Quasioptimal alternating projections and their use in low-rank approximation of matrices and tensors

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: Budzinskiy, Stanislav
Format: Preprint
Veröffentlicht: 2023
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914026661871616
author Budzinskiy, Stanislav
author_facet Budzinskiy, Stanislav
contents We study the convergence of specific inexact alternating projections for two non-convex sets in a Euclidean space. The $σ$-quasioptimal metric projection ($σ\geq 1$) of a point $x$ onto a set $A$ consists of points in $A$ the distance to which is at most $σ$ times larger than the minimal distance $\mathrm{dist}(x,A)$. We prove that quasioptimal alternating projections, when one or both projections are quasioptimal, converge locally and linearly for super-regular sets with transversal intersection. The theory is motivated by the successful application of alternating projections to low-rank matrix and tensor approximation. We focus on two problems -- nonnegative low-rank approximation and low-rank approximation in the maximum norm -- and develop fast alternating-projection algorithms for matrices and tensor trains based on cross approximation and acceleration techniques. The numerical experiments confirm that the proposed methods are efficient and suggest that they can be used to regularise various low-rank computational routines.
format Preprint
id arxiv_https___arxiv_org_abs_2308_16097
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Quasioptimal alternating projections and their use in low-rank approximation of matrices and tensors
Budzinskiy, Stanislav
Optimization and Control
Numerical Analysis
We study the convergence of specific inexact alternating projections for two non-convex sets in a Euclidean space. The $σ$-quasioptimal metric projection ($σ\geq 1$) of a point $x$ onto a set $A$ consists of points in $A$ the distance to which is at most $σ$ times larger than the minimal distance $\mathrm{dist}(x,A)$. We prove that quasioptimal alternating projections, when one or both projections are quasioptimal, converge locally and linearly for super-regular sets with transversal intersection. The theory is motivated by the successful application of alternating projections to low-rank matrix and tensor approximation. We focus on two problems -- nonnegative low-rank approximation and low-rank approximation in the maximum norm -- and develop fast alternating-projection algorithms for matrices and tensor trains based on cross approximation and acceleration techniques. The numerical experiments confirm that the proposed methods are efficient and suggest that they can be used to regularise various low-rank computational routines.
title Quasioptimal alternating projections and their use in low-rank approximation of matrices and tensors
topic Optimization and Control
Numerical Analysis
url https://arxiv.org/abs/2308.16097