STAR task force for evaluating tracking efficiency uncertainty

["Ma, Rongrong" <marr AT bnl.gov>]

Hello Gene

Looking at this resolution plot for primary tracks, the p1 value is about 0.003 between 1-3 GeV/c, which is compatible with embedding. So the numbers in your previous email kind of make sense now, as they are for global tracks while what I have been showing in my slides are for primary tracks which are expected to be much better. Am I missing something? 

So the Run18 study shows that the momentum resolution of global tracks in cosmic ray is similar to that of primary tracks in data. Is this expected?

Best

Rongrong

On Oct 20, 2022, at 2:47 PM, Van Buren, Gene <gene AT bnl.gov> wrote:

Hi, Rongrong

On Oct 20, 2022, at 1:56 PM, Ma, Rongrong <marr AT bnl.gov> wrote:

Hello Gene

Thanks for following up on this, and providing the additional information. 

My momentum resolution is the relative resolution, and therefore the "a" value is equivalent to p1. 

I am a bit surprised to see that the momentum resolution from MC with HFT has a p1 value of 0.008. Are those primary tracks with HFT hits?

No, that plot was labeled "Global". Here is the plot for primaries, which of course is much better:
https://www.star.bnl.gov/~jwebb/2016/BaseQA_April_13/evals1/pidPtiRPr_zy_1.png

In single particle embeddings, we have constantly seen p1 values of ~ 0.0035 over the years, which we know is way too good compared to data. 

Odd, as the S&C Team's QA did not show it to be that good. Jason Webb may have more such plots from over the years: the code to generate them was originally Yuri's for track-by-track comparisons, and I think Jason modified it for QA of Monte Carlo by setting one set of tracks to be the true simulation quantities.

Do we have a back-of-envelope estimate of what the TPC momentum resolution at 1 GeV/c should be? 

Yes, the colored lines on the first two plots of this page I sent earlier were generated via a model created by Jim Thomas:
https://drupal.star.bnl.gov/STAR/subsys/tpc/perf/tpc-pt-and-dca-resolution
The first plot assumes "nearly all" 45 padrows have hits, while the second plot has hits on every other padrow (i.e. half the hits). Reality is expected to fall somewhere between the two cases. 

The blue lines on the plots are for full field global tracks, and the coefficient is 0.009 in the first case, and 0.013 in the second case, as written on the web page. The relative momentum resolution at pT = 1 GeV/c appears to be about 0.013 in the first case and about 0.017 in the second case, from looking at the plots.

I think we can start with Run11 and Run14 cosmic ray data, and decide later on whether we need to produce Run19 cosmic ray. 

DataCarousel requests for Runs 11 and 14 are already submitted. Run 11 will appear on DD and accessible via FileCatalog queries. The Run 14 data was not catalogued (special production) and I am restoring them to a few locations on NFS. I'll let you know when they're complete.

Thanks,
-Gene

Best
Rongrong

On Oct 20, 2022, at 1:14 PM, Van Buren, Gene <gene AT bnl.gov> wrote:

Hi, all

On Oct 20, 2022, at 8:38 AM, Ma, Rongrong via Star-tf-trkeff-l <star-tf-trkeff-l AT lists.bnl.gov> wrote:

Hello Robert

Thanks a lot. The new fit looks very nice, and the p1 value is very close to what we have seen based on Jpsi width from Run18 Isobar data.

So the fit is actually p0 + p1 *pT, and the finding is that p1 is 0.0075.

If we focus on the high pT region, then I think we can all agree that the relative resolution of pT (the absolute resolution of pT, divided by pT) is equal to the relative resolution of (1/pT), and that these rise linearly in pT. This relation can be derived from the understanding that the TPC measures points in physical space with a generally pT-independent spatial resolution, such that the absolution resolution of (1/pT) is generally constant at high pT. That constant is p1:

dpT / pT = d(1/pT) / (1/pT)
=>
relative resolution = p1 / (1/pT) = p1 * pT

Rongrong's study has me a little confused. Is his momentum resolution the relative resolution, or the absolute resolution? If it is relative resolution, then his "a" parameter is the same as p1, determining the high pT behavior of the resolution.

Past studies have shown (see the links I sent)  that p1 for tracks without a primary vertex in their fit get values near 0.01.  The cosmic ray study I did got something like 0.014. I managed to find one of the Monte Carlo QA plots from 2016 when the S&C team was trying to get HFT tracking working:
https://www.star.bnl.gov/~jwebb/2016/BaseQA_April_13/evals1/pidPtiRGl_zy_1.png
In that plot we can see a rise of ~0.008 between pT = 2 GeV/c to 3 GeV/c, so perhaps as good as 0.008 with the HFT.

Robert's result of 0.0075 manages to get close to that even without the HFT. And Rongrong's past results of 0.0054 are much better, and his 0.0035 number from embedding blows that away. I don't see how we could have ever had it that good.

So, I'm confused about these numbers and findings. If someone understands something I've missed, please let me know.

So we have a good consistency here. I really appreciate if you can help analyze other datasets (Run11, 14, 19) as well. Hopefully, this won't take too much of your time. 

Gene: is it possible that you can restore cosmic ray data from Run11, Run14 and Run19? I think a quarter of the Run18 statistics used here for each dataset is probably sufficient. 

Run 11, I see P11id FullField and ReverseFullField MuDsts that I will restore.

Run 14, I see P14ih Half, Fulll, and ReverseFullField MuDsts that I will restore.

Run 19, I do not think we did an official production, but I believe Yuri handled a private production for use in TPC calibration work. We can do an official production if it is desired.

Thanks,
-Gene

Best
Rongrong

On Oct 20, 2022, at 8:21 AM, Robert Líčeník <licenik AT ujf.cas.cz> wrote:

Hello Rongrong,

since the y axis is log-scale (but the values are not), a simple linear function will not work, so I think you meant something like p0 + p1*10^x . It seems to do much better.
Here are the updated slides:
https://drupal.star.bnl.gov/STAR/event/2022/10/18/Tracking-Efficiency-Uncertainty-meeting/Cosmics-2018-update
Would it be helpful if I shared the file with the produced TTree with you? That way you can have a quick look by yourself and you don't have to wait for me in a different time zone every time.
However, I understand that you are very busy and I am of course more than happy to help.

Cheers,
Robert

On Wed, Oct 19, 2022 at 8:06 PM Ma, Rongrong <marr AT bnl.gov> wrote:
Hello Robert

Could you try a linear function y = p0 + p1*x to do the fit? Thanks. 

Best
Rongrong

On Oct 19, 2022, at 11:55 AM, Robert Líčeník <licenik AT ujf.cas.cz> wrote:

Hi Rongrong,

you are correct. The fit function is written on the first slide with momentum resolution and the range is the full histogram for the red fit and (-0.4 - 2) for the blue fit.

Cheers,
Robert

On Wed 19. 10. 2022 at 17:40 Ma, Rongrong <marr AT bnl.gov> wrote:
Hi Robert

Thanks for the update. However, the fit does not seem to work well. What function and fit range are you using? Thanks. 

Best
Rongrong

On Oct 19, 2022, at 9:05 AM, Robert Líčeník <licenik AT ujf.cas.cz> wrote:

Hi Rongrong,

no problem, here are the updated slides:
https://drupal.star.bnl.gov/STAR/event/2022/10/18/Tracking-Efficiency-Uncertainty-meeting/Cosmics-2018-update

Cheers,
Robert

On Tue, Oct 18, 2022 at 11:00 PM Ma, Rongrong <marr AT bnl.gov> wrote:
Hello Robert

Thanks. If possible, could you redo the fit starting from the lowest point, i.e. log(pt) ~ -0.4, and add the slope of the fit to the figures? 

Best
Rongrong

On Oct 18, 2022, at 2:38 PM, Robert Líčeník via Star-tf-trkeff-l <star-tf-trkeff-l AT lists.bnl.gov> wrote:

Hi everyone,
I uploaded my slides from today's meeting:
https://drupal.star.bnl.gov/STAR/event/2022/10/18/Tracking-Efficiency-Uncertainty-meeting/Cosmics-2018-update
The only difference is that I replotted the momentum resolution figures in log-log scale, as requested. 
Cheers,
Robert

On Mon, Oct 17, 2022 at 12:07 PM Petr Chaloupka via Star-tf-trkeff-l <star-tf-trkeff-l AT lists.bnl.gov> wrote:
Dear All,

Please join our meeting tomorrow at 10:00 New York time.

Please upload any material for discussion to the drupal page:
https://drupal.star.bnl.gov/STAR/event/2022/10/18/Tracking-Efficiency-Uncertainty-meeting

With best regards,
Petr

To join the meeting:

https://bnl.zoomgov.com/j/1607035037?pwd=WDVqaEdwa0RtTEM5a1VvZ24yU2gwdz09
Meeting ID: 160 703 5037
Passcode: 099371

One tap mobile
+16692545252,,1607035037#,,,,,,0#,,099371# US (San Jose)
+16468287666,,1607035037#,,,,,,0#,,099371# US (New York)

Dial by your location
      +1 669 254 5252 US (San Jose)
      +1 646 828 7666 US (New York)
Meeting ID: 160 703 5037
Passcode: 099371
Find your local number: https://bnl.zoomgov.com/u/aoARs6rfR

















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

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