Here I show the effect of removing the jet eta cut when selecting dijet candidates in the unbiased simu jet sample ...

Recently, a question came up about how to select the jets which make up a dijet. The theory code from deFlorian finds three jets (three because its NLO code) at any eta. Then the jets are orderd by their pts. After the two highest pt jets are found, the eta cuts and any other cuts are applied.

I have to select jets in data and at 3 different simulation levels:

Data: I find all jets in an eta range between -0.8 and 1.8 and in a detector eta range between -0.7 and 1.7. I then select the two highest pt jets. I then apply all other phase space cuts.

Detector level simulation: Same as for data.

Matched Simulation: The particle/parton level dijets which match to the detector level dijets are found simply by geometric closesness, no other conditions are put on these jets.

Unbiased Particle/Parton: I find all jets in the eta range -0.8 to 1.8. I then find the two highest pt jets. Then opening angle, asymmetric pt, and mid rapidity eta cuts are applied.

The initial eta cut (-0.8 to 1.8) is needed for data and detector level dijet selection because it selects the instrumented part of the detector. The question was whether or not the initial eta cut should be applied when finding dijets in the unbiased particle/parton sample. It was decided that to best match what the simulation did, the initial eta cut should be removed when finding dijets at the unbiased particle/parton level. So now when looking at the unbiased particle/parton level, I select the two highest pt jets from any eta.

The plots below show the differences in cross section, theory+UEH, and systematic error between the original calculations, and the new calculations without the initial eta cut.

Figure 1: This figure shows the ratio of the cross section calculated using the unbiased simu sample without the initial eta cut over the cross section calculated using the unbiased simu sample with the initial eta cut.

Figure 2: Same ratio as figure 1 but now for the UEH corrected theory. Note, the theory values from deFlorian's code are the same for both samples, so the difference comes from the effect that the initial eta cut has on the UEH correction.

Figure 3: This figure shows the combined systematic band (tracking inefficiency, track pt uncertainty, tower energy scale uncertainty, and time variation systematic) calculated with the initial eta cut () and without the initial eta cut.

Figure 4: This figure shows the two cross section money plots. The top figure is the old plot which was generated using the unbiased simu sample including the initial eta cut and the bottom plot used the unbiased simu sample whch did not have the initial eta cut.