J K wrong flavor decays

1
J/? K wrong flavor decays
Discussions of some common analysis techniques
in BaBar by Max Baak
2
Outline

• Why look at J/? K wrong flavor decays?
• Theoretical introduction
• BaBar in a nutshell
• Analysis Strategy
• BaBar data sample
• Fit Systematics
• Conclusion

3
CP Violation via the CKM matrix

• The CKM matrix is a complex unitary matrix,
  coupling between quark generations and W bosons.
• With 3 quark generations, it allows for 4
  independent, physical parameters
• 3 real numbers 1 complex non-trivial phase
• The existence of the complex coupling (phase)
  gives rise to CP violation.
• All CP violating observables are possible due to
  interference between different decay amplitudes
  involving a weak phase.

4
The CKM Matrix Wolfenstein parameterization

Complex phase
Unitarity
Wolfenstein parameterization uses the observed
hierarchy of the CKM elements and pushes the
complex phase to the smallest elements
? Vus sin(qcabbibo) 0.2205 0.0018 A Vcb/
?2 0.830.06

• Out of 6 unitarity triangles, this one
  practically interesting
• It has all sides O(?3)
• Large phases ?potentially large CP asymmetries

5
CP violation in the inference between mixing and
decay
Time evolution of initial B0 (or B0) mesons into
a final CP eigenstate
In order to have CP Violation
Amplitude ratio
Mixing Phase

• A single decay amplitude is sufficient
• -Mixed decay serves as 2nd amplitude
• -Thus, amplitudes comparable by construction
• -Large CP asymmetries are possible!

6
Golden Decay Mode B0 J/y K0S
K0 mixing

• Theoretically clean way (1) to measure the phase
  of l (i.e. sin2b)
• Clean experimental signature
• Branching fraction O(10-4) - large compared to
  other CP modes

Golden Modes

• hCP -1
• B0 ? J/? K0S
• B0 ? ?(2s) K0S
• B0 ? cc1 K0S

• hCP 1
• B0 ? J/? K0L

7
Can sin2bL and sin2bS be different?

• Normal assumption is that sin2bL-sin2bS .
• This holds to 1 in the Standard Model
• - Corrections from DG, ek, q/p?1, and
  suppressed penguins.
• Current value is S(J/Y Ks) S(J/ Y KL) 0.04
  0.17
• - Consistent with SM, but statistics limited.
  Can one do better? Yes!
• Violation of sin2bL-sin2bS requires (different)
  wrong-flavor amplitudes
  , forbidden in the
  Standard Model.
• How to check for these? Practically K0 mixes into
  CP states.
• At first order underlying physics for
  wrong-flavor K and K decays assumed to be
  similar.
• Use high-statistics sample
  to tag K0. ?
  Model-independent search for new physics.
• hep-ph/0204212 (Y. Grossman, A. Kagan, Z.
  Ligeti)

8
J/Y K Mixing pdfs

• Assume wrong-flavor decays are allowed. How do
  the pdfs change?
• Define the ratios ,
• For final state J/Y K0 this results in the
  mixing equations
• Where again , .
• For final state simply replace l by
  . One gets and .
• Equations add up to pure exponential ? need to
  determine initial flavor (t0 ps) of B meson to
  differentiate between mixed unmixed states.
• Time-dependent analysis gives
  coefficients at few level.

9
B meson production at BaBar

• Electron-Positron collider ee- ? ?(4s) ? B0B0
• Only ?(4s) resonance can produce B meson pair
• Low B0 production cross-section 1 nb (total
  hadron 4 nb)
• Clean environment, coherent B0B0 production

B-Factory approach
BB threshold
B0B0 threshold

• 81.3 /fb of BaBar data ? 88 million Bs

10
?(4S) Coherent B0B0 production
Incoherent (LHCb)

• B0B0 system evolves coherentlyuntil one of the
  particles decays
• Mixing-oscillation clock only starts ticking at
  the time of the first decay ?relevant time
  difference parameter Dt
• B mesons have opposite flavour at time Dt0
• Half of the time B of interest decays first
  (where Dtlt0)
• Integrated sine asymmetry is 0
• Coherent production requires time dependent
  analysis

At tcp0
B0
B0
Dt tB1 tB2
t(ps)
At Dt0
B0
B0
Dt(ps)
Coherent (BaBar)
11
A(-)symmetric collider for ?(4S) will (not) work

• Asymmetry is a time-dependent process
• ?t between two B decays of O(ps)
• In reality one measures decay distance between
  two B decays
• In symmetric energy ee- collider, where ?(4S)
  produced at rest, daughter Bs travel 20mm ?
  too small a distance to discern.
• Solution boost the CMS to increase distances in
  lab frame. Build an asymmetric collider!
• For BaBar
• High energy e- beam 9.0 GeV
• Low energy e beam 3.1 GeV
• ? ?? 0.56

12
In pictures
Z
Start the Clock
Coherent BB pair
This can be measured using a silicon vertex
detector!
13
Experimental technique
Exclusive B Meson Vertex Reconstruction

• Key strategies
• Exclusive B-reco for 1 meson
• Use other B to determine
• flavor-tag at Dt0.
• Determine vertices to get Dz.
• Question
• How to handle mistags?
• Limited vertex resolution ? need to disentangle
  resolution from physics.

Inclusive B-Flavor Tagging Vertex
Reconstruction
14
True Dt distributions of mixed and unmixed events
realistic mis-tagging finite time resolution
perfect flavor tagging time resolution
Dmd oscillation frequency
w the fraction of wrongly tagged events
Mistag rates need to be disentangled from C S
coefficients!
15
Splitting the Dilutions from the Coefficients

• To disentangle mistag fractions from (co)sine
  coefficients, a second, large data-sample is
  needed, having known coefficients.
• In BaBar uses the Breco sample, described with
  basic pdf
• Including the mistags the asymmetry then turns
  out as

Folded raw asymmetry
Sensitive to mistag fraction measurement because
the mixing has not started yetAt t0 the
observed mixed events are only due to wrongly
tagged events
Dt ps
16
Methods of B flavor tagging (1)

• In BaBar tagging is handled with Neural Nets.
• Many different physics processes can be used for
  tagging, primary information is listed below

17
B flavor tagging performance (2)

• 9 sub-taggers, using combinations of the various
  inputs, are combined in the Tagging Neural
  Network.
• The NN spits out 4 physics categories in which
  the data is cate-gorized, all with different
  tagging efficiencies and mistag-fractions.

Why? Number of events is prop. to e(1-2w).
Multiplication of CS with (1-2w) gives another
factor to Q.
BABAR 81.3 fb-1
18
Vertex and Dz reconstruction

• Convert from ?z to ?t, accounting for (small) B
  momentum in ?(4S) frame
• Note event multiplicity 10-12

Result s(?z)rms 180µm (?t0.6ps) versus lt?zgt
ß?ct 260µm
19
Actual Dt signal resolution function

• event-by-event s(Dt) from vertex errors
• Resolution Function (RF)
• Sum of 3 Gaussians (mixing CP analyses)
• Core correct vertex (90). Error systematically
  underestimated, so scaled up with Score (1.1).
• Tail nearly correct vertex (10). Reco. vertex
  picked up (a) track(s) from the tag B.
• Outliers (lt 0.1) wrong vertex. Outlier
  component serves as a vacuum cleaner.

sDz
0.6 ps
tracks from long-lived Ds in tag vertex?
asymmetric RF
Signal MC (B0)
high flexibility
Dt (meas-true)/sDt
20
Effect of charm tracks on Dt
Underlying principle tag vertex dominates
resolution. tag ?z110?m, reco ?z65?m
Bias mcore bcore sDt , mtail btail sDt ,
moutl 0
21
Correlation s(Dt) residual Dt bias
D flight direction
Monte Carlo
Charm tracks
ztag
Prompt B tracks
Charm tracks
ztag
Prompt B tracks
s(ztag)
D flight direction
s(Dt) smallest, Dt bias zero
s(ztag)
s(Dt) largest, Dt bias largest
z axis
bias
22
B reconstruction
For exclusive B reconstruction, two nearly
uncorrelated kinematic variables are employed to
cut on background. Both use the property that
Ebeam is well known
23
Example
mES
DE MeV
Typically, DE is fit for all events with mES gt
5.27 GeV. The entire mass spectrum is then
refit within the energy window to obtain bkg.
probablities, to be used as inputs in the
likelyhood fit.
signal region
DE
sidebands
mES GeV/c2
24
Breco Sample All
The Breco sample contains 24 reconstructed B0
open charm modes.
B open Charm decay modes
Charm decay modes
Prob(sig) 81.6 Prob(sig) 0 Gaussian ARGUS
function
BABAR 81.3 fb-1
Ntag 30977 Purity 81.6 Sigma 2.76 MeV
mES GeV/c2
25
Breco Sample Per tagging category (example)
Lepton
Lepton
Kaon
NT2
NT1
BABAR 29.7 fb-1
26
J/Y K data sample
Cleanest data sample in BaBar!
Yield 1641 events, Purity 97.3 , Mass
resolution 2.7 MeV Set tight K and p
selection, to minimize accidental swapping.
27
Fitting Technique

• Analysis performed blind to prevent
  experimenters bias.
• Simultaneous unbinned maximum log-likelihood fit
  to Dt spectra of both Breco and J/Y K samples.
  (Likelihood fit accounts for Poisson stats.)
• Fit for cosine and sine coefficients C, S, C, S.
• Signal model pdf for mixed and unmixed events
  (4) convolved with triple gaussian signal
  resolution function (8). Dilutions and
  dilution-diffs between B0 and B0 tags are
  incorporated for each tagging category (8). tB
  and Dmd fixed to PDG 2002 values.
• Background model prompt and lifetime components
  for mixed and unmixed data (5) convolved with
  double gaussian resolution function (5). Separate
  dilutions for background description (10).
• Assign probabilities for individual events per
  tagging category to be signal (probsig) or bkg
  (1-probsig), based on observed mES value and a
  global fit to the mES distribution.
• - Likelihood function Sum all signal and bkg
  pdfs for a combined fit with a total of 40 free
  parameters.

28
Background description
MC cocktail

• 4 types of background are accounted for in
  empirical Dt description
• Argus background (combinatorics)
• Prompt background no time dependence (70)
• Lifetime 1 pure double-sided exponential
• Lifetime 2 exponential mixing terms
• Peaking background (in signal probability)
• Lifetime 3 double-sided exponential, fixed to B
  lifetime.

peaking bkg

• For J/Y K data
• 2.5 Argus shape background 1.2 from
  inclusive J/Ys
• Peaking background (from incl. J/Y MC) 2.3 J/Y
  K-p (non resonant)
• 1.1 J/Y p0/r0/KS

29
Systematic errors on C(C) and S(S) (preliminary!)
30
Conclusions

• No conclusions yet analysis is still blinded.
• s(C) s(C) 0.055?0.031 s(S) s(S)
  0.094?0.033
• Max C correlation 29 max S correlation 9
• Expected error s(S(J/Y Ks) S(J/ Y KL)) 0.14
  (old 0.17)