sPHENIX MAPS tracker discussion

[Hideki Okawa <Hideki.Okawa AT cern.ch>]

Joint Heavy Flavor & Quarkonium Topical Group Meeting (2021/12/22)
https://indico.bnl.gov/event/14044/

[TODO & call for volunteers]

We want to ask for volunteers who can look into the multiple

track fitting with different PID assumptions and check the D0 and

Lc mass reconstruction.

=>Please contact us if interested.

Introduction

MDC2 samples consider full pileup simulation; the best guess

we have for the 2024 data taking. Please have a look at them

so that we can provide feedback to the tracking team.

Next meeting will be on Jan 12 or 19 after the Collaboration

meeting. We will choose the date based on our progress.

Toolset for analyzing MDC2 - Cameron Dean

Millions of simulation events already available for MDC2 (e.g.

8M Pythia8 Minbias). The list of samples was mentioned at

the previous meeting:

https://indico.bnl.gov/event/13884/contributions/57898/attachments/38713/63947/MDC2_toolset_C_Dean_20211208.pdf

Previously, we had to use different readers for calo info and

truth. Cameron changed the code to read DSTs as vectors.

It should be more transparent now.

Please use "-p" flag for pileup and no-pileup comparison.

cshell is not good at dealing with strings. The cshell option is

removed for MDC2; it automatically switches to bash.

MDC2 D0/D0bar separation - Thomas Marshall

p.4

Cameron: Is there combinatorial background here, or just

   the real Kpi?

Thomas: Yes, there is pipi contamination.

Cameron: How did you check whether it was the true D0?

Thomas: Jin suggested to use the track 1,2 ID and compute

   the invariant mass with the truth momenta.

Jin: HepMC record is not available for decay length > 1 micron.

   Beyond this, G4 handles the decay instead of Pythia8. Truth

   analysis chain is more complicated now.

Tony: What is the purpose of the study and what is the motivation

   of the truth selection.

Jin: For most analyses, we do not care about D0/D0bar separation.

   Exceptions are the D0/D0bar v1 flow and azymuthal asymmetry.

   With the truth selection, we want to see the cleanest events and

   show first of all that we can separate D0 and D0bar. Then, once

   we tune the cut with simulation, we move onto background studies.

p.6:

Marzia: Is the tail from the true D0 on the left plot?

Thomas: The plot has both D0 and D0bar.

Marzia, Jin: Please make plots for true D0 or D0bar only; namely

   4 plots in total.

Marzia: Can we see the tail in the real event?

Jin: We only use the peak for the analyses.

p.9:

Hideki: What does each point correspond to in the ROC curves?

Thomas: Each point has a different cut applied.

Jin: We need to think about how to do the 2D mass scan and

   improve the presentation of the ROC curves to match with the

   standard convention.

Cameron: Which KFParticle setup?

Thomas: The default.

Cameron: That would have a 1-GeV cut on D0 pT.

Wei: The daughters from D0 with pT=0 GeV have <pT>~0.7 GeV.

   It is better not to cut on D0 pT.

Cameron: The default cut has no physics reason; it was just an

   example.

Wei: No cut or pT>0.5 GeV is better.

Joe: Tracking efficiency drops at low pT. It is worth checking the

   impact of track fitting with different PID assumptions. How is

   the linkage of track fitting and KFParticle?

Cameron: KFParticle only reads one track container. I will need to

   look into handling multiple track maps.

Joe: Shall we consider this in the production?

Jin: We need to prove that it is useful with pre-production tests

   before we really include it in the production. How long does it

   take for the refitting?

Joe: 1-2 sec per AuAu event. Less for pp pileup.

Tony: Actually, quite less for pp pileup. How is the pT resolution?

Joe: It significantly drops at low pT. Wei's group showed such

   plots. We want to ask for volunteers for this self-contained study.

  (advertised at the top of the minutes)

Marzia: How is the PID capability with TPC?

Jin, Joe: dE/dx is not used. The TPC is not large enough.

Tony: TPC is optimized to suppress the ion backflow. We cannot

   sacrifice this for dE/dx. Though, there was an idea that we can

   use dE/dx at some level. It is worth pursuing. We should discuss

   this with Tom. (Contacted Tom after the meeting)

Jin: I have seen in the testbeam that dE/dx can improve with the

   gas choice. Poor dE/dx, but maybe still useful.

Wei: What is the best way to become familiar with KFParticle?

Cameron: (1) twiki: https://wiki.bnl.gov/sPHENIX/index.php/KFParticle

   (2) MatterMost. I am also happy to offer zoom conversations for help.

   (3) Grab example scripts and run