不完全博弈学习过程的虚拟行动规则

本文把博弈学习虚拟行动规则的学习时间一般化,研究不完全学习过程中参与人策略选择的收敛性及效用一致性问题. 分析结果表明：当学习时间不完全时，在学习一致到达的条件下,虚拟行动规则对严格纳什均衡是吸收的; 在学习时间充分及时与虚拟行动非频繁转换的条件下,各参与人的虚拟行动具有效用一致性.

Rule of Fictitious Play in the Learning Process with Incomplete Learning Times

In this paper we have generalized the time frequency of learning in the repeated games to study the problems of strategy converging and utility consistency under the fictitious play rule. We have showed that in the process of incomplete learning, the strict Nash Equilibria are absorbing for the fictitious play provided they are played by uniform learning and thefictitious play has the property of utility consistency when the learning times are frequent enough and the fictitious play exhibits infrequent switches.