摘要
首先讨论策略有约束的、策略无约束的和终端有约束的微分对策鞍点存在的必要条件；然后建立策略有约束的和策略无约束的微分对策数值解的梯度法，并对策略约束条件给出四个具体构造约束算子的方法；最后证明所给梯度迭代法的收敛性．
The paper first discusses the necessary conditions of saddle points’existence of thedifferential games whose strategies are unconstrained,constrained and terminally constrained,then establishes the gradient methods of differential games whose strategies are unconstrainedand constrained,and presents four specific examples of constructing constrained operator’smethods for strategies constrained,finally proves the numerical convergences of the gradientiterated methods presented.
出处
《大连理工大学学报》
CAS
CSCD
北大核心
1994年第4期456-462,共7页
Journal of Dalian University of Technology
基金
国家自然科学基金
关键词
对策论
梯度算法
微分对策
数值解
two-person games
saddle points
games theory
gradient algorithms
Convergene
