STAR HardProbes PWG

[Sooraj Radhakrishnan <skradhakrishnan AT lbl.gov>]

Hi Leszek,

   Thanks for preparing this well written and clear analysis note and for the paper draft. Sorry for the delay in sending the comments. Please find some comments/questions from me on the analysis note below:

p 12: Do you know what goes into the definition of 'flag' variable? How are the flag > 1000 and flag < 0 tracks defined? Is it something commonly used in p+p analyses?

p 12: would be good to include the definition of TofMatchFlag

p 12: the drop in TofMult vs BBC rate is attributed to occupancy. But the refmult values at these BBC rates are small, comparable to very peripheral Au+Au collisions. In Au+Au, the occupancy effects show up in central events. Is there another reason for the drop in TofMult?

p 12: This may not be the best way to evaluate if there is impact. The trigger requirements will boost number of events with Y. So a better estimate would be, with the production rates, whats the probability to have an Y in your sample after the selection cuts

Table 2.4: Is there a reason to reduce the nHits cut for TofMult definition?

p14: Is the R_SMD cut used in other electron analyses as well?

Fig 2.7,2.8: In all pT bins, the points around the 1S and 2S peaks fluctuate above the fits. Also the right side of the peak is below the fit in several cases. Is this a systematic effect, on background evaluation or signal shape? Would this have an impact on yield ratios? 

p17: Is the PYTHIA for the combinatorial background without any detector effects? Will there be any impact from the efficiency or resolution? Are the same electron kinematic cuts applied on PYTHIA as the data?

p19: Are the uncertainties on the fit parameters considered when doing the purity and efficiency corrections for the signal extraction?

p26: Section reference missing

p34: Could you give some details/links to the TOF embedding. Remember this was one part that went through a few revisions 

Fig.2.27: Sorry, I am a bit confused here. The analysis is using the triggered sample. The red distribution is the multiplicity distribution for Y events.  Are the final ratios relative to the red or the blue (MB) distribution?

Sec 2.6.1: As pointed at above, there is a fluctuation of the data points above the fit near the peak and balancing of it on the right side through fluctuations down, This small uncertainty could be coincidental 

Sec 2.6.4 This is much smaller than the 5% uncertainty typically quoted for tracking efficiency. Is this the value commonly used in p+p analyses?

Sec 2.6.14: Is this variation of the prior distribution in the unfolding? What is the variation studied here?

p 47: figure references missing

p 48: figure references missing

Sec 3.1.6 What physics do we learn from the xT scaling? Could you add a sentence or two? 

thanks

Sooraj

On Fri, Aug 12, 2022 at 2:12 PM Barbara Trzeciak via Star-hf-l <star-hf-l AT lists.bnl.gov> wrote:

Hi Leszek,

Sorry for the late reply.

Please find my comments/questions to the Analysis Note below.

Also, do you have checks on distributions that Rongrong asked ?

Cheers,

Barbara

--- Comments on the Analysis Note ----

- Sec 2.3.2: it would be good to add nSigmaE (and some projections) and E/p plots to see where your cut values sit w.r.t. to the distributions. 

- Sec. 2.3.3: it's not clear from this section that you form a pair and then one electron candidate is required to satisfy the trigger conditions (you only write that at least one track from each event has to satisfy DSM adc > 18 condition) and the other is the partner electrons for which you require pT > 1 GeV. Please make it more clear.

- Signal extraction: 

   - it would be good if you add an appendix (or include a link) to all the invariant mass fits, in all the pT, y and multiplicity bins.

   - You should mention somewhere and define the pull distributions that you also have. 

   - Correlated bkg: how did you define the fit function ?

- Sec. 2.5.1

   - Missing reference to the embedding section

   - Table 2.7: are those cuts only for the partner electron or for both electrons in a pair ? Are the track quality cuts the same as before used here ? 

   - It would be good to add the invariant mass distribution of unlike and like-sign pairs to judge the purity of the sample.

- Sec. 2.5.2

   - You assume a Gaussian shape of the y distribution ? I assume so from the first term of Eq. 2.5 (that I guess is for rapidity), but it's not explained in the text.

   - The fit was used to extrapolate the obtained values - do you mean the log fit from the previous sentence shown in FIg. 2.15b ? It's for Upsilon (1S), how do you get extrapolated values of the other states ? Also, can we determine the slope parameters with such precision and don't they have some uncertainties ?

- Sec. 2.5.3

   - Do you calculate efficiencies as a function of MC or reconstructed electron (Upsilon)  pT ? Please clarify in the note.

   - Why do you need single electron efficiencies if you later calculate efficiencies directly for different Upsilon states ?

- Sec. 2.5.4

   - Why do you start here from the level of L0 cut ? Also, does L0 include the pT cut for the partner electron ? Please clarify.

   - Explain how you calculate acceptance. 

   - nSigmaE cut is calculated from data for single electrons. Please explain here how you obtain nSigmaE efficiency for Upsilon. And do you consider rapidity dependence ?

   - Please also explain in the text each efficiency contribution that you present on the plots. 

   -  Most of the efficiencies are calculated based on the embedding. There should be added comparison of distributions from embedding and data (nFitPts, dca, E/p, Adc ...) so that one can judge how well the simulation described the data. Have you e.g. compare E/p efficiency from data and embedding ? I don't see any sys. unc. assigned to E/p.

   - Differences between efficiencies for different Upsilon states are quite big. Is it expected, why the trigger efficiency would differ so much ? It's hard to judge difference for other sources as they already contain L0 efficiency. Could you please add efficiencies for single sources, i.e. L0, Acc, Track etc. (not L0*Acc ... ) and compare them between Upsilon states.

- sec. 2.5.5

   - Is the a parameter the same for the whole pT range ? I think there can be some pT dependence of it. Is the chi^2 minimization done for the pT-integrated signal or vs J/psi pT, also what is the red line in Fig. 2.20b ?

- sec. 2.5.7

   - Do you assign any uncertainty related to the assumption of the power law shape ? 

- sec. 2.5.8

   - How did you determine 4 interactions for the unfolding and what did you use as a prior ? 

   - Do you vary the prior shape as part of your systematic unc. ? In your unfolded range you go beyond what is measured and therefore you're probably sensitive to the underlying prior shape, so you depend on Monte Carlo simulations here. 

   - Fig. 2.27a: what is the black distribution ?

- sec. 2.6.1

   - The fit also includes fit to the uncorrelated and correlated backgrounds which influence the signal extraction. It would be good to have some sys. unc. related to these fit parts of the fit.

- sec. 2.6.6

    - It's not explain how the 3% shift is taken into account and how the final unc. is estimated ?

- sec. 2.6.10

   - Could you please provide more detail here. It's not clear to what exactly you refer from Ref. 3.

- sec. 2.6.12

   - Is the reconstruction efficiency applied before or after unfolding ? 

   - The efficiency has quite a large uncertainty in the last TofMult bin. Is it due to low statistics in the embedding ? If so, isn't it a problem for creating a response matrix ?

- sec. 2.6.14

   - Which distribution is fitted with NBD ? Would be good to add these fits.

- sec. 2.6.15

   - What is the default setting ?

- sec. 2.6.16

   - What is "Fit" in the tables. Is it signal extraction unc. ? Please make it more descriptive. 

- sec. 3.1.2

   - How do you obtain the 2S+3S signal, i.e. at which level do you add the signals: at the raw level or after applying corrections ? If at the raw level, how do you calculate the combined efficiency ?

- sec. 3.1.3

   - "The cross section for the 1 < y < −0.5 and 0.5 < y < 1 is added together in 0.5 < |y| < 1 and reflected symmetrically.": I'm not sure if I understand - do you perform fits separately for 1 < y < −0.5 and 0.5 < y < 1, calculate cross sections and then add the cross sections or you extract the yield by adding 1 < y < −0.5 and 0.5 < y < 1 bins and then calculate one cross section ? It sounds like the former but I think the latter makes more sense. 

   - It would be good to add a plot of efficiency vs rapidity for each Upsilon state - in sec. 2.5.4. It will be easier to compare values in the two rapidity ranges. Could you also compare |y| < 0.5 and 0.5 < |y| < 1 efficiencies for single sources, i.e. L0, Acc, Track etc. (not L0*Acc ... ).

   - Figure numbers are missing in the result sections. 

- sec. 3.1.4

   - Do you correct for BR ?

   - "The STAR data are more consistent with the values up to pT = 6 GeV/c" - this is not clear to me

- sec. 3.1.5

   - " A pT dependence is observed and is indicated in the shift of ..." - you claim here a pT dependence, how many sigmas from the linear fit and from the data point for pT > 0 GeV/c, this last red point is ?

   - "This gives a hint of collective effects, which may be present at high multiplicity in p + p collisions" - what exactly points to collective effects ?

   - Some figures, like 3.7 right and 3.12 seem not to be discussed in the text. 

  

- It would be good if you added to the AN the most recent checks and discussions that we had at the PWG meetings.

On Fri, Aug 12, 2022 at 5:51 AM Yi Yang via Star-hf-l <star-hf-l AT lists.bnl.gov> wrote:

Dea Leszek,

(I added this to the HP list, probably we can discuss it there.) 

Sorry again for the late response. 

I have some comments/questions on your analysis note and I will send out my comments on your paper draft in a separate email. 

 

- Abstract: remove all "J/psi" related things. 

- p12: Any references for the "flag" information or definition? 

- p12: Should include the eta range in the second paragraph.   

- p13: Move Figure 24 to Figure 23 (since the current Fig. 23 was mentioned later). 

- p14: You mentioned several variables: R_SMD, E_TOW/E_CLU, E_CLU/p, ... how did you choose these cut values? Can you show some studies or references? 

- p17: The fitting of the Upsilon signals. Since STAR doesn't have a good resolution to separate Upsilon states, it is very tricky to fit the signal. You can take a look at how ATLAS did (they have similar mass separation power), just for your reference.  https://arxiv.org/pdf/1211.7255.pdf  (Section IV C.)

- p19: It is still strange to me that you do a bin counting from a range and then move the contaminations from other states from the fit. Why do you just integrate from the fit? (I understand you use this as the systematic, but the procedure is a bit strange to me.) 

- p23: Figure 2.10 (b): why the uncertainties on 2S and 3S only have up-forward or down-forward uncertainty? 

- Could you please provide the full function used in your fit? 

- All mass plots (especially for the plots used in paper), I would think the style can be improved a bit: all the font sizes are too small, the bottom panels can move up and combine to the top panel (share the same x-axis). Since you are using unbinned-likelihood fit, you can make the plots with coarser binning. Difficult to read the label on the y-axis in the bottom panels. 

 - p26: First paragraph. Section ??

 - p26: Table 2.7: invariant mass cut < 0.04 GeV/c^2 --> Should it be 0.3 GeV/c^2?  I think other heavy-flavor electron analyses used 0.2 GeV/c^2. Can you add the mass plot of your photonic electrons in the note? 

 - p29: Electron efficiency:  

          - Figure 2.16: there is an "acceptance" in the electron efficiency, can you define it? Naively, I thought the electron efficiency should only be calculated inside the TPC acceptance. 

          - Do you expect the efficiencies for electrons and positrons to be different? If not, why don't you combine them? 

 - p29: Upsilon efficiency: 

          - How do you define acceptance? Can you add it in the note? 

 - p30: The main concern I have is in the result of cross-section vs y and I remember we discussed intensively and the reason is that you have similar raw yields for 2S and 3S in |y| < 0.5 and 0.5 < |y| < 1, but the efficiencies has a factor of 2 difference (Fig. 2.17 (b) and (c). Do you know why L0 efficiencies are different in two rapidity ranges? I remember Ziyue and others found that the n_sigma_e has some dependency on eta. Could you please also check it on your dataset? 

 - p31: Did you also consider "shifting" the mean of the track resolution?   

 - p39: Systematics: Please add more details of your systematic uncertainty study. It would be good to have all the plots included. So far, you only have a short description and the final number for each source. 

 - p45: Chapter 3: Lots of links to Figures or Sections are missing, for example p47, Figure ?? and Figure ?? in the second paragraph. 

 - p50: Figure 3.5 (b): I am still worrying about the "dip" in 3S state, it seems that there is a ~3 sigma effect from "flat". We probably really need to understand it before we release this result, since it might cause lots of discussion from the theorists' side. (I am not saying it is wrong, but we will need some understanding and explanations. 

 - Similar comments on the results plots, the style probably can be improved. (But this can be done in the GPC stage.)

 - Figure 3.4 and 3.5: I still don't understand why we need to mirror the points to the negative rapidity. It won't give us any more information, you can just simply change the x-axis to |y| and plot to data points. 

 - You can make all your plots much larger. 

 - p56: NLO [?]

Cheers,

Yi

+++++++++++++++++++++++++++++++++++++++++++++++++++
Yi Yang, Associate Professor
Department of Physics
National Cheng Kung University
Tainan, 701 Taiwan
E-Mail: yiyang AT ncku.edu.tw
Tel: +886-6-2757575 ext.65237
Fax: +886-6-2747995
Group Web: http://phys.ncku.edu.tw/~yiyang
+++++++++++++++++++++++++++++++++++++++++++++++++++

On Wed, Feb 16, 2022 at 11:03 AM Leszek Kosarzewski via Star-hf-l <star-hf-l AT lists.bnl.gov> wrote:

Dear All

As I presented today, the Upsilon states in p+p 500 GeV paper is ready for review.

Paper page:

https://drupal.star.bnl.gov/STAR/blog/lkosarz/upsilon-production-pp-500-gev

Paper:

https://drupal.star.bnl.gov/STAR/system/files/Upsilon500_Paper_LK_v6.pdf

Technical notes:

https://drupal.star.bnl.gov/STAR/system/files/Upsilon500_TechNotes_LK.pdf

I also updated the presentation to answer today's questions (see slides 24-29):

https://drupal.star.bnl.gov/STAR/system/files/Upsilon_HF_LK_2022_2_15.pdf

The tracking efficiency uncertainty is actually correct. It affects the shape of N_ch distribution and also the shape of Upsilon N_ch distribution through unfolding correction. For Upsilons also the input distribution is affected by the efficiency from embedding and signal extraction. In the simulation for unfolding correction the uncertainty is estimated by changing the efficiency by +/-5%. For the Upsilon signal, the nFitPts is changed by +/-2 in embedding and data. The cumulative effect of all of that is large for the last Nch bin. Also if the shapes are changed this means that the effect on Nch/<Nch> will not cancel out enirely, though the effect is lower than the input +/-5%. In addition the bins are fixed with default <Nch>. The J/psi paper shows similar values for these uncertainties.

The 4Cx uncertainty also affects the data in this way. Upsilons are affected by the change in both Upsilon and min-bias Nch distributions, so again the effect is larger on Upsilons.

I also checked the eta and y distributions and efficiencies. There is a dip for triggering track, but this is not shown in y distribution of Upsilon candidates.

Best regards, Leszek

_______________________________________________
Star-hf-l mailing list
Star-hf-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/star-hf-l

_______________________________________________
Star-hf-l mailing list
Star-hf-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/star-hf-l

_______________________________________________
Star-hf-l mailing list
Star-hf-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/star-hf-l

--

Sooraj Radhakrishnan

Research Scientist,

Department of Physics

Kent State University

Kent, OH 44243

Physicist Postdoctoral Affiliate

Nuclear Science Division
Lawrence Berkeley National Lab
MS70R0319, One Cyclotron Road

Berkeley, CA 94720

Ph: 510-495-2473

Email: skradhakrishnan AT lbl.gov