1. Ellie Dobson e.dobson1@physics.ox.ac.uk Z→ee and missing Et resolution In collaboration with Paul Prichard (Liverpool), Tony Weidberg (Oxford) Files produced by Barry King (Liverpool), Alessandro Tricoli & Monika Wielers (RAL)  Reminder of methodology and results in v11  Testing of calibration of EtMiss variables  Preliminary v12 MET resolution results in RefFinal and Obj variables  Ideas for further plans UK standard model meeting 26/02/07 http://www-pnp.physics.ox.ac.uk/~dobson/MET/Z-ee/Liv_v12/MET_v12_nojets.pdf

2. 2 The boring stuff Details on study: - http://www-pnp.physics.ox.ac.uk/~dobson/MET/Z-ee/Liv_v12/MET_v12_nojets.pdf (details on MET study) - http://www-pnp.physics.ox.ac.uk/~dobson/MET/Z-ee/Diagnostic.pdf (details on diagnostics) - Oxford talk I http://www-pnp.physics.ox.ac.uk/~tseng/atlas (21-10-06) - W/Z+jets group talk http://indico.cern.ch/conferenceDisplay.py?confId=9738 (13-12-06) - JetEtMiss group talk http://indico.cern.ch/conferenceDisplay.py?confId=6201 (29-12-06) - Oxford talk II http://www-pnp.physics.ox.ac.uk/~tseng/atlas (13-02-07) - Upcoming presentations T&P week (Mid March), SUSY group? Files: 11.0.42 - RAL (A.Tricoli and M.Wielers) 1419 inclusive Z→ee events - Liverpool (Barry King) Ntuples from grid CSC files produced on castor (50,000 inclusive Z → ee events) 12.0.4 - Small samples produced at RAL for testing new EtMiss variables - Large scale production of 20k Z+Jet and 15k Z→ee events in collaboration with Liverpool http://www-pnp.physics.ox.ac.uk/~dobson/Ntupling/Ntupling.pdf Cuts: IsEM(e) == 0 Electron Pt > 10.0 GeV Electron |  < 2.4 and Electron algorithm == 1 Kt jet size 0.4

3. 3 Defining parallel and perpendicular axes e+ e- Parallel Bisector Perpendicular Bisector 0       Perpendicular axis defined to reduce the error on the measurement to an error in angle only.

4. 4 The benefits of the perpendicular axis… Resolution better along perpendicular than along parallel

5. 5 Again, resolution better along perpendicular than along parallel Measure of the calorimeter resolution in Z→ee events Note the offset in the mean value for the perpendicularly resolved variable (this should theoretically equal 0)…

6. 6 Version 11 Woes  The missing Et algorithm in version 11 is very sick  This plot is useful as a diagnostic tool  The problem involves the double counting of objects. The Et miss in versions before 12.0.2 has the problem that cells are used more than once in the final state objects (electrons, jets, photons), since the algorithms run independently from each other. Distribution of means should be a horizontal line

7. 7 Miscalibration in version 12  Health warning: variable MET_Ex(y)missRefFinal  must be used instead of MET_Ex(y)missFinal (the ‘standard missEt container’)  The new variable is calibrated by first producing a map for each cell and tag it if it was used already by another object. The package that is used in version 12 runs over this map and takes cells out if they were used already or gives a weight if two jet objects share them. MissEtDet MissEtDetRefFinal

8. 8 Other EtMiss variables MissETLocHadFinal=MissETLocHadTopoObjDet+ MissETMuonBoyDet Classification of large transverse momentum objects (electrons, photons, muons, tau jets and jets) Further classification of low transverse momentum objects into hadronic and electromagnetic depositions. Object Based Method Hadronic Calibration

9. 9 Statistical summary VersionETmiss variableNumber of events Mean (Et )σ (Et )ErrorN (deviations from mean) 11.0.42MissEtDet489-3.3556.1330.277 -12 11.0.42MissEtDet13529-3.1706.6190.057 -56 12.0.4MissEtDet1706-3.4486.7440.163 -21 12.0.4MissEtRefFinalDet1706-1.4766.2070.150 -10 12.0.4MissEtLocHadFinal1706-4.0696.7240.163 -25 12.0.4ObjMissEtDet1706-0.9697.3210.177 -5  Improvement is seen from version 11  The Obj and RefFinal variables in version 12 are the best calibrated seen yet  Testing is still in progress…  Further improvement in the calibration is expected

10. 10 Resolution variable with hadronic activity Negative due to miscalibration of EtMiss Want to parameterise how σ behaves with SumEt… Use cluster energy, not jet energy (see backup) Scalar sum of hadronic activity

11. 11 Parameterising increase in σ with SumEt RefFinal variable We see an offset – σ does not tend to zero with SumEt

12. 12 Statistical summary MissEt variableAxis of resolutionFitting functionProb MissEtRefFinalPerpendicularLinear1.540.066 MissEtObjPerpendicularLinear1.810.019 MissEtRefFinalParallelLinear1.540.063 MissEtObjParallelLinear1.590.050 MissEtRefFinalPerpendicularSqrt1.450.096 MissEtObjPerpendicularSqrt1.130.311 MissEtRefFinalParallelSqrt1.450.099 MissEtObjParallelSqrt1.720.030 MissEtRefFinalPerpendicularExponential1.630.048 MissEtObjPerpendicularExponential1.910.013 MissEtObjPerpendicularExponential1.810.026 MissEtObjParallelExponential1.640.046

13. 13 Non Gaussian Tails Percentage of events lying outside 1,2,3 σ… …with crack removal We need more statistics to investigate this further Significant non Gaussian tails seen

14. 14 Kt Jet η distribution for events with EtMiss lying outside 0,1,2,3 σ Significant evidence for unusual EtMiss signature corresponding to jets pointing to crack regions (although limited statistics)

15. 15 Status, conclusions and further plans The method works in principle but we must have better calibrated variables before work can progress. Testing in progress in collaboration with Donatella Cavalli, Bruce Mellado and Alessandro Tricoli on the new version 12 variables using the ‘diagnostic plot’. MissEtRefFinal and ObjMissEt seem the most promising so far. Z→μμ Ntuples from the Wisconsin group just available, and testing in ObjMissEt commencing. Will test if the problem in calibration is confined to the electron sector. I have added information into the ntuple from the RefFinal, Obj and LocHadTopo from the AOD file so that analysis may proceed with these v12 ntuples. We need ZJet files for further analysis on events lying outside 3σ (next task)

16. 16 Backup

17. 17 Method 2Method 1 || Method 1: Nomalise electrons before adding, and normalise the resultant Method 2: Normalise the resultant of the electron vector sum only

18. 18  Method 1 leads to a better resolution along the perpendicular then method 2.  Perpendicular axis yields a better resolution than parallel || RMS(meth.1) =2.8 GeV RMS(meth.1) =4.4 GeV

19. 19 Resolved electron energies Perpendicular always positive by construction

20. 20 Jet Energies Method b (clusters)Method a (jets) Smearing effect worse in method a than in method b Should always be negative by momentum conservation Method a: Take the vector sum of the jet transverse energies Method b: Take the vector sum of the energy in the hadronic topological clusters

21. 21 Distribution of hadronic activity

22. 22 Resolution variable against hadronic activity: scatter plots 0 <=Pt(hadronic activity)<=200GeV (reasonable statistics)

23. 23 Resolution Variable – profile plot Method b Method a Method b (clusters) is again more accurate and is thus used for all subsequent analysis

24. 24 Gaussian Fitting to profile plots Bin 0Bin 5 Bin 14Bin 9 Fits are good (although care must be taken to check fits at high hadronic energy)

25. 25  values of Gauss fits:

26. 26 Fraction of events outside 1,2,3 σ for the Obj variable

27. 27 Jet Parameters for the zee events

28. 28 Diagnostic plot in the events used

29. 29 Fitting of Gaussian parameters for the Obj variable

30. 30 Truth variable No negative offset along the ┴ in the truth

31. 31 Application of MET studies to SUSY e+ e- Jets MEt due to imperfect detector v v Jets MEt due to imperfect detector Pt(vv) + Pt(Jets) + MET(had miscalibration) = 0 MET(e miscalibration) + MET(had miscalibration) + Pt(e+e-) + Pt(Jets) = 0 Constant ratio If we assume that MET(had miscalibration) >> MET(e miscalibration) This gives us the Pt distribution for the channel Z→vv Important for SUSY background studies (presentations upcoming to SUSY1 CSC Note meeting)