Relativistic magnetohydrodynamics in dynamical spacetimes: A new AMR implementation

TL;DR

This paper introduces a new general relativistic magnetohydrodynamics (GRMHD) code with adaptive-mesh refinement capable of evolving MHD fluids in dynamical spacetimes, validated through various tests including black hole scenarios.

Contribution

The paper presents a novel GRMHD code that integrates AMR with a divergence-free magnetic field scheme in full 3+1 dimensions, validated against multiple complex tests.

Findings

Code accurately models magnetized shocks and waves.

Achieves convergence and matches analytic solutions.

Successfully simulates black hole related MHD phenomena.

Abstract

We have written and tested a new general relativistic magnetohydrodynamics (GRMHD) code, capable of evolving MHD fluids in dynamical spacetimes with adaptive-mesh refinement (AMR). Our code solves the Einstein-Maxwell-MHD system of coupled equations in full 3+1 dimensions, evolving the metric via the Baumgarte-Shapiro Shibata-Nakamura (BSSN) formalism and the MHD and magnetic induction equations via a conservative, high-resolution shock-capturing scheme. The induction equations are recast as an evolution equation for the magnetic vector potential, which exists on a grid that is staggered with respect to the hydrodynamic and metric variables. The divergenceless constraint div(B)=0 is enforced by the curl of the vector potential. Our MHD scheme is fully compatible with AMR, so that fluids at AMR refinement boundaries maintain div(B)=0. In simulations with uniform grid spacing, our MHD scheme is numerically equivalent to a commonly used, staggered-mesh constrained-transport scheme. We present code validation test results, both in Minkowski and curved spacetimes. They include magnetized shocks, nonlinear Alfv\'en waves, cylindrical explosions, cylindrical rotating disks, magnetized Bondi tests, and the collapse of a magnetized rotating star. Some of the more stringent tests involve black holes. We find good agreement between analytic and numerical solutions in these tests, and achieve convergence at the expected order.