Fine-grained Analysis and Faster Algorithms for Iteratively Solving Linear Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dereziński, Michał, LeJeune, Daniel, Needell, Deanna, Rebrova, Elizaveta
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908409485328384
author Dereziński, Michał
LeJeune, Daniel
Needell, Deanna
Rebrova, Elizaveta
author_facet Dereziński, Michał
LeJeune, Daniel
Needell, Deanna
Rebrova, Elizaveta
contents Despite being a key bottleneck in many machine learning tasks, the cost of solving large linear systems has proven challenging to quantify due to problem-dependent quantities such as condition numbers. To tackle this, we consider a fine-grained notion of complexity for solving linear systems, which is motivated by applications where the data exhibits low-dimensional structure, including spiked covariance models and kernel machines, and when the linear system is explicitly regularized, such as ridge regression. Concretely, let $κ_\ell$ be the ratio between the $\ell$th largest and the smallest singular value of $n\times n$ matrix $A$. We give a stochastic algorithm based on the Sketch-and-Project paradigm, that solves the linear system $Ax = b$, that is, finds $\bar{x}$ such that $\|A\bar{x} - b\| \le ε\|b\|$, in time $\bar O(κ_\ell\cdot n^2\log 1/ε)$, for any $\ell = O(n^{0.729})$. This is a direct improvement over preconditioned conjugate gradient, and it provides a stronger separation between stochastic linear solvers and algorithms accessing $A$ only through matrix-vector products. Our main technical contribution is the new analysis of the first and second moments of the random projection matrix that arises in Sketch-and-Project.
format Preprint
id arxiv_https___arxiv_org_abs_2405_05818
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Fine-grained Analysis and Faster Algorithms for Iteratively Solving Linear Systems
Dereziński, Michał
LeJeune, Daniel
Needell, Deanna
Rebrova, Elizaveta
Data Structures and Algorithms
Machine Learning
Numerical Analysis
Optimization and Control
Despite being a key bottleneck in many machine learning tasks, the cost of solving large linear systems has proven challenging to quantify due to problem-dependent quantities such as condition numbers. To tackle this, we consider a fine-grained notion of complexity for solving linear systems, which is motivated by applications where the data exhibits low-dimensional structure, including spiked covariance models and kernel machines, and when the linear system is explicitly regularized, such as ridge regression. Concretely, let $κ_\ell$ be the ratio between the $\ell$th largest and the smallest singular value of $n\times n$ matrix $A$. We give a stochastic algorithm based on the Sketch-and-Project paradigm, that solves the linear system $Ax = b$, that is, finds $\bar{x}$ such that $\|A\bar{x} - b\| \le ε\|b\|$, in time $\bar O(κ_\ell\cdot n^2\log 1/ε)$, for any $\ell = O(n^{0.729})$. This is a direct improvement over preconditioned conjugate gradient, and it provides a stronger separation between stochastic linear solvers and algorithms accessing $A$ only through matrix-vector products. Our main technical contribution is the new analysis of the first and second moments of the random projection matrix that arises in Sketch-and-Project.
title Fine-grained Analysis and Faster Algorithms for Iteratively Solving Linear Systems
topic Data Structures and Algorithms
Machine Learning
Numerical Analysis
Optimization and Control
url https://arxiv.org/abs/2405.05818