Solving Allen-Cahn equations with periodic and nonperiodic boundary conditions using mimetic finite-difference operators

Abstract

[EN] In this paper, we investigate and implement a numerical method that is based on the mimetic finite difference operator in order to solve the nonlinear Allen-Cahn equation with periodic and non-periodic boundary conditions. In addition, we also analyze the performance of this mimetic- based method by using the classical heat equation with a variety of boundary conditions. We assess the performance of the mimetic-based numerical method by comparing the errors of its solutions with those obtained by a classical finite difference method and the pdepde built-in Matlab function. We compute the errors by using the exact solutions when they are available or with reference solutions. We adapt and implement the mimetic-based numerical method by using the MOLE (Mimetic Operators Library Enhanced) library that includes some built-in functions that return representations of the curl, divergence and gradient operators, in order to deal with the Allen-Cahn and heat equations. We present several results with regard to errors and numerical convergence tests in order to provide insight into the accuracy of the mimetic-based numerical method. The results show that the numerical method based on the mimetic difference operator is a reliable method for solving the Allen-Cahn and heat equations with periodic and non- periodic boundary conditions. The numerical solutions generated by the mimetic-based method are relatively accurate. We also proposed a new method based on the mimetic finite difference operator and the convexity splitting approach to solve Allen-Cahn equation in 2D. We found that, for small time step sizes the solutions generated by the mimetic-based method are more accurate than the ones generated by the pdepe Matlab function and similar to the solutions given by a finite difference method.