ABSTRACT
Approximation to some of the largest singular values and associated vectors (largest singular triplets) of very large rectangular matrices is essential in many areas of data analysis, for example, dimension reduction, data mining, data visualization, and detection of patterns. Furthermore, the statistical procedure principal component analysis has a direct relationship to the singular value decomposition (SVD) [16,26,31]. The importance of computing the singular triplets of very large matrices has spurred numerous computational methods in the literature; see, for example, [1,3,4,6,9,12-15,18,20,22,24,27,29,30] and references therein.
