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.

A -mode approach for exponential integrators: actions of -functions of Kronecker sums

A $μ$-mode approach for exponential integrators: actions of $φ$-functions of Kronecker sums

Direction splitting of $φ$-functions in exponential integrators for $d$-dimensional problems in Kronecker form

Adaptive rational Krylov methods for exponential Runge--Kutta integrators

Efficient third order tensor-oriented directional splitting for exponential integrators

Adaptive Rational Krylov Methods for Exponential Runge–Kutta Integrators

KIOPS: A fast adaptive Krylov subspace solver for exponential integrators

EPIRK-W and EPIRK-K time discretization methods

Should exponential integrators be used for advection-dominated problems?

Low-synchronization Arnoldi Methods for the Matrix Exponential with Application to Exponential Integrators

Efficient simulation of complex Ginzburg--Landau equations using high-order exponential-type methods

Explicit Exponential Runge–Kutta Methods for Semilinear Time-Fractional Integro-Differential Equations

A second order directional split exponential integrator for systems of advection--diffusion--reaction equations

A second order directional split exponential integrator for systems of advection–diffusion–reaction equations

Third order, uniform in low to high oscillatory coefficients, exponential integrators for Klein-Gordon equations

Parallel-in-time high-order multiderivative IMEX solvers

Exponential integrators for second-order in time partial differential equations

Simplified explicit exponential Runge-Kutta methods without order reduction

A Direct Matrix Method for Computing Analytical Jacobians of Discretized Nonlinear Integro-differential Equations

Matrix-free weighted quadrature for a computationally efficient isogeometric $k$-method

High-order implicit time integration scheme based on Padé expansions