一种求解非线性互补问题非精确光滑化牛顿算法.pdf

中文摘要 互补问题是运筹学与计算数学的交叉研宄领域 . 由于与最优化、变分不等式、 平衡问题、对策论、不动点理论等数学分支的紧密联系，以及在实际生活中的广 泛应用，互补问题越来越显示其重要性.本文主要研宄非线性互补问题的非精确 牛顿算法 ;针对其算法中求解线性子系统的复杂性，提出了求解其近似解的方法， 从而减少计算量 . 基于光滑牛顿法的基本思想，利用光滑对称扰动FB-函数对光滑 牛顿算法做出了一些改进，提出了一种新的求解P r N C P的非精确光滑Newton算 法 . 然后对该算法进行了细致的收敛性分析，证明过程表明该算法在适当条件下 是全局超线性 (局部二次 )收敛的. 最后给出了一些利用非精确光滑牛顿算法求解 非线性互补问题的数值结果.数值结果证明该算法是适定的并且对于大规模问题 的求解是非常有效的 . 关键词：非线性互补问题；非精确牛顿算法；全局线性收敛；二次收敛 . Abstract The complementarity problem is an in terdisciplinary research area of Op- erations Research and Computational Mathematics. Because its close contact w ith the branches of mathematics optim ization, variational inequalities, balance problems, game theory, fixed point theory and extensive applications, the comple- m entarity problem became more and more im portant. In this paper, an inexact smoothing Newton method for solving the nonlinear complementarity problem was developed. This method was based on a smooth symmetric perturbation Fischer Burmeister function and aimed to overcome the complexity of the linear subsystems w ith reducing computation for obtaining the approximate solution of NCP. The new algorithm performed global and superlinear convergence w ithout stric t complementarity under m ild conditions. Furthermore, numerical experi- ments demonstrated the feasibility and high efficiency of the new algorithm for solving large-scale problems. K e y W o rd s : Nonlinear complementarity problems; Smoothing Newton method; Global convergence; Superlinear convergence; Quadratic convergence. 目 录 第 一 章 绪 论 ...............................................................................................................1 1.1 前 言 .................................................................................................................. 1 1.1.1 互补问题的引入 ............................................................................. 1 1 . 1 . 2 应用 背 景 ........................................................................................... 1 1