Skip to content
David Quigley edited this page Dec 17, 2021 · 6 revisions

Welcome to the hpc-reprohack wiki!

Planning

Support:

  • 3-4 Warwick mentors - 10 days time each
  • AK: 15 days
  • Admin support

Timescales

  • Event registration & comms:

    • Set up hub event
    • Ask mentors to promote paper submissions (pre-christmas)
    • Deadline for papers end of February
    • Filter out papers which are out of scope before asking students to select
    • DQ to set up registration for students (immediately in the new year, register before end Jan)
  • Documentation:

    • https://sulis-hpc.github.io/
      • accessing and using the system
      • recreating computational environment
        • modules
        • environments
        • containers: docker vs singularity
      • managing code: especially git
      • managing data.
  • Event: March: 21 March - 01 April

Event Format

  • Duration: 7 - 10 days

    • In person kick-off: 1 full day.
      • training in morning
      • start hack in afternoon
    • Closing event: 2 - 3 hrs
      • Re-group and present progress
      • Talks
    • Drop in support sessions 2-3 1-2 hr sessions in the week
  • Rest of the time asynchronous work

Documentation needs

  • Docs on using the HPC systems
  • Docs on translating workflows to HPC workflows
    • Configuring HPC computational environments
    • Building Singularity containers?
  • Docs for demonstrators

Student Cohorts

  • CDT: Modelling heterogeneous systems e.g plasma physics modelling
    • one of the leaders (James) uses a lot of containerisation. Other's just list software used. python, fortran, HPC, openMP MPI.
    • students will have some familiarity with HPC but likely with existing software or a lot of support
  • CDT: Mathematics of real world systems
    • less computational training but some lectures on git, HPC, software best practices. Python & Julia familiarity. Interest in machine learning. Big influence of Warwick epidemiology team.
  • CDT: AI, heavy users of tensorflow. Reproducibility in tensorflow of interest

Papers

  • Request mentor submissions
  • Request help from mentors
  • Papers with code? https://paperswithcode.com/
  • Papers using VASP - using open source version or free of package instead