Pile-up composition in data. The 'charged pile-up' is PF charged hadrons that are not associated to a primary vertex, and therefore are not removed by the charged hadron subtraction (CHS). Instead, they are handled by pile-up corrections together with neutral contributions.

This plot is produced with the 'Random Cone' method by taking pile-up at NPV=20 and subtracting pile-up at NPV=10 to estimate the average pile-up per vertex. The corrections for data show linear behavior as a function of NPV, while MC has slight quadratic behavior in the end cap regions around fabs(eta)=2.5-3.0.

Jet energy corrections shown separately for different jet flavors. Gluons (g) have a high color charge compared to quarks, which causes them to fragment to more particles than quarks and results in larger JEC. The high mass of heavy quarks (c, b) also results in fragmentation to more particles than for light quarks (uds), in addition to many heavy quarks being produced in gluon-splitting.

The neutrinos produced in semi-leptonic decays of heavy quarks are not included in the generator-level particle jets when computing the corrections.

Eta-dependent corrections determined from dijet data. The detector modeling is accure, and the eta-dependent residuals for all jet types are below 2.5% at fabs(eta)<2.4. The residual corrections in the end caps outside tracker coverage are slightly larger, and HF modeling requires a 5 to 10% correction.

The main recommended algorithm is PF(chs), but CALO and JPT are also supported.

Data/MC ratio for the absolute scale after combining the photon+jet sample with two Z+jet decay modes. The residual correction required for absolute scale is of the order of 1.5% for all subchannels, and shows no significant pT dependence. Possible pT dependence is covered by extrapolation systematics based on a comparison of Pythia and Herwig fragmentation models and on the uncertainty in single particle response propagated to jets.