限定广义纳什均衡及其控制罚算法  

Controlled Penalty Algorithm for Solving Restricted Generalized NASH Equilibrium Problem

作　　者：罗美菊[1] 吴欧[2] 

机构地区：[1]辽宁大学,沈阳110036 [2]解放军理工大学,南京210007

出　　处：《火力与指挥控制》2012年第6期200-202,共3页Fire Control & Command Control

摘　　要：广义纳什均衡问题通常有很多解,只有在相当严格的条件下,才有可能得到唯一解。如果任意选取其中之一作为该问题的解,显然是不合理的。为此,提出限定广义纳什均衡的概念。通过给共同约束相对应的拉格朗日乘子增加约束的方法,找出具有某些特性的解。为了求解限定广义纳什均衡,还进一步给出了控制罚算法。.It is well known that a generalized NASH equilibrium problem usually has many solutions and the uniqueness solution is expected only under very restrictive assumptions. Obviously, it is not reasonable if we find just an arbitrary generalized NASH equilibrium as its solution. Therefore, we introduce a new solution concept called a restricted generalized NASH equilibrium. By imposing a certain additional condition on the Lagrange Multipliers associated with the shared constraints, we may try to give a characterization of particular solution. Furthermore, to compute a restricted generalized NASH equilibrium, we propose a controlled penalty algorithm.

关 键 词：限定广义纳什均衡 KKT条件 罚算法 

分 类 号：O225[理学—运筹学与控制论]