HSF Weekly Meeting #174, 21 November, 2019
Present/Contributors: Graeme Stewart, Riccardo Maria
Bianchi, Chris Jones, Michel Jouvin, Andrea Rizzi, Caterina Doglioni,
Kyle Knoepfel, Andrea Valassi, Agnieszka Dziurda, Martin Ritter, Heather
Gray, Serhan Mete, Stefan Roiser, Josh McFayden, Paul Laycock
Apologies: Eduardo Rodrigues, Liz
Sexton-Kennedy
News, general matters
- New convenor positions - nominations open until 22
November, send to
hsf-search-committee@cern.ch:
- PyHEP
- Software Tools and Packaging
- Event Generators
- Software Training
- Detector Simulation (1 position)
- Data Analysis (1 position)
- We are happy to receive self-nominations,
especially from currently active convenors
- Graeme has been asked to give a presentation to
the CERN Scientific Policy
Committee
on “Goals and plans of HEP Software Foundation and related
areas”
- This is excellent news that we are asked to
provide input at this level
- Will need your contributions!
- Graeme gave a talk to the CEPC workshop on
HEP Software
R\&D
(we can add this to the list of general presentations)
Activity and Working Group Updates
Data Analysis
- Not yet a follow up for next meetings
- Analysis languages
- DOMA follow-up from pre-CHEP workshop
Detector Simulation
- Prepared slides for LAWS (see below)
- Trying to schedule a meeting for 4 December (16h)
on user feedback for geometry and description
- Let us know if you’d like to contribute a
talk
Reconstruction and Software Triggers
- Lots of ideas from CHEP!
- Great to see many plenaries/parallels in line with
HSF & HSF trigger+reco interests
- E.g. Birds of a Feather sessions (real-time
analysis in LHC experiments and beyond, organized by TJ Khoo
and S. Schramm, tracking in trigger organized by A.
Boveia)
- Sessions with talks on reconstruction with
accelerators
- Plenaries on real-time analysis
- We could have a “summary of CHEP & follow-ups” in
one of the next meetings, but priority is organizing the follow-up
to the optimization in CPU / multithreaded meeting with ATLAS and
CMS
- Possible dates: Wed Dec 4th, Dec 11th
- Hands on session with TAU tool, 20 January.
- Would be good to have a piece of GPU code that
could be used as a profiling example
PyHEP
- The PyHEP 2019 workshop took place in Abingdon,
U.K., 16-18 Oct.
- We were already asked by several colleagues when
the next workshop will take place!
- US seems a natural location, and US colleagues
(IRIS-HEP) are keen to help …
- Interest in having the event tag along with
SciPy, for cross-community exchange.
Training
- Software Carpentry at CERN is next week
- Graeme was asked to join the Scientific Organising
Committee of the ESCAPE school, which will happen early next
summer (June): outcome of the
JENAS
discussions
- Good opportunity to reach out beyond HEP for
basic training (ESCAPE grows out of a previous Astroparticle
school called
ASTERICS)
- Foreseen that we can have a HEP specific
specialisation at the end of the week
Event Generators
- Have not resumed activities yet after the summer
and CHEP. Will do so soon.
- Accounting and benchmarking to be followed
up
- Steve will give a report this evening at the LAWS
workshop,
https://indico.cern.ch/event/813325/contributions/3603456/
- Andrea gave a brief report yesterday at the LHCb
computing workshop,
(direct link to slides:
ppt,
pdf)
Packaging
- CHEP
presentation
on the group’s progress since Sofia and the current plans for the
Key4hep stack to be built with Spack was well received
- Heard news from the FAIR experiments that they
have moved to using Spack
- Belle II would like to also investigate this
direction
- Nice that this community consensus was covered in
the Track 5 highlights
talk
- Hope to have another meeting before the end of the
year
Frameworks
- We had a nice meetup over lunch at CHEP 2019
involving people from 10 different experiments. It was a chance
for each person to introduce themselves and bring up issues
associated with their experiments.
- We want to have another meeting before the Winter
break period.
Quantum Computing
- (reported by Riccardo M. Bianchi)
- New interest in the HEP community.
- Riccardo Bianchi had a discussion with Federico
Carminati: Openlab has contacts with major QC players (IBM,
Google, …) who would like to have pilot projects with “CERN” (for
industry partners, CERN is equal to HEP…); however, unfortunately,
we currently don’t have HEP use cases to establish partnerships
with industries.
- HSF QC & Openlab would like to organize a workshop
together; let’s say, next Spring. The draft idea is to have
vendors presenting their platforms. After that, there would be a
HEP-only session to discuss the potential uses and establish
realistic use-cases. Then, there will be a follow-up session,
where HEP will present their with vendors to set up joint
projects. Alberto Di Meglio, Federico Carminati, Riccardo M.
Bianchi, and Maurizio Pierini are now brainstorming on how to
organize HEP working groups on QC
- Federico and Riccardo think it is important to ask
vendors first about what they are interested in and what their
platforms can offer; then we can come up with realistic use cases
which are also interesting for vendors and feasible on their
infrastructure, avoiding dead ends.
- There was a Google Cirq Bootcamp at CERN,
organized by Google & OpenLab, last week. Riccardo went and
established some contacts with “Google QC Partnerships” team.
Also, a draft of a joint project has been prepared by Markus
(Google), Federico (Openlab), and Riccardo (HSF) to test on the
Openlab cluster a new port of Google Cirq to Intel. Google will
contact Intel to port their Cirq library to the Intel Quantum
simulator. Openlab will provide the cluster. HSF QC, and possibly
the CERN Theory Div., will be involved to test code on the new
platform.
- At the Google Bootcamp Riccardo shared the HSF QC
mailing lists with the participants and new members have joined
that.
- Also, Federico would like to invite the members of
the QC openlab mailing list to join the HSF list and close the
former.
- P.S. If interested, you can take a look at the
Bootcamp’s slides and hands-on instructions here:
https://sites.google.com/view/cirq-bootcamps/when-where/geneva
Visualisation
- (reported by Riccardo M. Bianchi)
- Phoenix:
- Phoenix
(https://github.com/HSF/phoenix)
is developed under the HSF aegis. Phoenix was started by ATLAS
(Ed Moyse), then LHCb (Ben Couturier) and CMS (Tom
McCauley) joined to integrate and test their data files and
geometries in the tool.
- Phoenix was awarded with a Google Summer
of Code student, Emilio Cortina Labra, last Summer. Emilio
worked very hard: he refactored the structure, updated the UI,
and added many features to the tool.
- Phoenix was presented at the Paris “Learning
to Discover” workshop
(https://indico.cern.ch/event/847626/),
where Emilio run a Phoenix hands-on tutorial.
- A new Topical Workshop?:
- I saw some activities in visualization within
the experiments recently (and at CHEP2019); I think it could
be the right time to have another topical workshop. I’ll ask
about potential interest on the Visualization list.
Workshops
HSF WLCG Workshop in Lund, 11-15 May 2020
- International organizing committee is in the
process of choosing the location…
- One location is the main University Building
(iconic and in the centre of Lund)
- Or a hotel, which would have very professional
setup and would be smooth
- Price will be about 300€ (+50€ for
conference dinner), but the university does not include
lunches
- lunches can be cheap in the centre of
town
- If the weather is nice, there are a lot of
places where one can take lunch to go and sit in the park
(Lundagård). We could inquire if we can put out some
standing tables but it’s unclear if we can do it as it’s a
public park.
- Full description of two locations here:
https://docs.google.com/document/d/1lLLA4NqyCa5MpYJVseflX-LiHwKFu3CvDp5j_83PgAM/edit?usp=sharing
- Opinions to:
wlcg-hsf-workshop-2020-organisation@cern.ch
Pre-CHEP, Analysis Systems: From Future Facilities to Final Plots (2-3 November)
- Workshop was judged a success - people rather
pleased with it
- 2 x ½ days was the correct length
- Plenary session on Saturday was good to set
the scene, but has the usual issue that <5 people contribute
about 80% of any discussion
- Interactive session was innovative and people
liked it, yet we could improve some aspects of the
organisation:
- People should know their challenge topic
earlier (we wanted to give people time after Saturday to
choose this, but this meant the assignments did go out too
late)
- It was important to let people choose
their group or groups if they wanted to do so
- The brainwriting was hard for some groups
(not all) to get started on - having a pre-discussion was
fine in these cases, see first point
- Sizing the groups to be about 6 was the
right decision, (our initial idea for larger groups would
not have worked as well)
- The group and supergroup discussions were
really useful and people liked them
- The final plenary feedback session was too
long (back to the problem of very few contributors in
these very big sessions)
- Organisers will meet next week with the aim of
providing a very short summary and outcomes document
Latin American Workshop on Software and Computing Challenges in HEP (Mexico, Nov 20-23)
- Ongoing, report next meeting
AOB
- Went back to Google Docs for this week’s minutes,
due to problems with CodiMD and people who did not have full CERN
accounts
- Positive feedback on the CodiMD idea, with
direct MD editing
- Any update on EduGAIN authorisation to write
to CERN CodiMD?
- Or should we use the
public CodiMD
instance?
- Decision: we can use the public CodiMD for now -
N.B. this is only for writing minutes, no long term need for
it after the minutes are posted
- Next meeting…
- Next week is Thanksgiving (28 November), so
reconvene 5 December
- We had some discussion about the 2 week cadence
for these coordination meetings
- This was decided, but the weeks themselves
remained quite dynamic (i.e. not regularly in even or odd
weeks)
- Hard for people to plan other
activities
- We probably do need to be fairly flexible, but
would it be preferred that we picked a baseline to prefer
(N.B. this is an odd week)
- Yes, this would be a good idea - we’ll do this
for next year’s meeting
- ECFA discussion of early-career researchers on
input to the European Strategy update
- Computing & Software discussion
slides.
- Should think about coordination list vs. people
who are active organisers for HSF
- Coordination list has become rather
broad
- Sometimes not clear how to contact HSF
- Could setup a smaller management/executive
list?
- We should think about this and decide if we
propose any changes.