Quick introduction and howto for glemur.

Introduction

GLEMuR is a finite difference Lagrangian code which uses mimetic differential operators and runs on Nvidia GPUs. GLEMuR stands for Gpu-based Lagrangian mimEtic Magnetic Relaxation. Its main purpose is to study the relaxation of magnetic relaxation in environments of zero resistivity and viscosity.

GLEMuR is not a general purpose equation solver, neither are the full magnetohydrodynamics equations implemented. For such problems refer to other numerical codes.

Quick Start

Requirements

• Nvidia graphic cards with computing capability 3.0 or higher
• CUDA Toolkit
• C++ compiler
• Fortran compiler
• Make

optional, but strongly recommended:

• IPython
• Paraview
• Bash

Setup the code

1. Obtain GLEMuR from its repository: git clone https://[email protected]/SimonCan/glemur.git This creates the directory glemur with its source files.

2. Add the binary and python directories to your PATH and PYTHONPATH environmental variables: export PATH=$PATH:glemur/bin, export PATH=$PYTHONPATH:glemur/python If you are using anything else then bash add the paths accordingly. For convenience, copy those lines into your ~/.bashrc.

3. Compile GLEMuR within the glemur/src directory by simply typing make.

4. Create the directory data.

5. Run the code with ./glemur

Visualize the data

Before we can plot we need to make sure that IPython knows about GLEMuR by changing a few configuration files. These can reside in different directories, depending on which operating system or distribution you have. The following will work with Ubutnu 14.04:

1. Add to your ~/.ipython/ipythonrc: import_all glemur.

2. Add to your ~/.ipython/profile_default/startup/init.py: import glemur as gm and import pylab as plt.

3. Go back to your simulation directory glemur/src and start ipython.

4. Load and plot the data: ts = gm.readTs(), plt.plot(ts.t, ts.B2).

Advanced setup and visualization

• Tidy up your simulation directory structure.
• Simulation parameters.
• Creating custom initial conditions.

Code internals

• Time stepping.
• Boundary conditions.
• Mimetic operators.