sPHENIX MAPS tracker discussion

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

Heavy Flavor Topical Group Meeting (2021/1/4)

Participants:

Hideki, Jin, Cameron, Guannan, Martin, Ming, Ping,

Sanghoon, Xuan, Yuanjing, Zhaozhong

Introduction - Jin

The same meeting time turned out to be optimal.

We may revisit the meeting time ~once a year

(maybe send another doodle poll in September).

New MatterMost channel for MDC1:

https://chat.sdcc.bnl.gov/sphenix/channels/hf-mdc1

Cameron Dean has agreed to be the production

manager from the Heavy Flavor TG. Thanks!

MDC1 samples focus on inclusive HF signals (50M

events for ccbar and bbbar, respectively) Volunteers

to look into diverse HF signal channels are welcome.

MDC1 Status & Plans - Cameron

One bug found in sPHENIX framework: Jet QA crashes

if KFParticle is enabled. Jet reconstruction itself is ok.

Ming: Why do we need Jet QA for MDC1?

Cameron: Actually, we do not need it for MDC1.

Jin: In the MDC1 samples, very low-pT jets will dominate.

No need to run the Jet QA. However, this crash implies

that something is wrong deeper in the framework, and

needs to be debugged (but not urgent for MDC1).

Slide 2

Jin: These are pp samples, so no minbias embedding

actually.

Sanghoon: Do cc samples contain bb as well?

Jin: The events are filtered, so cc and bb are produced

separately.

Slide 5

Cameron: The first root file is in the DST format. Please

have a look if you are interested.

Follow-up

Cameron will give an update at the simulation meeting (1/5):

https://indico.bnl.gov/event/10406/

B_s update - Zhaozhong

Slide 2

Jin: You should separately make signals and BGs.

Hideki: Why do you separate training and testing samples

like that? Usually, we pass one input file and specify the

percentages for training/testing.

Zhaozhong: This is a pure technical issue. The training

crashes for the higher track multiplicity somehow. I will try

to fix this in the next round.

Slide 6

Jin: Please remake the right plot in the log scale.

Slide 7

Jin: How do you get the BG number? Is this significance of

~60 true? Also, integ. lumi. =1.4 fb-1?

Zhaozhong: BG normalization to be checked. The 5-year run

plan was considered for the integ. lumi. here. I will also

update this.

Jin: Is the pre-cut signal efficiency considered in the table?

Zhaozhong: No, this is the BDT efficiency only.

Hideki: You worked on the same analysis in CMS with cut-based.

How much do you gain by switching to BDT?

Zhaozhong: Actually, I used the cut-based only for B+. For

B_s, BDT is used also in CMS. Anyway, the gain should be by

a factor of a few, not a factor of 10.

Slide 8, 9

Jin: You need to produce more BG samples.

General

Xuan: Did you consider PID in your studies? The PID capability

is not good with sPHENIX TPC.

Zhaozhong: No hadronic PID considered. For the J/psi phi

channel, I consider the electron PID.

Cameron: How do you plan to use TMVA? Online or offline?

If you have an xml file, we can run it online in MDC1.

Zhaozhong: I plan to run it offline.

Jin: We are not ready to produce such an xml file for MDC1.

Cameron: OK, just to let you know, a class to use xml files

is already implemented in KFParticle.