A dimensional splitting exponential time differencing scheme for multidimensional fractional Allen-Cahn equations (Q2660695)

The aim of this work is to provide efficient numerical methods for solving multidimensional space fractional Allen-Cahn (FAC) equations. Thanks to a dimensional splitting of the spatial discretization matrix, a new ETD (exponential time differencing)-type integrator for multidimensional FAC equations is proposed. It is shown that the proposed dimensional splitting ETD Runge-Kutta (DSETDRK2) scheme preserves the maximum principle and is a second order accurate. In addition to this, an error estimate in the maximum norm is given. The scheme leads to linear systems and matrix exponential associated with only 1D discretization matrix which possesses Toeplitz structure. Fast algorithms for Toeplitz inverse and Toeplitz matrix exponential are used in the implementation of the scheme. Numerical examples in two and three spatial dimensions are presented to illustrate the efficiency of the proposed scheme.