STAR task force for tuning PYTHIA

[Raghav Kunnawalkam Elayavalli <kunnawalkamraghav AT gmail.com>]

This is very interesting Zilong.

Normally i would say that we use the latest PDF but like you said i think it would be very interesting to see if the parameters we tune are dependent on the input PDFs…. 

I think thats something that we should discuss at our next meeting. 

Cheers

Raghav 

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

Raghav Kunnawalkam Elayavalli

email: raghav.ke AT yale.edu 

Associate Research Scientist  

Yale University and 

Brookhaven National Lab

<he/him/his>

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

On Apr 24, 2021, at 11:32 PM, Chang, Zilong <zchang AT bnl.gov> wrote:

Hi Raghav,

Thanks for the information. Indeed when I switched to CTEQ6L1, the PYTHIA8 produces a similar shape as PYTHIA6. The larger bump at jet p_T less than 10 GeV could be due to that I used the default cutoff p_T,0 without any modification. Also surprisingly PYTHIA 8 also reproduces my preliminary results well. From this observation, it’s very promising we further tune p_T,0 to an optimal value in order to have better matching.

https://drupal.star.bnl.gov/STAR/system/files/Pythia8tune_pp510.pdf

I’m not sure how this works in RIVET. But if you would be able to implement the changes and see how PYTHIA 8 reproduces our other data at 200 GeV. That would be an interesting thing to look into. If all works well, we may settle with tuning PYTHIA8 parameters (based on Montash tune) with CTEQ6L1 as input. I’m not sure how exactly PDF plays a role in PYTHIA tune, but this could be up to debate. I have a few parameters in mind to tune ISR and FSR overall strength in PYTHIA 6, but I have to find their counterpart in PYTHIA 8.

Zilong

On Apr 24, 2021, at 1:17 PM, Raghav Kunnawalkam Elayavalli <kunnawalkamraghav AT gmail.com> wrote:

Hi Renee and Zilong, 

The difference in the x-secs between pythia6 and pythia8 is something that we also observed and my understanding is that its primarily due to the PDF thats used.  see slide 6 here https://drupal.star.bnl.gov/STAR/system/files/jetcorr_embedding_update_raghav_11Feb2021_v0.pdf especially the inset plot which shows the ratio of jet spectra for pythia8 vs 6. As you said, i think this is primarily producing the shape differences in jet spectra and this cannot be explained by differences in jet shower algorithm etc… 

Cheers

Raghav 

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

Raghav Kunnawalkam Elayavalli

email: raghav.ke AT yale.edu 

Associate Research Scientist  

Yale University and 

Brookhaven National Lab

<he/him/his>

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

On Apr 24, 2021, at 11:57 AM, Chang, Zilong via Star-tf-tunepy-l <star-tf-tunepy-l AT lists.bnl.gov> wrote:

Hi Renee,

Yes I’m also surprised to see that the partonic cross-section from PYTHIA 8 is close to PYTHIA 6 at low partonic p_T, but larger at high partonic p_T. The difference could almost make up the difference in the jet cross-section at high p_T. I looked at the Pythia 8.2 webpage, and it says the default factorization and normalization scales are different between PYTHIA 8 and PYTHIA 6. I tried to set the scales to match PYTHIA 6 but the PYTHIA 8 cross-section remain unchanged.

mode  SigmaProcess:renormScale2   (default = 2; minimum = 1; maximum = 6)

option 3 : the arithmetic mean of the squared transverse masses of the two outgoing particles, i.e. (mT_3^2 + mT_4^2) / 2 = pT^2 + 0.5 * (m_3^2 + m_4^2). Useful for comparisons with PYTHIA 6, where this is the default. 

mode  SigmaProcess:factorScale2   (default = 1; minimum = 1; maximum = 6)

option 3 : the arithmetic mean of the squared transverse masses of the two outgoing particles. Useful for comparisons with PYTHIA 6, where this is the default.

http://home.thep.lu.se/~torbjorn/pythia82html/Welcome.html

I don’t know if it is a sqrt{s} scaling to cause this difference, I’m also suspecting it could be due to PDF difference (NNPDF2.3 vs CTEQ6l1). I’m trying to test this.

Zilong

On Apr 23, 2021, at 11:46 PM, Fatemi, Renee <rhfate2 AT g.uky.edu> wrote:

Hi Zilong,

This is really interesting. Clearly we need to tune away these shape changes. What I am really surprised by is the huge shift in absolute Xsec from PYTHIA 6->8.  I wonder if this is also due to a sqrt(s) scaling?

Thanks,

Renee

On Fri, Apr 23, 2021 at 5:24 PM Chang, Zilong via Star-tf-tunepy-l <star-tf-tunepy-l AT lists.bnl.gov> wrote:

Hi guys,

I used the PYTHIA8.2.35 default tune installed under StRoot (first link below) to generate the inclusive jet cross-section and compared it to the run12 pp510 GeV preliminary results. You can find my results at the second link below. The PYTHIA8 seems to under-predicate the measured cross-section at low jet p_T, but over-estimate it at high jet p_T.

https://www.star.bnl.gov/cgi-bin/protected/viewvc.cgi/cvsroot/StRoot/StarGenerator/Pythia8_2_35/

https://drupal.star.bnl.gov/STAR/system/files/Pythia8tune_pp510.pdf

Thanks,

Zilong

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

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