Please find below the PA responses to the referee reports from PRL for the U+U paper.

New Draft: https://drupal.star.bnl.gov/STAR/system/files/UU_v16.pdf

Comments from the Referrees and Responses from the PAs

PA responses are in bold-face

 

We thank both referees for comments and suggestions that we have found very valuable.
We've implemented the following changes to draft in response to the suggestions we received:

1) Cleaned up the presentation of Figures 1 and 2 

2) Added a mapping between %central and dNch/deta 

3) Merged figures 3 and 4 

4) Added a comparison to the constitutent quark Glauber model to figure 3 and in a new panel in figure 2.

The addition of the new model comparison requires modifications to the abstract, figure
discussion, conclusions, and some additional references. For more detailed responses to
the comments, please see the answers inserted within the reports below.

 

----------------------------------------------------------------------

Report of Referee A -- LE15801/Adamczyk

----------------------------------------------------------------------

 

The Letter addresses the interesting topic of the ability or inability

to control heavy ion geometry and utilizing such control for testing

particle production models and exotic chiral magnet effects. I believe

these are the first U+U at 193 GeV results submitted for journal

publication and are therefore quite interesting and potentially worthy

of publication in Physical Review Letters. The manuscript is general

well written and the points made in a straightforward manner. I have a

few comments that should be addressed before one could decide on

publication.

 

First, the manuscript compares data with a model described as

"Glauber-based models .. with a two-component model for multiplicity."

These two-component models may have been or still used in the

literature, and yet at this point they have been proven incorrect. I

recommend first a comparison with U+U dN_ch/deta in the manuscript for

both the Glauber and IP-Glasma based calculations. This is a key piece

missing.

 

=====================================================================

At this point, inclusion of the dN_ch/deta distributions and

complementary studies in this paper would make the paper far longer

than can be accomodated. More importantly, we found that uncertainties

in the multiplicity dependence of our tracking efficiency preclude

strong physics conclusions from the dN_ch/deta distributions. For that

reason we chose early on not to include those studies in this

draft. We are however cognizant of the need of modelers and theorists

to map between our corrected multiplicities (as shown in figure 1 and

2) and their models. For this reason we have provided analytic

expressions relating our corrected dNch/deta values to fractional

cross sections. They can be found in the text.

======================================================================

 

In the recent publication from the PHENIX collaboration, the

two-component model is dis-proven. The authors should consider a

constituent quark Glauber picture for particle production and check

the results.

 

http://journals.aps.org/prc/abstract/10.1103/PhysRevC.89.044905

 

Once the two-component model fails, it is unclear what more can be

derived from the additional comparisons. One did not need U+U to prove

this point.

 

=======================================================================

We have added results from a constituent quark Glauber model to figure

2 and 3. The constituent quark Glauber model can replicate some of what

Nbinary usually does because the probability for multiple quarks to

become participants goes as TA*TB. For that reason, it also predicts

that v2 will decrease with multiplicity for fully overlapping U+U

collisions. It predicts a smaller slope that is more in line with the

data. It does a quite a good job of describing the slope of v2 vs

multiplicity in Figure 3.  We have updated the text to reflect these

conclusions. The discussion of the physics conclusions has been made

more detailed. To provide space for the improved discussion we've

merged figures 3 and 4.

=======================================================================

 

Second, the IP-Glasma model in fact uses a Glauber picture of nucleon

configurations first. It would be helpful to include the Glauber

parameters used by the IP-Glasma authors and compare those to the

Glauber input or Ref. 26. Otherwise, one is mixing up different

changes in the nucleon configuration versus the effect of saturation.

Again, adding the Glauber with constituent quark production may

reconcile some of these differences as well, and should absolutely be

included.

 

=======================================================================

Indeed, the authors of ref 26 (who are a subset of the authors of this

paper) recommend that the new parameters be used. The IP Glasma model

made use of the more standard Woods-Saxon parameters. We've added

curves to show the effect of the old vs new parameters on the

Constituent Quark Glauber model and on the xhard Glauber model. The

changes to the curves in Figure 3 are visible but small. A larger

difference is seen in Figure 2 when we use the new parameters with the

Quark Glauber model.

=======================================================================

 

Lastly, the claim is that Figure 3 demonstrates the ability to select

geometry. The differences in v2 with +/-10% multiplicity changes are

quite small. The authors should quantify a selection of tip-tip and

body-body and state how much of these contributions are selected. At

first glance, one might assume a large admixture of geometries and

just selecting a slightly different sample at each extreme - which

then might not enable extensive studies specifically in body-body for

the chiral magnetic effect.

 

=======================================================================

The fraction of tip-tip vs body-body is a natural but surprisingly

difficult thing to quantify. First, just as there are exactly zero

events with a zero impact parameter, there are exactly zero tip-tip or

body-body collisions. The only thing we can look at is how well

aligned the angles of the nuclei are. The most useful ways we've

noticed of quantifying how aligned the nuclei are is to look at 2-D

distributions of various angles. In the end, the quantity that is the

most revealing is really the eccentricity which is what we've used in

Fig 3. What is important is that you can apparently achieve a

difference in v2 of roughly 10% for events that have identically small

B-fields. That is the lever arm that is sought.

=======================================================================

 

There are a few minor issues in addition. dN/deta is more commonly

written as dN_charge/deta. Figure 1 - there appears to be a large

vertical blue line at the left side - unclear what that is. In Figure

2 one might comment on why the Au+Au statistics are much worse than

U+U, probably from your centrality trigger. Can this be addressed with

future data sets - which year's Au+Au data is this analysis from?

Figure 3, the caption has some odd formatting issue.

 

=======================================================================

dN/deta has been switched to dN_ch/deta. The vertical line was an

error bar. That point was removed and the figure has been remade to

improve it's cosmetics including larger symbol sizes.  The reason

Au+Au appears to have larger error bars is because we've plotted our

data vs dNch/deta and the Au+Au distribution runs out at a lower

dNch/deta than U+U. The Au+Au data set (from Run 11) is actually

larger than the U+U data set. The formatting issues have been fixed.

=======================================================================

 

----------------------------------------------------------------------

Report of Referee B -- LE15801/Adamczyk

----------------------------------------------------------------------

 

The manuscript reports the STAR measurements of two and four particle

cumulants in U+U and Au+Au collisions. Congratulations for this very

nice analysis. The results are very significant to be published in

PRL. However there are some discrepancies as authors admitted that

Glauber calculations might have given an incorrect description. In

order to get a better picture, more details are needed to add clarity

of the text. Please see my comments below.

 

1/ ZDC resolution reported as 23+/2%. ZDC have been used for

centrality categorization? If yes, more explanation needed how this

limited resolution might effect the estimation of the Glauber

parameters and their uncertainties.

 

=======================================================================

The ZDC's are used only for figure 3 (with figure 4 now merged into

figure 3).  We describe the procedure for modeling the ZDC resolution

in the text now. "The ZDC response was modeled by calculating the

number of spectator neutrons from the Glauber model (accounting for

the charge to mass ratio of the nucleus) and folding each neutron with

the known ZDC resolution for a single neutron."

=======================================================================

 

2/ Please list all the Glauber parameters like woods saxon parameters,

deformation parameters etc. Also how much they were varied to

calculate the uncertainties, list them for each source? How those

uncertainties reflect in dn/deta?

 

=======================================================================

We use the same Woods-Saxon parameters as those listed in the

references and provide numbers for our models wherever the relevant

parameters are not listed in the reference. In general, we prefer that

this letter not turn into a long paper about Glauber modeling so we

haven't repeated the extensive investigations of these parameters that

have already been carried out in other references. Although we agree

that those studies are valuable, they do not fit within the scope of

this letter.

=======================================================================

 

3/ Simulation was tuned with the weights from eta and phi distribution

from data. One should be careful with such tuning. They could be not

understood detector effects which might alter the v2{2,4} results.

 

=======================================================================

We are not sure what is meant by "simulation" above. But indeed,

calculating the correct acceptance and efficiency corrected v2{2} and

v2{4} is non-trivial. We have employed a method that we developed

within STAR that was later seperately ellaborated and generalized by

Ante Bilandzic and published in PRC. The advent of this method allowed

us to achieve the most consistent results across many years of data

with different detector configurations and different tracking

algorithms. We performed extensive checks based on changing the

z-vertex range, artificially creating detector inefficiencies, varying

track selection criteria, comparing runs from differnt years, and

comparing runs from different periods within each year. All

uncertainty about the analysis method has been included in the

systematic errors.

=======================================================================

 

4/ Fig 1: Very hard to follow the legends. Given their small sizes,

circles and squares are both looks same. I would recommend increasing

the size of the legends. Maybe use circles and triangles for better

clarity.

 

=======================================================================

We've updated all the figures in the paper to improve the readability

as suggested.

=======================================================================

 

5/ Fig 1 (inset): You mentioned, v2^{4}{4}>0 in the most central

collisions is because the prolate shape of the Uranium collisions.

What v2^4{4} is also >0 in peripheral collisions in U+U collisions but

its zero in Au+Au collisions. What does that mean? Please add the

explanation in the paper.

 

=======================================================================

c{4} (v2^4{4}=pow(c{4},1/4)) is positive throughout the entire

centrality range for U+U. For Au+Au it becomes negative in the most

central collisions. This is because of npart fluctuations. This is

explored in some detail in the reference provided in the text. We do

not investigate very peripheral collisions in this manuscript but it

may be possible that c{4} could become negative in those collisions as

v2{4} is falling quickly for both U+U and Au+Au. Some of the trends in

peripheral collisions may have been unclear because of the difficultly

in reading the labels mentioned before. Please let us know if we've

misunderstood the question.

=======================================================================

 

6/ 2% of U+U is at dN/dy ~ 910 but in your paper 2% of U+U is at

dN/dy~780. Even the highest multiplicity achievable is fairly similar

between [13] and the measurement, dN/dy~1100. This is a caveat to the

statement "No knee structure is observed" because knee in [13] is

predicted above dN/dy~1000. So the Glauber calculations between [13]

and this manuscript are inconsistent. In that case, I would suggest

add more details about Glauber calculations to help theorists where

the inconsistency occurs.

 

=======================================================================

Because the multiplicity can easily vary in a Glauber model by 10-20%

by changing a few parameters, the most reliable way to compare a model

to the data is to convert everything into % centrality. In this case,

quantities like <eccentricity> and <Npart> are less dependent on the

exact parameters used (and also importantly on the efficiency

estimates for the detector). The knee in reference [13] is at 1%

central where there is no observed change in the data. We agree that

we need more details to make it possible to compare theories to the

data. In order to do so, we've provided a parameterization of

dNch/deta vs %centrality in the paper. This should make it possible

for theorists to compare their models to the data in Figs. 1 and 2.

=======================================================================

 

7/ Where dN/dy comes from? Is it measured or model? If it comes from

the two-component model then what is the uncertainty? Again, please

add more details bout Glauber calculations.

 

=======================================================================

The dNch/dy is measured in data and corrected for efficiency. The only

dependence on models comes from the need to correct for trigger

inefficiencies in peripheral collisions. Those corrections are very

insensitive to the model.

=======================================================================

 

8/ "No knee structure is observed" - this is a very strong statement.

Two caveats, first dN/dy don't match, second Glauber uncertainty. I

would still encourage the authors to consult with the authors of [13]

to figure out the inconsistencies in the Glauber calculations.

 

=======================================================================

This conclusion is robust. Please note that the author of reference 13

is a member of STAR and also one of the primary authors of this

paper. The model in reference 13 was not tuned to match the

multiplicity distributions (that were not yet measured). But the

Glauber model used in this paper is tuned and also exhibits a knee

structure which is a robust prediction. The only ways we know of to

remove the knee structure are to decrease the dependence on Nbin or

to increase multiplicity fluctuations as explored in references

contained in this draft.

=======================================================================

 

9/ You mentioned that new parameters has been tested to reduce the

mismatch? Please add more details like how much they were

varied/changed and how big the effect was in the paper.

 

=======================================================================

We used two sets of Woods-Saxon parameters (as listed in the two

references). We include an inset in Fig. 2 showing v2/e2 with a quark

Glauber model using the new set of parameters. We've also added curves

on Fig. 3 too show how much the model calculations change when using

the different Woods-Saxon parameters. A more complete set of

comparisons in Fig. 2 would include two kinds of Glauber models, two

different sets of Woods-Saxon parameters, and four different v2/e2

values. This would lead to 16 different data sets in a single figure.

=======================================================================

 

10/ I am surprised to see a negative slope in Au+Au case. Glauber

predicts "a slightly positive" slope for Au+Au collisions is

understandable. What make the IP-Glasma predict a slightly negative

slope for Au+Au?

 

=======================================================================

This is presumably a complicated interplay between zdc resolution (and

the width of the impact parameter distribution) and the actual shape

of Au nuclei. The shape of Au is very poorly understood. There are

almost no measurements and models indicate that the deformation of Au

may fluctuate from negative to positive (for this reason we don't rely

too heavily on Au to draw conclusions). We've also included more

discussion of the IP Glasma model in the paper explaining that the

correlation of v2 with geometry depends in a complicated way on the

transverse size of the system and the thickness of the

system. Presumably the weaker dependence of multiplicity on thickness,

causes the negative slope to be washed out by impact parameter

fluctuations.

=======================================================================

 

Please find below the PA responses to the Institutional reviews of the U+U paper. The updated paper is atttached below with changes highlighted in red.

https://drupal.star.bnl.gov/STAR/system/files/UU_v13.pdf

 

Comments from the Egyptian Group and Responses from the PAs

PA responses are in bold-face

Dear STAR-colleagues

 

please accept out sincere congratulations for this very interesting analyze! 

We have some remarks. Please excuse us if we prefer to list them out without categorization.

• In Abstract: The difference shape of U and Au should be mentioned (we think)

Done. We mention the prolate shape of U in the abstract.

• The results are not covering all what we concluded in this Letter, for instance, results about v2{4} are not mentioned.

We tried to highlight the most important conclusions which we think are 1) we have demonstrated that we can select tip-tip vs body-body collisions and 2) are data agree better with a gluon saturation model than a Glauber model.

• In page page 3, left-hand column, paragraph starting with "Even in nearly ..." one should mention the Lorentz contraction. This is entirely missing in the whole text.

We added this.

• The differences between binary and body-body collision should be explained

We added in an extra phrase about body-body collisions having fewer binary collisions to make it clearer.

• The paragraph starting with "In this Letter, we report ..." The mass number of Au is missing while that of U is given!

Fixed.

• More explanation of overlapping is needed 

We added a phrase “where most of the nucleons participate in the collision”

• In right-hand column, Q-Cumulant is used for v2{2} what about v2{2}? ZDCs and related corrections are implemented?

We specify now that v2{2} is calculated directly from particle pairs.

• In page 4, left-hand paragraph, eccentricity should have more explanation 

We put a phrase in one of the earlier paragraphs referring to eccentricity when we are describing the shape of the collisions.

• The two main results of turn-over and lack of knee structure needs more highlights

In principle, we would like to include more discussion but the paper length must adhere to the limits set by PRL.

• Bottom paragraph discusses origin of magnetic fields and their impacts. These are very essential and therefore need more details.

A similar comment here. We can only refer to the extensive literature on the topic.

• Page 5, left-hand paragraph, Lorentz contraction should be discussed/mentioned

We aren’t sure which paragraph this refers to. But by way of comment, the lorentz contraction does not impact on the initial geometry in the transverse direction. For this reason, and since the lorentz contraction is quite familiar in the field, we haven’t put in more discussion of it. We did add it to the previous paragraph though as mentioned above.

• Figure 4 summarizes the slope parameter as measured from the data. These should should be confronted to related models, the ones mentioned in Fig. 3, at least.

This is a good idea but we only have the two data points for the models so we can’t study the trend very well in this figure. We are relying on the work that was presented in the paper we reference.

• The finding that Glauber and two-component multiplicity calculations would arise the question whether other models are there?

There are other models and there are other ideas of how to tune the Glauber model. We did a lot of study on these but decided to limit the scope to avoid layering on many ad-hoc modifications to the Glauber model.

 

best regards from Cairo

 

Abdel

 

Thank you for your careful reading of the manuscript and thoughtful comments.

 

The P.A.s

 

Comments from the Tsinghua Group and Responses from the PAs

 

Please find below our comments to this nicely written paper.

Fig.1, the dashed lines indicating U+U centralities, can be drawn with the same color as UU data points.
Thanks, We will update the figure.

In the figure caption, should explicitly mention the pseudo rapidity range of charged tracks for v2 measurements.
OK. We added this.

line 77, how small the systematic uncertainty is? a quantitative value will be helpful here.
OK. We added "; less than 0.1\% absolute variation on $v_2\{2,4\}$."

line 89-92, would it be possible to elaborate more how the prolate shape of uranium affects the v2{4}^4? by changing fluctuations?
We modified the sentence as follows: the prolate shape of uranium __increases the anisotropy in__ the final momentum space distributions of the observed particles.

line 108-114, would it be possible that the measured v2{2} are affected more by other sources other than the eccentricity, for example, non-flow?
Non-flow is far less dominant in the measurements than affects related to geometry. Even if non-flow contributes some to some fraction of v2{2}, the knee structure should still be visible and only decreased slightly in how pronounced it is.

Fig. 3, it would be better to mention in the caption that, v2 in this figure is v2{2}.
It is now mentioned.

line 210-217, it would be better if some conclusions can be made explicitly from the comparison on v3{2} between AuAu and UU.
We don't have much space to go into a discussion of v3 and only include those numbers for some completeness.

Fig. 4, better to reverse the x-axis, so that, centrality will be more central towards the right (in agreement with Figs.1 & 2).
We'd rather keep it in this format which has been shown now for several years in conferences.

line 229-231, the sentence might be weakened a bit if effects like non-flow can not be proved to be negligible in v2{2} measurements.
Non-flow has been proven to be negligible. v2{2} and v2{4} are dominated by eccentricity and eccentricity fluctuations. Nonflow is sub-dominant.

Thank you for your careful reading of the paper.

Best regards,

Xianglei, on behalf of the Tsinghua group.

Comments from the LBNL group and responses from the PA's