Introduction H gamma Higgs boson production and decay
Introduction H gamma ¡Higgs boson production and decay ¡Backgrounds ¡Cut-based analysis ¡Applying Pi 0 discriminator ¡Outlook
SM Higgs at low mass
I. Higgs boson production and decay
I. A events display
H 120_gluon fusion Et Eta Theta E
II. Backgrounds irreducible Backgrounds with two real prompt high Et photons are call “irreducible”. Backgrounds in which at least one final state jet is interpreted as a photon are called “reducible”.
Signals and backgrounds
Mass Spectrum Int L = 1 fb-1
Expected Background and Signal Efficiency
Cut-based analysis ¡ ¡ ¡ Photon selection: photon candidates are reconstructed using the hybrid clustering algorithm in the barrel and the island clustering algorithm in the endcaps ET 1, ET 2 > 40, 35 Ge. V |η|<2. 5 Both photon candidates should match L 1 isolated triggers with ET > 12 Ge. V within ΔR < 0. 5 Track isolation No tracks with pt>1. 5 Ge. V present within ΔR<0. 3 around the direction of the photon candidate. We consider tracks with hits in at least two layers of the silicon pixel detector. Calorimeter isolation Sum of Et of the ECAL basic clusters within 0. 06<ΔR<0. 35 around the direction of the photon candidate <6 Ge. V in barrel, <3 Ge. V in endcaps Sum of Et of the HCAL towers within ΔR<0. 3 around the direction of the photon candidate<6 Ge. V(5 Ge. V) in barrel (endcaps) If one of the candidates has |eta|>1. 4442 the other has to satisfy also: Sum of Et of the ECAL<3, Sum of Et of the HCAL<5 Ge. V
Validation of cut-based analysis for signal MH=120 Ge. V CMS NOTE - 2006/112 Same selection as PTDR Efficiency Total Nevts/100 pb-1 Gluon-fusion 35. 0% 2. 8 IVB fusion 32. 7% 0. 3 WH, ZH, tt. H 26. 7% 0. 2 Total 31. 5% 3. 3 Do same selection as PTDR analysis, the efficiency slightly increases.
Vertex correction with CSA 07 samples M. Pieri presentation 14 mar 08 100 pb-1 calibration CMSSW_1_6_7 CSA 07 MC Our vertex corr. results
Validation of cut-based analysis for backgrounds CMS NOTE - 2006/112 Process 115/Ge. V/c 2 120/Ge. V/c 2 130/Ge. V/c 2 140/Ge. V/c 2 150/Ge. V/c 2 pp γγ(born) 42 +7 36 +6 34 +6 27 +5 26 +5 pp γγ(box) 35 +6 30 +6 22 +5 17 +4 15 +4 pp γ+ jet 92 +10 77 +9 56 +7 38 +6 34 +6 pp jets 43+18 43 +18 41 +16 29 +14 41 +16 The values of the two tables are consistent,but the numbers of the pp->jets channel has large error because of small statistics.
Mass Spectrum in Four Categories Barrel, min(R 91, R 92)>0. 93 Endcap, min(R 91, R 92)>0. 93 Barrel, min(R 91, R 92)<0. 93 Endcap, min(R 91, R 92)<0. 93
Results of cut-based analysis
Applying Pi 0 discriminator pass cutbased selection
Applying Pi 0 discriminator Not pass cut-based selection unconverted photons Not pass cut-based selection all photons (converted and unconverted)
NN cut value vs SC Et in Aris’ talk (Egamma meeting Apr. 21)
Apply NN on no-pass events ¡ Release photon Et’s to ‘safe values’: Et>25, Et>40 Ge. V with respect to gamma jet and jets preselection ¡ For passing events: the NN is applied as Aris' parameterization - 0. 1 (NN as a function of Et in barrel and endcap with preshower) in order to keep more signal. ¡ For no-pass events: two options the NN is applied as Aris’ parameterization, but it is also applied on converted photons (NN>0. 05).
Global relative performance improvement in case 2 interval 120+- 5 Ge. V for 1 fb-1 channel A) Events passing cut based analysis with Applying NN on unconverted photons as Aris’ suggested. The passing cut-based analysis events show in parentheses B) Events passing cutbased analysis + additional no-pass events: Photons with Et>25, 40 Ge. V and NN on unconverted photons as Aris’ suggestion and NN>0. 05 on converted photons. C) Total events Case A + additional no -pass events: Photons with Et>25, 40 Ge. V and NN on unconverted photons as Aris’ suggestion and NN>0. 05 on converted photons. Signal (gg, IVB, ZWtt. H) 25. 5+2. 9+1. 6 (31. 5) 30. 0+3. 7+2. 0 28. 7+3. 6+1. 9 pp γγ(born) 371. 7 (397. 3) 437. 7 412. 1 pp γγ(box) 293. 5 (309. 4) 353. 5 337. 6 pp γ+jet 666. 3 (793. 3) 1017. 7 890. 7 pp jets 374. 6 (444. 6) 580. 5 510. 7 Total 1706. 1 (1944. 6) 2389. 4 2151. 1 S/sqrt(B)_ 0. 726 (0. 713) 1. 018 0. 730 1. 024 0. 737 1. 033
Discrimination between converted photons and π0 s Giorgos Anagnostou Method based on finding the physics objects (e+, e-, γ) and geometric characteristics of the converted π0. Initially conversion identification with variables from the tracker & Ecal using likelihood. rejection of π0 with: asymmetry Ecal surface, distance e-e+, mγγ, E/P & shower shape. training with events in the whole mγγ range, testing around Higgs peak. Rejection ~57% for efficiency 90%.
Summary for Pi 0 dis ¡ Our first attempts with ‘safe’ relaxation of photon Et cuts and utilisation of pi 0 discrimination nn yield small but non-negligible improvements in S/sqrt(B) against backgrounds… ¡ Towards further optimization: Further investigate performance on converted photons We will study characteristics of photons from pi 0 and background samples performing well on nn
Outlook ¡ What I can do with the real data(40 pb-1) in the end of 2008, i. e. , my thesis? ¡ Are the H gamma data available in Beijing? ¡ When I give a starting report about my thesis?
- Slides: 23