The Heavy Quark Search at the LHC

BH, JHEP 0708 (2007) 069

BH, JHEP 0703 (2007) 063

The Heavy Quark Search at the LHC

Backgrounds: estimating and suppressing

• tt, multijets, ...

• jet mass technique

• NLO e!ects

• matrix elements for extra jets

• initial state radiation

• compare event generators

Basic picture

• sequential fourth family (with a heavy !) with at least some CKM mixing

• fourth family quarks with mass ! 600 GeV

pp " t!t! " W+W"bb

and/or

pp " b!b! " W+W"tt

• since colored fermions are involved, cross sections are decent at the LHC

"LO ! 900 fb

"NLO ! 1400 fb

Basic picture

• sequential fourth family (with a heavy !) with at least some CKM mixing

• fourth family quarks with mass ! 600 GeV

pp " t!t! " W+W"bb

and/or

pp " b!b! " W+W"tt

• since colored fermions are involved, cross sections are decent at the LHC

"LO ! 900 fb

"NLO ! 1400 fb

• 600 GeV roughly corresponds to when we stop talking about a light Higgs

Basic picture

• sequential fourth family (with a heavy !) with at least some CKM mixing

• fourth family quarks with mass ! 600 GeV

pp " t!t! " W+W"bb

and/or

pp " b!b! " W+W"tt

• since colored fermions are involved, cross sections are decent at the LHC

"LO ! 900 fb

"NLO ! 1400 fb

• concentrate on first process

• assuming mt! < mb! , or

• |Vt!b| > 0.01, 0.004, 0.001 for mt! ! mb! = 70, 50, 30

• 600 GeV roughly corresponds to when we stop talking about a light Higgs

• W ! jj from single jet invariant mass

energy deposit in calorimeter cell

pµ1 + pµ

2

!

i

pµi

J. M. Butterworth, B. E. Cox and

J. R. Forshaw 2002

P. Savard 1997

I. Borjanovic et al. 2005

Identifying energetic and isolated W ’s

• W ! jj from single jet invariant mass

energy deposit in calorimeter cell

pµ1 + pµ

2

!

i

pµi

b jetb jet

from boosted t from t!

J. M. Butterworth, B. E. Cox and

J. R. Forshaw 2002

P. Savard 1997

I. Borjanovic et al. 2005

Identifying energetic and isolated W ’s

• suppression of tt background

Sample jet mass plot

60 70 80 90 100 110

10

20

30

40

signal

background

GeV

Njet

jet mass

Sample jet mass plot

W-jet definition

• invariant mass of jet within ! 10 GeV of the peak

• peak can be experimentally determined—need not be exactly at MW

• “splash out” and “splash in” e!ects

60 70 80 90 100 110

10

20

30

40

signal

background

GeV

Njet

jet mass

Detector Simulation

• PGS4 with cone-based jet finder option

ATLAS ! parameter set name

81 ! eta cells in calorimeter

63 ! phi cells in calorimeter

0.1 ! eta width of calorimeter cells |eta| < 5

0.099733101 ! phi width of calorimeter cells

0.01 ! electromagnetic calorimeter resolution const

0.1 ! electromagnetic calorimeter resolution * sqrt(E)

0.8 ! hadronic calolrimeter resolution * sqrt(E)

0.2 ! MET resolution

0.00 ! calorimeter cell edge crack fraction

cone ! jet finding algorithm (cone or ktjet)

3.0 ! calorimeter trigger cluster finding seed threshold (GeV)

0.5 ! calorimeter trigger cluster finding shoulder threshold (GeV)

0.6 ! calorimeter kt cluster finder cone size (delta R)

1.0 ! outer radius of tracker (m)

2.0 ! magnetic field (T)

0.000005 ! sagitta resolution (m)

0.98 ! track finding efficiency

0.30 ! minimum track pt (GeV/c)

2.5 ! tracking eta coverage

3.0 ! e/gamma eta coverage

2.4 ! muon eta coverage

2.0 ! tau eta coverage

pp ! t"t" ! W+W#bb

Event Selection

• scalar pT sum HT > !HT

• one b-tag jet with pT > !b

• one W -jet

pp ! t"t" ! W+W#bb

Event Selection

• scalar pT sum HT > !HT

• one b-tag jet with pT > !b

• one W -jet

• choose !HT= 2mt! and !b = mt!/3

• for real data keep !HT/!b = 6 and scan to optimize signal

pp ! t"t" ! W+W#bb

b-tagging e!ciencies

• account for large pT ’s, where e!ciencies worsen

1/2 for b’s

1/10 for c’s

1/30 for light quarks and gluons

• assume to vanish for pseudorapidity |!| > 2

Event Selection

• scalar pT sum HT > !HT

• one b-tag jet with pT > !b

• one W -jet

• choose !HT= 2mt! and !b = mt!/3

• for real data keep !HT/!b = 6 and scan to optimize signal

HT distribution

• signal peaks at high HT ! 2mt!

• why not just look for the signal bump on the tail?

• t! mass reconstruction is essential

• consider invariant mass of all W -b pairs

• pairs of plots: W -mass plot and t!-mass plot

1200 1400 1600 1800 2000 2200

1000

2000

3000

4000

signal

HTGeV

HT distribution

• signal peaks at high HT ! 2mt!

• why not just look for the signal bump on the tail?

Event Generators

• MC@NLO-Herwig

• stand-alone Herwig

• Alpgen-Herwig

• Alpgen-Pythia

• stand-alone Pythia

Event Generators

• MC@NLO-Herwig

• stand-alone Herwig

• Alpgen-Herwig

• Alpgen-Pythia

• stand-alone Pythia

• in each case we use the same tool to calculate both signal and background

Backgrounds

• consider the tt irreducible background first

• then consider QCD multijet backgound

• dealing with this basically eliminates other backgrounds:(W/Z) + jets, bb + jets, (W/Z)bb, (WW/ZZ/WZ) + jets, and (t/t)(b/b)W

Event Generators

• MC@NLO-Herwig

• stand-alone Herwig

• Alpgen-Herwig

• Alpgen-Pythia

• stand-alone Pythia

• in each case we use the same tool to calculate both signal and background

MC@NLO-Herwig

• good for getting absolute rates

• no description of fourth family

• use mt = 600 GeV and t ! bW to model t! ! bW

• produces some negative weights (" 15%)

• use CTEQ6M and scale to 2.5 fb"1

MC@NLO-Herwig

• good for getting absolute rates

• no description of fourth family

• use mt = 600 GeV and t ! bW to model t! ! bW

• produces some negative weights (" 15%)

• use CTEQ6M and scale to 2.5 fb"1

stand-alone Herwig

• increases background and decreases signal!

!"#$%&'(

!"#$)*#%&'(

!"# "## $"# %### %!"# %"##

"####

%#####

%"####

!#####

!"####

+,-!"## !$## !%## !&## "### ""##

!###

"###

'###

$###

!"#$%&'(

!"#$)*#%&'(

+",-./

012

HT distribution

• NLO gives the K-factor enhancement

• but it also suppresses the tail, and thus the background

Alpgen-Herwig

• choose!

s/2 as the renormalization scale—ignore K-factors

• generate samples involving tt + 0, tt + 1 and tt + 2 partons

• use MLM parton-jet matching scheme

• model t! decay the same way and use CTEQ6M again

Alpgen-Herwig

• choose!

s/2 as the renormalization scale—ignore K-factors

• generate samples involving tt + 0, tt + 1 and tt + 2 partons

• use MLM parton-jet matching scheme

• model t! decay the same way and use CTEQ6M again

Alpgen-Pythia

• good agreement with Alpgen-Herwig and MC@NLO-Herwig

Alpgen-Pythia

• good agreement with Alpgen-Herwig and MC@NLO-Herwig

• either way of going beyond leading order increases signal to background

Stand-alone Pythia

• initial-state radiation becomes an issue

• models of underlying event(old with Q2 ordered showers and new with pT ordered showers)

• note that MSTP(68)=3 is default (“power showers”)

• ISR has very high phase space cuto! in tt production

• MSTP(68)=0 gives a phase space cuto! more in line with the hard process itself

HT

MSTP(68)=3

MSTP(68)=0

Stand-alone Pythia

• initial-state radiation becomes an issue

• models of underlying event(old with Q2 ordered showers and new with pT ordered showers)

• note that MSTP(68)=3 is default (“power showers”)

• ISR has very high phase space cuto! in tt production

• MSTP(68)=0 gives a phase space cuto! more in line with the hard process itself

DW (R. Field) and S0A (Sandho!-Skands) tunes

• share value of PARP(90)

mt! = 800 GeV

• increase !top5 and !b by 4/3

• MC@NLO-Herwig and Pythia(DW)

kT jet vs. cone jet finder

• Pythia(DW)

• Alpgen drastically reduces sensitivity to ISR

HT tails again: sensitivity to ISR

• turning ISR o! in stand-alone Pythia shows it large e!ect, and thus modeldependence

tt + jets with pminT

= 100 GeV W + jets with pminT

= 150 GeV

• Alpgen controls bulk of showering

• standardize on tune QW and CTEQ6.1

• let Alpgen generate samples for 0, 1 and 2 extra hard partons, using the MLMjet-parton matching scheme

• maximum jet pseudorapidity 2.5 and the minimum jet separation 0.7

QCD multijet background

• problem is that the cross section is enormous and only a tiny fraction survivesin the signal region—how to generate enough events?

• use Alpgen to “divide and conquer”

• generate 2, 3, and 4-jet samples, with relative sizes determined by a pminT

• 3-jet sample dominates 2-jet sample (in signal region) if pminT

is not toolarge

• cross sections are smaller if pminT

is large

• compromise by choosing pminT

= 200 GeV

• 2-jet sample still has enormous cross section

• remove b-tag to generate enough events, and then compare to the 3 + 4 jet

samples

• not bad then to only keep 3 + 4 jet samples with b-tag

dealing with QCD background

• !R < 2.5 between W and b

• look for the leptonic decay of the other W in the signal events

• consider a loose cut and a tight cut

• (isolated electron or muon) or (> 200 GeV missing energy)

• (isolated electron or muon) and (> 50 GeV missing energy)

dealing with QCD background

• !R < 2.5 between W and b

• look for the leptonic decay of the other W in the signal events

• consider a loose cut and a tight cut

• (isolated electron or muon) or (> 200 GeV missing energy)

• (isolated electron or muon) and (> 50 GeV missing energy)

try without b-tag

• made possible by the lepton/missing energy cuts

• may be useful for early running of LHC

• covers the possibility that b is not in the dominant decay

try without b-tag

• made possible by the lepton/missing energy cuts

• may be useful for early running of LHC

• covers the possibility that b is not in the dominant decay

b!b! production

• can add to previous signal through

pp " b!b! " W+W"t!t! " W+W"W+W"bb

• assume mb! = 700 GeV and mt! = 600 GeV

b!b! production

• can add to previous signal through

pp " b!b! " W+W"t!t! " W+W"W+W"bb

• assume mb! = 700 GeV and mt! = 600 GeV

pp ! b"b" ! W+W#tt

• use jet masses to identify both the t and the W—no b-tags again

• problem is that a single cone size is not optimal for either

• choose cone size 0.8

• t-jet: mjet > 100, pT > 300 GeV

• !R < 2 constraint between W and t jets

• assume mb! = 600 GeV

pp ! b"b" ! W+W#tt

• use jet masses to identify both the t and the W—no b-tags again

• problem is that a single cone size is not optimal for either

• choose cone size 0.8

• t-jet: mjet > 100, pT > 300 GeV

• !R < 2 constraint between W and t jets

• assume mb! = 600 GeV

Summary

• a sequential fourth family wrapped up with electroweak symmetry breaking isa predictive scenario

• the process pp ! t!t! ! W+W"bb is most promising

• jet mass technique is a promising way to identify W ’s and enhance S/B

• improved matrix elements from MC@NLO and Alpgen give results that are inagreement (and encouraging for S/B)

• Alpgen reduces sensitivity to initial state radiation

• why not develop tunes after incorporating the matrix elements?

Summary

• a sequential fourth family wrapped up with electroweak symmetry breaking isa predictive scenario

• the process pp ! t!t! ! W+W"bb is most promising

• jet mass technique is a promising way to identify W ’s and enhance S/B