【作者】 唐嘉；

【导师】 刘三阳；

【作者基本信息】 西安电子科技大学 ， 应用数学， 2010， 博士

【摘要】 互补问题是数学规划研究中较为活跃的分支.由于其应用背景的广泛性和与其他分支的交叉性,近年来受到越来越多的研究者的关注并取得了丰硕的成果.对互补问题的研究可以分为理论和算法两方面.前者主要研究问题的解的存在性、唯一性、稳定性以及灵敏度分析等性质,后者着重研究如何构造有效的算法.本文主要研究互补问题的数值算法.主要内容概括总结如下：1.基于Chen-Harler-Kanzow-Smale (CHKS)光滑函数,提出求解P0-函数混合互补问题的一种正则化的光滑方法.即：通过求得适定问题的一组解序列来代替求解原来不适定的问题,并且使得这组解序列收敛于原来问题的解.该算法中的正则参数和光滑参数都是彼此独立的变量,并且可以通过线性方程组的迭代很快得到.2.鉴于光滑函数在解决互补问题的光滑类算法中的重要性,根据混合互补问题的定义域本身的结构,构造了两种新的混合互补函数函数.它们具有一些很好的性质,即：保证了与之相关的光滑路径的存在性和连续性以及光滑类算法所产生的迭代序列的有界性.在此基础上,给出了求解P0-函数混合互补问题的三种光滑算法：预测校正光滑牛顿算法、一步光滑牛顿算法和Broyden-like拟牛顿算法.3.针对一般的(即：不要求是P0函数)混合互补问题提出了一种修正的光滑牛顿算法.该算法结合了梯度步.这使得该算法可以去掉F至少是一个P0-函数的假设.在适当条件下,该算法全局收敛性和局部超线性收敛性也得到了证明.4.构造了一种新的对称锥上的互补函数,并研究了这个函数的强制性和强半光滑性以及其雅可比矩阵的强半光滑性.这些良好的性质为利用该SCCP互补函数设计相关算法求解对称锥互补问题SCCP打下了一定的基础.更多还原

【Abstract】 Complementarity problem is a heated topic in the research of mathematical program-ming. Because of its wide application background and overlapping with other branch, Complementarity problem attracts more and more attentions from various fields of sci-ence and engineering in recent years. Also many achievements have been proposed in the field. Generally speaking, its research can be classified into two classes:theory and algorithms. The former is devoted to the existence, uniqueness, stability and sensitivity analysis of the solutions, while the latter is confined to solve the problems efficiently.This paper is primarily designed to solve the problems efficiently. The main contri-butions are listed as follows:1. Based on Chen-Harler-Kanzow-Smale (CHKS) smoothing function, we propose a regularized smoothing Newton method for mixed complementarity problems with a P0-function. This method is designed to handle ill-posed problems which substitutes the solution of original problem with the solution of a sequence of well-defined problems whose solutions converging to the solution of the original problem. And the regularization parameter and the smoothing parameter in our algorithm are independent variables and can be immediately obtained through iteration of linear system.2. In view of the significance of smooth functions in smoothing type algorithms for solving complementarity problems, we construct two new MCP functions by smoothing a perturbed mid function such that one can study the existence and continuity of the smooth path and boundedness of the iteration sequence. Then, three smoothing algorithms for solving the MCP with a P0-function, that is, predictor-corrector smoothing Newton algo-rithm and one-step smoothing Newton method and broyden-like quasi-Newton algorithm are proposed respectively.3. To solve general (not necessarily P0) mixed complementarity problems, a modified one-step smoothing Newton method is given. The algorithm incorporates a gradient step. Such procedure make the algorithm remove the condition that function F is required to be at least a P0-function. Under suitable assumptions, global convergence and locally superlinear convergence of the algorithm are established.4. A new C-function for symmetric cone complementarity problems is constructed. It is showed that the new C-function is coercive, strongly semismooth and its Jacobian is also strongly semismooth, which are important for the construction of the corresponding algorithm for solving symmetric cone complementarity problems with this C-function.更多还原