1. Search for 7-prong  Decays Ruben Ter-Antonyan on behalf of the BaBar Collaboration Tau04 Workshop, Sep 14, 2004, Nara, Japan Outline:  Introduction  Event Selection  Data - Monte Carlo Comparison  Background Estimate  Systematic Uncertainties  Preliminary Results

2. 1.5 T Solenoid Electromagnetic Calorimeter (EMC) Detector of Internally Recflected Cherenkov Light (DIRC) Instrumented Flux Return (IFR) Silicon Vertex Tracker (SVT) Drift Chamber (DCH) BaBar and  Physics BaBar is a great place for  physics    (e + e -   +  - ) = 0.89 nb at 10.58 GeV  Recorded luminosity: 244 fb -1  220 million  pairs!  Analyzed luminosity: 124.3 fb -1  110 million  pairs e - (9 GeV) e + (3.1 GeV) PEP-II Delivered 253 fb -1 BaBar Recorded 244 fb -1

3. 7-prong  decays MC 1-7 event 1-prong side 7-prong side  Experiment: BR(     )  < 2.4  10 -6 (CLEO, 1997, PRD 56, 5297)  Theory: BR(     ) < 6  10 -11 (assuming no substructure) (S. Nussinov, M. Purohit, 2002, PRD 65)  Motivation:  With 25 times CLEO’s statistics we hope for a first observation  More stringent bound on the  neutrino mass if the decay is observed  Search for possible substructure in decay products. Very rare – no observation to date. e-e- e+e+  tag  rec

4. MC Studies of Signal and Background Signal 7  (  0 )  :  generated using phase space Background:  generic  : -- generated using TAUOLA  biggest contribution from 5  0  mode (  - conversions)  hadronic: uds, cc, bb -- continuum qq simulated with JETSET  Bhabha,  -pair, 2-photon: negligible Signal region B A B AR preliminary Mass (GeV/c 2 ) Analysis proceeds “blinded”: events below 2 GeV/c 2 are removed from the data.

5. Pseudo-Mass Pseudo-mass was introduced by ARGUS in 1992 to measure the  lepton mass.  Assume neutrino is mass-less and takes zero energy   direction is approximated by 7 ch. tracks  m*  2 =2(E beam – E 7  )(E 7  – P 7  )+m 7  2 MC 7-prong Invariant and Pseudo-Mass Advantage of pseudo-mass:  Sharp cut-off at the  mass (1.777 GeV/c 2 ).  significant improvement of signal-background separation B A B AR preliminary BR=2.4×10 -6 B A B AR preliminary Invariant Mass (GeV/c 2 )Pseudo-Mass (GeV/c 2 ) B A B AR preliminary Mass (GeV/c 2 ) Events / 0.01 GeV/c 2 Events / 0.005 GeV/c 2 All plots on this slide show Monte Carlo simulated events

6. 1-prong tags: electron ID + 0 or 1  muon ID + 0 or 1   , 0  h, 0  Pre-Selection: Up to 10 charged tracks and 12 neutrals in event Thrust magnitude > 0.90 Reject  -conversions Select 8 “good” tracks in event:  distance of closest approach to the beam spot in XY-plane DOCA XY < 1.5 cm  distance of closest approach to the beam spot in Z-plane DOCA Z < 10 cm  5 tracks with ≥12 drift chamber hits and transverse momentum p T >100 MeV/c Topology cut: event is divided into two hemispheres perpendicular to thrust axis with 1 “good” track recoiling against 7 “good” tracks and zero net charge Event Selection Event and 7-prong cuts: Thrust magnitude > 0.93 Particle ID for  -mesons p T >100 MeV/c DOCA XY / p T < 0.7cm  c/GeV 1.3 < Pseudo-Mass (7-prong) < 1.8 GeV/c 2

7. Data-MC comparison  Both data and MC have smooth pseudo-mass distributions  Both can be fitted with a Gaussian function MC simulated qq events will be used as a check of bkg. estimate method. Pseudo-Mass (GeV/c 2 ) B A B AR preliminary MC qq is scaled to data qq above 2 GeV/c 2. Background from  events is small and is determined from MC. signal region  Quantitative disagreement between data and MC throughout the analysis  Data after all cuts contain 5 times larger sample of qq events than MC simulation predicts MC simulated qq events will not be used for bkg. estimate in data. Events / 0.025 GeV/c 2 Data above 2 GeV/c 2 will be used to estimate qq bkg. in signal region

8. Background Estimate Scenario Fit from 2 to 2.5 GeV/c 2 after thrust cut with a Gaussian function Extrapolate the fit below 2 GeV/c 2 Integrate from 1.3 to 1.8 GeV/c 2 Use these fit parameters on the pseudo-mass spectrum after all cuts. extrapolate integrate fit  Pseudo-Mass (GeV/c 2 ) DATA After thrust cutDATA After all cuts Mean and sigma do not vary significantly after thrust cut. B A B AR preliminary Events / 0.025 GeV/c 2 thrust cut Sigma Mean Cuts B A B AR preliminary

9. Background Estimate Validation: MC Pseudo-Mass (GeV/c 2 ) MC Hadronic Bkg. (75 fb -1 )  Pseudo-mass is fitted after thrust cut and fit parameters are used for bkg. estimate after each cut.  Good agreement between expected and observed number of bkg. events throughout the cuts.  After all cuts (1.3-1.8 GeV/c 2 ): -- expected: 1.8 ± 0.7 -- observed: 1 Pseudo-Mass (GeV/c 2 ) B A B AR preliminary Events / 0.025 GeV/c 2 Pre-selectionThrust cut 1-prong tagsDOCA XY / P T cut P T cut 7-prong  ID

10. Background Estimate Validation: 1-8 data Pseudo-Mass (GeV/c 2 ) 1-8 Data after thrust cut1-8 Data after all cuts B A B AR preliminary 1-8 Topology Data. (91 fb -1 )  Pure hadronic bkg.  Good agreement between expected and observed number of bkg. events in the signal region throughout the cuts. Events / 0.025 GeV/c 2 CutsExpected bkg.Observed evt. Thrust mag.41 ± 1057 7-prong  ID 29 ± 732 pTpT 19 ± 522 DOCA XY /p T 7.7 ± 2.38 1-prong tag2.0 ± 0.61 B A B A R p r e l i m i n a r y Events / 0.025 GeV/c 2 Pseudo-Mass (GeV/c 2 ) B A B AR preliminary

11. Preliminary Results Events in signal region -- expected bkg.: 11.9 ± 2.2 -- observed: 7 Pseudo-Mass (GeV/c 2 ) No evidence for signal ! After all cutsAfter thrust cut B A B AR preliminary B A B AR preliminary signal region Signal efficiency: -- 7   mode: 8.05% -- 7  0   mode: 8.04% extrapolation of fit Events / 0.025 GeV/c 2

12. Systematic Uncertainties Signal Efficiency (both modes have equivalent uncertainties) Tracking efficiency 5.2 % Particle ID 2.7 % 1-prong generic  BR 0.5 % Limited MC statistics 2.6 % Luminosity and     cross-section 2.3 %  background Limited  MC statistics (3 events out of 621 fb -1 ) 58 % 5  0  branching ratio 15 % qq background Fit parameters (%)18 % Fit range (%) 3 % Num. events fitted (%) 4 % Total uncertainty of signal efficiency (%) 6.8 % Total uncertainty of  background (%) 60% Total uncertainty of qq background (%) 19% B A B A R p r e l i m i n a r y

13. Preliminary Upper Limit   1.1 × 10 8     background0.6 ± 0.4 qq background11.3 ± 2.2 Total expected background11.9 ± 2.2    4   3  +  efficiency(8.05 ± 0.55) %    4   3  +  0  efficiency(8.04 ± 0.55) % BR (    4   3  + (  0 )  ) @ 90% CL < 2.7 × 10 -7 Experiment CLEO (1997) BaBar Luminosity (fb -1 ) 4.6 124.3 Observed (predicted) events 0 (2.8) 7 (11.9) BR (    4   3  + (  0 )  ) @ 90% CL < 2.4 × 10 -6 < 2.7 × 10 -7 B A B A R p r e l i m i n a r y using most conservative Bayesian approach

14. Summary  Pseudo-mass is a powerful tool for reducing qq background in the signal region  Hadronic background estimate completely done from data  No evidence for    4   3  + (  0 )  found; BR upper limit is 10 times better than previously set  Will finalize the analysis with doubled statistics soon.

15. Backup Slides

16. 1-7 Topology Event A typical example of a MC simulated 1-7 event: on the left plot 8 tracks are counted, but the right plot shows where the 1 additional track comes from.

17. Looper and Photon Conversion Rejection Looper candidate:  A pair of tracks with SVT hits  p T,LAB < 200 MeV/c for each track  |cos  LAB | < 0.18 for each track  |  p T,LAB | < 100 MeV/c Remove tracks with largest DOCA Z Photon Conversion candidate:  A pair of tracks with invariant mass < 5 MeV  Distance between tracks in XY-plane < 0.2 cm

18. Efficiency of the Cuts Cuts 77 7     bkg. uds cc bb Pre-selection (%)23.6%22.8%0.0006%0.01%0.006%0.0001% Pre-selection (#events)23.6%22.8% 62826093 9786 152 7-prong cuts13.4%12.8% 5.1 725 99 2.8 1-prong tags8.6%8.4% 2.3 143 13 0 Events in signal region 8.1% 8.3% 0 0 1.6 0  After pre-selection background is always dominated by qq events.  7-prong cuts suppress the background from generic  events.  Background from qq is suppressed after tagging the 1-prong and the pseudo-mass cut. B A B AR preliminary

19. Data-MC Comparison 68 3.8 0.59 0.37 0.03 85  Quantitative data-MC disagreement increasing with cuts for multi-prong events.  Domination of qq bkg. in multi- prong events, resulting in worse data- MC agreement. MC simulation of qq in 1-7 topology does not agree with data.  MC simulation of  events is reliable for an estimate. Data/MC ratio for various topologies B A B AR preliminary Numbers show MC simulated qq/  ratio for different topologies.

20. Background Estimate Validation  1-7 MC: expected and observed qq bkg. in the region (1.3-1.8) GeV/c   1-8 Data: expected and observed events in the region (1.3-2.0) GeV/c 2  1-7 Data: expected background (  and qq) in the region (1.3-1.8) GeV/c 2 --  bkg. is estimated using Monte Carlo simulation -- qq bkg. is estimated from the fits Cuts 1-7 MC 1-8 Data 1-7 Data exp.obs. exp.obs. exp. obs. Thrust mag. 89 ± 34 95 41 ± 10 57 257 ± 36 298 7-prong  ID 33 ± 11 29 29 ± 7 32 108 ± 18 98 7-prong pT 22 ± 8 23 19 ± 5 22 83 ± 14 79 DOCA XY /pT 10 ± 4 15 7.7 ± 2.3 8 47 ± 9 40 1-prong tag 1.8 ± 0.7 1 2.0 ± 0.6 1 11.9 ± 2.2 7 The agreement is quite good! B A B A R p r e l i m i n a r y

21. Upper Limit Calculation with Errors To obtain the BR upper limit calculation incorporating uncertainties, we integrate the Likelihood function of the experiment: n – number of events observed, sampled from Poisson,  = = f B + b b - number of bkg. expected, b* sample from normal N(b,  b ) f = 2 N  , f* sample from normal N(f,  f )