Energy Stable Multigrid Method for Local and Non-local Hydrodynamic Models for Freezing

TL;DR

This paper introduces an unconditionally energy stable, mass conserving implicit finite difference method for hydrodynamic models derived from density functional theories of freezing, applicable to local and non-local free energy functionals, with efficient nonlinear multigrid solvers.

Contribution

It develops a general, energy stable numerical scheme for complex hydrodynamic models involving local and non-local free energies, with proven stability and efficiency.

Findings

The method is unconditionally energy stable and mass conserving.

Numerical simulations confirm stability and accuracy.

The approach efficiently handles non-local convolution operators.

Abstract

In this paper we present a numerical method for hydrodynamic models that arise from time dependent density functional theories of freezing. The models take the form of compressible Navier-Stokes equations whose pressure is determined by the variational derivative of a free energy, which is a functional of the density field. We present unconditionally energy stable and mass conserving implicit finite difference methods for the models. The methods are based on a convex splitting of the free energy and that ensures that a discrete energy is non-increasing for any choice of time and space step. The methods are applicable to a large class of models, including both local and non-local free energy functionals. The theoretical basis for the numerical method is presented in a general context. The method is applied to problems using two specific free energy functionals: one local and one non-local functional. A nonlinear multigrid method is used to solve the numerical method, which is nonlinear at the implicit time step. The non-local functional, which is a convolution operator, is approximated using the Discrete Fourier Transform. Numerical simulations that confirm the stability and accuracy of the numerical method are presented.