<title>Recursive ULV decomposition</title>

Abstract

The ULV decomposition (ULVD) is an important member of a class of rank-revealing two-sided orthogonal decompositions used to approximate the singular value decomposition (SVD). The ULVD can be modified much faster than the SVD. The accurate computation of the subspaces is required in applications in signal processing. In this paper we introduce a recursive ULVD algorithm which is faster than all available stable SVD algorithms. Moreover, we present an alternative refinement algorithm.