Optimization: A Key EDF Competency at the Heart of the PGMO Programme
The PGMO programme brings together mathematicians from a wide range of backgrounds to work on a major topic for EDF: optimization. Involving, in particular, the Institut Polytechnique de Paris, this unique partnership is based on a philanthropic programme launched by EDF in 2012 under the auspices of the Fondation Mathématique Jacques-Hadamard.
“Optimization has always been a meeting ground for bilateral academic collaborations,” explains Sandrine Charousset, who initiated the programme in 2012. “As our needs in this area of expertise grew, we set up this programme to build long-term partnerships and open it up to other research teams.”
EDF’s R&D: A Partner of the FMJH and Institut Polytechnique de Paris
Duration: 4:43
Your browser does not support javascript.
To enable you to access the information, we suggest you view the video EDF’s R&D: A Partner of the FMJH and Institut Polytechnique de Paris in a new tab.
[This testimonial presents the partnership between EDF’s R&D, the FMJH and Institut Polytechnique de Paris through the PGMO programme, which brings together academic and industrial researchers to address energy-related optimization challenges. It features Stéphane Gaubert (Inria Research Director, CMAP, École Polytechnique and PGMO programme coordinator) and Wim Van Ackooij (expert research engineer at EDF’s R&D division and PGMO programme liaison officer).]
Our pitch
Stéphane Gaubert: Our aim is to bring academic and industrial research together in the field of optimization. We bring mathematicians and computer scientists together to address industrial challenges, particularly in the energy sector.
Our partnership is based on a unique philanthropic programme established in 2012 at EDF’s initiative, with the involvement of researchers from engineering schools that are now part of Institut Polytechnique de Paris. The programme is supported by the Fondation Mathématique Jacques-Hadamard.
In practical terms, we issue calls for proposals and organize conferences and advanced courses for PhD students and researchers, always with the aim of addressing genuine academic and industrial challenges.
Wim Van Ackooij: Today, our partners come not only from the French academic community, but also from across Europe and even further afield.
Stéphane Gaubert: We work on a wide range of topics related to energy management. For example, optimizing nuclear power plant refuelling schedules or, in the context of the energy transition, making the best possible use of renewable energy sources.
Our achievements
Wim Van Ackooij: The partnership has also helped us identify new research areas, such as game theory. Game theory can, for example, help us make better use of low-carbon energy sources.
Stéphane Gaubert: Mean field game theory, a relatively recent development, helps us understand the macroscopic effects of interactions among large numbers of individual agents. For example, it can help us assess the impact on the electricity grid of decisions about when to charge batteries.
Wim Van Ackooij: The PGMO partnership demonstrates the vitality of mathematical research and its potential for practical applications in industry.
Our challenges
Wim Van Ackooij: One of the main challenges in optimization clearly stems from practical constraints. For example, when we need to account for the operating status of a power plant, it is either switched off or running. This is a combinatorial, binary decision: there is no in-between, which makes the problem quite difficult.
Stéphane Gaubert: Indeed, a fundamental challenge is what we call combinatorial explosion, caused by the discrete nature of decisions. If the problem is not approached correctly, execution time can grow exponentially.
Wim Van Ackooij: Another challenge is accounting for uncertainty. In an optimization problem, we seek the best possible solution while ensuring that it remains robust in the face of uncertainties that may arise in real time.
Our solutions
Wim Van Ackooij: If EDF had to solve all the challenges posed by the energy transition on its own, using optimization alone, we would need a team almost ten times larger than our current one. The partnership’s open approach enables highly fruitful collaborations with the academic community, helping new solutions and ideas spread more quickly than they would through conventional channels.
Stéphane Gaubert: In particular, the PGMO programme helps introduce pure mathematicians to energy-related challenges. Some of the problems we face are longstanding, with some remaining open for 30 years. By drawing on fundamental mathematics, we hope to make progress.
Our partnerships
Stéphane Gaubert: We bring together researchers with very different backgrounds, ranging from pure mathematics to applied mathematics and computer science. Some work independently, with just paper and pencil, while others, particularly those working on applied topics, collaborate in teams, which is essential for software development, for example. One of the PGMO programme’s aims is precisely to bring these diverse profiles together.
Wim Van Ackooij: EDF, and its R&D division in particular, has a long-standing scientific tradition. One example is Michel Gondran, one of our most brilliant minds, who worked on seminal papers in operations research and combinatorial optimization as early as the 1970s.
We should also mention Marcel Boiteux, who first worked as an engineer in EDF’s R&D division before later becoming a senior executive at the company. He published seminal papers on marginal-cost electricity pricing.
Today, through the PGMO programme, we play an active role in fostering scientific collaboration and driving research in the field of optimization.
The Foundation promotes mathematical excellence. Established in 2011 on the Saclay plateau, it provides an ideal environment for fostering synergies between academic and industrial research. The principles of collaboration were quickly established: participation through calls for proposals and openness to mathematicians from all countries.
“The PGMO programme enables us to launch calls for proposals to address some of the major mathematical optimization challenges facing EDF, and select research teams interested in these topics,” explains Sandrine Charousset. “All results are publicly available and can be used by anyone, including EDF, of course.”
The need for optimization has always existed, but optimization problems have become increasingly complex over time. The energy transition across Europe, with the uncertainties associated with renewable energy sources and the need to account for different energy vectors, gives rise to enormous challenges. This is where stochastic optimization comes into play, aiming to calculate optimal decisions that remain robust in the face of uncertainty.
“When we want to simulate a system in which 80–90% of energy production comes from renewable sources, the resulting variability is enormous,” explains Sandrine Charousset. “How can we incorporate the management of a very large number of flexibility resources into our optimization problems, such as batteries or flexibility in residential energy use? We break the problem down and launch research projects on the different aspects.”
The programme covers the energy transition, of course, but also energy management, competition between players in energy markets, and support for the Group’s business activities. Its research areas are wide-ranging. “EDF cannot solve these problems on its own,” acknowledges Wim van Ackooij, expert research engineer at EDF’s R&D division and EDF’s liaison officer for the PGMO programme. “Collaboration with the academic community brings solutions, new ideas and more efficient algorithms.”
But new solutions are not the programme’s only benefit. “It attracts talented PhD students who produce high-quality theses,” says Wim van Ackooij. “It demonstrates both the vitality of research and the wealth of career opportunities in industry, where researchers can tackle major societal challenges. This is mathematics that is both rigorous and useful.” The selected projects, submitted by internationally renowned teams from Saclay to Scotland and New Zealand, also help raise the PGMO programme’s profile around the world.
Collaboration with the academic community brings solutions, new ideas and more efficient algorithms.
The PGMO Programme
The PGMO programme (Gaspard Monge Programme for Optimization, Operations Research and Their Interactions with Data Science) is a programme of the FMJH (Fondation Mathématique Jacques-Hadamard) that is open to corporate sponsorship.
The PGMO covers all areas of decision mathematics, including the following topics: modelling, continuous optimization, large-scale system optimization, combinatorial optimization and operations research, optimization under uncertainty, global optimization, game theory, constraint programming, and related fields such as statistical learning and data science.
EDF: A Longstanding Expert in Mathematical Optimization
EDF Group can pride itself on a strong tradition of mathematical excellence, particularly in optimization. “EDF’s first publications on optimization date back to the 1950s,” explains Sandrine Charousset. “At the time, the focus was on optimizing power plants, dams and major infrastructure.”
Today, EDF’s R&D division has developed this expertise to such an extent that it has been able to incorporate game theory into grid management and offer pricing. “This theory makes it possible to address multi-agent interactions, take customer responses into account when setting prices, and balance supply and demand,” explains Stéphane Gaubert, Research Director at Inria, based at École Polytechnique, Institut Polytechnique de Paris, and coordinator of the PGMO programme. “Its application in the energy sector is genuinely new and also draws on the most advanced theoretical developments.”
“Some of our brightest minds, such as mathematician Michel Gondran, who spent his entire career at EDF’s R&D division, have published seminal papers in this field,” adds Wim van Ackooij. “As such, we play an active role in bringing the scientific community together and fostering collaboration.”
The PGMO programme brings together a remarkably diverse range of expertise, “from pure mathematicians to applied mathematicians and computer scientists,” explains Stéphane Gaubert. “Its purpose is precisely to enable researchers from all these different backgrounds to exchange ideas.” An annual meeting gives researchers the opportunity to meet in person, complementing their day-to-day collaboration on various projects.
EDF’s investment also benefits the academic discipline. “We have helped open up new avenues of research,” says Wim van Ackooij. “Our work has helped generate and disseminate new academic knowledge in mathematical optimization,” which is now recognized internationally.
Misconception: Researchers Work in Isolation
The life of a researcher remains something of a mystery. “We tend to picture free spirits, accountable to no one but their own intellect, spending long hours wrestling with a problem on paper while drinking copious amounts of coffee,” Stéphane Gaubert acknowledges with a smile. “And that is certainly true. But even in fundamental research, progress is often only possible when such individual efforts are combined with a collective dynamic. This is all the more true for applied projects, which sometimes involve interdisciplinary or computer science components: teamwork is essential.”
The PGMO programme illustrates the many forms of cooperation that are essential to research. First, there is collaboration between industry and academia. Industrial partners provide case studies and data that researchers greatly value. Meetings are organized through dedicated events or simply as working sessions. The physical proximity of researchers and industry professionals, for example on the Saclay plateau, also encourages exchanges. Collaboration can also span generations: today's researchers inevitably build on the achievements of the great minds who came before them, such as Gaspard Monge, after whom the programme is named. Put simply, the deeper a tree's roots, the stronger it stands!
Finally, the PhD thesis, a widely used tool in these partnerships, is very much a collective endeavour. “Although the work centres on the doctoral student's research, they collaborate with their thesis supervisor, other PhD students, industrial mentors and other partners who can contribute to the project,” explains Ange Caruso, Partnerships Delegate at EDF's R&D division. “Sometimes, a thesis also draws on other areas of expertise, such as social sciences, physics or chemistry.” In short, there is always a good reason to collaborate in research.
