Overview of the High-Level Trigger Electron and Photon Selection for the ATLAS Experiment at the LHC Ricardo Gonçalo, Royal Holloway University of London On behalf of the ATLAS High-Level Trigger group NSS 2005 – Puerto Rico, October 2005

Ricardo Goncalo, Royal Holloway University of London2 ATLAS HLT e/gamma selection Outline ATLAS and the Large Hadron Collider The ATLAS High-Level Trigger Performance studies Summary and outlook

ATLAS and the LHC

Ricardo Goncalo, Royal Holloway University of London4 ATLAS HLT e/gamma selection The LHC The LHC will start operation in 2007 and will represent the high-energy frontier in collider physics Much is expected of the LHC and its experiments:  Study the origin of the electroweak symmetry breaking  Test models of physics beyond the Standard Model  Perform precision Standard Model measurements  … and still be able to detect unexpected new physics CM energy14 TeV Bunch crossing period25 ns Interactions/bunch crossing~5-25 Initial luminosity2x10 33 cm -2 s -1 Design luminosity10 34 cm -2 s -1

Ricardo Goncalo, Royal Holloway University of London5 ATLAS HLT e/gamma selection ATLAS Large angular coverage (|  |<5; tracking coverage up to  ~2.5) Liquid Argon electromagnetic calorimeter with accordion geometry Iron-scintillating tile hadronic calorimeter; tiles placed radially and staggered in depth Toroidal magnetic field in muon spectrometer (supercondutor air-core toroids)

The ATLAS High-Level Trigger

Ricardo Goncalo, Royal Holloway University of London7 ATLAS HLT e/gamma selection Challenges faced by the ATLAS trigger “Interesting” cross sections at least ~10 6 times smaller than total cross section 25ns bunch crossing interval (40 MHz) Up to ~25 proton-proton interactions per bunch crossing (depending on luminosity) Offline processing capability: ~200 Hz ~5 events selected per million bunch crossings

Ricardo Goncalo, Royal Holloway University of London8 ATLAS HLT e/gamma selection The ATLAS trigger Three trigger levels: Level 1:  Hardware based (FPGA/ASIC)  Coarse granularity detector data  Calorimeter and muon spectrometer only  Latency 2.2  s (buffer length)  Output rate ~75 kHz Level 2:  Software based  Only detector sub-regions processed (Regions of Interest) seeded by level 1  Full detector granularity in RoIs  Fast tracking and calorimetry  Average execution time ~10 ms  Output rate ~1 kHz Event Filter (EF):  Seeded by level 2  Full detector granularity  Potential full event access  Offline algorithms  Average execution time ~1 s  Output rate ~200 Hz High-Level Trigger

Ricardo Goncalo, Royal Holloway University of London9 ATLAS HLT e/gamma selection match? Selection method EMROI L2 calorim. pass? L2 tracking cluster? E.F.calorim. track? E.F.tracking cluster? E.F. pass? Level 2 seeded by Level 1 Fast reconstruction algorithms Reconstruction within RoI Level1 Region of Interest is found and position in EM calorimeter is passed to Level 2 Ev.Filter seeded by Level 2 Offline reconstruction algorithms Refined alignment and calibration Event rejection possible at each step Electromagnetic clusters

Performance studies

Ricardo Goncalo, Royal Holloway University of London11 ATLAS HLT e/gamma selection Signature example: single e  e25iEfficiency (*) Rate Level %5.6 kHz Level 2 Calorim.95.2 %1.2 kHz Level 2 Tracking88.7 %450 Hz Level 2 match88.2 %240 Hz Ev.Filter Calorim.85.4 %65 Hz Ev.Filter Tracking81.5 %43 Hz EF match80.0 %34 Hz (*)Barrel-endcap crack excluded Background composition after trigger: W  e 20% Z  ee 6% e from b and c decays8%  (quark brem & prompt) 14% Other (  0 , jets, etc) 52% Using fully simulated single electron Monte-Carlo samples with p T elec =25GeV Trigger rate studied using Standard- Model simulated sample: mostly QCD di-jet events Pileup (overlapped p-p events) and electronic noise included Initial (low) luminosity scenario “2e15i” and “e60” signatures also studied: efficiency above ~90% for E T above threshold Note: high uncertainty on rate, due to QCD jet cross section uncertainty e25i electron p T >25 GeV isolated

Ricardo Goncalo, Royal Holloway University of London12 ATLAS HLT e/gamma selection Efficiency optimization Efficiency must be balanced against the trigger output rate to optimize available bandwidth Every point in the plot corresponds to a set of selection cut values Efficiency vs. rate/jet rejection curve provides continuous set of working points Envelope is optimum rejection for each efficiency value Level2 Tracking

Ricardo Goncalo, Royal Holloway University of London13 ATLAS HLT e/gamma selection Photon menu optimization A different way of doing optimization: This study fixed Level 2 and Event Filter efficiency at 90% and 80%, respectively Optimize background rejection by minimizing function below using simplex method Efficiency 2  20i  60 Level 290% Event filter80% Rate 2  1 Hz9.8  0.5 Hz Barrel-endcap crack excluded 2  20i photon p T >20 GeV isolated double trigger

Ricardo Goncalo, Royal Holloway University of London14 ATLAS HLT e/gamma selection Physics applications Z  e + e - and W  e  Precision SM physics  Detector commissioning  Detector calibration  Luminosity measurement H   Relevant for low-mass Higgs boson searches (~100 – 150 GeV)  Very demanding on calorimeter resolution and linearity Efficiency numbers wrt the following kinematical cuts:  Z  e + e - : 2 electrons with E T >15 GeV, |  |<2.5  W  e : 1 electron with E T >25 GeV, |  |<2.5  H  (m H =120 GeV) 1 photon with E T >40 GeV, |  |<2.5 1 photon with E T >25 GeV, |  |<2.5 Efficiency Ze+e-Ze+e- W  e 2e15i67.2%  e25i92.9%79.6% e6020.4%6.9% all94.8%80.3% Barrel-endcap crack excluded H  (m H =120 GeV) Efficiency 2  20i  602  20i   60 Level 194%85%98% Level 284%81%94% Event filter78%69%89% Barrel-endcap crack excluded

Ricardo Goncalo, Royal Holloway University of London15 ATLAS HLT e/gamma selection Trigger efficiency from data Electron trigger efficiency from real Z  e + e - data: 1. Tag Z events with single electron trigger (e.g. e25i) 2. Count events with a second electron (2e25i) and m ee  m Z No dependence found on background level (5%, 20%, 50% tried) ~3% statistical uncertainty after 30 mins at initial luminosity Small estimated systematic uncertainty Method Ze+e-Ze+e- counting Level 2 efficiency87.0 % M Z (GeV)

Ricardo Goncalo, Royal Holloway University of London16 ATLAS HLT e/gamma selection Timing studies Timing of the trigger algorithms essential for performance Times estimated for a 8 GHz CPU and 1 RoI/event Level 2 latency is 10 ms; still work to do here but much progress made recently Most of the Level 2 time taken by unpacking of data Event Filter time small wrt allowed latency (~1s)

Ricardo Goncalo, Royal Holloway University of London17 ATLAS HLT e/gamma selection Test beam studies 2004 combined test beam used prototypes of all ATLAS subdetector systems Objective was to study e/  separation and electron efficiency with real data and real detector A good opportunity to test the tools Tracking algorithms used without modifications Tracking efficiency measured always above 95% Data sample Electron eff. (%)  fake rate (%) 20 GeV 95.3   GeV 94.9   0.2 Using calorimeter data only:

Conclusions and outlook

Ricardo Goncalo, Royal Holloway University of London19 ATLAS HLT e/gamma selection Conclusions and outlook The LHC will turn on in 2 years time (not such a long time to go) Much work still needed prior to data taking, but e/  signatures are well developed and should be able to cope with the harsh environment of the LHC Signatures exercised on fully simulated physics channels, and with real data in test beam Looking forward to triggering on real data at the LHC!