人工神经网络求一般对称矩阵特征向量的研究

Theoretical Feasibilty of Finding the Eigenvector of an Ormnary Symmetric Matrix Using an Artificlal Neural Network

讨论了对称矩阵特征向量的人工神经网络求法，特别是特征值重数大于１的情况。通过对已有人工神经网络数学模型的细致分析，得出了“只要输入电压适当，一般对称矩阵全部特征向量可由输出电压给出”的结论。

A method is proposed for finding the eigenvector of any symmetric matrix using a neuralnetwork,especially for the case in which the degree of the eigenvalue is greater than 1.Through a careful analysis of the existing mathematical model for the artificial neural net-work,it has been found that all the eigenvectors of an ordinary symmetric matrix are givenby the stable voltage output if the voItage input is appropriate.