OBSERVATION OF THE RARE CHARMED B DECAY, AT THE BABAR EXPERIMENT

Transcription

1 OBSERVATION OF THE RARE CHARMED B DECAY, AT THE BABAR EXPERIMENT ON FINDING A NEEDLE IN A HAYSTACK Virginia University HEP seminar

2 SEARCH FOR THE RARE B DECAY 2 Contents BaBar and CP violation Physics goals and production mechanism Analysis Optimization of signal selection Setup of three dimensional likelihood fit Results Interpretation and outlook

3 SLAC BaBar 3 Hella Snoek Nikhef PEPII s LINAC is1.9 miles long: longest building in the world!!

4 BABAR and antimatter 4 BaBar at PEPII collider, SLAC ( 99 08) Focus on measurements of CP violation: differences between matter/antimatter Run at e + e collisions at Y(4S): clean coherent Bmeson production (On tape million BB pairs)! (e + e " Hadrons) #(2S) BB threshold #(3S) Y(4S) # Hella Mass Snoek (GeV/c Nikhef 2 ) 10.54

5 CP violation in electroweak decays (1) 5 Quark changes flavor due to weak interaction u c t d s b 3x3 flavor changing currents CabibboKobayashiMaskawa (CKM) matrix

6 CP violation in electroweak decays (1) 6 CKM Matrix must be unitary (conservation of probability): 9 equations (6 triangles) 4 free parameters, (3 magnitudes + 1 complex phase) CabibboKobayashiMaskawa (CKM) matrix

7 CP violation in electroweak decays (1) 7 CKM Matrix must be unitary (conservation of probability): 9 equations (6 triangles) 4 free parameters, (3 magnitudes + 1 complex phase) CabibboKobayashiMaskawa (CKM) matrix relative magnitudes relative phases

8 CP violation in electroweak decays (2) 8 Im Unitarity Triangle CKM triangle γ α β Re CabibboKobayashiMaskawa (CKM) matrix

9 CP violation in electroweak decays (2) 9 Im Unitarity Triangle CKM triangle γ α CP β Re CabibboKobayashiMaskawa (CKM) matrix

10 CKM angles Im CKM triangle 10 γ α β Overconstraining CKM triangle powerful test of Standard Model Re

11 CKM angles Im CKM triangle 11 γ α β Overconstraining CKM triangle powerful test of Standard Model Re

12 12 CKM angles Overconstraining CKM triangle powerful test of Standard Model CKM angle γ least constrained by measurements α = 88 ± 5º, β = 22 ± 5º, γ = 77 ± 30º Measured through time dependent interference in B decays Current measurements through B 0 D* + π and B 0 D* + ρ Im B 0 D* + a 0 could be more sensitive γ α CKM triangle β Re

13 Sensitivity to CKM angle γ 1 13 Same initial and final state. Weak phase difference gives sin(2β+γ). Asymmetry amplitude to phase given by the amplitude ratio. I II Oscillation process where anti B meson changes in B meson CKM angle 2β CKM angle γ

14 Sensitivity to CKM angle γ 1 14 Same initial and final state. Weak phase difference gives sin(2β+γ). Asymmetry amplitude to phase given by the amplitude ratio. I Asym. amp. ~ II Weak decay constants (MeV): ~0.04 ~1 ~200 Asymmetry amplitude to CKM phase is large for B 0 D* + a 0! However, we also have less events because of low branching ratio. CKM angle γ

15 So how large is the branching ratio? I 15 I Using amplitude small due to Vub ~ 10 6

16 So how large is the branching ratio? II 16 I Using amplitude small due to Vub ~ 10 6 II W decay constants Using amplitude small due decay constant ~10 6

17 Using is that correct? 2 17 In short It means we use factorization principles No interactions between produced mesons XY B X Y Nonfactorizing terms (with interactions produced mesons) usually small.. B X Y. but not in case of B 0 D ( * )+ a 0 Naïve prediction: 3 x10 6 Including calculable nonfactorizing terms: 6 x10 6 Diehl, Hiller: JHEP 0106:067,2001

18 18 Recap Punch line is: We search for the B 0 D ( * )+ a 0 decay We are looking for a needle in a haystack. If we don t find it we can eliminate nonfactorizing QCD scenario s If we do we can eliminate other nonfactorizing QCD scenario s measure CKM angle γ with high precision (potentially)

19 Decay reconstruction 3 19 Lineshape width of a 0 uncertain: PDG MeV Setup selection no criteria linewidth: a priori select B 0 D ( * )+ ηπ events Use likelihood fit to discriminate resonant from nonresonant decay 6 D decay modes for statistics selection and fit separate; different S/B Nonresonant B 0 D ( * )+ ηπ decays not measured before: interesting on its own!

20 Signal selection optimization 20 Defined 30 discriminatory observables Particle properties m/p/e Decay lengths (D/K s ) Event shape Utilize angular properties in decay 27 for event selection / 3 for likelihood fit Cuts are optimized while branching ratio is unknown!! using significance: desired observation significance Simultaneous optimization searches for highest SL using rectangular box cuts signal selection efficiency background events Punzi (2003) physics/

21 Unbinned likelihood fit in 3 dimensions M ES beam energy substituted mass, signal peaks at B mass, background as Argus shape 2. ΔE energy difference, signal peaks at 0, background flat Y(4S) 3. Mηπ invariant mass, resonant signal is BreitWigner, non resonant as phase space with kinematics

22 PDF shapes signal 22 peaks at B mass Centered at 0 BreitWigner at a 0 mass M ES ΔE M ηπ

23 PDF shapes nonresonant signal 23 M ES ΔE M ηπ peaks at B mass Centered at 0 spin 0 phase space

24 PDF shapes background 24 M ES ΔE M ηπ peaking bkg not uniform combinatorics + kinematics green: peaking background blue: combinatorics green: peaking background Used as control sample!

25 PDF shapes summary 25 B 0 D + a 0 B 0 D + ηπ background Shape parameters determined on MC and Hella fixed Snoek in the Nikhef fit to data

26 Adding the different D decay modes 26 5 parameters fitted in likelihood fit for every D decay mode : 1. (# resonant signal events) 2. (# nonresonant signal events) 3. # Ds events 4. # BB background events 5. # qq background events BR(resonant) BR(nonresonant) Combine D + and D* + decay modes 1. Simultaneous fit of branching ratios of B 0 D + a 0 and B 0 D + ηπ 2. Other 3 parameters fitted individually.

27 Combined Fit Result B 0 D* + a 0 B 0 D* + ηπ B 0 D* + D s 27 Full projection

28 Combined Fit Result B 0 D* + a 0 B 0 D* + ηπ B 0 D* + D s 28 Full projection M ES signal reg

29 Combined Fit Result B 0 D* + a 0 B 0 D* + ηπ B 0 D* + D s 29 Full projection M ES signal reg M ηπ signal reg

30 Combined Fit Result B 0 D* + a 0 B 0 D* + ηπ B 0 D* + D s 30 Full projection M ES signal reg M ηπ signal reg M ES side region

31 Combined Fit Result B 0 D* + a 0 B 0 D* + ηπ B 0 D* + D s 31 Full projection M ES signal reg M ηπ signal reg M ES side region

32 Systematic uncertainties 32 Determined using: Full Monte Carlo studies Toy Monte Carlo studies Data sample itself Control sample B 0 D ( * )+ D s Largest uncertainties due to a 0 lineshape and selection validations

33 Systematic uncertainties 33 Determined using: Full Monte Carlo studies Toy Monte Carlo studies Data sample itself Control sample B 0 D ( * )+ D s Largest uncertainties due to a 0 lineshape and selection validations

34 Profile likelihood 34 D* meson modes

35 Combined Fit Result B 0 D + a 0 B 0 D + ηπ B 0 D + D s 35 Full projection

36 Combined Fit Result B 0 D + a 0 B 0 D + ηπ B 0 D + D s 36 Full projection M ES signal reg

37 Combined Fit Result B 0 D + a 0 B 0 D + ηπ B 0 D + D s 37 Full projection M ES signal reg M ηπ signal reg

38 Combined Fit Result B 0 D + a 0 B 0 D + ηπ B 0 D + D s 38 Full projection M ES signal reg M ηπ signal reg M ES side region

39 Combined Fit Result B 0 D + a 0 B 0 D + ηπ B 0 D + D s 39 Full projection M ES signal reg M ηπ signal reg M ES side region

40 Profile likelihood 40 D* meson modes D meson modes

41 Profile likelihood 41 D* meson modes D meson modes

42 Intermediate results 42 Branching ratio of B 0 D ( * )+ a 0 predicted (36) 10 6 Analysis setup o Data selection optimization using ~30 variables o Likelihood fit in 3 observables Measured branching ratios o Nonresonant B 0 D + ηπ at 4 σ (46 events) B 0 D* + ηπ at 8 σ (76 events) o Resonant B 0 D* + a 0 6.0x10 5 at 5 σ (30 events) B 0 D + a 0 > 2.3x10 5 at 90 % CL Not enough for timedependent analysis needed to measure γ (?)

43 Possible explanation higher BR 43 We measure B 0 D* + a 0 6.0x10 5 at 5 σ (30 events) Naïve factorization model predicts 3x 10 6 Including some QCD diagrams upper limit at 6x 10 6 Restrictions on both naïve factorizable diagrams strict! I II

44 Possible explanation higher BR 44 We measure B 0 D* + a 0 6.0x10 5 at 5 σ (30 events) Naïve factorization model: 3x 10 6 Adding QCD diagrams: upper limit 6x10 6 Restrictions on both naïve factorizable diagrams strict! For naïve calculation we need to know: CKM V ub V cd V cb V ud I II W decay f D fa 0 B decay F B D F B a All well known or strong upper limit Ratio I/II measured in timedependent analysis (cosine term)

45 Rescattering 45 + J P a 1 a 0 D J D* BF B 0 D + a 1 (6.0 ±3.3 ) 10 3 Possible rescattering through Da 1 D*a 0 Not calculated, comparison with Dρ D*π gives order 10 6 Kinematic arguments: easily would be at order 10 5 Same CKM phase gamma!

46 Conclusion 46 Branching ratio of B 0 D ( * )+ a 0 predicted (36) 10 6 Analysis setup o Data selection optimization using ~30 variables o Likelihood fit in 3 observables Measured branching ratios o Nonresonant B 0 D + ηπ at 4 σ (46 events) B 0 D* + ηπ at 8 σ (76 events) o Resonant B 0 D* + a 0 6.0x10 5 at 5 σ (30 events) B 0 D + a 0 > 2.3x10 5 at 90 % CL Not enough for timedependent analysis needed to measure γ (?) Time dependent analysis will give insight to high BF

47 47 Backup slides

48 CKM UT current status 48

49 Possible explanation higher BR 49 Restrictions on both naïve factorizable diagrams strict! QCD factorizable diagrams upper limit at , factor 10 too low! Possible rescattering through Da 1 D*a 0 not calculated, comparison with Dρ D*π gives order 10 6 but should be larger!! Easily would be at order I II

50 Optimization setup 50 Per cycle: Evaluates significance level in given search area with cuts placed on all other variables. New, optimized, cut is placed in direction of the highest SL. All variables are evaluated in single cycle Current New Optimal SL Search area

51 51

52 a 0 suppression mechanism 52 G parity suppressed/second class current Weak current has VA behaviour natural spinparity meson couple to V, unnatural to A G C e iπi 2 First class currents: G+ and V, or G and A Second class currents: G and V, or G+ and A a 0 doubly suppressed; by CVC and G parity violating

53 Punzi 53 S/sqrt(B) does not behave well for low numbers of S (prefers 0.1/0.01 over 10/2) S/sqrt(S+B) depends on BR!