STAR HardProbes PWG

[Andrew Tamis <andrew.tamis AT yale.edu>]

Hello Rongrong,

That is a good point that this method reworked that way would get absolute yields correct.  As you mentioned, the response matrix is very diagonal so this is most likely why it closed initially, and the physics messages i draw did not change between my preliminary and new method.  Because we want to draw conclusions on the substructure changing with pT, it is best to move away from this method going toward publication.  I will take a look into making the toy that you mentioned, i think it will be a good check that we are now avoiding these issues with multi-dimensional unfolding.

Best,

Andrew

On Wed, Nov 8, 2023 at 5:03 PM Ma, Rongrong <marr AT bnl.gov> wrote:

Hello Andrew

Thanks for adding those two extra slides explaining the conceptual issue you discussed about. Thinking a bit more on this, I feel there are probably two issues:

i) the formula used on s19 to calculate b_corrected is incorrect

ii) a dilute effect, but I am not sure if it is same as what you mean on s20. 

On s19, if you take the numbers in the table as yields, you will get b_corrected  = 11 using the formula while there are only 10 b_t jets. The correct procedure should be that you calculate the probability of each measured jet bin coming from true jet bin b, and multiply that with the measured jet yield. i.e. b_corrected = 20%a_m + 60%b_m + 20%c_m. Then if you plug in the measured jet yields, you will get b_corrected = 10 = b_t. Of course, this only woks of you know the truth spectrum shape, and is essentially a Bayesian unfolding with iteration = 1. I think what a full-blown unfolding procedure buys you is a weaker dependence on the prior. 

Another difference between your current approach and a full-blown multi-dimensional unfolding is that you assume the substructure of a measured jet at given pT is the same no matter at which pT it comes from, which is probably not true. In a proper multi-dimensional unfolding, it takes into account different shapes at different jet pT. This effect will not be significant if the response matrix is more or less diagonal, and the substructure does not vary much with jet pT, which is probably your case. If you are interested, you can design a toy MC where the substructure varies dramatically with jet pT, and see if you can get closure with different methods. 

Best

Rongrong 

On Nov 2, 2023, at 10:59 AM, Andrew Tamis via Star-hp-l <star-hp-l AT lists.bnl.gov> wrote:

Hello All,

As mentioned in the meeting i have uploaded a version of the slides with the example of the conceptual error in the old method to this blog page.

Best,

Andrew

On Thu, Nov 2, 2023 at 9:10 AM Andrew Tamis <andrew.tamis AT yale.edu> wrote:

Please find my slides for today's meeting attached

PWG Update 11_2_23 | The STAR experiment (bnl.gov)

Best,

Andrew

On Wed, Nov 1, 2023 at 2:30 PM Andrew Tamis <andrew.tamis AT yale.edu> wrote:

Hello All,

I would like to present an update on my EEC Measurement in pp Run 12 data and the associated paper tomorrow.  I will send my slides beforehand.

Best,

Andrew

On Mon, Oct 30, 2023 at 1:55 PM Mooney, Isaac via Star-hp-l <star-hp-l AT lists.bnl.gov> wrote:

Hi all,

We will also have our regular HP-PWG meeting this Thursday (November 2) at 10 AM EDT.
Please let us know if you want to present and discuss your analysis or any other business.
Please also send your presentation slides a few hours before the meeting.

HP working group weekly meeting info:
https://drupal.star.bnl.gov/STAR/pwg/Hard-Probes/Weekly-HP-PWG-meeting

Join ZoomGov Meeting
https://bnl.zoomgov.com/j/1611419615?pwd=VW1hNm43ZDd5d2EvK2R4aEJsQ2ZNZz09

Meeting ID: 161 141 9615
Passcode: 744968

Thanks,
Isaac, Nihar, and Yi

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

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