Eigenmode initialisation of 2D (magneto)hydrodynamic simulations
Date:
Contributed talk & poster presentation at AAPPS-DPP 2026, Busan, South Korea.
Abstract. The early evolution of unstable (magneto)hydrodynamic equilibria is often governed by a few dominant linear eigenmodes. We show that 2D (magneto)hydrodynamic simulations, performed with MPI-AMRVAC (Keppens et al. 2023), reach the non-linear stage faster when they are initialized with the system’s most unstable linear eigenmode (De Jonghe & Russell 2026), compared to simulations using traditional perturbation methods. To do this, we calculate the equilibrium’s eigenmodes with the Legolas code (Claes et al. 2020) and select a superposition of eigenmodes as the initial perturbation for our simulations. Three configurations were considered for this comparative study: (a) a flow-sheared fluid interface, unstable to the Kelvin-Helmholtz instability; (b) a Harris current sheet, unstable to the resistive tearing instability; and (c) a flow-sheared plasma interface, unstable to the magnetized Kelvin-Helmholtz instability. These cases represent different non-linear behaviours and are used to test the effect of the initial perturbation on the development of turbulence, on secondary instabilities, and on the non-linear suppression of the instability by magnetic tension, respectively. Benchmarking is done with respect to simulations initialized with velocity noise or analytic perturbation profiles. Across all three test cases, initializing with the instability allows the simulations to reach their late-evolution stage faster.
To make use of this numerical approach more accessible, we developed the open-source GIMLI framework, short for General Interface for MPI-AMRVAC / Legolas Interconnection. This toolkit, combining routines in Python and Fortran, facilitates setting up a single configuration consistently in both MPI-AMRVAC and Legolas by generating user files for both codes, as well as extracting the desired linear eigenmode solutions from Legolas for use in MPI-AMRVAC, and allowing for spectroscopic analysis of 1D simulation data (e.g. simulation slices across a current sheet) with Legolas.