Revisiting Asynchronous Linear Solvers
From MaRDI portal
Abstract: Asynchronous methods for solving systems of linear equations have been researched since Chazan and Miranker's pioneering 1969 paper on chaotic relaxation. The underlying idea of asynchronous methods is to avoid processor idle time by allowing the processors to continue to make progress even if not all progress made by other processors has been communicated to them. Historically, the applicability of asynchronous methods for solving linear equations was limited to certain restricted classes of matrices, such as diagonally dominant matrices. Furthermore, analysis of these methods focused on proving convergence in the limit. Comparison of the asynchronous convergence rate with its synchronous counterpart and its scaling with the number of processors were seldom studied, and are still not well understood. In this paper, we propose a randomized shared-memory asynchronous method for general symmetric positive definite matrices. We rigorously analyze the convergence rate and prove that it is linear, and is close to that of the method's synchronous counterpart if the processor count is not excessive relative to the size and sparsity of the matrix. We also present an algorithm for unsymmetric systems and overdetermined least-squares. Our work presents a significant improvement in the applicability of asynchronous linear solvers as well as in their convergence analysis, and suggests randomization as a key paradigm to serve as a foundation for asynchronous methods.
Recommendations
- Further results on convergence of asynchronous linear iterations
- Asynchronous iterative algorithms for nonexpansive linear systems
- On the convergence analysis of asynchronous SGD for solving consistent linear systems
- Fully asynchronous stochastic coordinate descent: a tight lower bound on the parallelism achieving linear speedup
- scientific article; zbMATH DE number 1421671
- Convergence analysis of asynchronous linear iterations with stochastic delays
- Asynchronous relaxed iterative methods for solving linear systems of equations
- Asynchronous Optimization Over Graphs: Linear Convergence Under Error Bound Conditions
- Asynchronous block-iterative methods for almost linear equations
- A Randomized Solver for Linear Systems with Exponential Convergence
Cited in
(19)- On unbounded delays in asynchronous parallel fixed-point algorithms
- Asynchronous Richardson iterations: theory and practice
- Fully asynchronous stochastic coordinate descent: a tight lower bound on the parallelism achieving linear speedup
- An asynchronous direct solver for banded linear systems
- Performance analysis of asynchronous parallel Jacobi
- Red light green light method for solving large Markov chains
- On the convergence of randomized and greedy relaxation schemes for solving nonsingular linear systems of equations
- On the convergence analysis of asynchronous SGD for solving consistent linear systems
- ARock: an algorithmic framework for asynchronous parallel coordinate updates
- Asynchronous stochastic coordinate descent: parallelism and convergence properties
- scientific article; zbMATH DE number 1760077 (Why is no real title available?)
- Perturbed iterate analysis for asynchronous stochastic optimization
- Sublinear column-wise actions of the matrix exponential on social networks
- Stochastic reformulations of linear systems: algorithms and convergence theory
- A sampling Kaczmarz-Motzkin algorithm for linear feasibility
- Distributed stochastic optimization with large delays
- Randomized numerical linear algebra: Foundations and algorithms
- Convergence of an asynchronous block-coordinate forward-backward algorithm for convex composite optimization
- Asynchronous Jacobi and randomized Gauss--Seidel methods in shared and distributed memory: A unified convergence-rate analysis
This page was built for publication: Revisiting Asynchronous Linear Solvers
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3177757)
