Domain #4: Optimisation

Home > Areas of competence > Optimisation

Shifting from the descriptive to the prescriptive

The understanding of the world through the lenses of modelling and simulation is a classic paradigm, renewed by the level of complexity of the systems that make it up today.
In order to move from the descriptive to the prescriptive, the exploration of modelling spaces and the identification of sub-spaces of interest are carried out using optimisation methods adapted to specific use cases.
These methods must be capable of handling large amounts of data, the uncertainties attached to them and the associated time constraints.

Challenges

Companies and public players must optimise the design and operations of complex systems such as transport or energy infrastructures, using simulation methods and algorithms.

Positioning of the institute

Optimisation irrigates many related scientific fields (scientific computing, machine learning, networks, etc.). IRT SystemX has a substantial research force in this field to provide state-of-the-art solutions, but also to carry out more upstream research on large systems, stochastic problems, sequential decision-making, solvers adapted to machine learning and finally on simulation by multi-agent systems, which are very well suited to the resolution of use cases in various fields (transport, logistics, energy, etc.).

Roadmap

Scientific and technological challenges	Related research fields
Optimization for large problems	• Coupling of solvers and simulation tools • Optimisation of resource constrained problems • Reduction of the computational burden • Design and automatic training of neural networks (solver)
Dynamic stochastic optimisation	• Time constrained solving • Sequential decision making and learning • Dynamic grouping of tasks • Hazard management
Simulation and optimisation using multi-agent systems	• Improvement of multi-agent simulation tools for scenario analysis • Usability and replicability of Multi-Agent Systems models