Postdoctoral researcher in accelerator physics: AI-based optimisation of the CIME cyclotron (European iRIS project) (M/F) - Ref. 478

Cnrs Inc; France; On-site; EUR 42,000 - 65,000; Full time

Benefits offered by this job

53 days leave per year

Collective supplementary health insur.

Meal cost contribution

Transport costs or mobility allowance

Social Action Committee

Job summary

GANIL, a CNRS-led national research facility in Caen, France, seeks a Researcher with a PhD in accelerator physics or related field and strong ML expertise. The role focuses on AI-based optimisation of accelerator performance and energy efficiency within the iRIS project.

You will work with EPICS-based control systems and existing accelerator codes, contributing to deliverables and conferences. The candidate will report to the Accelerator Physics Group, interact with WP2 partners, and operate in

Qualifications

• PhD in accelerator physics, applied physics or a related field.
• Proven experience in applying machine learning to accelerators.
• Ability to present results in meetings and write progress reports.

Responsibilities

• Lead ML-based optimisation of CIME settings and surrogate models.
• Coordinate with operations, GANIL control room and WP2 partners.
• Document results for IPAC/ICALEPCS/MALAPA and deliverables.

Skills

Python

Machine learning

Beam dynamics

English fluency

Education

PhD in accelerator physics, applied physics or related field

Tools

XSuite/Cheetah

TraceWin

MAD-X

EPICS

Job description

Organisation/Company CNRS Department Grand Accélérateur National d'Ions Lourds Research Field Physics Researcher Profile Recognised Researcher (R2) Application Deadline 26 Nov 2026 - 23:59 (UTC) Country France Type of Contract Temporary Job Status Full-time Hours Per Week 35 Offer Starting Date 4 Jan 2027 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

GANIL is a partner in the European iRIS project (intelligent Research Infrastructure Sustainability, Horizon Europe, coordinated by CERN) ( https://www.irisproject.eu ). The project aims to reduce the environmental footprint of large research infrastructures using artificial intelligence methods. CNRS leads work package WP2 GreenBeam, dedicated to the AI-based optimisation of particle accelerator sustainability. The missions for this project are:

• Develop and validate the machine learning and optimisation methods needed to reach the project's contractual objective for the site: reduce set-up times by 30-50 %, by building surrogate models of the machine and closing the optimisation loop on the control system.
• Lead GANIL's scientific contribution to the deliverable 'Operational accelerator data and instrumentation toolkit' (M13), for which CNRS is responsible, and contribute to the deliverables of the relevant work package.

1. Instrumentation and acquisition of CIME operational data (task T2.2, led by CNRS)
   • Define and implement the measurement procedures and instrumentation that link machine settings and beam performance on CIME; establish the pre-processing protocols and the associated feature sets.
   • Feed the project's FAIR (Findable, Accessible, Interoperable, Reusable) feature store and ensure the transferability of data and models to the other infrastructures of the consortium.
2. Surrogate models and AI-based optimisation of CIME settings (task T2.3)
   • Develop surrogate models of CIME from operational data and beam dynamics codes, and integrate them into the design (XSuite/Cheetah/in-house) and optimisation (GeOFF) frameworks selected by the project.
   • Implement optimisation algorithms (Bayesian optimisation, reinforcement learning) for assisted machine tuning, and close the loop with the control system following the scheme adopted by the iRIS project: the algorithms propose the settings, the operator validates, the machine adapts.
3. Reduction of CIME energy consumption and set-up times
   • Characterise and reduce the sources of energy consumption related to operation; quantify the gains against the baseline, aiming at the project's contractual targets for the machine.
   • Act as the interface with the operations teams and the GANIL control room; document and publish the results (IPAC/ICALEPCS/MALAPA conferences, journals); take part in the iRIS consortium meetings and ARTIFACT events (artifact-network.org).

The Grand Accélérateur National d'Ions Lourds (GANIL) is a national research infrastructure based on the use of ion beams. Research fields include fundamental research in nuclear physics and nuclear astrophysics, materials under irradiation and nanostructuring, molecular collisions and the interstellar medium, radiobiology, and innovative techniques for dosimetry and the treatment of certain cancers.
GANIL (about 350 staff) is located on the future EPOPEA site, a science and innovation park of the Caen la Mer urban community, in Caen, Normandy. It is jointly managed, within an Economic Interest Group (GIE), by the French Alternative Energies and Atomic Energy Commission (CEA/DRF) and the French National Centre for Scientific Research (CNRS/IN2P3). As a research infrastructure, GANIL serves a national, European and international scientific community of about a thousand users.

In an international working environment, the successful candidate will report to the Accelerator Physics Group of the Operations and Development Division, and will work in coordination with the WP2 partners.
The CIME cyclotron (Cyclotron à Ions de Moyennes Énergies, K = 265, post-accelerator of SPIRAL1, in operation since 1997, 1.7 to 25 MeV/A) forms, together with the CSS1 and CSS2 cyclotrons, GANIL's set of historical machines. Their age and their weight in the site's energy consumption make them the priority target for energy efficiency work. The project's contractual objective for the site is to reduce set-up times by 30-50 %.

Employee benefits:

• 53 days of leave per year (32 days of annual leave and 21 RTT days)
• Collective supplementary health insurance
• Contribution to meal costs
• Contribution to public transport costs or sustainable mobility allowance
• Social Action Committee

Field of expertise:

• Accelerator physics, beam dynamics, machine learning applied to accelerators (surrogate models, time series), optimisation (Bayesian, reinforcement learning). Experience in cyclotron tuning and operation is a plus but not mandatory.

Job level:

• PhD in accelerator physics, applied physics or a related field, with proven experience in machine learning applied to accelerators (or vice versa).

Know-how:

• Proficiency in Python and the ML ecosystem (PyTorch or equivalent, scikit-learn); optimisation tools applied to accelerators (Xopt, Bayesian optimisation, reinforcement learning). Knowledge of GeOFF/Badger and XSuite/Cheetah is a plus.
• Experience with beam transport and beam dynamics codes (TraceWin, MAD-X, XSuite or equivalent). Knowledge of accelerator control systems (EPICS) is a plus.
• Analysis of operational data (archives, diagnostics, energy consumption measurements); instrumentation and acquisition; rigorous validation in an operational environment.
• Ability to present results in meetings and at conferences, and to write progress reports, European project deliverables and articles. Fluency in English is mandatory (spoken and written); French is a plus (interactions with the operations teams).

Soft skills:

• Ability to work in a diverse (physicists, operators, software engineers) and international team; autonomy and adaptability.
• Organisation and sense of priorities, within the contractual deadlines of a European project; initiative and commitment to supporting operations.