search for new physics at cms using the razor kinematic variables

+

Search for new physics at CMS using the razor kinematic variables

Will Reece (CERN) for the CMS Collaboration

+Introduction

n  Presenting four analyses that use the Razor variables n  All use the 2011 CMS 7TeV pp dataset

n  Razor Inclusive: 4.4* fb-1: PAS-SUS-12-005 n  https://twiki.cern.ch/twiki/bin/view/CMSPublic/PhysicsResultsSUS12005

n  Razor-B: 4.6 fb-1: PAS-SUS-11-024 n  https://twiki.cern.ch/twiki/bin/view/CMSPublic/ResultSUS11024

n  Razor MultiJet: 5.0 fb-1: PAS-SUS-12-009 n  https://twiki.cern.ch/twiki/bin/view/CMSPublic/PhysicsResultsSUS12009

n  Razor 3rd-Gen. Scalar LQ: 1.8* fb-1: PAS EXO-11-030 n  https://twiki.cern.ch/twiki/bin/view/CMSPublic/PhysicsResultsEXO11030

W. Reece (CERN) - New results using the Razor at the LHC - ICHEP 2012

2

*Previous Lumi normalization

NEW

NEW

+The Razor Variables

n  No strong intuition in specifics of new physics models

n  Try to be model independent and inclusive

n  Canonical example: R-parity conserving SUSY

n  New heavy states & LSP as DM candidate

n  High pT jets, leptons, and MET in final state

n  Also in SM background: must discriminate


3


n  No strong intuition in specifics of new physics models

n  Try to be model independent and inclusive

n  Canonical example: R-parity conserving SUSY

n  New heavy states & LSP as DM candidate

n  High pT jets, leptons, and MET in final state

n  Also in SM background: must discriminate

n  We assume new physics is pair produced at the LHC

n  Reconstruct pseudo-dijet topology with hemispheres

n  Ignore some details of event; look just for high scale

n  Use MR to estimate scale event-by-event

n  For signal: Peaks at MΔ; For QCD di-jets: falls as

n  Transverse boost to di-squark rest frame: |ph1| = |ph2|


4


n  Studying MR distribution not enough: QCD dominates

n  Consider additional variable: MTR

n  Gives kinematic endpoint at MΔ

n  MR and MTR measure similar quantity

n  Use different information: transverse/global

n  Dimensionless ratio R: Approximate range 0-1

n  R very small for QCD; large for signal


5

+

[GeV]RM0 500 1000 1500 2000

2R

00.05

0.10.150.2

0.250.3

0.350.4

0.450.5

0

200

400

600

800

1000

1200

=7 TeVsCMS Simulation -1 L dt = 750 pb∫ V+jets

)+jetsννW+jets, Z(

[GeV]RM0 500 1000 1500 2000

2R

00.05

0.10.150.2

0.250.3

0.350.4

0.450.5

0

50

100

150

200

250

300

350

=7 TeVsCMS Simulation -1 L dt = 750 pb∫ +jetstt

[GeV]RM0 500 1000 1500 2000

2R

00.05

0.10.150.2

0.250.3

0.350.4

0.450.5

Even

ts / b

in

0

20

40

60

80

100

=7 TeVsCMS Simulation -1 L dt = 7.29 nb∫ QCD

The Razor Variables


6

n  Each background has a own shape in plane n  Defined by scale of process and intrinsic MET

n  MR peak moves to right as scale increases

n  Model shapes of SM background n  See signal as bump on exponentially falling tail

QCD V+Jets ttbar+Jets

[GeV]RM0 500 1000 1500 2000

2R

00.05

0.10.150.2

0.250.3

0.350.4

0.450.5

Even

ts / b

in

00.10.20.30.40.50.60.70.80.9

=7 TeVsCMS Simulation -1 L dt = 750 pb∫ SUSY LM6

SUSY (LM6) MΔ= 831 GeV

+

MR (GeV)200 400 600 800 1000 1200 1400

Even

ts /2

6 G

eV

1

10

210 > 0.042R > 0.062R > 0.092R > 0.122R > 0.162R > 0.202R > 0.252R > 0.302R

= 7 TeVsCMS Preliminary MU box zero b-tagged jet data(NR11 'W+jets' control data)

!"#$%&#

'()#$%&#

!"*!"#$%&#

+,+*!"#$%&#

+,+#$%&#

-./012#!"#3.#4#56#77#89:6#+,+#3.#4#;6<=#

-./012>,%%?@#!"#3.#4#5A>56<=#

-89:6>89BA#+,+#3.#4#;6>56<=#

+,+*+,+#$%&#-./012#!"#3.#4#56<=#

-89:6#+,+#3.#4#;6<=#

!"#

!"#

!"#

!"#

!"#

$%&#

$%&#

$%&#

$%&#

$%&#

Inclusive Razor


7

n  Very inclusive selection:

n  Razor triggers: hadronic + single lepton

n  Event quality filters, ≥2 jets (60 GeV)

n  Cut on offline MR/R to remove HLT turn-on

n  Standard lepton ID: tight & loose defines box

n  Classify events into orthogonal Boxes:

n  EleMu, MuMu, EleEle, Mu, Ele, Had

n  After MR/R cuts, QCD background negligible

n  Cross-check with shape from low MR control region

n  Data control sample for shape of each background:

n  ttbar: ≥1b-tags; ≥ 1 lepton

n  Z+jets: 0b-tags; ≥2 OS leptons; Z mass window

n  W+jets: 0b-tags; ≥1 lepton

No razor trigger in control samples: Independent from analysis data

+

n  2D shapes and normalization of SM background from ML fit n  Fit in low MR/R fit region; Extrapolate to signal region à SM prediction

n  Box-by-box and then combine for best limit

2R0.2 0.25 0.3 0.35 0.4 0.45 0.5

Even

ts/(0

.013

)

210

310DataSM totalMU-like effective

componentstV+jets 1 plus ELE-like effectivest+jets 1tt

= 7 TeVsCMS Preliminary -1 Ldt = 4.4 fb∫HAD box

[GeV]RM500 1000 1500 2000 2500 3000 3500 4000

2R

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

-310

-210

-110

1

S1 HAD box68% range [9.0,196.2]Mode 25.5Median 67.5observed 10p-value 0.42






Inclusive Razor 8

[GeV]RM400 600 800 1000 1200 1400 1600

Even

ts/(4

0 G

eV)

1

10

210

310DataSM totalMU-like effectivestV+jets 1

plus ELE-like effectivest+jets 1tt

= 7 TeVsCMS Preliminary -1 Ldt = 4.4 fb∫HAD box 1D projections

of 2D PDF


+

[GeV]0m500 1000 1500 2000 2500 3000

[GeV

]1/

2m

100200

300400

500

600700

800

9001000

± l~ LEP2

± 1!" LEP2

No EWSB

= L

SP#"

Non-Convergent RGE's) = 500g~m(

) = 1000g~m(

) = 1500g~m(

) = 2000g~m(

) = 1000q~m(

) = 1500q~m(

) = 2000

q~m(

) = 2500

q~m(

Median Expected Limit$1 ±Expected Limit

Observed LimitHAD Observed LimitLeptons Observed Limit

CMS Preliminary -1 L dt = 4.4 fb% = 7 TeV s

Hybrid CLs 95% C.L. LimitsRazor Inclusive

No EWSBNon-Convergent RGE's

)=10&tan( = 0 GeV0A

> 0µ = 173.2 GeVtm

= L

SP#"

Inclusive Razor


9

n  Model dependent result: Hybrid CLS with signal from MC n  Systematics included toy-by-toy

+

]2 [GeV/ct~m400 600 800 1000 1200

]2 [G

eV/c

!"m

100200300400500600700800900

1000

-210

-110

!" t + #t~, t~t~ #pp

CMS Preliminary-1 L dt = 4.7 fb$ = 7 TeV s

1 b-tag%Razor Inclusive ) s (p

b) (C

L&

95%

CL

uppe

r lim

it on

)NLO-QCD& = prod&observed limit ( (th.)& 1 ±observed limit

median expected limit (exp.)& 1 ±expected limit

]2 [GeV/cg~m500 600 700 800 900 1000 1100 1200

]2 [G

eV/c

!"m

100

200

300

400

500

600

700

800

-210

-110

)g~) >> m(t~; m(!" 2t + #g~, g~g~ #pp

CMS Preliminary-1 L dt = 4.7 fb$ = 7 TeV s

1 b-tag%Razor Inclusive ) s (p

b) (C

L&

95%

CL

uppe

r lim

it on

)NLO-QCD& = prod&observed limit (

(th.)& 1 ±observed limit median expected limit

(exp.)& 1 ±expected limit

Inclusive Razor with B-tag n  Natural extension to inclusive analysis: ≥1 b-tags (40GeV)

n  Dataset is a subset of the inclusive sample: same triggers & selection

n  Background model simplified: V+jets & ttbar shapes from control regions

n  Interpret model-dependent results in simplified models (SMS)


10

T1tttt T2tt

ISR sensitive points removed

+

n  Complementary to inclusive analyses: Probe low R region n  Use MultiJet triggers to select events: No online MET cut

n  Require ≥ 6 jets (30GeV); ≥ 1 b-tags; isolated lepton veto (inc. taus)

n  Very high statistics measurement: 71k events in search (BJet) box

n  Major backgrounds: QCD & ttbar: Shapes from control regions then float in BJet fit

n  QCD & ttbar hadronic: b-jet veto with loose tagger; Veto leptons as in BJet box

n  ttbar leptonic: b-jet veto with loose tagger; invert lepton veto à W+≥5 jets

Razor MultiJet:

W. Reece (CERN)

11

Extract SM background shape in low MR/R region

Predict background in signal region for limit

[GeV]RM500 1000 1500 2000 2500 3000 3500 4000

Entri

es

1

10

210

310

410

BJet Box

Data

Loose leptons

SM Total

QCD

W+jets

= 7 TeVsCMS Preliminary

-1 Ldt = 4.98 fb∫

2R0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Entri

es

1

10

210

310

410BJet Box

Data

Loose leptons

SM Total

QCD

W+jets = 7 TeVsCMS Preliminary

-1 Ldt = 4.98 fb∫

= 7 TeVsCMS Preliminary

-1 Ldt = 4.98 fb∫

M∆ =

�m2

t̃− (mt +mχ̃)2

m2t̃

+

gluino mass [GeV]500 600 700 800 900 1000 1100 1200

) S (p

b) (C

Lσ

95%

CL

uppe

r lim

it on

-210

-110

1

)g~) >> m(t~; m(χ∼ 2t + →g~, g~g~ →pp

CMS Preliminary

-1 L dt = 4.98 fb∫ = 7 TeV s

Razor MultiJet (BJet box)m(LSP) = 50 GeV

)g~g~ (NLO+NLLσ

Observed

Expected (68%)

t' mass [GeV]300 400 500 600 700 800 900 1000 1100 1200

) S (p

b) (C

Lσ

95%

CL

uppe

r lim

it on

-210

-110

1

χ t + → t't', t'→pp

CMS Preliminary-1 L dt = 4.98 fb∫ = 7 TeV s

Razor MultiJet (BJet box)m(LSP) = 50 GeV

)1/2

(t'NLO+NLLσ LOk

)0

(t'NLO+NLLσ

Observed

Expected (68%)

Razor Multijet

n  T1tttt: Set limits on gluino pair production in SUSY

n  T2tt: Set limits on t’0 mass in SUSY (stop) & t’1/2 in UED, 4th Gen, etc n  Treat t’1/2 and LSP masses as independent: Reweight SUSY σNLO with LO k-factor


12

T2tt T1tttt

+

2R0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6

Even

ts

0

20

40

60

80

100

120

140

160

180

DataTotal SM PredictionTotal SM Prediction +

= 250 GeV)LQ

LQ (M

-1 L dt = 1.8 fb∫=7 TeV sCMS Preliminary (GeV)RM400 500 600 700 800 900 1000 1100 1200 1300 1400

Even

ts /

( 40

GeV

)

1

10

210DataMultijetsW/Z + jets

+ jetsttTotal SM Prediction +

= 250 GeV)LQ

LQ (M

-1 L dt = 1.8 fb∫=7 TeV sCMS Preliminary

3rd Generation Scalar LeptoQuarks n  Search for LQ pair production: LQàbντ (2b, 0 leptons, & MET)

n  Use data from razor HLT trigger suite: hadronic + leptonic

n  Require ≥2 jets (60 GeV); 1,2 loose b-tags; lep. ID (e,µ)

n  Classifly events into Ele, Mu, & Had boxes

n  Describe background shapes with 1D function:

n  Find A1,2, B1,2 for from control regions/MC n  V+ heavy flavour jets: Madgraph matched to pythia

n  ttbar: ≥2 b-tags; ≥1tight µ; R2 > 0.14

n  QCD Multijets with HF: ≥1 b-tags; ≥1oose µ; R2 > 0.14

n  Find scale factors from sidebands in Ele & Had boxes n  Apply to Had box signal region to predict background

n  Cross-check in Ele box at high R2 (>0.25)


13

+3rd Generation Scalar LeptoQuarks


14

(GeV)LQM200 250 300 350 400 450 500

Cro

ss s

ectio

n (p

b)

-110

1

10

σExpected 1 σExpected 2

Observed limitTheory

)-1D0 exclusion (5 fb


(GeV)LQM200 220 240 260 280 300 320 340 360 380 400

) τν b

→Br

(LQ

0.4

0.5

0.6

0.7

0.8

0.9

1

)-1D0 exclusion (5.2 fb

)-1CMS 95% CL Limit (observed, 1.8 fb

)-1CMS 95% CL Limit (expected, 1.8 fb


Assuming Br(LQàbντ) is 1 Br(LQàbντ) free to vary

+Summary

n  The razor variables allow us to search inclusively for NP n  Make assumptions about underlying kinematics

n  Get peaking signal on exponentially falling background

n  Presented four analyses: n  Inclusive Razor: Stringent limits on CMSSM

n  Razor-B: New limits on di-stop and di-gluino production

n  Razor MultiJet: Sensitivity to di-stop production expected for 2012

n  3rd Generation LQ: Improves previous limits by 100 GeV

n  Looking forward to new results in 2012


15

+Backup Slides

W. Reece (CERN) - New results using the Razor at the LHC - ICHEP 2012 16

+Razor Cartoon


17

+Model Dependent Results


18

+Model Dependent Results


19

+Razor Multijet: T1tttt 2D limits


20

]2gluino mass [GeV/c500 600 700 800 900 1000 1100 1200

]2LS

P m

ass [

GeV

/c

100

200

300

400

500

600

700

800

-310

-210

-110

1

10)g~) >> m(t~; m(χ∼ 2t + →g~, g~g~ →pp


Razor MultiJet (BJet box) Observed

) s (p

b) (C

Lσ

95%

CL

uppe

r lim

it on

)g~g~ (NLO+NLLσ

)g~g~ (NLO+NLLσExpected

[GeV]RM500 1000 1500 2000 2500 3000 3500 4000

2R

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-310

-210

-110

1

= 7 TeVsCMS Preliminary -1 Ldt = 4.98 fb∫Razor Multijet (BJet box)

p-value

+Razor Multijet: T2tt 2D limits


21

]2t' mass [GeV/c200 400 600 800 1000 1200

]2LS

P m

ass [

GeV

/c

200

400

600

800

1000

-310

-210

-110

1

10χ t + → t't', t'→pp



) s (p

b) (C

Lσ

95%

CL

uppe

r lim

it on

)0

(t'NLO+NLLσ LO

Expected k

]2t' mass [GeV/c200 400 600 800 1000 1200

]2LS

P m

ass [

GeV

/c

200

400

600

800

1000

]2t' mass [GeV/c200 400 600 800 1000 1200

]2LS

P m

ass [

GeV

/c

200

400

600

800

1000

-310

-210

-110

1

10χ t + → t't', t'→pp



) s (p

b) (C

Lσ

95%

CL

uppe

r lim

it on

)1/2

(t'NLO+NLLσ LOk

)1/2

(t'NLO+NLLσ LO

Expected k

]2t' mass [GeV/c200 400 600 800 1000 1200

]2LS

P m

ass [

GeV

/c

200

400

600

800

1000

Scaler t’ (SUSY) Fermioninc t’ (UED, 4th Gen. etc)

search for new physics at cms using the razor kinematic variables

Documents