1. Status and Prospects of the H → γγ Analysis Jim Branson - Marco Pieri - Sean Simon UCSD Meeting March 11 th 2008

2. 11-Mar-08Marco Pieri 2 Introduction  H→ γγ analysis will start to be more important for Int L >~ 1 fb -1  UCSD has played a major role in the PTDR studies and is expected to play a major role in the next years  Other people/groups contributing are:  Caltech, Lyon, Notre Dame, Saclay, ….  For 2008 not much to be expected in H→ γγ channel  In addition the ECAL calibration will not be optimal  Related analyses: γ+jet, γγ from SM (except Higgs) – Should collaborate more with people working on them  Since about 1 month started revisiting the analysis framework to have it more flexible and common with other analyses  For now we ran over small MC samples: ~100k GamJet + ~100k Higgs + ~ 100k QCD + photonsJets + ~50k Dy  All what shown here very preliminary  News: In CMSSW 2_0_0 photons a 5 GeV E t cut an H/E cut at 0.2 will be applied for reconstructing photons

3. 11-Mar-08Marco Pieri 3 forward jets Photons from Higgs decay qqH → qq γγ M H = 120 GeV H→ γγ Signal SIGNAL: two isolated photons with large E t  Gluon-gluon fusion  WW and ZZ fusion (Weak Boson Fusion)  WH, ZH, ttH (additional leptons and MET)  Total σ x BR ~95 fb for M H = 110-130 GeV  Very good mass resolution H → γγ M H = 115 GeV Jets from qq are at high rapidity and large Δ η

4. 11-Mar-08Marco Pieri 4 BACKGROUND  ‘irreducible’ backgrounds, two real photons  gg → γγ (box diagram)  qq → γγ (born diagram)  pp → γ+jets (2 prompt γ)  ‘reducible’ backgrounds, at least one fake photons or electrons  pp → γ+jets (1 prompt γ + 1 fake γ)  pp → jets (2 fake γ)  pp → ee (Drell Yan) when electrons are mis-identified as photons  Handles for Irredicible BG – Kinematics  Handles for Reducible BG – Until now only Isolation Background to H→ γγ ProcessP that (GeV)Cross section (pb)Events/1 fb -1 pp → γγ (born)>258282K pp → γγ (box)>258282K pp → γ+jets>3090x10 4 90M pp → jets>251x10 8 1x10 11 Drell Yan ee-4x10 3 4M

5. 11-Mar-08Marco Pieri 5 Cross section and K-factors  Signal cross sections and BR used for the PTDR (NLO M. Spira)  K-factors for the background used for the PTDR (to be re-evaluated if needed) pp → γγ (born)1.5 pp → γγ (box)1.2 pp → γ+jets (2 prompt)1.72 pp → γ+ jets (1 prompt+ 1 fake)1 pp → jets1 M=115 GeVM=120 GeVM=130 GeVM=140 GeVM=150 GeV σ (gg fusion)(pb) 39.236.431.627.724.5 σ (IVB fusion) (pb) 4.74.54.13.83.6 σ (HW, HZ, Hqq) (pb) 3.83.32.62.11.7 Total (pb) 47.644.238.333.629.7 BR (H → γγ) 2.08x10 -3 2.21x10 -3 2.24x10 -3 1.95x10 -3 1.40x10 -3 Inclusive σ x BR (fb) 99.397.586.065.541.5

6. 11-Mar-08Marco Pieri 6 PTDR Mass Spectrum of Selected Events  All plots are normalized to an integrated luminosity of 1 fb -1 and the signal is scaled by a factor 10  Fraction of signal is very small (signal/background ~0.1)  Use of background MC can be avoided when we will have data  Data + signal MC can be used for optimizing cuts, training NN and precise BG estimation

7. 11-Mar-08Marco Pieri 7 MC Samples  Requests at: https://twiki.cern.ch/twiki//bin/view/CMS/HiggsWGMCRequestsForHiggsToGamGa m https://twiki.cern.ch/twiki//bin/view/CMS/HiggsWGMCRequestsForHiggsToGamGa m  Higgs Signal (Pythia) masses between 60 and 160 GeV (at Fnal, Cern, Lyon)  gluon-gluon fusion,  IVB fusion,  WH, ZH, ttH  Background came late at it is not yet complete  GamJet 1.9 M events Lyon, Cern, SanDiego - OK  Twophoton_Born 450 K events Lyon - 1/2 of requested  Twophoton_Box 950 K events Lyon - OK  Jets_pt50up 14 M events Cern - 1/6 of requested  DY – Enough I think  Twophoton Skims at UCSD, we will soon run on them processpythia lev cutsgen level cutsgen sigma sim sigma gen level cuts reduction factor # of gen evts # of sim evts Int L (fb-1) gg->gamgam (box) pthat>25 GeV none 36 pb 11M ~28 qq->gamgam (born) pthat>25 GeV none 45 pb 11M ~22 pp->gam +jetpthat>25 GeV Special cuts (~sel B' in CMS IN 2005/018) 90 nb0.6 nb~150300M2M~3.3 pp->jetspthat>50 GeV Special cuts (~sel C' CMS IN 2005/018) 24 ub4.8 nb~500050G10M~2.1

8. 11-Mar-08Marco Pieri 8 Important Points – Reconstruction Level  Trigger and Skims  L1 Trigger  HLT  Skims  Photon isolation  Primary Vertex estimation  Energy Measurement  Ecal crystal calibration  SuperCluster calibration  Photon energy scale  Energy Resolution and Error (maybe optional, was done before)  Photon conversion identification and π 0 rejection

9. 11-Mar-08Marco Pieri 9  Electromagnetic trigger towers are classified in two categories depending on the energy deposition in the calorimeter trigger towers: non-isolated, isolated.  Single isolated  E t >23 GeV  Double isolated  E t >12 GeV  Double non-isolated  E t >19 GeV  At startup thresholds lower  Total electron+photon Level-1 trigger rate ~ 4 kHz  Level-1 trigger efficiency for H→ γγ larger than 99%  Perhaps could still optimize the threshold at which all Isolation L1 cuts are removed, never done until now Level-1 Trigger

10. 11-Mar-08Marco Pieri 10  H → γγ signal has two isolated photons  Dominant background from di-jets and γ+jet has at least one candidate from jet fragmentation that is not well isolated  We keep early conversions in the double stream  HLT trigger efficiency 88% - almost 100% for events selected in the analysis  Trigger is relatively easy for H→ γγ because of high Et photons  Total rate for photons after HLT ~5 Hz  Need to make some improvements, particularly for pre-scaled triggers, I also would like to add the double from single L1 HTL paths (also for electrons?) HLT for Photons PTDR HLT photon selection (still the same I think)

11. 11-Mar-08Marco Pieri 11 Skim for H→ γγ  I made a very simple skim selection last summer  For now very simple:  Double Photon HLT.OR. Single Photon HLT with an additional SC – to easily study trigger efficiency  Will hopefully keep it simple forever  Skimmed datasets not too large ~1-3 Hz for photons  RECO format planned to be used for now  PDPhoton Skim higgsTo2Gamma files are at UCSD now  We should run on them  No veto for electrons – Stream can also be useful to study electrons

12. 11-Mar-08Marco Pieri 12  Reducible backrounds (π 0 ’s and mis-identified jets) have other particles near at least one photon candidate  We are in process of repeating and improving the study we carried out for the PTDR  Most of discriminating variables are built by summing up the E t or P t of calorimeter deposits or tracks within a cone ΔR =  (Δη 2 + Δφ 2 )  To study the performance of isolation variables we use individual photon candidates match or not within ΔR < 0.2 to a prompt generator level photon  Signal is: 120 GeV H → γγ gg-fusion reconstructed photon with E t >30 GeV matched with a generated photon within ΔR 40 GeV NOT matched with a generated photon  Low statistics for now, cannot really look at correlations  Trigger (L1 and HLT) not included Photon Isolation ΔRΔR

13. 11-Mar-08Marco Pieri 13 Photon Isolation – Barrel – QCD p that 80 – 120 GeV  Two possible views, first better for high purity, second better for high efficiency Trigger not included

14. 11-Mar-08Marco Pieri 14 Photon Isolation – Barrel – QCD p that 50 – 80 GeV Trigger not included

15. 11-Mar-08Marco Pieri 15 Photon Isolation – Endcaps – QCD p that 80 – 120 GeV Trigger not included

16. 11-Mar-08Marco Pieri 16 Photon Isolation – Endcaps – QCD p that 50 – 80 GeV Trigger not included

17. 11-Mar-08Marco Pieri 17 Photon Isolation II  For low pthat, isolation much less effective  Should study it better – need more statistics at low pthat  Note that pre-selected QCD events below 50 GeV pthat not simulated  Run on Gumbo skims – already at UCSD  Some more checks must still be carried out  Study the correlation between isolation variables and specify benchmark selections for photons  For the PTDR analysis we used a Neural Network with 2, 3 or 5 of following inputs:  ΔR of the 1 st track with P t >1.5 GeV/c  Sum ECAL E t within ΔR<0.3  The shower shape variable R 9  Sum HCAL E t within ΔR<0.35  Sum tracks E t within ΔR<0.2  We did not use kinematical information, easy to combine these variables with reconstructed mass and photons E t in an optimized H → γγ analysis  Repeat the study in the near future

18. 11-Mar-08Marco Pieri 18 Primary Vertex Determination  New longitudinal interaction spread σ~7.5 cm (was 5 cm)  Vertex estimated from the underlying event and recoiling jet  In PTDR analysis the efficiency of determining the right vertex was ~83% for H→ γγ events after selection  Efficiency for the different types of background is similar and basically irrelevant  First check of usage of identified converted photons – very preliminary  Currently we have datasets with no pileup  Efficiency of reconstructing the right primary vertex ~98% on all generated H→ γγ events  Must be compared with minimum bias events

19. 11-Mar-08Marco Pieri 19 Primary Vertex Determination II ProcessEff (%) H → γγ (gg fusion)82 H → γγ (IVB fusion)89 pp → γγ (born)71 pp → γγ (box)72 pp → γ+jet (2 prompt)78 pp → γ+jet (1 prompt + 1fake)86 pp →jets 90 PTDR low luminosity Efficiency of determining the primary vertex within 5 mm from the true one PTDR analysis

20. 11-Mar-08Marco Pieri 20 Primary Vertex Determination III  Generator level plots for different track pt cuts are provided in the Extra slides

21. 11-Mar-08Marco Pieri 21 Primary Vertex From Photon Conversions At least 1 convpho identified At least 1 selected convpho identified All 35.0%15.7% Vtx within 1 cm 19.7%12.9% Vtx within 1 mm 10.2%9.3%  Choose ConvPho with best e/p  Selected convpho have e/p>0.3, 3DR_vtx < 30 cm  Use all generated H→ γγ events (should apply selection)

22. 11-Mar-08Marco Pieri 22 Primary Vertex Studies  Wider longitudinal beam spot will:  Worsen the Mass resolution for events with the wrong primary vertex or no vertex  Make easier the discrimination between different vertices with converted photons  When we want to optimize Primary Vertex finding we can also use the direction of the total tracks transverse momentum that should be opposite to the Higgs pt

23. 11-Mar-08Marco Pieri 23 PTDR Selection for Cut-Based Inclusive Analysis  Photon selection: photon candidates are reconstructed using the hybrid clustering algorithm in the barrel and the island clustering algorithm in the endcaps  E T1, E T2 > 40, 35 GeV  |η|<2.5  Both photon candidates should match L1 isolated triggers with E T > 12 GeV within ΔR < 0.5  Track isolation  No tracks with p t >1.5 GeV present within ΔR<0.3 around the direction of the photon candidate  Calorimeter isolation  Sum of Et of the ECAL basic clusters within 0.06<ΔR<0.35 around the direction of the photon candidate <6 GeV in barrel, <3 GeV in endcaps  Sum of Et of the HCAL towers within ΔR<0.3 around the direction of the photon candidate<6 GeV(5 GeV) in barrel (endcaps)  If one of the candidate has |eta|>1.4442 the other has to satisfy also: Sum of Et of the ECAL<3, Sum of Et of the HCAL<6 GeV  L1 + HLT inefficiency negligible after selection

24. 11-Mar-08Marco Pieri 24 Higgs M H =120 GeV  PTDR Selection for cut based analysis applied now  We will soon improve the photon selection  Results are in basic agreement with PTDR  Still no pileup, efficiency will be somewhat lower  May provide a first estimate adding minimum bias events  BG still to be evaluated Efficiency Nevt/100 pb-1 Gluon-fusion33.7%2.7 IVB fusion31.9%0.3 WH, ZH, ttH24.3%0.2 Total30.0%3.2

25. 11-Mar-08Marco Pieri 25 Higgs Photons Efficiency Plots  Top plots photon finding efficiency  Bottom plots photon isolation efficiency (PTDR cuts)

26. 11-Mar-08Marco Pieri 26 Higgs Mass Resolution Barrel Endcaps R9>0.93 R9<0.93  Ecal calibration for 100 pb -1  Resolution ~1.5 GeV all (1.25 GeV Barrel, 2.1 GeV Endcaps)

27. 11-Mar-08Marco Pieri 27 Fake Photons from Jets  We ran on very low BG statistics, did not yet estimate the two photon BG  Start studying the single photon efficiency and fake rate  Will compare between QCD and γ + jets

28. 11-Mar-08Marco Pieri 28 QCD Fake Photon Rate – 1 pb -1 Trigger not included Fake Photon RateFake Photon Rate after isolation

29. 11-Mar-08Marco Pieri 29 Photon+jet Fake Photon Rate – 1 pb -1 Trigger not included ??? Should check Fake Photon RateFake Photon Rate after isolation

30. 11-Mar-08Marco Pieri 30 Fake Photon Isolation Efficiency Trigger not included

31. 11-Mar-08Marco Pieri 31 One Photon Rate – 1 pb -1 Trigger not included

32. 11-Mar-08Marco Pieri 32 ECAL Calibration and Photon Energy Scale  Crystal Intercalibration  Electrons from W →e ν decays will be used  Also π 0 and/or η will be used  In CMSSW 2_0_0 there will only be SC corrections, no photon nor electron corrections anymore  Photon energy scale being studied from μμγ by Lyon, Florida State University and Kansas State University  μμγ events can also be used for efficiency studies

33. 11-Mar-08Marco Pieri 33 Photon Conversions  Most of the work carried out by Nancy Marinelli and Notre Dame University  They are currently trying to choose the best candidate  Some changes Photon Objects in CMSSW 2_0_0  In my opinion much more word needed in order to use them for photon identification

34. 11-Mar-08Marco Pieri 34 π 0 Rejection  Start looking at the π 0 rejection NN variables provided in CMSSW  Et photon 1 > 40 GeV  Et photon 2 > 35 GeV  Use unmatched photons for γ + jet  Plots are normalized to unity  No isolation cuts applied  Some discrimination power seen at this stage  R1 and R9 also show discrimination power at this stage

35. 11-Mar-08Marco Pieri 35 Π 0 NN Variable BarrelEndcaps BarrelEndcaps

36. 11-Mar-08Marco Pieri 36 Other Shower Shape Variables – R1 and R9 Barrel Endcaps Barrel Endcaps

37. 11-Mar-08Marco Pieri 37 Important points – Analysis Level  We are restarting on this, just a few hints, we will address these issues in a future presentation  Background simulation  We previously studied generator level preselection for fake photons  The Lyon group is working with DiPhox authors to have a full NLO irreducible BG simulation (ATLAS is using ResBos)  Anyway we should be able to carry out the analysis basically using the BG from data, enough events from sidebands  Signal Simulation (common with other Higgs Channels)  We should get NNLO calculations ad rescale Pythia and MC@NLO to those in order to exploit at best the signal topology  Real analysis – as much as possible from data  Efficiency from data (Z->ee, Z->eeγ, Z->μμγ)  Fake rate from data (not very useful for H → γγ)  Use data (sidebands) to optimize the selection and to estimate the BG properties  Study of systematic errors  Optimized Analysis  Exploit the Different Production Modes (signatures 1l, 2l, MET, VBF)  See how to avoid using MC background also for these  Carry out optimized multivariate/multicategorized analysis  Related Analyses (to be studied since the beginning)  Photon fake rate  Gamma + jet cross section (Fake rate)  Gamma-gamma cross section (Fake rate)

38. 11-Mar-08Marco Pieri 38 Effect of Systematic Errors Input for CL calculation is:  Background expectation from fit to the data (sidebands)  Signal expectation from MC Origin of systematic errors  Error on the BG estimation (statistical from fit of sidebands + uncertainty of the form of the fitted function)  Error on the signal (theoretical σxBR, integrated luminosity, detector + selection efficiency) Effect of systematic errors  Systematic errors on the signal do not change the expected discovery CL  Systematic error on the signal makes exclusion more difficult  Systematic error on the BG makes exclusion and discovery more difficult

39. 11-Mar-08Marco Pieri 39 Main Systematic Errors SIGNAL  Theoretical error on cross section times BR (~15%)  Integrated luminosity (~5%)  Higgs Q t distribution – effect to be evaluated  Selection efficiency (~10%)  Can assume a total of 20% (anyway not important in case of discovery)  Nevertheless systematic errors on the signal may cause the analysis to be less optimized BACKGROUND  Statistical error on the fit of the sidebands (~0.3% for ~20 fb -1 )  Systematic error on the shape of the fitted function (~0.3%)  No other errors when data available

40. 11-Mar-08Marco Pieri 40 Outlook  We started revising the H→ γγ analysis framework so that it can also be used for all other analyses  We only ran over small samples for now  We can now run on larger samples  We are also trying to organize the CMS-wide effort in order not to be alone in the analysis as it was for the PTDR  Getting other groups to contribute to the H→ γγ analysis NEXT STEPS  Continue the studies presented here  Include HLT (and re-optimize it) in our analysis  Need to re-optimize the basic selection for the cut-based analysis  Study more converted photons and π 0 rejection to see if they can be used in the analysis  Get NNLO description of the signal and rescale Pythia – Also check MC@NLO  Look at all issues of the real analysis on data  Look again at the optimization of the analysis

41. 11-Mar-08Marco Pieri 41 End of the talk

42. 11-Mar-08Marco Pieri 42 EXTRA

43. 11-Mar-08Marco Pieri 43 Barrel – p that 80 – 120 GeV Trigger not included

44. 11-Mar-08Marco Pieri 44 Barrel – p that 80 – 120 GeV Trigger not included

45. 11-Mar-08Marco Pieri 45 Track Isolation Barrel Trigger not included

46. 11-Mar-08Marco Pieri 46 Track Isolation Endcaps Trigger not included

47. 11-Mar-08Marco Pieri 47 Ecal Isolation Barrel Trigger not included

48. 11-Mar-08Marco Pieri 48 Ecal Isolation Endcaps Trigger not included

49. 11-Mar-08Marco Pieri 49 Hcal Isolation Barrel Trigger not included

50. 11-Mar-08Marco Pieri 50 Hcal Isolation Endcaps Trigger not included

51. 11-Mar-08Marco Pieri 51 Generator Level, charged pt>1.5 GeV |eta|<2.5

52. 11-Mar-08Marco Pieri 52 Generator Level, charged pt>0.3 GeV |eta|<2.5

53. 11-Mar-08Marco Pieri 53 Z + , Z   A clean source of photons, Z + , Z   A clean source of photons, can determine, with real data: Efficiency of photon triggers Efficiency of photon triggers Determination of photon energy scale Determination of photon energy scale Determination of photon id efficiency Determination of photon id efficiency Determination of photon energy corrections Determination of photon energy corrections PTDR I: Y. Gershtein, AN 2005/040: Results based on fully- reconstructed Z+jets background (ORCA) and smeared generator-level signal Current objective: Validate AN 2005/040 selection and event rate predictions in CMSSW with all fully-simulated and reconstructed signal and background samples, including those backgrounds not previously studied Susan Gascon Shotkin General Interest of Z + , Z  ll « Inner Brem »

54. 11-Mar-08Marco Pieri 54 Z  μμγ ALPGEN Summary Z   Signal Z + jets + jets (*) Bbarttbar Nevents (/pb-1)26.22750.31.706E08 (**)7.8E06 (**)561 After (1)19.1336.30.127 (**)990.8 (**)14.6 (2)16.741.00.127(**)381.7 (**)12.3 (4)13.73.280.121(**)136.6 (**)5.5 (5)6.20.2340.116(**)70.4 (**).82 (6)4.0+ 0.6 + 0.03=4.63--- (7)---0.029<0.03(**)8.74 + 3.02 + 0.24=12 (**)0.08

55. 11-Mar-08Marco Pieri 55 Since 170 up until 180_pre9 sizeable reduction in the reconstruction efficiency I could not find out why 167 180_pre4/5 Conversion radius eta (1/pt rec -1/pt sim )/1/pt sim 167 180_pre10/180 MUCH better in 180_pre10 and 180. No intervention from my side ….. Still would wish to know what happened because in 200_pre2 the situation is as before 180_pre10 eta (1/pt rec -1/pt sim )/1/pt sim Conversion radius Photon Conversions – Nancy Marinelli Last Egamma Meeting