SUSY06, UC Irvine, June 2006Filip Moortgat, ETH Zurich SUSY06 Higgs (to bb) final states (including the Hemisphere Separation Algorithm) top final states

SUSY06, UC Irvine, June 2006 Filip Moortgat, ETH Zurich

SUSY06

• Higgs (to bb) final states (including the Hemisphere Separation Algorithm)

• top final states

• tau final states

New SUSY studies using 3rd generation tagging with the CMS detector

New CMS Physics TDR full simulation results on supersymmetry signatures with

Filip Moortgat, ETH Zurich


Neutralino decays in mSUGRA

• dilepton final state(e, , ) through sleptonintermediate stateLarger tan -> more taus

• higgs final stateDominant h0 decay to bbLarge area in parameter space

• dilepton final state through Z0

Neutralino 2 decays in mSUGRA:


Higgs final state

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.


Selection

• At least 4 jets with Et > 30 GeV • of which at least 2 b-tagged jets (discriminator > 1.5)• MET > 200 GeV• Highest jet Pt > 200 GeV• Second highest jet Pt > 150 GeV• Third highest jet Pt > 50 GeV• B-jets in the same hemisphere• Smallest R of b-jet pair (among R < 1.5)

for SUSY Higgs final states:

basic SUSY

basic SUSY

Higgs bb

optimizingS/B for SUSY

optimizingS/B for Higgs

Trigger stream = Jet + MET (L1 & HLT) : ~80% efficient HLT thresholds: 180 + 123 GeV


Kinematics : jets

• at least 4 jets with Et > 30 GeV • hard jets …

Jets with Et > 30 GeV, GammaJet calibration

SUSY SM backgrounds


Kinematics: MET

MET > 200 GeV

Missing transverse energy (calculated from the jets):


B-tagging

working point

• using Combined Secondary VertexAlgorithm, which combines track and secondary vertex properties into one discriminator : vertex mass, flight path, narrowness, track multiplicity, energy fraction, track impact parameters, …

• performance in multi-jet environment@ chosen working point:

b-tagging: 55% c-jets: 12% udsg-jets: 1.6%


“Hemisphere” separation

Inspired by “thrust/sphericity” methods, but now 2 axes per event due to LSP’s

Axis 1

Axis 2

Collect objects in 2 groups with their axis (iterative procedure)

Group1

Group2

• 2 primary sparticles in event

• they both decay into a cascade

separate both cascades willreduce pairing combinatorics!


• 3 association methods: assign object to the axis for which

1. the scalar product is maximal (|A| = 1) = pure angular test:

2. the hemisphere masses are minimal:

3. the minimal Lund distance:

Hemisphere association

APrr

•

)cos()cos( jkjjikii pEpE θ−≤θ−

2222jkijik mmmm +≤+

Axis i

Axis jJet k

22 )()cos(

)()cos(

kj

jjkjj

ki

iikii EE

EpE

EE

EpE

+−≤

+− θθ

=>

jkik θ≥θ coscos

• 3 seeding methods:

1. 1st axis: highest momentum object 2nd axis: object with largest P x ΔR (wrt A1)2. pair of objects with maximal invariant mass

• recalculate the axes as sum of objects; iterate till no object changes group


Hemisphere efficiencies

All jets quark jets

gluon jets

q from squark

q from gluino

LM1 81% 81% 79% 87% 73%

LM5 77% 77% 74% 87% 70%

LM9 74% 75% 69% --- 76%

(using seeding method 2, association method 3)

For jets, the probability that the jet is assigned to the “correct” hemisphere =

different CMS study points

Cleans up invariant mass plots (hq, hqq, lqq, …) significantly Does not work well for leptons (~massless) but there are tricks …


Angular separation of b-jets

SUSY SM

within one hemisphere, take only that combination of reconstructed b tagged jets for which the space angle R is smallest (among those with R < 1.5)

p p

b

b

q

q

q

h002÷

q%

g%q%

…

hemi1

hemi2


Cut flow

DATA TRIGGER NJET+NBJET MET PTJETCUTS DR+JET PAIRING SIGNAL 100 43.0 30.7 24.7 8.1 SUSYBG 100 89.8 11.9 9.1 1.4 TTBAR 100 19.0 3.6 1.1 0.1 Z+JET 100 0.7 0.1 0.02 < 10-4 W+JET 100 0.004 0.001 0.004 < 10-4

Selection efficiency for signal and main backgrounds (in percent), starting after the trigger (L1 + HLT Jet+MET).

Trigger efficiency: SUSY : 79% ttbar : 4%


Mass reconstruction

Background 5th order polynomial(with fixed parameters from off peak fit)

Signal gaussian G(mh, 18.2)

= fraction of signal in the plot

( ) ⎥⎦

⎤⎢⎣

⎡−⎟

⎠

⎞⎜⎝

⎛ ++= NSNB

NS1logNBNS2Scl

Scl(1 fb-1) = 4.5Scl = 5 for 1.5 fb-1

= 0.28 ± 0.08mh = 112.9 ± 6.6= 18.2Scl(1 fb-1) = 4.5

MhMC = 116 GeV/c²

extract bb width from other measurements (top, ZZ, …)

Mass fit:


Mass reconstruction(2)

²=0.6 = 0.28 ± 0.04mh = 118 ± 4.5= 18.2Scl(1 fb-1) = 7

MhMC = 116 GeV/c²

²= 1.5 = 0.24 ± 0.02mh = 118.5 ± 2.6= 18.2Scl(1 fb-1) = 11.5


Systematics

• Main systematic: Jet energy scale + MET scale (recomputed from corrected Jets):

• for 1 fb-1 (10 fb-1), JES uncertainty goes linear from 15%(10%) at 20 GeV to 5%(3%) at 50 GeV;

flat 5 %(3%) above 50 GeV;• leads to the following systematic uncertainties: 15 % (7 %) on SUSY event selection ;

17 % (10 %) on tt background rejection ;• Effect on fitted parameters estimated to be : ± 7.5 (5) GeV/c² on mh (and ± 0.04 (0.01) on ).

• Tracker misalignment : applying the “short term”

misalignment scenario (misalignment of about 100 µm on strips, 20 µm on pixels) => no effect on the position of the invariant mass distribution ; => observed a small drop in number of selected signal events due to

the reduced b-tagging efficiency


Reach in mSUGRA

SUSY Higgs bb


Top final states

2nd CMS study: inclusive top + MET signature for SUSY

• goal: examine low mass SUSY observability

• target stop and sbottom : often they are light; a lot of top quarks are generated in their decays

• different backgrounds than the inclusive jet + MET search

• use CMS test point LM1 where gluino cross section is high (35 pb)


Working point LM1

• LM1 parameters: M1/2 = 250, M0 = 60, tan = 10, sign() = +


Selection summary


Kinematic fit (2C)

• The purpose of this analysis is not to measure the top mass top mass is used with W mass as the 2 constraints to find the best jet combination.

=> To reject non-SUSY backgrounds (W+X) => To reject SUSY combinatorial backgrounds

• Only energy of Jets is smeared in the detector ( checked that directional errors have a small effect.)

• Last 2 terms: take into account the width of the particles. (Breit-Wigner approximated by Gaussian.)

• mW and mTop computed from jets. The third jet is a b-jet.


Kinematic Fit Probability

cut here

Peak at low probability ( = high 2 ) even for ttbar is due to hadronisation/fragmentation and jetfinding


Selection (2): kinematic fit


Reach


Di-tau final states

LM2 Point m0 = 185 GeV m1/2 = 350 GeVtan β = 35 A0 = 0; μ > 0

Always one very soft

tau per event

3rd CMS study: inclusive di-tau + MET signature for SUSY

For moderate and high values of tan , taus are dominating the di-lepton modes

In the following we consider only hardonic tau decays (1- or 3-prong)


Tau-tagging

•Need to get tau-jet at energy as low as possible Et Jet > 5 GeV

•Need to get as much as possible of visible energy: R<0.6

•Tau reco. based on standard Tau reco. algo.

with regional tracking signal cone R(lead-track,track)<0.1

and R(lead-track,Jet)<0.17

•1 or 3 tracks in signal cone, •No track in isolation cone (smallest fake tau rate) with Pt>1GeV (all tau candidates) & with Pt>0.1 (0.2) Ptleadtrack for single track candidate with Et>60GeV (Et<60GeV)•Charge assigned by taking sum of track charge•Lead track Pt>5GeV•Lead track transverse I.P.<0.7mm

Gives purity of 64% and efficiencies of 17% (Et<60 GeV) and 59% (Et>60GeV)


Selection

GammaJet) (IC5 jets from 1. miss GeVE t 150>

GeVEt 150>jet withjets 2least At 3.

2. At least two "good" tau reco. candidates

2), <( 21R if only considered cand. tau-di 4.

Jets obtained with Iterative Cone R<0.5 + GammaJet calibration


Selection: results


Reach


Conclusions

Several full simulation studies using 3rd generation tagging in CMShave been completed in the framework of the Physics TDR (out soon):

• Higgs + MET: b-tagging & hemisphere separation significant mSUGRA reach for 10 fb-1 (could be Higgs discovery channel)

• top + MET: kinematic fit for reconstructing hadronic tops mSUGRA reach for 10 fb-1 upto m1/2 ~ 500 GeV

• di-taus + MET: dedicated reconstruction for soft taus mSUGRA reach for 10 fb-1 upto m1/2 ~ 500 GeV


References/Acknowledgements

SUSY with h bb final states: F.M., A. Romeyer, P. Olbrechts, L. Pape (CMS Note 2006/090)Hemisphere separation algorithm: F.M., L. Pape SUSY with top final states: S. Paktinat, L.Pape, M. Spiropulu (CMS Note 2006/102) SUSY with tau final states: D. Mangeol, L. Houchu, U. Goerlach (CMS Note 2006/096)

Many thanks to the CMS Thesis Award Committee for the financial support.


Inclusive reach for 10 fb-1


Extra


Laser pointers, batteries & naturalness

How to put batteries in a laser pointer? (in a “natural” way)

simplified toy model :

• remember batteries have a + and a - side (for weak electro potential)• open your pen (turn upper part) & look inside• you see one electrical contact in the middle• no contact on the other side in the middle => need to take one contact from the side• batteries are + on top AND at the side !!!!• so - side needs to touch the inner electrical contact

there is only one solution (= ok with naturalness) i.e. + sign up.

Documents

SUSY06, UC Irvine, June 2006Filip Moortgat, ETH Zurich SUSY06 Higgs (to bb) final states (including the Hemisphere Separation Algorithm) top final states