Intro

Method

Plots

Results

Conclusions

Backup

Search for direct top squark pair productionin events with two tau leptons

with the ATLAS detector

Ewan Hill

University of Victoria + TRIUMF

June 15 2016

Ewan Hill

1 / 14

Intro

Why

SUSY

ATLAS SUSY

Signal

Method

Plots

Results

Conclusions

Backup

ATLAS is searching for “new physics” to explainsome open questions about the Standard Model

Dark matter

Hierarchy problemEwan Hill

2 / 14

Intro

Why

SUSY

ATLAS SUSY

Signal

Method

Plots

Results

Conclusions

Backup

Supersymmetry to the rescue

Supersymmetry fixing SM:

I G̃ = dark matter candidate

I top squark (“stop”, t̃) helps with hierarchy problemEwan Hill

3 / 14

Intro

Why

SUSY

ATLAS SUSY

Signal

Method

Plots

Results

Conclusions

Backup

Many SUSY searches performed by ATLAS,including for top squarks

https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/CombinedSummaryPlots/SUSY/

Ewan Hill

4 / 14

Intro

Why

SUSY

ATLAS SUSY

Signal

Method

Plots

Results

Conclusions

Backup

Search for top squark to tau slepton signal over tt̄background

Signal (GMSB)

t̃

t̃

τ̃

τ̃

p

p

b ν

G̃

τ

νb

G̃

τ

σ(t̃t̃∗) ∼ 0.5 pb, m(t̃) = 500 GeV

Dominant background

t

t

W

W

p

p

b `

ν

b

ν

τhad

σ(tt̄) ∼ 8× 102 pb

I G̃/ν = undetected → EmissT

I Final state objects:

I 2 b-jetsI Emiss

TI 0− 2 e or µI 2− 0 τhad

Channels:

I lepton–lepton = 2 e or µ

I lepton–hadron =1 τhad and (1 e or µ)

I hardon–hadron = 2 τhadEwan Hill

5 / 14

Intro

Method

Analysis method

Variables

Signal regions

SR cuts

Plots

Results

Conclusions

Backup

Selection cuts isolate regions of phase space tomaximize/minimize signal to background

Just count # events in phase space regions:

I Control: scale simulation to data for dominant backgrounds

I Validation: check scaling

I Signal: maximized signal to background

Analysis blinded to reduce human bias

Data set from 2012: 8 TeV, 20fb−1Ewan Hill

6 / 14

Intro

Method

Analysis method

Variables

Signal regions

SR cuts

Plots

Results

Conclusions

Backup

Several kinematic variables separate signal frombackground

Signal

t̃

t̃

τ̃

τ̃

p

p

b ν

G̃

τ

νb

G̃

τ

Dominant background

t

t

W

W

p

p

b `

ν

b

ν

τhad

I Number of b-jets

I Probes of decaying particles’ masses, e.g. mT2 (b`,bτ)

I mT2(a,b)=

√min

qaT

+qbT

=pmissT

(max[m2T(pa

T,qaT),m2

T(pbT,qb

T)])

I Sums/ratios of momenta e.g.

I HT = pjet1T + pjet2T

Ip`T+pτT∑alli pT(i)Ewan Hill

7 / 14

Intro

Method

Analysis method

Variables

Signal regions

SR cuts

Plots

Results

Conclusions

Backup

Design analyses around unknown signal masses:m(t̃) and m(τ̃)

m(t̃) =? m(τ̃) =? m(G̃)� 1 GeVEwan Hill

8 / 14

Intro

Method

Analysis method

Variables

Signal regions

SR cuts

Plots

Results

Conclusions

Backup

Main cuts separating each channel/SR aremT2(t)→ m(t̃) and mT2(W )→ m(τ̃)

I SRHM mT2(W ) > 120 GeV vs SRHH mT2(W ) > 50 GeV

I SRHM mT2(t) > 180 GeV vs SRHH all top squark masses

I 2 leptons mT2(W ) and jet pT & multiplicity

I SRLM mT2(t) < 60 GeV vs SRHH all top squark massesEwan Hill

9 / 14

Intro

Method

Plots

mT2 (b`, bτ)SRL

mT2 (`, τ)SRH

Results

Conclusions

Backup

t mass-probing variable

notNeededQuestionMark

0 100 200 300 400 500 600

Eve

nts

/ 20

GeV

-310

-210

-110

1

10

210

310

410

510

610-1 = 8 TeV, 20.3 fbs

ATLAS

Preselection SRLM

Data 2012SM Backgroundtop fake taustop true taus

ν l→W Other

)=(391,148) GeV1τ∼,1t~m(

)=(195,87) GeV1τ∼,1t~m(

) [GeV]hadτ(bl,bT2m

0 100 200 300 400 500 600Dat

a / M

C

00.5

11.5

2

≥

mT2(t) = mT2 (b`,bτ)

Eur. Phys. J. C 76, 1-30, 2016

Ewan Hill

10 / 14

Intro

Method

Plots

mT2 (b`, bτ)SRL

mT2 (`, τ)SRH

Results

Conclusions

Backup

W mass-probing variable

notNeededQuestionMark

0 50 100 150 200 250 300

Eve

nts

/ 20

GeV

-210

-110

1

10

210

310

410

510

610-1 = 8 TeV, 20.3 fbs

ATLAS

Preselection SRHM

Data 2012SM Backgroundtop fake taustop true taus

ν l→W Other

)=(391,148) GeV1τ∼,1t~m(

)=(195,87) GeV1τ∼,1t~m(

) [GeV]hadτ(l,T2m

0 50 100 150 200 250 300Dat

a / M

C

00.5

11.5

2

≥

mT2(W ) = mT2 (`, τ)

Eur. Phys. J. C 76, 1-30, 2016

Ewan Hill

11 / 14

Intro

Method

Plots

Results

SR counts

CombinedExclusion

Conclusions

Backup

No excesses seen in any signal region

Analysis Expected bkgd Observed(nom. ± stat. & syst.)

Lepton–hadron low massSRLM 22.1± 4.7 20Lepton–hadron high massSRHM 2.1± 1.5 3Hadron–hadronSRHH 3.1± 1.2 3

2-leptons channel has many SRs : no excess.

Eur. Phys. J. C 76, 1-30, 2016Ewan Hill

12 / 14

Intro

Method

Plots

Results

SR counts

CombinedExclusion

Conclusions

Backup

Use SR/channel that gives best expected CLs forcombination

[GeV]1

t~m

200 300 400 500 600 700

[GeV

]1τ∼

m

100

200

300

400

500

600

700-1 = 8 TeV, 20 fbs

) = 1G~

τ →1

τ∼) = 1, BR(νb1τ∼ → 1t~

production, BR(1t~1t

~

ATLAS

)theory

SUSYσ1 ±Observed limit (

)expσ1 ±Expected limit (

LEP limit

All limits at 95% CL

Combination

Num

bers

indi

cate

bes

t exp

ecte

d S

R

1τ∼ + m

b < m1t~m

1 2 2 2 2 2 2 2 5 2 2

2

2 2 5 5 2 2 2 2 5 2

35 5 5 5 5 2 2 5 5 53

35 5 5 2 5 5 5 53

33 5 5 5 5 5 5 5

35 5 5 5 5 5

35 5 5 5

3 5 5 5 5

5

5 5

5

1 2 2 2 2 2 2 2 5 2 2

2

2 2 5 5 2 2 2 2 5 2

35 5 5 5 5 2 2 5 5 53

35 5 5 2 5 5 5 53

33 5 5 5 5 5 5 5

35 5 5 5 5 5

35 5 5 5

3 5 5 5 5

5

5 5

5

3=lepton–lepton 5=lepton–hadron high mass1=lepton–hadron low mass 2=hadron–hadron

Eur. Phys. J. C 76, 1-30, 2016

Ewan Hill

13 / 14

Intro

Method

Plots

Results

Conclusions

Summary

Backup

Summary

I Top squark could help fix an open problem in the StandardModel

I Low and high t̃ masses probed as a function of the τ̃ mass

I No excesses observed.

I Set limits on m(t̃) as a function of m(τ̃)

I Heaviest top squark mass excluded ∼ 660 GeV

I Expecting ∼ 25fb−1 this year

I Should be sensitive to m(t̃) ∼ 800 GeV

Ewan Hill

14 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

Region cuts are all orthogonal to each other

Region N⌧had Nµ Njet Nb�jet EmissT ��(j1,2, p

missT ) mT2(⌧had, `) msum

T (⌧had, `)

SRHH 2 0 � 2 � 1 > 150 GeV � 0.5 > 50 GeV > 160 GeVCRHHTop 1 1 � 2 � 1 > 100 GeV � 0.5 - [70,120] GeV

CRHHWjets 1 1 � 2 0 > 100 GeV � 0.5 < 40 GeV [80,120] GeVVRHHTop 1 1 � 2 � 1 > 120 GeV � 0.5 < 40 GeV [120,140] GeV

VRHHWjets 1 1 � 2 0 > 120 GeV � 0.5 < 40 GeV [120,150] GeV

CRHHQCD � 2a 0 � 2 � 1 > 150 GeV 0.5b - -

aFor the multi-jet control region (CRHHQCD), no identification criteria are applied to tau leptons.bThe �� requirement only applies to the sub-leading jet j2.

Region Nb−jet HT/meffp`T+p

τhadT

meffmT2(b`, b) mT2(b`, bτhad) mT(`, p

missT ) meff

SRLM ≥ 2 < 0.5 > 0.2 < 100 GeV < 60 GeV - -CRTtLM ≥ 2 - > 0.2 < 100 GeV 110− 160 GeV > 100 GeV -CRTfLM ≥ 2 - > 0.2 < 100 GeV 110− 160 GeV < 100 GeV -CRWLM 0 < 0.5 > 0.2 - - > 40 GeV < 400 GeVVRTLM ≥ 2 > 0.5 > 0.2 < 100 GeV 60− 110 GeV - -

Region Nb−jet EmissT meff HT/meff mT2(b`, bτhad) mT2(`, τhad) mT(`, p

missT )

SRHM ≥ 1 > 150 GeV > 400 GeV < 0.5 > 180 GeV > 120 GeV -CRTtHM ≥ 1 > 150 GeV > 400 GeV < 0.5 > 180 GeV 20-80 GeV > 120 GeVCRTfHM ≥ 1 > 150 GeV > 400 GeV < 0.5 > 180 GeV 20-80 GeV < 120 GeVCRWHM 0 > 150 GeV > 400 GeV < 0.5 - 20-80 GeV 40-100 GeVVRHM ≥ 1 < 150 GeV > 400 GeV < 0.5 > 180 GeV > 80 GeV -

Ewan Hill

15 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

mT2 Upper Limits

t̃

t̃

τ̃

τ̃

p

p

b ν

G̃

τ

νb

G̃

τ

mT2 Max Value for Max Value forSignal tt̄ background

mT2 (b`,bτ) top squark mass top mass

amT2 (b`,b) top mass

mT2 (`, τ) If chargino not virtual W mass(not true here):chargino mass

Ewan Hill

16 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

Fake and τ background estimation done with MC

I Fit for tt̄(τ true), tt̄(τfake), and W+jets seperately in thebackground-only fit with a CR for each.

I W+jets CRs: Use b-veto to isolate.

I tt̄ CRs: Use an mT2 variable to isolate tt̄ and then m`T to distinguish

true and fake taus.

I The same-sign method (tau has same sign charge as e/µ) was alsotested

Summary of results:

I Both methods give similar results to within the uncertainties.

I MC method was chosen because of a lack of statistics in thesame-sign method.

Ewan Hill

17 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

Exclusion plot for lepton-hadron Low Mass channel

[GeV]1

t~m

160 170 180 190 200 210 220 230 240 250

[GeV

]1τ∼

m

50

100

150

200

250

300

350

-1 = 8 TeV, 20.3 fbs

) = 1G~

τ →1

τ∼) = 1, BR(νb1τ∼ → 1t~

production, BR(1t~1t

~

ATLAS

)theory

SUSYσ1 ±Observed limit (

)expσ1 ±Expected limit (

LEP limit

All limits at 95% CL

Lepton-hadron SRLM

1τ∼ + mb < m

1t~m

Eur. Phys. J. C 76, 1-30, 2016Ewan Hill

18 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

Exclusion plot for lepton-hadron High Mass channel

[GeV]1

t~m

200 300 400 500 600 700

[GeV

]1τ∼

m

100

200

300

400

500

600

700-1 = 8 TeV, 20.3 fbs

) = 1G~

τ →1

τ∼) = 1, BR(νb1τ∼ → 1t~

production, BR(1t~1t

~

ATLAS

)theory

SUSYσ1 ±Observed limit (

)expσ1 ±Expected limit (

LEP limit

All limits at 95% CL

Lepton-hadron SRHM

1τ∼ + m

b < m1t~m

Eur. Phys. J. C 76, 1-30, 2016Ewan Hill

19 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

Exclusion plot for hadron-hadron channel

[GeV]1

t~m

200 300 400 500 600 700

[GeV

]1τ∼

m

100

200

300

400

500

600

700-1 = 8 TeV, 20.1 fbs

) = 1G~

τ →1

τ∼) = 1, BR(νb1τ∼ → 1t~

production, BR(1t~1t

~

ATLAS

)theory

SUSYσ1 ±Observed limit (

)expσ1 ±Expected limit (

LEP limit

All limits at 95% CL

Hadron-hadron

1τ∼ + m

b < m1t~m

Eur. Phys. J. C 76, 1-30, 2016Ewan Hill

20 / 14

Intro

Method

Plots

Results

Conclusions

Backup

Regional Cuts

mT2 UpperLimits

Real/fake τEstimation

SRLM Exclusion

SRHM Exclusion

SRHH Exclusion

SRLL Exclusion

Exclusion plot for lepton-lepton channel

[GeV]1

t~m

200 300 400 500 600 700

[GeV

]1τ∼

m

100

200

300

400

500

600

700-1 = 8 TeV, 20.3 fbs

) = 1G~

τ →1

τ∼) = 1, BR(νb1τ∼ → 1t~

production, BR(1t~1t

~

ATLAS

)theory

SUSYσ1 ±Observed limit (

)expσ1 ±Expected limit (

LEP limit

All limits at 95% CL

Lepton-lepton

1τ∼ + m

b < m1t~m

Eur. Phys. J. C 76, 1-30, 2016Ewan Hill

21 / 14