Stochastic models, multi-agent models and social simulations

This theme focuses on modeling and simulation approaches, particularly multi-agent, for complex human, social and economic systems.

Scientific referents

Coordinator  : Julien Randon-Furling

 

Presentation


Many human systems present challenges in mathematical modeling and scientific analysis similar to those encountered for complex physical, natural and biological systems. The work in this theme therefore focuses, for objects in the human and social sciences, on developing and studying approaches to stochastic modeling and numerical simulations similar to those used, for example, in statistical physics.

We focus in particular on establishing a formal framework for determining, analytically or by simulation, phase diagrams describing the behavior of complex systems of human, social, historical and/or economic interactions.

We also seek to define a theoretical framework for the use of multi-agent models or other types of simulations for prediction or causal inference purposes. We also model network diffusion dynamics and socio-geographic phenomena such as segregation.

Finally, we study the properties of some stochastic processes (Markov chains, random walks, Lévy processes) involved in our models, in particular questions of extreme values, first passage time, rare events and stochastic geometry.

Key words

Complexity; Stochastic models and processes; Multi-agent models; Simulations; Collective dynamics; Causal inference; Stochastic geometry

Key facts

Applications

The work carried out in this theme provides a theoretical basis for modeling real socio-economic systems, predicting their behavior, testing possible intervention scenarios and thus being able to offer assistance in decision-making in public policy and/or in the management of collective systems in the fields of Public Health and Prevention, Guidance Sciences and Educational Transitions and in Mathematics and human and social science interactions.

Topics covered

  • Complex systems: phase diagrams, phase transitions, criticality, emergence, simulations, multi-agent models, hybrid models

  • Stochastic processes: reaching and first passage times, extreme values, rare events, random convex hulls

International partners