Abstract:SIAM Journal on Matrix Analysis and Applications, Volume 45, Issue 1, Page 744-770, March 2024. We consider the solution of large stiff systems of ODEs with explicit exponential Runge–Kutta integrators. These problems arise from semidiscretized semilinear parabolic PDEs on continuous domains or on inherently discrete graph domains. A series of results reduces the requirement of computing linear combinations of [math]-functions in exponential integrators to the approximation of the action of a smaller number of matrix exponentials on certain vectors. State-of-the-art computational methods use polynomial Krylov subspaces of adaptive size for this task. They have the drawback that the required number of Krylov subspace iterations to obtain a desired tolerance increases drastically with the spectral radius of the discrete linear differential operator, e.g., the problem size. We present an approach that leverages rational Krylov subspace methods promising superior approximation qualities. We prove a novel a posteriori error estimate of rational Krylov approximations to the action of the matrix exponential on vectors for single time points, which allows for an adaptive approach similar to existing polynomial Krylov techniques. We discuss pole selection and the efficient solution of the arising sequences of shifted linear systems by direct and preconditioned iterative solvers. Numerical experiments show that our method outperforms the state of the art for sufficiently large spectral radii of the discrete linear differential operators. The key to this are approximately constant numbers of rational Krylov iterations, which enable a near-linear scaling of the runtime with respect to the problem size.

Tensorized block rational Krylov methods for tensor Sylvester equations

An efficient block rational Krylov solver for Sylvester equations with adaptive pole selection

A rational Krylov methods for large scale linear multidimensional dynamical systems

Tensor Krylov subspace methods via the T-product for large Sylvester tensor equations

Krylov methods for low-rank commuting generalized Sylvester equations

Randomized sketched TT-GMRES for linear systems with tensor structure

Krylov subspace methods for large multidimensional eigenvalue computation

A Randomized Block Krylov Method for Tensor Train Approximation

A tensor-based approach to solving linear systems involving Kronecker sum of matrices

A class of Petrov-Galerkin Krylov methods for algebraic Riccati equations

Block Full Rank Linearizations of Rational Matrices

A nested divide-and-conquer method for tensor Sylvester equations with positive definite hierarchically semiseparable coefficients

Rational Krylov methods for functions of matrices with applications to fractional partial differential equations

Adaptive rational Krylov methods for exponential Runge--Kutta integrators

Sketched and Truncated Polynomial Krylov Subspace Methods: Matrix Sylvester Equations

Randomized extended average block Kaczmarz method for inconsistent tensor equations under t-product

Developing iterative algorithms to solve Sylvester tensor equations

Randomized Kaczmarz for tensor linear systems

Adaptive Rational Krylov Methods for Exponential Runge–Kutta Integrators

On Greedy Randomized Kaczmarz-type Methods for Solving the System of Tensor Equations

Rational Krylov methods for fractional diffusion problems on graphs