Higher-order monotone iterative methods for finite difference systems of nonlinear reaction-diffusion-convection equations (Q2271416)

The authors consider the monotone iterative solution of a discretized parabolic equation or of its elliptic counterpart. The discretization in space is done by finite differences (not just carefully: a first-order approximation in case of a variable diffusion coefficient, or if the boundary conditions are not of the first kind) and by backward Euler in time. They work with lower and upper solutions (and the intervals defined by both), with the aim of convergence acceleration by including the maximal values of the derivatives of the nonlinear right-hand sides (computed on those intervals) into the linear system which is solved by a simple (Picard) iteration and started from the lower or upper solution. The authors show that the two sequences obtained converge monotonously and that their mean value converges with order \(p+2\), \(p\) being the number of simple iterations. Finally, results of numerical experiments are presented (including CPUs) illustrating the effectiveness of their methods.