-- AmnonHarel - 17-Oct-2010

Question

During and after final reading Nikos showed preliminary, internal plots from the inclusive jet cross-section analysis, that indicated that the data/theory curve as a function of jet pT has a downward slope, independent of jet eta (at least out to |eta| of 1.5). Due to the different kinematics, such an effect translates to a lower R_eta. He asked whether we ought to account for that.

The plots

Abs(eta) range:	0 - 0.5	0.5 - 1.0	1.0-1.5	1.5-2.0
data/theory plot:	pflow_Rap0_LUM.pdf	pflow_Rap1_LUM.pdf	pflow_Rap2_LUM.pdf	pflow_Rap3_LUM.pdf

Answer

In the paper we state exactly the assumption made to derive the limits on contact interactions. Should future research show that predictions of R_eta from QCD @ NLO with non perturbative corrections (CNLO for corrected NLO) are way off, it will be natural, and completely beyond reproach for CMS to say "we now need to rederive the limits - here are the new ones". Furthermore, between Fig 1a, Fig 2, and Table 1 we provide the reader with all the inputs from CMS data needed to do his own statistical analysis, and enough information on the systematics to approximate them well. So the community won't even have to wait for CMS' answer - the facts will all be out there.

The above assumes any discrepancy is on the theory side. We make this assumption since our cross checks of multijet pT balance in events one leading jet is inner and the other outer showed a good balance, as mentioned in the final reading. This severely limits the size of the relevant possible detector effects.

So the only question is: do we currently know the CNLO predictions are wrong?

Taking the plots at face value

As Greg pointed out - even if we take these plots at face values, their uncertainties are large enough that they are consistent with the theory.

First, we note that as far as we know, these plots show the experimental uncertainty centered around the ansatz, which we find misleading.

As demonstrated by dividing the 1st and 3rd plots, the slope is basically the same at the relevant |eta|s, and so we can focus on the plot for the inner jets (*). The limit setting around 4TeV is dominated by the high Mjj events. Therefore we should focus on the difference between 550 and 800GeV. The ansatz is a downward slope of 1 in this region, which is not supported by the data. Keeping in mind the experimental uncertainties, we see no evidence of a slope in this range.

(*) - as the JES is calibrated across etas to match the absolute calibration in the central calorimeter, the same 1sigma JES variation can apply to all 4 plots.

We also that the kinematic effect that connects these plots to the R_eta measurement works both ways - some of the downward curve is accounted for by the low R_eta at high Mjj. As such, this is already accounted for in our analysis by the CLs method's built-in pessimism to lucky fluctuations.

External reasons to mistrust these conclusions

On the other hand, there are reasons beyond those evident in the paper to believe the current analysis is stable w.r.t. such effects, if they exist

• our own preliminary, internal plots using 14pb-1 of data, no longer show a deficit in R_eta. It was mostly a statistical fluctuation.
• our cross checks of possible detector effects, mentioned above.

Things to ask the analyzers that produced these plots

• what is the theory that underlies this curve?
  • why are the theory uncertainties only around 7%? From Klaus' latest numbers for our own spectra, their the uncertainty is twice as large.
• how much of this slope can be attributed to a pT-dependent variation in JES within its uncertainties?
  • it seems that would put these results in line with other measurements that find lower yields across all kinematics
• we still haven't heard a definitive answer on the CMSSW used for the RECO. CMSSW_3_8 is known to remove true ECal energy from hard jets.
• what do they about the plots validity to draw such conclusions?