1 of 15

Software Training

With a focus on cross-experiment training

Kilian Lieret1,2, Wouter Deconinck8, Peter Elmer1,2, Sudhir Malik3, Alexander Moreno4, Stefan Roiser5, Jason Vaetch6, Michel Hernandez Villanueva7

for the IRIS-HEP/HSF Training Group

1

1 Princeton University

2IRIS-HEP

3University of Puerto Rico

4Universidad Antonio Nariño

5CERN

6Universität Göttingen

7DESY

8University of Manitoba

Confidential

Customized for Lorem Ipsum LLC

Version 1.0

2 of 15

The need for software training

  • As we collect more and more data and perform increasingly complex analyses, our software is mission critical �→ need to follow industry standards & best practices
  • The broader data analysis ecosystem is evolving faster than ever, but these changes are driven by industry → we must keep pace

  • (Almost) all scientists write software but few have formal software education�→ almost every scientist needs to be trained

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

3 of 15

Much is being done already…

Experiment onboarding

Recent talk at CHEP from the HSF Data Analysis and HSF Training WGs (proceedings in progress)

Common challenges for all initiatives:

  • Supporting self-study and events (virtual, in-person, hybrid)
  • Maintaining up-to-date materials (unit testing and regular updates)
  • Consolidate different resources (Sphinx, doxygen, twiki, forums, …)
  • Interactivity & active learning
  • Sustaining educator workforce

No two experiments have the same strategy; every experiment excels at certain aspects; not the focus of this talk

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

4 of 15

Much is being done already…

Cross-experiment training schools

Unsurpassed for hands-on training and networking, but limited #participants

and self-study support

… and more

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

5 of 15

Much is being done already… but we need to scale!

Basic Programming Skills

Intermediate/HEP Domain

Advanced

Developers

O(6k)

Undergrads

O(3k)

Ph.D. students

O(1k)

Postdocs

O(😭)

Faculty + RSEs

We need a unified, scalable, and sustainable software training framework

Need to educate

O(10k)

HEPex people worldwide

Let’s collaborate and do this efficiently!

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

6 of 15

6

We need a unified, scalable, and sustainable software training framework

Unified

  • Material and events should be centrally listed & discoverable
  • Concentrate efforts by developing cross-experiment content
  • A community must guide, support, and coordinate

Scalable

  • Material must be teachable by multiple instructors
  • Self-study must not be an afterthought

Sustainable

  • Material must be open source and maintained collaboratively
  • Incentives & recognition important motivators

The IRIS-HEP/HSF Training group is building a community around these principles

7 of 15

A unified Training Center for HEP

7

  • HSF Training Center currently lists 25 training modules of various authors
  • Want to become focal point for all HEP Training activities
  • Central list of training events (everyone can add)

25 modules

O(500) views/month

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

8 of 15

Building a community

Monthly Hackathons

Community pages

Recognition

How-to guides

Platforms

Increasing our reach

Weekly meetings

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

9 of 15

Scalable & sustainable trainings

Example 1: HSF Analysis preservation training (5 individual modules)

9

Many of our modules embrace the framework of The Carpentries

  • Built from markdown files (easy & maintainable!)
  • Rendered as a webpage with Jekyll
  • Verbose and self-study ready
  • Events use combination of self study with videos + live lectures + small-group virtual mentoring sessions

Bundling 5 Modules

Video walkthroughs

Clear prerequisites

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

10 of 15

Scalable & sustainable trainings

Example 2: HSF C++ course

10

  • Slides generated from LaTeX source
  • Has been taught in-person, virtual and hybrid
  • Life lectures and exercise sessions
  • Full videos available from events

548 slides, 695 pages, > 1k commits

Originally developed by S. Ponce, now community effort driven by B. Gruber, S. Hageboeck et. al.

8 events till now

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

11 of 15

We scale!

11

Our material is �

  • verbose

Onboarding instructors is easy!

→ Significant self-study readership

  • Easy-to-edit and open-source �→ Can be adapted and extended for different contexts

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

12 of 15

Sustainability

12

Training is mission-critical and costs significant resources Efforts must be incentivized:

  • Acknowledged: HSF credits contributors both on our community page and on individual lessons, but could more be done?
  • Encouraged: Supervisors must motivate and push their students/ employees to contribute
  • Rewarded:
    • Service work credits should be given for all contributions (including cross-experiment training)
    • Hiring discussions should acknowledge and push for contributions

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

13 of 15

Conclusion

  • Software training efforts are central
  • Experiment-specific training initiatives differ vastly between experiments (let’s learn from each other’s strengths!)
  • Experiment-independent topics are the foundation on which to build; we need to scale up training in this area
  • The HSF Training WG built a community of educators around training material that is open source, verbose and sustainably maintainable
  • Scaled to > 2000 learners and more could be done
  • Our training center aims to be a central entry point to all relevant material
  • We need buy-in from the experiments to grow and sustain our efforts:
    • Need instructors & mentors
    • We welcome & support new material
    • Cross-linking our training center/ material will increase readership
  • Contributions to training must be
    • encouraged (by supervisors)
    • rewarded (service work credits, hiring, …)

13

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training

14 of 15

Thank you!

14

15 of 15

Much is being done already…

Impart best practices & HEP domain knowledge: �HSF Intermediate Training, experiment StarterKits, …

Democratize science: �HSF software basic training, University courses, Self-study, …

Mentoring & hands-on experience: TAC-HEP, WATCHEP, IRIS-HEP, GSoC, other internships

Networking & catch up with recent developments:CSC, GridKA, CoDaS-HEP, MLHEP, INFN ESC, …

Basic Programming Skills

Intermediate/HEP Domain

Advanced

Developers

🚧 [WIP] WILL BE REPLACED BY COMPLETELY DIFFERENT SLIDE 🚧

(Cross-experiment) Software TrainingKilian Lieret for IRIS-HEP/HSF Training