Single Top Search

~ New Kinematics Fitter and ttbar bkg~

K Nakamura

Pointed out previous meeting… MET fit function Top specific correction factor W boson direction at the W* rest frame lepton direction at the Top rest frame Use Kinematics likelihood to the event selection Mbb distribution for the event selection

ttbar background

ttopvl: Mtop=175 GeV Generated by Pythia MC version 5.3.3 spring summer2005 version

Contents

S-channel single top event @ CM Frame

W*

t

b

b

l

ambiguity

Selection of bjet from top quark

t-channel : tagged jet -> 95%

s-channel : tagged jet -> 53%

CDFNOTE

top mass resolution +15 GeV

Selection of neutrino Pz solution30% … no solution

Mlb : for 4 cases

This Top Mass constraint gave some bias to Mlnb distribution??

Likelihood Function for neutrino Pz

To define a Likelihood as the function of Neutrino Pz

Move three parameters: METx METy PZ

2 solution case

0 or 1 solution case

Likelihood Function for bbbar ambiguity

is the angle of b-jet @ Wstar Rest Frame

MET

Double Gaussian with single mean value (5 parameter)

Top Specific Correlation

gen bPt [GeV]

rec

bPt [

GeV

]Rec Pgen P

Top Specific Correlation Factor

distribution

p0=2.12

p1=23.3

p2=0.057

p0=1.83

p1=24.8

p2=0.032

p0=1.31

p1=25.0

p2=0.081

Rec Pgen P – Rec P

Xaxis value of each bin is weighted mean

From takeuchi-san From takeuchi-san From takeuchi-san

Mlnb distributionSim Lev

-- top-- other

W boson direction at the W* rest frame

Dominant channel of Wbb background is t-channel like upper right feynman diagram.-> W boson direction is close to p or pbar direction

P PW*-

t

b

b

W

W*+

t

b+Z

@ Wstar Rest Frame

W

b

du

u d

Signal

Wbb

@ Wbb Rest Frame

P PW

b

+Z

b

d u

b b

W

u dd uu d

+-

+-

Signal

Wbb

e

ttbarBkgs

??

W+W-

lepton direction at the Top rest frame

P P

b

l-

b

l+

+Z

t tu dud

@ Top Rest Frame

Top quark is polarizing the d-type-quark direction at the top rest frame.-> lepton is boosted d-type-quark direction

W*+W*-

correctwrong

cos lep distribution

Use Kinematics likelihood to the event selection

Similar distribution…

Mbb distribution for the event selection

-> bb invariant mass in the Wbb background looks much smaller than signal…

Mbb distribution

We can use this variable for event selection !

ttbar backgroundMlnb distribution normalized by # of events of CDFnote

At least 1 b-tag

B-tag

64.3%

29.5%

52.4%

19.7%

45.9% 37.6%

Signal W(mu)bb

57.2%

20.9%

36.5%

ttbar

At least 1 b-tag S+S S+J double tag

Double tag…

61.21

7.02

20.67}

8.41.9

0.493.3

----

0.6

S/B=7.02/96.6=0.07

S/sq(B)=0.71

S/B=15.3/534.5=0.03

S/sq(B)=0.66

improvement

Backup

S-channel single top event @ CM Frame

W*

t

b

b

l

ambiguity

Selection of bjet from top quark

t-channel : tagged jet -> 95%

s-channel : tagged jet -> 53%

CDFNOTE

top mass resolution +15 GeV

Selection of neutrino Pz solution30% … no solution

Mlb : for 4 cases

This Top Mass constraint gave some bias to Mlnb distribution??

Neutrino solution

Likelihood Function for neutrino Pz

To define a Likelihood as the function of Neutrino Pz

Move three parameters: METx METy PZ

2 solution case

0 or 1 solution case

PZ @GenLev

---- W*+

---- W*-

To have pronounced tendency to separate W*+ and W*-

P PW*-

t

b

b

l-

W*+

t

b

b

l+

+Z

correct wrong

PZ *Qlep

@Sim

---- correct---- wrong

W*

t

b

b

l

MW*2

> Mt2+Mb

2~ Mt2

MW* =q2=sW* has the mass as small as possible to keep larger than Mt…??

W* mass

MW*

@SimUsing L5 jet

No b jet correction

Likelihood … correlation check

MW*

PZ *Qlep

MET

Gaussian

DoubleGaussian

We use this one !

Result of the Likelihood fitting

bbbar ambiguity

Likelihood Function for bbbar ambiguity

is the angle of b-jet @ Wstar Rest Frame

B-jet Pt distribution

W*

t

b

b

l

Using L5 jet

No b jet correction

Gen Lev Sim Lev

-- top-- other

-- top-- other

Use these distribution as the Probability Density Function

Mlb distribution

W*

t

b

b

l

Using L5 jet

No b jet correction

Gen Lev Sim Lev

-- top-- other

-- top-- other

Use these distribution as the Probability Density Function

cosb x Qlep

P P

W*-t

b

b

l-

W*+t

b

b

l+

+Z

tb production process has the V-A coupling, so it is easy for the b-jet to be produced in the direction of quark in proton, and bbar-jet to be produced in the direction of anti-quark in anti-proton @ the Wstar rest frame.

@ Wstar Rest Frame

cos angle shown in upper figure times lepton charge is the powerful variable to select b-jet which is directory producedform Wstar.

cosb distribution

Gen Lev

Sim Lev

top other

top other

---- W*+

---- W*-

cosb x Qlep distribution

Gen Lev

-- top-- other

Sim Lev-- top-- other

Use these distribution as the Probability Density Function

Parameter correlation

These figure shows some correlations between the Likelihood parameters……But Likelihood is only used as the choosing bbbar ambiguity, so we use this likelihood !

Top Specific Correction for b-jet

Correlation Factor as the function of Pt Eta

After Correction check

Top Mass DistributionLinear Scale Log Scale

Wbb background Shape

---- Wbb---- Signal

Normalization is not correct…

Use MC Wbb+0j bkg

---- Wbb---- Signal

Mlnb distribution normalized by # of events of CDFnote

Wbb 198.5 events s-channel 15.00 events

At least 1 b-tag

64.3%

29.5%

52.4%

19.7%

B-tag

45.9%37.6%

Signal W(mu)bb

---- Wbb---- Signal

At least 1 b-tag S+S S+J double tag

Double tag…

---- Wbb---- Signal

Enhancement !?

ConclusionWe developed new likelihood fitter without using top mass constraintMake Likelihood fitter like blow :

For the 75.2% events of the 2 solution events we can choose the correct solution.After decided neutrino Pz, we can choose the correct b-jet from the top quark for the 81% of the total events.But Mlnub distribution is not so improved…We will try to check correlation between the likelihood variables and top mass

2 solution case

0 or 1 solution case

Backup

W*

t

b

b

l

wrong

correct

PW*Z

@GenLev

---- W*+

---- W*-

Bug is Fixed !

PZ *Qlep

@GenLev

---- correct---- wrong

W*

t

b

b

l

wrong

correct

MW*

@GenLev

Probability Density Function

Neutrino Solution Check

@GenLev @SimChoose CorrectSolution

Result of the Likelihood fitting What??

PZ=0 ??

-2Log L distribution as the function of PZ First 6 events…

Likelihood distribution for the case of PZ=0　

Most likely PZ is really 0 ?? Is this correct??

---- correct---- wrong

corr

wrong

72.7%75.2%

Likelihood for Solution

corr

wrong