A globally convergent trust region algorithm for optimization with general constraints and simple bounds

1.IntroductionInthispaper,weconsiderthefollowingnonlinearprogr~ngproblemwherec(x)=(c,(x),c2(2),',We(.))',i(x)andci(x)(i=1,2,',m)arerealfunctions*ThisworkissupPOrtedbytheNationalNaturalScienceFOundationofChinaandtheManagement,DecisionandinformationSystemLab,theChineseAcademyofSciences.definedinD={xEReIISx5u}.Weassumethath相似文献   

解半定规划的二次摄动方法

半定规划在系统论,控制论,组合优化,和特征值优化等领域有着广泛的应用。本文将半定规划摄动成二次半定规划,它的唯一解恰为原问题的解,并且对其偶问题等价于一个线性对称的投影方程,可方便地用投影收缩方法求解,从而获得原半定规划问题的解。文章给出了算法及其收敛性分析,数值试验结果表明摄动方法是解半定规划的一种有效的方法。     

解非线性最小二乘问题的锥模型算法

在自然科学研究、经济、统计等领域,非线性最小二乘有着广泛的应用,因而寻找快捷有效的算法有着十分重要的意义。它首先是一个最优化问题,同时又有自身的结构特点,充分利用其结构特点,是寻找更有效算法的关键。     

解非线性最小二乘问题的锥模型算法的局部收敛性

1 引言 对于非线性最小二乘问题 minf(x)=(1/2)sum from x=1 to m (r_i(x))~2=(1/2)R(x)~TR(x), (1.1)其中R(x)=(r_1(x),…,r_m(x))~T:DR~n→R~m,m≥n,有 g(x)=f'(x)=J(x)~TR(x), (1.2) H(x)=f(x)=J(x)~TJ(x)+sum from x=1 to m r_i(x)r_i(x), (1.3)其中J(x)=((r_i(x))/x_j)·Gauss-Newton方法,及Dennis等的改进方法,都是采用二次模型     

ON SOME PROPERTIES OF SOLUTIONS TO SEMIDEFINITE PROGRAMMING

It is well known that for symmetric linear programming there exists a strictly complementary solution if the primal and the dual problems are both feasible. However, this is not necessary true for symmetric or general semide finite programming even if both the primal problem and its dual problem are strictly feasible. Some other properties are also concerned.