解凸约束非线性单调方程组的无导数低存储Broyden族投影法  

A DERIVATIVE-FREE MEMORYLESS BROYDEN FAMILY PROJECTION METHOD FOR SOLVING NONLINEAR MONOTONE SYSTEMS WITH CONVEX CONSTRAINS

作　　者：饶佳运 黄娜 Rao Jiayun;Huang Na(Department of Applied Mathematics,College of Science,China Agricultural Univeristy,Beijing 100083,China)

机构地区：[1]中国农业大学理学院应用数学系,北京100083

出　　处：《计算数学》2023年第2期197-214,共18页Mathematica Numerica Sinica

摘　　要：拟牛顿法是求解非线性方程组的一类有效方法.相较于经典的牛顿法,拟牛顿法不需要计算Jacobian矩阵且仍具有超线性收敛性.本文基于BFGS和DFP的迭代公式,构造了新的充分下降方向.将该搜索方向和投影技术相结合,本文提出了无导数低存储的投影算法求解带凸约束的非线性单调方程组并证明了该算法是全局且R-线性收敛的.最后,将该算法用于求解压缩感知问题.实验结果表明,本文所提出的算法具有良好的计算效率和稳定性.Quasi-Newton methods are a class of effective methods for solving nonlinear equations.Compared with the classical Newton method,Quasi-Newton methods do not require computation of the Jacobian matrix and still possess superlinear convergence.Based on the BFGS and DFP iterative schemes,we construct a new sufficient descent direction.Combining this direction with some projection techniques,we propose the derivative-free memoryless projection method for solving nonlinear monotone systems with convex constrains and derive its global R-linear convergence.Finally,we use this method to solve the compressive sensing problems.Numerical results show the efficiency and stability of our new method.

关 键 词：非线性单调方程组 凸约束 BROYDEN族 投影法 无导数 

分 类 号：O241.7[理学—计算数学]