摘要

Dynamical process simulation of complex real-life problems often requires the use of modern algorithms, which automatically adapt both time and space discretization in order to get error-controlled approximations of the solution. In this paper, a combination of linearly implicit time integrators of Rosenbrock type and adaptive multilevel finite elements based on a posteriori error estimates is presented. This approach has proven to work quite satisfactorily for a wide range of challenging practical problems. We show the performance of our adaptive method for a brine transport model in porous media.