Markowitz模型的一种神经网络解法

A Neural Network for Solving Markowitz Model

对不允许卖空情况下的Markowitz模型的求解 ,在金融学里面 ,一直是个很棘手的问题 .提出了一种反馈神经网络算法 .该算法计算步骤简单 ,收敛速度快 ,特别针对大规模问题非常有效 .首先提出了这种算法的神经网络模型 ,给出了它的能量函数 .接着证明了它在Lyapunov意义下的稳定性 .最后证明了一定存在一个收敛序列 {Xk} Rn,使得

It has been a very difficult problem to solve Markowitz model with no short sale in finance. In this paper, a feedback neural network algorithm is presented for this problem. The features of this algorithm are simple step, high convergence rate and efficiency. It is especially effective for large_scale problem. Firstly, the neural network model and its energy function is provided. Secondly, the neural network is proven to be Lyapunov stable. Finally, it is proven that there must exist a convergent sequence, so that each limit point of the sequence is the solution of Markowitz model.