STAR Flow, Chirality and Vorticity PWG

[Huan Huang <huang AT physics.ucla.edu>]

Hi Evan,

   Thanks for the comments.

  We have been trying to find a good description of our results which reflect what we have learned about the background subtraction and also physical limitations of the measurements. Possible model dependence is one aspect. Our plan is to finalize the analysis details first and also work with the GPC to finalize the physics message. 

  It is our intention to be on the conservative side. The over-subtraction for the AMPT model with our method could have a few possibilities: one is that AMPT has non-hydro-driven background which also exists in real data and our method would over-subtract in the dg-112 calculation; the other is that AMPT has non-hydro-driven background which does not exist in real data and we should not worry about the over-subtraction in AMPT. When Zhiwan/Gang compared the dg-112 background with v2-Delta-delta background, they found that AVFD actually did a better job in describing the relationship between these correlators. But we also considered the possibility that the non-hydro-driven background is real and we did over-subtract the signal. Then in this case, we are a little too conservative. The charge separation signal would be larger than what we have reported. Personally I do not mind if that is the situation. For the short paper, we may be limited by the length. I welcome you and other PWG members to help us formulate an appropriate way to describe such a possibility in the paper. 

  What I objected to in Fuqiang's paper is arbitrarily mocking up a toy model without being constrained by basic measurements from experiments to show the ESS method does not work while ignoring the fact that the ESE method would not work either in those cases. The strong conclusion for the ESE method in his paper cannot be justified by the data reported in the paper. I hope that you read the paper and my comments. 

  Event shape analysis will have some model dependence as I have admitted in my first comment to Fuqiang. We hope to have that reflected in the papers prepared, certainly in the long paper where there should be more space to discuss. I welcome your specific suggestion on the text of the papers.

  Thanks. Regards,

  Huan

On Thu, Aug 1, 2024 at 6:07 PM Evan Finch <levanfinch AT gmail.com> wrote:

Hi Huan, all,

Thanks for sharing the discussion with us.  If I can interject a comment/question:

My chief concern for the short version of the ESS paper is that the model dependence isn’t discussed.  In my view, a key argument for the overall method and the details (e.g. using q2 pairs, v2 of singles) is that it performs background subtraction very well with AVFD.  The over-subtraction in AMPT is an argument against the background subtraction being correct, but from the methods and long papers and your discussions (and Gang and Zhiwan, of course) I understand you feel that there are strong reasons from data/AVFD/AMPT comparisons not to trust AMPT for this purpose.  I don’t understand that argument in detail and am somewhat worried that if it using CME-sensitive correlations then it may not be independent of the finding that the AVFD and data intercepts are both zero. 

Anyway, I’m advocating for trying to fit a model discussion into the short paper as I think it is a crucial part of the argument for the method, and also I think that succinctly laying out the argument for why AVFD is the correct comparison would be helpful for the group (well, for me at least) to follow the argument better.  

Thanks,

-Evan

On Jul 31, 2024, at 6:18 PM, Huan Huang <huang AT physics.ucla.edu> wrote:

Dear All,

  I had some major concerns about a paper Fuqiang recently posted (2407.14489). I initially hoped that I would convince Fuqiang that we had legitimate concerns for the conclusions in his paper and he may consider revising his paper.  But my wishful hope very quickly faded. 

  Since the paper draft on the CME search with the ESS method is in the pwg review and we need to move forward with the GPC review, I think that it is necessary for the people in the pwg to be aware of the discussions surrounding Fuqiang's paper.  

  I iterate that both ESE and ESS can perform event shape analysis. The ESE seems to be sensitive to eccentricity and longitudinal correlations. The ESS is sensitive to a combination of eccentricity and local emission pattern fluctuations. We found that ESS can suppress the CME background more effectively. We have done studies with AVFD (hydro-inspired) and AMPT simulations to evaluate the residual background and possible deficiencies in the ESS approach as noted in the analysis note and the method paper published. We do agree that the ESS method will not work for situations where non-hydro-driven correlated background is arbitrarily marked up. In such situations all event shape analysis will fail. 

  The ESS method is not "mathematically proven" or entirely based on "my logical thinking". It is largely empirical based on many trials. We welcome suggestions to improve the approach to suppress the CME background. And we are working on how to formulate the physical limitations on our ESS results. If you claim that you have a much superior method to suppress the CME background, let us use AVFD and AMPT simulations to evaluate. I do not believe that the ESS approach is mathematically proven to be THE approach for CME search. I can be happily convinced with simulated data or real STAR data.

   Fuqiang sent a reply to my last email to him. I will let him forward to the pwg. He may have the last word on the email exchanges we had. In any case, we have laid out our major concerns and objections to his paper. 

   In retrospect, I admit that there is no need for me to call his work "sloppy".  I regret the prompt use of the word in my email. I do urge people to look at the paper, especially the attached figures, and draw your own conclusions before you read the conclusions in the paper.

  Sorry for the long email and if you read the whole email thread, you are a true hero of CME physics. 

  Regards,

  Huan

  

---------- Forwarded message ---------
From: Huan Huang <huang AT physics.ucla.edu>
Date: Wed, Jul 31, 2024 at 12:32 PM
Subject: Re: Concerns for your posted paper
To: Wang, Fuqiang <fqwang AT purdue.edu>
Cc: Frank Geurts <geurts AT rice.edu>, Lijuan Ruan <ruan AT bnl.gov>, ptribedy AT bnl.gov <ptribedy AT bnl.gov>, Sooraj Radhakrishnan <skradhakrishnan AT lbl.gov>, subhash <subhash AT rcf.rhic.bnl.gov>, Gang Wang <gwang AT physics.ucla.edu>, Wang, Fuqiang <fqwang AT purdue.edu>

Fuqiang,

  What you said essentially is that because "ESE is mathematically proven to be the right event-shape method. All LHC experiments used it.", therefore , you could ignore your own simulation results and the quality of your simulated data, and draw a strong conclusion to support the ESE method.

  For the ESS method, despite what we have done for the AVFD and AMPT simulations, and what you showed in the paper for the hydro-model, all indicated that ESS worked well to remove hydro-driven background. You ignored these results, you insist that the ESS has to work on all kinds of other models which have large non-hydro components or have nothing to do with hydro to begin with. Because of the way the event shape works, for those models the ESS and ESE will fail to some extent equally. Yet you choose to ignore our arguments and perhaps based on your mathematics-based statement of belief, you continue to promote the ESE and argue against the ESS.

  Please apply your ESE to the EPOS model and show me the simulation data to demonstrate that the ESE works. That is all we are asking. 

  I will be happy to show your Figs. A.4 and A.6, and the conclusions you have drawn from these Figs about the ESE method at any meeting or conference. I do not need to comment on the quality of work. The Figs will speak for themselves. In fact, I attach these Figures in this email. Please note that the conclusion was that the ESE method is the right event-shape method which will help to get the intercepts to be ZERO and the CME background can be fully suppressed with this ESE method. Of course, we disagree on two aspects: 1) if your model has non-hydro driven background, it is not guaranteed that the ESE or the ESS methods can remove all the background; 2) with the quality of data shown in the Figs, the ESE does not shown to have sensitivity on shape selection and NO conclusion about the ESE method should be drawn from the simulated data in the paper.  You may be disappointed that I called your paper "sloppy" and "unscientific", and you are not "embarrassed" by the paper. Just let the Figs and your paper conclusions show, and I do not need to say any words.

  If nature and physics can be proven by "mathematically" alone, I guess that we experimentalists will soon lose our jobs.

  We have no intention and cannot prove that the ESS method "is the correct event-shape method for all possibilities of data makeup". The ESS method was intended for suppressing CME search background. We have already stated that if you want to compare with hydro calculations the ESE method will have better sensitivity for eccentricity selection. We want to suppress the v2-driven background in CME searches which came from combinations of eccentricity and emission pattern fluctuations. You can arbitrarily increase the non-v2-driven background to show that the ESS method cannot remove these backgrounds. We agree with this and you do not need to make toy models to show that. What we stated was that neither the ESE nor the ESS method will be able to remove that kind of background. If you claim otherwise for the ESE method, you NEED to show it. You did not show that in your paper, yet you insisted on the claim. In real analysis, we use the spectator plane from the EPD to suppress those non-v2-driven backgrounds. 

  You also had a question about emission pattern fluctuation and you confused it with statistical fluctuation. By statistical fluctuation we refer to the variations in the number of particles involved. By emission pattern fluctuation we refer to the variations in the azimuthal angular distribution of particles while the number of particles does not change much. In our analysis of data with a small centrality window, the number of particles does not vary greatly. But the azimuthal angular distributions could vary which will give us the range of v2. So please do correct your description about this. If you read our method paper with an open mind, perhaps you would not continue to be confused between statistical and emission pattern fluctuations.

  Huan

  

  

On Wed, Jul 31, 2024 at 10:52 AM Wang, Fuqiang <fqwang AT purdue.edu> wrote:

Hi Huan,

Thanks for the message. Please see my reply inline below.

Best regards,

Fuqiang

---

From: Huan Huang <huang AT physics.ucla.edu>
Sent: Tuesday, July 30, 2024 5:35 PM
To: Wang, Fuqiang <fqwang AT purdue.edu>
Cc: Frank Geurts <geurts AT rice.edu>; Lijuan Ruan <ruan AT bnl.gov>; ptribedy AT bnl.gov <ptribedy AT bnl.gov>; Sooraj Radhakrishnan <skradhakrishnan AT lbl.gov>; subhash <subhash AT rcf.rhic.bnl.gov>; Huan Huang <huang AT physics.ucla.edu>; Gang Wang <gwang AT physics.ucla.edu>
Subject: Concerns for your posted paper

 

---- External Email: Use caution with attachments, links, or sharing data ----

Fuqiang,

  I had the opportunity to talk with several people at the Chirality 2024 conference. Your hastily posted paper 2407.14489 created more confusion and misunderstanding, and your conclusion is so biased that I had a hard time believing such sloppy work was from you. I am concerned that your paper would make the STAR collaboration spend so much effort to explain mistakes/deficiencies in your paper, instead of physics and limitations in our recent CME search results from the BES II. Here I list some major concerns.

As you know, I have been continuously pointing out over years that the ESS method is entangled and we do not know what the ESS intercept means. We have been trying over the past year to understand the ESS method more quantitatively by simulations.

1. In the paper and many presentations, you claimed that Event Shape Engineering (ESE, your favored method) is sensitive to dynamical fluctuations, and our Event Shape Selection (ESS) method to statistical fluctuations in event shape. This is your personal biased description because, as far as I know, we do not have a way in ESS to select statistical fluctuations alone without including the geometrical eccentricity contributions. “Dynamical” is a word used too often to describe something that you cannot identify well. In your description of “dynamical” here, do you mean “geometrical shape” as in eccentricity calculation or “longitudinal correlation/decorrelation” as in complicated time evolutions or both and many other effects? In any case, the cartoon in the ESS paper and in many of our presentations made it clear that the ESS method was intended to take both the geometrical eccentricity shape and the emission pattern fluctuation into account to select azimuthally round events to minimize the CME background. You chose to ignore our description and invented your own description of our method, which missed important points related to our method.

Event-by-event q2 (or any quantity) has large fluctuations, dominated by statistical fluctuations. This is true for both the ESS and ESE methods. In ESS, you select events by q2 and correlate Delta gamma with v2, both of which are calculated from the same particles of interest as for q2, so your v2 (which is closely related to q2) is also affected by statistical fluctuations. In ESE, Delta gamma and v2 are calculated from particles independent of q2, so the statistical fluctuations in q2 are canceled in v2. This is just a fact, mathematically proven. The variations in v2 is due to dynamical fluctuations in q2. By dynamical fluctuations, I mean all correlations due to physics in the underlying event, not due to statistical fluctuations. The dominate dynamical fluctuation source in this case is geometry eccentricity.

2. 
   
3. You failed to recognize that the CME background has contributions from both eccentricity-inspired hydrodynamic flow and emission-pattern-fluctuation-related v2. We always consider ESE a good method to select eccentricity shapes and very useful for hydro-inspired comparisons. It appears that ESE is not very effective in selecting round events with a finite magnetic field (mid-central collisions).

I'm not sure what emission-pattern-fluctuation is; I think you mean statistical fluctuation. If you concentrate on those "round" events, it is tricky: this was the idea of Quan Wang's thesis work and his PRC 2014 publication. The critique from people (also from your group) was that those round events were very biased so we don't know what was really measured. It's certainly true that those events are very biased, but as long as the background turns out to be zero, then the intercept would indeed be a measure of CME, albeit in those biased events only. We later realized [PRC 95, 051901(R) (2017)] that the problem was not because of the bias, but because the v2 of the background contributing sources (like resonances) do not equal to zero even when you require the v2 of all final-state particles to be zero (round events). This is because the v2 of all types of particles are positive in our events and therefore the background source v2's are doomed to be positive in those round events (we illustrate it in the middle cartoon of Fig.1 in the preprint). 

As I replied in 1), the ESE is proven to give zero intercept in the absence of CME. The ESS method is more complex and I believe it is somewhere in between the ESE method and Quan's method. It is so complex that I don't know what it measures, and that's why we try to understand it better and more quantitatively by simulations. 

3. In our ESS approach, we aimed at suppressing all v2-related backgrounds. The underlying driving dynamics can be modeled by hydrodynamic calculations. We removed protons in our CME correlator calculations because protons have contributions from the baryon transport mechanism which may not be described by hydrodynamic calculations. We used pure pions and kaons for our correlator calculations and only included particles of pT < 1.4 GeV/c. 

4. For AVFD Au+Au 200 GeV simulations, we can calculate the expected CME term <a1a1> from the model exactly.   The ESS method has proved to successfully reproduce the true signal a1^2 using the dg112 intercept, so we know that the ESS intercept is the true CME signal. However, the ESE intercept is systematically lower than the true signal (see Fig 9 of our method paper). Does this indicate that the ESE is not working accurately in removing the hydro bkg? You CANNOT ignore the accuracy and uncertainties in the measured numbers and draw your conclusions from your prejudice. 

As mentioned above, one can prove that ESE intercept removes all v2-induced background if v2 and q2 are completely decoupled from each other. When the eta gap is not large enough so that some nonflow correlations exist between v2 and q2, then one needs to carefully consider that, but the ESE method is conceptually clear and clean. On the other hand, it is not proven that the intercept in ESS is background free because everything is entangled with everything else. In principle, the ESS intercept can be anything, you do not know what you get. The AVFD closure test is accidental; we show that some other models fail, and we also show that the ESS intercept depends on all the details of the event. This is not hard to understand—the makeup of the event (resonance abundances, spectra, etc.) changes your q2, v2, and Delta gamma in different and entangled ways, your ESS intercept is going to change, it's not going to be always fixed at zero (in the absence of CME).

5. It is clear that both AMPT and EPOS models have a non-hydrodynamic driven background. So the ESS method could have slight over-subtraction or under-subtraction of the v2-related background. In our ESS method paper, we described the AMPT results and considered the difference between AVFD and AMPT. In your paper, Figs. A.3 and A.5 showed that the calculated background level with the ESS method does follow the expected v2 dependence. The residual intercepts need to be understood better. The EPOS model has a very different baryon-stopping mechanism, and perhaps you need to remove protons in the ESS method as we have done in our analysis.

We do not know what is in the real data. I am sure none of the models can completely describe the real data. 

6. When you apply the ESE method to AMPT and EPOS models as shown in Figs. A.4 and A.6, the dependence of Delta-g versus v2 is totally unexpected. We would conclude that the ESE method lost its sensitivity to the shape selection because of non-hydro driven v2. However, you insisted on the conclusion that ESE is a better method for CME analysis. That is not very convincing to us and is a textbook example of bias, intentionally or unintentionally.

As I said above, ESE is mathematically proven to be the right event-shape method. All LHC experiments used it.

7. We have nothing to comment on your Toy model simulations. Toy models should respect certain physics principles/boundary conditions. Often, it is more useful to learn why certain toy models do not work because of what dynamics missing in the model play important roles. If I could disregard physics and pick up whatever I wish to prove someone else wrong, I can easily make up a toy model to prove QCD wrong. I just point out that if you do not have hydro-driven v2 and correlations in your model, both ESE and ESS will equally fail in shape selection.

I think the burden is on you to prove ESS is the correct event-shape method for all possibilities of data makeup. 

8. Your “V. Summary” section is full of biased descriptions of ESS and ESE. Some of the non-hydro-driven dynamics in models will fail ESS and ESE in similar ways. You showed NO evidence that ESE can work better than ESS in any of the cases where you criticized the ESS results. In Figs 5, 6 and 7 of your paper where you showed the intercept values from ESS and ESE method for all the models, the values from ESS and ESE agree within the statistical errors though the errors are much larger from the ESE method. Yet it does not prevent you from highlighting your belief that the ESE would be able to remove CME background and, unbelievably, you further postulated in Fig. 8 what the ESS method would yield without giving any scales. I can only conclude that with the garbage in and garbage out process, indeed any figure/number is possible in simulations. Please note we have a measurement of ESS result for 200 GeV Au+Au data, and toy models should at least attempt to match the measured features. or models can already be falsified by comparing with the 200 GeV data.

The conclusion is that ESS is murky and highly model dependent, and ESE is clean but more statistically challenging. I don't see any bias. That Fig.8 doesn't have scales is exactly because one does not know those scales—it illustrates the point that it can be negative, zero, or positive, depending on the details of the events.

  It will take the collaboration too much effort to clear the confusion from your sloppy work and biased conclusions. I would like to request that you withdraw the current version of the paper and you may update with a new version in the future if you have enough simulation data to support your ESE method. The paper as it stands now should be an embarrassment to you, a waste of STAR time to deal with it.

 There is nothing that I will do if you decide to reject my request. We will not waste our time to write a paper to rebut your claims in the paper. It should be obvious to most experimentalists though we may need to point out your inconsistencies to theorists at conferences. It is not a STAR policy issue either.  STAR policy is not meant to prevent anyone from posting sloppy work and drawing unscientific conclusions in their personal papers. My request was just one from a colleague who is not prepared to waste too much effort to deal with the confusion created by your paper. I hope that you at least think about the concerns we have raised here and do the right thing.

  Huan 

I'm disappointed that you used strong words like "sloppy", "embarrassment", "unscientific". I don't know what inconsistencies you mean that I have. 

Since you mentioned STAR policy, it is rather unfortunate that scientific discussions/debate are sometimes suppressed in STAR. There have been several cases in STAR that erroneous results or overclaims are published that I think is very damaging to STAR's credibility. I'm, for one, exercising my scientific consciousness.

Best regards,

Fuqiang

<2407.14489v1 15.pdf><2407.14489v1 17.pdf>