HSF Weekly Meeting #172, 26 September, 2019
Present/Contributors: Graeme Stewart, Attila
Krasznahorkay, Pere Mato, Marco Clemencic, Daniel Elvira, Paul Laycock,
Gloria Corti, Eduardo Rodrigues, Andrea Rizzi, Liz Sexton-Kennedy, David
Lange, Ben Morgan, Chris Jones, Guilherme Amadio, Michel Jouvin,
Caterina Doglioni, Heather Gray, Sudhir Malik, Stefan Roiser, Kyle
Knopfel
Apologies: Agnieszka (at a conference), Serhan Mete
(ATLAS Software Tutorial)
News, general matters
- Post-JENAS HSF meeting on October 16th can take
place, 16:00-17:30, at LAL and on Vidyo:
- JENAS: Meeting ECFA + NuPECC (nuclear) + APPEC
(astro),
https://jenas-2019.lal.in2p3.fr
- Idea of this meeting is to start the
discussion with other communities, opening the road to
discussions on more practical problems.
- Example of successful tool sharing: Rucio,
also more in GW
- Practicalities
- Speakers
- We don’t mean to have an exhaustive
list of speakers (would be difficult to do), this is a
first brainstorming that can reach a relatively large
audience due to the context.
- Both APPEC and ECFA presidents will be
present.
- “Intro-to-HSF” speaker, any
(remote/live) volunteers?
- Better that it’s in person -
Graeme can do it.
- Plenty of options from astroparticle,
none from nuclear:
- Astroparticle options:
(prioritize people @
workshop)
- CTA (Stefan Schlensted,
software responsible)
- SKA/DUNE (maybe invite Andy
McNab)
- ESCAPE (Giovanni
Lamanna)
- Speaking, not registered →
emailed first
- DM (Chris Tunnell, was at JLab
WS)
- Neutrino:
- IceCube (Anna Frankoviak,
analysis coordinator)
- KM3NeT/ANTARES (Antoine
Kouchner)
- Michel could ask for
contact, if not himself → emailed
- Nuclear: any suggestions?
- FAIR people → HSF/Graeme’s
contacts
- Jerome Lauret -
BNL/STAR
- Michel can ask as well
- Graeme and Caterina can send
invitations (today)
- How to announce?
- JENAS website once speakers.
- HSF mailing list.
- [Caterina at dark matter conference]:
Astroparticle community very positive to this, but need to sit
down together to work on something in practice as
data/software needs are usually quite different
- UK RSE
conference last week,
Stefan Roiser presented the idea of a
scientific software
institute,
with the idea of establishing better connections between natural
science and computer science research. The presentation was well
received though no concrete actions emerged from it immediately.
From earlier brainstorming this year a
one page
statement
with the main ideas of this initiative is available.
- After the Computing seminar on Wed several met
with the speaker who is also author of Alpaka
(https://arxiv.org/abs/1602.08477).
The possibility of holding a workshop on Alpaka as a concrete
example of computer science and natural science collaboration
(also including photon science & others) was discussed.
- Alpaka training at CERN foreseen (would be
useful for core software and reconstruction
developers).
- What we really want is for things to get into
the C++ standard!
- GDB Chair (Ian Collier) has invited HSF to have a
member on the GDB steering group.
- Meets after each GDB and provides advice to
the GDB chair about topics for future meetings.
- We are positive about this.
- Important to have an increasing recognition
for software and the role of the HSF.
Activity and Working Group Updates
Data Analysis
Detector Simulation
- We will resume the topical meetings. The current
plan is to have one before CHEP and another one at the end of
November/beginning of December.
- We are discussing potential topics for them as
well as for next year.
- Geant4 collaboration meeting this week has a
session on Computing performance with reports from all
experiments,
https://indico.cern.ch/event/825306/sessions/314167/#20190926.
Reconstruction and Software Triggers
- Plans for this Fall:
- Originally planned meeting on ATLAS/CMS/ALICE
SW optimization in trigger and reconstruction on October 2nd,
but may be postponed as many ATLAS trigger experts are
organizing a hackathon. ATLAS speaker available on October
9th, no conflicts in the calendar if we do 5 to 6 pm
(sacrificing 1h of ATLAS week) but need to check CMS/ALICE
availability.
- Next meetings: include more reconstruction, so
talked to ATLAS reco experts, main interests to add to our
original ones in the July
report:
- Priority is run-4 tracking
- Particle flow cross-talk between ATLAS and
CMS (and future colliders?)
- Calorimeter reconstruction on ML
- More futuristic: tracking on quantum
computers
- More ambitious plans for the future: would be
interested in thinking about a HSF COST action for the April
deadline:
https://docs.google.com/presentation/d/1facB4BKuN4zm66C9JtoOhDa68rqenIz-cWWqdhLpSfY/edit?usp=sharing.
- Needs more people than us to be on board /
think it’s useful.
- Main pros:
- Very open funding tool.
- Once a consortium gets it, others can
join afterwards as full members & access funding as
well.
- Could fund HSF (WLCG) workshops.
- Would encourage student exchanges / stays
at CERN to work with experts.
- Can join up with other COST actions and
work together on software
- Main cons
- Not much money compared to other
grants.
- Can consider it as “seed money” for
bigger things.
- Non-negligible paperwork, so we’d have to
crowdsource…
- Maybe something we discuss in a dedicated
meeting?
- TODO: post HSF trigger&reco meeting live notes to
our own webpage…
- Reminder: TAU Seminar
- Monday October 14th at 5pm CERN time Room:
40/S2-D01
- https://indico.cern.ch/event/845622/
- Graeme and Serhan will meet (w/b 7 October) to
discuss the details of installing the tool to
/cvmfs/hsf.org.
- Thanks to Gordon, pinged Marc Goodner from VS Code
about Visual Studio (waiting response).
- Could do something more targeted towards our
community - building on CVMFS and Docker. Should interact a
bit in advance.
- Ben - this would be very useful for DUNE and
packaging.
- Planning to contact Luca and Guillermo from IT for
a possible seminar/tutorial for using Intel Parallel Studio XE if
people think that’d be useful.
PyHEP
- PyHEP 2019 workshop basically ready (just small
things to be done at this stage).
- Final agenda is now online and will be advertised
soon.
Training
- Training meeting on Monday discussed organisation
for CERN Software Carpentries workshop at the end of
November.
- Training “pre” blueprint meeting planned (to
assess scaling of past carpentries material and plan
future)..
- LAWSCHEP workshop in Mexico City, November 20-23,
- 25+5 min presentation on “Software training needs and
challenges”
(https://indico.cern.ch/event/847294/)</span>
- 19 Aug - 23 Aug, 2019 -
ATLAS Software
Carpentries Training
(LBNL).
- Training meetings are now Mondays 15h30. Reminder
Packaging
- Meeting regularly each month, practical work
focused on Spack as a prototype tool (work at FNAL and
CERN).
- Next meeting 9 October.
Frameworks
- Trying to get a foothold for the moment…
- After some discussions with Chris and Kyle, we
will have a first meeting for the group next Wednesday.
- In the first few meetings we will invite
experiments to talk about their current efforts with their
software frameworks, mainly about the issues that they want to
solve in the non-too-distant future.
- Will start with more topical meetings
afterwards
- Plan to have 2 meetings until CHEP (with next
Wednesday’s being one of them).
- So that we may have something to talk about in
a possible parallel meeting there. But on this we’ll just have
to see what we have by the time we get there.
Workshops
Next HSF/WLCG Workshop
- First, many thanks to all of the people who prepared bids. This is non-trivial work and we really appreciate all the effort.
- All bids had very attractive aspects!
- The feedback from the community highlighted
positive aspects from every bid; overall it was very useful in
helping us to make a choice.
- Based on logistical constraints and on the
feedback we received we decided that the next workshop will be in
Lund from 11-15 May 2020.
- Congratulations to Caterina, Mattias and Oxana - we look forward to this a lot.
- We have a few logistical points we need to work on
pretty soon.
- It would be good to reduce costs for Early
Career Researchers over more senior people (reduced
registration and priority for less expensive
accommodation).
- (Pre-)booked 29 single rooms for ~60
EUR/night/person and 22 doubles for 30
EUR/night/person.
- The Friday before the workshop (8 May) is a
public holiday in France and in the UK.
- Do we start later on Monday to allow
people to travel to Lund on Monday, rather than
Sunday?
- What time to finish on Friday to allow for
reasonable travel home?
- From personal experience (Caterina):
leave for airport ~3.5h before flight if no checked
bags, there is a very convenient flight back to Geneva
at 20:50. Finishing at lunchtime would allow for this
and earlier flights.
- Input on these points is already
welcome!
Pre-CHEP, Analysis Systems: From Future Facilities to Final Plots (2-3 November)
- Workshop will be divided into two:
- Saturday PM, 8 speakers laying out challenges
and opportunities for analysis in the future.
- Sunday AM, an interactive session with
everyone involved in gathering ideas to tackle 4 key questions
for the future of analysis.
- Details to be circulated in the next few
weeks.
- Timetable is in place:
- Local organisers have done a great job to provide
us with suitable rooms for the Sunday activities - thanks to
them.
Latin American Workshop on Software and Computing Challenges in HEP (Mexico, Nov 20-23)
- Latin American PIs have been able to find funding,
so they will be present.
- Strong international presence.
- IRIS-HEP are participating and sponsoring (can
fund US participants).
- Aim to have all of the sections of the CWP roadmap
represented.
- Workshop will be small, but should have all key
people and that can be a way to start cooperation (within Latin
America and with the wider community).
- CMS and ATLAS invited.
- Hope that registration will be usable to support
LA students to present their work (posters).
- Sponsorship from AWS fell though, unfortunately.
AOB
- Important experiment week dates seem to be a bit
weak in the calendar for 2020.
- Can some names individuals take charge of LHC + Belle II + Dune + …?
- CMS - Liz can take care.
- DUNE - not clear what state they are
in.
- Belle II - Paul will follow up
- ATLAS - Graeme
- ALICE and LHCb: ask computing
coordinators?
- Next Rucio workshop will be at FNAL. Want to avoid
clash with CTD (in Princeton).
- David - CTD will now probably be at the end of
the week April 20-24.