1 / 94

Hadronic Moments in Semileptonic B decays Preblessing CDF 6754, 6972, 6973

Hadronic Moments in Semileptonic B decays Preblessing CDF 6754, 6972, 6973. Alex, Hung-Chung, Laurent, Marjorie, Ramon. Contents. What are these moments? Why are they interesting? Analysis strategy Selection of l  D*/l  D + samples Montecarlo validation ** selection & optimization

Download Presentation

Hadronic Moments in Semileptonic B decays Preblessing CDF 6754, 6972, 6973

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Hadronic Moments in Semileptonic B decaysPreblessingCDF 6754, 6972, 6973 Alex, Hung-Chung, Laurent, Marjorie, Ramon

  2. Contents • What are these moments? Why are they interesting? • Analysis strategy • Selection of lD*/lD+ samples • Montecarlo validation • ** selection & optimization • Raw mass distributions • Background modeling • Background Subtraction • Acceptance corrections • Final fit to m** • Extraction of QCD parameters

  3. Vcb connected to BXcl Xc=anything(c)Inclusive Xc=D0/*/+Exclusive Hadronic mass moments: Hadronic mass distribution from semi-leptonic decays: BXc l  D, D*, D** only D** component needs to be measured Spectroscopy of D mesons Introduction

  4. Inclusive Vcb Determination and hadronic moments • Inclusive semi-leptonic B decays: (BXcl) = |Vcb|2 f(,1,2,…) • Moments: g(,1,2,…) • one can measure the moments to improve the knowledge on Vcb • currently the theory uncertainties dominate • general test of non-perturbative aspects of HQET • measuring ,1 in several ways and finding consistency would be a powerful test of the OPE treatment of HQET • Experimentally: • CLEO, BABAR: inclusive technique with fully reconstructed B on the away side • DELPHI: inspired our approach

  5. Moments Definition • Spectral Moments: • lepton energy: En(d/dE)dE / (d/dE)dE • photon energy in bs • hadronic mass: dsH sHn (d/dsH) /dsH (d/dsH) where sH = mX2 • usually sH=mX2-m2Dspin (mDspin = 0.25mD+0.75m D* spin averaged mass)

  6. G G æ ö G G G ( ) ( ) 1 d ( ) ç = d - + * d - + - - * 2 2 * * SL D D s m D s m 1 D f s ç H D H H * G G G G G D ds è ø SL H SL SL SL SL Hadronic Mass • Hadronic mass spectrum: • Explicitly measure only the D** component, f**(sH), normalized to 1. Only the shape is needed. • PDG values for D and D* masses and b.r. will be inserted.

  7. Channels with charged B • B- D**0 l- • D**0 D+- OK • D**0 D00 Not reconstructed. Half the rate of D+- • D**0 D*+- • D*+ D0+OK • D*+ D+0 Not reconstructed. Feed-down to D+- • bckgd shape from channel above (D0-), rate is half • D**0 D*00 Not reconstructed. Half the rate of D*+- We can reconstruct all the Xc spectrum Neutral B would add statistics but involve neutrals

  8. Event Topology K- + • D0, D+, D*+: 3D vertex of K() • Lepton +D: 3D vertex • Additional track (**) for D** • use the track’s d0 w.r.t. the B and Primary vertices to tell ** from prompt tracks + PV D+ B- D**0l- - (aka **) l-

  9. Add another **D** Measure <m**2>, <m**4> Reconstruct D*/D+ The strategy CDF6972/6973 CDF6754 Correct for (m**), (D+)/(D*) • Selection: • Optimize on MC+WS combinations • Cross check on * • ** Background • Combinatorial • D’ • BDD • cc • … • Measure selection bias on m** from: • MC • D* candidates • Rely on MC (& PDG) for: • (D+)/(D*) • Unseen modes (Isospin) • Lepton spectrum acceptance • Collect as many modes as possible: • (K)* • (K)* • (K0)* • K • Check yields • Validate MC • Subtract backgrounds • Use PDG to go m**m** • Compute <m**2> & <m**4> • Include D(*)0 • Extract , 1 • Systematics

  10. Reconstruct D*/D+ D(*)+ Reconstruction

  11. Dataset & Initial Selection • Dataset: • Jbot2h/0i: muon + SVT • Jbot8h/4i: electron + SVT • Refit: • G3X (standard B, phantom layer), beamline 19 • ISL, L00 hits dropped • COT scaling: (curv,d0,0,,z0)=(5.33,3.01,3.7,0.58,0.653) • LeptonSvtSel: default cuts Thru run 165297

  12. Track & Vertex Cuts • TrackSelector: • COT hits: >20 Ax, >20 St • Si hits: 3 Ax • K, : pT > 0.4 GeV/c • leptons: pT > 4 GeV/c (from LeptonSvtSel) • D vertex: • 3D fit • one track has to be matched to the SVT track • Lepton+D vertex: • 3D fit • **: • 20+20 COT hits • Si hits: 3 Ax, 3 SAS+Z (-30% stat, x2 S/B) • Pt>0.4 GeV/c

  13. K K0 e   e K  e K e 

  14. MC samples and validation

  15. Montecarlo Generation • Bgenerator/EvtGen/CdfSim/TRGSim++ • “realistic simulation” with representative run number • Different samples: • MC Validation • D samples  inclusive BXcl • * tracking (** proxy)  exclusive BD*l • Optimization inclusive BlD** • Efficiency, M** bias individual D** mesons • (e.g. BD1l, D1D*, D*D0, D0K)

  16. Approach to MC validation • Cross-check kinematic variables • B spectrum modeling • Trigger emulation • Compare many data/MC distributions using binned 2 • Every possible decay mode • Sideband subtracted before comparison • Duplicate removal (D0K)

  17. Kinematic Comparisons: D*, D0K

  18. Can we “predict” relative yields? • Based on inclusive bD(*)+l • Based on exclusive BD(*)+l, D**l Two methods (a,b) to derive this BR +PDG BR + MC efficiency ratios Assume MC predictions and use 13% systematics

  19. Add another **D** D**

  20. Optimization • The relevant discriminants are: • Pt • R • d0PV • d0BV • d0DV • LxyBD • Signal model: MC • Background model: • WS **l charge • Optimize significance d0PV PV d0BV d0DV BV D0D+ DV **

  21. Discriminating Variables R (**-lD) ** 3d IP signif.WRT BV ** Pt (GeV) ** 3d IP signif.WRT DV ** 2d IP signif.WRT PV

  22. Optimization! • Generate D** montecarlo (the shape) • Normalize D** MC to data with reasonable cuts • Now we can turn the crank and optimize… • But what? • a is the ratio of background events between signal and sideband region (a<1, usually) • S is the MC signal (right sign combinations, signal region) • SWS is the WS data in the signal region • SBRS is the RS data in the sideband region • SBWS is the WS data in the sideband region

  23. Optimal point: • Pt(**)>0.4 GeV • R<1.0 • |d0PV/|>2.5 • |d0BV/|<3.0 D* • S/sqrt(…)8 D+ • |d0DV/|>0.8 • Lxy(BD)>0.05 cm • S/sqrt(…)6.6 • We have to live with different selections for D**D+ and D**D*(K)

  24. Measure <m**2>, <m**4> m**

  25. Current Mass Distributions • DELPHI: • ~80 (K) • ~80 (D+)

  26. Background

  27. Backgrounds • Background from B decays: • Know how to model • Study using Bgenerator/EvtGen/TRGSim++/CdfSim • “Feeddown” • Combinatorial background under D peaks: • side-band subtraction • Prompt pions in D(*)+-l-: • Mostly from fragmentation • wrong-sign combination D++l- • cc • D0 impact parameter distribution

  28. Physics Background • Physics background studied with BD(*)+Ds- • Size wrt signal: • 100% uncertainty Other modes ~7% ~1 ~7%

  29. Background: Feed Down • Irreducible D**0 D*+( D+p0)p-background toD**0 D+p-subtracted statistically: • Mshape of D+p- combination above is like D0p-from D**0 D*+( D0p+)p- • Rate is one half (isospin) times the relative efficiency in both channels time the ratio of the D0 and D+ B.R.’s used in the analysis

  30. Pollution from ccbar? • We are cutting hard on Lxy(B) (500m), this is known to “solve” the lifetime problem • Look at D+/D0 impact parameter for evidence of prompt objects: we do not see any

  31. Measure <m**2>, <m**4> Efficiency Corrections

  32. Pl* • Theory prediction depends on Pl* cuts. We cannot do much but: • see how our efficiency as a function of Pl* looks like • Use a threshold-like correction • Evaluate systematics for different threshold values

  33. MC efficiencies • (M) is dependent on: • D** decay Model • Pl* cut • Use different models/cuts to evaluate systematics: • Individual resonances • Goity-Roberts • Phase space (not shown) • Baseline: BR weighted • (EvtGen) average of modes

  34. Corrections to MC efficiencies • Worried about possible shortcomings of the MC simulation: • Efficiency for requiring Si hits • ** separation variables not perfectly reproduced by MC • Use * candidates from data to derive corrections

  35. * probes Si hit efficiency • D0/+ reco is based on trigger tracks • Si requirements can bias (D0**) • (D0)m** • Take COT-only * • Apply 3Ax+3 (Stereo+SAS) • Measure fraction as a function m** • Use slope (error) as correction

  36. ** Separation Variables 124/49 137/43 • DataMC comparison • Behavior is similar, but there are discrepancies • Derive corrections from this comparison 146/45

  37. Data vs MC efficiencies • Apply to * the same tracking requirements we usually apply to ** (including Si hits) • Measure efficiencies for ** selection cuts applied to * • Take the ratio of MC and Data efficiencies as a function of Pt

  38. Measure <m**2>, <m**4> Moments ExtractionProcedure

  39. Computing the D** Moments • All the pieces are put together in an unbinned procedure using weighted events • Signal right sign (SRS)w = +1 • Signal wrong sign (SWS)w = -1 • Sideband right sign (SBRS)w = -ai • Sideband wrong sign (SBWS)w = +ai • Apply efficiency corrections: efficiencies are propagated on weights W=w (aPt**+b)/[MC(m**)(cm**+d)]

  40. More Backgrounds • Feed-down pseudo-events are formed from the D0p-** mass in Kp events (in SRS,SWS,SBRS,SBWS). The weight is • Physics background events are generated and assigned a weight where e=0.07 is the efficiency of the background relative to the signal. The weight is then corrected with the efficiency factor from MC.

  41. D+, D*+ Relative Normalization • Relative normalization of D* channels is irrelevant since they all have the same underlying M distribution. • D+ channel has a different M distribution. All D+ events have their weights modified as: • Systematics: • BR uncertainties from PDG (7%) •  ratio from studies on data (13%)

  42. Resulting m** distribution

  43. Computing the Xc Moments • The D0 and D*0 pieces have to be added to the D**0 moments, according to where the fi are the fractions of Dil events above the pl*cut. Only ratios of fi’s enter the final result. f

  44. Measure <m**2>, <m**4> Moments ExtractionSystematics

  45. Systematicsshopping list • Mass scale and resolution • Efficiency corrections • From MC • from data • Lepton momentum cut • Background model • Radial excitations • Physics background • Relative D+/D* normalization • Semileptonic B branching ratios • D** mass cut

  46. Mass scale/resolution • We are measuring m** and then adding the PDG masses for D+/D* • Basically insensitive to absolute scale issues • Mass resolution matters • The sample with the worst resolution is K0 • Re-smear K0 with 60MeV gaussian and use this as systematic

  47. Efficiency corrections from MC • Uncertainty comes from lack of knowledge on the D** BR and phase space structures… • Two possible MC models: • BR weighted EvtGen admixture, to the best of today’s knowledge • Plain phase space • Switch to evaluate • systematics…

  48. Efficiency Corrections from data • Efficiencies measured on data have modeling uncertainties/stat. Errors • Float parameters within ranges and compute the effect on the moments • Mass-dependent: use stat error on slope • Pt-dependent: use 0th/1st order polynomial difference

  49. Lepton momentum cut-off • We are not “literally” cutting on Pl* (it is not accessible, experimentally) • Detector implicitly cuts on it • Assume a baseline cut-off • Vary in a reasonable range to evaluate systematics • We use f to derive f**, given f0, f* • f=f(,1) • We use experimental prior knowledge on ,1 to evaluate systematics • Effect is negligible

More Related