Latest results on B (s)0 and other very rare decays A. Sarti on - PowerPoint PPT Presentation

Latest results on B (s)0 ➞ µµ and other very rare decays A. Sarti on behalf of the LHCb collaboration Dipartimento di Scienze di Base e Applicate per l’Ingegneria (Università di Roma “La Sapienza”) and LNF - INFN, Italy

A rare beauty! ➡ LHCb pursues a physics program aiming to a precise validation of SM predictions and indirect NP searches in the heavy flavor sector, which is fully complementary to direct NP searches done with GP experiments ➡ A key role is played by B decays occurring trough Δ B = Δ S = 1 transitions, which are highly suppressed in SM (FCNC) and are thus sensitive probes of NP Two distinct theoretical approaches can be followed: model-independent model-dependent Δχ 2 CMSSM including pre HCP constraint on limit use rare decays to set constraints on Wilson coefficients 03/03/13 A. Sarti Moriond EW 2013 2

LHCb: a golden mine for rare B decays ➡ Rare decays: – Radiative decays: (K* 0 훾 , φ 훾 ) [ Nucl. Phys., Sect. B 867 (2013), pp. 1-18 ] – B ➞ Xll decays: B ➞ K (*) µµ , B ➞ φ µµ [ Phys. Rev. Lett. 110, 031801 (2013), JHEP 02 (2013) 105 ]; B ➞ πµµ [ JHEP 12 (2012) 125 ] 1 fb -1 – Ongoing analyses: D 0 ➞ µµ , t ➞ 3 µ , B (s)0 ➞ e µ collected but not used yet ➡ In this presentation: – B ➞ 4 µ and K s ➞ µµ : latest results on 1 fb -1 2011 dataset – B (s)0 ➞ µµ . LHC combination (Jun ‘12): BR(B 0s ➞ µ + µ − ) < 4.2 × 10 -9 @ 95% CL. 2012 2011 Today I present here the latest result on 2.1 fb -1 2011/12 dataset (published in Jan ‘13) 1 fb -1 at 7 TeV (2011) 2 fb -1 at 8 TeV (2012): only ~50% (1.1 fb -1 ) used for published results so far 03/03/13 A. Sarti Moriond EW 2013 3

B (s)0 ➞ 4 µ LHCb-PAPER-2012-049 ➡ B ➞ 4 µ decays in SM: B (s)0 ➞ J/ ψ K* 0 – Non resonant BR(B (s)0 ➞ µ + µ - 훾 *( µ + µ - )) < 10 -10 Preliminary [D. Melikhov and N. Nikitin, Phys. Rev. D 70 (2004) 114028] – B s0 ➞ J/ ψφ (control) channel = 2.3 ± 0.9 10 -8 [ Phys. Rev. D86 (2012) 010001 ] ➡ In MSSM: sensitive to new scalar (S) and 32k candidates pseudoscalar (P) sGoldstino particles ➡ Normalization on B 0 ➞ J/ ψ ( ➞ µµ)K ∗ 0 ( ➞ K π ) ➡ Result: observed 1 event in B 0 window, Preliminary 0 in B s0 . Consistent with expected bkg. 1 fb -1 ➡ [preliminary] Limits at 95(90)% C.L.: 7 TeV – BR (B s0 ➞ 4µ) < 1.6 (1.2) ·10 − 8 data – BR (B 0 ➞ 4µ) < 6.6 (5.3) ·10 − 9 Paper in preparation First experimental limit to date 03/03/13 A. Sarti Moriond EW 2013 4

K s ➞ µµ [JHEP 01 (2013) 090] ➡ The SM prediction for BR(K s ➞ µ + µ - ) is 5.1 ± 1.5 10 -12 [NuPh B366(1991) 189; JHEP 0401 (2004) 009] . Best exp. limit (’73) < 3.2 10 -7 @ 90% CL [ PL B44 (1973) 217 ] – Comparison with K L ➞ µ + µ - can reveal effects due to new light scalars and bounds at 10 -11 level constrain CP violating phase from s ➞ d ll (E.g.: K ➞ π νν ) Limit is computed using the CL s (modified frequentist) approach [J. Phys. G28 (2002) 2693] combinatorial (exp) + K s ➞ ππ Normalization channel: K s ➞ ππ misidentified bkg from mass sidebands 1.0 fb -1 1.0 fb -1 7 TeV Expected 7 TeV data data Observed Results: BR (K S0 ➞ µ + µ − ) < 11(9) · 10 − 9 at 95(90)% C.L. 03/03/13 A. Sarti Moriond EW 2013 5

B (s)0 ➞ µµ in the Standard Model FCNC process → very small branching fraction: t =0 Buras et al., arXiv:1208.0934 t =0 The authors used f Bs = (227±8) MeV, averaging from recent lattice inputs Bazavov et al., arXiv:1112.3051 Mc Neile et al., PRD 85 (2012) 031503 Na et al., arXiv:1202.4914 To compare with experiment need a time integrated branching fraction, taking into account the finite width of the B 0s system: 〈 t 〉 t =0 y s = ΔΓ s /2 Γ s De Bruyn et al., PRL 109, 041801 (2012) uses LHCb-CONF-2012-002 03/03/13 A. Sarti Moriond EW 2013 6

B (s)0 ➞ µµ analysis Performed on full 2011 [@ 7 TeV] data (reanalyzed, with improved bkg evaluation), and 1.1 fb -1 of 2012 [@ 8 TeV] sample (~50% of available statistics) : 8 TeV data signal region kept blind until analysis completion Probability - Signal/Background separation by invariant 1 LHCb di- µ mass (IM) and a multivariate (MVA) -1 10 classifier (Boosted Decision Trees, BDT) -2 10 BDT training on MC signal and bkg BDT calibration on data: for signal used -3 10 Signal exclusive B 0(s) ➞ h + h ′ − channels (h= π , K) Background -4 10 - Normalization with B ± ➞ J/ ψ K ± & B 0 ➞ K + π − 0 0.2 0.4 0.6 0.8 1 BDT B 0s ➞ J/ ψφ was dropped for 2012 data as 3 10 × ) 2 B ± ➞ J/ ψ K ± 35 c third normalization channel, but used to Candidates / (2 MeV/ LHCb 30 check √ s dependence of f s /f d 25 - Analysis performed in 7(8) bins of BDT and 20 15 9 bins of IM for the analysis of 8(7) TeV 10 data 5 0 5200 5250 5300 5350 03/03/13 A. Sarti m [MeV/ c 2 ] Moriond EW 2013 7 J/ K ψ

Normalization Evaluated from MC, Measured Ratio of probabilities for a b quark cross-checked with data on data to hadronize to a given meson checked on 2012 data: no [ PRD85 (2012) 032008; Combined result at 7 TeV significant change ➞ used same LHCB-PAPER-2012-037 submitted to JHEP ] value for 2011 and 2012 B ± ➞ J/ ψ K ± and B 0 ➞ K + π − channels give consistent results, and are averaged 8 TeV data Assuming SM rates, after selection we expect in 7 TeV + 8 TeV data (1.0 + 1.1 fb -1 ) ~11+13 B 0s ➞ µ + µ − and ~1.3+1.5 B 0 ➞ µ + µ − in signal region ( m(B 0(s) )±60 MeV/c 2 ) 03/03/13 A. Sarti Moriond EW 2013 8

Combinatorial background The main background source in the B 0s ➞ µ + µ − signal window, m(B 0s )±60 MeV/c 2 , is combinatorial from bb ➞ µ + µ − X For the CL s computation, the expected background yield in the signal region is evaluated from a fit to the mass sidebands , for each BDT bin separately An exponential shape is used to model the combinatorial bkg BDT<0.25 For BDT values <0.5 this is by far the dominant bkg source B 0s window B 0 window 03/03/13 A. Sarti Moriond EW 2013 9

Exclusive background sources Various exclusive decays have been studied which are able to fake a signal by misID of either one or two hadrons or by two muons coming from the same vertex: B 0 ➞ π − µ + ν µ B 0s ➞ K − µ + ν µ B +(0) ➞ π +(0) µ + µ − Λ 0b ➞ pµ − ν µ B 0(s) ➞ h + h’ − B +c ➞ J/ ψ (µ + µ − )µ+ ν µ (other channels like B → (D → µX)µX, B → ττ X being negligible in [4900-6000] MeV/c 2 ...) These background sources can affect the result in two ways: 1) non negligible contribution in the signal mass window, m(B 0(s) )±60 MeV/c 2 only B 0(s) ➞ h + h’ − has to be accounted for (mainly for B 0 ): take K ➞ µ and π ➞ µ from data, fold with MC spectra. In the full BDT range, for 8 TeV data we get: Events in B 0s Events in B 0d 0.76 +0.26-0.18 4.1 +1.7-0.8 mass window mass window 2) mass shape different from exponential → bias in the combinatorial background interpolation from mass sidebands Three dominant components have been added: B 0 ➞ π − µ + ν µ B +(0) ➞ π +(0) µ + µ − B 0(s) ➞ h + h’ − 03/03/13 A. Sarti Moriond EW 2013 10

Unblinded 8 TeV data R. Aaij et al. (LHCb Collaboration) Phys. Rev. Lett. 110 , 021801 (2013) B 0s window B 0 window 03/03/13 A. Sarti Moriond EW 2013 11

B 0s ➞ µ + µ − candidate R. Aaij et al. (LHCb Collaboration) Phys. Rev. Lett. 110 , 021801 (2013) 8 TeV data p Tµ- = 4.2 GeV/c [0.2 mm ticks] τ = 2.84 ps p Tµ+ = 2.3 GeV/c p T (B)= 4.1 GeV/c B candidate: m µµ = 5353.4 MeV/c 2 BDT = 0.826 p T = 4077.4 MeV/c t = 2.84 ps [mm] 03/03/13 A. Sarti Moriond EW 2013 12

BR (B 0(s) ➞ µ + µ − ) results R. Aaij et al. (LHCb Collaboration) Phys. Rev. Lett. 110 , 021801 (2013) Use CLs method to evaluate compatibility with background only (CL b ) and signal + background hypotheses (CL s+b ); the 95% CL upper limit is defined at CL s = CL s+b /CL b =0.05 1 s CL B 0 ➞ µ + µ - : B 0s ➞ µ + µ − obs BR limit < 9.4 x 10 -10 @ 95% CL LHCb 0.8 exp BR limit < 7.1 x 10 -10 @ 95% CL 0.6 Compatibility with bkg only observed expected hypothesis: p-value = 1-CL b = 0.11 0.4 bkg+SM expected B 0s ➞ µ + µ − : 0.2 bkg bkg only p-value= 5.3x10 -4 0 2 4 6 8 (3.5 σ excess) 0 -9 - + B( B ) [10 ] → µ µ s double-sided limit: 1.1 × 10 − 9 < B(B 0s ➞ µ + µ - ) < 6.4 × 10 − 9 at 95% CL where the lower and upper limits are evaluated at CL s+b = 0.975 and CL s+b = 0.025, respectively 03/03/13 A. Sarti Moriond EW 2013 13

B 0s ➞ µ + µ − : BR fit R. Aaij et al. (LHCb Collaboration) Phys. Rev. Lett. 110 , 021801 (2013) ➡ Unbinned maximum likelihood fit to the mass spectra – 7 TeV and 8 TeV data are treated simultaneously – mass range [4900-6000] MeV/c 2 ➡ Free parameters: – BR(B 0s ➞ µ + µ - ), BR(B 0 ➞ µ + µ - ) and combinatorial background – The relative signal yield in each BDT bin is constrained to the expectation from B 0(s) ➞ h + h ′ − calibration , – The yields and pdf’s for all of the relevant exclusive backgrounds are constrained to their expectations ➡ Additional systematic studies on background composition/ parameterization: – add the B 0s ➞ K − µ + ν µ component to the exclusive background – change the combinatorial pdf from single to double exponential, to account for possible residual contributions from Λ 0b and B +c decays 03/03/13 A. Sarti Moriond EW 2013 14