线性互补问题的有效算法研究.pdfVIP

Abstract Abstract The paper studies some efficient iterated algorithms for linear complementarity problem and their theories. Large sparse matrices are accounted in application. Seeking the efficient numerical algorithms which can keep the sparsity of coefficient matrix has been an emphasis in recent years. The emphasis of the paper is to extend some efficient methods to solve large LCPs with unsymmetric coefficient matrices. Firstly, the author proposes the two-stage multi-splitting algorithm that is efficient for LCP. Unsymmetric problem is extended from symmetric problem, so the paper mainly discusses the convergence theory the LCPs with large sparse unsymmetric coefficient matrix. The emphasis is base on the complexity of unsymmetric matrix establishing and proving the convergence theories about the different kinds LCPs. After all, extend the two-stage multi-splitting algorithm. Secondly, learn from other class algorithms strong points to offset the weakness and construct the two-step iterative algorithm. The paper introduces the LCPs with general matrix, symmetric matrix and unsymmetric matrix. The emphasis of this part is the algorithm of LCPs with unsymmetric matrix and its convergence theory. Then makes an extension of the two-step iterative algorithm. Finally, the algorithms proposed in the thesis are tested on computer. During the test of two-stage multi-splitting algorithm, contrast the numerical results that are received under different conditions. During the test of two-step iterative algorithms, compare the results with which is from class SORP. Roughl