Search for supersymmetry in hadronic final states with MT2

We present the results of a search for physics beyond the Standard Model (BSM) using data of 1.1 /fb integrated luminosity collected by the CMS experiment at the LHC. Fully hadronic final states were selected based on the"stransverse"mass variable MT2 and interpreted in various models of supersymmetry (SUSY). Two complementary analyses were performed targeting different areas of the SUSY phase space. All backgrounds were estimated using both simulation and data-driven methods. As no excess of events over the expected background was observed exclusion limits were derived.


Introduction
We describe a search [1] for physics beyond the Standard Model in pp collisions collected by the Compact Muon Solenoid (CMS) detector [2] at the Large Hadron Collider (LHC) at a centre-of-mass energy of 7 TeV. The results are based on a data sample of 1.1 fb −1 of integrated luminosity collected in 2011. We use the "stransverse mass" variable M T 2 [3] to select new physics candidates out of fully hadronic events. We divide our search into two channels: one targets high squark and gluino masses with a high M T 2 cut, the other heavy squarks and light gluinos with a medium M T 2 cut but including a b-tag and high jet multiplicities. In the following we describe the properties of M T 2 in sec. 2, our analyis strategy and event selection in sec. 3, and the background estimation in sec. 4. In sec. 5 we state the results of our search, and draw a conclusion in sec. 6.

The search variable M T 2
The variable M T 2 was introduced to measure the mass of primary pair-produced particles where both particles decay into detected and undetected particles (e.g. the lightest supersymmetric particle (LSP)). It is a generalization of the transverse mass m T in case of two identical decay chains each containing unobserved particles. The variable M T 2 is defined as where m T is the transverse mass of the visible system and the corresponding LSP χ of the decaying sparticle: In this analysis the stransverse mass M T 2 is not used for mass measurements but rather as a discovery variable [4].
In order to associate all visible decay products to the decay chains of the two sparticles we cluster the jets of an event into two "pseudojets" using a hemisphere algorithm a e-mail: hannsjorg.artur.weber@cern.ch described in [5], Sect. 13.4. As seeds (inital axes) the direction of the two (massless) jets are chosen which have the largest invariant mass. We then associate a jet k to the pseudojet i rather j if the Lund distance is minimal:

Advantages of M T 2
In order to gain a better understanding of the behaviour of M T 2 we take the case where we set all masses to zero and assume no initial-state radiation (ISR) or upstream transverse momentum 1 . In this simple case M T 2 becomes where p vis(i) T is the transverse momentum of pseudojet i, and φ 12 the angle between the two pseudojets in the transverse plane. It can be observed that for symmetric events (p vis(1) T = p vis(2) T ) with large acoplanarity M T 2 behaves like the missing transverse momentum (MET). Thus SUSY with expected large MET will accumulate in the high M T 2 region. However, back-to-back systems or balanced events will populate the region with small M T 2 . Thus M T 2 is robust against QCD jet mismeasurements: Mismeasurements along one of the pseudojets results in M T 2 ≈ 0 GeV, while for asymmetric mismeasurements still M T 2 < MET.

Analysis Strategy and Event Selection
In this analysis we have established two search channels in order to be sensitive to different regions in the SUSY phase space. One approach, the High M T 2 analysis, targets events resulting from heavy sparticle production which is characterized by large MET and M T 2 . The second approach, the Low M T 2 analysis, is designed to be sensitive to the region where squarks are heavy and gluinos relatively light. Here EPJ Web of Conferences gluino-gluino production is dominant, the gluinos giving rise to three-body decays with small MET. Also, as stops and sbottoms are expected to be relatively light, these events can be enriched with b-quarks. Thus the two strategies require two different sets of selection cuts stated in table 1.
Besides this channel specific selection we also require: -Lepton (e, µ) veto to reduce W+Jets and tt background.
-|MHT − MET| < 70 GeV to minimize the influence of unclustered energy (e.g. ISR) to the M T 2 shape 4 . -MET tail cleaning cuts (e.g. noise filters) to filter out events with unphysical MET. For the selection of data we require the data to pass H T trigger paths.

Background Estimation Strategy
For each type of background data-driven estimation methods have been designed: QCD is estimated from the bulk of the M T 2 distribution as described in sec. 4.1. In order to reduce the effect of signal contamination and statistical fluctuations the electroweak and top background is estimated from an adjacent control region in M T 2 . The prediction is taken from data in the control region scaled by Monte-Carlo (MC) ratio of the event yield in the signal region over the yield in the control region. Similarly the uncertainties are scaled by a MC ratio. The control region for High M T 2 is defined as 200 GeV < M T 2 < 400 GeV, for Low M T 2 it is defined as 100 GeV < M T 2 < 150 GeV.

QCD background estimation
The QCD estimation method is based on the two variables M T 2 and ∆φ min = min ∆φ(MET, any jet). These two variables are strongly correlated, but a factorization method can be applied if the functional form for the ratio r(M T 2 ) = N(∆φ min ≥ 0.3)/N(∆φ min ≤ 0.2) is known. From simulation studies it has been found that for M T 2 > 50 GeV the ratio falls exponantially: This behaviour was confirmed in data. The estimate has been performed by a fit from data in the QCD dominated region of 50 GeV < M T 2 < 80 GeV to extract the parameters a and b. In order to get also parameter c from data its value was fixed to the value of the ratio at M T 2 = 200 GeV where the ratio still falls exponentially. 2 The jet selection requires p T > 20 GeV, |η| < 2.4 3 H T is the scalar sum of all jet-p T . 4 MHT is the negative vectorial sum of all jet-p T .

Z → νν background estimation
Z → νν is an irreducible background because the produced neutrinos leave the detector unmeasured and thus generate real MET. The number of Z → νν+Jets events passing the event selection can be estimated from W → µν+Jets via where W(µν) is the number of W → µν events passing the event selection with additionally requiring one muon, R ZW is the ratio of Z → νν events to W → µν events, acc is the acceptance, and reco/iso is the combined reconstruction and isolation efficiency. In order to reduce the tt background in the W → µν selection a b-tag veto has been applied, while the residual tt background has been estimated from the b-tagged region. Furthermore reco/iso has been calculated from Tag & Probe studies on Z → ll events while the acceptance and the ratio R ZW are taken from MC.

W and Top background estimation
The background due to W and Top has two sources: Either a lepton (e, µ) from a W has been unobserved due to acceptance cuts, or has been "lost" due to failing either identification, isolation, or reconstruction criteria. The other source are W decays into neutrinos and taus which decay hadronically.
The number of events with a "lost" lepton has been estimated from the number of events with one lepton found in data. This number is then corrected for the probability to loose a lepton via the formula where N reco e,µ is the number of events containing a lepton, N bg e,µ is the expected background from processes other than W or Top, and ε e,µ is the probability for a W → lν (l = e, µ, or τ → e, µ) passing all selection and reconstruction cuts. The number of events with hadronic tau decays are taken from simulation and validated by data: The W → lν (l = e, µ, τ) kinematics in simulation has been validated in data with one muon. Furthermore, it has been shown that tau decays are well modelled in the simulation [6] justifying the use of simulation.

Results
The M T 2 distributions for the High M T 2 analysis and Low M T 2 analysis are shown in Figs. 1 and 2, table 2 summarizes the results of the two analysis strategies. As no excess over background has been found limits have been set.

Exclusion Limits
First, model independent limits on σ × BR within our acceptance has been derived by computing a 95% upper limit on the number of events using a CL s formulation [7]. These   Table 3. Observed and expected limits on σ × BR within the acceptance ot the two analysis strategies.
σ× BR (pb) observed limit expected limit High

Conclusion
We conducted a search for supersymmetry in hadronic final states using the M T 2 variable calculated from massless pseudojets. A data set containing 1.1 fb −1 of integrated luminosity in √ s = 7 TeV pp collisions recorded by the CMS detector during the 2011 LHC run was analyzed. Two complementary analyses were performed to probe a larger SUSY phase space. In both analyses the tail of the M T 2 is sensitive to a possible SUSY signal. As no evidence for a signal was found, we set upper limits on the cross section times branching ratio within our acceptance. Exclusions limits were established in the mSUGRA/CMSSM parameter space, as well as in a Simplified Model topology.