FSQ-12-013: Answers to the ARC questions:

Livio's and Rick's questions (1st iteration)

Q: Why MC/DATA agreement is more fluctuating in pt>50 than pT>20 Jets at large eta for both MC (fig. 1 and 2) ?

A: This behaviour is not explained by statistical fluctuation but it is basically due to the different spectra predicteed by Herwig++ and Pythia6 for the hard jet pairs. In particular, Herwig++ predicts a flatter distribution, while Pythia6 exhibits a shape, more peaked in the central region. This can be already observed in the corrected distributions (figure 7(a), 7(b) in the PAS), where the predictions obtained with PYTHIA6 and HERWIG++ at the hadron level are compared to the data: they show a bigger discrepancy (~20%) around |eta|>2, present also at the detector level. These discrepancies are not present for the third and the forth jet. Further studies have been performed to understand the origin of this different shape: in the following, a study with Powheg interfaced with Pythia6 is shown; the different ingredients of the parton evolution have been switched on in turn to see the single effects.

The points we would like to emphasize is that the sample with only FSR on and the one with only ISR on have rather different shapes and that the main contribution to the 4-jet final state comes from the ISR. We suspect that the effect of the shape difference between the Herwig++ and Pythia6 samples might be due to the details of the generated parton shower in each of the generators, in particular of the angular ordering used in Pythia6 and Herwig++. Unfortunately, Pythia6 does not allow to switch off the angular ordering for the ISR and we can not check in this way. A different approach is used along the same line. In particular, different samples for Pythia6 and Herwig++ have been compared at the hadron level with only ISR or FSR on. The distributions for these samples are shown along with the shapes of the inclusive samples where the underlying event is completely switched on. The plots in the following represent the pseudorapidity shapes of the leading and the third jet.

We observe that for the leading jet, the discrepancy at the edge of the distributions is pronounced (PYTHIA6 Inclusive and HERWIG Inclusive) while for the third jet, there is basically no difference along the phase space. For the leading, we observe the fact that, while for Herwig++ the distributions for initial and final state radiation are very similar in shape, for Pythia6, the ISR contributes mainly in the central region, while the FSR has a more flat distribution like the ones observed for Herwig++. Since the ISR has the bigger contribution in the 4-jet final state, the final shape mainly reflects the shape of the ISR sample and the difference seen at the detector level seen between Herwig++ and Pythia6 can be explained by the different parton shower used in the generators.

Link to all the plots: http://desy.de/~gunnep/ETASTUDY/ , http://desy.de/~gunnep/POWHEGSTUDY/

Q: L186-188. Peculiar flat low-pT spectrum. Can you justify ? different production with lower thresholds ?

A: A different sample at the hadron level has been generated with Pythia6 when removing the asymmetric threshold for the jet pairs: a sample that requests a pT threshold of 20 GeV for all the jets has been compared with the nominal selection. The shape distributions for the leading, subleading, 3rd and 4th jet are shown in the following.

It can be observed that the distributions between the two samples are different: by setting a threshold of 20 GeV, both the distributions are rapidly decreasing, while for the nominal sample, the shape for the 3rd jet decreases of less than one order of magnitude over a wide pT range, while the 4th jet of about two orders of magnitude. We suspect that the fact that the third jet has a more constant behaviour in the low pt region is due to the DGLAP evolution of the partons produced in the hard scattering. The emission of a third jet at a scale close to the hardest one is favoured, while the emission of a forth jet is highly suppressed. This behaviour is typical of a scenario where asymmetric thresholds for the two pairs are set.

Link to all the plots: http://desy.de/~gunnep/PTSTUDY1/

Q: What's the merging limit of jets for collinear parton emission ? (L201)

A: The jets are well separated between them and this fact is assured by the cluster algorithm that is used for the jet reconstruction (anti-kT 0.5). The distance in the eta-phi space between the jets is always above 0.45 due to this. Even in a collinear emission via gluon splitting, the jets are basically well separated and reconstructed.

Q: Is it possible to extract and with the same sensitivity f_DPS from normalized distribution ?

A: In principle yes, but since the uncertainty in the higher for the cross section distributions, it would be preferred to extract the fraction from the shapes. The sensitivity of the correlation observables and a preliminary study about the content of DPS already implemented in the generators is provided in the Analysis Note.

Q: jet definition, eta cut at 2.5 and tracker acceptance at 2.4 (used for correction). Any bias if a jet axis is near the pseudorapidity limit ?

A: The bias introduced by the selected eta region is negligible. Here shown the plots for a selection where the eta region was restricted to |eta|<1.9. No strange behaviours are observed and the agreement between data and Montecarlo is very good, as it was in the nominal selection.

Link to the plots: http://desy.de/~gunnep/EtaRangeStudy/

Q: L128 can you also quantify a rejection ?

A: In the references, rejection values are found to be around 99.98% for single jets using the tight selection that is applied in our analysis. The references are https://twiki.cern.ch/twiki/bin/view/CMS/JetID, https://indico.cern.ch/getFile.py/access?contribId=0&resId=0&materialId=slides&confId=89919 and http://cms.cern.ch/iCMS/jsp/openfile.jsp?tp=drafts&files=AN_2010_003_v3.pdf

Q: Figures: can you use the same pT scale for 3rd and 4th jets ?

A: Done.

Q: L148: what you mean with "threshold effects" ?

A: They are migration effects from outside the phase space into or out of the selection. If the jet pT resolution can determine the fact that either at the hadron level or at the detector level, one jet is above (below) threshold in one case and below (above) threshold in the other.

Q: L179 and 180: JETMET and JET sample are not defined

A: they are defined in the table where all the datasets are listed.

Q: L215-217: is this visible in your distribution ?

A: is it visible from the fact that the shape of the pT of the third and forth jet is higher at high pT for the sample without MPI with respect to the one with MPI. A comment has been added in the PAS.

Suvadeep's questions (1st iteration)

Q: Change tune for POWHEG, since Z2 overestimates the data

A: P11 has been used now and it shows a better agreement for POWHEG when the MPI contribution is included.

Link to the plots: http://desy.de/~gunnep/POWHEGTUNES/

Q: Peculiar feature of the third jet

A: see previous list (with some plots for the explanation)

Q: Why do the MPI contribute so big? (by looking at the leading jet pt)

A: The MPI can have a double effect: it may add additional jets and more events might be selected or it may contribute with a "pedestal" energy for the jets produced in the hard scattering; this might determine the fact that jets that w/o MPI were below threshold are then above threshold with this additional momentum. Anyway, the excess in the leading jet pt does not mean that the leading jet comes from the MPI but that one event that was not selected w/o MPI has passed the selection cuts with the introduction of the MPI.

Q: More details in the discussion of the results.

A: Done.

Q: More specific in the discussion of the plots.

A: Done.

Q: Details of the unfolding.

A: Analysis note section 4.9.2 (Purity, Stability, Background and Acceptance)

Q: Need of the systematic uncertainty due to NP correction

A: There is no need of applying NP corrections since we are comparing data and MC at the hadron level. We are not going to the parton level. In that case, we would have needed a further correction with a consequent systematic uncertainty

Q: Set of other generators.

A: Madgraph and Sherpa have also been studied.

Link to the plots: http://desy.de/~gunnep/MADGRAPH/ , http://desy.de/~gunnep/SHERPA/

Suvadeep's questions (2nd iteration)

Q: - In Fig. 8 ratio plots- (a) and (b) - why Powheg+Pythia6 no MPI line fluctuates up for \Delta \Phi between 1 and 2.0?

A: This is due to the absence of the MPI itself..the contribution of the MPI is expected to be at high DeltaPhi and a missing component in that region determines a change in the shape that is reflected along the whole phase space. This behaviour is basically present in all the correlation variables where Powheg+pythia6 with MPI describes better the shapes, with respect to the sample without MPI.

Q: - In the description between lines 211 and 223 - it would be great if the figures are referred and text points to the features of the plots being mentioned.

A: Done.