The systolic execution of one of the most famous and also most complex algorithms in matrix computations: the QR iteration or algorithm is considered. This algorithm allows to find the spectral decomposition of a matrix A (n, n), its eigenvalues and its eigenvectors. A new problem-size dependent Systolic Algorithm for its execution on a one dimensional array is first presented. Then, a block algorithm is deduced, in order to extract parallelism, and finally, a new problem-size independent Systolic Algorithm is considered.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados