atlas su per sy mmetry wg journée de réflexion – sept. 14 th 2007

24
ATLAS Supersymmetry WG Journée de réflexion – Sept. 14 th 2007 Till Eifert DPNC – ATLAS group

Upload: maeve

Post on 11-Jan-2016

16 views

Category:

Documents


0 download

DESCRIPTION

ATLAS Su per sy mmetry WG Journée de réflexion – Sept. 14 th 2007. Till Eifert DPNC – ATLAS group. What’s going on there ?. Till. ?. Andree, Tuan Clemencia, Moritz. * Computing System Commissioning. Currently, people concentrate on the so-called CSC * notes … - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

ATLAS Supersymmetry WGJournée de réflexion – Sept. 14th 2007

Till EifertDPNC – ATLAS group

Page 2: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

What’s going on there ?■ Currently, people concentrate on the so-called CSC* notes …

○ 1 & 2: Data-driven estimations of Z/W & top backgrounds Generator & detector uncertainties Many analyses, most data-driven

○ 3: Data-driven estimations of QCD backgrounds Fake MET rejection MC, data-driven estimates

○ 4: Estimation of Heavy Flavor backgrounds and associated systematic○ 5: Searches and inclusive SUSY studies

RPC, no GMSB, no split SUSY Study signatures; scan parameter space

○ 6: Exclusive measurements for SUSY events DiLepton edge, lepton+jet edge Mass reconstruction Extract susy parameters

○ 7: Gaugino direct productions○ 8: ‘Studies for Gauge mediated SUSY’ -> ‘Photonic and long-lived SUSY signatures’

2007 Sept 14 ATLAS SUSY WG 2

* Computing System Commissioning

Till

Andree, TuanClemencia, Moritz

Page 3: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Supersymmetry (SUSY)■ The light scalar Higgs boson is unprotected at GUT/ Planck scales■ On the contrary, all the other light particles of the SM are protected against large scales:

○ Due to chiral symmetry, their mass corrections are logarithmic in E (and not quadratic)

○ Gauge symmetry protects the bosons (no correction to photon or gluon masses)

■ Fermion and boson loops contribute with different signs to the Higgs radiative corrections:if there existed a symmetry relating these two, this could protect the masses of the scalar !

■ Supersymmetry realises this by transforming bosons fermions○ SUSY transforms for example a scalar boson into a spin-½ fermion, whose mass is protected

○ Hence, the scalar mass is also protected

○ This solves the naturalness and the hierarchy problems of the SM

■ Local gauge invariance of SUSY requires existence of spin-3/2 and spin-2 particles○ This naturally introduces the spin-2 graviton, assumed to mediate the gravitational force

2007 Sept 14 ATLAS SUSY WG 3

Fermion loop

Boson loop

Page 4: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Minimal Supersymmetric Standard Model (MSSM)

■ To create supermultiplets, we need to add one superpartner to each SM particle

■ Need to introduce an additional Higgs doublet to the non-SUSY side■ Mutual superpartners have equal masses and couplings

Spin 0 Spin 1/2 Spin 1 Spin 3/2 Spin 2

Higgs Higgsino Gravitino Graviton

sLepton Lepton

sQuark Quark

Gluino Gluon

Photino Photon

Zino Z

Wino W

2007 Sept 14 ATLAS SUSY WG 4

SM

SUSY

Page 5: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Minimal SuperGravity (mSUGRA)

Reduce the ~ 105 parameters of MSSM to 5 !

mSUGRA assumes that at the GUT scale all scalars (squarks, sleptons, and Higgs bosons) have a common mass m0,

all gauginos and Higgsinos hava a common mass m1/2,

and all the trilinear Higgs-sfermion-sfermion couplings have a common value A0

Remaining two parameters (at GUT scale): SUSY conserving Higgs mass => sign Ratio of Higgs vacuum expectation values tan = 1/2

Renormalisation group equations (RGEs) govern the running to the EW scale

Lightest neutralino is LSP

R-parity is conserved R = (-1)( 3(B-L) + 2S) where B, L, and S are the baryon number, lepton number, and spin respectively.=> R=+1 for SM particles

R=-1 for SUSY particles

2007 Sept 14 ATLAS SUSY WG 5

RG evolution of unified mSUGRA mass parameters

Page 6: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

2007 Sept 14 ATLAS SUSY WG 6

Characteristic SUSY “Cascades” at the LHC

Conserved R-parity requires existence of a lightest stable SUSY particle = “LSP”. Since no exotic strong or EM bound states (isotopes) have been observed, the LSP should be neutral and colourless WIMP !

The experimental signature of the LSP would be just as the one of a heavy neutrino !

The LSP is typically found to be a spin-½ “neutralino”, a linear combination of gauginos (in much of the SUSY parameter space the neutralino is a mixture of photino and zino)

p pg

Lq 02

R

01

qq

“Typical” SUSY decay chain at the LHC

escapes detection missing ET

qX

Page 7: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

2007 Sept 14 ATLAS SUSY WG 7

Inclusive SUSY Searches

The precise signatures of the SUSY “cascades” are driven by the masses of the SUSY particles

To good generality we can expect:

High-pT jets from squark & gluino decays

Leptons from gaugino & slepton decays

Missing energy from LSPs

This lays out an inclusive search strategy

Detector requirements:

Excellent jet-energy measurement

Excellent lepton identification

Hermeticity of the detector (good acceptance)

Run II V. Shary @ CALOR04

Measuring missing energy is a tough task !

Page 8: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

2007 Sept 14 ATLAS SUSY WG 8

( , ) 1.5 TeVm q g ( , ) 1.0 TeVm q g

fully inclusive 1 Lepton

Inclusive SUSY Searches … continued

A sensitive variable to detect SUSY decays is the “effective mass”: eff ,missjets, leptons

T TM E p

Requiring at least one lepton reduces QCD background by factor of 20–30, with signal loss of only factor of ~3 better signal-to-background ratio than fully inclusive analysis

Meff

Eve

nts 10 fb1

Page 9: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

2007 Sept 14 ATLAS SUSY WG 9

Inclusive SUSY Searches … continued

Most SUSY searches are prepared by studying few “characteristic” points:

At the limit of experimental exclusion (SU4)

“Typical” point (SU3)

Special-feature points (SU1, SU2, SU6) m0

(GeV

)

SU1

SU1

SU3

SU4

SU6

SU3

SU4

no neutral LSP

SU6

“focus point”

“funnel region”

“coannihilation point”

“bulk region”

“low mass point”

SU2

SU2

m½ (GeV)

Idea of this study:

Simulate MC signals for a grid in the m0,

m1/2 space

Require ≥ 1 lepton (inclusive 1 lepton)

Find 1 optimal set of cuts for the whole grid

Page 10: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

TDR SUSY analysis■ ATLAS TDR vol. II, page 820

■ Reach for S/sqrt(S+B) > 5 for various SUSY signatures in the mSugra parameter space

■ TDR Selection○ Transverse mass (l, MET)

≥ 100 GeV “..reduce W+jet bkg..”

○ Jet cut ≥ 2 Jets pT ≥ 100 GeV optimize pT cut for each point

○ MET ≥ 100 GeV optimize cut for each point

○ transverse sphericity > 0.2 “ .. To reduce dijet background .. “

○ Lepton pT > 20 GeV Eta < 2.5

■ Integrated lumi = 10 fb-1

Page 11: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

■ Each point is separately optimized to yield the min p-value (max sigma)

■ As in TDR analysis, except for the missing ST cut ..

■ .. different datasets■ .. different detector

simulations …■ .. different isajet

version -> different susy spectra

■ Can we do better□ Other/more

variables ?□ Other methods ?

All opt resultAfter preSelection

2007 Sept 14 ATLAS SUSY WG 11

Page 12: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

■ Start from pre-selection (as before)■ Choice of variables for NN

○ MET

○ TransverseMass (l, MET)

○ JetLepPt = ΣEl_pT+ΣMu_pT+ΣJet_pT

… less correlated to MET as allMeff

○ Jet_C4_N … total number of jets

○ TopInvMass … ttbar-veto analysis t -> jet + W -> jet + lepton + nu (MET)

1 lep case: assume lep is boosted -> η(lep) = η(nu)

2 lep case: share MET b/w 2 nu, η, φ from lep

Future: use kinematic fit (HITFIT) Split analysis into 1, 2 lepton

channel

New analysis1 lep channel

2 lep channel

2007 Sept 14 ATLAS SUSY WG 12

Page 13: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

All points optimized

■ L=10fb-1

■ Each point is optimized!

■ Opt. against T1, W bkgs

Sign-plot from TDR (box-cuts on JetPt 1,2, MET)

Sign-plot (NN on MET, TM, Jet_N, JetLepPt, TopMass)

2007 Sept 14 ATLAS SUSY WG 13

Page 14: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Conclusions

■ Contribution to CSC 5 note○ Lep (electron) ID in SUSY environment○ mSugra study (presented here)○ SM background validation with first data ○ common tools developments

■ Need to find out best (most sensitive) cut approach (single cut, cut as function of integrated lumi, multiple cut regions) including systematics

■ Also follow non-box-cut approaches

2007 Sept 14 ATLAS SUSY WG 14

Page 15: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Backup slides

2007 Sept 14 ATLAS SUSY WG 15

Page 16: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Data samples■ mSugra signal

□ Grid in parameter space● A0 = 0● tan = 10● sign ● scalar mass m0 = 0 .. 3TeV● Gauginos mass m1/2 = 0 .. 1.5 TeV

□ 5k events on each par. Point□ All AtlFast 12.0.6

■ SM Backgrounds□ Consider various SM bkg samples,

see next slide□ All AtlFast 12.0.6+

■ Software□ Isajet 7.75 (for the mSugra

spectra) + HERWIG/Jimmy□ AtlFast (Athena) 12.0.6□ HighPtView

■ Production□ LCG grid□ Private productions

2007 Sept 14 ATLAS SUSY WG 16

Page 17: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

SM Background Samplesprocess description gen. Vers. σgen (pb) EventFiler εEF σEF (pb) # evtsdisk

Wenu 8270 Wenu + q/gckin(3)=80 (lower pt W)

Pythia 12.0.6.1 343 MissingEtFilter MET≥ 80 & TruthJetFilter: N≥2, minPt≥80,40, maxEta=5

14.3% 49 48k

Wmunu 8271 Wmunu + q/gckin(3)=80 (lower pt W)

Pythia 12.0.6.1 343 as above 8.35% 29 54k

Wtaunu 8272 Wtaunu + q/gckin(3)=80 (lower pt W)

Pythia 12.0.6.1 343 as above 16.3% 56 54k

Znunu 8190 Znunu + q/gckin(3)=80 (lower pt Z)

Pythia 12.0.6.1 246 as above 16.8% 41 26k

Zee 8194 Zee + q/gckin(3)=80 (lower pt Z)

Pythia 12.0.6.1 46.2 none 100% 46.2 20k

Zmumu 8195 Zmumu + q/gckin(3)=80 (lower pt Z)

Pythia 12.0.6.1 46.4 TruthJetFilter as above 20.7% 9.6 61k

Ztautau 8191 Ztautau + q/gckin(3)=80 (lower pt Z)

Pythia 12.0.6.1 46.3 MET & TruthJetFilter as above

9.72% 4.5 47k

T1 5200 ttbar "l+jets" (l=e,mu,tau) MC@NLO 12.0.6 854 TTbarLepton : 1 charged lepton from W decay

54% 461 720k

J4 8090 ckin(3)=140, ckin(4)=280 Pythia 12.0.6.1 3.16x105 JetMETEstimator > 100 & TruthJetFilter

0.29% 917 32k

J5 8091 ckin(3)=280, ckin(4)=560 Pythia 12.0.6.1 1.25x104 as above 2.85% 356 8k

J6 8092 ckin(3)=560, ckin(4)=1120

Pythia 12.0.6.1 344 as above 19.6% 67 18k

J7 8093 ckin(3)=1120, ckin(4)=2240

Pythia 12.0.6.1 5.3 none 100% 5.3 26k

J8 8094 ckin(3)=2240 Pythia 12.0.6.1 2.21x10-2 none 100% 0.02 42k

?

2007 Sept 14 ATLAS SUSY WG 17

Page 18: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

PreSelection■ Put samples on an equal

basis & reduce #evts○ Lepton cut

≥ 1 lepton (El / Mu) pT ≥ 20GeV

○ Jet cut ≥ 2 Jets pT ≥ 80, 40 GeV

○ MET ≥ 100 GeV

■ Add some variables○ AllMeff = MET+ΣJet_pT○ TransverseMass of

hardest lepton + MET

proc. Lep ε Jet ε MET ε Tot ε σPS(pb) # evtsPS #evts100pb-1

Wenu 62.8% 41.5% 68.5% 17.8% 8.7 8k 874

Wmunu 50.8% 88.0% 68.9% 30.8% 8.8 16k 881

Wtaunu 12.6% 62.6% 75.2% 5.9% 3.3 3k 330

Znunu 0.005 % 33% 100% 0.002% 7x10-4 1 0.1

Zee 83.0% 28.7% 0.2% 0.05% 0.02 12 2

Ztautau 30.5% 70.9% 66.8% 14.4% 0.65 3k 65

Zmumu 83.7% 84.0% 2.4% 1.7% 0.16 1k 16

T1 54.5% 60.7% 20.3% 6.7% 30.9 48k 3092

J4 0.1% 90.0% 3.8% ~0.003% 0.03 1 3

J5 0.05% 100 % 67 % ~ 0.12 0.12 2 12

J6 0.02% 100 % 0 % 0

J7 0.01 % 100 % 0 % 0

J8 0.02 % 100 % 40 % ~0.009% 2-6 4 0

out of statistics

out of statistics

Background efficiencies

2007 Sept 14 ATLAS SUSY WG 18

Page 19: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Optimizing each point ?

■ Optimizing each point separately effectively means having one analysis per point…○ decreases rate of the statistical type-II error

(missing a true signal) ○ increases the rate of the statistical type-I error

(finding a wrong signal)

■ One needs to find a balance○ Divide parameter region into regions with

different signatures => optimize on as few points as possible… ?

2007 Sept 14 ATLAS SUSY WG 19

Page 20: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

A single optimization point■ Apply set of optimized cuts of signal @

■ m0=300, m1/2=150

■ 5-sigma region smaller, see sigma plot

■ High-sigma points stay

■ Low-sigma points gone

Ratio of significance w.r.t. “all optimized points” plots

2007 Sept 14 ATLAS SUSY WG 20

Page 21: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

A single optimization point .. II■ Try lower-sigma point:

■ Apply set of optimized cuts of signal@

■ m0=1500 m1/2=450

■ High-sigma points go down, but …

■ Keep some more low-sigma points

Ratio of significance w.r.t. “all optimized points” plots

2007 Sept 14 ATLAS SUSY WG 21

Page 22: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Details @ m0=300 m1/2=150

Signal & Bkg variable dists

NN output variable => we run out of stats!

Input vars not strongly correlated

Sample TMVA NN eff. Events

Signal @ m0=300, m12=150 0.34 69420

Wenu 0.049 4242

Wmunu 0.053 4630

Wtaunu 0.065 2156

Zee 0.18 35

Zmumu 0.019 30

Ztautau 0.1 649

Znunu 0.0 0

Js 0.0 0

T1 0.11 34287

Bkg sum 46029

2007 Sept 14 ATLAS SUSY WG 22

Page 23: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

Details @ m0=1500 m1/2=750

Signal & Bkg variable distsJet_N & JetLepPt 75% corr.

NN output variable =>better seperation power

Sample TMVA NN eff. Events

Signal @ m0=1500, m12=750 0.078 7

Wenu 0.0 0

Wmunu 0.0 0

Wtaunu 0.0 0

Zee 0.0 0

Zmumu 0.0 0

Ztautau 0.0 0

Znunu 0.0 0

Js 0.0 0

T1 0.0 0

Bkg sum ~ 0

2007 Sept 14 ATLAS SUSY WG 23

Page 24: ATLAS  Su per sy mmetry WG Journée de réflexion – Sept. 14 th  2007

A single optimization point

■ Apply set of optimized cuts of signal @

■ m0=300, m1/2=150 (left)

■ m0=1500, m1/2=750 (right)

■ Net result: quite good coverage with 2 optimized NN (w.r.t. all points opt.)

Study of systematics -> need more stats

2007 Sept 14 ATLAS SUSY WG 24