---

HSF Weekly Meeting #173, 17 October, 2019

Present/Contributors: Graeme Stewart, Eduardo Rodrigues, Andrea Valassi, David Lange, Attila Krasznahorkay, Agnieszka Dziurda, Chris Jones, Daniel Elvira, Witek Pokorski, Stefan Roiser, Serhan Mete, Liz Sexton-Kennedy, Mark Neubauer, Kyle Knopfel, Nikos Kasioumis

Apologies: Pere Mato, Caterina Doglioni

News and General Matters

Working Groups in 2020

• We discussed the future of working groups in 2020 in the coordination group
• The decision was that for the current appointed groups we allow convenors to carry on for another year, as we are very pleased with the outcomes of the groups
  • Most convenors have said yes to this proposal (thank you very much!)
  • However, Danilo (Analysis) and Heather (Simulation) have had to step down, due to other commitments
  • We shall therefore call for nominations to replace them
• We also decided that the nomination and appointment process is a successful one and that it should be used more consistently for working groups from now on
  • So we are also going to call for nominations for convenors in the following groups
    • PyHEP
    • Software Tools and Packaging
    • Physics Event Generators
    • Training
  • Current, active, convenors in these areas are stongly encouraged to be nominated, including self-nomination
  • Note that we decided that Software Tools and the Packaging group can be usefully merged into one
  • Three convenors are sought for each group
• Areas where the HSF activity has been more sporadic and informal will be rebadged as Activity Areas
  • These will not have formally appointed convenors, but carry on with the input from motivated members of the community
  • This type of activity is very much encouraged when people would like to start a new interest group (evolution to a working group would be possible, in time)
  • At the moment we propose that Visulaisation and Quantum Computing become activity areas
• Timeline:
  • Call for nominations next week until mid-November (post CHEP)
  • Comments for the community for ~2 weeks after that (end November?)
  • Appointments to be made in December, if we can, for 2020
    • These will be as now, 1 year, renewable
• Meeting approved this plan

Post-JENAS Meeting

• As discussed in previous HSF coordination meetings & announced on the mailing list, Caterina organised a meeting with the Nuclear, Dark Matter, Astroparticle communities as a follow-on from the JENAS Workshop
  • We presented an overview of the HSF
  • Followed by contributions from nuclear, neutrino, dark matter and ESCAPE communities
    • HSF-ESCAPE interaction encouraged by APPEC in this context
  • Very much a first getting to know you meeting
  • The level was quite senior
  • Feedback was positive, but not immediately clear how to take next steps
  • Two interesting concrete things came from the discussions with ESCAPE
    • Training event at LAPP that HEP could be involved in
    • Software directory that is being built as a deliverable for ESCAPE and could be promoted for HEP software projects
  • Caterina and Graeme are thinking we can probably continue interactions at the Lund workshop next year, inviting a few key people (not that we do nothing else untl then!)

Activity and Working Group Updates

Detector Simulation

• Next meeting will be on Dec 4th due to CHEP and other experiments’ meetings between now and then. We plan it will be on feed-back on use of new geometry tools by experiments. We will contact them soon to fix the agenda. For first meeting next year we would like to invite non-LHC experiments.
• There was a GeantV dedicated meeting on Tuesday. There will be a formal write-up of the results obtained. At the end Pere Mato and Philippe Canal presented directions for future activities. We should follow this up regularly next year.
• Directions on accelerators are now interesting. Will the GeantV group look at this? No, it will be done with new R&D projects. We know that this is hard and will require new approaches.
• Geant Exascale project is exploring this area.
• Contact between ORNL and CERN recently

Reconstruction and Software Triggers

• Next meeting, targeting “Algorithmic approaches and data structures to efficiently exploit many-core architectures” moved from 9th October to “after CHEP” due to unavailability of the speaker.

Software Tools

• In case you missed it, there was a TAU talk last Monday (monitoring tool)
• To be followed up with a hands-on session, most probably will take place on January 20th at CERN:
  • Imperative to focus on HEP use cases.
  • It might be nice to give priority to accelerator related examples.
  • Please let us know if you have a workload that you’d like us to focus on.

PyHEP

• PyHEP 2019 workshop now taking place in Abingdon, U.K., 16-18 Oct., see https://indico.cern.ch/e/PyHEP2019

Frameworks

• Will have a meeting next Wednesday. Plan to have presentation from DUNE. Workflow is quite different from HEP collider experiments. Speaker TBC.

Training

• Software Carpentry at CERN, 27-29 November
• A lot of progress in organising the event
• Registration opens tomorrow…
• Advertise this also via HSF lists, particularly to call for helpers

Event Generators

• No meetings yet after the summer
• Have not pinged or heard back from the Argonne team working on Madgraph on GPUs
• Josh McFayden is working on a setup to allow standalone runs of several generators, which eventually may be incorporated into the HEP-workloads benchmarking suite using Docker containers
• Steve Mrenna will give an overview of activities at the Latin American workshop in Mexico

Packaging

• Meeting last week had a very useful discussion on best practices for package relocatability from Ben Morgan
• There are a few packaging contributions we know of at CHEP
• Plan to have another meeting before the end of the year

Workshops

Next HSF/WLCG Workshop

• Caterina is offline today, but happy to start the discussion on progress with logistics, program… if we want a dedicated kick-off meeting that we then summarize, we could do next Thu in the HSF slot?
  • Will poll on coordination list
  • We need to setup an IOC

Pre-CHEP, Analysis Systems: From Future Facilities to Final Plots (2-3 November)

• Event is now very well planned
• Just putting the finishing touches to the interactive sessions to try and get these to go smoothly
  • All topics have confirmed animators to help the discussions
  • We will try to have group ‘coordinators’ (someone who guides the smaller group writing process and discussion)
  • We will circulate information shortly to all participants so that people know what to expect

Latin American Workshop on Software and Computing Challenges in HEP (Mexico, Nov 20-23)

• Good attendance expected (32 people registered): Latin American PIs, international speakers, students
• Workshop URL https://indico.cern.ch/event/813325
• Giving final touches to the agenda
• We are missing three important topics/speakers:
  • Detector simulation (Pere, any progress identifying a speaker? HSF simulation converners consulted, possibility of a CERN speaker)
  • Reconstruction (IRIS-HEP will support travel if speaker is US-based). D. Lange, P. Elmer, G. Watts trying to find a suitable speaker from an NSF-funded institution. Alternatively, Giuseppe Cerati is available and willing to do it.
  • “HPC lanscape for science applications”. We are trying to get someone from Oakridge. Not clear we will succeed. Ideas?
• Everyone: you still have time to register and attend! Excellent opportunity to network with Latin American collaborators.

AOB

• We experimented with switching to CERN’s collaborative Markdown editor, CodiMD, for the meeting minutes this week
  • More convenient to write notes in the format needed by the website than to have to convert from Google Docs
  • Unfortunately it seems that lightweight accounts cannot edit these documents on the CERN service
  • Options:
    • Get CERN to allow lightweight editing
    • Switch back to Google Docs
    • Use the CodiMD demo service, e.g. with a note like this one
      • Notes can be set to be editable by anyone at all (freely) or by logged in people (editable) - to login a GitHub account can be used
  • Nikos (CERN IT): want to understand the needs
    • In a nutshell, we have HSF people who do not have full CERN accounts, so we do need something that everyone can contributed to
    • Federated identity is advancing fast - eduGAIN would be an option, as all of our people should have institutional credentials of some kind
• Next coordination meeting…
  • CHEP is very soon, do people want to have a meeting on 31 October? Or wait till afterwards, 14 November
  • Agreed next meeting is after CHEP

---