English

Structural Convergence Results for Approximation of Dominant Subspaces from Block Krylov Spaces

Numerical Analysis 2017-05-11 v2 Numerical Analysis

Abstract

This paper is concerned with approximating the dominant left singular vector space of a real matrix AA of arbitrary dimension, from block Krylov spaces generated by the matrix AATAA^T and the block vector AXAX. Two classes of results are presented. First are bounds on the distance, in the two and Frobenius norms, between the Krylov space and the target space. The distance is expressed in terms of principal angles. Second are quality of approximation bounds, relative to the best approximation in the Frobenius norm. For starting guesses XX of full column-rank, the bounds depend on the tangent of the principal angles between XX and the dominant right singular vector space of AA. The results presented here form the structural foundation for the analysis of randomized Krylov space methods. The innovative feature is a combination of traditional Lanczos convergence analysis with optimal approximations via least squares problems.

Keywords

Cite

@article{arxiv.1609.00671,
  title  = {Structural Convergence Results for Approximation of Dominant Subspaces from Block Krylov Spaces},
  author = {Petros Drineas and Ilse Ipsen and Eugenia-Maria Kontopoulou and Malik Magdon-Ismail},
  journal= {arXiv preprint arXiv:1609.00671},
  year   = {2017}
}