180 likes | 295 Views
(Di-)Photon + MET Status and Plans for 5 fb -1 Analysis. Bruce Schumm 3 Nov 2011. Who plans to contribute? DESY (Ehrenfeld, Wildt, Vankov) Annecy (Przysiezniak-Frey) Penn (Williams, Bradmiller-Feld) Santa Cruz (Damiani, Kim, Schumm) La Plata (Dova, Alonso) Tokyo (Jinnouchi).
E N D
(Di-)Photon + MET Status and Plans for 5 fb-1 Analysis Bruce Schumm 3 Nov 2011 Who plans to contribute? • DESY (Ehrenfeld, Wildt, Vankov) • Annecy (Przysiezniak-Frey) • Penn (Williams, Bradmiller-Feld) • Santa Cruz (Damiani, Kim, Schumm) • La Plata (Dova, Alonso) • Tokyo (Jinnouchi)
1 fb-1 Analysis: Thumbnail Sketch • (First-order) signal selection straightforward: • 2 tight isolated photons with ET 25 GeV • ETmiss 125 GeV • Optimization based only on ETmiss cut value. Compare to CMS analysis: • At least 2 isolated photons withET 45,30 GeV • ETmiss 100 GeV • At least one jet with ET 30 GeV
Much time spent in deconstructing this difference, but at the end of the day, it all comes down to expected limit. How could these numbers be consistent?
The answer to the conundrum lies in the acceptance*efficiency: e.g. at (mg,mB,mq) = (880,375,1520) ATLAS 27% CMS 12% But let’s look at backgrounds… “Instrumental” “Genuine”
Our estimated backgrounds: But CMS made their cut at 100 GeV; for that cut CMS significantly reduced the “genuine” component (jet requirement? e gamma fake rate?) at expense of efficiency.
In addition, examination of our high-PT events suggested • e fakes do indeed dominate • alignment of ETMiss with photon • This motivates • Studies to reduce backgrounds • Use of additional discriminating variables • Since we have a little time, we’ve tried to open things up a bit (but we are now beginning to refocus on optimization for final event selection)
0 conversion 1 conversion 2 conversions 1 conversion 0 conversion 2 conversions
Bino-Like Grid Points 2-d Gluino vs Bino grid Gluino masses: 800, 850, 900, 950, 1000, 1050, 1100, 1200 Bino masses: 50, 100, 150, 300, 450, … , mgl – 20, mgl – 10 79 points total – 5000 events per point 2-d Squark vs Bino grid Mainly care about this grid to measure acceptance differences with respect to the gluino-bino grid. Same basic structure of points as above. Status: Submitted; awaiting processing. 6 Oct 2011 SUSY Photon Meeting 10
So where do we go from here? • Penn/La Plata: Explore isolation w/ W sample; devise pragmatic suggestion for optimization: • ETCone_20,30,40? • Cut energy? • Energy-dependent cuts? • Leading vs. sub-leading? • La Plata/UCSC: Discrimination of other observables: • Separate conversions categories, pixel hits for conversions, ETMiss (scaled?), Photon-ETMiss , Photon ET, Photon-ETMiss transverse mass, Meff/HT, jet activity • DESY: Contamination in QCD control sample (W, signal…)
Rough Calendar (Feedback?) Preliminary studies (isolation, discriminating variables, QCD control sample) finish this week Optimization over next two weeks (next week through 21-Nov?) Un-blind (Through 1-Dec?) Systematics (Through 15-Dec?) What should target for support note be?
Photon + X + ETMiss Starting to ask: what should “X” be? CMS: X = 3 jets above 30 GeV Have started to look at signal, background for single tight photon selection (MC; no trigger selection yet) Signal: 800 GeV gluino, 400 GeV Bino, all else at ~
Photon + X + ETMiss Next Steps Look at full 1 fb-1 (5fb-1?) loose-tight ETMiss distributions to gauge QCD backgrounds from data. Look at W+jets and ttbar for other possible contributions To me, backgrounds look a little less daunting than I had feared (but this is all MC, and only a partial sampling so far)