-We thank the referee for is/her comments. Our reply together with respective
-changes to the manuscript can be found below.
+We thank the referee for his/her very constructive report. Our reply to the questions raised by referee, together with changes made to the manuscript, can be found below.
> The plots are difficult (impossible?) to read in black and white. The authors
> should consider addressing this issue.
-It is true that the plots are not very legible in black and white. However, we
-have tried plotting the histograms in different linestyles and it does not
+It is true that the plots are not very legible in black and white. We tried plotting the histograms in different linestyles, but unfortunately this does not
improve the legibility.
> The manuscript naturally leads the reader to wonder what JEWEL would
> predict for di-jet asymmetry in high multiplicity pp and pA collisions. Providing
> predictions for the asymmetry in these collision systems would dramatically
> strengthen the paper, especially as the authors conclude that the di-jet
> asymmetry is only weakly correlated with jet pathlength.
-High multiplicity pp and pA collisions are clearly very interesting. Unfortunately,
-Jewel in its current form is not contructed to address the physics of these small
-systems (in particular the correlations between the hard process and the soft
-component of the events).
+We fully concur with the referee in that high multiplicity pp and pA collisions are
+of great and topical interest. Unfortunately, in its current form, Jewel is not constructed to address the physics of these small systems. The background (the QGP) with which jets interact in Jewel is provided through a parametrisation of Bjorken hydrodynamic evolution. Addressing, in a meaningful manner, the high multiplicity pp and pA cases would require the availability of analogous parametrisations (alternatively, profiles obtained from full hydro simulations) for these cases where the applicability of hydrodynamics remains a matter of contention.
+
+We note that we do not conclude that the di-jet asymmetry is weakly correlated with average path-length but rather that the di-jet asymmetry is weakly correlated with the difference of path-length of the two jets in the di-jet pair.
+
> The manuscript, and the Introduction in particular, is significantly lacking
> in citations. This lack of citations must be adequately addressed before
> publication.
-
-
+::INCLUDE REFERENCEDS ::
> When the authors write “JEWEL—that has been validated for a wide set of
> observables and a specific observable,” what wide set of observables and which
> specific observable do the authors mean?
-Part of the confusion is due to an unclear formulation. We have added a list of
-observables and reformulated the sentence, it now reads
+Part of the confusion generated by this statement is due to what we realize now was a rather complex sentence structure. To improve clarity, we have added a list of
+observables for which Jewel has been validated and reformulated the sentence to read as:
"By considering an event generator - JEWEL - that has been validated for a wide
set of observables (jet rates and shapes, fragmentation functions, di-jet observables,
leading hadron suppression etc.) and the di-jet asymmetry as an example for a jet
observable, we illustrate a generic strategy for achieving such identification.
+
+
> In the next paragraph, the authors write “we attempt to qualify common
> assumptions” [emphasis mine]. Do the authors rather mean identify, quantify,
> or ...?
-
-
-
-
+We use the word 'qualify' in one of its standard meanings as 'making (a statement or assertion) less absolute; add reservations to'. We believe the word to reflect precisely the discussion we carry out in the manuscript aimed at challenging common assumptions made in the literature (e.g. that a sub-leading jet in a di-jet pair necessarily crosses a longer in-medium path-length than its leading counterpart).
> In fig 1, what do the yellow boxes in the Ratio plot represent? In several
> subsequent plots, the yellow boxes (presumably some kind of uncertainty
> estimation) do not extend fully across the x-axis (e.g. figs 2, 5, 6, 10, 11,
> and 14); why?
-The yellow band in the ratio plots shows the uncertainty on the reference (i.e. the
+The yellow band in all ratio plots shows the uncertainty on the reference (i.e. the
data points in fig. 1 and the red histogram in all other figures). We have added
-this information to the figure captions. It is then also clear that there is
-no yellow error band in empty bins.
+this information to the figure captions. This statistical cannot defined for empty bins (i.e. that contain no events) and, as such, no yellow band is assigned for these cases.
> How does JEWEL’s pp AJ compare to experiment? E.g. fig 1 (or a new
> figure) would benefit the manuscript tremendously by comparing JEWEL pp to
> data explicitly.
-There is not measurement in pp with the same cuts as fig. 1, we therefore had to add
-a separate plot with a comparison of Jewel to pp data.
+A direct comparison with pp data was not included in the original version of the manuscript as no data is available for the experimental cuts used for the PbPb data shown in the figure.
+
+We have included a new figure (new fig. 1) with an explicit comparison of Jewel pp results with CMS pp data for cuts in which a measurement is available.
+
> The authors may find it beneficial to their readers to include a clarifying
> statement in the first paragraph of Section 2.2 noting that, although b = 0, jets
> are distributed according to the binary distribution produced by the Glauber
> model.
-We added a sentence clarifying this.
+We added a sentence clarifying this point.
> In the second paragraph of Section 2.2, the authors appear to state that jets
> and subleading jets are reconstructed with |η| < 2. But the authors state in
> the second to last paragraph of 2.2 that subleading jets are reconstructed with
> |eta| < 5. Could the authors please clarify the text?
+
The former refers to the di-jet sample and the latter to the gamma-jet sample. To
-avoid confusion we have changed the text in the second to last paragraph to
-"In addition to the jet cuts, which are $|\eta| < 5$ and $\pt > \unit[20]{GeV}$ in
-the $\gamma$-jet sample, the initial parton is required to be within
-$|\eta| < 2.5$ and ..."
+avoid confusion we have re-written subsection 2.2 and, to stress that two samples are used in the study, changed its title to 'The generated di-jet and $\gamma$-jet samples'.
+
> At the start of Section 3, the authors state that in pp collisions, the dijet
> asymmetry is induced entirely by fluctuations in the fragmentation pattern.
> This statement is true at leading order, but what about the case of 3+ jet events,
> which make up a sizable fraction of jet events?
-Exactly this is explained in the accompanying footnote, we don't know how this could
-be made clearer.
+This point is addressed in the footnote. We refer to all emissions beyond a core 2->2 process as fluctuations of the fragmentation pattern. As explained in the footnote, this can be done irrespectively of whether multi-jet configurations arise from LO+parton shower or multi-leg matrix elements matches to a parton shower.
+
> Throughout the paper, the authors appear to use “fluctuations” as a shorthand for
> “fluctuations in the medium-induced jet energy loss for a given pathlength.” The
> authors should clarify throughout the paper what precisely they
> mean when they write simply “fluctuations,” either by making a clear definition
> of what they mean by “fluctuations” near the start of their manuscript or by
> providing clearer text at each instance of use. As an example, in the second
> paragraph of Section 3, the authors write “we shall argue that the di-jet
> asymmetry in heavy ion collisions is dominated by fluctuations and that the effect
> of path-length difference is small”: the reader does not know what specific
> fluctuations the author has in mind. Similarly, in the last paragraph before
> Section 3.1, the authors write “hard fluctuations” but leave the term undefined.
We have re-arranged the text in the respective paragraph to make it clear what kind
of fluctuations we are referring to. We have also checked that in the rest of the
text we indicate which kind of fluctuations are meant.
> The authors show in fig 2 that the di-jet asymmetry is surprisingly insensitive
> to the exploration of the full geometry of the collision. What is the physics
> origin for this insensitivity? Perhaps the Glauber model biasing the production
> points to be mostly near the origin? Is there a greater pathlength dependence for
> non-central collisions? Additionally, the authors correctly note in the Summary
> that other energy loss models (e.g. YaJEM) have significant surface bias, which
> should lead to a much larger difference in paths traversed by the subleading jet
> compared to the results shown in fig 3. The authors should comment on the
> model dependence of their conclusions regarding the insensitivity to pathlength
> fluctuations.
We claim that the insensitivity to the geometry is due to fluctuations in the
fragmentation pattern and the medium induced energy loss. Some of the other models
(e.g. YaJEM) do not contain both these sources of fluctuations, so their different
findings are no contradiction to our work. We have added a comment to the discussion
"...this is not in contradiction to our findings, as this model does not contain
energy loss fluctuations"
We already comment on the model dependence in the discussion section, namely
"Although our analysis was carried out in a specific implementation of jet-medium
interactions, namely JEWEL, it relies on rather generic features and we believe
that the main findings should hold in general. "
> Section 3.2 introduces a discussion of initial state emissions. What exactly
> do the authors mean by initial state emissions? Do the authors mean nearly
> perpendicular emissions from the parton lines moving down the beampipe prior to
> a hard scattering? It’s hard to believe that a significant fraction of the measured
> jet events come from such nearly perpendicular emissions from partons moving
> down the beampipe. Additionally, if initial state radiation leads to jets that
> are identified experimentally, shouldn’t these initial emission jets be included
> in the study? And, if it is the case that one cannot distinguish experimentally
> the difference between initial emission jets and jets from the hard scatter, then
> why investigate the difference in di-jet asymmetry when including or excluding
> these initial state jets?
It is known that initial state jets contribute significantly even around central
rapidities. The reason for removing these jets is that we want to carve out the
systematic effects of pt-loss in vacuum and medium, which is only possible when
-looking at (predominantly) final state jets. For any data comparisons etc. initial
+looking at (predominantly) final state jets. For any data comparisons initial
state jets are of course included.
> The authors write in Section 3.2 that “The procedure works reasonably well
> in practice. . . ” What is the metric by which the authors determine that the
> procedure works reasonably well?
As stated in the text there is no unambiguous way of doing this and the procedure
we employ is only approximate. What we mean here is that we varied the cut and
looked at many distributions to see whether the result makes sense or not.
> What do the authors mean by the “ME+PS level” label in fig 6?
We changed this to "initial asymmetry".
> In Section 3.2, why do the authors perform the gamma-jet analysis to determine
> the initial momentum of a jet when they are able to track the individual
> partons and the effect of recoil?
There is no way of matching a jet to an initial parton in di-jet events. Already
at parton level initial partons may fragment into several jets and jets may receive
contributions from different initial partons. After hadronisation it is utterly
impossible the say which hadron originated from which parton.
We have expanded the respective explanation:
"The effect of transverse momentum loss when going from initial parton to
reconstructed jet cannot be studied in the di-jet sample, as it is impossible to
match a jet to an initial parton in this case. It can, however, be isolated by
considering a sample of $\gamma$-jet events with initial state parton showering
disabled. In this case there is only one initial parton and no initial state jets,
so all observed jets must originate from this initial state parton. Here, we
associate the hardest final state jet with the initial parton and study the $\pt$
difference between the two, $\Delta \pt = p_\perp^\text{(in)} - p_\perp^\text{(jet)}$.
> Similarly, how do the authors compute the
> initial values they quote, such as the jet asymmetry A_J^(in)?
We added a footnote and a sentence explaining this:
"\footnote{This configuration is easily accessible in \jewel, since there is no
a posteriori reshuffeling of momenta in the parton shower. It is known as soon
as the scale of the first splitting on each of the outgoing partons from the
matrix element has been determined.} The asymmetry of this initial configuration
$A_J^\text{(in)}$ can be computed by substituting the jet $\pt$'s $p_{\perp,1}$
and $p_{\perp,2}$ by the $\pt$'s of these initial partons $p_{\perp,1}^\text{(in)}$
and $p_{\perp,2}^\text{(in)}$ in equation (1)."
> What is the formula used by the authors to compute the jet mass?
The jet mass does not appear in this paper.
> As above, how do the authors determine the initial jet mass, m^(in)?
This should be clear after we have clarified how the initial pt is determined. Also,
we refer to this quantity as the initial parton's mass and not the initial jet
mass to avoid confusion.
> I would think either
> the authors could determine the initial jet quantities, such as A_J^(in) , m^(in) ,
> etc. and thus would not need to perform a gamma-jet analysis, or the authors would
> not have the ability to determine these initial quantities.
The answer is again, that the partonic initial configuration is known in the MC, but
there is no way of matching a jet to an initial parton in di-jet events.
> In the fifth paragraph of 3.2, the authors write that the final asymmetry is
> “larger” than the initial one. That does not generally appear to be the case
> according to fig 7. In particular, the 0.2 to 0.4, 0.4 to 0.6, and even 0.6 to 1
> curves all gain weight for smaller asymmetries. Rather, the distributions are
> broader. The authors provide a nice physical picture for why the asymmetry
> may increase. But what is the physical picture that leads to a decrease in
> asymmetry? How is this physical picture for a decrease in asymmetry consistent
> with or contradictory to the picture for an increasing asymmetry?
Yes, the distributions get wider, but they are markedly asymmetric so that the average
-increases. This just means that it is more likely for the asymmetry in increase,
+increases. This just means that it is more likely for the asymmetry to increase,
but of course it is also possible that the leading jet loses more energy than the
sub-leading and the asymmetry decreases. We have added a sentence to clarify this:
"This is a statement about averages and does not preclude the possibility that the
leading jet loses more energy than the sub-leading one, leading to a decrease of
the asymmetry (figure (8) shows that this does indeed happen)."
> What is the precise definition of Delta p_t?
This was fixed together with another point above.
> In Eq 3, and in general, the authors should probably note that A_m/pt is A_m/pt^(in).
We changed this.
> Below Eq 3, figures 10 and 11 are labelled out of order.
Thanks for pointing this out, we have repaired that.
> In Section 3.3, while I naively accept the authors’ intuitive arguments, the
> authors should quantify their statements that “nuclear modification. . . leads to
> small differences” and that the first final state emission is unmodified by the
> generation of a medium in heavy ion collisions.
Concerning pdf effects we added
"...(at reasonably high $\pt$ and $\sqrt{s}$ nuclear pdf effects are typically of
the order of a few percent, cf.\ e.g.~\cite{Helenius:2012wd}, and have a very limited
impact on our argument)."
The timescale for the first splitting in this kinematics is about 0.01 fm, which renders
it very unlikely that it can be modified by the medium. We have added this number to
the text to support our argument.
> In the fourth paragraph of Section 3.3, the authors write “the dependence
> on the min/pperpin ratio is qualitatively similar to the pp case and hence a
> result of vacuum-like dynamics.” The authors should be cautious about mixing
> correlation with causation.
We rephrased the statement to avoid confusion.
-"...Second, the dependence on the $m^\text{(in)}/\pt^\text{(in)}$ ratio is
-qualitatively similar to the p+p case, which indicates vacuum-like dynamics. ..."
+"...Second, the dependence on the $m^\text{(in)}/\pt^\text{(in)}$ ratio is qualitatively similar to the p+p case, which indicates that vacuum-like dynamics plays an important role also in the presence of a medium. ..."
> In the second to last paragraph of Section 3, the authors should provide
> (as they have done so well throughout the paper) a physical picture to understand
> why medium induced energy loss fluctuations lead to a reduction in the
> final measured asymmetry compared to the initial asymmetry when that initial
> asymmetry is large.
We have expanded the discussion:
"Compared to the p+p case, the distributions are broader and, for large initial
asymmetries, there is a tendency for the final asymmetry to be smaller. Both
features are a direct consequence of medium related fluctuations (these can both
increase and decrease the initial asymmetry, but when in a configuration with large
initial asymmetry becomes even more asymmetric it is likely to fail the di-jet cuts
and thus disappear from the sample)."
> In the second to last paragraph of the Summary & outlook, the connection
> the authors draw between their work and strong coupling seems spurious as the
> comparison is apples (quasiparticles) to oranges (soup).
+We agree with the referee that the fundamental objects with which one deals in the two approaches are very different and a connection between then is at best elusive. However, we believe that a comparison of the qualitative pictures that result from the approaches is meaningful. We find (in a purely perturbative picture) that the mass of the hard parton from which a jet originates plays a very important role on how much p_t it will loose while traversing a QGP-length. The authors of Ref. [29] of our original manuscript find (in a purely non-perturbative computation) that a energy loss of a 'jet' (the closest to a real jet that can defined in such a construction) is entirely driven by its 'initial opening angle' (a non-perturbative concept that can be qualitatively related to the mass of the initial parton in our perturbative description). In this sense, two very different formulations of the same problem result in a very similar physical outcome. This is a parallel we believe should be mentioned in our discussion. We have expanded the text in the manuscript as to convey the sense in which we believe the results of these two very different approaches can be compared. I know reads:
+'...Our finding that the fractional $\pt$ loss depends largely on the initial mass to $\pt$ ratio and only weakly on the initial $\pt$ is in qualitative agreement with the observation made in~\cite{Chesler:2015nqz}. There, within a purely non-perturbative scenario where objects analogous to QCD jets were considered, the authors found that the fractional energy loss of a jet depends only on the jet opening angle. Since the jet opening angle (an angular shape variable) can be related, at least qualitatively, to the initial mass used in our work, we believe that both observations point towards the same physical picture... '
> The English in the manuscript is perfectly understandable.