A Smoothing Inexact Newton Method For Inverse Quadratic Programming Problems

In recent years, the inverse mathematical programming problem has become a very important field of research.Study the inverse problem of quadratic programming problems and their solutions with a wide range of application value. For a class of inverse quadratic programming problems have a lot of decision variables, in order to reduce the complexity of the problem, the inverse quadratic programming problem transform it into a relatively small number of decision variables dual problem, the Newton algorithm for solving the dual problem are proposed, the global convergence of the algorithm and feasibility proved in a certain hypothetical conditions.For the long running time for Newton's algorithm solving the the inverse problem Inexact smoothing newton algorithms is proposed,the algorithm by introducing a smooth function of the dual problem into sub-problems continuous unconstrained optimization problems is proposed inverse problem solving quadratic programming inexact smoothing newton Method. Numerical results show that:the method is feasible and effective that compared with Newton's method, high speed, its'running time is short.