Overview

The ASCII team investigates stochastic particle systems which behave like agents interacting in a stochastic and intelligent manner. These agents strive to cooperate optimally towards a common objective by solving strategic optimisation problems. ASCII aims to develop models in this direction, as well as their theoretical and numerical analyses, with target applications in neuroscience, physics, biology, networks and stochastic numerics.

In addition to the modelling challenges raised by our innovating approaches for handling intelligent multi-agent interactions, we will have to develop new mathematical theories and numerical methods to deal with interactions which, in most of the interesting cases we have in mind, are irregular, non Markovian, and dynamical. Original and non standard stochastic control and stochastic optimization methodologies are being developed, combined with original calibration methodologies.

To reach our objectives, we combine techniques at the intersection of stochastic analysis, partial differential equation, numerical probability, optimization theory, and stochastic control theory.

Research directions

• Stochastic cooperative numerical methods for singularly interacting particle systems and uncertainty analyses
• Particle filtering for Bayesian estimation
• Networks with interacting agents
• The ICI project: agent-based epidemic propagation simulator relying on detailed geographic and sociological data