Efficient Accelerated Coordinate Descent Methods and Faster Algorithms for Solving Linear Systems

TL;DR

This paper introduces accelerated randomized coordinate descent methods that improve convergence rates and computational efficiency, leading to faster algorithms for solving linear systems across various regimes and applications.

Contribution

It generalizes and implements a Nesterov-inspired acceleration technique for coordinate descent, achieving faster asymptotic convergence and optimality in solving linear systems.

Findings

Achieves faster asymptotic runtime than conjugate gradient for certain positive definite systems.

Improves convergence guarantees for Kaczmarz methods in image reconstruction.

Attains the best known running time for solving SDD systems in the RAM model.

Abstract

In this paper we show how to accelerate randomized coordinate descent methods and achieve faster convergence rates without paying per-iteration costs in asymptotic running time. In particular, we show how to generalize and efficiently implement a method proposed by Nesterov, giving faster asymptotic running times for various algorithms that use standard coordinate descent as a black box. In addition to providing a proof of convergence for this new general method, we show that it is numerically stable, efficiently implementable, and in certain regimes, asymptotically optimal. To highlight the computational power of this algorithm, we show how it can used to create faster linear system solvers in several regimes: - We show how this method achieves a faster asymptotic runtime than conjugate gradient for solving a broad class of symmetric positive definite systems of equations. - We improve the best known asymptotic convergence guarantees for Kaczmarz methods, a popular technique for image reconstruction and solving overdetermined systems of equations, by accelerating a randomized algorithm of Strohmer and Vershynin. - We achieve the best known running time for solving Symmetric Diagonally Dominant (SDD) system of equations in the unit-cost RAM model, obtaining an O(m log^{3/2} n (log log n)^{1/2} log (log n / eps)) asymptotic running time by accelerating a recent solver by Kelner et al. Beyond the independent interest of these solvers, we believe they highlight the versatility of the approach of this paper and we hope that they will open the door for further algorithmic improvements in the future.