FROM PIETRO:

Abstract, introduction and conclusion: I expect that these sections will be substantially improved. The second half of the abstract ("A more differential measurement...") about the details of how many bins of how many variables and in what combinations have been considered is a bit repetitive and confusing. It should be replaced by something interesting, answering the question: why is this new measurements interesting and why should I read this paper? What does it add or improve with respect to existing measurements? The question is addressed in the introduction, but in a rather disorganized way. I find the introduction of the AN note much nicer. I will wait for a more final version to give more specific comments. In the conclusion (now missing), what kind of comparison do you expect to include? Will the results be compared to other energies, as in the AN, or even to other conditions and experiments? I think that it is important to find a comparison that attracts interest on the new measurement. What makes the new measurement more than a boring sequence of plots? What is the main new piece of information it brings? For example, is it expected that there is practically no energy dependence with respect to 2.76 TeV results? Was a similar energy dependence found in other measurements? Does the higher pT reached by the new measurement contain any physical information? Is the suppression increasing, remaining constant or decreasing with pT at high pT?

Event selection: I find the definition of centrality incomprehensible: "centrality is defined using fractions of the inelastic hadronic cross section determined from the HF distributions, with 0% denoting the most central collisions". Since I don't know anything, as many readers, I ask: cannot you rather tell WHAT is centrality defined to be? How are those "fractions of the inelastic hadronic cross section" defined, i.e. with respect to what TOTAL cross section? A calculated inelastic cross section? Is therefore this a model-dependent number? How are then the values of these (how many?) fractions converted to THE centrality value measured for one event? "Using fractions" is very vague.

We have updated the text to:

The \PbPb sample is split in bins of collision centrality, defined using fractions of the inelastic hadronic cross section where 0\% denotes the most central collisions. This fraction is determined from the HF energy distribution~\cite{Chatrchyan:2011pb}. The most central (highest HF energy deposit) and most peripheral (lowest HF energy deposit) centrality bins used in the analysis are 0--5\% and 70--100\%.

All pp figures: The plots for pp collisions have the label "0-100%" in all legend lines, presumably a remnant of the Pb-Pb macros.

We plan to improve the style of the figures between green light and approval, and we will fix this at that time.

Acceptance and efficiency: referring to the point raised by Andrea during the meeting, it is actually not written in the PAS (it is only in the AN) that there is a difference of treatment for the acceptance-efficiency correction of the b fractions with respect to all other measurements reported in the paper. Given the information that will be published, it is important to uniform the treatment, applying also to the b fractions the acceptance x efficiency correction and evaluating the systematic uncertainty in a proper way. The adoption of raw-minus-corrected as uncertainty would make sense for a measurement where the acceptance and efficiency have been determined as a global factor of the kind 0.98, or 0.97, etc., so that, instead of bothering with systematic tests, you simply assume that the uncertainty in a correction of 2 or 3% cannot be bigger that the correction itself. However it does not seem the case here. Moreover, you apply proper uncertainties when these B fractions are used to derive the prompt and non prompt yields, and in the PAS there is no mention of the b-fraction-specific procedure.

This is correct. We are now correcting the b fractions for acceptance and efficiency.

7.2: Eq. 1 is a bit useless and misleading. To determine Nprompt and Nnonprompt in this way is what you did, but this reflects only your chosen parametrization of the signal fits (not discussed in any formula before in the PAS text). An average reader would consider this equation as a DEFINITION of the b fraction, not as a derivation of Nprompt and Nnonprompt, since this derivation could be done directly in the fit by using Nprompt and Nnonprompt as parameters, as it would be natural in a fit aimed at measuring particle yields. I think that Nprompt and Nnonprompt can be defined in words: they are the prompt and nonprompt signal yields. Quickly browsing, I did not find a definition of f_b, and the figures use "nonprompt fraction", so it seems that it is also useless to have a formula defining f_b (and is "f_b" ever used by the way?). Also the "nonprompt fraction" can be defined in words.

We agree that this equation is useless / misleading and have removed it. Also, the nonprompt fraction has been better defined in the signal extraction paragraph, where it was already introduced.

Fig. 1-right: -- The horizontal scale is very wide and the majority of the plot is occupied by the "lack of data", i.e. zeroes. Maybe showing from -1 to 4 would be more informative. In fact the scales are very compressed. -- The pulls contain only one point deviating significantly, but when we try to see in the distribution where the point is we are surprised that the point seem to be overestimated by at most 2 sigma, not by more than 4, by the curve. I confess I don't know what a pull is. It the pull a square of the residual? -- Does the fit use the function at the bin centre or the integral of the function? With such steep function, by shifting the bin centre only a bit the fit quality and therefore the pull would change a lot. -- It is not nice that all points at zero (which means: no event) are plotted. For two reasons: they are not plotted in the panel above, which is in log scale (so that it is difficult to look for correspondence of points between the two panels) and, above all, they contain no information, especially for the pulls plot: if they contained information, they would contribute to the degrees of freedom in the chisquare there written, which is certainly not the case (I hope). So I think that the balls for the bins with zero events should be removed.

We plan to work on the style of the figures between the green light and the approval. We will reduce the X-axis range of the right panel of Fig.1 as you suggest, as well as remove the pulls, among other improvements.

All figures: somewhere in legends or captions the uncertainties should be described (what is systematic and what is statistical).

done

The sentence "The bin boundaries are indicated by small horizontal lines" in several captions can be improved. Actually the bin boundaries are indicated by the small vertical lines, the tick marks at the extremities. The horizontal lines can be huge if the bin size is huge. Above all, I think that the meaning of the points with bars should be discussed better. In fact

1) it is many times impossible to see the error bars, which are hidden by the points. In particular, statistical and systematic uncertainties are often not distinguishable. By the way, are they statistical and systematic or statistical and total? If the first is the case, it is essential that they can be "read from the figures" quite clearly and individually.

2) the big points have actually no added meaning at all: all the information is contained by the interval boundary and by the value of the ordinate, i.e. by the horizontal line. On the contrary, the position of the point is misleading since it is always the middle of the bin. In most cases in this paper (except maybe the rapidity dependences at mid rapidity) the middle of the bin is rather far from something physically relevant. It would lead to big mistakes of interpretation to fit the pT distributions using such points.

In summary, the ONLY relevant information is given by the horizontal lines and, to avoid long discussions about how to calculate better bin "centres" (not easy, especially for ratios), I think that the points should be removed. Anyone can draw a little square in the middle of each bin and to have the squares pre-drawn "for dummies" is certainly not essential, especially given that there is no physical meaning attached to them. To use the data rigorously, only the integral counting within one bin (or their ratios, in the case of the ratios) should be used.

The ideal would be to have in each bin two horizontal bands (with no points), for statistical and systematic (or total) uncertainties. In this way, we would continue to see the uncertainties and distinguish them even when they are very small, because the two bands would always remain visible (if the thinner band is drawn on top).

The way the bin boundaries are explicited visually in the plots was used previously in published CMS papers (including the previous J/psi RAA paper, HIN-14-005), and seems fairly standard in other experiments too. We would prefer to keep the current way to present the bin boundaries, as well as the statistical and systematic uncertainties, unless a majority of the ARC members feels strongly against it.

Note also that it is recommended to put the data points at the midpoint of the bin (see https://twiki.cern.ch/twiki/bin/view/CMS/StatComWideBins). In the case of the comparison with theory, the theory should be binned to be compared to the data in a meaningful way.

Eq. 3: the comma at the end looks like a "prime" for the "centrality bin fraction". You can put some /;/; before the comma.

fixed

There is a LaTeX mistake in the last figure.

we couldn't find it? In any case the style of the figures will be improved. -- EmilienChapon - 2017-06-23

Edit | Attach | ~~Watch~~ | Print version | History: r1 | Backlinks | Raw View | WYSIWYG | More topic actions

Topic revision: r1 - 2017-06-23 - EmilienChapon

Copyright &© 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.

or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback

or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback