bb cross section: blessing

Sofia VallecorsaUniversity of Geneva

CDF note 8589

Outline

• Jet Corrections• Tagging efficiency• bb-fraction• Scale factor• Unfolding factors• Systematics• bb cross section

« SVT » JET• JetClu cone 0.4 SecVtx TIGHT tagged jet

• corrected ET>30 GeV, |η|<1.2• Matched to L3 jet with ET>20 GeV (δR<0.4)• Matched to L2 cluster ET>15 GeV, |η|<1.5 (δR<0.4)• Matched to SVT track, |d0|>120 µm,

pT>2GeV/c, |η|<1.2 (δR<0.4)• Matched to Si track |d0|>120 µm (δφ, δcurv)

On JET20: only 0.1% events TWO « SVT » jets have HIGH_PT_BJET bit off

Negligible effect compared to other systematics

bb cross section

!

d"

dEtd#=

N2SVT

$ f2SVT

2b

%# $ %ET$L $ &

2SVT $ SFjet1 $ SFjet2

• N2SVT: Events with two « SVT -tag » jets:• Leading jet: ET

corr>35 |η|<1.2• Second jet: ET

corr>32 |η|<1.2 (changed from pre-blessing!)• f2b

2SVT: bb-jets fraction− Use Secondary vertex masses− TFractionFitter standard routine− Templates for Σjet Mjet (Mjet : SecVtx Mass)

• ε2svt: Efficiency for 2 SVT-jets− Measured from PYTHIA bb

• SFjet1, SFjet2: Official scale factor, (run a check first):SF = 1.029 ± 0.009 (stat) ± 0.034 (syst)

Standard vertex quality cuts

Jet Energy CorrectionsNeed additional correction for b jets:

• Calculate average Et correction (relative to L5 generic corrections):• Match CAL jets to HAD jets (ΔR<0.4)• Fit correlation Et(CAL) vs Et(HAD)• Check fit consistency

R= (Et(HAD) - Et(CAL)corr) / Et(HAD)

• Calculate unfolding factor to correct Et smearing effect

Fit - SVT tagged jets

!

R =Et

HAD" Et

CORR

EtHAD

Correction is of the order of 5%:Cut on the SVT-tag jet L5 ET >30is raised to take it into account:

(L5+ add.c.) ET>35 GeV, (L5+ add.c.) ET>32 GeV (second jet)

Taggingefficiency

Efficiency for 2SVT-tag jets in event:• Calculate directly « per event »

efficiency• Systematics: b-quark content ofthe jet

bb fraction• Two SVT tagged jets• Sum-up the SecVtx masses of the two jets: one entry per event• Mass template:

• bb template• Bkgd template (bc, cc, cl, ll, bl ..)

• Fit fraction using TFractionFitter in different di-jet average Et bins

0.88 ± 0.04 %ET> 320.75 ± 0.09 %ET> 1000.84 ± 0.06 %(80,100)0.78 ± 0.05 %(60,80)0.86 ± 0.04 %(50,60)0.88 ± 0.03 %(40,50)0.87 ± 0.04 %(32,40)

B fractionET bin (GeV)

bb fraction

Systematics:• Background composition (light vs charm)• Track reconstruction ineff.

UnfoldingUse dijet PYTHIA bb-HEPG filtered sample( pt>10, pt >18, pt>40, pt>60, pt>90, pt>120 GeV) - Apply L5 jet energy corrections + additional corr. - Build JetClu calorimeter/hadronic jets - Identify each as a b-jet (match to a B-hadron) - Only central jet (|η|<1.2) Et>35 GeV,Et>32 GeVat the pre-blessing: Et>30, Et>30 GeV

Unfolding factors:

Invariant mass Δφ

Leading jet Et

Systematics

• Jet Corrections:– Standard systematics

• Tagging:– Efficiency: quark multiplicity in the jet– Fraction:

– Quark multiplicity in the jet– Background template composition– Track reconstruction efficiency– b-jets correlation in templates

• Unfolding:– Et dependence

Total Jet Corr. systematics• Use Jet Energygroup standard routine• Add systematics from specific corrections

(-1 σ) (+1 σ)ET bin (GeV)

14%17%(47,54)

20%17%(180,230)14%14%(140,180)7%18%(110,140)15%10%(99,110)18%18%(88,99)10%11%(78,88)13%18%(69,78)11%13%(61,69)13%15%(54,61)

15%20%(41,47)

17%19%(35,41)

Total Jet Corr. systematics (2)Invariant Mass & Δφ

Mass Δφ

Unfolding

• Compare Et spectrum indata and MC.

• Ratio Data/MC afterunfolding data crosssection

• Fit ratio to third orderpolinomial

• Reweight MC• Calculate new unfolding• Take the difference as a

syst. err.

Unfolding

Total systematics

Add all uncertaintiesin quadrature6% from lumi is not

included

ET bin (GeV)

-23%23%(47,54)

-21%24%(180,23)-19%23%(140,180)-21%25%(110,140)-16%19%(99,110)-21%24%(88,99)-18%20%(78,88)-19%24%(69,78)-19%21%(61,69)-22%23%(54,61)

-24%26%(41,47)

-23%25%(35,41)

NLO predictionUse MC@NLO + Herwig v6.510 + Jimmy v4.3

• bb events with:• CTEQ6m PDFs, µR = µF= sqrt(p2 +m2)• Pt (bquark) > 10 GeV/c (before pt >22 GeV/c)• |η|<1.75

• Full simulation 6.1.4mc (redecay using EvtGen)

• Stntuple dev243• Hadron level jet cross sections

UnfoldedCross section - Leading jet ET

UnfoldedCross section - Invariant Mass

UnfoldedCross section - Δφ

Summary

• Questions and remarks:• Changed Et cut• Unfolding systematics• MC@NLO generation cut• Total cross section

2259 ± 44 (stat)MC@NLO

2201 ± 29 (stat)Herwig

2140 ± 22 (stat)Pythia2360 ± 70 (stat) ± 133 (syst)Data

σ(bb) [pb]|η1,2|<1.2,ET,1>35 GeV, ET,2>32 GeV

CDF Run II Preliminary

More..

Cross section- linear scale

Reweighted MC samples

• Scale each ptHat bin forrelative luminosity

• Normalize to highest ptsample luminosity

3.97(nb)

20.5 (nb)

0.178

1.33

49.8

570

σ (µb)

Pt > 120

Pt > 90

Pt > 60

Pt > 40

Pt > 18

Pt > 10

PYTHIA bb

!

ˆ p t

HAD jet Et

CAL jet Et

Tagging efficiencyFound a correlation between SVT

tagged jetsSelect events with:

>2 jets ET>30 GeV, |η|<1.2>2 good SVT tracks

Order jets (ET,φ,η.. etc..)

Count:1. First jet is SVT-tagged -> Look at the second jet (RED)2. First jet is NOT SVT-tagged -> Look at the second jet (BLACK)

-->DIFFERENCE!NOT the same problem for« simple tag »

SVT tag

simple tag

Tagging efficiencyMain difference: matching to the track• Select « good » tracks (|d0|>120µm,pt>2 GeV/c)• Track-Track and Jet-Track show the SVT tracks

are « grouped »

ΔR(jet-track)ΔR(track-track)

bb fraction - fit32<ET<40 GeV 40<ET<50 GeV 50<ET<60 GeV

60<ET<80 GeV 80<ET<100 GeV ET>100 GeV

Scale Factor

• Use muon jets ( jet Et - up to 100-120 GeV)• MC: btop7a, btop8a Herwig dijet + HEPG filter e>8 GeV, µ pt> 9 GeV

• Run TrigSim. with correct beamline• good run list - but exclude some run (beamline)

• Data: inclusive µ sample (bmcl0d,bmcl0h)• goodrun list CMU+CMP+Si+SVT

• Measure efficiency for « SVT - TAG » jets (JetClu 0.4 - tight SecVtx - SVT track (d0>120 µm)

Scale factor• Select « good muon »:

• Pt > 9 GeV/c• CMU stub |dx|<3cm, CMP |dx|<5cm• Track |z0|<60cm, |z0 -Zvtx|<5cm• Iso(cone 0.4) > 0.1• Require track to hit all SVX layers

• Match good muon to jet (taggable, Et>20, |η|<1.2)• Correct jet Et, and P

• Select only highest Pt muon• Search for away jet (Et>20, |η|<2, |Δφ|>2rad, mass>1.5)

!

Ecorr = EL5

Eraw "2gev

Eraw

+ pt,µ

!

r P corr =

r P +

r p µ 1"

2gev

pµ

#

$ % %

&

' ( (

Scale factor: single tag

!

" =Naway

e

Naway

#fafter

fbefore

• Reconstruct muon pt relative to corrected jet axis• Fit b fraction in muon jet before and after tagging• Two componets FIT: b jets and

• c templates from µ sample (btop7a, btop8a) (or light jet templates from dijet pythia (jqcd1f, jqcd2f))

Scale factor

Use official value:• 1.029 ± 0.009 ± 0.034 (tight SecVtx + SVT track)• (1.066 ± 0.045) - (0.00050 ± 0.00107)* ET

MC:N (µ+,away+) = 1795N (away+) = 7952

ε = 0.225 +/- 0 .007

DATA:N (µ+,away+) = 1191N (away+) = 7485

ε = 0.22 +/- 0 .01

SF = 0.98 +/- 0.05

bb fraction - templatesTo have enough statistics in the templatesCombine couple of jets coming from different events.

CHECK: 2 jets in different events vs 2 jets in the same event32<ET<40 GeV 40<ET<50 GeV 50<ET<60 GeV

60<ET<80 GeV 80<ET<100 GeV ET>100 GeV

Effect is small+Kolmogorov test

Templates: cc content

0.86 ± 0.04Et>30

0.76 ± 0.09Et>100

0.83 ± 0.0780<Et<100

0.80 ± 0.0760<Et<80

0.86 ± 0.0550<Et<60

0.87 ± 0.0340<Et<50

0.88 ± 0.0432<Et<40

b fractionEt bin (GeV)

0.87 ± 0.04Et>30

0.79 ± 0.09Et>100

0.84 ± 0.0780<Et<100

0.79 ± 0.0760<Et<80

0.87 ± 0.0550<Et<60

0.88 ± 0.0340<Et<50

0.88 ± 0.0432<Et<40

b fractionEt bin (GeV)

Double cc content Half cc content

b multiplicity in the jetEff. is different according to b

content in the jet

In 2SVT-tagged jets eventsselect couples with:• One jet including 2b quarks• Two jets including 2b quarks

< 1.5 %

< 3%

Quark multiplicity:b fraction (mass templates):

0.86 ± 0.04 (0.88) Et>30

0.79 ± 0.09 (0.75) Et>100

0.83 ± 0.07 (0.84)(80,100)

0.78 ± 0.07 (0.78)(60,80)

0.86 ± 0.05 (0.86)(50,60)

0.88 ± 0.03 (0.88)(40,50)

0.87 ± 0.04 (0.87)(32,40)

b fractionET bin(GeV)

Double fraction of bb jets events in mass templates and fit::(same check on jet including >1c quark)

32<ET<40 GeV 40<ET<50 GeV 50<ET<60 GeV

60<ET<80 GeV 80<ET<100 GeV ET>100 GeV

Templatescomposition

• Light jets contribution tomass templates

For light jet to c jetproportion rely on Pythia

-> change light jet fraction

2%(180,230)

2%(140,180

2%(110,140)

2%(99,110)

1%(88,99)

1%(78,88)

0.5%(69,78)

0.5%(61,69)

0.7%(54,61)

0.5%(47,54)

0.5%(41,47)

0.5%(35,41)

Syst.err.ET bin (GeV)

Other systematicsfor b fraction• Track reconstruction

efficiency (CDF 6894)-> 3% shift in mass template

7%(180,230)

7%(140,180

8%(110,140)

8%(99,110)

8%(88,99)

8%(78,88)

7%(69,78)

6%(61,69)

6%(54,61)

5%(47,54)

5%(41,47)

5%(35,41)

Syst.err.ET bin (GeV)