The Research On The Algorithms Of Some Quadratic Programming

This thesis deals with the topoic of algorithms of some Quadratic Programming problems.We analyse some Quadratic Programming algorithms and get the improved algorithms efficiency with matlab.Based on the introduced of basic knowledag of Quadratic Programming,we analyse some Quadratic Programming algorithms,specially Lemke algorithm and Newton interior algorithms. Based on the analysis of solving convex quadratic programming with the complementary rotating axis algorithm of classical Lemke, we find the localization of the Lemke algorithm.According to the analysis of solving convex quadratic programming by classical Lemke pivot algorithm, we give the steps of computation with improvement Lemke's algorithm and procedure. Making use of the new algorithm to solve the convex quadratic programming, we get the improved Lemke algorithm efficiency. On the basis of calculate examples, the new algorithm could effectively overcome the localization of solution by the experimental results, decreasing iterative process of the convex quadratic programming, improving the algorithmic efficiency.We presented alogarithm for the convex quadratic programs problem. The logarithmic is based on Interior penalty function method and Quasi-Newton method. We use augmented lagrange function and logarithmic barrier function method to translate the constrained problem into unconstrained problem on the assumption that the linear search satiesfies Wolfe-Powell rules, then use quasi-Newton methods to solve the unconstrained problem. Finally, we use numerical emulate test indicating the efficiency of the algoritbm.