Search for Supersymmetry at CMS in all-hadronic final states

Searches for Supersymmetry in final states with jets and missing transverse energy are presented. They were performed with data samples corresponding to up to 1.1 fb−1 of integrated luminosity collected by the CMS experiment at the LHC in pp-collisions at a center-of-mass energy of 7 TeV. No excess of events was observed above the expected Standard Model background, which has been measured using various data-driven techniques. Exclusion limits on the parameters of the constrained minimal supersymmetric extension to the Standard Model were derived. Presented at PANIC11: 19th Particles and Nuclei International Conference Search for Supersymmetry at CMS in All-Hadronic Final States Matthias Schröder, on behalf of the CMS collaboration Institut für Experimentalphysik, Universität Hamburg Abstract. Searches for Supersymmetry in final states with jets and missing transverse energy are presented. They were performed with data samples corresponding to up to 1.1 fb−1 of integrated luminosity collected by the CMS experiment at the LHC in pp-collisions at a center-of-mass energy of 7 TeV. No excess of events was observed above the expected Standard Model background, which has been measured using various data-driven techniques. Exclusion limits on the parameters of the constrained minimal supersymmetric extension to the Standard Model were derived.

Supersymmetry (SUSY) [1] is a promising theory addressing various shortcomings of the Standard Model (SM) of particle physics.At the Large Hadron Collider (LHC) at CERN, SUSY particles can be produced, predominantly via the strong interaction.Due to their high masses, they subsequently decay in cascades into SM particles with large momenta.In case R-parity is conserved, SUSY particles will be produced in pairs and the lightest SUSY particle will be stable.If only weakly interacting, it can not be detected directly but will result in missing transverse energy (/ E T ).
In the following, searches conducted by the CMS experiment [2] in all-hadronic final states are discussed, which are characterized by several high-p T jets, large / E T , and no isolated leptons.This channel has the largest branching ratio but at the same time faces a huge background from SM processes with the same signature, such as Z → ν ν+jets events as well as t t+jets and W +jets events, where the W either decays into light leptons which are not identified or into hadronically decaying τ leptons.In those cases, genuine / E T is caused by neutrinos.A different major background arises from QCD multi-jet events where one or more jets are severely mismeasured because of leptonically decaying heavy flavor hadrons inside the jet or instrumental effects.
Several complementary analyses are performed, all of which employ data-driven techniques to predict the SM background with minimal dependence on the Monte Carlo (MC) simulation.The inclusive, model-independent search with jets and missing momentum [3] aims at a detailed understanding of the different background contributions, while both the razor [4] and the α T analysis [5] use kinematic variables to identify events originating in decays of pairproduced heavy particles.The first two searches were conducted as cut-and-count experiments on the data collected in 2010 corresponding to an integrated luminosity of 36 pb −1 ; the α T analysis was performed as a shape analysis on a data set of 1.1 fb −1 collected in 2011.Further analyses are presented in [6].

JETS AND MISSING MOMENTUM ANALYSIS
The strategy of this analysis is to precisely understand the different SM backgrounds in order to be most inclusive w. r. t. new physics resulting in all-hadronic like final states.The central observables are H T , the scalar sum of the jet transverse momenta, and / H T , the magnitude of the negative vectorial sum of jet transverse momenta.The event selection starts from a baseline region requiring at least three jets with p T > 50 GeV and |η| < 2.5, H T > 300 GeV, / H T > 150 GeV, and no isolated leptons.Additionally, the / H T is to be non-aligned with any of the leading jets to suppress QCD.Two search regions were defined by tightening the baseline selection requirements in order to increase the sensitivity for different SUSY models: a high-/ H T region at / H T > 250 GeV and a high-H T region at H T > 500.The QCD background contribution was estimated from a sample of inclusive multi-jet data using the "rebal-ance+smear" method: an estimator of the particle level / H T distribution is obtained by rebalancing the jet momenta in each event, taking into account the jet p T resolution.Finally the rebalanced momenta are smeared by the full jet p T response distribution to predict the multi-jet kinematics.The jet response distribution was taken from the MC simulation and adjusted to the data by factors obtained from photon+jet and di-jet data [7].Background contributions from t t and W +jets events were estimated from a muon+jets control sample: The number of events with a lost lepton was predicted by scaling the control sample with the lepton reconstruction and identification efficiencies as well as acceptance factors, and the number of events hadronically decaying tau leptons by replacing the muon by a simulated jet using a tau response template.The Z → ν ν background contribution was estimated from a photon+jets control sample, where the photon was ignored to model the impact of the neutrinos and corrected by the ratio of photon+jets and Z+jets distributions as predicted by the theory [8].In the high-H T region, 40 events have been observed and 15 in the high-/ H T region, which is compatible with the SM expectations of 43.8 ± 9.2 and 18.8 ± 3.5, respectively.

RAZOR ANALYSIS
This analysis aims at a kinematic identification of events with pair-produced, heavy particles.Hence, two or more jets were required for the baseline selection, and they were combined into two "mega-jets" using an event hemisphere algorithm in order to approximate the visible part of the decay products.FIGURE 1. Left: M R distribution for R > 0.5 in data (circles) compared to the SM backgrounds predicted from data (solid lines) and two SUSY benchmark scenarios (dashed lines) [4].Right: α T distribution measured in data (circles) and expected from the MC simulation for the SM backgrounds (solid lines) and two SUSY benchmark scenarios (dashed lines) [5].
In each event, the kinematic variables R ("razor") and M R are computed from the energies and momenta of the "mega-jets" as well as the / E T .For signal events, R is expected to peak at ≈ 0.5 and M R at a characteristic scale, which depends on the mass difference of the initial heavy particle and the invisible part of the decay products.In contrast, SM events, in particular QCD events, feature lower values of R and M R [10].Above a certain value that depends mostly on the trigger efficiencies, the M R distribution of QCD multi-jet events with R < R max drops exponentially, as shown in Fig. 1 (left).The slope of the exponential function was found to be proportional to R 2 max , which is the key property exploited in the background prediction: the slopes for low R max were fitted with a linear function of R 2 max and extrapolated into the search region, defined as R > 0.5, M R > 500 GeV.The overall normalization was obtained by fitting the predicted M R distribution to the data in a control region with 80 < M R < 400 GeV.The other SM backgrounds were estimated similarly.Seven events were observed in the search region, in agreement with the SM expectation of 5.5 ± 1.4.

α T ANALYSIS
In this analysis, the QCD multi-jet background is strongly reduced using the kinematic variable α T [9] as a discriminator between events with real and fake / E T .Again, two pseudo-jets were constructed in each event from all jets with E T > 50 GeV and |η| < 3, and α T is computed from their energies and momenta.In case of QCD multi-jet events, where the transverse jet momenta balance, α T is expected to be < 0.5 while it can have larger values for processes with genuine / E T , as shown in Fig. 1 (right).Therefore, the search region is defined as α T > 0.55.The residual QCD contribution and the other SM backgrounds in the search region were predicted by scaling the yields for α T < 0.55 in different control samples by R α T , the ratio of the number of events with α T > 0.55 and α T < 0.55.In case of QCD multi-jet events, where / H T is caused by jet mismeasurements, R α T is expected to fall with increasing H T because the jet p T resolution improves with H T .However, no dependence of R α T on H T has been observed in This is expected for events with real / E T , demonstrating the QCD depletion in the search region.Hence, R α T was modeled exponentially falling with H T for the QCD contribution and constant for the other SM processes.The final background prediction was obtained by a maximum-likelihood fit of the expected background yields in each H T bin to the data.A muon+jets sample was used as control sample for the t t and W +jets contributions, a photon+jets sample for the Z+jets contribution, and the QCD contribution was estimated from a multi-jet sample collected without requirements on / H T at trigger level.Good agreement between the measured yields and the best fit was found with a p-value of 0.56, indicating an appropriate parametrization of R α T and compatibility of number of events observed in data with the SM expectation.

RESULTS AND CONCLUSIONS
Complementary searches for SUSY in the all-hadronic final states have been performed by CMS.The observed event yields are consistent with the SM background expectations, which have been predicted from data.Consequently, limits have been set on the parameters of the constrained minimal supersymmetric extension to the SM (CMSSM).They are shown in Fig. 2. The CMS searches exceed the sensitivity of previous collider experiments already with the 36 pb −1 of data collected in 2010; with 1.1 fb −1 , squark masses lower than 1.1 TeV are excluded at the 95% confidence level.

FIGURE 2 .
FIGURE 2. Exclusion limits in the m 0 − m 1/2 parameter space of the CMSSM at the 95% confidence level.Left: observed limits of the 2010 analyses[6].Right: observed (solid and dashed lines) and expected (dotted line) limits of the 2011 α T analysis[5].Regions excluded by previous collider experiments are shown for comparison.