Fourier spectral methods based on restricted Padé approximations for space fractional reaction-diffusion systems

Abstract

By utilizing the power of the Fourier spectral approach and the restricted Padé rational approximations, we have devised two third-order numerical methods to investigate the complex phenomena that arise in multi-dimensional space fractional reaction-diffusion models. The Fourier spectral approach yields a fully diagonal representation of the fractional Laplacian with the ability to extend the methods to multi-dimensional cases with the same computational complexity as one-dimensional and makes it possible to attain spectral convergence. Third-order single-pole restricted Padé approximations of the matrix exponential are utilized in developing the time stepping methods. We also use sophisticated mathematical techniques, namely, discrete sine and cosine transforms, to improve the computational efficiency of the methods. Algorithms are derived from these methods for straight-forward implementation in one- and multidimensional models, accommodating both homogeneous Dirichlet and homogeneous Neumann boundary conditions. The third-order accuracy of these methods is proved analytically and demonstrated numerically. Linear error analysis of these methods is presented, stability regions of both methods are computed, and their graphs are plotted. The computational efficiency, reliability, and effectiveness of the presented methods are demonstrated through numerical experiments. The convergence results are computed to support the theoretical findings.