提出了向最值函数的锥D－s凸，锥D－s拟凸，s右导数及锥D－s伪凸等新概念，讨论了锥D－s凸函数的有关性质，建立了约束向量极值问题（VP）的最优性必要条件与涉及锥D－s凸（拟凸，伪凸）函数的约束极值问题（VP）的最优性充分条件，揭示了（VP）的局部最优解与整体最优解，（VP）的弱有效解与有效解的关系，所得结果推广了凸规划及部分广义凸规划的相关结论。

Some new concepts of coneD_s convexity,coneD-s quasiconvexity,cone D-s pseudoconvexity,s right derivative of vector value mapping are proposed;and their related properties are discussed;the optimality necessary conditions and sufficient conditions for constrained extremum problem(VP)involving in cone D-s convex mapping are extablished;the relation between local optimal solution and global optimal solution;as well as the relation between weakly efficient solution and efficient solution for (VP) are derived;the corresponding results generalize the related results of classic convex programming.

O221.2

国家自然科学基金资助项目 (199710 92，1940 10 40 )