W Charge Asymmetry at CDF

W Charge Asymmetry at CDF

Kevin McFarlandUniversity of Rochester

on behalf of CDF collaboration

Ph.D. thesis of Bo-Young Han

DIS 2008, UC-London, 8 April 2008DIS 2008, UC-London, 8 April 2008

K. McFarland, W Charge Asymmetry 28 April 2008

Introduction and Analysis Technique Signal, Backgrounds and Corrections Uncertainties Preliminary results

Outline


W Charge Asymmetry and PDFs At Tevatron, W± are pro

duced primarily by uproto

n (danti-proton) and uanti-proton (dproton)

ud

/ /( )

/ /( ) ( ) ( ) ( )

( ) ( ) ( ) ( ) where W W

W WW

W W

M y

s

d dy d dyA y

d dy d dyu x d x d x u x

u x d x d x u xx e

W charge W charge AsymmetryAsymmetry

u quarks carry more momentum than d quarks


Lepton ChargeAsymmetry Previous work

from TeVatron has measured the charge asymmetry vs. lepton η in W→ℓν

This observable is aconvolution of the Wcharge asymmetryand V-A W decayangular distribution

CDFII, 170 pb-1

Lepton charge asymmetry

W charge asymmetry


Lepton and W Rapidity

2**

( )1 cos

d W e

d

Lepton prefers to decay against boost


Analysis Technique

Measure W± rapidity

Use W mass constraint solve eqn. answer :

Weight the two solutions

This method must be iterated to remove input bias shown it does not depend on assumed charge asymmetry

222 )()( PPEEM llW

)(),,(cos)(),,(cos

)(),,(cos

22*211

*1

2,12,1*

2,12,1

ypyPypyP

ypyPF W

TW

T

WT

z

zW PE

PEy ln

2

1 z

lz

Wz PPP

can’t measure !!!

21 , zz PP

Probability of angular distribution

Differential W cross section


Complications:Detector Effects

Response matrix to correct for detector acceptance, smearing and final state effects QED, W→τν→eννν

Y1

Y2

weightinweighting factorg factor

wt1

wt2

response response matrixmatrix

YW

true true rapidirapidi

tytyYYWW

FSRFSR

depends on physics input →should be iterated

doesn’t depends on physics input →do not need to be iterated


Complications:W Production from the sea Sign of V-A angular bias flips

when W± is produced fromanti-quarks take this fraction as an input

ratio of two angulardistributions at each rapidity

* * 2

* 2

(cos , , ) (1 cos )

( , )(1 cos )

WW t

WW t

P y p

Q y p

q P q P

q P q P


What is input, and what is measured? Inputs from theory:

and

Output from iteration:

u d

u d

( ) ( )

W W

d pp W X d pp W X

dy dy

( ) ( )

( ) ( )W W

W W

d pp W X d pp W X

dy dy

d pp W X d pp W X

dy dy

WA y

method documented in B.Y.Han et al., arXiv:hep-ph/0711.2859



Outline


Summary of Data Issues Require one high ET

electron and missing ET

to tag the neutrino ET

e>25(20) GeV incentral (plug) detector

missing ET>25 GeV Detector response,

including missing ET,tuned on Z→e+e- sample

Highlight backgrounds (primarilyjets faking electrons) and charge mismeasurement also, trigger and detector acceptance and smearing


Scale and resolution tuned on Z→e+e-

Red : MC

Blue : data

Data and Simulation Kinematic Distributions


Backgrounds

Measure jet backgrounds directly in data use extra energy “isolation” around electron to separate, and

fit shape to background fraction

Illustrative fit (one of many)shown at right use jet sample to predict measured

“yW” and charge from this sample

uncertainty is ~0.15% of total sample

A number of minor backgrounds(real Ws) from simulation

electron


Electron ChargeIdentification Charge identification is

crucial for this measurement

Forward tracking has fewer points at shorter lever arm

Determine directly from (background subtracted)

sign samesign opposite

sign same

NN

Nfmis



Outline


Systematic Effects

Detector response energy scale and smearing for electron and recoil efficiency to find electron and pass missing ET cut

Inputs PDF uncertainties (CTEQ6 PDF error sets) for total W

production and quark/anti-quark fractions


Evaluation Derive result with shifted parameters Systematics dominate only if yW<0.2 or >2.6

largest detector effect: energy scale largest PDF effect: qbar/q

largest selection effect: elec. cuts charge fake rate uncertainty



Outline


Uncertainties

Unfolding correlates (strongly) the statistical errors innearby bins

Systematics also strongly correlated… PDF fitters, please exercise caution…


Input PDFs We are in the process of documenting how

result depends on input PDFs PDF fitters can explicitly put this in (or ignore if small)

effect of increasing valence u+d by +5%

effect of increasing sea u+d by +5%


CDF Preliminary Result (1fb-1)

Positive and negative yW agree, so fold Compare to NNLO (MRST)

NLO error PDFs (CTEQ)


CDF Preliminary Result (1fb-1) Compare CTEQ6M and MRST2006 with PDFs and their

uncertainties


First direct measurement of W charge asymmetrydespite additional

complication of multiplesolutions, it works!

appears that it will haveimpact on d/u of proton

Looking forward toworking with PDF fitting groups to incorporate

Conclusions


Introduction and Analysis Technique Signal, Backgrounds and Corrections Uncertainties Preliminary results Backup

Outline


Results


Systematics


Acceptance

W

WW dyd

ywtya

/

)()(


Plug Electron Tracking Efficiency


QCD higher order corrections Our weight factor is initialized by one prediction.

The parameterization of angular distribution : The ratio of anti-quark and quark in proton : Q(yW,PW

T) using MC@NLO

Differential cross-sections : dσ±/dyW

KNNLO = (dσNNLO/dyW) / (dσLO/dyW)


f(PT) and g(PT)

3)(

)4(3)2,1,(0)(

polPg

PpExppppPLandaupPfW

T

WT

WT

WT


NNLO K-factor: (dσNNLO/dyW)/(dσLO/dyW)

NNLO dσ±/dyW (mrst2002 PDFs) from theoretical prediction.


NNLO K-factor

W charge asymmetry measurement using K-factor.


Z PT at LO (Pythia)

check if LO (Pythia) is tuned by DATA. Using Zee samples : zewk6d(Pythia) vs bhel0d(data) Looks MC agrees with data.


W Pt tune at MC@NLO Comparison of Pt distribution between NLO(Herwig) and LO(Pythia)

Using W→eν samples Obtain the re-weight factor from the ratio of both We are going to re-weight the event of NLO.


Q(yQ(yWW,P,PWWTT) (cont.)) (cont.)

10 < PT and 10 < PT < 20



20 < PT < 40 and 40 < PT < 60



60 < PT < 80 and 80 < PT < 200


• Q factor + K-factor

Even if QCD higher order correction makes small effect on the asymmetry, we believe that it describes better differential cross-section and angular distribution.


• Q(yW,PWT) using MC@NLOMC@NLO

W charge asymmetry measurement using NLO Q factor Compare it with previous result using LO Q factor

W Charge Asymmetry at CDF

Documents

Transcript of W Charge Asymmetry at CDF