Operator-splitting methods in respect of eigenvalue problems for nonlinear equations and applications for Burgers equations (Q2271969)

The author derives, from a standard method, new methods to solve complicated mixed coupled partial differential equations and reorders the operators for different scales on the basis of the approximated eigenvalues of the linearized systems obtained by a fixed-point iterative scheme; thus reducing the decomposition error. The increased hyperbolic behavior of the equations leads to an increase in the iteration steps of the new method, but a second-order method is obtained. Such iterative splitting methods can balance the different types of behavior of the underlying operators, so that one operator smooths out the solution process, while the other operator decreases the smoothness. A balance between the implicit and explicit discretizations helps overcome the mixed behavior in an unsplitted method. Numerical applications are to Burgers equations.