Alain Romeyer - January 20051 Light higgs decay in SUSY cascade - Status report Introduction Trigger efficiency B tagging Jet calibration Invariant mass.

Alain Romeyer - January 2005 1

Light higgs decay in SUSY cascade - Status report

Introduction

Trigger efficiency

B tagging

Jet calibration

Invariant mass reconstruction

Likelihood fit

Conclusion

Filip Moortgat (CERN)Alain Romeyer (Mons - Belgium)

Alain Romeyer - January 2005 2

Introduction

Signature :h0

01χ LSP

MET

b

bB tagging

02χ

q / g

Benchmark point LM5Tan = 10AO = 0µ > 0M1/2 = 360 GeVM0 = 230 GeV

MSUGRA 5 parameters : m0, m1/2, A, tan , sign(µ)

S. Abdullin – L. Pape Update II of the new MSUGRA test points proposal

0 02 1

0 0

g

q

χ χ

H h

m ~ 860GeV

m ~ 800 GeV

m ~ 273 GeV m ~ 142 GeV

m ~ 570 GeV m ~ 116 GeV

Alain Romeyer - January 2005 3

Simulated events Production of signal events and pile-

up

Signal events : all SUSY channel (easier for later event reweighting)

~20% contains our decay chain

Digitisation : merge pile-up and signal

official 100 000 evts production at CERNavailable

our reconstruction strategy launched on digis

First results

Isajet 7.69 +

Pythia 6220

+ CMKIN

OSCAR2_4_6

ORCA7_6_1

generator

simulation

digitisationFU

LL S

IMU

LA

TIO

N C

HA

IN

Alain Romeyer - January 2005 4

« Reconstruction » strategy ORCA_8_6_1

Simulatedevents I – L1 +HLT

ReconstructionII - Jet Reconstruction :

• Iterative Cone Algo.• cone size : 0.5• tower Et thr. : 0.5 GeV • seed Et thr. : 1 GeV

III - B tagging : • By Track Counting• Number of tracks above minimum significance : 2

MET : • MET from EcalHcalTower

Analysis

• event selection• IV - b ̅b inv mass reco.

Efficiency ?

Jet sel.Pt > 10 GeV

Alain Romeyer - January 2005 5

Trigger efficiencyGenerated HLT

100 000 96 462 (19 338)

L1 : 99.9 % - HLT : 96 %

~ all events !!!

Trigger Condition (GeV) % of evts

1 jet Et > 571 23.5

3 jets Et > 209 21.7

4 jets Et > 122 32.7

B jet 1 Jet Et > 237 77.4

MET calo1 jet > 180 GeV + MET

> 123 79.5

MET jets1 jet > 180 GeV + MET

> 123 79.8

Low

Lum

i.

Caveat : ~ same efficiency for SUSY bkg.

Non cumulative Trigger !!!

Alain Romeyer - January 2005 6

Btagging

Number of jets

Number of tagged jets

Efficiency /

Inefficicency (%)

b 95593 47310 49.5

u 123 807 9408 7.6

d 97409 7957 8.2

c 50728 9103 17.9

Global efficiency : 49.5 % Global inefficiency : 9.4 %

Nhiggs = Nhiggs after HLT * ( Btag Eff. )² ~ 4700 evts

Track Counting Algo. (default setting)

Need to be optimised Filip

Alain Romeyer - January 2005 7

Efficiency / inefficiency

Alain Romeyer - January 2005 8

Efficiency / inefficiency…

Alain Romeyer - January 2005 9

Select b jets from h0

Select only b jets coming from h0 decay

No jet energy calibration (from parton content in a 0.3 cone)

Alain Romeyer - January 2005 10

Jet reconstruction accuracy

Alain Romeyer - January 2005 11

Jet calibration

Jet calibration method following CMS-note 1999/066

Association between rec. Jets and generated partons before hadro. (cone < 0.3)

2 types of jets : b jets and other Etrue

jet = sum(Equarks,gluons) Correction Factor C = Etrue

jets / Erecjets

Fit of the correction factor surface in the E vs || plane :

21 a

1 3 43

22

f(|η|)C η, E = +

E

f(|η|) = + |η|+ |η|

a

b b b b+ |η|

Alain Romeyer - January 2005 12

E correction for b jets

Alain Romeyer - January 2005 13

E correction for non b jets

Same operation is done with the jet momentum

Alain Romeyer - January 2005 14

Impact on Minv Select only b jets coming from h0 decay

Calibrated jets

Alain Romeyer - January 2005 15

Impact on jet reco. accuracy

Alain Romeyer - January 2005 16

Invariant mass (no cuts)

Alain Romeyer - January 2005 17

Likelihood fit Follow the Z’ µµ strategy (CMS AN 2004/018) Unbinned method

No constraints on the absolute background level only on the background shape

3 parameters : mh, and

S Sh s h bg h

tot. tot.

N NP(m )= P (m ,σ) + [1 - ] P (m )

N N

Fraction of signal NormalisedGaussian

Extract from background fit

S

tot.

N

N

Alain Romeyer - January 2005 18

Fit result (full stat)

S+BS = 2 ln 14.1 B

LL

Unbinned fi

t

Binned o

nly w

hen plo

tted

Alain Romeyer - January 2005 19

Various running time

50 000 evts – S = 10.9

25 000 evts – S = 7.9515 000 evts – S = 5.8

100 000 evts – S = 14.1

Alain Romeyer - January 2005 20

Conclusion For the first time the full analysis chain is working

higgs peak visible with significance above 5 ~15000 evts (~ 40 days of understood data)

Future : Include SM background Improve S/B ratio (cuts, b tagging optimisation…) Reduce the width of the invariant mass Run the code on the Zbb sample for cross check and

optimisation of the reconstruction strategy Fast simulation (FAMOS) in order to explore the SUSY

parameter phase space Contour plot for the CMS reach