Presentation on the normalization channels

Search for signal in B->D(*)ππlν

channel in BaBar data

- Ruturaj ApteIIT Bombay,

IndiaWorking under

Prof. Robert KowalewskiThomas Lueck

Outline

● Motivation for studying semileptonic decays

● Motivation for our search for the D(*)ππlν channel

● Some details about the experiment

● Reconstruction of the Btag

meson

● Introduction to “normalization modes” and how they help in reducing the uncertainties in the final branching fraction calculation

● Multi variate fisher analysis

● Fitting of components

Motivation for studying semileptonic decays

● In the Standard Model, the CKM matrix helps describe CP violation in weak interactions involving quarks of different flavors.

● Precise determination of these matrix elements is crucial for a stringent test of the SM and for reducing theoretical uncertainties in the many New Physics searches with flavor.

● Semileptonic B decays are used to measure the matrix element |Vcb

|

● According to Babar measurements, the experimental estimate for the quantity

BF(B->D(*)τν)/BF(B->D(*)lν) shows a deviation of 3.3σ from the SM prediction. The channel we are looking for is a background for the above search and depending on our result the above 3.3σ deviation can increase or decrease.

Motivation for our search

● The sum of the Branching Fractions for all the known exclusive semileptonic decays of the B meson do not match the inclusive B->X

clν Branching Fraction. The Branching fractions are :

B -> Dlν : 2.30 (+/-) 0.1

B -> D*lν : 5.34 (+/-) 0.12

B -> D**lν : 1.64 (+/-) 0.18

Inclusive B->Xclν : 10.91 (+/-) 0.14

● Gap between exclusive and inclusive : 1.63 (+/-) 0.3

Y(4S) resonance and need for a Btag

● The asymmetric e+e- B factories operate at a center of momentum(CM) energy of E = 10.58 GeV

● This energy corresponds to the mass of the Y(4S) resonance, a

bound state and it then decays almost exclusively to approx equal numbers of and B+B- pairs.

● The B mesons are almost at rest in the Y(4S) frame

● No other additional particles are produced.

B0 B̄0

b b̄

Y(4S) resonance and need for a Btag

● The decay products of the B and antiB overlap completely in the detector.

● The vertex resolution is not good enough to unambiguously assign charged tracks

● Hence a kinematic reconstruction one of the B mesons in a fully reconstructed decay mode is needed to assign particles to the B and anti-B

● The characteristic semileptonic decay is searched by identifying a reconstructed track left by an electron or a muon with CM momentum > 0.6 GeV

Event Reconstruction Technique

● A total of 2968 separate decay chains are considered in the reconstruction of the B

tag.

● One of the independant variables used to study the Btag

reconstruction is “delta E” which is the energy difference between the reconstructed and expected energy of the Bmeson.

Delta E = ErecB

– (Ecms

of the colliding beams)/2

● Energy substituted B mass (mESB) =

mESB = sqrt(E2/4 – PB

2)

E = CM energy of the colliding beams

PB = three vector momentum of the B meson.

Event Reconstruction Technique

● After that a D(*) meson is reconstructed using a subset of D meson decay modes.

● While looking for decays with an added pion, we look for an extra pion track which has not been used for any reconstruction.

● The percentage of Y(4S) decays with a lepton that give an acceptable B

tag reconstruction is only about 0.5%

● We do not want to rely on the MC to estimate our Btag

reconstruction

efficiency

Delta E Plot for B->Dlν signal side

MESB Plot for B->Dlν

Normalization Modes

● The estimate of the systematic uncertainty related to differences in the efficiency for reconstructing the B

tag in data and MC is

non trivial.

● Thus we use a sample of B->D(*)lnu as normalization modes in order to cancel out these uncertainties.

Efficiency calculation

● Signal Efficiency is defined as the number of signal events reconstructed divided by the number of signal events actually produced in the detector.

● We estimate this efficiency from the MC samples

● Efficiency = Nsig,mc

/ (NBBbar,MC

*bf*2)

Nsig,mc

= Signal yield from MC

NBBbar,MC

= Total BBbar events generated in MC

bf = Branching Fraction of the signal decay mode put in during MC generation of the events

Calculation of the Branching Fraction

● Once we extract the signal yield from the fit, we can continue to estimate the branching fraction.

● BF = Nsig,data

/ ( εtag

*εsig

* NBBbar,data

)

εtag

= efficiency of the Btag

reconstruction

εsig

= efficiency of reconstruction on signal side

Nsig,data

= signal yield from the fit

NBBbar,data

= Number of B,Bbar events from the data

How it will help in our final BF calculation

● We need to cancel out the uncertainties that arise due to systematic uncertainty related to differences in the efficiency for reconstructing the B

tag i.e. ε

tag

● BF(B->DPiPilnu)/BF(B->D(*)lnu) = Nsig

* εsig,norm

/ εsig

* Nnorm

● The BF(B->D(*)lnu) is taken from the well measured BFs of these modes.

● We thus cancel out the εtag

assuming it is the same for both,

the signal and normalization modes.

Variables of Interest● (Missing mass)2 : missing mass square for a particular decay say

B->(D(*)lν + nπ) is defined as :

(mmiss

)2 = (Py – P

Btag - P

D(*) – P

l - P

pions )2

should peak at 0 for signal

● plep_cms : momentum of the lepton in the cms frame.

● mESB : sqrt(E2/4 - PB

2)

where E is the total CM energy and PB

is the momentum of the B meson.

Variables of Interest

● Delta E = ErecB

– (Ecms

of the colliding beams)/2 should peak around 0 for correctly reconstructed B

tag.

● Extra Energy : The total energy of all the neutral particles and photons that have not been used for reconstruction of any particle

● Unmatched neutral : Number of neutral particles that have not been used in any reconstruction.

● Charged multiplicity : number of charged particles used in the reconstruction of the B

tag candidate.

Variables of Interest

● CosThrustB : the cosine of the thrust angle of the Btag

candidate

with respect to the rest of the event.

● MoltNB : number of neutral particles that have been used to reconstruct the B

tag.

● Fox2CT : A variable that measures the event shape. It is close to 1

if the event is jetlike and close to 0 for a spherical event.

Why they are useful

Cuts applied

● Flavor correlation between D and B.

● Charge Flavor correlation between the D and the lepton

● 5.27GeV < mESB < 5.29GeV

● Momentum of lepton in cms frame > 0.6GeV

● -2GeV < mm2 < 2GeV

● Total charge of the event has to be 0.

● Total charge of the D meson and added pions has to be <=1

Multivariate Analysis

● The method of Fisher discriminants was used to reduce the background levels

● The variables used for the fisher tuning were

ExtraEnergy,mESB,unmatched neutral multiplicity, multiplicity of charged particles in B

tag reconstruction, absolute value of cosine of

Thrust angle of Bmeson with the rest of the event ,

Fox2CT, multiplicity of neutral particles in Btag.

● A different Fisher cut expression was obtained for each decay

mode increasing the significance of the data.

(Missing mass)2 plot for B->Dlν

Improvement in signal to background ratio

● We quantify the signal/background ratio using a quantity called

significance

Significance = S/ sqrt(S+B)

S : number of signal events

B : number of background events

● Significance of the mm2 plot without the fisher tuning = 164.432

● Significance after the fisher tuning = 171.106

● Improvement is not very drastic since these decay modes are already quite clean

Similar analysis done for the B->Dπlν

Improvement due to Fisher Tuning

● Initial significance with D** that is the D1, D

1' , D

0 , D

2 as the

signal and all other components as background is

= 15.80

● Significance after fisher tuning

= 21.38

● An increase of 35.31%

● This analysis was repeated for B -> D(*)πlν and

B -> D(*)ππlν for charged as well as neutral D mesons.

B ->D*+ππlν

Improvement due to Fisher

● Significance before the Fisher cuts :

19.6025

● Significance after the Fisher cuts :

22.76

● An increase by 16%

Fitting to the (missing mass)2

● Used the RooFit package for fitting

● Did an unbinned maximum likelihood fit for the data.

● The components that were fitted were:

1. Dlnu

2. D*lnu

3. D**lnu

4. Other BBbar decays

5. Continuum events ( e+e- -> qqbar ; q != b)

✔ Used 2 types of fit methods : one by extracting histogram pdfs from the MC histograms and other using RooKeysPdf which gives a smooth pdf

Fit Plots for B -> D+lν

Fit Plots for B -> D+lν

Final Fit for B -> D+lν

Results of the fit● Signal Yields for the normalization mode calculated from the

fit are :

D0lν = 8365.84 +/- 135.485

D+lν = 3671.78 +/- 86.834

D*0lν = 12052.4 +/- 143.713

D*+lν = 5498.93 +/- 84.4648

● Signal reconstruction efficiencies :

εsig

(D0lν) = 0.077 %

εsig

(D+lν) = 0.014%

εsig

(D*0lν)= 0.123%

εsig

(D*+lν) = 0.022%

Conclusion

● The normalization modes have been measured to get their signal yields and to get the signal efficiencies for the individual decay modes.

● We now need to fit to the 2 pion cases using the MC generated samples for DΠΠlν to get the signal yield and the efficiencies for this channel.

● We then use the double ratio formula to get our final Branching Fraction.