Geometric convergence of fictitious play for stochastic differential games
Adam Andersson (Saab AB and Chalmers University of Technology)
Abstract: Stochastic Differential Games (SDG) are stochastic control problems, where multiple players can influence or control a joint SDE with possibly different objectives. Fictitious Play (FP) is a classic Piccard type approximation technique from game theory that has been applied in the literature also to SDG. In this talk I introduce SDG, the synthesis through dynamic programing and systems of PDE and FBSDE, and present ongoing work on theoretical convergence properties of FP for SDG. A numerical example confirms experimentally the established geometric convergence. Joint work with Kristoffer Andersson (univ. of Verona) and Per Ljung (Saab AB & the Dept. of Electrical Engineering, Chalmers).
numerical analysisoptimization and control
Audience: researchers in the topic
Series comments: Online streaming via zoom on exceptional cases if requested. Please contact the organizers at the latest Monday 11:45.
| Organizers: | David Cohen*, Annika Lang* |
| *contact for this listing |