一类复变量优化问题的次梯度投影算法

A complex-valued subgradient projection method( CSPM) for a class of complex variables non-smooth convex optimization problems

利用CR微分理论,提出求解一类线性等式约束的复变量非光滑凸优化问题的复值次梯度投影算法(CSPM),该算法能完全基于复域上运行。在较弱的条件下证明了算法的全局收敛性,数值实验进一步表明了CSPM的可行性和有效性,该算法尤其适合大规模优化问题的求解。

A complex-valued subgradient projection method（ CSPM） based on CR calculus theory is presented to solve a class of complex variables non-smooth convex optimization problems with linear equality constraints,which can be completely implemented in the complex domain. The proposed method is proved to be globally convergent under mild conditions. Numerical experiments show that CSPM is feasible and effective and suitable for solving large-scale optimization problems.