Nonlinear grid mapping applied to an FDTD-based, multi-center 3D Schrodinger equation solver

We developed a straightforward yet effective method of increasing the accuracy of grid-based partial differential equation (PDE) solvers by condensing computational grid points near centers of interest. We applied this "nonlinear mapping" of grid points to a finite-differenced explicit implementation of a time-dependent Schrodinger equation solver in three dimensions. A particular multi-center mapping was developed for systems with multiple Coulomb potentials, allowing the solver to be used in complex configurations where symmetry cannot be used for simplification. We verified our method by finding the eigenstates and eigenenergies of the hydrogen atom and the hydrogen molecular ion (H-2(+)) and comparing them to known solutions. We demonstrated that our nonlinear mapping scheme - which can be readily added to existing PDE solvers - results in a marked increase in accuracy versus a linear mapping with the same number of (or even much fewer) grid points, thus reducing memory and computational requirements by orders of magnitude. (c) 2011 Elsevier B.V. All rights reserved.