二阶锥规划在工程、控制、金融等领域具有广泛的应用.本文研究一种求解二阶锥规划的非精确不可行内点法.该算法的基本思想是首先定义不可行中心路径及其邻域,然后通过求解一个非线性方程组得到非精确的搜索方向,再取一个合适的步长,使得新的迭代点落在不可行中心路径的邻域内.该算法不要求初始点和迭代点位于严格可行解集内.在适当的假设条件下证明了算法只需迭代O(√n ln(1/ε))次就可以找到问题的ε-近似解.
Second-order cone programming has a wide range of applications in many fields, including engineering, control, finance and etc.. A primal-dual inexact infeasible interior-point algorithm for solving the second-order cone programming is investigated. The basic idea of the algorithm is to first give an infeasible central path and its neighborhood, then find the inexact search directions by solving a nonlinear equation set and choose a suitable step-size which ensures the new iteration point in the neighborhood of the infeasible central path. This algorithm does not require the initial points and iteration points to be in the sets of the strictly feasible solutions. Under certain assumptions, we prove that the ε-approximate solution can be obtained in O(√ n ln( 1 /ε )) iterations.