Dijet Mass Resolution of the CMS Calorimeter

We have investigated the jet resolution of the CMS calorimeter using the CMSIM GEANT simulation. Dijet decays of Z and Z 0 (mZ0 = 1 TeV) bosons were used as a benchmark process. 1 Simulation of the Calorimeter The calorimeter was simlated using GEANT3 within the framework of the CMS general simulation program CMSIM (version 111). The electromagnetic calorimeter (ECAL) consisted of 23 cm PbWO4 crystals followed by the support structure. The hadron calorimeter (HCAL) consisted of layers of copper absorber, 5.0 cm thick (7.9 cm in the endcap) separated by air gaps containing the scintillator packages. Outside the solenoid, two tail catcher (HO) layers are implemented in the muon system. The thickness of the calorimeter inside the colenoid was 6:9 at = 0. The detector model follows the TDR-2 design described in the HCAL Technical Design Report. Hadronic showers were simulated using the GCALOR package, with energy cutoff values of 1 MeV for electrons and photons, and 10 MeV for hadrons. The energy deposit in the crystals and scintillators was stored in an array of cells of size = 0:087 0:087 for j j < 2:262 and 0:174 0:174 above this value. 2 Dijet Resolutions As benchmark processes for moderate and high-ET jets, we used Z ! jj and Z 0 ! jj (mZ0 = 1 TeV). Events were generated using PYTHIA processes 1 and 141, and were then run through the calorimeter simulation. Jets were reconstructed using a simple cone algorithm. The ECAL and HCAL responses were first summed into towers using the transverse segmentation given above. Then the tower with the largest ET was used as the initial seed of a jet, and energies within a cone of radius R = p 2 + 2 around the seed tower were added to the jet. The jet axis was recalculated after the addition of each new tower. Once all the towers within the cone radius were merged, those towers were removed from the list to be clustered, and the procedure was repeated until all the towers had been associated with a jet. We required at least two jets with ET > 20 GeV and j j < 2:0. The dijet mass was then defined as the invariant mass of the leading two jets satisfying these cuts. The jets were assumed to be massless. 2.1 Results Without Gluon Radiation The simplest system for study is one where initial and final state radiation are turned off in PYTHIA, and the Z decays cleanly to two jets. For additional simplicity, we also turned off multiple-parton scattering and Z ! bb decays (also Z 0 ! tt) in these events, leaving only decays to light quarks. The reconstructed mass of the two leading jets in the calorimeter then gives a direct measure of calorimeter performance. The jet reconstruction was first verified by comparing the direction of the found jets with those of the partons from Z decay. Figure 1 shows the separation R = p 2 + 2 between jet and parton for Z events with R = 0:7 cone jets. R is typically < 0:1 and almost always < 0:2, so the reconstruction appears to be finding the correct jets. For these events, the reconstructed jet-jet mass divided by the true Z mass is shown in Fig. 2. The mean is 0.87 and the resolution is about 11%. The reconstructed jet ET divided by the corresponding parton ET is shown in Fig. 3, and also has a mean about 0.87 (indicating that the shift in reconstructed mass is due to the jet energy scale). For the 1 TeV Z 0 events, the corresponding plots are shown in Figs. 4 and 5. For these high-ET jets the mean reconstructed mass is equal to the input mass, and the resolution is about 7%. 2.2 Effects of Gluon Radiation In reality of course it is not possible to “turn off” gluon radiation and this adds unavoidable broadening to the reconstructed jet-jet mass. The study was therefore repeated with initial and final state radiation turned “on” in PYTHIA. Z ! bb decays and multiple parton scattering were also included in this sample. For these events with gluon radiation, the reconstructed jet-jet mass divided by the true Z mass is shown in Fig. 6. The mean is 0.82 and the resolution is 19%. For the 1 TeV Z 0 events, the corresponding plot is shown in Fig. 7. For these higher-ET jets the mean reconstructed mass is larger, 0.90 of the input mass, and the mass resolution is about 14%. The mass resolution for both Z and Z 0 is substantially increased by the addition of gluon radiation. In section 3 below we will use PYTHIA to show that the magnitude of the increase agrees with expectations.