Description of the event

CEMRACS (Centre d’Été Mathématique de Recherche Avancée en Calcul Scientifique) is an annual summer school and research program held in July and August at the Centre International de Rencontres Mathématiques (CIRM) in Marseille, each year focusing on a specific theme. The event consists of a one-week summer school followed by a five-week research program, during which selected participants work on research projects supervised by academic researchers or industry professionals.

The 2026 edition focused on modelling and AI for the environmental transition, with a multidisciplinary approach combining climate and environmental economics, climate science, ecology, green finance, insurance, and applied mathematics. It was organized by a committee comprising Aurélie Fischer (Université Paris Cité), Pierre Gruet (EDF R&D), Ziad Kobeissi (INRIA), Olivier Lopez (ENSAE), Gauthier Vermandel (École Polytechnique), and Olivier David Zerbib (ENSAE), and chaired by Peter Tankov (ENSAE).

Summer school

The summer school consisted of five mini-courses and for special invited lectures delivered by the leading researchers:

• Mathieu Barbier: Mathematical Modeling for Biodiversity (mini-cours)
• Freddy Bouchet: Mathematical Modeling for Climate: Stochastic Dynamics, Extreme Events, Large Deviations (mini-cours)
• Olivier Lopez: Climate Risks, Insurance, Insurability and Rare Events (mini-cours)
• Claire Monteleoni: Machine Learning for Climate Change (mini-cours)
• Simon Scheidegger: Computational and AI Methods in Climate Economics (mini-cours)
• René Aïd: Mathematical Models for Carbon Emission Markets (special invited lecture)
• Luc Doyen: Mathematical Bioeconomics for Sustainable Fisheries (special invited lecture)
• Roxana Dumitrescu: Mean-Field Game Models for Energy Transition (special invited lecture)
• Jan Polcher: Modeling of the Water Cycle at the Interface Between Climate, Hydrology, and Socio-Economy (special invited lecture)

The talks have been recorded and are now available online. 

Research session

Around half of the participants of the summer school (38 participants in total) were selected to stay for the 5-week research session. They worked in groups of 2 or 3 on the following research projects, proposed by French or international researchers, research chairs and industrial partners:  

• Characterizing predator-, prey- and self-control in complex food webs
• Modeling of extremes of electricity demand
• Dynamic Top-Down Model for Financial Loss Distribution Under Climate, Credit & Market Risks
• Enhancing the temporal resolution of climate model outputs represented by diffusion processes
• Sequential Stochastic Optimization of Shipping Routes Under Time-Evolving Metocean Forecasts
• Mean-field models for fishery management
• Inverse Problem for a Production–Consumption Market Model
• Numerical Methods for Stackelberg Mean Field Games with Common Noise
• Dynamic Contracts for Credible Climate Impact: Principal–Agent and Mean-Field Methods for Blended Finance
• Optimizing reinsurance protection in case of loss of mutualization
• Generative AI and scenario generation in insurance
• Numerical Methods for Credit Risk Management under Stochastic Climate Scenarios and Model Uncertainty
• Localized impacts of transition risks: causal evidence and prospective analysis
• Quantifying Economic & Financial Impacts of Droughts
• Neural-Network Meta-Models for Reduced-Complexity Climate Models

The results of the projects will be published in a special issue of ESAIM Proceedings and Surveys