Highly efficient exponential scalar auxiliary variable approaches with relaxation (RE-SAV) for gradient flows

TL;DR

This paper introduces a novel, highly efficient exponential scalar auxiliary variable (E-SAV) method with relaxation (RE-SAV) for gradient flows, achieving high-order accuracy and unconditional energy stability without boundedness assumptions.

Contribution

The paper develops the RE-SAV approach that enhances existing SAV methods by improving accuracy and efficiency, and allows easy construction of high-order energy stable schemes.

Findings

Demonstrates improved efficiency and accuracy through numerical examples.

Provides a framework for constructing high-order unconditionally energy stable schemes.

Removes the need for bounded-from-below assumptions for free energy potentials.

Abstract

For the past few years, scalar auxiliary variable (SAV) and SAV-type approaches became very hot and efficient methods to simulate various gradient flows. Inspired by the new SAV approach in \cite{huang2020highly}, we propose a novel technique to construct a new exponential scalar auxiliary variable (E-SAV) approach to construct high-order numerical energy stable schemes for gradient flows. To improve its accuracy and consistency noticeably, we propose an E-SAV approach with relaxation, which we named the relaxed E-SAV (RE-SAV) method for gradient flows. The RE-SAV approach preserves all the advantages of the traditional SAV approach. In addition, we do not need any the bounded-from-below assumptions for the free energy potential or nonlinear term. Besides, the first-order, second-order and higher-order unconditionally energy stable time-stepping schemes are easy to construct. Several numerical examples are provided to demonstrate the improved efficiency and accuracy of the proposed method.