HCOL seminar in KEK. Review of recent results from LHC.

1. Search for an excess of events with an identical
flavour lepton pair and signicant missing transverse
momentum
* ATLAS collaboration
* ArXiv:1103.6208
* Submitted to EPJC
—
* √s=7TeV
∫Ldt=35pb-1 were collected between March and November
| | | | | >
t
2009 2010 2011

2. Aims
a search for the supersymmetric (SUSY) particles in events with
exactly two leptons of identical flavour (e or μ) and opposite
charge, and signicant missing transverse momentum
lepton pair invariant mass
> { distribution }
l+/- l+
>
~0 ~0 l-
>
χ2 l-/+ χ2
• •
~
l
•
~0 ~0
χ1 > χ1
miss >
ET

3. Decay chains
-
q
~0
χ1
q
>
• > -
q
• ~± •
>
> χ1 W±
q
~ • -
~
g q
>
q q
>
• > •
> ~0
χ2 l+
~
•
>
q
g •
>
>
~
l • > l-
q
~0
χ1

4. End-points in mll distribution
(m 2̃ −ml2̃ )(m 2̃ −m2̃ )
χ0
l χ 0
2 edge R R
(m ) ll = 2
2
1
m l̃R
OSOF subtraction
In signal region
Model:SPS1a
* Gjelsten, Hisano, Kawagoe, Lytken, Miller, Nojiri, Osland & Polesello
(in LHC/IC study group) 04
* SUSY Parameter and Mass Determination at the LHC, C. Sander, Cambridge Phenomenology
Seminar, 2010

5. Features
* one of the best routes to model-independent measurements of the
masses of SUSY particles via end-points in the lepton pair invariant
mass distribution
* Standard Model background is (almost) equal for lepton pairs of
identical and different flavour (in the signal region):
Bg(e+e-) = Bg(μ+μ-) = Bg(e+μ-)= Bg(μ+e-)
* SM bg. can be removed with a ‘flavour subtraction’ procedure:
Signal(e+e-Vμ+μ-)=Data(e+e-)+Data(μ+μ-)-Data(e+μ-Vμ+e-)

6. Monte-Carlo
is used to develop analysis procedure and estimate residual SM bg.
}
* QCD jets PYTHIA, LO-PDF: MRST2007LO*
>
* Drell-Yan
}
* top quark pairs MC@NLO, NLO-PDF: CTEQ6.6
>
* single top
* W and Z/γ* production > ALPGEN
}
* Diboson production (WW, WZ, ZZ)
> HERWIG
* Fragmentation and hadronization
* Underlying event > JIMMY
* Parameter tune > ATLAS MC09
* Detector simulation > GEANT4

7. Electron identification (general)
ATLAS, JHEP12(2010)060
starts in the high-granularity liquid-argon sampling electromagnetic (EM)
calorimeters. Further, there are three reference sets of requirements:
“Loose”: uses EM shower shape information and discriminant variables from
hadronic calorimeters
“Medium”: full information from EM + some from the inner tracking detector (ID)
(track quality variables + cluster-track matching variable)
“Tight”: exploits the full electron identication potential of the ATLAS detector (full
information from ID and EM)

8. Electron identification (ArXiv:1103.6208)
* pass “tight” electron selection criteria
Pseudorapidity:
η =-lnTan(θ/2)
* have pT>20GeV and |η|<2.47
* additional E/P cut (E is the shower energy in the EM and p the track
momentum in the ID) and TRT (transition radiation tracker) cut to provide
additional rejection against conversions and fakes from hadrons.
* pass isolation criteria: total transverse energy within a cone ΔR=√(Δη)2+(Δφ)2
around the electron is less than 0.15 of the electron pT
* Veto if a “medium” if a medium electron is found in the transition region
between the barrel and end-cap EM : 1.37<|Δη|<1.52

9. Muon identification
* have pT>20GeV and |η|<2.4
There are two possibilities
“Combined muons” are identified in matching between an extrapolated ID
both the ID and MS (muon track and one or more track segments
spectrometer) systems in the MS
* ID track quality test – several (or at least one) silicon pixel detector hits,
silicon microstrip detector (SCT) hits, TRT hits
* a good match between ID and MS tracks
* pT measured by these two systems must be compatible within the resolution
* isolation: ΣpT<1.8GeV for other ID tracks above 500 MeV within ΔR<0.2 around
the muon track
* Δz<10mm between the primary vertex and the extrapolated muon track

10. Jet identification
* have pT>20GeV and |η|<2.5
* reconstruction using the anti-kT-algorithm with a distance parameter D=0.4
* Jets are corrected for calorimeter non-compensation, material and other effects
using pT- and η-dependent calibration factors obtained from Monte Carlo and
validated with test-beam and collision-data studies

11. Common criteria
* identified “medium” electrons or muons are only considered if they satisfy
ΔR>0.4 with respect to the closest remaining jet
* if a jet and a “medium” electron are both identified within a distance ΔR<0.2 of
each other, the jet is discarded (?)
* ET is the modulus of the vector sum of pT of the reconstructed objects (jets with
pT>20GeV but over the full calorimeter coverage |η|<4.9, and selected leptons), any
additional non-isolated muons, and the calorimeter clusters not belonging to
reconstructed objects.

12. Signal region
* events that contain a lepton pair of identical or different flavour
* signs of the leptons are opposite
* invariant mass mll>5GeV
* missing ET>100GeV in order to reject SM Z+jets events whilst maintaining
effciency for a range of SUSY models.
* events must also possess at least one reconstructed primary vertex with at least
five associated tracks (?)

13. Flavour subtraction
Using the quantity S defined as follows:
N(e+e-) βN(μ+μ-) N(e+μ-Ve-μ+)
S= + -
β(1-(1-τe)2) (1-(1-τμ)2) (1-(1-τe)(1-τμ))
* electron plateau trigger efficiency τe=(98.5±1.1)%
* muon plateau trigger efficiency τμ=(83.7±1.9)%
* the ratio of electron to muon effciency times acceptance β=0.69±0.03
* the value of S obtained from selected identical-flavour and different-flavour
lepton SM events is expected to be small but non-zero, due primarily to Z/γ*
boson production

14. Data vs MC
N(e+e-) βN(μ+μ-) N(e+μ-Ve-μ+)
+ and
β(1-(1-τe ) 2) 1-(1-τμ )2 1-(1-τe)(1-τμ)
* weighted invariant mass distribution of
e+e- or μ+μ- pairs prior to applying the
missing ET requirement
* the distribution for different flavour
pairs
* in the region with mll < 100 GeV, the
dominant contributions to the different
flavour data events are expected to come
-
from tt, QCD and Z/γ*+jets events

15. Data vs SM background
e+e- e+μ- or e-μ+ μ-μ+ dominated, but
cancel out in S,
Data 4 13 12 but signal-free RMS
dominated by stat.
>
Z/γ*+jets 0.40±0.46 0.36±0.20 0.91±0.67 fluctuations in number
of tt events
Diboson 0.30±0.11 0.36±0.10 0.61±0.10
> Actually dominated in S
tt 2.50±1.02 6.61±2.68 4.71±1.91
Single top 0.13±0.09 0.76±0.25 0.67±0.33
> negative is an artifact
Fakes 0.31±0.21 -0.15±0.08 0.01±0.01
Total SM 3.64±1.24 8.08±2.78 6.91±2.20
Sobs=1.91±0.15(β)±0.02(τe)±0.06(τμ) SSM=2.06±0.79(stat.)±0.78(sys.)

16. Contibution to S from SM
* from single top and diboson events are estimated using the MC samples described
above, scaled to the luminosity of the data sample
* from Z/γ*+jets, tt and events containing fake leptons (from QCD jets and W+jets
events) are estimated using MC samples normalised to data in an appropriate
control region
-
Z/γ* tt fake leptons
* (ET)miss<20GeV * “top-tagged” lepton * electron with pT>30GeV,
* 81<mll<101GeV pair (ET)miss<60GeV,
* 60<(ET)miss<80GeV
Δφ<0.1 between a jet and the (ET)miss vec.
* ≥2 jets with
* muon with pT<40GeV, (ET)miss<30GeV,
pT>20GeV
mT(μ,ET)<30GeV
* A loose-tight matrix method is used to estimate the number of events with
fake leptons in the signal region

17. Consistency between Sobs and SSM
* generating pseudo-experiments
using the estimated mean numbers
of background events from Table 1
as input
* The resulting total mean number of
background events in each channel is
then used to construct a Poisson
* The probability of observing a
distribution from which the observed
value of S at least as large as Sobs
number of events in that channel is
is 49.7% and hence no evidence
drawn
of an excess of identical flavour
events beyond SM expectations is
* 106 pseudo experiments
observed.

18. Model-independent constrains on Ssignal
* adding signal event contributions to the input mean numbers of
background events in each channel
* assumption about the relative branching ratio of new physics events into
identical flavour and different flavour channels
* new set of signal-plus-background pseudo-experiments
* If Brnew physics(eμ)=0, then * If Brnew physics(eμ)=½Brnew physics(ee+μμ),
Ssignal<8.8 at 95% confidence level Ssignal<12.6 at 95% confidence level

19. Model-dependent constrains
* mean numbers of signal events added to each channel are sampled according to
the expectations from each point in the parameter space of the model together
with the uncertainties in these expectations
* 24 parameter MSSM model: mA=1TeV, μ=1.5minP(mq,mg), tanβ=4, At=μ/tanβ,
Ab=Al=μtanβ. The masses of the 3d generation sfermions are set to 2 TeV, and
common squark mass and slepton mass parameters are assumed for the first two
generations
* Two grids in the (mq,mg) plane are considered (MSSM PhenoGrid2):
“compressed spectrum”: “light neutralino”:
m χ̃ =M −50GeV
0 m χ̃ =M −50 GeV
0
2 2
m χ̃ =M −150GeV
0 m χ̃ =100GeV
0
1 1
ml̃ = M −100 GeV
L
m l̃ =M /2
L
M =min( mq , m g )
̃ ̃ M =min( m q , m g )
̃ ̃

20. For “compressed spectrum” (“light neutralino”) models and
mg = mq + 10 GeV, the 95% confidence lower limit on mq is 503 (558) GeV

21. Conclusion
* a flavour subtraction technique has been used to search for an excess
beyond SM expectations of high missing transverse momentum events
containing opposite charge identical flavour lepton pairs
* no signicant excess has been observed, allowing limits to be set on the
model-independent quantity Ssignal, which measures the mean excess from
new physics taking into account flavour-dependent acceptances and
effciencies.