STAR HardProbes PWG

[Gabe Dale-Gau <gdaleg2 AT uic.edu>]

Hi Helen,

I'm glad this follow up helped! I will not plan on presenting in the meeting tomorrow morning unless anyone has further questions. I will attend as always so if anyone has further follow up just let me know.

One minor correction to my explanation from yesterday: the leading jet requirement in pseudo-embedding is applied only after after embedding, not both before and after as it reads now. We inject all jet seeds found in p+p to Au+Au events, and if the leading jet after embedding matches a p+p jet seed, then that jet is accepted. for example, if a two-jet p+p event has jet1: 7 GeV and jet 2: 5 GeV, we embed the entire event, if after embedding the resulting jet 1 has 8 GeV and jet 2 has 10 GeV, jet 2 is accepted as the leading jet.

Thanks,

Gabe

On Tue, May 21, 2024 at 6:58 PM Helen Caines <helen.caines AT yale.edu> wrote:

Thanks Gabe, I think I have understood now and indeed it seems you have a complete separation from the background you infer from the pseudo-embedding and that from mixed events.  Nice study

Helen

***********************

Yale University
Physics Dept. - Wright Lab.
PO Box 208120
New Haven, CT 06520
203-432-5831
***********************
she/her/hers
"Life is not about waiting for the storms to pass.
It's about learning how to dance in the rain." - Vivian Greene

On May 21, 2024, at 5:01 PM, Gabe Dale-Gau <gdaleg2 AT uic.edu> wrote:

Hi Helen,

Thank you for your attention to this method, and the follow up questions.

I have provided a few more responses below. We can still discuss in the meeting this week if you (or others) would like, but I don't have much more to add beyond what we have gone over in writing, and maybe email is a better way to lay out our logic on this cut.

Let me know if you have any further thoughts on this matter.

Best,

Gabe

This I now get, indeed 1 track with 9 GeV (you jet cut-off I assume) can be taken as having to have come from a hard scattering. Its a functional definition that I suspect most/all can agree with.  I think you also stated that there’s a high pT cut off for tracks you use in the mixed events. Do I remember correctly and what cut-ff is that?

 

 

Yes, I had a cut in place preventing any track above 9 GeV on it’s own from entering the mixed event. This cut was redundant. I removed it today and ran the mixed event again with all tracks above the constituent minimum, and produced essentially the same result (0.02% difference). The nTracks in-jet cut alone is sufficient to remove 1 and 2 track jets.

 

 

Doesn’t this  assertion depend on how low in pT you define a hard scattering i.e. what pT you inject PYHTIA events down to,  and what’s the lowest pT track you put into the mixed events? Do you inject PYTHIA “jets” down to a single track with pT  = 4.5 GeV? And mix event with only 4.5 GeV and below?

 

We inject p+p jets all the way down to our constituent pT minimum. The caveat being that we only inject leading jets (highest pT in event), and then they must be matched in phi and eta space to the leading jet post-embedding to be accepted. We made a mistake when we first developed the pseudo-embedding procedure related to this point. At first we assumed that the only significant contribution to pseudo-embedding would come from jets that pick up a single extra track from Au+Au background. So initially we embedded p+p jets with jet pT > 6 GeV. Last year Joern pointed out that there is significant contribution from fluctuations involving multiple low pT tracks, so we adjusted our method to inject p+p jets all the way down to the constituent minimum, and found that there is indeed a contribution from low pT jets that pick up multiple tracks from Au+Au background. I have included the plot of pseudo-embedding jet spectra before and after embedding to demonstrate this point. This plot has identical counts between the two spectra, as each jet must be matched to appear in the sample. you can see that in this particular example, p+p jets are embedded starting as low as 2 GeV.

Our aim here is to design these two methods to be separate by construction, rather than to separate them by a pT cut that represents hard scattering.

<pseudo_Spect_Comp_Linear.gif>

On Tue, May 21, 2024 at 7:13 AM Helen Caines <helen.caines AT yale.edu> wrote:

Hi Gabe,

  Thanks for putting in writing what you were telling us last Thursday, having the chance to chew over the explanation has helped me understand your logic. I have a couple more questions below. You can feel free to wait until Thursday to answer them, but I wanted to write them down while I was thinking about it.

Helen

 Compare rate at which jets are found per event to the same rate in data
Q: Do you have any comparison plot?

 

This was Helen’s question in the meeting yesterday, and it is a good question. I plan on presenting again next week with a few more plots to explain our logic for the choice of the 3 track cut.

For now I will state our general thinking again:

Pseudo-embedding and Mixed Event methods aim to capture two different sources of correlated background. Pseudo-embedding looks for additional background signal that gets pulled along due to upward fluctuation around a real jet signal. Mixed Event looks for completely combinatorial objects that appear as jets on their own. While these are two distinct sources, there was originally some overlap in the resulting corrections. The 3 track minimum for combinatorial jets aims to eliminate this double-counting. The two cases this eliminates are 1 and 2 track jets.

1 track jets (tracks above 9 Gev/c) are real jets according to our definition. Thus, they must be excluded from a study aiming to identify fake jets.

This I now get, indeed 1 track with 9 GeV (you jet cut-off I assume) can be taken as having to have come from a hard scattering. Its a functional definition that I suspect most/all can agree with.  I think you also stated that there’s a high pT cut off for tracks you use in the mixed events. Do I remember correctly and what cut-ff is that?

2 track jets are accounted for in the pseudo-embedding correction. If a 1 track jet is found in p+p and embedded into Au+Au background, it has no correlation with any other track in the event, so if it forms a 2 track jet that reaches our 9GeV threshold, this is identical to a combinatorial 2 track fake jet. Thus we must exclude these from one of the corrections to avoid double-counting.

Doesn’t this  assertion depend on how low in pT you define a hard scattering i.e. what pT you inject PYHTIA events down to,  and what’s the lowest pT track you put into the mixed events? Do you inject PYTHIA “jets” down to a single track with pT  = 4.5 GeV? And mix event with only 4.5 GeV and below?

3 Track jets are where the corrections become distinct. If a 2 track jet is found in p+p and embedded into Au+Au background, it retains association between the two original tracks, forming a jet cone on its own. If it picks up a third track from Au+Au background, this is exactly what we are looking for with pseudo-embedding. It is distinct from combinatorial jets due to the presence of a “real” jet.

The same holds true for all >3 track jets. This is the regime where collections of low p_T tracks can form fake, fully combinatorial jets.

Agreed