多人非合作随机自适应博弈(英文)

Non-cooperative stochastic adaptive multi-player games

本文考虑系数未知的离散时间线性随机系统多人非合作的自适应博弈问题,每个参与者运用最小二乘算法和"必然等价原则"来设计博弈策略组合,目的是自适应优化自身的一步超前收益函数.本文证明此自适应策略组合使得闭环系统全局稳定,并且在一定意义下是该博弈问题的渐近纳什均衡解.

In this paper, we consider non-cooperative stochastic adaptive multi-player games described by linear discrete-time stochastic systems with unknown parameters. The least-squares algorithm together with the certainty equivalence principle is used by each player in designing the strategy for optimizing its own one-step-ahead payoff function.It will be shown that the resulting adaptive strategy profile can make the closed-loop system globally stable and at the same time, the profile converges to an asymptotic Nash equilibrium in some sense.