Belle B τν 1
KEKB collider and Belle detector Located(at(KEK(in(Tsukuba. e + e Υ(S) BB Resistive plate chamber + Fe (detection of μ and K L ) CsI(Tl) calorimeter (detection of e ± and γ) e! e! Four layers of doublesided-silicon-strip (vertexing) Drift chamber (tracking and de/dx) Time-of-flight counter" (K/π separation)" Aerogel Cherenkov counter" (K/π separation)
Introduction for B τν In the SM, annihilation process mediated by W ±. Branching fraction proportional to f B V ub. f B : B meson decay constant. (190 ± 13) MeV from HPQCD. V ub : CKM matrix element. (.15 ± 0.9) x 10 3 from PDG. Expected branching fraction = (1.10 ± 0.30) x 10. PRD80, 01503 (009) From b ulν transitions. 3
Effect of charged Higgs for B τν Branching fraction of B τν could be affected by charged Higgs. H An example of the modifications is: where B(B )=B(B ) SM r H r H = 1 m B m H tan Type II of two Higgs doublet model, W. S. Hou, PRD8, 3 (1993)
Methods for analyzing B τν Exploit that a B meson pair is generated by e + e Υ(S) BB. K π D (*) B Υ(S) B ν s τ μ, e, π, etc. ν Tag π s, etc. Signal Two independent tags are used. Hadronic tag: tag B in hadoronic decays B D (*) π, etc. Semileptonic tag: tag B in semileptonic decays B D (*) lν. Signal extraction using extra energy in electromagnetic calorimeter, which corresponds to detected energy for neutrinos (~0 for signal). 5
First evidence for B τν First evidence for B τν signal obtained by Belle using hadronic tag for 9M BB data (3.5σ). 60% of full data B =[1.79 +0.56 0.9 (stat)+0.6 0.51 (syst)] 10 τ decay Syst. from BG PDF, tag efficiency, etc. signal background Fitted by smooth PDFs. PRL 97, 5180 (006) 6
B τν by semileptonic tag Using 657 M BB (85% of full data). Evidence of signal (3.6σ). Precision better than hadronic-tag result. signal B =[1.5 +0.38 0.37 (stat)+0.9 0.31 (syst)] 10 Syst. from BG PDF, tag efficiency, etc. background Fitted by histogram PDFs. PRD 8, 071101(R) (010) 7
Summary for B τν as of winter 01 Belle, hadronic tag, 9M BB BaBar, hadronic tag, 68M BB Belle, semileptonic tag, 657M BB BaBar, semileptonic tag, 59M BB B =[1.79 +0.56 0.9 (stat)+0.6 0.51 (syst)] 10 B =[1.80 +0.57 0.5 (stat) ± 0.6(syst)] 10 B =[1.5 +0.38 0.37 (stat)+0.9 0.31 (syst)] 10 B =[1.7 ± 0.8(stat) ± 0.(syst)] 10 0 0.5 1 1.5.5 3 HFAG average: B =(1.67 ± 0.30) 10 BaBar also obtained results for hadronic and semileptonic tags. The results are in good agreement while all results are slightly higher than a SM expectation: (1.10 ± 0.30) x 10. 8
Tension with CKM-fit prediction V ub can be obtained from a global CKM fit. 0.7 0.6 0.5 0. 0.3 0. excluded area has CL > 0.95 sin 1 K 3 m d m d & m s K V ub sol. w/ cos < 0 1 (excl. at CL > 0.95) 0.1 V ub SL 1 3 0.0-0. -0. 0.0 0. 0. 0.6 0.8 1.0 CKM f i t t e r Winter 1 Using this V ub, we obtain a tension with a significance of.8σ. p-value 1.0 0.8 0.6 0. 0. CKM CKM fit w/o BR(B ) f i t t e r Winter 1 Measurements (WA) Discrepancy by.8σ. BR(B ) 0.30 0.5 0.0 0.15 0.10 0.05-3 10 Relation with sin Φ 1. Direct Meas. (±1σ) CKM f i t t e r Winter 1 CKM fit. p-value 1.0 0.9 0.8 0.7 0.6 0.5 0. 0.3 0. 0.1 0.0 0.6 0.8 1.0 1. 1. 1.6 1.8.0.. BR(B ) x 10 0.00 0.5 0.6 0.7 0.8 0.9 1.0 sin 1 0.0 9
Data sizes used for B τν Semileptonic tag (~85% of full data) Hadronic tag (~60% of full data) Hadronic and semileptonic (full data) 10
Data sizes used for B τν Update on hadronic tag (full data), ICHEP01 Semileptonic tag (~85% of full data) Hadronic tag (~60% of full data) Hadronic and semileptonic (full data) From next page, will explain about recent update on hadronic-tag analysis. 11
Improvements for new hadronic-tag analysis at Belle Data size x 1.7. Improved hadronic tag: efficiency x ~. Improved signal extraction: sensitivity x ~5%.... Expected sensitivity: x ~. 1
Improved hadronic tag More decay modes. NIMA 65, 3 (011) Event selection by using NeuroBayes (neural network). New Classical M bc : mass of tagged B obtained using e + e energy. Efficiency improved by a factor of ~ (at ~same purity). 13
Improvement for signal extraction Previous analyses (including BaBar) used single variable E ECL for signal extraction. Belle, PRD8. signal This analysis uses two variables E ECL and M miss for the signal extraction. Arbitrary MC background E ECL : extra energy detected at ECL after removing all detected particles ( detected energy of neutrinos). 0 5 10 15 0 5 30 (GeV /c ) M miss M miss : missing mass squared in an event (mass squared for neutrinos). 1
Improvement for signal extraction Previous analyses (including BaBar) used single variable E ECL for signal extraction. This analysis uses two variables E ECL and M miss for the signal extraction. Belle, PRD8. signal Arbitrary MC Improvement on statistical sensitivity: ~0%. Method less sensitive to peaking BGs in E ECL. background E ECL : extra energy detected at ECL after removing all detected particles ( detected energy of neutrinos). 0 5 10 15 0 5 30 (GeV /c ) M miss M miss : missing mass squared in an event (mass squared for neutrinos). 15
Another improvement: K L veto If a K L exists, we reject the events. Efficiency difference in data and MC calibrated by D 0 ΦK S, Φ K S K L (normalized by Φ K + K ). Validity checked using B 0 D * π +, D * Dπ, D K L π 0. Check done also for B τν BG in E ECL sideband data. w/ K L veto, w/o K L veto efficiency correction w/ K L veto, w/ K L veto efficiency correction 16
Another improvement: K L veto If a K L exists, we reject the events. Efficiency difference in data and MC calibrated by D 0 ΦK S, Φ K S K L (normalized by Φ K + K ). Validity checked using B 0 D * π +, D * Dπ, D K L π 0. Check done also for B τν BG in E ECL sideband data. Improvement on statistical sensitivity: ~5%. w/ K L veto, w/o K L veto efficiency correction w/ K L veto, w/ K L veto efficiency correction Method less sensitive to peaking BGs including K L. 17
Efficiency and expected signal yield Efficiencies and expected signal yields are listed depending on signal τ decays following to B τν. Preliminary Cross feeds taken as signal (ratios of efficiencies fixed in the fit). Expected signal yields obtained for BR(B τν) = 1.65 x 10. Corresponding expected statistical significance is 6.7σ! 18
Box opened. Preliminary Events / 0.05 GeV 10 100 80 60 0 0 background signal 0 0 0. 0. 0.6 0.8 1 1. (GeV) E ECL E ECL in all M miss region. ) /c Events / 1 (GeV 5 0 15 10 5 πν,ρν 0 0 5 10 15 0 5 30 (GeV /c ) M miss lνν M miss in E ECL < 0. GeV. Signal yield: 6.3 +3.1 1.7. BR(B τν)= [0.7 +0.7 0.5] 10. 3.σ (stat only) (Only statistical errors are shown.) 19
Systematic uncertainties Preliminary Syst. for signal yield Syst. for branching fraction Significance for signal yield: 3.0σ (including syst) 0
Comparison of the results for B τν Belle, hadronic tag, 77M BB B =[0.7 +0.7 0.5 Preliminary ± 0.11] 10 WA as of winter 01 0 0.5 1 1.5.5 3 BR(B τν) [ 10 ] Belle, hadronic tag, 9M BB BaBar, hadronic tag, 68M BB Belle, semileptonic tag, 657M BB BaBar, semileptonic tag, 59M BB New result lower than the previous results. Preliminary Combining the results for hadronic and semileptonic tags of Belle, we obtain B(B τν) = (0.96±0.6)x10 -. 1
Comparison of the result with CKM-fit prediction p-value 1.0 0.8 0.6 0. 0. 0.0 CKM CKM fit w/o BR(B ) f i t t e r Winter 1 Measurements (WA) Discrepancy by.8σ up to winter 01. Belle, ICHEP1 preliminary 0.6 0.8 1.0 1. 1. 1.6 1.8.0.. BR(B ) x 10 BR(B ) 0.30 0.5 0.0 0.15 0.10 0.05-3 10 Direct Meas. (±1σ) up to winter 01 CKM f i t t e r Winter 1 0.00 0.5 0.6 0.7 0.8 0.9 1.0 sin 1 CKM fit Belle, ICHEP1 preliminary p-value 1.0 0.9 0.8 0.7 0.6 0.5 0. 0.3 0. 0.1 0.0 New result is in good agreement with CKM-fit prediction.
Constraint on Type II HDM From B(B τν) = (0.96±0.6)x10 -, we constrain r H for Type II of HDM. 3 Preliminary Excluded (95% C.L.) B(B )=B(B ) SM r H r H r H = 1 m B m H tan 1 preferred preferred 0 0 0.1 0. 0.3 tan / m (GeV H ± -1 c ) Do not agree with recent results on B D (*) τν from BaBar... arxiv:105.5 preferred Type II disfavored...? Would need more statistics. preferred 3
Summary Recent update on B τν at Belle. Sensitivity x ~ by data increase, improved hadronic tag, improved signal extraction, etc. Result consistent with SM expectations. Relation between B τν and B D (*) τν: interesting topic at super B factories. Note: error still dominated by the statistical error. (Also most of the systematics related to data size.)
Backup slides 5
Event selection Tag side Signal side ν s μ, e, π, ππ 0 K π D (*) B ν B τ ( signal modes) ππ 0 mode in ρ mass region (±0.15 GeV) π s, etc. Charged particle: π 0 : e ±, μ ±, or π ± with Δr < 0.5 cm and Δz < 3 cm. Hadronic tag: 0.08 GeV < ΔE < 0.06 GeV 5.7 GeV/c < M bc < 5.9 GeV/c Tag quality (neural net) > 0.03 0.1178 GeV/c < M π < 0.150 GeV/c E γ, forward > 0.05-0.15 GeV depending on angle. No extra tracks in Δr < 15 cm and Δz < 75 cm. No extra π 0 s. No K L. 0.86 < cosine of missing 3-momentum in e + e frame < 0.95. Signal extraction by extra energy E ECL and M miss. 6
Tag efficiency correction Efficiency correction by fitting M bc for E ECL sideband data. Validity of tag efficiency correction using B D (*) lν control sample. ) 50 00 Events / ( 0.001 GeV/c 00 150 100 50 wrong tag correct tag Number of events 150 100 50 MC data 0 5. 5.5 5.6 5.7 5.8 5.9 (GeV/c ) M bc 0 0 0.5 1 E ECL (GeV) After applying selection for the signal side. PDG: 7
K L veto Background rejection using K L is introduced. Effective to reduce peaking backgrounds. Improves the statistical significance about 5%. B 0 -tagged total without reconstructed K L with reconstructed K L 8
Signal PDFs for E ECL and M miss Signal PDF for E ECL is calibrated using D * lν control sample. Signal PDF for M miss is affected by momentum resolutions. Since M miss for B τν has wide distribution, do not apply correction. Number of events 160 10 10 100 80 60 0 0 D * lν data D * lν MC Number of events 80 70 60 50 0 30 0 10 D * lν data D * lν MC 0 0 0.5 1 E ECL (GeV) E ECL (GeV) 0-0. -0. 0 0. 0. MM (GeV /c ) M miss (GeV /c ) 9
Peaking backgrounds At least one of E ECL and M miss distributions have difference from signal. Result is less sensitive to peaking backgrounds. If BR is known, error of BR and MC statistics in Syst. MC:$data$x$10$ $$$$$$$$$for$b c. MC:$data$x$0$ $$$$$$$$$for$b u. D (*) lν E ECL *******M miss π 0 lν E ECL (((((((M miss If BR is not known, assume SM value in the nominal fit. SM value ±50% and MC statistics in Syst. MC:$data$x$50$ for$rare$modes. MC:$data$x$50$ for$rare$modes. lνγ E ECL '''''''M miss X s νν E ECL '''''''M miss 30
In the nominal fit, ratios for different τ modes are fixed. Here we test a fit by floating yields for the four τ modes. Events / 0.05 GeV /c Events / 1 (GeV 5 0 35 30 5 0 15 10 5 0 0 0. 0. 0.6 0.8 1 1. (GeV) 10 8 6 E ECL 0 0 5 10 15 0 5 30 (GeV /c ) M miss Events / 0.05 GeV /c Events / 1 (GeV 0 35 30 5 0 15 10 5 0 0 0. 0. 0.6 0.8 1 1. (GeV) 1 10 8 6 E ECL 0 0 5 10 15 0 5 30 (GeV /c ) M miss Events / 0.05 GeV /c Events / 1 (GeV 50 0 30 0 10 0 0 0. 0. 0.6 0.8 1 1. (GeV) 16 1 1 10 8 6 E ECL 0 0 5 10 15 0 5 30 (GeV /c ) M miss Events / 0.05 GeV /c Events / 1 (GeV 5 0 15 10 5 ) ) ) ) 0 0 0. 0. 0.6 0.8 1 1. (GeV) 10 8 6 E ECL 0 0 5 10 15 0 5 30 (GeV /c ) M miss Belle Preliminary (stat. errors only) ICHEP 01 τ ρν +0.7 0.5-0.6 τ πν 0.57 +0.70-0.59 τ µνν +0.63 1.06-0.58 τ eνν 0.68 +0.9-0.1 Combined Fit 0.7 +0.7-0.5 0 1 3 B(B τ ν) 10 Consistent results obtained. 31
Comparison with the previous hadronic-tag result New$analysis$based$on$improved$tag,$loose$ event$selec6on,$and$reprocessed$data.$ Most$of$the$data$a;er$the$selec6on$are$ independent$from$old$analysis.$ Assuming$that$all$the$events$in$old$analysis$ are$included$in$new$analysis,$the$remaining$ data$sample$in$n BB $=$.9$x$10 8 $provides$ BR$~$(0.6±0.)$x$10 L $(1.9σ$from$old$result).$ BR(B ) 0.5 10 0. 0.15 0.1 0.05-3 *conserva6ve. 0 8.9 (old).9 (new) 3.3 [ 10 BB] 3
Combining results of hadronic and semileptonic tags Simultaneously fit the events of hadronic and semileptonic tags using single floated BR(B τν). Correlated and uncorrelated Syst. taken into account. 33
B τν by hadronic tag from BaBar Using 68 M BB. Evidence of signal (3.3σ). B =[1.80 +0.57 0.5 (stat) ± 0.6(syst)] 10 Syst. from BG PDF, tag efficiency, etc. signal background Fitted by histogram PDFs. arxiv:1008.010 3
B τν by semileptonic tag from BaBar Using 59 M BB. Excess of signal (.3σ). expected signal x 10 B =[1.7 ± 0.8(stat) ± 0.(syst)] 10 Syst. from BG yield, tag efficiency, etc. Counted in signal region. (Region depends on τ modes.) PRD 81, 051101 (010) 35
Guglielmo De Nardo, ICHEP01 arxiv:1008.010: 36
B D (*) τν from BaBar arxiv:105.5 R(D (*) ): ratio btw tau and l modes. Blue: this result, red: Type II of HDM. Type II of HDM is excluded by 99.8%... 37