1. 1 Di-Boson Physics Study Suen Hou for Diboson working group Oct. 12, 2007, ATLAS Week at CERN Aristotle University of Thessaloniki Brookhaven National Laboratory Cambridge University Duke University University of Michigan Oxford University Institute of Physics, Belgrade Southern Methodist University Taiwan Academia Sinica Univ. of Sci. and Tech. of China

2. 2 Standard Model diagrams for W , Z , WW, WZ, ZZ Productions includes t-channels and Triple Gauge Boson couplings (TGC) Leptonic decays to be measured : W   l , Z  l + l - Standard Model LO: W  : ISR + WW  Z  : ISR only WW : t-channel + WW  + WWZ WZ : t-channel + WWZ ZZ : t-channel only di-Boson Production at Hadron Collider W Z0Z0 Z0Z0 WW  WWZ / WW  WWZ

3. 3 Charged TGC of Standard Model : WW , WWZ  = Z =0 proportional to s of q - q interaction g 1 Z =   =  Z =1 proportional to  s Neutral TGC : ZZZ, ZZ , Z ,  forbidden (Z and  carries no charge nor weak isospin) Triple Gauge Boson coupling ,,

4. CSC datasets for di-Boson Study CSC12.0.6.4 and CSC11.0.42 are in use Final results are given based on 12.0.6.4 (all diboson signals) W , Z  signal: Pythia W/Z inclusive (1.4 fb -1 total) Pythia W/ZPhoton (ISR events, 2 fb -1 total) Background: Pythia W/Z inclusive Photon tagged with MCtruth for ISR, FSR, fake in jets, underlying WW, WZ, ZZ signal: MC@NLO 12.0.6.4 (W and Z mass widthes are not included in MC for WZ and ZZ) Background: MC@NLO 12.0.6.4 (ttbar, Wjet, Zjet, inclusive W, Z) + Pythia 11.0.42 W/Zjet, W/Zgamma, DY (30 M events) + gg2WW 12.0.6.5 TGC MC generator: BHO 4

5. 5 Analysis using Boosted Decision-Trees (BDT) Z+ISR training of Z+ISR against Z+FSR and Z+fake_γ Z(ee)γ: S/B= 289/160 for 1 fb -1 (significance: 23) Z(μμ)γ: S/B= 650/343 for 1 fb -1 (significance: 35) BDT analysis based on 19 variables to distinguish ISR photon from bkgd Final cuts chosen at 60% efficiency Zγ → l + l -  event selection Z-FSR Z-ISR Z-fake γ

6. 6 WW → l  l  selection P t (l + l - ) WW ZZ ZZ Zjet P t (l + l - ) cut to reject fate events Vector∑E t (jet) tt WW Vector∑E t (jet) to reject jet Straight-cut, and BDT analyses Cut-based analysis: S/B= 103/17 for 1 fb -1 (significance: 25) BDT analysis: S/B= 571/159 for 1 fb -1 (significance: 45) BDT analysis based on 15 variables WWZ WW 

7. Diboson modeSignalBackgroundS/√BAnalysis W   e  3780±611186±34110 BDT ( ε=50%) W    5864±771752±42140 BDT ( ε=50%) Z   e + e -  289±17 160±13 23 BDT ( ε=60%) Z    +  -  650±25 343±19 35 BDT ( ε=60%) W + W -  e + e - 72.6 36.2 12BDT>220 W + W -   +  - 90.0 20.1 20BDT>290 W + W -  e +  - 406±3.4 103±19 40BDT >220 W Z  l l + l - 153±1.7 16±2.5 38BDT >200 ZZ  4 l 11.0 2.2 7.6Straight cuts ZZ  l + l - 10.2 5.2 4.5Straight cuts di-Boson Sensitivity with 1 fb -1 7 Di-Boson observation will be established with ~ 100 pb -1 for Wγ, and up to ~ 1 fb -1 for ZZ

8. 8 Systematic Uncertainties  Signal systematics ~9% Luminosity measurement 6% PDF assumption 3% NLO scaling 5% Particle ID 3%  Background systematics ~18% ( in addition to the above) MC sample statistics 15% (may drop to 10%) Calibration on lepton, jet energy 5%

9. 9 Estimate Sensitivity to Anomalous Couplings non-SM couplings BHO NLO used to produce diboson events with non-SM couplings d  (AC)/d  (SM) CSC Diboson event P T (Z) or M T (VV) distributons reweighted by d  (AC)/d  (SM) Binned maximum likelihood fit on M T or P T for TGC sensitivities (95% CL interval, or contour) 1 fb -1 M T (WZ) background Binned likelihood Binned likelihood applied on the M T (WZ) of CSC selected P T (Z) (GeV) Anomalous coupling investigated for Charged TGC vertices : WWγ, WWZ using Wγ, WW, WZ events Neutral TGC vertecis : ZZZ, ZZγ using ZZ events  AC signature is in high P T, M T tails

10. 10 WW  anomalous coupling 95% Confidence interval W(eν) +ISR γ W(μν) +ISR γ λ  Δκ  λ  Δκ  1 fb -1 [-0.13,0.07] [-0.55,0.31] [-0.13,0.04] [-0.51,0.29] 10 fb -1 [-0.04,0.04] [-0.34,0.15] [-0.05,0.01] [-0.28,0.02] Wγ → l ± ν γ analysis Binned likelihood on P T (γ) 95% CL interval for TGC Δκ γ and λ γ W(eν)+ISRγ Tevatron WWγ (D0): -0.20 <λ γ < 0.20, -0.88 < Δκ γ < 0.96

11. WWZ anomalous coupling WZ → l ± ν l + l - analysis WWZ contribution has three parameters Binned likelihood on M T (WZ), 95% CL intervals 95% CL contour of Z vs  Z =  g 1 Z  Z Z  g 1 Z 11 Tevatron WWZ (CDF): -0.82 < Δκ Z < 1.27 -0.13 <λ Z < 0.14

12. 12 TGC sensitivities with WW analysis HISZ assumption: (2 free parameters)  g     / (c-s)   = 2  Z c/(c-s) Z =  Assume Z,γ universality: (3 free parameters)  Z =   and Z =  0.1/fb 1/fb 10/fb 30/fb WW has coupling of WWZ and WWγ Tevatron (D0) WWZ=WWγ: -0.36 < Δκ< 0.33 -0.31<λ< 0.33

13. 13 ZZZ, ZZγ anomalous coupling ZZ→ l + l - νν analysis ZZZ, ZZγ are SM forbidden AC contribution lead by f 4 Z Lumi / fb -1 95% C.L. 10.023 100.011 300.0088 LEP: |f 4 Z | < 0.3 Binned likelihood on p T (Z)

14. 14 Summary  Physicists from ten institutes contributed to Di-Boson Physics CSC note. First draft is ready.  Analysis tools, such as BDT, are developed and tested in our studies.  With fully simulated MC events (both signal and background) we show that ATLAS will establish the WW, WZ, Wγ and Zγ signals with significance better than 5 with the first 100 pb -1 data. ZZ signal will be established with the first 1 fb -1 integrated luminosity.  Cross-section measurements, with 5-10 fb -1 integrated luminosity, the systematic errors will be the dominant uncertaintites.  Charged TGC sensitivity will be significantly improved with 100 pb -1 data to the Tevatron limits  Neutral TGC sensitivity will be much tight compared to the limit from LEP and Tevatron for 1 fb -1 data.

15. 15 Additional Slides

16. Lepton Reconstruction Efficiency Leptons of W, Z decay Reconstruction efficiency of oIsEM 0x7FF with a track matching MuID P T distributions η distributions 16

17. 17 Tevatron di-Boson update Cross sectionSignal / BkgdMeasurements CDF, 0.2 fb -1, W  ( l   ) 323 / 114  21  =18.1  3.1  1.2 pb (  SM =19.3  1.4 pb) D0, 1.0 fb -1, Z  ( l + l -  ) 968 / 117  12  =4.96  0.30  0.30 pb (  SM =4.74  0.22 pb) CDF, 0.83 fb -1, WW(ee, ,e  ) 96 / 38±5  =13.6  2.3  1.6  1.2 pb (  SM =12.4  0.8 pb) CDF, 1.9 fb -1, WZ( l , l + l - ) 25 / 5  0.5  =4.3  1.3  0.2  0.3 pb (  SM =3.7  0.3 pb) D0, 1.0 fb -1, WZ( l , l + l - )  =2.7+1.7-1.3 pb CDF, 1.5 fb -1, ZZ( l + l -, l + l - ) 1 / 0.03 Duke Lepton, photon efficiency by Z events background by QCD jets Theory by Pythia (LO), Baur (LO, NLO, k-factor) TGC95% CL Comment D0, 0.16 fb -1, WW  from W  -0.88 <   < 0.96 -0.20 <  < 0.20 D0, 0.25 fb -1, WW  =WWZ from WW-0.36 <  < 0.33 -0.31 < < 0.33 D0, 1.0 fb -1, WWZ from WZ -0.12 <  Z < 0.29 -0.17 < Z < 0.21 (  Z =  g 1 Z ) CDF, 1.9 fb -1, WWZ from WZ -0.82 <  Z < 1.27 -0.13 < Z < 0.14

18. 18 Boosted Decision Tree method Choose many event histogram to differentiate signal/noise One cut on each histogram separating signal/noise Train 1000 trees, score each event correctly counted in trees Boosted training to optimize correctness Advantage: 1. cuts are parallel thus maximize yield 2. final cut on Score thus retain study statistics Michigan

19. 19 Binned likelihood for Anomalous Coupling Binned likelihood for Anomalous contribution References are the SM selected of CSC data bin contents: f s ν s, f b ν b, for total n events Gaussian convolution to signal, background for measurement errors 1. Mock data spectrum: CSC final selection multiplied by Poisson fluctuation 2. Theory prediction: CSC final selection reweighted with by σ(g 1,Δκ,λ)/σ(SM) Scale to 1, 10, 30 fb-1 expectation