---

HSF Weekly Meeting #127, 22 February, 2018

Present/Contributors: Graeme Stewart, Benedikt Hegner, Marco Clemencic, Marco Petric, Jim Pivarski, Mike Sokoloff, Mohammad Al-Turany, Sudhir Malik, Daniel Elvira, Helge Meinhard, Pere Mato, Martin Ritter, Peter Elmer, David Lange, Liz Sexton-Kennedy, Giulherme Amadio, Michel Jouvin, Oli Gutsche, Simone Campana

News, general matters

• Graeme returned the message sent about the FOSS Backstage event, but no response.

HSF/WLCG Workshop

• General points

  • 139 participants registered so far
  • Extended early registration finishes 28 February (in 6 days), so now is a good time to get people to sign up to come.
  • The organisers confirmed that there’s one room with better vidyo - simulation session will benefit from this. Anyone else with a particular need?
  • For US institutes there is a possibility of support from S2I2 conceptualisation funds, contact Mike Sokoloff if this is interesting.
  • Liz is trying to encourage the head of LArSoft (Erica) to come for the neutrino field. Will try to get others (Tom, Heidi, Andrew).
  • KEK/Belle II are coming, but could have more (Michel).
  • BES/IHEP participation - we think so, Pete and Graeme to follow up. (Perhaps Xiaomei Xhang or Wenjing Wu, both IHEP)
  • Discussion on schedule:
    • Simulation and Frameworks could well do with late afternoon sessions to maximise remote participation. Decided to: swap simulation AM session with software development; swap frameworks early PM session with Costs & Performance late PM session. Simulation gains 30’, software development loses 30’.

• Monday

  • Opening Plenary
    • Finalised
  • Technology Watch
    • Skeleton is filled in.
    • Interesting work on networking offered - would be difficult to have if we the vendor talks.
      • This is very interesting and we should try and make time this, need 10’.
    • Still need to follow up with the organisers re. the vendor presentations? Yes. Graeme will take this.
  • HEP Use Cases
    • Speakers contacted, waiting for responses. No particular concerns right now.

• Tuesday

  • Data Management/Lakes
    • Getting meaningful requests for talks, so need to re-think agenda to capture all the good ideas, but also have discussion. To be sorted on Monday.
  • Frameworks and Infrastructure
    • 2 confirmed talks, last one still to 100% pin down. So in principle this is done.
  • Training (Parallel)
    • Will have 2 hours, 90 min talks, 30 min discussion. Foresee a lot of contributions. Want to reach out to future experiments too. 9 lightning talks of 10’. 8/9 have confirmed. Will add this to Indico now.
  • DPHEP (Parallel)
    • Follow up offline.
  • Workload Management
    • Torre: “still need to set talks/speakers for workload management, f2f with Frank at CERN in a week”.
  • Analysis Facilities and Use Cases
    • Agenda almost final. Just a last speaker to confirm.

• Wednesday

  • Programming for Concurrency and Co-Processors
    • 3 or 4 speakers confirmed. Last one should be ok. Overlap with simulation fixed.
    • Did we find a second convenor yet (Vincenzo Innocente?). Yes!
  • Visualisation (Parallel)
    • From Ric:
    • “The Naples agenda is growing:
      • “we will have updates from experiments (ATLAS, CMS so far; hopefully, others as well)
      • “plus, we will have discussions on potential technologies for common tools, and experiences/best practices”
  • Simulation
    • We have assigned 11 talks, some of which will be given remotely via Vidyo.
    • We have requested that our morning session be shifted to late afternoon for the benefit of those connecting from west coast USA. Hopefully we will be able to swap with another session. Done (see general points).
    • Live attendance is looking better, thanks to S2I2 support.
  • Security (Parallel)
    • David: “The topics for the security session are still the same (albeit working on relative emphases), and we’re discussing speakers (Dave Kelsey has agreed to speak to some policy areas, and Hannah, Romain and Liviu have offered to help prepare materials). At least some of us (including Dave and Hannah) will be at the WISE meeting in Abingdon next week and I’m hoping to continue discussions there. I still have it in mind to keep 30 minutes for discussion with 90 minutes for talks.”
    • Did we find a second ‘live’ convenor yet? Could ask DK.
  • Performance and Cost Modeling (Parallel)
    • Looks in good shape.
  • Technical Discussion on Frameworks (Parallel)
    • Some ideas on topics - looking for speakers.
    • Did we find a second convenor yet?
    • Outcome from the ROOT I/O workshop could be used as input
    • Remote participation from the US would help - pivot with P&C technical session.
  • Software Development (Parallel)
    • Preliminary idea of topics, need to work out the details.
• Thursday
  • Closing Plenary
    • Pere and Liz met on Monday. They are proposing to organize the session as a collection of summary talks (10’) given by the chairpersons with the main messages for each of the parallel sessions and finalize with a summary of summaries (20’) as a conclusion of the workshop with a bit of time for the final discussion and next steps.
    • Some of the San Diego session summaries were very successful, could use those as a model. We are in a different phase now, which should make things easier. Concentrate on concrete outcomes and on collaboration across boundaries. What is the future for the HSF given the outcomes of the sessions?
    • We should have live notebooks for all of the sessions, which will help with the summary of summaries. Can create templates for all of the summaries, like in Annecy.

CWP

• General Matters and Roadmap

  • Michel will give a presentation about lessons from the CWP and future work at the 4th CERN Scientific Computing Forum, on Friday 23 (tomorrow). Draft slides available for comments.
  • Updated author list version (v3) is live on arXiv.
  • Discussions with the Editors of CSBS advanced - they cannot change the author list after submission (Springer policy). So a ‘pre-review’ is happening, based on arXiv v3. We can keep the author list open until the Naples meeting.
  • Noticed that the CWP inputs are now hidden behind many clicks, so we should make sure they are easier to find. Graeme TODO.

• Publication strategy for Individual WG Papers

  • Simulation

    • Ready to go to arXiv now, with a finalised author list.

  • Software Trigger and Event Reconstruction

    • Planning to upload both full doc and executive summary tomorrow. Current versions are https://github.com/davidlange6/recoCWP/blob/master/recoCWP.pdf and https://github.com/davidlange6/recoCWP/blob/master/recoCWPFull.pdf so please make any further requests today.
    • We asked all the authors to confirm their entry in our author list and got quite a bit of feedback. We’ve fed this back to the main CWP doc github where relevant.

  • Data Analysis and Interpretation

    • Current version (forked from in HSF/documents, syn’d with origin repo)
      https://github.com/msneubauer/documents/blob/master/CWP/papers/HSF-CWP-2017-05_analysis/
    • Progress on text this past week, plan to have version uploaded sometime next week

PyHEP Workshop

• No major update (just 1 extra talk+speaker confirmed since last week).

Activity updates

• Eduardo suggested that we should get Zenodo DOIs for the HSF technical notes and also highlight a bit more the TNs (which are a bit hidden). Feedback for this? Also worth for the CWP and CWP chapters?
  • Generally sounds useful for the DOIs. The website could generally do with some reorganisation to reflect the more active areas and highlight them more.

Licensing

• HepMCv2 is relicenced as LGPL2.1+. Need to also do HepMC3 and follow up with main authors.
• Should follow up with FastJet at some point.

Training

• No news.

Packaging

• Meeting yesterday focused on experiences with aliBuild and the lessons learned. Next meeting in 2 weeks time (7 March) will discuss best use of and feature development in Spack and using the test stack for tool comparison.

GSoC

• Progressing well, a lot of proposals and applications received.
• We have many projects but we need to decide how to decide the prioritization of the project: should not be left to the GSoC coordinators to decide which projects are selected if Google is funding less students than the total number who applied.
• Student numbers are increasing - perhaps we will hit a limit. Do we need 2 organizations, one on each side of the Atlantic? Not sure… To be rediscussed later, after we know the numbers from this year.

Visualisation

News from Riccardo Maria Bianchi:

• As announced at last week's meeting, we started an exploratory CMS-ATLAS joint project with the Computing Science department of the University of Lubjana, to explore possibilities offered from advanced visualization techniques used in other fields, such in Medicine.
• There is a first draft of the project. More content and background has to be added.
  • This is an attempt at exploring new potential technologies for common tools, which could be shared among the experiments. It is a follow-up from what we stated and proposed in the Visualization CWP.
• Of course, all interested people are more than welcome to join!

AOB

• EP R&D Software Group will have its second session of Lightning Talks on Monday 26.
• Should poll for dates to be entered in the community calendar - Benedikt to follow up.
• Postal addresses for institutes for, e.g., CHEP - it would be useful to have this.
  • Pete volunteered to help collate some of these based of Gordon Watt’s ACAT lists.

---