sPHENIX discussion of physics

[Anthony Frawley <afrawley AT fsu.edu>]

Hi All,

Below are my notes from the HF/Q topical group meeting yesterday.

Tony

Slides:

https://indico.bnl.gov/event/14736/

MDC2 update - Cameron Dean

-----------------------------------------

Fixed a bug in the pp setup that inadvertently caused jet production. The memory use is now down, so more jobs can be run concurrently.

Last slide: What is nEventTracks? How many tracks are in the track node for a reconstructed event. These include pileup tracks.

Wei: Do you plan to produce D0 exclusive events? Yes, we have a dedicated Pythia8 config file. But we will use EvtGen if we can. The magnetic bend is not done before the decay if we use Pythia decays.  EvtGen fixes that so is preferable. The charm events contain D0 as subevents, but statistics are limited.

Wei: AuAu timeline? Not clear, there is a long queue ahead. Maybe request a smaller Hijing sample to use for testing.

Hideki: Jet sample? The wrong config file was producing jet samples, we still have them. But Jin pointed out these were embedded in jet pileup events. We need to organize a list.

D0/D0_bar MLP and BDT training and performance - Thomas Marshal

---------------------------------------------------------------------------------------------

Fixed overtraining, ran new performance tests. Looks more sensible.

What is used for the background? The background signal for D0 is D0_bar. A more general background would be better.

The BDT has the best performance in the ROC curves, but more/different background could change that.

Wei: Does TPOT give time info useful for PID? Jin: No, the TPOT acceptance is small. The best tool is precision tracking. We may also be able to use dE/dx. Will have to do statistical separation in any case.

Jin: Have you tried adding momentum info of the pi and K to the MLP, BDT models? Thomas: Looked at it, but not pursued yet. They are highly correlated. It is interesting to look further at it.

Discussion of using D0 from D^(*+) decay to D0 + pi. Can use pi charge to separate D0 and D0_bar. The reduced branching fraction produces a hit in yield, but could use for data driven systematic study.

Hideki: What causes the spike in the middle on slide 5? These are events where both assumptions give the mass of the D0. Could be removed by adding more variables (momenta and rapidity were mentioned).

Zhaozhong: Did you check the correlation matrix? Yes, but was overtraining at the time. Will do it again.

Zhaozhong Shi - EvtGen decayer

---------------------------------------------

Working on implementing EvtGen decayer in sPHENIX.

Jin: Time frame for finishing? Zhaozhong: In the next few weeks is possible.

Jin: We want to enable the user to configure EvtGen from the macro. Maybe generate config file from thenFun4All macro.

Next meeting moved:

-----------------------------

The next scheduled date will occur during WWND2022. The conveners decided to slide the schedule by a week to avoid that. So the next meeting will be on March 9.