Abstract:SIAM Journal on Matrix Analysis and Applications, Volume 45, Issue 4, Page 1902-1928, December 2024. In this article, we derive fast and robust parallel-in-time preconditioned iterative methods for the all-at-once linear systems arising upon discretization of time-dependent PDEs. The discretization we employ is based on a Runge–Kutta method in time, for which the development of parallel solvers is an emerging research area in the literature of numerical methods for time-dependent PDEs. By making use of classical theory of block matrices, one is able to derive a preconditioner for the systems considered. The block structure of the preconditioner allows for parallelism in the time variable, as long as one is able to provide a robust solver for the system of the stages of the method. We thus propose a preconditioner for the latter system based on a singular value decomposition (SVD) of the (real) Runge–Kutta matrix [math]. Supposing [math] is invertible and the discretization of the differential operator in space is symmetric positive definite, we prove that the spectrum of the system for the stages preconditioned by our SVD-based preconditioner is contained within the right-half of the unit circle, under suitable assumptions on the matrix [math] (the assumptions are well posed due to the polar decomposition of [math]). We show the numerical efficiency of our approach by solving the system of the stages arising from the discretization of the heat equation and the Stokes equations, with sequential time-stepping. Finally, we provide numerical results of the all-at-once approach for both problems, showing the speedup achieved on a parallel architecture.

A parallel method for numerical solution of delay differential equations

Convergence Aspects of Step-Parallel Iteration of Runge-Kutta Methods for Delay Differential Equations

A Parallel Iteration of Runge-Kutta Method for Delay Differential Equations

Stability Analysis of a Class of Parallel Runge-Kutta Methods for Delay Differential Equations

The convergence theorem of parallel Runge-Kutta methods for delay differential equation

A Perturbation-Incremental Method for Delay Differential Equations

Parallel-in-Time Solver for the All-at-Once Runge--Kutta Discretization

Fast and high-order approximation of parabolic equations using hierarchical direct solvers and implicit Runge-Kutta methods

Convergence analysis of an Arnoldi order reduced Runge-Kutta method for integro-differential equations of pantograph type

Diagonalization Based Parallel-in-Time Method for a Class of Fourth Order Time Dependent PDEs

Parallel-in-Time Solver for the All-at-Once Runge–Kutta Discretization

A Randomized Runge-Kutta Method for time-irregular delay differential equations

Algebraically Stable SDIRK Methods with Controllable Numerical Dissipation for First/second-Order Time-Dependent Problems

Very High-Order A-stable Stiffly Accurate Diagonally Implicit Runge-Kutta Methods with Error Estimators

A Parallel-in-Time Algorithm for High-Order BDF Methods for Diffusion and Subdiffusion Equations

The Parallel Waveform Relaxation Stochastic Runge–Kutta Method for Stochastic Differential Equations

A Five-Step Block Method Coupled with Symmetric Compact Finite Difference Scheme for Solving Time-Dependent Partial Differential Equations

Multistep finite difference schemes for the variable coefficient delay parabolic equations

A New Multi-Step Method for Solving Delay Differential Equations using Lagrange Interpolation

Stability Analysis of Runge-Kutta Methods for Delay Integro-Differential Equations

Parallel-in-time Multilevel Krylov Methods: A Prototype