sPHENIX MAPS tracker discussion

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

Joint Heavy Flavor & Quarkonium Topical Group Meeting (2021/8/18)

https://indico.bnl.gov/event/12703/

Introduction - Hideki

We plan to migrate to joint bi-weekly meetings from today, and

try to keep them within an hour. If we have many contributions,

we will switch to weekly.

Hideki: What is the status of the upcoming production?

Cameron: B field map is implemented this early Summer. I will

  check with Chris whether we can proceed with the 2D field map.

Tony: ACTS requires the 3D field map and it cannot load

   the 2D field map. The vertexing is still not working. What

   is your plan: truth or reco vertexing?

Cameron: As realistic as possible, but our tools are transparent

   for either condition.

Tony & Joe: We suggest the truth vertexing for now.

Joe: If you want a bit more realistic vertexing, you can rerun 

   it from tracks that are fit with the truth vertexing.

Tony: The 3D field map does not yet work in the simulation. Can

   you use the 2D for the simulation and the constant field map

   for ACTS?

Cameron: We saw that such an option creates weird feature in

   the momentum reconstruction. So, to be consistent, we should

   use the constant field map for both the simulation and ACTS

   for now.

HF Tools - Cameron

Tony: When you run the secondary vertexing, on what basis do you

   do? Decay kinematics or MVTX?

Cameron: Sort of both. We do use the DCA. Inputs are the track

   parameters in the track map.

Wei: Is the ancestor information available in KFParticle?

Cameron: There is a truth history option. If you enable it, you can

   go up to the parton level, and the information will be saved as a

   vector.

Tony: Will you consider pileup in HF?

Cameron: No for MDC1, but yes for MDC2.

Electron ID - Weihu

Zhaozhong: pion rejection for the electron ID would be helpful for

   the B meson analyses.

Weihu: Yes, the rejection performance is shown on p.7.

Cameron: What does the electron pair module write to? The node tree?

Tony: The node tree and output file.

Hideki: Why is the electron ID performance better with the embedding?

Weihu: Actually the pT range is different (p.12). You can see the pT

   dependence on the far right plots on p.7.

Sebastian: The pT dependence is rather small. It would be better to

   make two sets of plots with only one cut applied at a time.

Zhaozhong: Maybe a MVA-based electron ID would improve the

   performance.

Marzia: E/p dominantly contributes to the ID performance, and

   E_HCAL/E_ECAL is somewhat redundant. I am not sure how much

   improvement we can obtain from MVA.

Hideki: Simple cuts should be fine for Upsilon measurements, as the

   highest priority is the efficiency. We should check whether we need

   to consider other working points for the HF analyses.

Tony: Our ID already provides several levels of cuts, and can reject

   the combinatorial background.

Sebastian: What do you exactly use for E_HCAL?

Weihu: 3x3 clusters.

MVTX Alignment - Jaehyun

Tony: How do you consider the misalignment in simulation?

Cameron: We copy over the MVTX classes. Layer 2 is shifted

   in the GEANT4 simulation.

Tony: Are these features in the G4 area?

Cameron: Right now no. It is in my local repository and I

   recompile locally.

Tony: The objective is to test the alignment in an internal code?

Cameron: Yes. Whether to add in the daily build would require

   discussions with Chris.

Joe: So, the surface is moving.

Yasser: I thought the same thing.

Tony: It is a different way of thinking about it; it is a modification

   of transformation associated with the surface.

Jaehyun: What we determine is the correction function once a run.

Sebastian: What level of misalignment?

Jaehyun: Maximum of 500 microns.

Hideki: What do you exactly use when you say AI?

Jaehyun: Multi-layer perceptron.