Systematic Comparison between Constrained Transport and Mixed Divergence Cleaning Methods for Astrophysical Magnetohydrodynamic Simulations

TL;DR

This paper systematically compares constrained transport and divergence cleaning methods in astrophysical MHD simulations, revealing that CT is generally more accurate and reliable, while divergence cleaning can produce artifacts under certain conditions.

Contribution

The study provides a detailed comparison of CT and divergence cleaning schemes, identifies limitations of divergence cleaning, and proposes modifications to enhance its robustness.

Findings

Divergence cleaning can produce artifacts in localized magnetic fields and during timestep changes.

Original divergence cleaning schemes may be inaccurate in certain astrophysical scenarios.

Constrained transport generally offers more accurate and reliable results.

Abstract

Magnetohydrodynamic (MHD) simulations are indispensable research infrastructure in astrophysics today. In order to satisfy the solenoidal constraint of the MHD equations on discretized grids, modern simulation codes often employ either constrained transport (CT) with a staggered grid or divergence cleaning using an additional variable. We compare CT and Dedner's mixed divergence cleaning schemes systematically, and find that the divergence cleaning scheme can produce substantial artifacts in certain situations. Through numerical experiments including both idealized tests and practical applications, we show that the original implementation of Dedner's scheme becomes inaccurate when magnetic fields are strongly localized or when the timestep suddenly changes. We find that some previous results, such as the extremely rapid growth of magnetic fields during star formation in the early Universe, may be affected by the spurious behavior of the divergence cleaning scheme. We propose a few modifications to improve the robustness of the divergence cleaning method. Nevertheless, we find that the CT scheme is more accurate and reliable in many situations.