A parallel algorithm for the dense symmetric eigenvalue problem on a transputer array (Q1186180)

The problem of calculating the eigenvalues of the system \(Ax=\lambda x\) where \(A\) is a real symmetric fully populated matrix of order \(n\times n\) is considered. A parallel implementation of Householder's algorithm on a transputer array for the transformation of a dense matrix \(A\) to tridiagonal form is presented. The algorithm is based on a sequence of similarity transformations on a matrix \(A\). The results of numerical experiments obtained from array consisted of 16 transputers connected in a ring topology are presented.