Post-Doctoral Research Visit F/M Postdoctoral Fellowship – PGMO Project: From Complementarity to Multi-Agent Learning for Electricity Markets

Descrizione

Testo originale dell'annuncio (in inglese)

This postdoctoral position is part of a project funded by the PGMO program (Gaspard Monge Program for Optimization and Operations Research) . It aims to develop new modeling and optimization methods to address the challenges of the energy transition. The project lies at the intersection of operations research, mathematical optimization, game theory, and artificial intelligence, with applications to electricity markets. The work will be conducted in collaboration with academic researchers and industrial partners, notably EDF, to develop decision-support tools for energy policy. This project examines the investment decisions of electricity generators in decentralized markets subject to uncertainty. It models interactions among generators using stochastic equilibrium games that incorporate long-term contracts, such as capacity mechanisms and contracts for difference (CfDs). The goal is to better understand how these contracts influence investments and the composition of the energy mix. The project combines two complementary approaches: optimization methods based on complementarity problems and multi-agent learning techniques. The latter allow for approximating equilibria when models become too complex for classical methods. The research also takes into account market power, risk aversion, and the presence of multiple equilibria. Finally, the models will be applied to real-world cases, particularly island power systems, to assess the economic impacts and implications for energy policies. Here is a summary of the work program in a few points: WP1 – Model Development : Incorporate long-term contracts, uncertainty, risk, and market power into a stochastic equilibrium model. WP2 – Algorithm Development : Design and compare multi-agent learning methods to calculate equilibria in complex electricity markets. WP3 – Economic analysis : Evaluate the impact of different contracts on investments, the energy mix, and market efficiency using real-world case studies. Validation : Compare the new approaches to traditional optimization methods in terms of accuracy, robustness, and scalability. Dissemination : Publish the results in scientific conferences and journals, in collaboration with industrial partners such as EDF.