## Nash Certainty Equivalence in Large Population Stochastic Dynamic Games: Connections with the Physics of Interacting Particle Systems |
Back |

Abstract:We consider large population stochastic dynamic games of the form which occur, for instance, in decentralized power control in wireless communication systems. Beginning with the simpler class of LQG problems, it is shown by fixed point arguments that the dynamics of each individual of the mass can be modeled by the McKean-Vlasov equation found in the statistical physics of interacting particle systems. Based upon this large population modelling, a so-called Nash Certainty Equivalence (NCE) Methodology is introduced for specifying the feedback control law of a given agent within the Nash equilibrium setting. The crucial new feature in the controlled agent setting is that each individual's control law is computed via a generalized amilton-Jacobi-Bellman equation where this equation includes a distribution function representing the behaviour of the mass of the other (self-optimizing) agents. Work with Minyi Huang, ANU, Canberra, and Roland Malhame, Ecole Poly.and GERAD, Mtl. http://www.cim.mcgill.ca/~peterc/ |
Type of Seminar:Public Seminar |

Speaker:Prof. Peter E. Caines Department of Electrical and Computer Engineering and Centre for Intelligent Machines, McGill University, Montreal | |

Date/Time:Sep 25, 2006 17:15 | |

Location:ETH Zentrum, Gloriastrasse 35, Building ETZ, Room E 6 | |

Contact Person:Prof. Morari | |

File Download:Request a copy of this publication. | |

Biographical Sketch: |