摘要
本文提出了一个求解非线性半定规划的可行序列半定规划(SSDP)算法。该算法的初始点和迭代点均是可行点,在每次迭代中通过求解两个二次半定规划子问题确定搜索方向,步长由满足目标函数下降性和约束函数可行性的线搜索产生,在某些假设条件下本文证明了算法的全局收敛性。
This paper proposes a feasible SSDP algorithm for solving nonlinear semidefinite programming. The initial point and iteration points are feasible. The search direction is determined by solving two quadratic semidefinite programming subproblems. The step size is obtained by calculating the line search that satisfies the descent property of the objective function and the feasibility of the constraint function. The global convergence of the algorithm is proved under mild conditions.
出处
《应用数学进展》
2020年第2期238-243,共6页
Advances in Applied Mathematics
基金
获国家自然科学基金(No. 11561005),广西自然科学基金(No. 2016GXNSFAA380248)资助。
