Reduced basis approximation for nonlinear parametrized evolution equations based on empirical operator interpolation (Q2904819)

The paper deals with the numerical solution of parametrized partial differential equations. Usually, to find such a numerical solution is a very time-consuming task especially if many parameter constellations have to be simulated. Reduced basis methods that represent an efficient tool in solving parametrized problems with stationary elliptic and instationary parabolic problems are extended in this paper for nonlinear parametrized evolution equations. The main idea is based on an empirical interpolation of the nonlinear differential operators and their Fréchet derivatives. An efficient offline/online decomposition is obtained for the discrete operators that allow an efficient evaluation for a certain set of interpolation functionals. In such a way a new reduced basis method is established which can be applicable to a large class of discretization schemes that are based on evaluations of discretized operators and their directional derivatives only.NEWLINENEWLINEIn this paper, the finite volume method is used and Newton iterations are used for the solution of resulting nonlinear systems. In such a way, the new so-called POD-GREEDE algorithm is proposed. An a posteriori error estimate for this new method is derived. The experimental section of the paper demonstrates the applicability of the resulting method also for discontinuous data subject to nonlinear convection and diffusion. The approximation properties and run-time for the new method are compared to the full finite volume scheme and confirm the efficiency of proposed scheme.