基于可信度的多次重复博弈研究

Research on Multiple Repeat Games Based on Credibility

提出了一种基于可信度的多次重复博弈算法,通过参与人在每轮博弈中采取的行动策略统计其可信度,参与人下一轮的行动策略与其可信度有关,根据可信度建立相应奖惩机制,单方面提高参与人收益会造成可信度下降,同时证明了在重复博弈中最终趋向于帕累托最优.

This paper proposes a multiple repeat game algorithm based on credibility, calculates the credibility through the acting strategy of participators in each round of game, indicates that the acting strategy of the participators of next round is related to their credibility and that the unilaterally increasing of the participators＇ gain can result in the decrease of their credibility according to the relative rewards and punishment mechanism based on the credibility, and proves that the repeat games ultimately trend to Pareto Optimality.