Prepared for submission to JINS International Conference on Instrumentation for Colliding Beam Physics 7 February - March, 7 Budker Institute of Nuclear Physics, Novosibirsk, Russia ALAS jet and missing energy reconstruction, calibration and performance in LHC Run- A. Hrynevich on behalf of the ALAS collaboration Institute for Nuclear Problems of Belarusian State University, Bobrujskaya Str., Bdg., Minsk, Belarus E-mail: aliaksei.hrynevich@cern.ch AL-PHYS-PROC-7-45 May 7 Abstract: he performance of the reconstruction and calibration of the jet energy scale and missing transverse energy scale with the ALAS detector at the LHC is a key component to realize the ALAS full physics potential, both in the searches for new physics and in precision measurements. New algorithms used for the reconstruction and calibration of jets and missing energy with the ALAS detector during LHC Run- are presented. Measurements of the performance and uncertainties are derived from data. he results from the 6 pp collision data set at s = ev are reported. Keywords: Analysis and statistical methods, Missing ransverse Energy studies
Contents Introduction Jet reconstruction and calibration Jet energy scale uncertainty 4 Missing transverse energy performance 4 5 Conclusions 6 Introduction Jets, collimated sprays of hadrons, are the dominant physics objects arising in proton-proton collisions at the LHC. Jets play a key role in a variety of Standard Model physics analyses and searches for new phenomena in the ALAS experiment. he precision of the jet energy measurement directly affects the performance of the missing transverse energy reconstruction. Missing transverse energy is a measure of the transverse momentum imbalance created by detected and well measured objects in an event. A good performance of missing transverse energy reconstruction is crucial for many SUSY and dark matter searches. he algorithms for the jet and missing transverse energy reconstructions, as well as for the jet energy scale (JES) calibration were developed and validated in ALAS during the LHC Run-. he proton-proton collisions at the centre-of-mass energy of s = ev at the LHC Run- force the evolution of these algorithms in presence of multi-ev final states towards the highest precision. he main algorithms for the jets and missing transverse energy reconstructions are presented and their performances at Run- ALAS data and Monte Carlo simulation are reported. Jet reconstruction and calibration he main jet identification algorithm used by the ALAS collaboration is the anti-k t algorithm [] with a distance parameter R =.4. Various objects can be used as inputs to this algorithm: calorimeter energy deposits, inner detector tracks [] or a combination of both [4]. Stable particles with a lifetime longer than ps, excluding muons and neutrinos, are used for the jet identification in the Monte Carlo simulation. Jets reconstructed from tracks, also referred to as the track jets, have low dependence on the pile-up activity since only tracks originating from the primary vertex are used for the jet finding. However, the reconstruction of track jets is limited by the ALAS tracker acceptance to η <.5. herefore, the majority of ALAS analyses uses a jet reconstruction based on calorimeter deposits, calorimeter jets. he inputs for calorimeter jets reconstruction are the topologically clustered calorimeter cells, so-called topo-clusters [5]. he topological clustering algorithm groups cells with the significant
energy deposits aiming for effective noise suppression. he total noise in calorimeter cells, σ, is calculated as the quadratic sum of the measured electronics and pile-up noise. he clustering algorithm starts from the seed cells with the energy deposits above 4σ. All neighbor cells are iteratively added to the topo-cluster if their energy is above σ. his is followed by the addition of all adjacent cells. As the final step, the cluster splitting algorithm separates produced topo-clusters based on local energy maxima to avoid overlap. opo-clusters are considered to be massless and only those with positive energy are used for jet reconstruction. he energy of the calorimeter cells is measured at electromagnetic (EM) scale, established using electrons at test beams. he local cell weighting (LCW) calibration [6] can be applied to topo-clusters classified as hadronic to account for the difference in the detector response to electromagnetic and hadronic particles. In addition, it improves the jet energy resolution. he LCW calibration is derived using Monte Carlo simulation of single pion events. he jet energy scale calibration [7] restores the energy scale of reconstructed jets to that of simulated truth jets. Different sets of correction factors are developed for jets reconstructed using the EM and LCW topo-clusters. he JES calibration includes origin correction, pile-up correction, absolute correction of the detector response based on Monte Carlo simulation, global sequential correction and residual in situ calibration. he origin correction forces the four-momentum of the jet to point to the hard-scatter primary vertex rather than to the center of the detector while keeping the jet energy constant. he centre-of-mass energy in proton-proton collisions of s = ev and the reduced bunch spacing intervals (5 ns) at Run- increased the pile-up, biasing the measured jet transverse momentum. A two step procedure is used to subtract the pile-up contribution. First, it removes the effect of pile-up exploiting the average energy density and the area of the jet. Second, a residual correction removes the remaining dependence of the jet on the number of reconstructed primary vertices (N PV ) and the expected average number of interactions per bunch crossing ( µ ). he performance of the pile-up correction is shown in Figure (a). he jet energy scale and η calibration correct the reconstructed jet to the particle-level energy scale to account for the difference of the calorimeter energy response using Monte Carlo simulation. In addition, it corrects any bias of the reconstructed jet η caused by the transition between different calorimeter regions and the difference in calorimeter granularity. he energy response as a function of detector η for jets of different simulated truth energy is shown in Figure (b). he global sequential correction (GSC) [8] is designed to reduce the jet response dependence on the flavour of the initiated-jet parton. he correction uses global properties of the jets such as the portion of the jet energy measured in the first layer of the hadronic calorimeter, the portion of the jet energy measured in the third layer of the electromagnetic calorimeter, the average p -weighted transverse distance in the η-φ space between the jet axis and all tracks associated to the jets, the number of tracks associated to the jet and the number of muon track segments associated to the jet, accounting for punch-through correction. he correction removes the jet response dependence on the listed observables. he in situ JES calibration is applied to jets measured in data. he correction is calculated as the jet response difference between data and Monte Carlo simulation using the transverse momentum balance of a jet and a well-measured reference object.
[GeV] p / N PV. ALAS Simulation Preliminary Pythia Dijet s = ev.8 anti-k t EM R=.4.6.4. -. -.4 Before any correction After ρ A subtraction After residual correction.5.5.5.5 4 4.5 η Energy Response..9.8.7.6.5.4.. E true = GeV E true = 6 GeV E true = GeV E true = 4 GeV E true = GeV ALAS Simulation Preliminary Pythia Dijet s = ev anti-k t EM jets R=.4 true > GeV -4 - - - 4 η det E (a) (b) Figure. (a) he dependence of EM scale jet transverse momentum on N PV as a function of η before pile-up corrections (blue), after the area-based correction (violet), and after the residual correction (red). (b) he average jet energy response as a function of the detector pseudorapidity for different truth energies [9]. Jets from the central detector region ( η det <.8) are used to derive a residual calibration for jets in the forward region (.8 < η det < 4.5) exploiting dijet events (η-intercalibration). Figure (a) shows the relative jet response measured using Run- ALAS data with the integrated luminosity L =.6 fb and in Monte Carlo simulation as a function of detector pseudorapidity in a single jet p bin. he lower panel depicts the η-intercalibration correction. wo methods to calibrate jets with p up to 95 GeV using Z+jet and γ+jet events are employed. he Direct Balance method relies on perfect transverse-momentum balance of a jet and a reference Z-boson or γ in the process at leading order. However, the measurement can be affected by additional parton radiation contributing to the recoil of a boson and appear as subleading jets, or due to energy loss outside of the considered jet cone. he Missing Projection Fraction method utilizes the full hadronic recoil of the event rather than a single reconstructed jet, and thus less sensitive to the jet definition and out-of-cone radiation. However, it does not directly reflect the energy within the reconstructed jet cone. hese two methods are complementary and both are pursued to check the compatibility of the measured response. opologies with three or more jets are used to calibrate high-p jets (up to ev) using a recoil system composed of several lower-p jets (Multijet Balance). he combined in situ corrections are shown in Figure (b) as a function of the jet p measured using Z+jet, γ+jet and multijet events. he listed calibrations are derived and applied sequentially and the systematic uncertainties are propagated through the calibration scheme. Jet energy scale uncertainty he jet energy is measured with the precision of up to % using Run- ALAS data with the integrated luminosity L =.6 fb. he measurement is described by a set of 8 independent sources of systematic uncertainties. he total systematic uncertainty is shown in Figure as a function of jet transverse momentum at η =. he total uncertainty at low jet p reaching up to 6% is driven by the in situ methods uncertainties and the difference of the jet flavour composition
Relative jet response (/c) MC / data...9 ALAS Preliminary anti-k t R =.4, EM+JES - s = ev,. fb avg 85 p < 5 GeV Data 5 Powheg+Pythia8 Sherpa.5.95.9 4 4 η det MC /Response Data Response..5.95.9.85.8 R=.4, EM+JES anti-k t Data 5 γ+jet Z+jet Multijet ALAS Preliminary - s = ev,. fb η <.8 otal uncertainty Statistical component jet p [GeV] (a) (b) Figure. (a) Relative jet response measured using Run- ALAS data and Monte Carlo simulation as a function of detector pseudorapidity in a single jet p bin. he lower panel depicts the η-intercalibration correction. (b) Combined in situ corrections as a function of the jet p measured using Z+jet, γ+jet and multijet events. he total uncertainty and its statistical component are shown with shaded bands. []. between Monte Carlo generators. he in situ methods uncertainties of about % solely dominate at high jet transverse momentum p > GeV. Jets with transverse momenta above ev are not covered by Multijet Balance calibration and a larger uncertainty is taken from the single particle response measurement. Each uncertainty is independent from the others and fully correlated across p and η. A reduced set of systematic uncertainties is available for physics analyses while minimising the loss of correlations. Fractional JES uncertainty..8.6.4 Data 5, s = ev ALAS Preliminary anti-k t R =.4, EM+JES + in situ correction η =. otal uncertainty Absolute in situ JES Relative in situ JES Flav. composition, unknown composition Flav. response, unknown composition Pileup, average 5 conditions Punch-through, average 5 conditions. 4 jet p [GeV] Figure. Relative jet energy scale uncertainty as a function of jet transverse momentum at η = []. 4 Missing transverse energy performance he missing transverse momentum ( E miss ) is reconstructed as the negative vector sum of transverse momenta ( p ) of reconstructed physics objects. he magnitude of the missing transverse energy is denoted by E miss. he physics objects considered in the E miss calculation are electrons, photons, muons, τ-leptons and jets (hard terms). he reconstructed momentum not associated to any of the hard terms is referred as the soft term and is also considered in the E miss calculation. Several algorithms can be used to reconstruct the E miss soft term using calorimeter energy deposits or tracks []. he main algorithm for the soft term reconstruction used by ALAS at Run- fully relies on tracks, the so-called rack Soft erm (S). he algorithm is very robust against varying pile-up condition, but it misses the contribution from neutral particles. 4
he removal of pile-up jets is essential for E miss resolution. his is done with the jet-vertextagger (JV) technique which extracts the pile-up jets using track-to-vertex association method []. In addition, a novel forward pileup tagging technique (fjv) that exploits the correlation between central and forward jets originating from pileup interactions is developed []. he fjv improves the E miss resolution in high pile-up conditions as shown in Figure 4. Resolution [GeV] miss, E y miss E x 5 5 5 ALAS Simulation Preliminary Powheg+Pythia8 Z µµ s = ev S forward jet S and p > GeV S and fjv 5 5 5 5 4 Number of primary vertices N PV Figure 4. he rack Soft erm E miss resolution as a function of N PV measured in Monte Carlo simulated Z µµ events using different strategies for pile-up suppression [4]. he performance of S E miss is validated using events with small E miss such as Z ll, W lν and t t events. Good agreement of S E miss measured using 8.5 fb Run- ALAS data and Monte Carlo simulation is observed as shown in Figure 5 (a) with Z ee events. he S systematic uncertainties are evaluated exploiting the differences between data and Monte Carlo using the balance of a soft term and a calibrated physics objects. he systematic uncertainties of each hard term are propagated to the E miss. he mean of the S distribution as a function of the hard term p measured using 6.5 fb Run- ALAS data agrees with Monte Carlo simulation within the systematic uncertainty as shown in Figure 5 (b). Events / GeV Data/MC 8 7 6 5 4 ALAS Preliminary Data 6, s = ev - Z ee, 8.5 fb Data Z( ee) Z( ττ) Diboson.5.5 5 5 5 5 4 45 5 miss S E [GeV] tt Single t MC Stat. [GeV] miss,softerm E MC / Data 5 ALAS Preliminary Data 5+6, s= ev - Z ee, 6.5 fb jets, p > GeV Data 5 Powheg+Pythia8 Madgraph+Pythia8 Powheg+Pythia8 AFII Sherpa S syst. uncert.. 4 5 6 7 8 9.8 4 5 6 7 8 9 hard p [GeV] (a) (b) Figure 5. (a) S E miss distribution for a selection of Z boson decays to a pair of electrons at Run- ALAS data. he expectation is superimposed by Powheg+Pythia8 Monte Carlo simulated events for the relevant signal physics processes including some background processes while diboson backgrounds use Sherpa. he shaded band represents the statistical uncertainty of Monte Carlo simulations [5]. (b) he mean of the S distribution projected in the direction longitudinal to the hard term p for Z ee events measured using Run- ALAS data and Monte Carlo simulation. he shaded band represent the systematic uncertainty [4]. 5
5 Conclusions he LHC opens new frontiers in particle physics using proton-proton collisions at the centre-of-mass energy of s = ev at Run-. he multi step jet energy scale calibration brings reconstructed jet energies to that of particle level reaching about % precision over a wide jet p range. he reconstructed missing transverse energy is validated using Run- ALAS data showing a good agreement with Monte Carlo simulation. References [] ALAS collaboration, he ALAS experiment at the CERN LHC, JINS (8) S8. [] M. Cacciari, G. P. Salam, and G. Soyez, he anti-kt jet clustering algorithm, JHEP 4 (8) 6. [] ALAS Collaboration, Jet energy measurement and its systematic uncertainty in proton proton collisions at s = 7 ev, Eur. Phys. J. C 75 (5) 7. [4] ALAS Collaboration, Jet reconstruction and performance using particle flow with the ALAS Detector, arxiv:7.485. [5] ALAS Collaboration, opological cell clustering in the ALAS calorimeters and its performance in LHC Run, arxiv:6.94. [6] ALAS Collaboration, Local Hadronic Calibration, AL-LARG-PUB-9-, URL: https://cds.cern.ch/record/5. [7] ALAS Collaboration, Jet energy scale measurements and their systematic uncertainties in proton-proton collisions at s = ev with the ALAS detector, arxiv:7.9665. [8] ALAS Collaboration, Jet global sequential corrections with the ALAS detector in proton-proton collisions at s = 8 ev, ALAS-CONF-5-, URL: https://cds.cern.ch/record/68. [9] ALAS Collaboration, Jet Calibration and Systematic Uncertainties for Jets Reconstructed in the ALAS Detector at s = ev, ALAS-PHYS-PUB-5-5, URL: https://cds.cern.ch/record/76 [] Public plots: Jet energy scale uncertainties updated for ICHEP 6 using full ev 5 dataset, URL: https://atlas.web.cern.ch/atlas/groups/physics/plos/jem-6-/ [] ALAS Collaboration, Performance of algorithms that reconstruct missing transverse momentum in s = 8 ev proton-proton collisions in the ALAS detector. Eur. Phys. J. C 77 (7) 4. [] ALAS Collaboration, Performance of pile-up mitigation techniques for jets in pp collisions at s = 8 ev using the ALAS detector Eur. Phys. J. C 76 (6) 58. [] ALAS Collaboration, Forward Jet Vertex agging: A new technique for the identification and rejection of forward pileup jets, AL-PHYS-PUB-5-4, URL: https://cds.cern.ch/record/498. [4] 5-6 S Systematic and Forward Pileup Suppression in ME, URL: https://atlas.web.cern.ch/atlas/groups/physics/plos/jem-7-/. [5] Public Plots: Missing ransverse Momentum Distribution and Performance in 6 data, URL: https://atlas.web.cern.ch/atlas/groups/physics/plos/jem-6-8/ 6