國立東華大學應用數學系

專    題    演    講

主講人：翟健

          浙江大學數學系

講  題：Stochastic evolutionary models in game

時  間：103 年 09月 26 日 (星期五)  15:00-16:30

地  點：理工一館 A324會議室

摘   要

    This talk examines the problem of equilibrium selection in games by decomposing the corresponding Markov processes into cycles iteratively. Since the pioneering work of Foster and Young (1990) and Young (1993, 1998), stochastically stability is used to explain that among multiple pure strategy Nash equilibria of a game with noise in players' actions, only some equilibria have positive probability as the noise goes to zero. They are known as stochastically stable equilibria. In their papers, the stochastically stable equilibria are computed by solving a series of shortest path problems in a graph building on work of Freidlin and Wentzell (1984) and Kandori, Mailath and Rob (1993) for perturbed Markov processes.

   This talk uses the theory of large deviations to analyse the large time behaviors of the perturbed Markov processes arisen in games with noise as noise goes to zero, as well as the problem of stochastically stable equilibria selection by decomposition the perturbed Markov processes into cycles iteratively. Similar iterative cycles decomposition method was used in simulated annealing problem by Hwang and Sheu (1990), Chiang and Chow (1989,1998).

※※※   歡      迎      參      加   ※※※se1030926   附檔：