1. Using direct photons for L1Calo monitoring + looking at data09 Hardeep Bansil University of Birmingham Birmingham ATLAS Weekly Meeting February 18, 2010

2. Talk Outline Direct Photons L1Calo Motivation Analysis – Monte Carlo Initial Monte Carlo studies Analysis – data09 data09 studies Next steps Summary 2

3. Direct Photons ‘Direct’ or ‘prompt’ photons originating directly from hard pp interaction Leading order processes – Compton (qg) – Annihilation (qq) 3 Francesca Bucci, 10/2008, Direct Photons at ATLAS

4. Direct Photons Have a γj final state Background from ‘fake’ photons created in the jet Long term strategy – Select events that were triggered by jet and get efficiency for γ (and vice versa) Independent triggers 4 Direct photon event signature seen in ATLANTIS Event Display, Mark Stockton, Production of Direct Photons at ATLAS, 6-8/4/2009

5. Direct Photons Jet and photon in event – Both EM and HAD scales are involved Higher cross section than Z → ee (approx 2 orders of magnitude) γ complementary to e but still isolated 5 L. Carminati, 28/8/2009

6. L1Calo Extremely Quick Guide Modules in Calorimeter Trigger for deciding good events based on EM objects, Tau, Jets etc. Sends data and Region of Interest (RoI) info out to other places I am interesting in using EmTau RoIs Eric Eisenhandler, Level 1 Trigger Overview 6

7. Motivation Using direct photons for: – Efficiency determination of electromagnetic and jet triggers – Monitoring and calibration of L1Calo with physics objects 7

8. Analysis – Monte Carlo Athena version 15.5.1 TrigT1CaloAnalysisExamples package – Package by JM for access to Trigger in Athena MC Direct Photon data files – PYTHIA, √s = 10 TeV, p T > 150 GeV (Lumi?) Get reconstructed offline photon candidates in each event from PhotonAODCollection Select the photon with highest E T Compare with EmTau RoIs 8

9. Initial Monte Carlo studies Distributions for highest E T offline reconstructed photons No other selection criteria MC p T cut at 150 GeV – v. high Want to look at lower energies 9

10. Initial Monte Carlo studies Resolution in η, Δη = η OfflinePhoton – η EmTauRoI 10 All PhotonsDirect Photons (> 50 GeV)

11. Initial Monte Carlo studies Resolution in φ, Δφ = φ OfflinePhoton – φ EmTauRoI 11 All PhotonsDirect Photons (> 50 GeV)

12. Initial Monte Carlo studies Calculate Δr = √(Δη) 2 +(Δφ) 2 using offline photon and EmTau RoI Good photons have Δr < 0.4, ignore others (< 5%) 12

13. Initial Monte Carlo studies Efficiency in E T for offline photons = Photon Δr < 0.4 & EM/TAU Threshold passed Photon Δr < 0.4 Affected by poor statistics at low and high E T (not the best MC) 13 Bit 02 – No Isolation, E T > 14 GeV Bit 06 – Isolation 24 GeV

14. Analysis – data09 Athena Tier0-15.5.4.9 TrigT1CaloAnalysisExamples package Using DESDs of the following data09 runs (DESD_COLLCAND) – 141707 141746 141749 141811 141994 141999 142042 142065 142144 142149 142154 142157 142161 142165 142166 142171 142174 142189 142191 142193 142195 142308 142383 142390 142397 142400 142402 142406 Trigger threshold L1_MBTS_1_1 Get reconstructed offline photon / electron candidates from respective AODCollections Compare with EmTau RoIs Merged results together to improve statistics 14 EM timing for L1Calo adjustedWhole L1Calo timing adjusted2 TeV runs (others 900 GeV)

15. Analysis – data09 Are the electrons the photons and vice versa? Check author (reconstruction algorithm) Use isEM Flag – Wanted to check at Tight isEm level 211 tight photons, 35 have ROI, 32 triggered 18 tight electrons, 1 has ROI, 0 triggered – Numbers of γ candidates > e candidates as expected – Better to use Loose isEm than to use nothing 15 Electron LoosePhoton Loose Electron MediumNo Photon Medium Electron TightPhoton Tight

16. data09 studies Loose Offline Photon Candidate Distributions – 811 loose photons, 161 have ROI, 146 triggered Get considerably fewer EmTau RoIs than photons – Opposite to MC results, investigation ongoing 16

17. data09 studies 146 triggered /161 RoIs Could study other 15 event-by-event Set requirement on ΔE T ? Ask for tighter Δr cut? ΔE T Δη Δφ 17 ΔrΔr

18. data09 studies Loose isEM should cut all candidates with |η| < 2.49 but still get an entry? 18

19. data09 studies Look at which candidates match with RoI and which do not match – Those not matching generally have lower E T – Some with high enough E T which should match? Not really e/γ? Hardware missing them? Mistimed towers? 19

20. data09 studies E T for Offline Photons against EmTau RoIs Majority match well in E T, a few don’t – Different regions of behaviour 20

21. data09 studies Efficiency = Low statistics so used variable binning Not as good as MC but need more data 21 Bit 00 – No Isolation, E T > ~3 GeV Bit 05 – Isolation ~15 GeV

22. data09 studies Loose isEM electrons (standard egamma) – Get more candidates than for photons - 1234 loose electrons, 233 have ROI, 171 triggered - unexpected? Background? Better ID for e than γ? 22

23. Next steps Understand why so few EmTau RoIs are created compared to numbers of e/γ Look in more detail at the selection of events – Balance between having enough statistics and filtering out misidentified electrons/photons Advice from Juergen for improving electron selection – No more cosmics shown in plots hopefully! Cut out all bad lumi blocks Use data10 Check with other simulated direct photon datasets – Direct photons generated with lower E T threshold Use most up to date version of Analysis Package 23

24. Summary Direct photons Monitor / calibrate Good results with existing MC data Interesting results with real data – More effort required to select useful EM objects Lots to look at 24

25. Thank you for listening Any questions / comments / suggestions? Thanks to Paul Newman, Juraj Bracinik, Miriam Watson and Juergen Thomas for their help