1. 1 Search for ZZZ Anomalous Triple Gauge Couplings at CDF Run II Matthew Norman, Shih-Chieh Hsu, Elliot Lipeles, Mark Neubauer, Frank Würthwein University of California San Diego

2. 2 ZZZ aTGC ZZ Branching Fractions 2.8% 0.48% 9.7% ● The ZZZ Anomalous Triple Gauge Couplings (aTGC) does not occur in the SM. Only the t-channel is allowed. ● aTGC allows s-channel production of ZZ at very high Z pT, allowing us to use the ZZ lljj channel beyond the reach of the Z + jets background.

3. 3 Previous Limits Previous limits on the ZZZ aTGC were established by combining results from the LEP experiments. LEP limits by experiment (above) LEP limits combined (below) For the remainder of this talk, we will assume the ZZ vertex is 0.

4. 4 Physics Strategy ● Measurement on data – Use Zll to probe Z pT – For given pT, plot dijet mass and fit for ZZ yield using templates derived from the MC. ● Use this to derive limits on yield in a Z pT range ● Does not depend on MC background normalization ● Comparison to aTGC – Similar approach as that of WZ aTGC – Demonstrate “Universal Efficiency” as a function of Z p T, independently of aTGC coupling values. – Generate spectra at generator level using Baur MC – Apply universal curve to find expected yield per bin in Z p T

5. 5 Generator MC Because Pythia does not contain the mechanisms necessary for ZZZ ATGC, we used the MC Generator created by Baur and Rainwater (hep-ph/0008063). This is an LO Generator, and does not generate the * component. Projection for Dilepton p T at the Tevatron from hep-ph/0008063 Using the Baur MC, we stepped through anomalous coupling space and generated a sample for all possible values of the anomalous couplings f 4 and f 5 in increments of 0.05.

6. 6 Baur Verification Dilepton Mass for fully-reconstructed particles, from Baur(blue) and SM PYTHIA ZZ (magenta) with a mass cut on the dilepton between 76 and 106 GeV. Baur does not produce Zs outside of those regions, but in the region it does produce, the masses are essentially identical. We have produced four fully simulated samples using four different values of aTGCs and compared them to the SM results.

7. 7 Selection and Background This analysis uses the same Diboson framework as the WZ and ZZ analyses. We select events with: ● 2 same flavor, opposite-sign leptons ● p T threshold for leptons: 20,10 ● 2 or more jets with the p T threshold for jets: 15 GeV ● Dilepton mass: [76,106] Our primary background is Z + jets. To study it, we have created a high p T Z sample (run weighted through period 11), with a generator-level filter at 100 GeV for the Z p T. We also looked at both the conventional Z + jets and SM ZZ MC: ● Zee : ze0s6d (0d), ze0sad (0h + 0i) ● Z : ze0s6m (0d), ze0s9m (0h + 0i) ● Z : ze0s8t (0d) ● ZZ : we0s7d (0d), we0sdd (0h + 0i)

8. 8 Dataset

9. 9 Projected Yields Projected yield for Drell-Yan background, SM ZZ, and aTGC ZZ for f 4 = 0.3 per fb -1 as a function of dilepton p T

10. 10 LO vs. NLO Baur is strictly a LO generator. Hence we have no information about NLO cross-sections. This prompts us to make a guess. We make the assumption that the NLO/LO k-factor is approximately the same between SM ZZ and aTGC ZZ. We can generate the NLO/LO k-factor via MCFM. First we use MCFM and Baur to both generate the LO cross-section for SM ZZ, which match to ~0.5%. Then we use MCFM to generate the NLO cross-section, and multiply all Baur results by the MCFM NLO/LO ratio. End result: ATGC (ZZlljj) = 79 fb with k-factor 1.315 We hope to eventually create a template for NLO/LO as a function of dilepton p T. We're also looking into the possibility of an NLO Baur generator.

11. 11 Signal Templates Strategy: Split into three bins in dilepton p T, a background bin for comparison purposes, a bin where signal and background are mixed, and a signal dominated bin. Bin 1: Dilepton p T 0-140 GeV Bin 2: Dilepton p T 140-210 GeV Bin 3: Dilepton p T 210+ GeV Dijet Mass for bin 2 and bin 3 (mixed, and signal dominated) from left to right for f 4 = 0.3.

12. 12 Background Templates Background Fits in the two signal bins.

13. 13 Signal vs. Background We expect background to be dominated by Z Drell-Yan. Signal vs. Drell-Yan background for dielectron channel only in the two signal bins.

14. 14 Universal Efficiency Curve Efficiency curve as a function of dilepton pT for several different values of aTGC

15. 15 Comparison between aTGCs We can only use our planned approach if the shape of the dijet mass distribution of aTGCs is independent of the value of the aTGC. Fortunately, this appears to be the case: Dijet Mass distribution in signal bins for f 4 = 0.3 (blue) and f 5 = 0.3 (green)

16. 16 Likelihood Fitting Comparing Likelihood ratios for a sample with both signal and background, and one with background only for f 4 = 0.3. To determine whether an event is aTGC or background we calculate a likelihood ratio for each event. Calculate the likelihood ratio = -2Log[ ℒ (S+B)] + 2Log[ ℒ (B)] where ℒ (S+B) = Likelihood of a signal + background fit. and ℒ (B) = Likelihood of a background only fit. and Signal is aTGC ZZ. Then, in toy MC, we simulate a full dataset, and sum of for the set. From this we can assemble a distribution of values for signal+background and background only.

17. 17 Likelihood Fitting II Once we have the full Likelihood distributions: ● We can find the 95% limit on signal, where 95% of the signal has a more negative log likelihood ratio (is more likely to be signal). ● We can then calculate the probability of exclusion based upon the percentage of background events that have a more negative likelihood than this limit. ● We can generate this over the entire grid of anomalous coupling points.

18. 18 Dilepton Mass Comparison Comparison of Dilepton Mass with data for 1.1fb -1 (to be updated).

19. 19 Dilepton Mass II Dielectron mass compared to data Dimuon mass compared to data Data comparisons done for the Zll study

20. 20 Data Comparisons Comparisons to data for 1.1fb -1 (full selection applied) Dijet Mass Dilepton p T

21. 21 Systematics We still have to consider how to properly do the systematics. Systematic effects that we have to consider are: ● Jet Energy Scale ● Variance in the Universal Efficiency Curve ● Variance in the Fitter

22. 22 Summary Currently we have several pieces: ● M jj background for one half of the Drell-Yan (at ~300x data) ● M jj templates for the signal ● Generator-level plots for a grid of points in aTGC space ● A Universal Efficiency Curve We have to do the following: ● Compare our non-signal region with data (can be done for dielectron channel today) ● Finish generating the Z background sample ● Generate the expected sensitivities (code already written) ● Compute systematics ● Finish analysis