) First sampling of the LAr calorimeter, uses fine granularity to apply cuts based on shower shape Included in tight/medium electron, but excluded from loose electron In Eta: explains the shape for differential efficiency of tight/medium electron (drops in endcaps and central eta) Sept 24 th 2007 ATLAS UK SM Meeting 9

Understanding Pt Differential Efficiecy § Can correct for the energy scale between true and offline electron: plot MC results in bins of Pt Offline Electron § Corrects for discrepancy at high Pt § Discrepancy at low Pt remains: Combiantiorial Background? Before En. Scale Correction Sept 24 th 2007 After En. Scale Correction ATLAS UK SM Meeting 10

Differential Efficiency in Pt • Data • MC Efficiency systematically underestimated at low Pt (after correction for Energy Scale) Sept 24 th 2007 ATLAS UK SM Meeting 11

Global Efficiecy (for Pt>25 Ge. V) Is. EM With cracks No cracks Sept 24 th 2007 Efficiency (Data) Tight (no TRT) 78. 87± 0. 13 Efficiency (MC) 78. 94± 0. 09 Medium 82. 30± 0. 12 82. 53± 0. 08 Loose 98. 27± 0. 04 98. 73± 0. 03 Tight (no TRT) 79. 73± 0. 13 79. 92± 0. 09 Medium 82. 61± 0. 12 82. 86± 0. 08 Loose 98. 83± 0. 03 99. 32± 0. 02 ATLAS UK SM Meeting 12

Tight Electron: with/without TRT No TRT Global Efficiency drops by more than 10 % with TRT (including crakcs) Sept 24 th 2007 With TRT Is. EM Eff (data) Eff (MC) Tight (no TRT) 78. 87 ± 0. 13 78. 94 ± 0. 09 Tight (with TRT) 67. 66 ± 0. 17 68. 01 ± 0. 10 ATLAS UK SM Meeting 13

Conclusions n n Differential Efficiency in Eta, Pt and Global Efficiency calculated for Z→ee (signal only) Differential Efficiency in Eta: n n n Differential Efficiency in Pt: n n n Agreement data-MC Cuts based on 1 st Sampling of LAr Calo affect medium/tight electron efficiency Need to correct for energy scale Still discrepancy between data and MC at low Pt (combinatorial background) Tight Electron with/without TRT bit: n TRT bit decreases the efficiency by more than 10% Further Studies n Study QCD background (Dijet samples, Pythia): methodology is ready (Mike Flowerdew and Ellie Dobson) n Other studies on offline electron and trigger efficiencies from Z→ee: Guillaume Kirsch (offline efficiency), Ellie Dobson and Mike Flowerdew (trigger efficiency) Sept 24 th 2007 ATLAS UK SM Meeting 14