Dear all,

I removed the trivial textual comments that were addressed. The remaining list below (I left the questions that were answered in, such that one can check whether the answer is sufficient or needs further changes to the paper, as well as the textual suggestions he did not implement, in case anyone feels strongly about those - that leaves quite a long list still, unfortunately). I will collect reactions on Friday 13/02 then.

Thanks,

Pieter

Physics: - I miss a section explaining bremsstrahlung correction. It is not clear if the numbers you give in Section 3 are before or after correction.

*** This is done by the efficiency \epsilon_{Kin.}, as explained in lines 333-340. I think perhaps the confusion is arising from the use of the word "uncorrected" in l.336. I will address this.

- (385-393) since \epsilon_kin includes bremsstrahlung, it must also be sensitive to the simulation of detector material. How is this accounted for?

*** This is exactly what is being tested by allowing the parameter \alpha to vary, and testing simulation against data.

- (443-444) Shouldn't the integrals be identical? It should not matter whether efficiencies vary more rapidly. Does it?

*** Not quite because the phi* distribution is truncated at phi*=4. But also, since the efficiencies very with yZ, and the simulation doesn't perfectly model the yZ distribution, then the efficiencies will not necessarily be correct in bins of phi*. Of course the simulation is consequently reweighted to take account of this, but it does in principle involve extra assumptions and uncertainties.

- (Tab.2) I am completely confused: doesn't FSR lead to migration between bins in rapidity? Why are all numbers smaller than 1?

*** No, FSR (like bremsstrahlung) is largely collinear with the electrons. So the effect is largely to decrease the pT of the leptons, with little effect on their direction. Rapidity is largely dependent on the directions.

- You use pseudorapidity and rapidity interchangeably, which is correct at large momenta, but give limits 2<eta<4.5 and the say the efficiency is zero above y_Z>4.25. What do you do with Z between 4.25 and 4.5?

*** Not interchangeably, I hope. We use pseudorapidity for the electrons (find because their mass is negligible) but Rapidity for the Z. The Z has a large mass, so its rapidity and pseudorapidity are VERY different.

*** Nothing is done for y>4.25. But the NNLO QCD calculations predict that this bin contains a negligible x-section.

General editorial comments: - You could add many plots in additional material that would be useful for talks. A mass distribution for instance, but also some more you have shown in the approval talk.

*** I could.

- Cite the relevant Atlas and CMS papers.

*** Which ones are relevant?

- At several places (abstract, L241, L481) you write that you know the luminosity approximately. This is strange as we know the luminosity to an amazing accuracy of 1%. Please remove the word “approximately”. In addition, please quote the lumi with the same number of digits everywhere (preferably 2.0 instead of just 2).

*** Anything that has an uncertainty is approximate. If we don't include this word we'd need to quote the number to 1% accuracy along with its error, which is possible, but maybe not necessary.

Textual:

214: "The effective *geometric acceptance* of the LHCb detector". I would not put [1] here.

*** Nor would I, but this was a request from a reviewer.

220: Z->mumu is not "straightforward". Rephrase.

*** No, but it is the "most straightforward" as stated - any other decay mode is more challenging.

230: Add comma after “Therefore”

*** No, not necessary in English.

237: either remove "respectively" - there is no ambiguity possible - or put commas around it

*** I was asked to add "respectively". Commas are not correct English usage.

Footnote 1: shouldn't the second Z be Z^0?

*** I don't greatly care. I just use the LHCb symbol, which gives Z.

254-5: Remove “covering the pseudorapidity range 2 < eta < 5”, since you have already discussed the effective range previously. This sentence only confuses the reader (you do no want to explain why it is defined differently here)

*** This is standard EB text. Here we are describing the spectrometer in general, while earlier we described the effective acceptance for the present analysis. This seems clear to me.

260 : “provides a measurement of THE momentum"

*** Again, I wouldn't dare to change approved EB text.

272 : “with a segment of A charged particle track”

*** OK.

273: remove “charged-particle”. This is implied by track.

*** I think this was included by a previous request.

278 : “… to evaluate SOME efficiencies.” -> sloppy! unclear if all relevant ones have been simulated, or all necessary ones, etc. “… to evaluate relevant efficiencies that couldn’t be determined by other means."

*** "some" seems perfectly clear and accurate to me. But I can just say "are used in the analysis."

286: "refine the sample of candidates for analysis" is jargon. "to reduce the same size used in the analysis" ?

*** Sorry - I don't understand your suggestion.

287: "should have induced a *single-electron trigger*" sounds like jargon. Maybe change it into "trigger decision"?

*** Surely "trigger decision" is just as much jargon?

292: is (propose to write ) measured before or after bremsstrahlung correction? Maybe adding "curvature significance" can make this clearer

*** after

296-8: The clause “where E_ECAL, … respectively” can be removed without any loss of information. The abbreviations were already introduced.

*** I was explicitly asked to include this. I agree with you.

Eq.1: Add comma at the end of the equation.

*** Maybe

Eq.2: No need to have periods at the end of Kin and Trig

*** In English usage, when you abbreviate a word you put a full stop (not a period!) at the end unless the last lketter of the word is part of the abbreviation (this Prof. but Dr)

p9: you can remove some of the forward references to Sect. 4 (the one on line 325 is sufficinet)

*** Only trying to help the reader...

309: events -> candidates

*** Surely "candidates" is LHCb jargon. Events is likely to be more widely understood.

312: “The choice of binning is informed ….” -> “The choice of binning is DETERMINED by ….”. Does this mean that the phi* and eta resolutions are small wrt the bin widths (and hence migration due to e.g. bremsstrahlung negligible)?

*** Yes, migration is negligible. "Determined" is too strong - the statistics are one consideration, but so is the desire to have tidy bin boundaries which aligh with other studies. "Informed" implies this (to me).

316: propose to remove the first sentence: " , (2) where eff_track is [...] quality requirements, eff_kin gives [...] and \eta, eff_pid is [...] as electrons, eff_GEC is [...].", then it is still very clear what each efficiency exactly means (including the "cumulative" determination)

*** I think the sentence should remain because reviewers were concerned about this point being emphasised.

327-8: Not clear what is meant with “electron and positron yield reconstructed tracks”. Not needed either. Suggest: “both tracks satisfy the selection requirements”.

*** The electron and positron may not have been reconstructed as tracks at all. The word "yield" tells you that they did.

The discussion of FSR comes a bit confusing at this point. Could it be introduced earlier, when the fiducial region is defined, or in the description of the simulated samples?

*** Maybe, but this is the point at which the correction for FSR is made, so it seemed quite natural.

345: "which varies significantly with y_Z". “…; this part of the efficiency should be modelled reliably.” -> I’d assume that holds for all parts. Remove entire sentence, or write what is really meant (e.g. needs to be modelled with a better than average precision, or something)

*** Certainly not true for the calorimeter energy cuts, since the calorimeter response is not at all well modelled. The acceptance effect only relies on the geometry being accurately modelled.

347: “the calorimeter energy requirements” -> the calorimeter acceptance”

*** No. This refers to the cuts in l.296.

349: Would it be possible to avoid the "GEC" acronym? It is rather specific jargon

*** It is carefully defined in l.274. What do you propose?

354: “apart from the contribution from the leptons”: why not write electrons? Suggest to be more clear: “… after correcting for the additional SPD hits of the electrons”.

*** We are talking about the difference between the muon and electron contributions, so "leptons" is correct.

360: maybe you can refer to the Z->mu+mu- paper for the treatment of the SPD multiplicity systematic. Why fit a gamma function?

*** Well, the Z->mumu paper isn't yet in circulation. And anyway, it's a different channel, so the systematics are not the same. A gamma function because it works. But alternatives are used too.

394: "acceptance is purely geometrical". Well, yes, that's the definition -> acceptance is modelled reliably in the simulation (that's a fact, not an assumption).

*** We don't know that. There must be some uncertainty in how accurately the detector geometry is known, and therefore the simulation isn't perfectly accurate. We assume it's accurate enough (and have tested it of course as mentioned in the ANA note.)

Fig.2: LHCb missing. A pull plot would help to judge the quality. Why "background-subtracted" in parentheses? That's quite crucial.

*** LHCb is not missing. A pull plot is possible.

397: Remove quotes around probe.

*** Defining a piece of jargon. Quotes indicate this.

398: what do you mean by the "energy recorded in each of the calorimeters in turn"? I guess it is once without the E_ECAL/p and once without the E_HCAL/p cut, but the formulation should be more explicit

*** Yes, that's what the text says. I can try to spell it out in gory detail.

413: “such backgrounds are intended to be …” -> such background are EXPECTED to be …”

*** I think "expected" is too strong - that's why we need to test it.

431: “statistical error” should be “systematic error”.

*** No, it's included in the statistical error, as stated.

437: remove "assumed to be"

*** It is an assumption

449: it seems like most (?) systematic errors are added linearly. From fig. 3 it seems that the results is systematics limited. Is the systematic error evaluation not too cautious?

*** Well yes, the largest systematics are assumed fully correlated. Maybe this is cautious, but in the absence of any better knowledge, what else to do?

454: “included some LHC data” sounds as if they purposely neglected available LHC data. Suggest to delete “some”.

*** I guess they included LHC data which were available at the time they starting doing their fits, and they selected the data for which they had sufficient information about covariance matrices etc. It's not for us to speculate on this. I think "some" is fair.

Fig.3: what do you mean with "integration errors"?

*** The errors arising from numerical integration

Table 2 and 3: error -> uncertainty (twice in both tables). Why are there no spaces around \pm?

*** Why should there be? Can be done, of course.

Pg.21: I’m somewhat surprised that the correlation matrix has only positive values. If there is some smearing effect going on, at least some negative values should appear here and there. Is there a single systematic error driving the correlation (e.g. luminosity)?

*** As said before, smearing isn't an issue (that's why we've used these observables). The systematic errors are positively correlated Luminosity is the largest.

Op woensdag 11 februari 2015 17:48:04 schreef Patrick Koppenburg:

Hi Pieter,

[ re-sending what I sent privately ]

I am not particularly happy with the replies. Can you try to compactify his replies into a list with only those where he disagrees? That would be helpful to start a reply.

The EB reading is on 16/2, so we need to react before that.

Thanks

Patrick

Hi all,

Wouter's list covers the most important point.

I would add that candidates is maybe jargon, but found in all our papers. Events are something different. There can be several Z candidates in the same event.

Cheers,

Patrick

On 12/02/15 10:52, Wouter Hulsbergen wrote:

Hi Pieter,

I completely agree with Patrick that some of David's answers are not entirely satisfying and that this asks for some sort of reply. Some of it will come back in the EB reading (like his use of the word 'some':-), so we should probably concentrate on a few obvious once and the physics. Here are a few suggestions:

• concerning the bremstrahlung efficiency: the shape of the pT

distribution can be changed by Z kinematics, by bremstrahliung and by other effects on the electron efficiency. it is not obvious that the combination of these effects can be described by a simple scaling of the pT. consequently, we don't understand why the variation of alpha as described in the draft (vary alpha as long as data-MC agreement in pT is acceptable) accounts for this.

• what is the fraction of events with phi*>4 in the simulation?

• the mass distribution may not look very convincing, but for one thing

it nicely illustrates the problem with brehmstrahlung

[if we want to have more plots, we need to be explicit. which one would we like?]

• for ATLAS/CMS references look at their web sites:

https://twiki.cern.ch/twiki/bin/view/CMSPublic/PhysicsResultsSMP#Vector_boso...

https://twiki.cern.ch/twiki/bin/view/AtlasPublic/StandardModelPublicResults#...

a small, non-exhaustive list for relevant results at 7TeV:

ATLAS: http://arxiv.org/abs/1109.5141
CMS: http://arxiv.org/abs/1107.4789,

http://arxiv.org/abs/1110.4973

• concerning 'intended' versus 'expected'. Propose to rewrite as:

"These backgrounds are partially accounted for by the subtraction of the same-sign Z → e±e± candidate events. To estimate to what extend this procedure indeed removes all the hadronic background, ..." or something along these lines

• the phrase "by extrapolation using a gamma function fit to the

distribution" should just be removed

• the way 'Therefore' is used here, requires a trailing comma in written

English. see for instance http://www.grammar-monster.com/lessons/commas_after_a_transitional_phrase.ht...

• in other papers we have used Z0 for the SM vector boson, so that

should also be done in footnote 1

• in LHCb papers we usually precede 'respectively' by a comma. it is

also in the aps style guide. in this case the word could really just be removed, though.

• 'errors arsing from numerical integration' is definitely better than

"integration errors"

• yes, there should be spaces around the \pm in the tables

I am probably missing a few more of the physics issues, so I hope others can look as well ...

Cheers, Wouter

On 11/02/15 18:13, Pieter David wrote:

...

Dear all,

I removed the trivial textual comments that were addressed. The remaining list below (I left the questions that were answered in, such that one can check whether the answer is sufficient or needs further changes to the paper, as well as the textual suggestions he did not implement, in case anyone feels strongly about those - that leaves quite a long list still, unfortunately). I will collect reactions on Friday 13/02 then.

Thanks,

Pieter

Physics:

• I miss a section explaining bremsstrahlung correction. It is not clear if

the numbers you give in Section 3 are before or after correction.

*** This is done by the efficiency \epsilon_{Kin.}, as explained in lines 333-340. I think perhaps the confusion is arising from the use of the word "uncorrected" in l.336. I will address this.

• (385-393) since \epsilon_kin includes bremsstrahlung, it must also be

sensitive to the simulation of detector material. How is this accounted for?

*** This is exactly what is being tested by allowing the parameter \alpha to vary, and testing simulation against data.

• (443-444) Shouldn't the integrals be identical? It should not matter whether

efficiencies vary more rapidly. Does it?

*** Not quite because the phi* distribution is truncated at phi*=4. But also, since the efficiencies very with yZ, and the simulation doesn't perfectly model the yZ distribution, then the efficiencies will not necessarily be correct in bins of phi*. Of course the simulation is consequently reweighted to take account of this, but it does in principle involve extra assumptions and uncertainties.

• (Tab.2) I am completely confused: doesn't FSR lead to migration between

bins in rapidity? Why are all numbers smaller than 1?

*** No, FSR (like bremsstrahlung) is largely collinear with the electrons. So the effect is largely to decrease the pT of the leptons, with little effect on their direction. Rapidity is largely dependent on the directions.

• You use pseudorapidity and rapidity interchangeably, which is correct at

large momenta, but give limits 2<eta<4.5 and the say the efficiency is zero above y_Z>4.25. What do you do with Z between 4.25 and 4.5?

*** Not interchangeably, I hope. We use pseudorapidity for the electrons (find because their mass is negligible) but Rapidity for the Z. The Z has a large mass, so its rapidity and pseudorapidity are VERY different.

*** Nothing is done for y>4.25. But the NNLO QCD calculations predict that this bin contains a negligible x-section.

General editorial comments:

• You could add many plots in additional material that would be useful for

talks. A mass distribution for instance, but also some more you have shown in the approval talk.

*** I could.

• Cite the relevant Atlas and CMS papers.

*** Which ones are relevant?

• At several places (abstract, L241, L481) you write that you know the

luminosity approximately. This is strange as we know the luminosity to an amazing accuracy of 1%. Please remove the word “approximately”. In addition, please quote the lumi with the same number of digits everywhere (preferably 2.0 instead of just 2).

*** Anything that has an uncertainty is approximate. If we don't include this word we'd need to quote the number to 1% accuracy along with its error, which is possible, but maybe not necessary.

Textual:

214: "The effective *geometric acceptance* of the LHCb detector". I would not put [1] here.

*** Nor would I, but this was a request from a reviewer.

220: Z->mumu is not "straightforward". Rephrase.

*** No, but it is the "most straightforward" as stated - any other decay mode is more challenging.

230: Add comma after “Therefore”

*** No, not necessary in English.

237: either remove "respectively" - there is no ambiguity possible - or put commas around it

*** I was asked to add "respectively". Commas are not correct English usage.

Footnote 1: shouldn't the second Z be Z^0?

*** I don't greatly care. I just use the LHCb symbol, which gives Z.

254-5: Remove “covering the pseudorapidity range 2 < eta < 5”, since you have already discussed the effective range previously. This sentence only confuses the reader (you do no want to explain why it is defined differently here)

*** This is standard EB text. Here we are describing the spectrometer in general, while earlier we described the effective acceptance for the present analysis. This seems clear to me.

260 : “provides a measurement of THE momentum"

*** Again, I wouldn't dare to change approved EB text.

272 : “with a segment of A charged particle track”

*** OK.

273: remove “charged-particle”. This is implied by track.

*** I think this was included by a previous request.

278 : “… to evaluate SOME efficiencies.” -> sloppy! unclear if all relevant ones have been simulated, or all necessary ones, etc. “… to evaluate relevant efficiencies that couldn’t be determined by other means."

*** "some" seems perfectly clear and accurate to me. But I can just say "are used in the analysis."

286: "refine the sample of candidates for analysis" is jargon. "to reduce the same size used in the analysis" ?

*** Sorry - I don't understand your suggestion.

287: "should have induced a *single-electron trigger*" sounds like jargon. Maybe change it into "trigger decision"?

*** Surely "trigger decision" is just as much jargon?

292: is (propose to write ) measured before or after bremsstrahlung correction? Maybe adding "curvature significance" can make this clearer

*** after

296-8: The clause “where E_ECAL, … respectively” can be removed without any loss of information. The abbreviations were already introduced.

*** I was explicitly asked to include this. I agree with you.

Eq.1: Add comma at the end of the equation.

*** Maybe

Eq.2: No need to have periods at the end of Kin and Trig

*** In English usage, when you abbreviate a word you put a full stop (not a period!) at the end unless the last lketter of the word is part of the abbreviation (this Prof. but Dr)

p9: you can remove some of the forward references to Sect. 4 (the one on line 325 is sufficinet)

*** Only trying to help the reader...

309: events -> candidates

*** Surely "candidates" is LHCb jargon. Events is likely to be more widely understood.

312: “The choice of binning is informed ….” -> “The choice of binning is DETERMINED by ….”. Does this mean that the phi* and eta resolutions are small wrt the bin widths (and hence migration due to e.g. bremsstrahlung negligible)?

*** Yes, migration is negligible. "Determined" is too strong - the statistics are one consideration, but so is the desire to have tidy bin boundaries which aligh with other studies. "Informed" implies this (to me).

316: propose to remove the first sentence: " , (2) where eff_track is [...] quality requirements, eff_kin gives [...] and \eta, eff_pid is [...] as electrons, eff_GEC is [...].", then it is still very clear what each efficiency exactly means (including the "cumulative" determination)

*** I think the sentence should remain because reviewers were concerned about this point being emphasised.

327-8: Not clear what is meant with “electron and positron yield reconstructed tracks”. Not needed either. Suggest: “both tracks satisfy the selection requirements”.

*** The electron and positron may not have been reconstructed as tracks at all. The word "yield" tells you that they did.

The discussion of FSR comes a bit confusing at this point. Could it be introduced earlier, when the fiducial region is defined, or in the description of the simulated samples?

*** Maybe, but this is the point at which the correction for FSR is made, so it seemed quite natural.

345: "which varies significantly with y_Z". “…; this part of the efficiency should be modelled reliably.” -> I’d assume that holds for all parts. Remove entire sentence, or write what is really meant (e.g. needs to be modelled with a better than average precision, or something)

*** Certainly not true for the calorimeter energy cuts, since the calorimeter response is not at all well modelled. The acceptance effect only relies on the geometry being accurately modelled.

347: “the calorimeter energy requirements” -> the calorimeter acceptance”

*** No. This refers to the cuts in l.296.

349: Would it be possible to avoid the "GEC" acronym? It is rather specific jargon

*** It is carefully defined in l.274. What do you propose?

354: “apart from the contribution from the leptons”: why not write electrons? Suggest to be more clear: “… after correcting for the additional SPD hits of the electrons”.

*** We are talking about the difference between the muon and electron contributions, so "leptons" is correct.

360: maybe you can refer to the Z->mu+mu- paper for the treatment of the SPD multiplicity systematic. Why fit a gamma function?

*** Well, the Z->mumu paper isn't yet in circulation. And anyway, it's a different channel, so the systematics are not the same. A gamma function because it works. But alternatives are used too.

394: "acceptance is purely geometrical". Well, yes, that's the definition -> acceptance is modelled reliably in the simulation (that's a fact, not an assumption).

*** We don't know that. There must be some uncertainty in how accurately the detector geometry is known, and therefore the simulation isn't perfectly accurate. We assume it's accurate enough (and have tested it of course as mentioned in the ANA note.)

Fig.2: LHCb missing. A pull plot would help to judge the quality. Why "background-subtracted" in parentheses? That's quite crucial.

*** LHCb is not missing. A pull plot is possible.

397: Remove quotes around probe.

*** Defining a piece of jargon. Quotes indicate this.

398: what do you mean by the "energy recorded in each of the calorimeters in turn"? I guess it is once without the E_ECAL/p and once without the E_HCAL/p cut, but the formulation should be more explicit

*** Yes, that's what the text says. I can try to spell it out in gory detail.

413: “such backgrounds are intended to be …” -> such background are EXPECTED to be …”

*** I think "expected" is too strong - that's why we need to test it.

431: “statistical error” should be “systematic error”.

*** No, it's included in the statistical error, as stated.

437: remove "assumed to be"

*** It is an assumption

449: it seems like most (?) systematic errors are added linearly. From fig. 3 it seems that the results is systematics limited. Is the systematic error evaluation not too cautious?

*** Well yes, the largest systematics are assumed fully correlated. Maybe this is cautious, but in the absence of any better knowledge, what else to do?

454: “included some LHC data” sounds as if they purposely neglected available LHC data. Suggest to delete “some”.

*** I guess they included LHC data which were available at the time they starting doing their fits, and they selected the data for which they had sufficient information about covariance matrices etc. It's not for us to speculate on this. I think "some" is fair.

Fig.3: what do you mean with "integration errors"?

*** The errors arising from numerical integration

Table 2 and 3: error -> uncertainty (twice in both tables). Why are there no spaces around \pm?

*** Why should there be? Can be done, of course.

Pg.21: I’m somewhat surprised that the correlation matrix has only positive values. If there is some smearing effect going on, at least some negative values should appear here and there. Is there a single systematic error driving the correlation (e.g. luminosity)?

*** As said before, smearing isn't an issue (that's why we've used these observables). The systematic errors are positively correlated Luminosity is the largest.

Op woensdag 11 februari 2015 17:48:04 schreef Patrick Koppenburg:

...

Hi Pieter,

[ re-sending what I sent privately ]

I am not particularly happy with the replies. Can you try to compactify his replies into a list with only those where he disagrees? That would be helpful to start a reply.

The EB reading is on 16/2, so we need to react before that.

Thanks

Patrick

---

Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics

---

Bfys-physics mailing list Bfys-physics@nikhef.nl https://mailman.nikhef.nl/mailman/listinfo/bfys-physics

Dear all,

I combined the replies from Wouter, Patrick and Jeroen, removed the other ones, and submitted them to cds (to appear soon as LHCB-PAPER-2015-003-001- COMMENT-016, http://cds.cern.ch/record/1989727)

Thanks, Pieter