Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://qfthep.sinp.msu.ru/talks2011/qfthep2011_popov_v3.pdf Äàòà èçìåíåíèÿ: Wed Oct 5 13:49:10 2011 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 19:53:27 2012 Êîäèðîâêà: ISO8859-5

Measurement of the single top production with the CMS detector

Andrey Popov
On behalf of the CMS collaboration

QFTHEP 2011, Sochi, Russia 26.09.2011
1 / 29


Outline
Ç Intro duction CMS detector Physical motivation for single top studies Single top processes Ç Single top t - & tW -channels General description of the process Event selection Data-driven estimations Analysis procedure Systematics Results Ç Conclusions

2 / 29


CMS detector

3 / 29


Why is single top interesting?
Single top probes the W -t -b interaction and can be sensitive to Ç additional non-standard particles
extra quarks extra gauge bosons additional scalar bosons Ç mo dified t -quark interactions FCNC contribution right-handed interactions

4 / 29


Single top processes
s -channel q ? W q t g ? b b tW -channel Wg t tb b t b W q t -channel q W t

Different rates and different kinematical regions
x-sec, pb
mt = 173 GeV TEVATRON, p p @ 1.96 TeV ? LHC, pp @ 7 TeV

s -channel 1.04 4.59

tW -channel 0.22 15.6

t -channel 2.08 63.2

t? t 7.2 165

N. Kidonakis, Phys. Rev. D 81, 054028 (2010), Phys. Rev. D 82, 054018 (2010), Phys. Rev. D 83, 091503(R) (2011), arXiv:0909.0037v1

5 / 29


Single top t -channel
LO q W b NLO q W b g
6 / 29

q

Ç Analysis fo cused on leptonic decays of W :

t Wb l b t
Ç Signal signature: exactly one isolated lepton, forward light-quark jet, central b -jet from top decay, additional b -jet with small pT is outside acceptance Ç Main backgrounds: t ?: similar kinematics, high rate t W ( l ) + jets QCD: extreme kinematical region, data-driven estimation needed

q t ? b


Analysis strategy
Two independent and complementary analyses performed:
Ç "2D" Exploits signal-sp ecific angular prop erties performing a 2D fit

in the corresponding variables
Ç "BDT" Multivariate approach with b o osted decision trees.

Combines many distinctive variables into a single powerful discriminator Results in t -channel presented here correspond to 36 pb-1 collected in pp run of 2010. The updated results are in preparation Details in Phys. Rev. Lett. 107, 091802 (2011) and arXiv:1106.3052

7 / 29


Event selection (1/2)
Very similar in 2D and BDT analyses

Leptons
Ç Exactly one "tight" muon (electron) with pT (ET ) > 20 (30) GeV/c , | | < 2.1 (2.5) relative isolation Ir = Iabs /pT (ET ) < 0.05 (0.1), absolute isolation Iabs is sum of pT in cone of radius 0.3 around the lepton, excluding the lepton itself no jet within cone of radius 0.3 around the lepton Ç Veto additional "loose" leptons with pT (ET ) > 10 (15) GeV Ir < 0.2

8 / 29


Event selection (2/2)
Jets
Ç Exactly two anti-kT 0.5 jets with pT > 30 GeV/c , | | < 5 Ç Exactly one b -tagged jet (The b -tagging algorithm exploits the

impact parameter of the tracks associated with the jet)
Ç (2D analysis only) Exactly one b -veto ed jet Ç (BDT analysis only) (j1 , j2 ) < 3: excludes back-to-back

W +jets events poorly modeled with Pythia D6T tune

Invariant mass
Ç MT > 40 (50) GeV/c in muon (electron) channel, where

MT =
9 / 29

(pT (l ) + pT ( ))2 - (px (l ) + px ( ))2 - (py (l ) + py ( ))2


Data-driven QCD estimation
The analysis probes for a very specific kinematical region populated by the tails of QCD distribution only
Ç Amount of QCD estimated with

maximum likelihood fit to MT : F (MT ) = a Ç S (MT ) + b Ç B (MT )
Ç Shap e of non-QCD S (MT ) taken from

Å

simulation
Ç Shap e of QCD B (MT ) taken from

orthogonal data sample with inverted cut in Ir
10 / 29

e


Event yields after selection

11 / 29


2D analysis variables
Å+e
lj cos lj light jet Ç Due to V - A structure of weak interaction t -quark is almost 100% left-handed polarized

Ç This feature propagates to signal

Å+e

asymmetry in distribution over cos lj (angle calculated in t -quark rest frame)



lj

Ç Light-quark jet recoiling against much

more massive t -quark has non-central pseudorapidity distribution
12 / 29

lepton


2D signal extraction
The x-section is determined with extended maximum likelihood fit to (cos lj , |lj |) distribution Å

e

13 / 29


BDT analysis variables
Ç 37

variables used in total. They reflect different event properties:
kinematics and properties of the lepton and jets correlations between these objects properties of derived objects (t -quark, W , etc.) angular distributions between the original and derived objects global event characteristics (sphericity, total transverse energy, etc.)

Ç The most sensitive variables: lepton's pT ^ (invariant mass of system s W + j + j) dijet pT pT of the most b -tagged jet reconstructed t -quark mass
14 / 29

pT (Å)


BDT signal extraction
The x-section is determined from a binned likelihood fit to BDT output within a Bayesian approach. The systematics is treated as nuisance parameters, they are marginalized out through Markov chain Monte Carlo (MCMC)

Å

e

15 / 29


Systematics
Impact estimated through pseudoexperiments Main sources of systematics: Ç b -tagging
Ç Q 2 scaling Ç jet energy

scale

16 / 29


Combination and results in t -channel
Ç 2D and BDT analyses combined

through BLUE method
Ç The combined x-section:

83.6Á29.8 (stat.+syst.)Á3.3 (lumi.) pb
Ç Stat. significance is 3.7 (3.5)

w.r.t. expected significance +1.0 2.1-1.1 2.9+1.0 for 2D (BDT) -0.9
Ç 2D (BDT) sets 95% CL lower limit

|Vtb | > 0.62 (0.68), where |Vtd |, |Vts | |Vtb | and |Vtb | [0, 1] assumed
17 / 29


Single top tW -channel
b t g g t t W ? b W
Ç Analysis exploits leptonic decays of W:

t W Wb W l b l Ç Signal signature:
exactly two isolated leptons of opposite charge exactly one b -jet within acceptance large enough ET due to neutrinos Ç Main backgrounds: t ?: very similar kinematics (interference at NLO!), t high rate Z / ( ll ) + jets

g

18 / 29


Analysis strategy
Ç Three leptonic final states: ee , ÅÅ, e Å Ç Pure counting experiment Ç Rates for t ? and Z / ( ll ) + jets backgrounds estimated from t

data Analysis in tW -channel is based on the integrated luminosity of 2.1 fb-1 recorded during the first data-taking period of 2011 Details in CMS PAS TOP-11-022

19 / 29


Event selection (1/3)
Leptons
Ç Exactly two "tight" leptons of opp osite charge with pT > 20 GeV/c , | | < 2.4 (2.5) for a muon (electron) relative isolation Ir < 0.15 Ç Veto additional "loose" muons (electrons) with pT > 10 (15) GeV/c , | | < 2.5 Ir < 0.2 Ç Reject events with leptons of the same flavor and dilepton mass 81 < mll < 101 GeV/c 2 : Z veto mll < 20 GeV/c 2 : poor data-MC agreement Ç ET > 30 GeV in ee and ÅÅ channels (to reduce Z / ( ll ) + jets)

20 / 29


Event selection (2/3)
Jets
Ç Anti-kT algorithm with cones 0.5 Ç Veto "tight" leptons inside cone 0.3 around a jet Ç Exactly one jet with pT > 30 GeV/c , | | < 2.4 Ç The jet must b e b -tagged (the b -tagging algorithm exploits

reconstructed secondary vertex)
Ç Veto additional b -tagged jets with pT > 20 GeV/c

21 / 29


Event selection (3/3)
Transverse balance
Ç pT of system l1 + l2 + j + ET must b e less than 60 GeV/c Ç In e Å channel HT = pT (l1 ) + pT (l2 ) + pT (j )+ ET > 160 GeV

system pT

H

T

e Å final state, events passing leptonic step of selection and containing exactly one jet

22 / 29


Data-driven estimation of Z / and t ? t
Ç Z / + jets normalization estimated with the events failing Z veto:

N

estimated ll ,out

=

N

MC ll ,out MC Nll ,in

ÇN

observed ll ,in

1 - k ÇN 2

observed e Å,in

observed Ne Å,in accounts for non-peaking backgrounds (e.g. t ?) t factor k is responsible for difference in e /Å acceptance (taken from data from numbers of near-peak ee and ÅÅ events after leptonic selection only)

Ç t ? rate is related to one of the largest sources of uncertainty t two control regions are defined: 2 jets, 1 tag and 2 jets, 2 tags; they are by far dominated by t ? t numbers of events in these regions are fed into the statistical procedure constraining t ? normalization and b -tagging efficiency t
23 / 29


Event yields after selection (1/2)

24 / 29


Event yields after selection (2/2)
Event yields in signal and the two control regions
200 180 160 events / 2.1 fb 140 120 100 80 60 40 100 20 1 jet 1 tag 2 jet 1 tag 2 jet 2 tag 1 jet 1 tag 2 jet 1 tag 2 jet 2 tag 50
-1

CMS Preliminary, s = 7 TeV 2.1 fb-1, ee channel

data tW tt Z/ *+jets Other
-1

700 600 500 events / 2.1 fb 400 300 200

CMS Preliminary, s = 7 TeV 2.1 fb-1, eÅ channel

data tW tt Z/ *+jets Other

350 300
-1

CMS Preliminary, s = 7 TeV 2.1 fb-1, ÅÅ channel

data tW tt Z/ *+jets Other

250 200 150 100

events / 2.1 fb

1 jet 1 tag

2 jet 1 tag

2 jet 2 tag

Ç Z / + jets scaled to data-driven estimate Ç t ? scaled to result of the statistical pro cedure t

25 / 29


Systematics

"-" - systematics doesn't apply, " " - negligible

26 / 29


Results
Ç Measured x-section:

22+9 (stat.+syst.) pb -7
Ç Observed significance is 2.7 consistent with the exp ected

significance 1.8 Á 0.9

27 / 29


Conclusions
Ç Single top in the t -channel has been searched for with two

complementary analyses: one exploiting two characteristic angular variables, another one using a multivariate technique
Ç Both t -channel analyses (re)found an evidence of the signal

providing compatible results
Ç The analyses were combined to obtain a more precise measurement

of t -channel x-section
Ç The x-section was translated into the lower limit on |Vtb | Ç The tW -channel x-section was measured with a counting

experiment
Ç All the results are in go od agreement with the SM exp ectations

28 / 29


Thank you for your attention!

29 / 29


BACKUP SLIDES

30 / 29


Particle flow reconstruction
Ç Attempts to reconstruct every particle on individual basis Ç Better p erformance thanks to using all the appropriate sub detector

systems (e.g. jet performance can profit a lot from using tracker info)
Ç Every track or calorimeter energy dep osit is guaranteed to b e

associated with one particle at maximum no double counting

31 / 29


BDT validation in W-enriched (zero tag) sample

Å

e

32 / 29


Golden candidate of t -channel process, muon

33 / 29


Golden candidate of t -channel process, electron

34 / 29


system tW -analysis: pT and HT distributions
system pT (up) and HT (down) in ee -, e Å-, and ÅÅ-channels (from left to right)

Full selection except for requirements on these two variables 35 / 29