随机二阶锥二次规划逆问题的SAA方法

本文讨论一类随机的二阶锥二次规划逆问题, 该模型是一个含有二阶锥互补约束的随机二次规划模型, 对解释部分实际问题有着一定的优势。为了求解该模型, 本文引入了随机抽样技术和互补约束光滑化近似技术, 得到问题的近似子问题。本文证明, 只要子问题的解是存在且收敛的, 则该极限以概率一是原问题的C-稳定点; 若严格互补条件和二阶必要性条件成立, 则该极限以概率1是原问题的M-稳定点。一个简单的数值实验验证了该算法具有一定的可行性。

An SAA approach for a class of second-order cone stochastic inverse quadratic programming problem

In this paper, we consider a class of stochastic inverse quadratic second-order cone programming problem. This stochastic model contains complementarity constraints, and is more proper to model some class of real world problems. By employing the techniques of stochastic sampling and smoothing, we construct auxiliary approximate sub-problems to solve the original model. In addition, we proved that if the solutions of the approximate sub-problems converge, then with probability one the limit is the C-stationary point of the original problem. If strict complementarity condition and the second order necessary condition hold, then with probability one the limit is an M-stationary point. A simple numerical test verified the applicability of our approach.