M4CAST Annual meeting
Bat 102 - Salle DA
IJCLab
M4CAST (Multiphysics Modelling, Machine learning and Model-based Control in Accelerator Sciences and Technologies) is a French network dedicated to applications of artificial intelligence to accelerator sciences and technologies.
This year's annual meeting will group institutional contributors to share their work and discuss synergies and collaborations.
Many subjects will be discussed including but not limited to :
- Computing resources
- Storage resources
- Data catalogues structuration
- Multi-physics simulation
- Anomalies detection
- Optimization techniques
- European opportunities
This year's annual meeting will be held at IJCLab in Orsay, Building 102, Room DA.
-
-
08:30
→
08:45
Accueil café 15m
- 08:45 → 08:55
-
09:00
→
09:40
Laser Wakefield Acceleration: AI for Laser WakeField Acceleration
Session dedicated to IA techniques used for laser wakefield acceleration (LWFA) in the French community
Président de session: Kevin CASSOU (IJClab - CNRS)- 09:00
-
09:20
AI in LWFA control and operation 20mOrateur: Kevin CASSOU (IJClab - CNRS)
-
09:40
→
10:20
SR Light sources: AI for Synchrotron Light SourcesPrésident de session: Laurent NADOLSKI (Synchrotron SOLEIL)
- 09:40
-
10:00
PyAML-Net EU COST Action 20m
The Python Accelerator Middle Layer Network (PyAML-Net) Action has been submitted recently. We are presently waiting for evaluation. If approved we would receive networking money for workshop, trainings, short term scientific missions and schools. Also, if the Action is approved any one will be able to join the network via a simple action on the cost.eu webpage.
The project aims at the developments of a joint technology platform for accelerator tuning, commissioning simulations and digital twin/shadows in python. The platform will be agnostic of the specific accelerator and control system, thus usable (after adequate configuration) by any laboratory.Orateur: Dr Simone Maria-Liuzzio (ESRF)
-
10:20
→
10:35
Pause 15m
-
10:35
→
12:05
Colliders, LINACs and Cyclotrons: AI for colliders, LINACs and cyclotronsPrésident de session: Barbara Dalena (IRFU)
- 10:35
- 10:50
- 11:05
-
11:20
Optimisation d'un jumeau numérique par utilisation des techniques d'inférence Bayésienne 15m
On propose d’explorer une méthode d'amélioration continue d'un jumeau numérique d'accélérateur linéaire en s’inspirant des techniques d'inférence Bayésienne. Initialement, les liens entre les paramètres physiques et de contrôle sont estimés avec des incertitudes. En s'appuyant sur les résultats expérimentaux et les simulations numériques, nous ajustons progressivement ces liens pour minimiser l'écart entre mesures réelles et simulées. Chaque expérience permet d'affiner le modèle numérique, en tenant compte des incertitudes et en améliorant la correspondance avec le comportement réel du linac. Cette approche itérative optimise le jumeau numérique et permet d'intégrer de nouveaux paramètres sans perdre les connaissances acquises.
Orateur: Dr Jonathan Dumas (CEA) -
11:35
Anomalies detection @ ARRONAX 15mOrateur: Freddy Poirier (CNRS/Arronax)
-
11:50
Session questions 15m
-
12:05
→
13:05
Technologies, computing and storagePrésident de session: Hayg Guler (IJCLAB)
-
12:05
ALESIA: Superconducting magnet design through multi-physics optimisation 20m
Designing superconducting magnets presents a challenge due to their multi-physics complexity, diverse analytical tools, and often imprecise specifications. To streamline this process, we introduce ALESIA, a novel optimisation and data management toolbox developed at CEA-IRFU.
ALESIA leverages advanced algorithms, including nonlinear programming techniques, evolutionary algorithms, active learning strategies, and surrogate modelling, to accelerate the design process. By intelligently exploring the parameter space, ALESIA enables rapid convergence towards optimal solutions while minimizing computational cost.
ALESIA's flexible architecture allows integration with any physics simulation software, encompassing magnetic field calculations (OPERA), and mechanical analysis (CAST3M), but its applicability can be broadening beyond magnet design. Crucially, ALESIA's automated optimisation loop simultaneously considers all stages - magnetism, conductor properties, mechanics, and quench behaviour - ensuring holistic and robust design solutions.
Orateur: Damien Minenna (CEA, Irfu) - 12:25
-
12:45
Computing and storage @ CC IN2P3 20mOrateur: Thomas Kachelhoffer (CC-IN2P3-CNRS)
-
12:05
-
13:05
→
14:05
Repas 1h
Paniers repas sur place
-
14:05
→
15:35
Discussion session: Discussions : computing and storagePrésident de session: Kevin CASSOU (IJClab - CNRS)
-
15:35
→
16:00
Discussion session: Discussion storage : IN2P3 transverse project 1/2Président de session: Dr Adnan Ghribi ({CNRS}UPR3266)
-
16:00
→
16:15
Pause 15m
-
16:15
→
16:50
Discussion session: Discussion session : IN2P3 transverse projectPrésident de session: Dr Adnan Ghribi ({CNRS}UPR3266)
-
16:50
→
17:50
Discussion session: Discussion session : 2025 prospectsPrésident de session: Hayg Guler (IJCLAB)
-
17:50
→
18:10
Cloture 20mOrateur: Dr Adnan Ghribi ({CNRS}UPR3266)
-
18:10
→
20:10
Dîner 2h
-
08:30
→
08:45