scientific article; zbMATH DE number 4043151

From MaRDI portal
Publication:3780353





[For the entire collection see Zbl 0636.00016.] Two algorithms for singular value decomposition are studied in a unifying framework: the power method and the Chebyshev iteration method. They allow to compute adaptively the partial SVD (dominant, smallest or intermediate triplets) of matrices that are slowly time varying and perturbed by noise. The only routines used are matrix vector multiplication and orthogonalization procedures. Some original convergence expressions are derived and convergence rates of both methods are compared. Some backward error theorems permit the control of the convergence level with respect to the notes level.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3780353)