HSF Weekly Meeting #200, 21 January 2021

Present: Graeme Stewart, Josh McFayden, Serhan Mete, Ben Morgan, Kyle Knoepfel, Paul Laycock, Efe Yazgan, Attila Krasznahorkay, Meirin Oan Evans, Michel Jouvin, Andreas Salzburger, Benedikt Hegner, David Lange, Eduardo Rodrigues, Peter Elmer, Stefan Roiser, Gloria Corti, Kevin Pedro, Krzysztof Genser, Liz Sexton-Kennedy, Marc Paterno, Mason Proffitt, Michel Jouvin, Philippe Canal, Stefano Piano, Sudhir Malik, Daniel Elvira, Torre Weanus

Apologies: Dorothea vom Bruch, Caterina Doglioni, Michel Villanueva

News, general matters, announcements

CLARIPHY Letter of Collaboration

We are asked for a letter of collaboration from the CLARIPHY project for their NSF funding application. See the agenda links for background documents.

Graeme Stewart has agreed to coordinate the collaboration between the HEP Software Foundation (HSF) and CLARIPHY. This collaboration will include co-hosting training and community events of mutual interest

The meeting approved the letter, Graeme will send the signed copy to Pete.


We in CERN and Fermilab IT wanted to leverage the HSF community to discuss what to do about the policy changes at RedHat/IBM regarding CentOS. While GitHub “Discussions” are still in beta, we thought it was worth a try to start one there in order to hear from as many people as possible who wish to contribute.


Waiting to add content to the site before advertising it more widely.

Google Summer of Code

Remember to submit your project ideas by 15 February! Also check the instructions for changes to the projects this year.

HEP C++ Course and Hands-on Training

Working Group Updates

Data Analysis

DAWG plans for early-mid 2021 (updated w/b 25 Feb to be more complete)

Ongoing series of metadata discussions seems like a workable format to repeat on other topics. Some themes that could follow:

Besides these, we discussed commonalities with Training WG. One q: how much should HSF branch into (experiment-agnostic) analysis tutorials?

DAWG Metadata Discussion 1 — Takeaways:

Detector Simulation

Reconstruction and Software Triggers

Summary of Hand-off meeting

More details… Brainstorming on Topical Meetings

Possible formats:

Next steps:


Software Tools and Packaging

We kicked off this year’s meetings w/ a fruitful discussion on perf yesterday:

We agreed to continue the discussion in an upcoming meeting:

We’re going to have a conveners’ meeting, most likely first week of February, to iron out plans for this year. Please let us know if you have any topics/subjects you’d like to see us cover…

Software Training

Plans for 2021:

Hackathon follow up Friday 22nd January 17:00 CERN time https://indico.cern.ch/event/997485/

Event generators

Thoughts from 2020:

Plans for 2021:

Complete the paper at last! (should be ready in ~few weeks)

Arrange meetings to discuss work done by others as in the plans laid out in detail in the paper, many areas of work:

  1. Understanding of the current CPU costs by accounting and profiling.
    • There has been separate work on e.g. profiling MG5_aMC and Sherpa which actually show some common conclusions
      • Time to bring this together and compare more apples-to-apples
    • Look ahead to NNLO codes?
  2. Moving to GPUs/accelerators.
    • Continue effort on the MG5_aMC GPU port
    • Try to follow-up better on similar activities for other generators (e.g. Sherpa and Pythia8)
  3. Optimisation of phase space sampling and integration algorithms, including the use of ML
    • This is somewhere we have not yet made large inroads - it will become a larger focus.
      • Topical meeting for the WG.
  4. Reducing the cost associated with negative weight events
    • Quite some discussions on this already in the WG
      • In particular at the November WS
    • Next steps…?
      • Trying to get existing tools more integrated into generator/experiment/ workflows
  5. Promote collaboration, training, funding and career opportunities in the generator area:


    • Build contacts with relevant projects:
      • E.g. Excalibur/SWIFT-HEP (UK) and HEP-CCE (US)
      • Offer a forum for communication between the two groups and help avoid overlaps?
    • Engage more with the Experiments?
      • There is work that has already been done but doesn’t always make it into the experiment workflows
    • Engage more with Nuclear people? Use contacts established via S&C Roundtable
    • Neutrino community - Liz has some contacts

    Career opportunities and funding

    • Not sure exactly what the WG can do here currently other than raise the visibility of the work through e.g. the paper and presentation at LHCC…


    • So far we have not been active in providing training
      • Should discuss this with HSF Training folks!
    • Need to think what that would look like…
      • E.g. would generator-specific code training be possible/make sense?
      • If not, do we need more “generic” training on e.g. programming for GPUs, profiling etc.?
        • This might be more the right emphasis from HSF.

Frameworks (updated 2 February with WG plans)


IRIS-HEP Feedback

We have been asked to give another round of feedback to IRIS-HEP at their next steering board meeting on 16 February. Can you think about feedback (the good, the bad, the ugly) to send via Graeme?

For reference, the feedback given last year is here: https://indico.cern.ch/event/809181/.


A few recent updates:

Next Meeting(s)