Go to the Indico Home Page Oldest event Older event Up to category Newer event Newest event Printable version
  • Compact style
  • Indico style
  • Indico style - inline minutes
  • Indico style - numbered
  • Indico style - numbered + minutes
  • Indico Weeks View
    • Switch to the management area of this event
    Login

    EP R&D Software Working Group Meeting

    → Europe/Zurich
    The Zoomiverse

    The Zoomiverse
    Live Notes
    EP R&D Software Working Group Meeting
    Zoom Meeting ID
    62893115044
    Host
    Graeme A Stewart
    Alternative host
    Andre Sailer
    Useful links
    Join via phone
    Zoom URL
    Hide

    Software R&D Working Meeting Minutes

    Introduction

    • R&D Report - please send your name to be included on the author list
    • ML4Sim - should add this to the Simulation page on the R&D website
    • ACAT - quite few submissions planned from different tasks

    Calorimetry and TICL

    • Q. Slide 24 - are these single particles?

      • A. Yes, single pions. The non-interacting particles are important to quantify. Performance plots with 140 and 200 pu. are prepared for the CMS TDR (to be approved today). Results are encouraging, both for core and tails. But some things to still improve, like jet energy resolution.

    • Q. To integrate into Key4hep what are the dependencies?

      • A. CLUE is completely stand-alone, with no unusual dependencies. TICL is loosly coupled to CMSSW, so will be a bit more complicated to make stand-alone.

    • How to make sure friction is minimised on integration, e.g., how is logging handled? how is it configured? Topic for future discussion.

      • CLUE standalone also relies on Alpaka
      • CLUE CMSSW is C++ only

    • Key4hep is using Gaudi for a lot of services, this does take some work then to integarte things. No clear decision on hetergeneous packages - for future colliders CPUs are working well (for now). Is CLUE an external library?

      • Foresee a special version to integrate with Key4hep; this is not a big overhead (algorithms all factorised). This is a point to follow-up on

    • Q. Slide 17, how is the algorithm robust against noise?

      • Because it works with local densities; pure noise would reconstuct 0 clusters, this is part of the kernel definition and tuning; also know that scaling with noise should not be a problem
      • For the parameter tuning this was not too difficult, so st should not take too long to retune for a different detector
    There are minutes attached to this event. Show them.