Efficient implementation of partitioned stiff exponential Runge-Kutta methods (Q2301422)

This paper is devoted to the development of exponential Runge-Kutta methods for the numerical solution of partitioned systems of differential equations. Such partitioned systems arise naturally in the study of multi-physics models, i.e., models with coupling and interaction of various physical fields, often leading to different time and spatial scales. The authors exploit the partitioned structure of this class of systems to construct computationally efficient integrators based on partitioned exponential Runge-Kutta methods. The key step in the numerical implementation is the splitting of linear and nonlinear terms of the additive partitioned system, thus allowing a suitable approximation of the resulting Jacobian in block form. In this way, the authors introduce the specific numerical implementation of this scheme of order two, three and four. This numerical approach is tested on an example describing reaction-diffusion phenomena of a chemical process in a multi-physics framework. In this example, both convergence and expected order are verified.