Dominant subspace and low-rank approximations from block Krylov subspaces without a gap

arXiv2107.01990MaRDI QIDQ6372005FDOQ6372005

Publication date: 5 July 2021

Abstract: In this work we obtain results related to the approximation of

h

-dimensional dominant subspaces and low rank approximations of matrices

A i n m a t h b b K^{m i m e s n}

(where

m a t h b b K = m a t h b b R

or

m a t h b b C)

in case there is no singular gap at the index

h

, i.e. if

s i g m a_{h} = s i g m a_{h + 1}

(where

s i g m a_{1} g e q l d o t s g e q s i g m a_{p} g e q 0

denote the singular values of

A

, and

p = m i n m, n

). In order to do this, we develop a novel perspective for the convergence analysis of the classical deterministic block Krylov methods in this context. Indeed, starting with a matrix

X i n m a t h b b K^{n i m e s r}

with

r g e q h

satisfying a compatibility assumption with some

h

-dimensional right dominant subspace, we show that block Krylov methods produce arbitrarily good approximations for both problems mentioned above. Our approach is based on recent work by Drineas, Ipsen, Kontopoulou and Magdon-Ismail on approximation of structural left dominant subspaces. The main difference between our work and previous work on this topic is that instead of exploiting a singular gap at

h

(which is zero in this case) we exploit the nearest existing singular gaps.

Mathematics Subject Classification ID

Eigenvalues, singular values, and eigenvectors (15A18)

This page was built for publication: Dominant subspace and low-rank approximations from block Krylov subspaces without a gap

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6372005)