b. resende top wg 28/10/05 polarization studies in ttbar events 1 polarization studies in tt events...

B. Resende Top WG 28/10/05

Polarization studies in ttbar events 1

Polarization studies in tt events Polarization studies in tt events with full simulationwith full simulation

1. Physics motivations

2. Full simulation objects (leptons, jets, ETmiss)3. tt event reconstruction: masses, angles4. W and top polarization measurements5. Conclusions

F. Hubaut, E. Monnier, P. Pralavorio, B. Resende – CPPM

C. Zhu – CPPM/SDU



Motivations: Motivations: test top production and test top production and decaydecay

LHC σ ~300 pb

Weak Interaction single top

Tevatron σ ~3 pb

Top polarization

BR(tWb)~100%

No top hadronization

tWb vertex

Top production Top decay

W polarization

Full hadronic (3.7M): jets Dileptonic (0.4M): 2 l + 2 b + 2

Semileptonic (2.5M): l + + 2b + jets

tt final states (10 fb-1)

Test SM, Search for anomalous couplings

LHC σ ~850 pb

Tevatron σ ~7 pb

Strong Interaction gg,qq tt pairs

ATLAS sensitivity to polarization with Atlfast in SN-ATLAS-2005-052

Accepted for publication in EPJC



Angle between: •lepton in W rest frame and •W in top rest frame

Standard Model (Mtop=175 GeV) 0.703 0.297 0.000

W polarization in top decayW polarization in top decay Test the top decay (in fully reconstructed tt) with W polarization ...

Longitudinal W+ (F0)

22

2

2 Wt

t

MMM

22

2

22

Wt

W

MMM

Left-handed W+ (FL) Right-handed W+ (FR)

NLO 0.695 0.304 0.001

...by measuring angular distribution of charged lepton in W rest frame

222

0 2cos1

2cos1

2sin

23

cos1

RL FFFddN

N

Sensitive to EWSB Test of V-A structure

1/2 1 1/2

cos

1/N

dN

/dco

s

W+b

l+

t

spin



tt spin correlationtt spin correlation Test the top production … t and t not polarised in tt pairs, but correlations between spins of t

and t0

)()()()()()()()tt(

LRRLRRLL

LRRLRRLL

ttttttttttttttA

… by measuring angular distributions of daughter particles in top rest frames

ZYXD AAAA AD=-0.29

Mtt<550 GeV

AD=-0.24

)coscos1(41

)(cos)(cos1

212121

2

AddNd

N

)cos1(21

cos1

21

DAddN

NAngle between spin analysers

Angle between t(t) and spin analysers

top spin ≠1/2, anomalous coupling, tH+b

A=0.42A=0.33



Measurement methodMeasurement method

apply weights event by event

Reconstruction and cuts

Correction function :parton level

Selection cuts and reconstruction distort the parton level distribution Use an independent sample to parametrize these effects

Apply this method for W polarization and spin correlation

cos

1/w

eigh

tcos

cos

1/N

dN

/dco

s1/

N d

N/d

cos



Scope of full simulation studiesScope of full simulation studies

SN-ATLAS-2005-052: with 10 fb-1 and combining semilep+dilep (Atlfast):

Spin correlation W polarizationA AD F0 FR

SM 0.422 -0.290 0.703 0.000± Stat ± 0.014 ± 0.008 ± 0.004 ± 0.003± Syst ± 0.023 ± 0.010 ± 0.015 ± 0.024

F0, FR driven by semileptonic and A, AD by dileptonic

Precision between 1% and 5% dominated by systematics. Challenging !

Perform the same study in full simulation



Full simulation Rome samplesFull simulation Rome samplesSignal samples – semileptonic tt, lepton = e, :TopReX+PYTHIA AOD/CBNT 1500 pb-1 375k # 4520, 4521MC@NLO+HERWIG

AOD/CBNT 500 pb-1 130k # T1

Background samples (most important with Atlfast):

W+4jets AOD 75 pb-1 180k # A7Single Top AOD 950 pb-1

1800 pb-1 9200 pb-1

90k 70k 19k

# 4511 # 4530, 4531 # 4540

Understand performance for each object: extensive comparison AOD/CBNT very useful

Quantify NLO effects on polarization measurements (angle resolution): extensive comparison MC@NLO/TopReX



Understand objects in tt events Electron Muon Jets (light and b)

ETmiss

Reconstruct tt events Masses

Background tt(), W+4jets, single top

Reconstruct angles in W and top rest frames Understand pT, , resolutions and spectra of W and tops

lepjetandangle resolutions Results on W and top polarizations

Topics of the presentationTopics of the presentation

Refine selection (isolation, calibration, ...)

Understand pT, , resolutions and spectra



pT(GeV)

Electron selectionElectron selection

Algorithm: author = 1 (high pT electron : egamma)

Identification with simple cuts: IsEM&2047=0 + hasTrack

Isolation: ET (EM+HAD in R=0.3*) < 10 GeV around electron (loose cut)

Rec. Eff. = 81.4%

Selection: 1 e pT>20 GeV, |<2.5

Efficiency = 60.5%

Purity = 98.9%

AOD = CBNT *:0.2 preferred but is not filled in AODs



Algorithm: author = 1 (high pT muon)

AOD (Moore) ≠ CBNT (MuonBoy)

Identification (combined information): hasCombinedTrack=1 with 2/ndof<20

Reject muons from and K decays (bad 2)

Isolation: ET (EM in R=0.2) <10 GeV around muon track (loose cut)

Reject muons from b and c-jets

Rec. Eff. (MuID) = 85.2%

Selection: 1 pT>20 GeV, |<2.5

Efficiency = 58.9%

Purity = 99.2% pT(GeV)

Muon selectionMuon selection



Selected lepton resolution in ttSelected lepton resolution in tt

e EM

pT scale is correct to 0.4%

=2.0 %

e ID =3.2 %

com

b

=2.6 %

e EM =20.10-3

e ID e ID

=0.6.10-3 =0.13 mrad

=0.6.10-3 =0.21 mradNo correction of track curvature

com

b

com

bpT scale is correct to 0.4%

Electron: E in EM calorimeter (brem. recovery) and , in ID (Not AOD default !) Comparison with Atlfast : pT OK, () not gaussian in Atlfast Same resolutions with MC@NLO

pT/pT resolution resolution resolution =1 mrade EM



NTR

THits

Electron : problem with # of TRTHits ?Electron : problem with # of TRTHits ? 10% of electron tracks with no TRT hits ? Muons OK Those tracks degrade pT)/pT by 40% and resolutions degraded by 10% and 20% e+ 2 times more affected than e-

NTR

THits



JetsJets Algorithm : cone with R size of 0.7

Electron/muon in jets :

if R(isolated el. - jet)<0.2 remove the jet (80%)

if 0.2<R(isolated el. - jet)<0.7 subtract electron 4-momentum from jet (15%)

Add muon 4-momentum if R(muon-jet)<0.2

In situ calibration : the 2 jets coming from W

Reduce combinatorial background

Selection : ≥ 4jets with pT>30 GeV, |<2.5

Efficiency = 37.6%

AOD ≠ CBNT : same , but pT (slightly) different... ?



b-tagging, pb-tagging, pTTmissmiss

CBNT:

SV2 weight > 3

Efficiency = 60%

Ruds = 84

Rc = 8

AOD:

SV1 + IP3D weight > 3

Efficiency = 53%

Ruds = 93

Rc = 11

B-tagging selection : ≥ 2 b-jets

pTmiss : >20 GeV

Efficiency = 91.2%



Jet/pJet/pTTmiss resolution in tt eventsmiss resolution in tt events

light jet

pT scale correct to 1%

= 9.8 %

b-jet = 8.6 %

= 18 %

= 0.03 = 22 mrad

= 0.27 = 0.22 rad

= 0.03 = 18 mrad

Large negative shift for pTmiss?? Resolutions as expected.

Same resolutions (and pTmiss shift !) with MC@NLO

resolution resolution

pT/pT resolution

pT scale correct to 1%

shifted by 9%

b-jet b-jet

light jet

light jet



Reconstructed masses, BackgroundReconstructed masses, Background

Kinematic cuts efficiency: 6.5 %. With Mtops<35 GeV: 5.0 %

Agreement with Atlfast and MC@NLO

Non tt background (W+4jets, single top) increases by a factor 2-3 wrt Atlfast (still preliminary)

S/B~10 ; still dominated by internal background tt + X

M=175.5 ± 0.3 GeV =7.5 ± 0.4 GeV

M=172.7± 0.3 GeV =7.2 ± 0.6 GeV

hadronic top leptonic top



Ws resolution: Ws resolution: FullFull vs vs fast simfast simreco – true reco – true(pTreco–pTtrue)/pTtrue

Good overall agreement between TopReX Full simulation and Atlfast

Difference in leptonic W pT resolution explained by pTmiss miscalibration

W had.

W lept.

W had.W had.

W lept. W lept.-1. 1. -1. 1. -1. 1.

-1. 1. -1. 1. -1. 1.



Resolution on Resolution on angle in W rest frame angle in W rest frame

CosCostrue

Full sim ATLfast

TopReX MC@NLO

Very good agreement between TopReX Full simulation and Atlfast ...

... and between TopReX and MC@NLO

CosCostrue

Angle between: •lepton in W rest frame and •W in top rest frame

1/2 1 1/2W+b

l+

t

spin



Results: W polarizationResults: W polarization

TopReX Atlfast10 fb-1

TopReX full sim1.5 fb-1

F0=0.699 ± 0.005FL=0.299 ± 0.003FR=0.002 ± 0.003

F0=0.677 ± 0.015FL=0.309 ± 0.009FR=0.014 ± 0.009

MC@NLO full sim0.5 fb-1

Values compatible with SM expectations in each case

F0=0.684 ± 0.021FL=0.307 ± 0.011FR=0.009 ± 0.011



Top resolution: Top resolution: FullFull vs vs fast simfast sim

Good overall agreement between TopReX Full simulation and Atlfast

Difference in leptonic top pT resolution explained by pTmiss miscalibration

top had.

top lept.

top had.top had.

top lept. top lept.

reco – true reco – true(pTreco–pTtrue)/pTtrue



Resolution on angles in top rest frameResolution on angles in top rest frame

Full sim ATLfast

TopReX MC@NLO

coslep-lejcoslep cosjet

Very good agreement between TopReX Full simulation and Atlfast

Small differences between TopReX and MC@NLO



Results: top spin correlationResults: top spin correlation

A (±stat) AD (±stat)

SM values 0.42 -0.29

TopReX Atlfast 0.42 0.02 -0.29 0.01

TopReX Full sim 0.45 0.06 -0.29 0.03

MC@NLO Full sim 0.02 0.09 0.00 0.05

Toprex values compatible with SM expectations

MC@NLO have no spin correlation included ... Check only it gives 0



ConclusionsConclusions

Fast simulation (SN-ATLAS-2005-052) shown that ATLAS can measure precisely W polarization ~1-2% and top spin correlation ~5% after 10fb-1

This precision requires a deep understanding of all objects:

Electrons: OK except N(TRT). Combine E from calo and angles from tracker

Muons: switch to more recent version of Moore/MuID

Jets: switch to cone size 0.4 for jets

Shift in pTmiss scale : ?

Good agreement between Fast/Full sim. and TopReX/MC@NLO

Next step: systematics update with full simulation

light/b-jet miscalibration, b-tagging efficiency, …



Spare: W, t performances : Spare: W, t performances : TopReXTopReX vs vs MC@NLOMC@NLO

Good overall agreement between TopReX and MC@NLO for the Ws

Difference in tops under invetsigation (b-jet association?)

top had. top had. top had.

W had.

W lept.

top lept.top lept.top lept.

W had.W had.

W lept. W lept.

pT/pT



Spare: Correction functionsSpare: Correction functions

tt spin correlation : AD

tt spin correlation : AW polarization

Good agreement within statistics with Atlfast

Agrees also with MC@NLO

• Full sim

ATLfast

b. resende top wg 28/10/05 polarization studies in ttbar events 1 polarization studies in tt events...

Documents