On the Global Convergence of Stochastic Fictitious Play

Abstract

We establish global convergence results for stochastic fictitious play for four classes of games: games with an interior ESS, zero sum games, potential games, and supermodular games. We do so by appealing to techniques from stochastic approximation theory, which relate the limit behavior of a stochastic process to the limit behavior of a differential equation defined by the expected motion of the process. The key result in our analysis of supermodular games is that the relevant differential equation defines a strongly monotone dynamical system. Our analyses of the other cases combine Lyapunov function arguments with a discrete choice theory result: that the choice probabilities generated by any additive random utility model can be derived from a deterministic model based on payoff perturbations that depend nonlinearly on the vector of choice probabilities.

Multiple equilibria and limit cycles in evolutionary games with Logit Dynamics

“…A large body of the research on evolutionary game dynamics has focused on identifying classes of games and dynamics that ensure convergence to point-attractors such as the Nash equilibrium (Hofbauer and Weibull (1996), Hofbauer and Sandholm (2002), Sandholm (2005)). …”

Evolution of social networks

“…If i gets a revision opportunity, he draws an action from the distribution b ε i (·|x) ∈ ∆(S i ). The mapping b ε i : X → ∆(S i ) is called the choice function Hofbauer and Sandholm (2002) of player i. Choice functions are basic objects in game 26 Hence, we think of the mappings σ and γ as the projection mappings onto their relevant factors.…”

Non-cooperative power control for wireless ad hoc networks with repeated games

“…Given the different time-varying filter parameters in (21) and (22), the algorithm inherently a two-time-scale dynamics with average utility estimate dynamics (fast dynamics) and probability distribution dynamics (slow dynamics).…”