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

    BLonD code development meeting

    → Europe/Zurich
    CERN

    CERN

    Video conferencing only
    Hide

    General

    Add docs and developer info to new website

    BLonD-common

    Helga: Abstract toy model

    • Keep data type common for Ring and RF
    • Common Ring, inheriting RingTracking, RingDesigning
    • RFSystems in RFStations?

    Alex: Codimd for Ring object

    • same data_type object for Ring and RF
    • synchronous data type via key-word argument?  => kwargs PREFERRED (consensus)
      • conversion functions automatic if input data is tagged
      • if not tagged, can call conversion by hand
    • Simon/Heiko: can input a single number or a numpy array for simple cases; no need to define an object => SIMPLE CASES SHOULD BE INTUITIVE FOR USERS (consensus)
    • YAML input? Helga: might be too ambiguous
    • Heiko: input data_program(data, time=time) and tag it when inputting to Ring
      • Simon: don’t want to discard the numpy array direct input option -> TO BE DISCUSSED NEXT TIME
      • KEEP OPTION C of codimd (consensus)
    • Markus: use a single input type for clarity

    Kostis: Synchrotron Radiation Implementation

    • Unittests have 1 % difference between C++/Python implementations
      • Statistically the same, but different random number sequences
      • Seed option now available in both
    • MPI issue with multi-bunch distribution
      • Problem with profile.reduce() before beam.split()
      • Now beam object knows whether in reduced mode or not => no flag needed
    • - Improved MPI testing
      • Skip MPI tests if MPI not installed
      • validate_mpi.py comparing MPI and regular BLonD outputs
      • time limit -> some examples (integration tests, not considered in coverage) are skipped, check with Helga

    Panagiotis: GPU implementation

    • moving data between CPU/GPU being investigated
      • FFT, tracking etc. all done on GPU
      • single transfer at the beginning/end of simulation -> should be investigated first (no interaction with tracking block)
      • keep ratio of macroparticles/slices constant for scaling
    • complex simulations being assembled
    • beam object functions implemented in GPU

    Next meeting exceptionally on Thursday

     

    There are minutes attached to this event. Show them.
      • 10:30 → 10:40
        News 10m
        Speakers: Aaron Farricker (CERN), Alexandre Lasheen (CERN), Giulia Papotti (CERN), Heiko Damerau (CERN), Ivan Karpov (CERN), Konstantinos Iliakis (CERN), Luis Eduardo Medina Medrano (CERN), Markus Schwarz (KIT), Panagiotis Tsapatsaris (NTUA), Simon Albright (CERN), Theodoros Argyropoulos (CERN)
      • 10:40 → 11:20
        BLonD common 40m

        Ring object definition

        Speakers: Alexandre Lasheen (CERN), Markus Schwarz (KIT), Simon Albright (CERN)
        BLonD-common, Ring and datatypes input.html
        Ring and datatypes input
        Toy model
      • 11:00 → 11:20
        MPI and GPU development 20m
        Speakers: Konstantinos Iliakis (CERN), Panagiotis Tsapatsaris (NTUA)
        blond_meeting_03042020.pdf