1. Tracey BerryExotics, Dilepton, Diphoton, 14 th November 06 Search for andall- undrum Gravitons in the  Channel at CDF Tracey Berry Royal Holloway University of London

2. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 2/32 Outline RS Model – Motivation for ED searches CDF Detector  Search Analysis –Selection Criteria –Efficiency and Acceptance from MC –Systematic Uncertainties –Backgrounds –Systematic Uncertainties on Backgrounds Limits Future Prospects

3. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 3/32 Extra Dimensions: Motivations  if compact space (R  ) is large M Pl 2 ~ R  M Pl(4+  ) (2+  ) Effective M Pl ~ 1TeV if warp factor kR c ~11-12 Planck TeV brane Arkani-Hamed, Dimopoulos, Dvali, Phys Lett B429 (98) Randall, Sundrum, Phys Rev Lett 83 (99) M EW (1 TeV) << M Planck (10 19 GeV)? Many (  ) large compactified EDs In which G can propagate 1 highly curved ED Gravity localised in the ED In the late 90’s Large Extra Dimensions (LED) were proposed as a solution to the hierarchy problem Some of these models can be/have been experimentally tested at high energy colliders Since then, new Extra Dimensional models have been developed and been used to solved other problems: Dark Matter, Dark Energy, SUSY Breaking, etc   = M pl e -kR c    ~ TeV

4. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 4/32 Experimental Signature for Model (Add BRs here?!) (Why not diphoton ?!!) K/M Pl 6 700 GeV KK Graviton at the Tevatron k/M Pl = 1,0.7,0.5,0.3,0.2,0.1 from top to bottom M ll (GeV) RS model Dilepton channel 400 600 800 1000 M ll (GeV) 10 -2 10 -4 10 –6 10 -8 10 -2 10 -4 10 -6 10 -8 10 -10 Tevatron 700 GeV KK graviton d  /dM (pb/GeV) Couplings of each individual KK excitation are determined by the scale,   = M pl e -kRc  ~ TeV Massive eigenstates only suppressed by   -1 ~ TeV -1 RS modes can be excited individually on resonance Via Virtual exchange but different expected cross sections/distributions from ADD model Detect effects of Virtual Graviton exchange Davoudiasl, Hewett, Rizzo hep-ph0006041 700 GeV KK Graviton at the Tevatron k/M Pl = 1,0.7,0.5,0.3,0.2,0.1 from top to bottom M ll (GeV) 1 0.5 0.1 0.05 0.01 RS model 1 0.7 0.5 0.3 0.2 0.1 Dilepton channel 200 400 600 800 1000 1200 M ll (GeV) 10 -2 10 -4 10 –6 10 -8 10 -2 10 -4 10 -6 10 -8 10 -10 Tevatron 700 GeV graviton LHC 1500 GeV graviton d  /dM (pb/GeV) 1000 3000 5000 1 0.5 0.1 0.05 0.01 1 0.7 0.5 0.3 0.2 0.1 KK excitations can be excited individually on resonance 1500 GeV G KK and subsequent tower states K/M Pl New parameters: 1.First graviton excitation mass: m 1 2.A ratio: k/ M pl   = m 1 M pl /kx 1,,  1 =  m 1 x 1 2 ( k/ M pl ) 2 LHC Signature: Narrow, high-mass resonance states in dilepton/dijet/diboson channels Model parameters: Gravity Scale: 1 st graviton excitation mass: m 1   = m 1 M pl /kx 1, & m n =kx n e krc  (J 1 (x n )=0) Coupling constant: c= k/M Pl  1 =  m 1 x 1 2 (k/M pl ) 2  width  position Resonance   = M pl e -kR c  k = curvature, R = compactification radius

5. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 5/32 CDF @ Tevatron p p Collider Highest energy operational collider in the world! and ability to set improved limits on new physics D0 p p CDF √ s  1.96 TeV Ldelivered ~400/pb Run II (2001-2009) √ s  1.96 TeV Physics Analyses use ~ 200/pb collected between 03/02 and 09/03 diphoton search 300/pb p Muon System COT Plug (P) Calorimeter Time-of-Flight Central (C) Calorimeters Solenoid Silicon Tracker

6. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 6/32 Data Sample Data Triggers : DIPHOTON_12/_18 ULTRA PHOTON_50 and SUPER PHOTON_70 EM & JET GoodRun List Luminosity: 1155 pb -1 CC 1070 pb -1 CP Selection Cuts E T > 15 GeV M  > 30 GeV 2 highest E T photons selected to assign region (CC, CP, PP) Good vertex Standard high-P T selection cuts…. Additional corrections applied from Stntuple/photon/ TPhotonUtil.cc (TPhotonUtil::CorrectPhotonEnergy and TPhotonUtil::CorrectPhotonIsolation) To keep high efficiency at high mass: Had/Em inefficient at high E T as EM E saturates, so is miscalculated. PHOTON 70 has no HAD/EM cut

7. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 7/32 Selection Cuts CEM selection cuts PEM selection cuts High-P T photon cuts in order applied. Chi2 dependence: (CDF note 8302- blessed) CEM  2 cut modified due to ET dependence of MC

8. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 8/32 Mass Spectra Highest mass CC event: 602 GeV

9. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 9/32 Highest Mass CC event

10. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 10/32 Photon ID Efficiencies Calculated using Monte Carlo (MC) Scaled by factors derived from Z  ee MC & Data study CEM scaled by 1.00 and PEM by 0.91 Monte Carlo: Randall-Sundrum G  MC generated using HERWIG version 6.510 with k/M Pl = 0.1 and CTEQ5L pdf. 10000 events generated between 200GeV<M G <1050GeV in 50 GeV steps Scaled Photon ID Efficiencies Applied sequentially

11. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 11/32 Scaled Total Efficiency and Acceptance From Monte Carlo

12. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 12/32 Systematic Uncertainties Acceptance Systematic Uncertainty –E T scale (0.2% CEM and 0.8% PEM) –PDF (4%) –FISR (4%) –Luminosity (6%) Efficiency Systematic Uncertainty –Z efficiency scale factor (1% per photon leg) –Trigger efficiency –Z vertex (0.2%) –Photon conversion (2%) Background Systematic Uncertainty –Diphox NLO MC –QCD fake photons from sideband variations

13. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 13/32 Expected Diphoton Spectrum Search selection 2 isolated  E T >15 GeV, |  |<1 Standard Model Diphoton Production estimated using Diphox (NLO), absolutely normalised mass spectrum corrected for efficiency from Pythia SM  prod. Jets Faking Photons:  - jet and jet-jet estimated using sidebands - loosen  selection criteria - exclude tight  events - apply various tighter cuts - fit to the spectrum Normalised in the low mass region ∫ 30 N data = ∫ 30 N diphox +∫ 30 N SB 10 0 e + e - Production

14. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 14/32 SM Diphoton Background Estimated from Diphox NLO MC. Gives mass distribution for events passing E T and  selection. Cross section modified by mass-dependant photon ID efficiency (SM MC PYTHIA). Efficiency = number of reconstructed events passing all cuts divided by number at generator level passing acc. Shows fit and uncertainty. Uncertainty from fit = 0.0002*mass Mass distribution for SM diphoton production fit to general function: y = (x 0.1 +  5 x  6 )(e x/  0 +  1 e x/  2 +  3 e x/  4 ) Where x = mass - mass threshold cut (30 GeV/c 2 )  = free fit parameter

15. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 15/32 Fake Photon Background Jets faking photons Fakes estimated by loosening selection criteria and removing events with 2 good photons. The remaining events constitute the sidebands (M  > 30 GeV/c 2 also required). CEM loose selection criteria PEM loose selection criteria

16. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 16/32 Fake Photon Background –There will be some true diphoton contribution in the sidebands. –Estimate the ratio of photons in data to sidebands from pythia diphoton MC. –Sidebands fit with same function but set  3 to zero and include predicted contribution from true events by including fixed number of events from Diphox distribution. Normalised in the low mass region ∫ 30 N data = ∫ 30 N diphox +∫ 30 N SB 10 0

17. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 17/32 Background Systematic Uncertainties From Diphox: Varying Q by a factor of 2 gives ~ 20% variation in normalisation from cross section. Uncertainty from fit: Translates to ~ 0.0002*mass (uncertainties on efficiency small in comparison). Uncertainties on SB: Estimated from varying SB cuts. Relative difference between variations and standard SBs is ~20 % These apply to display background only and are not used in setting limits. Uncertainties on SB: Estimated from varying SB cuts, for each cut in turn apply signal selection. Also allow looser isolation selection cut. Relative difference between variations and standard SBs is ~20 %

18. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 18/32 e + e - production Background from e + e - production: – Where e tracks have not been reconstructed. – Probability of losing a track is ~ 1% – From e analysis estimate background contribution: –100 events above 200 GeV for 450 pb -1 – Scale by 2 for luminosity –Add probability of 0.01 for losing a track (twice for CC, once for CP) –Leaves 0.5 events out of 198 – Background is less than 0.1% bkg for CC and 1% for CP above mass of 200 GeV. –Probability of losing a track is ~ 1% (cdf note 8220)

19. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 19/32 CC and CP Mass Distributions CC and CP mass distributions with a priori background overlayed and uncertainty. Blue: DiPhox true diphoton events. Red: fake photons predicted from sidebands. Low mass region

20. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 20/32 Limits Likelihood that binned diphoton data ( N d i ) is described by a predicted background (N b i ) and hypothetical signal ( N s i ): 95 % confidence level obtained by integrating likelihood wrt  such that ∫ L(  ) = 0.95  =0 ∞  95  =0 CC and CP channels combined by multiplying individual likelihoods. Compare observed limit to that expected if only background was present: 10 000 pseudo-experiments generated for each mass point and limit calculated. Median taken as expected limit. incorporate systematics by smearing the likelihood smear by integrating over all possible cross-sections weighted by a Gaussian with a mean and sigma equal to the given cross- section and its error

21. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 21/32 CC, CP and Combined Limits CC CP Combined channels

22. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 22/32 CC, CP and Combined Limits CC CP Combined channels Compare observed limit to that expected if only background was present: 10 000 pseudo-experiments generated for each mass point and limit calculated. Median taken as expected limit. Compare observed limit to that expected if only background was present: 10 000 pseudo-experiments generated for each mass point and limit calculated. Median taken as expected limit. Compare observed limit to that expected if only background was present: 10 000 pseudo-experiments generated for each mass point and limit calculated. Median taken as expected limit.

23. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 23/32 Lower Limits on M G K/M Pl Lower mass limit (GeV/c 2 ) 0.1850 0.07784 0.05694 0.025500 0.01230 Upper limit on cross-section shown with theoretical predicted cross- sections from scaling HERWIG and using correction factor K f = 1.3

24. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 24/32 Excluded Region 95% C.L. excluded region on the plane for M G vs k/M Pl for diphoton decay mode

25. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 25/32 Systematic Errors 100% Correlated Errors (all errors on acceptance or lumi) –luminosity –ISR –energy scale –energy resolution –efficiency uncertainty (inflated errors to 2% for CEM/PEM) –PDFs (di-electron changed error to 3.9% CC, 5.2% CP) –photon conversion uncertainty –z vertex efficiency uncertainty Uncorrelated (all bkg errors) –QCD ele bkg shape + norm (ele bkg) –Z MC norm (ele bkg) –EWK cross section (ele bkg) –Pho bkg fit uncertainties (pho bkg) Combining di-photon + di-electron analysis Identical to combining CC+CP channels: just multiply the likelihoods Only issue is the systematic errors combining in the presence of non-systematics is trivial just need to sort out the correlations di-electron analysis (note 8385)

26. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 26/32 Combined Limits

27. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 27/32 Present Limits on RS Model c>0.1 disfavoured as bulk curvature becomes to large (larger than the 5- dim Planck scale) Theoretically preferred   <10TeV assures no new hierarchy appears between m EW and   Present Experimental Limits Theoretical Constraints CDF performed ee &  search, then combine D0 perform a di-EM object search

28. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 28/32 Future Prospects Best channels to search in are G(1)  e+e- and G(1)  due to the energy and angular resolutions of the LHC detectors G(1)  e+e- best chance of discovery due to relatively small bkdg, from Drell-Yan * Sensitive at 5  up to 2080 GeV G 1  c>0.1 disfavoured as bulk curvature becomes to large (larger than the 5-dim Planck scale) Theoretically preferred   <10TeV Allenach et al, hep-ph0006114 Allenach et al, hep-ph0211205 M.-C. Lemaire et al. CMS NOTE 2006/051 CMS PTDR 2006 LHC completely covers the region of interest k/M Pl =0.01

29. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 29/32 Backup

30. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 30/32 Limit Fits CC and CP mass spectra fit with same general function, removing one of the exponentials and including predicted contribution from true diphotons using number of events from DiPhox distribution, the shape of which is allowed to float.

31. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 31/32 Acceptance Systematics E T scale: – E T threshold varied by ± 1% (from Z mass MC/ data comparison) –0.2% CEM and 0.8% PEM PDF: –Relative change calculated for acceptance values generated using different PDFs (high/ low gluon and high/ low  s ) –4% ISR: –Parameter varied in Pythia with CTEQ5L PDFs for more and less ISR –4% Luminosity: –6%

32. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 32/32 Efficiency Systematics Z efficiency scale factor: – from Z MC/ data comparison for photon selection criteria –Systematic error of 1% per photon leg Trigger efficiency: –L2 eff given by offline cuts + L2 trigger/ offline cuts –Had/Em inefficient at high E T as EM E saturated, so miscalculated. 50 (E T and Had/Em) and 70 (E T ) triggers therefore included to prevent loss of events. Trigger eff taken as 100% efficient 41/11088 CC signal events from 50/70 triggers (23 pass both, 18 pass 50 only) and 84/20933 CP signal events (22 pass both, 62 pass 50 only).

33. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 33/32 Efficiency Systematics Z Vertex: –Event selection requires Z vertex falls within ± 60 cm of centre of detector. –Some events may occur outside this region, efficiency for this measured in MC, 97% eff. –Q. from full status: MC does not include run dependent variations of generated Z vertex position- generated using single run. –Efficiency of Z vertexing 96% +/- 0.2% (syst) +/- 0.04% (stat) (cdf note 8318, W. Sakumoto) –0.2% uncertainty and 0.96/0.97 (=0.99) Z vertex eff. Photon Conversion: –10 % conversion probability, amount of detector material known to ~ 10% –take 1% uncertainty on probability of converting per leg, total of 2%.

34. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 34/32 Allenach et al, hep-ph0006114 M G =1.5 TeV 100 fb -1  HERWIG  Main Bkdg: Drell-Yan  Model-independent analysis  RS model with k/M Pl =0.01 as a reference (pessimisitc scenario)  Fast Simulation Sensitive at 5  up to 2080 GeV RS1 Discovery Limit Best channels to search in are G(1)  e+e- and G(1)  due to the energy and angular resolutions of the LHC detectors G(1)  e+e- best chance of discovery due to relatively small bkdg, from Drell-Yan * Di-electron * Reach goes up to 3.5 TeV for c=0.1 for a 20% measurement of the coupling. Allenach et al, hep-ph0211205 *

35. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 35/32 Allenach et al, hep-ph0211205  e BR(G →  ) = 2 * BR(G → ee ) Also the size (R) of the ED could also be estimated from mass and cross- section measurements. RS1 Model Parameters Allenach et al, JHEP 9 19 (2000), JHEP 0212 39 (2002) A resonance could be seen in many other channels: , , jj, bbbar, ttbar, WW, ZZ, hence allowing to check universality of its couplings: Relative precision achievable (in %) for measurements of .B in each channel for fixed points in the M G,   plane. Points with errors above 100% are not shown.

36. Tracey Berry Exotics, Dilepton, Diphoton, 14 th Novemeber 2006 36/32 RS1 Model Determination 100 fb-1 Spin-2 could be determined (spin-1 ruled out) with 90% C.L. up to M G = 1720 GeV Allanach et al, hep-ph 0006114 Note: acceptance at large pseudo-rapidities is essential for spin discrimination (1.5<|eta|<2.5) e+e-e+e- LHC M C = 1.5 TeV M G =1.5 TeV 100 fb -1 Stacked histograms Spin-2 nature of the G(1) can be measured : For masses up to 2.3 TeV (c=0.1) there is a 90 % chance that the spin-2 nature of the graviton can be determined with a 95 % C.L.