1. Update: High energy photon pairs Search for RS-1 Gravitons
Vladimir Litvin, Toyoko Orimoto
Caltech
QCD Meeting
11 December 2007

2. Introduction Feasiblity study for RS-1 G* Relevance to QCD group:
QCD backgrounds are amongst our largest
MC Samples:
Produced privately at Caltech & Fermilab (no pileup)
CMSSW_1_6_7
Signal RS-1 G:
MG=750, 1250, 1000, & 1500 GeV (with c = 0.1 & 0.01)
Main Backgrounds:
Irreducible qqbar & gg (Born & Box)
Gamma+jets (Brem), in which one photon comes from the hard interaction, and the second comes from FSR from an outgoing quark OR the decay of a neutral hadron
QCD jets: where we get photons from the decay of neutral hadrons

3. MC Samples Signal Samples Background Samples
Box and Born produced with CKIN(3)=200 (low limit of the invariant mass of the two outgoing photons, with high end is set to infinity).
Filters were used when producing Brem and QCD samples, with selection rates 1:80 and 1:1200, respectively.

4. HLT EG Efficiency e- related HLT paths
HLT1EMHighEt: 1 SC with pt>80 GeV and loose isolation:
ECAL isol: < 5 GeV in R<0.3 + H/E: <5% , HCAL isol + <8 GeV in 0.15<R<0.3 + tracker iso: <4 tracks in R<0.3
HLT1EMVeryHighEt: 1 Super-Cluster (SC) with pt>200 GeV

5. Selection:  Invariant Mass
Signal Samples
750
=15
1000
=19
1250
=22
1500
=28
Background Samples
Born
Box
Brem
QCD
Only requirement here: photon SC pt > 100 GeV… “final” cut is a 3window around MG

6. Selection: photon pt
750
1000
1250
1500
Each photon candidate is identified as a SC in ECAL. We require:
SC pt > 100 GeV
Born
Box
Brem
QCD
750
1000
1250
1500
Higher pt photon
Lower pt photon
Born
Box
Brem
QCD
Red = All candidates
Blue = MC Truth required

7. Selection: ECAL Isolation
750
1000
1250
1500
ECAL Iso =  ET within cone of R < 0.5
ECAL Iso < 2%
Plotted here are the Log of Ecal Iso
Born
Box
Brem
QCD
750
1000
1250
1500
Higher pt photon
Lower pt photon
Born
Box
Brem
QCD
Red = All candidates
Blue = MC Truth required

8. Selection: H/E
750
1000
1250
1500
H/E =  E of HCAL RecHits within cone of R < 0.3 / ET of SC
H/E < 5%
Plotted here are the Log of H/E
Born
Box
Brem
QCD
750
1000
1250
1500
Higher pt photon
Lower pt photon
Born
Box
Brem
QCD
Red = All candidates
Blue = MC Truth required

9. Selection: Tracker Isolation, nTracks
750
1000
1250
1500
Tracker Isolation = number of tracks within cone of R < 0.3
nTracks <= 3
Plotted here are the Log of nTracks
Born
Box
Brem
QCD
750
1000
1250
1500
Higher pt photon
Lower pt photon
Born
Box
Brem
QCD
Red = All candidates
Blue = MC Truth required

10. Selection: Tracker Isolation, pt Tracks
750
1000
1250
1500
Tracker Isolation =
pT of tracks within cone of R < 0.3 / pT of SC
pT of tracks / SC pT< 1%
Plotted here is the Log of this variable.
Born
Box
Brem
QCD
750
1000
1250
1500
Higher pt photon
Lower pt photon
Born
Box
Brem
QCD
Red = All candidates
Blue = MC Truth required

11. “Final” Selection Rates
Selection rates after applying cascade of afore-mentioned cuts (%)
With HLT1EMHighEt requirement
Signal efficiencies ~40-50%, low (all <0.2%) background effiiency

12. “Final” Selection Rates
Selection rates after applying cascade of afore-mentioned cuts (%)
With HLT1EMVeryHighEt requirement
Signal efficiencies ~40-50%, low (all <0.2%) background effiiency

13. Mass Distributions After Selection
750
1000
1250
1500
Born
Box
Brem
QCD
Numbers of events scaled to 1 fb-1

14. Next Steps Analysis: Production:
Finalizing analysis and working on note based on MC studies
Systematic Errors
Comparison with ORCA results
Currently, this is a simple MC analysis but we will eventually develop a more robust strategy for dealing with real data
Production:
167: have produced signal with c=0.01, will include next
171: produced a couple samples before realizing huge memory usage in this release.. Noticed some large differences in efficiencies and selection variable distributions..
172: just produced a few samples, currently checking
Working on gamma+jets and 2gamma+jets Alpgen or CompHEP samples to better emulate backgrounds

15. Backup Slides

16. Comparison of HLT EG in 167 vs 171
E-gamma HLT path efficiencies
Born
Box
167
171
167
171
Appears that EG efficiencies in HLT have changed significantly from 167 to 171

17. 167 vs 171 Selection: photon pt
Born bkg
Box bkg
Higher Et photon
Red = 167
Blue = 171
Lower Et photon

18. 167 vs 171 Selection: ECAL Iso Born bkg Box bkg Higher Et photon
Strange peaks
Higher Et photon
Red = 167
Blue = 171
Lower Et photon

19. 167 vs 171 Selection: H/E Born bkg Box bkg Higher Et photon Red = 167
Blue = 171
Lower Et photon
More narrow H/E distributions

20. 167 vs 171 Selection: Tracker Isolation (pt_tracks)
Born bkg
Box bkg
Significant change in tracker isolation distributions
Higher Et photon
Red = 167
Blue = 171
Lower Et photon

21. 167 vs 171 Selection: Tracker Isolation (N_tracks)
Born bkg
Box bkg
Decrease in tracking efficiency?
Higher Et photon
Red = 167
Blue = 171
Lower Et photon