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

Comments from Suvadeep and Livio concerning the text, have been already implemented in the new draft. Here we would like to discuss some further points.

Answers to the CWR https://twiki.cern.ch/twiki/bin/view/CMSPublic/CWRPaperFSQ12013

Livio's questions (1st iteration)

Q: This Z2' is a new tune proposal or an ad-hoc tune just for your analysis ? If Z2' tune for powheg is a general suggestion, some additional characterization should be performed...i.e.: Did you cross-checked with a process where the contribution of an additional hard emission is minimized ? (i.e. a 2-jet process). It is correct that lowering down the energy scale of the emission you will get a softer spectra with higher multiplicity ? It could be interesting to check independently the multiplicity of jets (i.e. studying the event shape and jet multiplicity in 2 high-pT jet).

A: The tune Z2' is mainly a general suggestion, since it is based on the observation of a good agreement between predictions and inclusive jet cross sections and uncerlying event observables. The concept is mainly based on the fact that POWHEG, that implements already a hard emission in the matrix element, does not need further emissions coming from the parton shower. This is done only by lowering the scales of the initial and final state radiation through the switches PARP(67) and PARP(71), respectively. I guess with these new settings, the multiplicity should decrease since the shower is somehow suppressed (but a part of it is already in the matrix element, as said). For this analysis, what is interesting is the decrease of the multiplicity at high pT; the difference between the standard tune and the modified tune is around 20% in the total cross section.

Attached are some plots that show the description of the some observables for default and modified tunes.

Q: L115-117 I don't understand this point: you used sherpa+p6 configuration ? and you used a different tune from Z2 ? the model is the same (there is only 1 MPI model in P6)

A: In the paper, SHERPA standalone is used: it generates the matrix element and has its own parton shower. The MPI model is based on the same that PYTHIA uses (this is the meaning of the sentence in the paper). What changes, are some parameters of the MPI model (so, basically what is different is the tune implemented in SHERPA).

Q: why different parameter for MPI in case of sherpa ? which parameters ?

A: The different parameters used in SHERPA have been set after some interactions with the CMS SHERPA experts and the set of used parameters has been addressed as the recommended one: this is based on the following parameters:

PROFILE_PARAMETERS 0.884 0.828 (PARP(83)=0.356, PARP(84)=0.651 in Pythia6 Z2 tune) RESCALE_EXPONENT 0.244 (PARP(90) 0.275 in Pythia6 Z2 tune) SCALE_MIN 2.65 (PARP(82) 1.832 in Pythia6 Z2 tune)

Basically, what it is mainly different, is the overlap between the two interacting protons through the profile parameters. This increases the contribution of the MPI.

Q: PDF uncertainties

A: Comparison with another CMS measurement: The PDF uncertainty for the dijet mass measurement (http://arxiv.org/pdf/1104.1693v1.pdf) is relatively low compared to the jet energy scale: "The PDF variation introduces a 5% (30%) uncertainty on the theoretical prediction at a dijet mass of 0.2 TeV (3 TeV), while the variation of alphaS (M_Z) by 0.002 introduces an additional 2–4% uncertainty." Regarding the jet energy scale, it "introduces a 15% (60%) uncertainty on the cross section at M_JJ = 0.2 TeV (3 TeV)". How does it need to be done for this analysis? If one should use MADGRAPH for the PDF uncertainties, this could take a while, since complete lhe files need to be generated.

Q: bin at 1 (Delta_pt rel soft) means that in a high fraction of events the soft jets within the same soft couple are "decorreleted" in pT. On the other hand in figure (a) there is an high fraction of event which are back-to-back (Deltaphi~3)...so pT correlated ? am I wrong ?

A: You are right. A high value of Delta_pt rel soft means a decorrelation between the jets belonging to the soft pair. The fact is that it is difficult to connect this high decorrelation with the values of DeltaPhi because it is dependent on the pt of the jets themselves. If a high pt jet pair is collinear, they can be balanced in pT without being exactly back-to-back in DeltaPhi, while a low pT jet with a high Delta_pt rel, it is more probable to be also back-to-back in the transverse plane. To confirm this explanation, the distribution of DeltaPhi for the soft jets when Delta_pt rel soft is greater than 0.9 (last bin) is shown below. It can be seen that a pt-balanced pair does not correspond exactly to a peaked distribution in deltaPhi but it shows a broad distribution around 0.5-0.8.

Livio's comments (2nd iteration)

1) L149-151 "Small differences are observed between data and simulation only at very high η values for the leading and subleading jets; however large scale uncertainties are associated with high-pT jets in the forward region" I think you need to better clarify, "small" difference are not so small in the last eta bin (80%)...In addition to the large scale uncertainty can we also assume a contribution for a poor MC description of the forward activity or this is in contrast with the data/MC agreement for 3rd and 4th jet ? I think you should mention both contribution. i.e. form CMS-FWD-11-002 (and also from your recent additional studies) you have large uncertainty for high-pT but also large disagreement with MC for pT>40 in the forward regione (3.2<eta<4.7).

> I agree with Suvadeep that we should offer few possible explanations, like the JES and the difference between generators. A further discussion is going on in the forward group together with the JETMET group but nothing is conclusive and they are trying to play with the corrections.

https://indico.cern.ch/getFile.py/access?contribId=2&resId=0&materialId=slides&confId=232847

When something is more defined, I will also try to test whether this new correction might change the results but I really doubt. I guess it can be done during the CWR.

2) Is there any additional follow up from the discussion with Steve and Suvadeep ?

> From my point of view, I have also seen that basically the discrepancy has little to do with the jet multiplicity, so whether inclusive, dijet or four jet exclusive or inclusive selections are set.

3) did you solve (and include) the "best way" to indicate shape measurement ? I think this is the only open question with Vivian

I think Vivian should comment on that (I did not receive a final answer) but it might have been misleading that in the paper version she looked at, the Y axis label were wrong (no normalization written). Sorry for that..Now they are fixed (version3).

Suvadeep's comments (2nd iteration)

About the inclusion of L249-252 : I welcome the inclusion and I have a few suggestions about text part of it. In L250: Either you write ".. is also shown for the normalized cross section in Figure 4." - or - you write ".. is also shown." I dont like the ":" in L251 - better to start a fresh line as "While for \delta \phi^{soft} .. "

> Sentence changed.

Vivian's comments (Language editor)

More problematic in my mind is that the tense of the paper is not consistent all the way through. In general, the paper should be written in present tense except for items like "The data were collected by the CMS detector...". The description of the analysis should mostly be in the present tense. In the detailed comments below, I don't point out the tense errors, since there were so many of them, but you should go through the paper and fix the tenses.

> The past tense has been left only for the description of the data collection and

Also, please use the tex macros for Herwig++, Pythia6, etc.

> DONE throughtout the paper

Please check the paper also for redundancy -- I think JES / JER is talked about in more than one place, for example.

> JES and JER are quoted only in the table of the systematic uncertainties

Below are some more detailed comments:

Abstract: 4-jet -> four-jet. The use of the number and the full word is not consistent in the paper. The first sentence is also a little awkward. Perhaps something more like: "Measurements are ... transverse momentum pT and pseudorapidity eta, as well as the correlations in ... jets."

DONE

line 4: "... at the LHC, and corresponds to an integrated ..."

DONE

Section 1: l 8 4 jets -> four jets

DONE

l 10 "... the strong coupling constant, alpha_s ."

DONE

ll 24-25 "... 7 TeV using the full 36 pb-1 data sample collected with the CMS detector at the LHC during 2010."

DONE

l 31 "... as a function of the transverse momentum and pseudorapidity of the four jets ..."

DONE

l 33 "... and soft pair of jets as follows: "

DONE

add a semicolon after equation 1 add a comma after equation 2

DONE

l 34 "... vectorial momenta of the soft jets; "

DONE

3rd bullet: "... between the two dijet pairs delta S defined as "

DONE

add a comma after equation 3

DONE

l 35 you might want to say "j^{soft1}" (etc) instead of just soft1.

CHANGED

l 37-38 I don't understand why normalizing distributions give smaller systematic uncertainties...

> They give a smaller systematic uncertainties because for the normalized distributions, it is related only to the change of shape for the correlation observables not to the absolute normalization; for the cross section, the uncertainties are big because more events pass the selection due to the different momentum (pt +/- pt uncertainty) of the jets and the total number of events is quite different but, since the normalized cross section are divided by the number of selected events, this effect is washed out and only how these events change the shape of the observables is contributing. And this contribution is much smaller than the one due to the increase/decrease of the total number of events.

l 41 "... in the future..."

DONE

l 44 " as follows: In section ..."

DONE

l 78 at -> during

DONE

l 79 "... uses the full 2010 data sample."

CHANGED

l 80-81 " ... is determined from simulation as a function of ... eta."

CHANGED

ll 89-90 "The free parameters describing MPI are..."

DONE

l 93-94 You say that the Z2* tune is the same as the Z1 tune except for the choice of PDFs. I think you mean that the Z2* tune is the same as the Z2 tune except for the choice of PDFs. Also you never again mention Z2* but talk a lot about Z2...

> Yes, you are right. It was a refuse from older versions. Also the label of the plot has been updated.

ll 96-97 "... HERWIG++ tuned to LHC data ..." (and use the Herwig++ tex macro)

DONE

ll 102 - 105 I can't understand the sentence starting with "Even if a good representation ..."

> We want to say that already POWHEG interfaced with PYTHIA6 tune Z2 describes the data quite well, but if we apply the modified tune (Z2'), the description improves with respect to the default tune.

ll 106-107 "... and PARP(71) to one from the default value of four. These parameters regulate the upper scale of the initial and final state radiation respectively."

CHANGED

l 108 "... and is labelled with Z2' in this paper." Also you never again use Z2' in the text of the paper.

> You are right. The expression "modified Z2 tune" has been replaced simply with Z2' "tune"

l 110 "The parton shower for MADGRAPH is provided by ..."

CHANGED

l 14 4 -> four "Predictions from the SHERPA ..."

DONE

l 115 "... obtained; This event generator produces ..."

CHANGED

ll 116-117 "MPI is implemented in the simulation [27], based on the model used in the PYTHIA6 underlying event, but different..."

CHANGED

ll 118-119 I think you want a new paragraph here.

CHANGED

l 126 "Every data sample..." you only have one data sample, so what does this mean?

> It is a refuse of the PAS where the different data subsamples from 2010 run were widely described. Here it makes no sense and the sentence has been removed.

l 142 "... should be less than..." I think this is a requirement of the analysis? so it should be "... is required to be less than..."

Yes, CHANGED

l 149 I would say "leading jets" instead of "first jets" also "... jets of the hard and soft pairs..."

CHANGED

ll 152-153 "...is observed for |eta| > 2 ..."

CHANGED

l 154 "... shower models implemented in ..."

CHANGED

l 160 with -> using

DONE

l 161 I don't understand this. Are you unfolding Herwig++ using phythia 6? Or are you unfolding data using Herwig++ ?

> The sentence is refering to Herwig++ unfolded with the response matrix obtained with pythia6 as a closure test. The sentence has been rephrased and hopefully it is more clear.

l 171 You should be specific if you are talking about Pythia6 or Pythia8 and Herwig or Herwig++. as a side note, although we commonly use the difference between pythia and herwig as a systematic uncertainty, there is a concern here, as e.g. POWHEG with NLO give a larger difference in some distributions.

> The fact that we are quoting as model uncertainty the difference between Pythia6 and Herwig++ is due to the fact that we unfolded the data with these two Mcs and took the difference of the unfolding results as the uncertainty. NLO simulations at the detector level are not available and we can not unfold with them.

l 177 "... and is the cominant uncertainty." Why is the JES uncertainty smaller with "normalized" data (I don't even know what normalized data is). I have the same question about JER.

> See comments of line 37-38

ll 187-188 "... on the luminosity of the 2010 data sample adds an additional uncertainty of 4% to the cross section."

CHANGED

l 189 effects -> uncertainties

DONE

l 192 "... are presented in Fig. 2."

CHANGED

l 196 ".. sections fall rapidly with ..."

CHANGED

l 199 "... as a function of eta (Fig. 2) is very similar..."

CHANGED

l 200 "In particular, hard jets are..."

CHANGED

l 201 "... after that the cross section for hard jets rapidly decreases by about... edges. Soft jets..."

CHANGED

l 203 ".. between eta ~ 0 and the forward region.."

CHANGED

l 204 "Measured cross sections ..."

CHANGED

ll 206-207 "... with respect to measurements: "

CHANGED

l 207 "modified Z2 tune is in good agreement... " Also, here are you referring to the Z2' tune?

Yes, CHANGED

l 209 "... excess and lack of events..." What does this mean?

> It is refering to the predictions of POWHEG and MADGRAPH that respectively overshoot (they have an excess of selected events) and undershoot (they have a lack of selected events)

l 215 "... PYTHIA 8 tune 4C..."

CHANGED

l 216 ".... PYTHIA 6 tune Z2 including MPI..." ( leave out the "and" and also I think you are referring to the Z2' tune?)

CHANGED

l 218 "... lower value and SHERPA, which uses different..."

CHANGED

l 221 "... coming from MPI, while..."

CHANGED

l 223 "... between hard and soft ..."

CHANGED

l 227 de-correlated -> decorrelated

DONE

l 228 "... by theoretical..."

CHANGED

l 234 "... exhibits another maximum around 0.2." Is there a conclusion or something to draw from this?

> The point we want to underline is that we have a maximum at 1 that refers to decorrelated jets, but also a maximum at 0.2 corresponding to a back-to-back-like configuration.

l 238 "... are reasonably well ... by all the predictions we considered. Overall SHERPA gives the best..."

CHANGED

l 240 "... achieves good agreement... less than 30% over the entire phase space..."

CHANGED

l 245-246 ".. measurements by a factor of ..."

CHANGED

ll 250 - 252 Sentence starting with "Interesting is to..." I don't understand this sentence.

> We want to underline the fact that the shapes between different matrix elements do not change much and even at LO, the description is quite good.

l 254 di-jet -> dijet

DONE

l 259 "Measurements of exclusive four-jet observables have been performed..."

CHANGED

l 262 "... has been measured to be sigma(four jet)..."

CHANGED

ll 265 - 266 "... predictions essentially agree with the ..."

CHANGED

l 269 "... is predicted by POWHEG..."

CHANGED

l 273 leave out the comma

CHANGED