Photons and Neutrons Realistic Artificial Intelligence Datasets (PaNRAID)

DIADEM Academy — Training course on synthetic data generation for supervised learning
📅 21–25 September 2026 | 📍 CNRS CAES Village, St-Pierre d'Oléron

Objective

This course develops an integrated approach to generating synthetic data for supervised learning, combining multi-scale material simulations (DFT, MD, XAS spectroscopy) with comprehensive digital twins of experimental X-ray and neutron facilities — including instrumental effects and experimental artefacts.

See Data_Generation_Pipeline.md for how a single McStas/McXtrace simulation run, made throughout Days 2–3, becomes one labelled record in a Day 4/5 training dataset — the explicit bridge between the instrument-side and AI-side halves of the programme below.

Audience

• Doctoral and post-doctoral students
• Teachers-researchers
• Researchers / research engineers

Prerequisites

• English proficiency at B2 level (course delivered in English)
• Basic knowledge of X-ray and/or neutron instrumentation
• Familiarity with scientific data processing tools: Python, numerical computation, simulation
• Personal laptop with McStas and McXtrace pre-installed, at best via these instructions.

Programme Overview

Day	Date	Session	Topic
1	Mon 21 Sept	14:00-17:00	Introduction
		lecture	21>01 Seeking for AI (PDF slides) (pptx)
		lecture	21>02 Intro and General Concepts (PDF slides) (pptx)
			21>02.1 vibe-code examples (PDF) (pptx)
		setting goals	PaNRAID team challenge
		software/tech status?	5-minute software/tech status
2	Tue 22 Sept	09:00-12:00	Sources, Detectors & Optics
		lecture	22>03 Sources and Monitors (PDF slides) (pptx)
		practicals	22>03 Sources and Monitors
		lecture	22>03 Optics (PDF slides) (pptx)
		practicals	22>03 Optics
2	Tue 22 Sept	14:00-17:00	Intro to AI
		lecture	22>04 Into Deep Learning (PDF slides)
		practicals	22>05 Samples: SANS / Data generation
3	Wed 23 Sept	09:00-12:00	Samples 1: Diffraction and Imaging
		lecture	23>06 Samples (PDF slides) (pptx)
		practicals	23>06 Samples: Powder Diffraction
		practicals	23>06 Samples: Imaging
		practicals	23>07 Samples: Absorption
3	Wed 23 Sept	14:00-17:00	Samples 2: X-ray Spectroscopy and SANS
		practicals	23>07 Samples: Fluorescence
		practicals	22>05 (cont.) SANS Inverse problem classifier Notebook
4	Thu 24 Sept	09:00-12:00	AI Applications 1
		practical	24>08 Optimisation
		practical	24>08 Image processing - De-noising/de-convolution/segmentation
		practical	24>08 Surrogate (approximator)
4	Thu 24 Sept	14:00-17:00	AI Applications 2
		practical	Team Challenge work
		excursion	Social activity: Fort Boyard ⛵ ➡️ 🏰
5	Fri 25 Sept	08:30:11:30	Team Challenge wrap-up

Practical Details

• Duration: 5 days — 8 half-days (28 hours total)
• Modality: In-person (présentiel)
• Price: €700 which includes return shuttle to La Rochelle, 4 nights accommodation, meals 21st evening – 25th midday, bike hire, group outing at Fort Boyard.

🚌 The shuttle on Sept 21st is planned at 1pm from La Rochelle (right side from the train station exit, Bd Joffre next to Hotel B&B - bus will be labelled 'SPECIAL').

🚌 The shuttle on Sept 25th is planned at 11:45am to La Rochelle train station.

⛵ The boat trip on Sept 24th in planned at 5:45pm from Boyardville (2 km away by bike), 1h sea cruise around Fort Boyard.

Trainers

Name	Affiliation
Peter Willendrup	Senior Research Engineer, DTU Physics / ESS DMSC — McStas lead developer since 2002, McXtrace co-developer since 2009
Emmanuel Farhi	Head of Data Reduction and Analysis, SOLEIL Synchrotron — McStas/McXtrace contributor
José Robledo	Researcher, CONICET / Balseiro Institute, Argentina — AI & neutron science, ML for neutron data analysis

Registration

Contact: Elodie ISTE
📞 05 87 50 23 32
📧 diadem-formationcontinuecontact@unilim.fr
🌐 https://formation.pepr-diadem.fr/les-formations

This school receives moral support from GDR 2123 IAMAT