A Systolic Array for SVD Updating

From MaRDI portal





A singular value decomposition (SVD) updating algorithm supplemented with a certain re-orthogonalization scheme is implemented on a systolic array with \(O(n^ 2)\) parallelism for \(O(n^ 2)\) complexity, by combining systolic implementations for the matrix-vector product, the QR updating and the SVD. It is shown that a main computational bottleneck for the array implementation can be overcome with a square root-free SVD algorithm based on modified Givens rotations and approximate schemes for the computation of rotation angles in the SVD steps.











This page was built for publication: A Systolic Array for SVD Updating

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5286279)