二阶锥规划基于新价值函数的新光滑牛顿算法

A New Smoothing Newton Algorithm Based on the New Valued Function for Se-cond-Order Cone Programming

本文在Fischer-Burmeister (FB)函数的基础上,提出一种求解二阶锥规划(SOCP)问题的新光滑牛顿函数,并采用一种新的价值函数。同时用一个带扰动的牛顿方程组去获得搜索方向,在适当假设下,证明新算法具有全局收敛和局部二次收敛。

Based on the Fischer-Burmeister function,a new Newton method is proposed for solving the Se-cond-order cone programming.This algorithm adapts a new smoothing value function and proposes a Newton equation with disturbance to gain the search direction.Under suitable assumptions,we prove that the proposed new method is globally and locally quadratically convergent.