Unleaching the power of digital transformation in accelerator physics

a new collaborative approach

Adnan Ghribi
adnan.ghribi@ganil.fr

GANIL / CNRS

September 29, 2023

Outline

  • Introduction
    What are accelerators ? Why are they important ?
  • Scattered developments
    Separate ways of dealing with the same picture
  • A new synergetic approach
    When FAIR and OPEN become a catalyser
  • Crossing boundaries
    Phyiscs, science, industry and society

Introduction

What are accelerators ? Why are they important ?

What are accelerators ?

  • Magnificient machines built to explore the boundaries of the human knowledge ;
    • From the Rutherford1 experiment, the Cockroft2 and the Van Der Graaf3 machines to Lawrence’s Cyclotron4 ;
    • Accelerators have now penetrated every aspect of our lives and given us the chance to expand our knowledge.

Why are they important (1/2) ?

  • Their applications span accross different disciplines :
    • Particle physics
    • Nuclear physics
    • Light sources
    • Medicine
    • …

Why are they important (2/2) ?

More than 30 000 accelerators operational world wide1

More than 99% used in industry and medicine :

  • Industrial applications > 20 000 ;
  • Medical applications > 10 000.

Less than 1% used in research and discovery science :

  • Cyclotrons ;
  • FFAG ;
  • Synchrotrons ;
  • Synchrotron light sources ;
  • Linear and circular accelerators/colliders.

Challenges

  • Accelerators pose quasi-industrial challenges in terms of operation and reliability ;
    • Detecting, preventing anomalies ;
    • Optimising beam time ;
  • Frugal complex physics simulation is another important aspect for future accelerators :
    • New digital twins ;
  • Several groups/labs and RI and are trying to meet these challenges.

The AI Context

  • Artificial Intelligence applications for physics and society is revolutionizing the way we do and think our roles ;
  • The acclerator community is catching up fast ;
  • However, synergy is much needed for data as well as for methods and applications.

The AI big picture

#Disrupt 4.0

and the ML picture

Pugliese, Regondi, and Marini (2021)

Scattered developments

Separate ways of dealing with the same picture

A complex picture

A mostly scattered developments landscape

  • Mosty individual and scattered efforts
  • Collaborations
    • Large RI infrastructures or accelerator projects
    • Specific tools for optimization and operation

Few examples

Tools

https://slac-ml.github.io/Badger

https://gitlab.cern.ch/geoff

Few examples

Projects

  • Real-Time Edge AI for Distributed Systems (READS)1
  • Artificial Intelligence for the Electron Ion Collider (AI4EIC)2
  • EUROpean Laboratories for Accelerator Based Science (EURO-LABS)3
  • CERN accelerator Machine Learning platform4

But there are limitations …

There are many limitations to scattered developments :

  • Numerous tools doing almost the same thing :
    • Risk of non sustainable development ;
  • Large projects focused developments :
    • Limited impact on smaller RI ;
  • Lots of redudancies :
    • Ressources intensive calculation, storage, machine studies ;

A new synergetic approach

When FAIR and OPEN become a catalyser

General Purpose

  • We want to unlock the use of artificial intelligence in particle and nuclear accelerators as well as in light/neutron sources ;
  • We want to tackle all challenges of particle accelerators.

The network

Two key workshops that led to the creation of two networks :

  1. A French network - M4CAST1
  2. A European Network - TRAINABLE2

and target Horizon Europe project - ARTIFACT3

The network

9 countries, 18 partners and even more direct and indirect beneficiaries.

What came out of it …

  1. We need guidelines to unclock the use of AI in acclerators ;
  2. We need to standardize and open our data and approaches ;
  3. We need to structure ourselves to work in a fertile collaborative space ;
  4. We need to keep an open mind (astrop, HEP, medecine, …) ;
  5. We need to train people : students as well as professionals ;
  6. We need a common goal : one to rule them all !

In other words

flowchart TD
    subgraph Transverse flow
    t1(Training and knowledge transfer)
    t2(Industrial applications and technology transfer)
    t3(Interdisciplinarity and outreach)
    t1-.-t2
    t2-.-t3
    end
    subgraph Logical flow
    A(1. Collecting inputs from the community)
    B(2. Bringing FAIRness)
    C(3. Developing specific tools and methods)
    D(3.1 Simulations)
    E(3.2 Anomalies)
    F(3.3 Optimisation)
    G(4. Integrating into a pilot facility)
    A --> B
    B --> C
    C -.-> D
    C -.-> E
    C -.-> F
    D <-.-> E
    D <-.-> F
    E <-.-> F
    F --> G
    E --> G
    D --> G
    end

And in numbers

  • 4 years
  • 10 M€
  • 86e+11 neurons1
    • To kick-off a European roadmap on the matter ;

Crossing boundaries

Phyiscs, science, industry and society

The obvious boundaries

  • We are taking a way others have taken boefore us ;
  • But we learned along the way that our way is somehow different … yet complimentary ;
  • We see that some of our weaknesses in research can be compensated by strong industrial partnership ;

\(\Rightarrow\) Boundaries are meant to be crossed.

The hidden boundaries

Cultures, countries, languages, religions, genders are also boundaries that are not meant to limit us, but to enrich us so that we can go further.

\(\Rightarrow\) The TRAINABLE network should work hand in hand with its US, African and Asian counterparts ;

  • If not, help pave the road for their creation through common scientific programs.

Conclusion

  • Exciting times ahead for the field ;
  • But we need everyone to make progress ;
  • And hope that AFRICA will play a significant role in the challenges ahead.

Thank you

References

3rd African Conference on Fundamental and Applied Physics | 25-29 Sep. 2023

Allaire, C, R Ammendola, E-C Aschenauer, M Balandat, M Battaglieri, J Bernauer, M Bondı̀, et al. 2023. “Artificial Intelligence for the Electron Ion Collider (AI4EIC).” arXiv Preprint arXiv:2307.08593.
Chou, Weiren, and Alexander Wu Chao. 2019. Reviews of Accelerator Science and Technology-Volume 10: The Future of Accelerators. World Scientific.
Cockcroft, John D., and Ernest T. S. Walton. 1932. “Experiments with High Velocity Positive Ions: II. The Disintegration of Lithium Nuclei by Protons.” Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 136 (830): 619–30. https://doi.org/10.1098/rspa.1932.0078.
Edelen, Auralee L, Jonathan P Edelen, L RadiaSoft, Sandra G Biedron, Stephen V Milton, and PJ van der Slot. 2017. “Using Neural Network Control Policies for Rapid Switching Between Beam Parameters in a Free-Electron Laser.” In Proceedings of the 31st Conference on Neural Information Processing Systems (NIPS), Long Beach, CA, USA. Vol. 8.
Graaff, Robert J. Van de. 1931. “A Convenient Type of Electrostatic Generator Which Produces Large Quantities of High Voltage.” Physical Review 38: 1919. https://doi.org/10.1103/PhysRev.38.1919.
Hazelwood, Kyle, Mark Austin, Michelle Ibrahim, Han Liu, Seda Memik, Vladimir Nagaslaev, Aakaash Narayanan, et al. 2021. “Real-Time Edge AI for Distributed Systems (READS): Progress on Beam Loss de-Blending for the Fermilab Main Injector and Recycler.” Fermi National Accelerator Lab.(FNAL), Batavia, IL (United States).
Lawrence, Ernest O., M. S. Livingston, H. J. O’Neill, and David Sloan. 1932. “Production of High Speed Light Ions Without the Use of High Voltages.” Physical Review 40 (1): 19–35. https://doi.org/10.1103/PhysRev.40.19.
Martel, Jean-Baptiste de, Roman Gorbonosov, and Nico Madysa. 2022. “Machine Learning Platform: Deploying and Managing Models in the CERN Control System.” JACoW, 50–55.
Pugliese, Raffaele, Stefano Regondi, and Riccardo Marini. 2021. “Machine Learning-Based Approach: Global Trends, Research Directions, and Regulatory Standpoints.” Data Science and Management 4: 19–29.
Rutherford, Ernest. 1909. “The Scattering of \(\alpha\) and \(\beta\) Particles by Matter and the Structure of the Atom.” Philosophical Magazine 27 (6): 488–98. https://doi.org/10.1080/14786440908637080.
Skowroński, Piotr, Matthew Fraser, and Isabella Vojskovic. 2022. “Experience with Computer-Aided Optimizations in LINAC4 and PSB at CERN.” JACoW IPAC 2022: 945–48.
Zhang, Z, AL Edelen, JR Garrahan, Y Hidaka, CE Mayes, SA Miskovich, DF Ratner, et al. 2022. “Badger: The Missing Optimizer in ACR.” SLAC National Accelerator Lab., Menlo Park, CA (United States).