一类随机规划问题的近似Lagrange-Newton算法

Approximate Lagrange-Newton Algorithms for a Class of Stochastic Programming Problems

通过利用MonteCarlo模拟方法近似目标函数及其一(二)阶信息,给出了带有补偿的随机二次规划问题的一个近似不可行Lagrange-Newton算法,并在依概率1条件下证明了它的全局收敛性和局部超线性收敛性。

By using Monte Carlo simulation-based approximate objective function and its first (second) derivative information, an infeasible approximate Lagrange-Newton algorithm is proposed for stochastic quadratic programs with recourse property. With probability one, the global convergence and local super-linear convergence of the algorithm are shown.[JP2]