INTT+Calo Tracking Meeting

→ Asia/Tokyo
Description

TWilki: https://wiki.sphenix.bnl.gov/index.php?title=INTT_Analysis_Software#INTT.2BCalo_Tracking

Zoom information: https://wiki.sphenix.bnl.gov/index.php?title=INTT_Internal_Meeting

Quick Recap:


In this meeting, TOMOKI reported on a machine-learning approach to electron particle identification and discussed how AI can improve inclusive electron separation power. Participants discussed increasing the statistics, classifying by momentum, and creating plots comparing AI with the conventional method. Takashi also reported on preparations of Si+Calo nanoDST data, using the same runs as used for the recent J/ψ peak analysis by Bade. At the end of the meeting, the role of the production manager and volunteer support were discussed.

Next Steps

TOMOKI:
  • Prepare slides describing the research theme in more concrete terms by the next presentation.
  • For slide 9 (BDT score distribution), create distribution plots sliced by particle pT.
  • Plan PHYSIA simulations to increase statistics, and measure and document the required computing time and disk space.
  • Prepare data for the conventional method without machine learning (manual cuts) so that the separation power can be compared with the AI-based approach.
  • Evaluate which parameters contribute most to electron identification (feature importance analysis).
  • Add comments to the code and prepare to explain the code in detail to Yasuyuki next week.
  • Check the current disk space usage.

Takashi:
  • Proceed with the production converting the Run 3 dataset to NanoDST, and notify the members once it is complete.
  • Consider requesting that the collaboration (upper management) create a DST containing all cluster information.
Akitomo:
  • Talk with the members at RIKEN and decide on volunteers to help with Takashi’s production work (job management and output checking).

Yui & Mizuki: 
  • Prepare to present status reports at next week’s meeting.
There are minutes attached to this event. Show them.