star-fcv-l AT lists.bnl.gov

Subject: STAR Flow, Chirality and Vorticity PWG

List archive

Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark Matter 2022 submitted for review

From: Jiangyong Jia <jiangyong.jia AT stonybrook.edu>
To: star-fcv-l AT lists.bnl.gov
Subject: Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark Matter 2022 submitted for review
Date: Tue, 26 Apr 2022 15:35:00 -0400

Hi, Huan, haojie etc.

Though this investigation is interesting and should continue, I believe the 40 um shift can not be responsible for the 1.5 track or 0.8% difference between Ru+Ru and Zr+Zr.

The reason: if any, the change dNch should be proportional to the difference in vertex dr if this is vertex dependent tracking efficiency . If dr=40um leads to dNch/Nch=0.8%, then dr=4mm would lead to dNch/Nch=80%, clearly not possible! Besides, we also know the p(Nch)_Ru and p(Nch)_Zr has different shapes, and one can not remove the difference by a Re-scaling.

I also do not understand why one think vertex shift should cause Nch to be shifted by a constant independent of centrality.

Maybe I
Cheers,
Jiangyong

On 4/26/22 3:28 AM, haojiexu via Star-fcv-l wrote:

Hi Huan,

I am sorry I may not get your point.

If we think the vertex variations come from real beam position variations, why do we just focus on the few outlier runs, but ignore the variations for the rest of the runs? These variations could be also larger than 40 um. It would be a good question why the first few runs (in blue, see the attached file) have large offsets in refmult, but this small piece of the dataset will not change our conclusions. Actually, if we put a tough cut on refmult,i.e 2sigma, most of these outliers can be regarded as bad runs, see figure 2 in attachment.

with best regards,
Haojie

On 2022-04-26 13:27, Huan Zhong Huang wrote:

Hi Haojie,
Even if I looked at your Fig. 2, I would conclude that the data
support the same hypothesis.
Fig.1 showed that <Vx> = -540 um with RMS 20 um for RuRu, and -582
um with RMS 18 um. I do not know enough about the beam properties to
interpret whether the RMS about 20 um is entirely due to beam position
variation or combination of beam position and vertex reconstruction
resolutions. So if I want to make sure that I look at events truly
from beam position variation, then I need to look at runs out side of
the central blob. You can see these <Vx> outlier runs in both Fig. 1
and Fig. 2 for RuRu collisions. Clearly you observe a multiplicity
variation with the beam position. Since we examine the same RuRu
collisions, the difference may be due to the beam position change.
Then you see that there is a 42 um difference in beam position
between RuRu and ZrZr. There could be multiplicity variations due to
the position difference.
When you make your interesting ratios with different regions, please
state your assumptions. Then we can learn whether the assumptions are
valid or need further justification.
Thanks. Regards,
Huan

-----Original Message-----
From: haojiexu <haojiexu AT rcf.rhic.bnl.gov>
Sent: Monday, April 25, 2022 7:39 PM
To: Huan Zhong Huang <huang AT physics.ucla.edu>
Cc: 'STAR Flow, Chirality and Vorticity PWG'
<star-fcv-l AT lists.bnl.gov>; 'ShinIchi Esumi'
<esumi.shinichi.gn AT u.tsukuba.ac.jp>
Subject: Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark
Matter 2022 submitted for review

Hi Huan,

Thank you for the comment.
The results show in Fig.3 are just two specific small run periods. I
think we should focus on Fig. 2 (the full runIds) instead of Fig. 3.
The obvious Ref-vx correlations shown in Fig.3 is only because we
select few runIds that have large offset, see the Ref-vx distribution
in Fig.2, and rest of the runIds are quite symmety around their
center. Actually, if we select two other subsets of runIds in Figures
2, we can get any correlaitons that we want. For example, if we chose
some other run periods in RuRu, say 300-330 and 350-380, we will get
Ref-vx correlations opposite to Fig.3.

I don't see any issue for the comparing in a signle system or two
system. The only different is that we only need two sets of
comparisons instead of four show in Fig.5. If there is any difference
in one system, it will directly affect the ratios between the two
systems. Since our goal is the ratio between two system, I think it is
better to show all the comparisons.

with best regards,
Haojie

On 2022-04-26 09:33, Huan Zhong Huang wrote:

Hi Haojie,
Thanks for the update.
In order to see any multiplicity shift due to vertex position
difference, one has to focus on the same collision system, not
comparing the two isobar systems. The figure that points to potential
vertex related multiplicity effect is the Figure 3. If you examine the
Ru+Ru and Zr+Zr data separately, one conclusion may be that a slight
shift in <Vx> changes the multiplicity (middle panel Ru+Ru), when Vx
change small corresponding multiplicity shift small too (Zr+Zr); the
<Vy> change does not cause multiplicity shift (bottom panel Zr+Zr).
Then the multiplicity shift of Ru+Ru as a function of <Vr> and <Vy>
was actually caused by the underlying <Vx> shift. This interpretation
would be consistent with your measurements.
I tend to think that these are the more sensitive aspects of your
results in these figures.
Thanks.
Huan

-----Original Message-----
From: Star-fcv-l <star-fcv-l-bounces AT lists.bnl.gov> On Behalf Of
haojiexu via Star-fcv-l
Sent: Monday, April 25, 2022 4:24 PM
To: ShinIchi Esumi <esumi.shinichi.gn AT u.tsukuba.ac.jp>; STAR Flow,
Chirality and Vorticity PWG <star-fcv-l AT lists.bnl.gov>
Subject: Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark
Matter 2022 submitted for review

Hi SchinIchi,

Thank you for the comments. Yes, the plots on page 4 are all e-by-e
quantities, I have corrected the titles.
I also add two plots on the same page under your suggestion. Attached
please find the new version.

with best regards,
Haojie

On 2022-04-25 11:23, ShinIchi Esumi via Star-fcv-l wrote:

Dear Haijie
Thank you very much for the plots.
Could you please re-plot the page 4 figures (x-/y- reversed axis) and
as profile plot, like <V_r,x,y,z> vs RefMult? And please expand the
vertical axis at least a few mm range...
You have indicated "< >” in the x- and y- axis labels for all 4
plots, but this is not the average, they are all event-by-event
quantities, right?
Best regards, ShinIchi

2022/04/25 8:03、haojiexu via Star-fcv-l
<star-fcv-l AT lists.bnl.gov>のメール:

Dear ShinIchi, Huan, Jie, and all,

Thank you for the suggestions. The comparison between two different
run periods is shown on page 3 of the attached file. The correlation
in RuRu collision is just a coincidence, as we do not see any
obvious correlations for the rest of the runs.

The second page is for run-by-run distributions. I also made a plot
for e-by-e distributions on page 4. As I have mentioned in the last
FCV meeting, these distributions are due to the
multiplicity-dependent vertex resolutions.

Therefore, as mentioned by Fuqiang, we can not re-weight or cut
events at the event level due to resolutions. I then apply the cut
on run level, i.e., divide the events into two groups by run-by-run
mean vx and vy, the results are shown on page 5. The distributions
are almost the same in two different <vx> and <vy> groups, which
confirm that no efficiency difference over vx, vy variations on this level.

The mean multiplicity difference between Ru+Ru and ZrZr is about
0.2% in the high multiplicity range, as I have shown in my QM talk.
One bin shift will cause ~0.16% differences in the mean multiplicity
ratio. To achieve our goal on determine the slope parameter of
symmetry energy, if any effect causes a difference above ~0.02%, I
will consider such effect seriously. The effect shown on page 5 is
only about ~0.002%, negligible in our measurements.

Yes, the shape differences are very important in our study and very
sensitive at high multiplicity range. That’s the reason we put more
effort into the refined centrality in isobar collisions. We found
the shape corrections from previous official centrality definition
are quite large, which will cause large sys uncertainties in our study.
We then improved vz corrections and now such uncertainty is
negligibly small. I think this may answer Huan’s question about the
efficiency effect on multiplicity distribution shape at the very
tail region, as the efficiency do have vz dependent but only affects
the high-end-point value. Here are my slides from the last
collaboration meeting for your reference

https://drupal.star.bnl.gov/STAR/system/files/FCV20220216.pdf

Again, thank you for the suggestions.

With best regards,
Haojie

On 2022-04-24 00:39, Wang, Fuqiang via Star-fcv-l wrote:

Hi, Huan, Jie,
Yes, cutting or reweighting events will be biasing the events.
Haojie’s highend point multiplicity vs Vx, Vy already showed that.
The vx vy distributions are mainly due to the vertex position
resolutions therefore dependent on event multiplicity. Cutting on
them is cutting on resolution. One cannot apply cuts or selection
on event level, but only on runs or run periods.
Best regards,
Fuqiang
On Apr 23, 2022, at 3:02 AM, Huan Zhong Huang via Star-fcv-l
<star-fcv-l AT lists.bnl.gov> wrote:

Hi Jie,
That is an interesting idea. I do not know enough about beam
physics to have a firm answer. I assume that the actual beam size
is much smaller than the measured Vx and Vy distributions, but the
center value should be correct beam position. Therefore, the
measured Vx, Vy width may be mostly due to track resolutions. Then
if you only take events on one side of the resolution function in
order to match the RuRu and ZrZr distributions, I do not know if
it will introduce any bias.
Regards,
Huan
From: Jie Zhao <jiezhao1119 AT hotmail.com>
Sent: Friday, April 22, 2022 5:21 PM
To: Huan Zhong Huang <huang AT physics.ucla.edu>; STAR Flow,
Chirality and Vorticity PWG <star-fcv-l AT lists.bnl.gov>
Cc: ShinIchi Esumi <esumi.shinichi.gn AT u.tsukuba.ac.jp>
Subject: Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark
Matter 2022 submitted for review Dear Huan, ShinIchi, Haojie, i am
wondering, whether a data driven method would be helpful or not,
for example, artificially weight or throw away part of the RuRu
data to match the ZrZr vx/vy/vz distribution, and see how big the
difference?
Best regards,
Jie
On Apr 23, 2022, at 3:56 AM, Huan Zhong Huang via Star-fcv-l
<star-fcv-l AT lists.bnl.gov> wrote:
Dear Haojie and ShinIchi,
Thanks for the update and sorry for the late response. It is very
typical nowadays that I lag behind in my email responses.
On a gross scale, I do not have any problem if you claim that
there is no effect related to vertex and/or efficiency
differences. But when you focus on the ratio at the very tail
region, we know that the sensitivity is very much enhanced. So I
have not seen any quantitative estimate that demonstrated that
there is nothing to worry about.
When you examine the multiplicity and the vertex positions to see
if there is any correlations, you did it as a function of run
number. I am afraid that this approach may not be a sensitive way.
As you may know the vertex position averaged over a run contains
too many effects including the z vertex and the slope parameters
which may be run dependent. You may be better off to focus on data
within one run. I understand that you have already done the Vz
dependent corrections. So you may ask the question whether you
correctly attributed the physical reason for the Vz correction and
whether the correction method (for the turning point, instead of
the falling shape for example) is well justified.
When I suggested you to carry out quantitative systematic
estimate, it probably does not make sense to you. Let me try to
elaborate a little. Since you focus on the ratio of multiplicity
in the tail region, you may check how the ratio changes with
possible multiplicity shift (more detailed and finer steps than Yu did).
Then if you decide that in order to constrain nuclear shape
parameters well using the ratios in the tail region, you may
decide our goal of systematic error allowed. That would be our
quantitative goal for controlling the systematics for this physics
measurement.
Once you know the goal that you need to achieve, you may be able
to examine if your Vz/Vx/Vy studies are compatible with achieving
the goal. We know that the average multiplicity between Ru+Ru and
Zr+Zr differ by a few counts while the average Vx and Vy are
slightly different. I wondered that among these two refmult
difference how much should be attributed to true shape difference
and how much to run condition difference. What makes you so
confident that there is no effect from vertex difference at the
accuracy level that we need to achieve. If we trust the simulation
package, we may have to simulate the same collisions with a slight
shift in vertex positions in order to guage the sensitivity.
In any case, thanks for the hard work.
Regards,
Huan
-----Original Message-----
From: Star-fcv-l <star-fcv-l-bounces AT lists.bnl.gov> On Behalf Of
ShinIchi Esumi via Star-fcv-l
Sent: Thursday, April 21, 2022 8:58 PM
To: STAR Flow, Chirality and Vorticity PWG
<star-fcv-l AT lists.bnl.gov>
Subject: Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark
Matter 2022 submitted for review Dear Haojie Thanks for the plots.
I have one question.
Is the second page for the entire run and event for both RuRu and
ZrZr?
Could you plot the 2nd page correlation just for the two run
ranges for each of 2 species, where you draw 2 black lines (run
ranges) in each panel of the 1st page, and plot one range for red
and another range for blue points?
Could you also plot each point event-by-event, not run-by-run for
these two run ranges.
Best regards, ShinIchi
On Apr 22, 2022, at 10:33, haojiexu <haojiexu AT rcf.rhic.bnl.gov>
wrote:
Dear ShinIchi, Huan, and all,
Under Jiangyong's suggestion, I have plotted the run-by-run vertex
distributions in both Ru+Ru and Zr+Zr collisions, attached please
find the plots. The vertex variations in each system are larger
than the average difference between the two systems. I also plot
the <RefMult> vs <vx>, <vy> and <vr> in each collision system.
There is no evidence of efficiency difference over vx, vy
variations on this level.
with best regards,
Haojie
On 2022-04-21 20:49, ShinIchi Esumi via Star-fcv-l wrote:
Dear Huan and all
There were some interesting discussions about this in the PWG
meeting yesterday, the results presented by Haojie in the meeting
were not enough to fully answer your question (you can see his
slide in the usual PWG agenda page at :
https://drupal.star.bnl.gov/STAR/blog/jjiastar/bulkcorr). One
difficulty is the vertex resolution changes with centrality (track
multiplicity in the tpc), but there were few more suggestions
made during the meeting, so let’s see his future updates. Do we
just worry about the actual vertex position dependence of the tpc
efficiency?, which is basically the homework for Haojie to see the
vertex position (3D xyz vertex) dependence of efficiency
(effectively number of reconstructed track for a given
acceptance), Or do we also need to worry about beam
tuning/focusing differences, for example, beam crossing angle
difference etc between the species?
One thing I forgot to ask yesterday was about the simulation test,
where we used say our accuracy of tracking efficiency is of the
order of 5% for the systematic error evaluation in the absolute
yield measurements, which is clearly not enough for these studies,
so the most of the people seem to be given up in this direction,
but I remember there were some task force formed sometime ago, to
revisit our accuracy of 5% on the absolute efficiency in the tpc,
and to see if we can improve this or not. We need to see if there
is any progress in this direction or not, too… Best regards,
ShinIchi On Apr 15, 2022, at 11:42, Huan Zhong Huang
<huang AT physics.ucla.edu>
wrote:
Hi ShinIchi and Haojie,
Thanks for the discussion. To pursue the physics topic, we need to
demonstrate that we have done the systematic study to the accuracy
of much better than 1/(300-400) in multiplicity measurement. We
may have done lots of systematic studies, but I am not aware one
which showed the accuracy matched what is needed for this physics topic.
It will be useful to keep this requirement in mind when you
consider more studies. It would certainly be very useful to have a
full geant simulation of Ru collisions with realistic beam profile
and with the correct magnitude of position shifts to see any
effect.
Haojie: if you think that your previous results have the accuracy
needed, please summarize your results/arguments in a few slides to
circulate. In particular, we will be interested in why you think
that you have the quantitative accuracy of 1/(300-400).
Thanks. Regards,
Huan
-----Original Message-----
From: Star-fcv-l <star-fcv-l-bounces AT lists.bnl.gov> On Behalf Of
ShinIchi Esumi via Star-fcv-l
Sent: Wednesday, April 13, 2022 7:50 PM
To: STAR Flow, Chirality and Vorticity PWG
<star-fcv-l AT lists.bnl.gov>
Subject: Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark
Matter 2022 submitted for review Dear Haojie Thank you for the
confirmation. The question is how accurately we do know the
relative efficiency difference between species, compared to the
observed level of multiplicity difference of ~1/(a few 100). TPC
condition itself is unchanged as we filled two beams
alternatively, but the beam optics/position seems to be different.
The effect coming from less than mm difference of beam using the
meter size detector would be small, but we just need to make sure
the effect is at lease much smaller than ~1/(a few 100) in the
realistic experimental condition.
Best regards, ShinIchi
On Apr 14, 2022, at 11:35, haojiexu <haojiexu AT rcf.rhic.bnl.gov>
wrote:
Dear ShinIchi,
The efficiency corrections are not implemented yet. As we have
discussed during my presentation, we plan to use the same
efficiency for two isobar systems, as the accuracy may not be good
enough to do it separately.
with best regards,
Haojie
On 2022-04-14 09:29, ShinIchi Esumi via Star-fcv-l wrote:
Dear Haojie and all
I think Huan has a point, if the tracking efficiency is really
different between the two species caused by the small beam shift
(or any beam related systematic difference), the beam luminosity
and/or zvertex corrections for the refmult would not correct such
difference, since
“97 vs 98” difference remain unchanged after the corrections and
this difference of “1 out of 500” that we believe that it is
coming from the nuclear structure, but I‘m not sure we have ruled
out any small fraction of “1” that might be coming from the beam
systematics or not.
Do we correct for the tracking efficiency (for the refmult)
independently between two species?
This might be a question of accuracy of our embedding simulation,
that does use the real data, but how precisely we can reproduce
the beam quality/position difference for the embedded track to be
combined into the real events for each species independently. Or
we might need to test this with full geant simulation with
realistic beam optics/profile, to see if our TPC (with realistic
holes of inactive RDOs etc) is sensitive to this or not. This is
to ask ourselves, if we already know our TPC efficiency (better
than
~1/500) relatively between two species, that is being questioned.
Best regards, ShinIchi
2022/04/14 9:41、haojiexu via Star-fcv-l <star-fcv-l AT lists.bnl.gov>
のメール：
Hi Huan,
Thank you for your comments and interest in my QM talk. And also
thank Yu for the testing on multiplicity shift, the effect of one
bin offset is large but I don’t think this shift is reliable.
The multiplicity ratios shown in my QM slides are not the raw
multiplicity distributions. As it was done in the general
centrality definition procedure, we have corrected the luminosity
dependent and the multiplicity distributions in different vz bins
are corrected to vz=0. I think the effect you mentioned has been
taken care of by the procedures. More details of the procedure can
be found in my presentation given in the centrality definition
meeting:
https://drupal.star.bnl.gov/STAR/system/files/Centrality20220301.
pdf Yes, the neutron skin effect can also be obvious at peripheral
collisions. One of the observables is the net charge ratios I have
shown in my QM presentation.
with best regards,
Haojie
On 2022-04-12 03:19, Huan Zhong Huang wrote:
Hi Haojie et al,
This is an interesting approach. I am concerned whether you have
done the systematic checks to demonstrate the sensitivity of
potential systematic bias to the physics conclusion. Yu Hu helped
to plot the multiplicity ratio of Ru+Ru/Zr+Zr if the multiplicity
of the Zr+Zr collisions is systematically offset by a few tracks.
When the Zr multiplicity is shifted by one track out of 300-400
tracks, the ratio changes very significantly. Therefore, it is
critical that we show that there is no systematic bias between two
isobar collisions even at one particle level.
Because of the isobar charge difference, I tried to ask CAD what
the possible magnitude of beam difference (position and collision
axis).
Bill Christie suggested that we should be able to get the data
from reconstructed vertex distributions as a function of Z. Gene
Van Buren has some data on this. He indicated that the beam
position between Ru and Zr could be shifted by 40-50 microns. We
may need to examine this shift with the full isobar data. In order
to examine if this magnitude of beam shift will cause any
systematic bias in the measured multiplicity, we may need to use
GEANT simulations of the Ru+Ru collisions and shift the beam
position to measure potential change in the TPC multiplicity.
That would ensure that we have a good control of the systematics.
There may be other approaches to use experimental data to examine
the potential shift due to beam variations. But I do not know how
well we can control the systematics with the experimental approach.
If this potential systematic shift is real, we may have to revisit
the model used for centrality definition as well.
Naively I would expect that the peripheral collisions would be
more sensitive to the shape and distribution of the neutron skin.
In any case, this is an interesting topic. It would be good if you
help evaluate these sensitivity issues. I am sorry that it took me
long to catch up with many interesting QM talks and did not
comment sooner.
Thanks. Regards,
Huan

Re: [Star-fcv-l] STAR presentation by Haojie Xu for Quark Matter 2022 submitted for review , (continued)