1. Diphoton + MET Analysis Update Bruce Schumm UC Santa Cruz / SCIPP 24 January 2014 Editorial Board Meeting

2. 19 December 20122 Gauge Mediation Grids for 7 TeV Analysis gluino/bino grid squark/bino grid “SPS8” Trajectory For 2012 (8 TeV) Data: Replace “constrained” SPS8 grid with wino/bino grid Bino =  1 0 Wino = degenerate triplet  1  and  2 0 Production through  1   2 0 and  1 +  1 -

3. 19 December 20123 2011 Signal Regions E T miss = LocHadTopo H T = Scalar sum of all transverse energy (no E T miss )   = minimum angle between photon and MET Strong production; high-mass bino Strong production; low-mass bino Electroweak production For 2012: Include additional observables, for certain signal regions: MEFF (or “HT-prime”) = H T + E T miss  j = minimum angle between jet and E T miss Also: E T miss is now EGamma10NoTauLoosePhotonRef

4. 19 December 20124 2012 Optimization Strategy Strong Production (SP1, SP2) Largely the same as for 2011 analysis Explore MEFF,  j, removal of no-pixel hit requirement for conversion tracks Tweak cuts for higher mass scales MEFF somewhat preferable;  j provides no apparent advantage As for 2011, optimal point is largely background-free Optimize for (m gluino,m bino ) = (1300,1050) and (1300,150) Electroweak Production (WP1, WP2) No real preference for MEFF, but highly correlated with E T miss so use H T. As for strong production, two SRs (low-mass, high-mass bino)   helps for high-mass bino SR;  j helps for low-mass bino SR Optimal point will have few-event background Optimize for (m wino,m bino ) = (600,500) and (600,100) NEW: Model-Independent Selection (MIS) Cut at MET for which EW background begins to dominate QCD background

5. 19 December 20125 for strong-production optimization Used wino_bino_600_500, wino_bino_500_100 for weak production > Used background distributions alone for model-independent selection

6. 19 December 20126 Figure of merit for 1300_150 Point Figure of merit for 1300_150 Point; Removing events for which converted tracks have pixel hits Re-Examination of No Pixel-Hit Requirement  Remove no-pixel-hit requirement

7. 19 December 20127

8. 8 1300_1050 (SP1) Optimization Number of signal events (20 fb -1 ) Figure of Merit Chosen point Avoid pitfalls: Rapidly falling signal Fluctuating backgrounds (above “dips” in FOM) SP1 Optimum: (MET,MEFF) > (250,1500)

9. 19 December 20129 Explore  ( ,jet)MET Cuts for SP1 Selection  ,MET > 0.5  jet,MET > 0.5 Statistics not so good, but see Improvement with  ,MET cut Degradation with  jet,MET cut SP1  Choose (MET,MEFF) = (1500,250) with  ,MET > 0.5

10. 19 December 201210 Model-Independent Selection Apply cut  jet,MET > 0.5 EW backgroundQCD background No cut on HT or MEFF; just choose point on horizontal (MET) axis for which the QCD background plunges below the EW background  MET > 250 GeV

11. 19 December 201211 Signal Selection Results

12. 19 December 201212 Status of Background Estimation QCD Backgrounds EW Backgrounds from e  Fakes Irreducible Backgrounds Irreducible Backgrounds (for now SP1, SP2 only): From W , (Z  )  events Estimate from MC samples, scaled to 20 fb -1 Very small for SP1, SP2 (before K factors that are < 3) For both, before K factors: 0.03  0.01 (stat)

13. 19 December 201213 QCD Backgrounds Estimates derived from scaling observed low-MET signal rates to high-MET using control samples Scale to number of signal events Integral above cut provides background estimate

14. 19 December 201214 QCD Backgrounds Continued For each SR, accumulate 8 control samples  In principle, eight independent estimates; establishes systematic range Define Pseudo-photon = loose, plus fail two shower-shape requirements (“FracS2” and “Weta”). Each control sample contains at least one pseudophoton. In addition, it must (QCDtg) or must not (QCDg) contain a tight isolated photon. For each of these two choices, we can further have A cut of either 50 or 75 GeV on the photons (tight and pseudo) Pseudophoton may be isolated or not  8 combinations  5 signal selections  40 QCD background estimates “Undefined”  no control-sample events OR no signal sample events below MET = 60 GeV If no control-sample events above MET cut  set 90% UCL NOTE: In 2011 only one control sample (QCDg 50 non-isolated I recall)

15. 19 December 201215 Et CutIsolated?QCDgQCDtg 75no.07+-.07< 2.3 75yes< 9.2undef 50no.17+-.10< 1.15 50yes< 4.6undef Et CutIsolated?QCDgQCDtg 75no0.5+- 0.11.4+-0.7 75yes< 0.653.7+-3.7 50no.3+-.050.6+-0.2 50yes0.5+-0.20.5+-0.5 SP1 SP2: Undefined (no signal at any MET!) MIS Et CutIsolated?QCDgQCDtg 75no.9 +-.11.4+-0.8 75yes.6 +-.68.4+-8.4 50no.8 +-.11.2+-0.5 50yes.7 +-.54.0+-4.0 WP1 Et CutIsolated?QCDgQCDtg 75no2.7+-0.30.6+-0.4 75yes1.1+-1.1< 22 50no2.6+-0.21.0+-0.4 50yes3.5+-1.6< 11 WP2 Does not exhibit consistency!! Propose QCDtg 50 GeV no-isolation for “nominal” estimate

16. 19 December 201216 Notes on QCD Background Estimates SP2 backgrounds undefined since no signal at any MET. However, if you take this as < 2.3 low-MET signal events at 90% CL, control-sample extrapolations yield < 0.5 events at 90% CL 2011 strong-production estimates were also undefined  developed extrapolation technique (underway now) Strong-production estimates suggest very small backgrounds, systematics under control WP1 and MIS estimates larger, but consistent from sample to sample WP2 estimates inconsistent between QCDg and QCDtg samples. Will require further thought (Osamu Jinnouchi)  Unblind SP1, SP2 first (and perhaps also MIS for the EW grid?) BUT WAIT…

17. 19 December 201217 Late-Breaking News: SP2 Extrapolation Use QCDtg with Et cut of 50 GeV, and do not require “g” to be isolated Best guess as to most representative control sample Will try others (with higher statistics) to check Expected QCD background (linear fit): N back QCD = -0.25 +- 0.49 (or less than 0.55 events at 90%CL)

18. 19 December 201218 e  EW Backgrounds Start with determination of the e  fake rate In bins of  Separately for converted, unconverted photons e  final state ee final state Fake rate is (roughly) the background-subtracted ratio of these two yields

19. 19 December 201219 Note: All 2012 numbers must be multiplied by x2 (only half of Z  ee electrons are above applied 50 GeV cut (table will be fixed soon) Fake Rate Results

20. 19 December 201220 SP1, SP2 Backgrounds at First Blush AnalysisQCD e  EW Irred.Total SP1< 0.5< 0.07~ 0.05Small SP2< 0.50.03~ 0.05Small EW background from e   fakes determined by scaling observed e  events by measured e  fake rate. SP1: no e  events observed  < 0.07 events expected background SP2: 1 e  event observed  0.03 +- 0.03 expected backgrounds Once QCD extrapolation studies are done, SP1, SP2 backgrounds should be in good shape.  Document and request unblinding next week?  Might we also want to finish MIS background studies? (No extrapolations needed for QCD

21. 19 December 201221