Potential to Discover Supersymmetry in Events with Same-sign Dimuon, Jets and Missing Energy

Yuriy Pakhotin SUSY08, Seoul, Korea, June 19th, 2008 1/16

Potential to Discover SupersymmetryPotential to Discover Supersymmetryin Events with Same-sign Dimuon,in Events with Same-sign Dimuon,

Jets and Missing EnergyJets and Missing Energyat LHCat LHC

Yuriy Pakhotin ([email protected])for CMS and ATLAS Collaboration

SUSY08, COEX, Seoul, KoreaJune 19th, 2008


LHC: Large Hadron ColliderLHC: Large Hadron ColliderLHC is pp-collider

nominal energy: E=14 TeV

design luminosity: L=1034 cm-2s-1

first physics run: (Fall 2008) E = 10 TeV, L=2·1033 cm-

2s-1


CMS: Compact Muon SolenoidCMS: Compact Muon Solenoid

Muons:muon system acceptance: |η|<2.4muons momentum resolution: dpT/pT~1% (pT~25 GeV) Calorimetry:HCAL |η|<5.0, δE/E ~ 70% / √E + 8%ECAL |η|<3.0, δE/E ~ 2.8% / √E + 0.3% + 12% / E

MUON BARRELDrift TubeChambers ( DT )

Resistive PlateChambers ( RPC )

SUPERCONDUCTINGCOIL

IRON YOKE

Silicon MicrostripsPixels

TRACKER

Cathode Strip Chambers (CSC )

Resistive Plate Chambers (RPC)

MUONENDCAPS

CALORIMETERSECAL

Scintillating PbWO4 crystals

HCALPlastic scintillator/brasssandwich

Total weight: 12,500 tDiameter: 15 mOverall Length: 22 mMagnetic field: 4 Tesla


ATLAS: A Toroidal LHC ApparatuS ATLAS: A Toroidal LHC ApparatuS

Muons:muon system acceptance: |η|<3.0muons momentum resolution: δpT/pT~2% (pTµ~30 GeV) Calorimetry:HCAL |η|<4.9, δE/E ~ 45% / √E + 3%ECAL |η|<3.2, δE/E ~ 10% / √E + 0.6% + 0.5% / E

Total weight: 7000 tDiameter: 25 mOverall Length: 46 mMagnetic field: 2 Tesla


Physics MotivationPhysics MotivationInclusive topological search for physics beyond SM

Despite many successes of Standard Model, there are strong indications that this theory is only an effective low-energy model and new physics must be present at a higher energy scaleAn excellent signature to search for deviations from the SM is production of two leptons with both leptons of two leptons with both leptons of the same electric chargethe same electric chargeThis signature occurs naturally in many extensions to the SM and occurs rather rarely in SM interactionsDifferent theory models can be used as templates for searches. Don’t need to believe in any of them, though well motivated examples are to be preferred:• SUSY (used in this analysis)• Majorana neutrino• etc.


Same-sign Dimuon SignatureSame-sign Dimuon SignatureThe key signature for SUSY in analyses presented hereMET - Missing Transverse Energy (R-parity conservation, neutral LSP)+ number of jets with high ET+ muon

clean trigger+ second like-sign muon

even cleaner signature with low background due to the same-sign muon requirementcomplementary to trilepton searches: more diagrams, for example

q

p

Xp

q

q

q

1

g

q

01

01

1

two same-sign muons analysis is able to distinguish SUSY diagrams with good efficiency and purity by applying muon isolation & tight quality cuts


Brief History of the AnalysesBrief History of the Analyses1.Theoretical studies

H. Baer et al. Phys. Rev. D41, #3 (1990)R.Barnett et al. Phys. Lett. B315 (1993), 349K. Matchev, D. Pierce hep-ph/9904282 (1999)and others

2.Experimental study at TevatronCDF Phys. Rev. Lett. 98, 221803 (2007)D0 Phys. Rev. Lett. 97, 151804 (2006)


Strategy of the Inclusive AnalysesStrategy of the Inclusive AnalysesThe main goal of the presented analysis is to prepare for the start-up running of LHC by searching for extensions of Standard Model using nSUGRA model for optimizationThe strategy is to search for excess in number of events over expected number from SM background events (counting experiment)Apply muon triggers: which are expected to be robust even at the LHC start-upApply quality cuts: pre-select well reconstructed quantitiesApply selection cuts: efficiently suppress SM backgroundOptimize cuts for benchmark mSUGRA sample to maximize significanceEstimate backgrounds from data

The main assumptions for the current analyses:low luminosity (1 fb-1) – initial data collected by LHCuse mSUGRA as a guide to find optimized parameters region where the SM background is expected to be significantly small in comparison to signal.optimized cuts applied to some other theories may also yield significant excess over SM backgrounds


Full detector simulation was

performed in 10 mSUGRA test points

for CMS

Some of these points are post-WMAP

benchmark points

Positions of the test points in the m0-

m1/2 plane are shown on the right plot

(as stars)

A0=0, tan(β)=10, sign(µ)=+1

Fast generation and simulation was also performed in order to scan the

plane of m0-m1/2

Other mSUGRA parameters are fixed: A0 = 0, tan(β) = 10, sign(µ) = +1

Points were generated on a coarse grid with: δm0 = 100 GeV and δm1/2 =

100 GeV

For validation purposes 7 benchmark points were also produced with fast

simulation and compared with fully simulated data

mSUGRA Benchmark Test Points for mSUGRA Benchmark Test Points for CMSCMS


mSUGRA Benchmark Test Points for mSUGRA Benchmark Test Points for AtlasAtlas

1. Set of SU benchmark points: fully simulated with Geant

2. LST1 Benchmark point (hep-ph/0512284): Atlas fast simulationLight stop production from gluino pair productiongluino is Majorana, so following decays have equal probabilities:

g->t ~t1, g-> tbar ~t1pair-produced gluinos give the signature

2b-jets + 2 same-sign leptons + jets + MET

M0 (GeV) M1/2 (GeV) A0 (GeV)tan(β)

sign(µ) σLO (pb) Nevents

SU1 Coannihilation 70 350 0 10 +1 8.15 200 000SU2 Focus point 3550 300 0 10 +1 5.17 50 000SU3 Bulk 100 300 -300 6 +1 20.85 500 000SU4 Low mass 200 160 -400 10 +1 294.46 200 000SU6 Funnel 320 375 0 50 +1 4.47 30 000SU8.1 Coannihilation

210 360 0 40 +1 6.48 50 000


Standard Model BackgroundsStandard Model BackgroundsTo prepare for Model Independent analysis it is imperative to understand the Standard Model background as well as

possible.

Full detector (both, Atlas and CMS) simulation was performed for following SM backgrounds:tt+jets sampleW+jets, Z+jetsWW+jets, WZ+jets, ZZ+jetsQCD jets

Example of 2 same-sign muons production in t tbar event

p

p

tX

t

W

b

W

c

b

W

Data driven methods to estimate SM backgrounds are currently elaborating


Atlas: Selection CutsAtlas: Selection Cuts

Effective Mass:

Meff = MET + ∑pTjets

Event Selection (not optimized):

2 same-sign isolated leptons (muons + electrons)lepton pT > 20 GeV4 jetsjet pT > 50 GeVMET > 100GeVEasy to see excess events to SM

Simple scaling to low luminosity still gives excess over SM

30 fb-1

LST1 mSUGRA benchmark poin


CMS: Selection CutsCMS: Selection Cuts

pT of leading muons

QCD

pT of leading muons

SUSY

Example: transverse momentum (pT) of leading

muonSignal (SUSY LM1) muons have muon pT spectrum:

harder than QCDsimilar to W/Z+jets and TTbar

Pre-selection cut pTµ > 10 GeV to avoid problems with low-pT range where efficiency is not very good

Other cut variables:1.muon trigger (di-muon High-Level Trigger)2.transverse momentum of muons3.combined (calorimeter + tracker) muons isolation4.muons track parameters5.jet multiplicity6.pT of 3 leading jets7.large missing transverse energy


LEPTevatron

A0=0, tan(β)=10, sign(µ)=+1

Optimized cuts for 10 fb-1 luminosity

mSUGRA 5σ reach contours (Monte Carlo simulation) of the same-sign dimuon analysis, including systematic uncertainties, for different integrated luminosities and assuming no re-optimization of the selection cuts

CMS: Same-sign Dimuon Reach CMS: Same-sign Dimuon Reach ContourContour

Sample

QCD ≤0.01

TTbar ≤0.3

W/Z+jets ≤0.05

mSUGRA LM1 17 (S ≥ 5.5)

CMS Preliminary: Number of expected

events after selection cuts applied for

∫L=1 fb-1


Atlas: In-situ Background Atlas: In-situ Background EstimationEstimation

2D Side Band (SB) is chosen:pTjet2 [40 GeV ~ 80 GeV] MET [50 GeV ~ 80 GeV]

ASR=ASB*BSR/BSB

True background events in 1 fb-1=14.8Side Band Content

Estimated BG in 1 fb-1

All BG 15.2 ± 4.0BG+SU1 15.4 ± 4.1BG+SU2 15.2 ± 4.0BG+SU3 15.3 ± 4.1BG+SU4 21.4 ± 4.8BG+SU6 15.3 ± 4.1

Transverse mass (MT) of leading lepton and MET is a good candidate as a variable for Side Band to Signal Region (SR) normalization:

Transverse mass distributions for SB and SR are in a very good agreement.


ConclusionConclusion1. The preparation for topological search of new physics

(excess over SM background) with same-sign dimuon in LHC experiments is presented

2. Assuming unknown signal, mSUGRA model for cut optimization is used

3. It is shown that SM background can be suppressed with optimized cuts almost to zero with luminosity less than 1 fb-1 (early running of LHC). Significant number of signal events are survived

4. Small number of expected SM background events leads to less dependency on the statistical and systematical uncertainties, which is crucial for initial experiment data. Then even relatively small excess (a few events) over expected SM background may be safely interpreted as a discovery of physics beyond SM

5. The same set of cuts may be applied to search for different theoretical predictions beyond SM

6. Data driven methods to estimate SM backgrounds for low luminosity (1 fb-1) are currently elaborating


Back-up slides


Early running of LHCEarly running of LHC…it was decided to push for collisions at an energy of 10 TeV this year, as quickly as possible, with full commissioning to 14 TeV to follow over the winter shutdown.

Robert Aymar , CERN Director-generalCERN Bulletin, Issue No. 14-15/2008 - Monday 31 March 2008

Reduction in all cross sections is expectedConsider two particular parton combinations, q qbar (e.g. for Z, Z', etc) and gg (for Higgs, ttbar, etc). Thus, the reduction in cross section

•for a 200 > GeV Higgs boson is almost exactly a factor of 2

•for W,Z the > reduction factor is less (70%)


SUSY: SupersymmetrySUSY: Supersymmetry

Avoids fine-tuning of SM, can lead to GUTs Generally assume LSP is stable (R-parity conservation) possible

dark matter candidate SUSY breaking mechanism is unknown many parameters

mSUGRA:supergravity inspired model5 free parameters: m0, m½, A0, tan(β) and sign(µ)

A possible symmetry between fermions and bosons|S=0 or S=1⟩ ↔ |S=½⟩


mSUGRAmSUGRAmSUGRA stands for minimal supergravity. The construction of a realistic model of interactions within N = 1 supergravity framework where supersymmetry breaking (Gravity Mediated Supersymmetry Breaking ) is communicated through the supergravity interactions.Parameters of mSUGRA:

1.m0 – the universal scalar mass

2.m1/2 – the universal gaugino mass

3.tan(β) – the ration of the vacuum expectation values of the two Higgs fields

4.A0 – the Higgs-squark-squark trilinear coupling constant

5.sign(µ) – where µ is the unmixed Higgsino mass or the SUSY-conserving Higgs mass parameter


Physics motivation: other Physics motivation: other theoriestheories

Another model is one with a Majorana particle that decays through SM-like bosons into leptons:

pp -> pp -> ννMM l X l X

Heavy Majorana neutrinos (νM) can be produced in pp

collisions in association with a lepton through a virtual W boson [T. Han and B. Zhang, Phys. Rev. Lett. 97, 171804 (2006)]. This new particle can subsequently decay to a W and another lepton:

ννMM -> W l -> W lGiven the Majorana nature of this neutrino, i.e., that it is its own

antiparticle, more than half of such events will contain like-sign more than half of such events will contain like-sign dileptons in the final statedileptons in the final state


MuonsMuonsTwo different collections of muons are storedSTA - Standalone (muon system)GMR - Global reconstructed (muon system + tracker)

Kinematics distributions for GMR:

Eta distribution of leading muon (Pt>5 GeV)

LM1

Eta distribution of leading muon (Pt>5 GeV)

LM1


MuonsMuonsSignal (LM1) muons have muon Pt spectrum:

harder than QCDsimilar to W/Z+jets and TTbar

Pre-selection cut PTµ > 10 GeV to avoid problems with low-PT range where efficiency is not very good

Pt of leading muons

QCD

Pt of leading muons

LM1


Muons MultiplicityMuons MultiplicitySUSY events have a more of muons, hence we can require 2 same sign muons. This cut efficiently kills QCD, W+jets and Z+jets, because these backgrounds typically have less than 2 muons in most of events

Nmuons = 2 same sign muons

Number of muons

LM1

Number of muons

W/Z + jets

TTbar

Number of muons

QCD


2SS muons from hadron decay2SS prompt SUSY muons Efficiency

Muons isolationMuons isolationIn order to distinguish SUSY diagrams combined isolation was used in the dimuon analysis:Combined Isolation = Tracker_Iso + 0.75 * Calorimeter_Iso < 10 GeV

65% efficient at identifying SUSY diagrams, 90% pure w.r.t. SUSY hadrons decay and SM backgrounds

Isoµ < 10 GeV

prompt SUSY muons

muons from hadrons decay


2 Same Sign Muons2 Same Sign MuonsNo remarkable difference in the PT distribution for muons between signal and background (see plots below) Pt of leading muon in

2SS LM1Pt of leading muon in

2SS LM1

Pt of second muon in 2SS

W/Z+jets

TTbar

Pt of second muon in 2SS

W/Z+jets

TTbar


JetsJetsBecause of large squark masses we expect high energy jets in SUSY events, hence we can cut on their ETTransverse energy of jets:

• ET1st > 175 GeV• ET2nd > 130 GeV• ET3rd > 55 GeV

Et of leading jet

W/Z+jets

TTbar

Et of leading jet

QCD

Et of leading jet

LM1


Multiplicity of jetsMultiplicity of jets

Number of jets

W/Z+jets

TTbar

Number of jets

LM1

Number of jets

QCD

SUSY events have a lot of jets, hence we can cut that number at 3 This cut efficiently kills QCD, W+jets and Z+jets, because these backgrounds typically have 2 jets or less in most of events

Njets >= 3


MET

QCD

METMET

MET

W/Z+jets

TTbar

MET

LM1

Missing energy is one of the most important cuts, because large value of missing energy is an inherent SUSY signature due to LSP which escape detection (R-Parity is conserved in mSUGRA).

MET > 200 GeV


m0m1/2

m1/2

m0

no ewsb

no ewsb

1 LSP

1 LSP

Gluino and squark isomassGluino and squark isomass


no ewsb

1 LSP

no ewsb

1 LSP

no ewsb

1 LSP σ(p+ + p+ → ~q + ~q + X) σ(p+ + p+ → ~q + ~g + X)

σ(p+ + p+ → ~g + ~g + X)

Cross section iso-contoursCross section iso-contours


no ewsb

1 LSP

m(~χ20) = m(~lL) m(~χ2

0) = m(~lR)

reach contour for 10fb-1

x-section isolines

Same-Sign di-muonsSame-Sign di-muons: : Contour Contour behaviorbehavior


Significance as a measure of meritSignificance as a measure of meritIf for a particular set of cuts we observe ns (signal) and nb (background) MC events, is it better of worse than a different set of event numbers for a different set of cuts?Common approach: expected number of signal events in an experiment is s = ws ns

… actually should be = ws (ns+1)

expected number of bkgd events in an experiment is b = wb nb

… actually should be = wb (nb+1)estimate significance S of observing (b+s) events, when ones expect b events for background

Significance estimators implemented in Garçon?

popular choices, but very poor estimators…

work only in the limit of very large b>>1…

the bias is even greater when nb is not large…

Yes, but

NOT RECOMMENDDED!!!

correct significance for any b, as long as nb>>1,

but would be an overestimate for a smaller nb

Yes, see Sc below

very close approximation for ScP Yes, OK when nb>>1

correct significance, takes into account statistical uncertainties arising from limited MC statistics; identical to ScP for nb>>1

Yes,

Strongly recommended

1

sS

b

2

sS

s b

12 2S s b b

cPS

2( ) ln(1 / ) 2cLS s b s b s

cS


Significance: SSignificance: Scc (one bkgd) (one bkgd)

number of MC events before cuts N event weight w number of MC events after cuts n

1) pdf (probability distribution function) for the cut efficiency ε

2) probability to observe in experiment exactly k events

3) calculate probability to observe k0 events and convert it to significance

Sc

we use this approximation (easy to integrate)

very good for large N>>1

on a conservative side for smaller N

( 1)k w n side note:

( )( ) ( 1) (1 )

!

nn n N n NN

NN C N e

n

1

10

( ) 1 ( )( ) ( )

! (1 ) ! ! 1

kkN

n

N k n wp k e d

k w k n w

2

0

21

( )2

cS

x

k k

p k e dx


Significance: SSignificance: Scc vs. S vs. ScLcL

Blue = ScL ~ ScP

b=w(n+1)

Red = Sc

w=0.1

n=0w=0.5

n=0

w=0.1

n=10w=0.5

n=10

0 10 20 30 400

2

4

6

8

10

Observed events k0

Sig

nifi

can

ce

0 10 20 30 400

2

4

6

8

10

Observed events k0

Sig

nifi

can

ce

0 10 20 30 400

2

4

6

8

10

Observed events k0

Sig

nifi

can

ce

0 10 20 30 400

2

4

6

8

10

Observed events k0

Sig

nifi

can

ce


Systematic uncertaintiesSystematic uncertainties

Systematic Uncertainty dNB/NB

Single muon analysisSame sign dimuon

analysis

Theory 13% 13%

Luminosity 5% 5%

Jet energy scale 10% 15%

Jet energy resolution 5% 10%

Muon efficiency/resolution negligible negligible

Fast vs. full simulation 2% 2%

Total systematic uncertainty

18% 23%

36

Summary table:


Results for fully simulated pointsResults for fully simulated points 37

Summary table: total number of background and signal events which pass the optimized selection cuts for 10 fb-1, together with the corresponding significance (with and without systematic uncertainties) to discover different signal benchmark points.

Potential to Discover Supersymmetry in Events with Same-sign Dimuon, Jets and Missing Energy

Documents