top mass in t-tbar 6jetsStatus Report

Physics and Astronomy University of Victoria

British Columbia, Canada

Top meetingCERN, 21 Feb 2007

Michel Lefebvre

Keith Edmonds M.Sc. project Much help from Rolf Seuster

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 2

Introduction and goalsFully hadronic t-tbar decays

Attempt at measuring the top mass First see how far one can go without b-tagging

• only calorimetry

Huge combinatorics and QCD backgrounds

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 3

SamplesSignal sample

5204, athena 11.0.5 AOD• 270k events• Cone4TowerParticleJets

Background samples QCD multijet, alpgen 2.06, atlfast, athena 12.0.31

• home production– generation: pTmin=25 GeV and Rmin=0.5

– matching: ETmin=25 GeV and Rjet=0.5

• 2, 3, 4, 5, 6(inc.) jets

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 4

SelectionsEvent selection

at least 6 jets in || < 3 at least 6 jets satisfying graded pT cuts

• 85, 65, 50, 40, 40, 40 GeV

Hypotheses for each event only consider the 6 highest pT jets

consider all 90 combinations for t-tbar topology• 10 distinct triplet pairs, each triplet with 3 possible W

– assumes one cannot distinguish charges and quark types

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 5

SelectionsHypotheses treatment

build both W and top 4-vectors from jets

Hypotheses selection MW window cut: 40 < MW < 120 GeV

Mtop window cut: |Mtop| < 100 GeV

top pT cut: pT > 100 GeV

Perform kinematic fitHypotheses selection

valid kinematic fit full accurate cov matrix kinematic fit 2 cut: 2 < 10

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 6

PT distribution: signal

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 7

PT distribution: 6 partons background

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 8

Kinematic fit: input parametersExperimental jets

initial 4-vectors we set

• should not do this when considering merged partons

each jet therefore characterized by the triplet

• for now assume no b-jet tagging

,m m mj j jp E p

0,1...5j

0mjm

, ,m m mj j jE m m

j jp E

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 9

Kinematic fit: simple chi2consider a simple chi2 fit

assume jet direction exact constraint both W masses demand that both tops have the same mass consider Gaussian for W and top masses

• easy to improve to likelihood and Breit-Wigner if needed

for a given jet hypothesis consider the chi2

• the free parameters are the six fit jet energies

2 22 25

2

0

W W t tbarW W

W W t

-

2

mj j

Ejj

M M M ME E M M

E

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 10

Kinematic fit start each fit with in the fit use

for use the same values as used in the generation

use the best possible estimate of the jet energy error

• Here just use

mj jE E

ˆ, mj j j jp E E p

20 1WM p p 22 3W

M p p

20 1 4tM p p p 22 3 5tbarM p p p

, ,W W t M

,Ej j jE 100% 10%

GeV( )E

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 11

Cut flow

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 12

Jet-parton matchingStudy events with “true” jet hypothesisMatching criteria

For each parton, look for a matching jet• restrict search in a region limited by Rmax = 0.2

• keep the closest jet in this region

Demand that a jet be matched only once• matching efficiencies depends on interparton distances

– same top combinationsR(u-b) = 2.220 ± 0.002 R(ubar-bbar) = 2.219 ± 0.002

R(u-dbar) = 2.008 ± 0.002 R(ubar-d) = 2.010 ± 0.002

R(dbar-b) = 2.037 ± 0.002 R(d-bbar) = 2.036 ± 0.002

– other 9 combinations R 2.40

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 13

Jet-parton matchingMatching efficiencies

• u 76.1% ± 0.1% ubar 76.2% ± 0.1%• dbar 70.2% ± 0.1% d 70.2% ± 0.1%• b 79.1% ± 0.1% bbar 79.2% ± 0.1%

Jet-parton distances

R < 0.2 is a reasonable criteria

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 14

Need to characterize kinematics uncertainties required for the kinematic fit

Consider 1-to-1 matched jet-parton pairs study the quantities

here focus on a few aspects

Jet kinematics uncertainties

mE E E

m

m

E rmsE E

rms

rms

statistics in the E, plane

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 15

Jet kinematics: energy

energy bias energy resolution

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 16

Jet kinematics: energy bias

?

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 17

Jet kinematics: resolution

0 2

1 GeVGeV

p pp

E EE

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 18

Signal truth kinematic fit results

2 ×1104 entries

2 ×1038 entries

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 19

Signal truth kinematic fit results

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 20

Cut flow: signal

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 21

Selections: signal top massall hypotheses valid fit hypotheses

1-to-1 matched jet-parton pairs (overlaid, not stacked)

with combinatorics recall that 10% of all hypotheses combinatorics have correct top triplets

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 22

Selections: signal top massfinal hypotheses (after chi2 cut)valid fit hypotheses

chi2 cut

1-to-1 matched jet-parton pairs (overlaid, not stacked)

with combinatorics

Need to revisit combinatorics strategyperhaps keep only one (best?) combination per event?

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 23

Selections: signal W massall hypotheses valid fit hypotheses

1-to-1 matched jet-parton pairs (overlaid, not stacked)

with combinatorics recall that 10% of all hypotheses combinatorics has correct top triplets

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 24

Cut flow: all samples

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 25

Selections: top massall hypotheses valid fit hypotheses

signal 1-to-1 matched jet-parton pairs (overlaid, not stacked)

signal with combinatorics QCD 6 partons QCD 5 partons QCD 4 partons

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 26

Selections: top massvalid fit hypotheses

signal 1-to-1 matched jet-parton pairs (overlaid, not stacked)

signal with combinatorics QCD 6 partons QCD 5 partons QCD 4 partons

chi2 cut

final hypotheses (after chi2 cut)

Need larger background samples!!

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 27

Selections: W massall hypotheses valid fit hypotheses

signal 1-to-1 matched jet-parton pairs (overlaid, not stacked)

signal with combinatorics QCD 6 partons QCD 5 partons QCD 4 partons

Michel Lefebvre, 2007/02/21 Status report: top mass in t-tbar -> 6 jets 28

Possible way forward... need b-tagging explore existing background samples

SUSY samples?

revisit jet-parton hypotheses strategy study energy biases

can they be extracted from the study of semileptonic decays? how do they affect the kinematic fits?

improve kinematic fit better energy resolution parameterization

• obtain this in situ?

allow jets direction to vary move to likelihood with Breit-Wigner

tackle b-jet energy scale obtain this in situ?