Data-driven Harsanyi Transformation And Its Application In Penalty Game Simulations

The classical Harsanyi transformation transforms the standard analysis framework of the incomplete information game,laying the foundation for the generation of information economics,occupying an irreplaceable importance in the history of economic development,but it also encounters the criticism of transcend Rational demands and beliefs update complexity.In this paper,the initial goal is to improve the Harsanyi transform from the viewpoint of data driven together with application in a theoretical model and simulation.Specifically,three aspects of work have been done in this article.First,the idea of data-driven Harsanyi transform is presented,and its logical thinking and implementation framework is analyzed to bring the equilibrium results closer to reality.Second,in order to test the effect of data-driven Harsanyi transform method,the penalty game is analyzed from the perspective of incomplete information repetitive game.At first,the parameterization of the hit rate is used to convert the incomplete information into a parametric game,and then complete information according to the available data,and at the end its mixed Nash equilibrium is analyzed.Followed by the data-driven Harsanyi transform,the incomplete information penalty game is converted into a complete but imperfect information dynamic game,with the analysis of its Bayesian Nash equilibrium.Third,the principle of type assignment randomization and mixed Nash equilibrium simulation o is discussed via Monte Carlo simulation method.Using the triangular distribution to carry the available data,we try the simulation practice of incomplete information penalty game under data-driven Harsanyi transform.The 500 simulation analysis of the penalty game model considering the hit rate shows that the data-driven Harsanyi transform can easily realize the real-time guidance of the players,and the accuracy rate is 95%.