mhd-ray-tracing

A python repository of ray tracing methods (e.g. WKB) and plotting tools for MHD wave trajectories.

Current functionality

Currently consists of the generalised WKB solution for fast waves in a zero beta plasma for a user-specified or predefined magnetic field. The intention is to further generalise the code in future (e.g. different modes, finite-beta, nonlinearity, etc)

The mathematics is discribed McLaughlin, Thurgood, Botha & Wiggs 2019.

Using the code

The best place to get started might be to read the Jupyter examples notebook . If you view it in github, it should render as a (static) html file (note, it can take a while to load). If you download the repository and have Jupyter running, you should be able to interact with it.

Otherwise, there are some instructions below

External packages needed

Requires a number of not-too-exotic packages (matplotlib, numpy, scipy). If you are used to python it should be easy to figure out, if you are not, the anaconda distribution should come pre-packaged with everything you need.

magnetic_field.py

This is where new magnetic fields should be added, as unique classes. No checks are made on the physicality of the field specified. This module also contains pre-defined classes for different magnetic fields, currently:

Linear2dNull B = [y,x,0] a arbitarilly-rotated version of the field in e.g. McLaughlin & Hood 2004
Dipole3dCase1 a 3D dipolar magnetic 'dome' embedded within a uniform vertical magnetic field. Described McLaughlin, Thurgood, Botha & Wiggs 2019.

The main solver imports one of these classes under the alias magnetic_field to act as a function to return the fields components and derivatives at a set of coordinates.

wkb_fast_zerobeta.py

This is essentially the main solver.

It contains the data structure for an individual ray path (the class 'Ray'), and functions related to advancing and sampling the path in time. Also contains classes and functions relating to collections of rays ('Swarms') and their advancement and manipulation en-masse.

Usage

Typically import the module into a script or the interactive interpreter under alias wkb.

import wkb_fast_zerobeta as wkb

However, it requires foreknowledge of the class of magnetic field you want to use (from magnetic_field.py). To avoid having to edit wkb_fast_zerobeta.py every time we want to use a different field, and to keep scripts relating to different problems self-contained, we can proceed as follows. When the module is first imported it immediately looks for a module called tmp_config.py , which tells it the desired class. If this file does not exist it will throw an error. As such you can initialise your splotting script / interactive session / whatever with the following

# Create a tmp file to pass the chosen field class to wkb class
f = open('tmp_config.py','w')
f.write('fieldclass=\'Dipole3dCase1\'')
f.close()
# Import
import wkb_fast_zerobeta as wkb
# Remove the temporary file
import os
os.remove('tmp_config.py')

where you mayreplace 'Dipole3dCase1' with the class name of whatever field is appropriate for your problem. After or before this block, you will likely want to import whatever else you need (e.g. matplotlib).

This snippet initialises a ray at time 0 at (x0,y0,z0) and (p0,q0,r0)

import wkb_fast_zerobeta as wkb
x0 = 
# y0,z0,p0,q0,r0 etc..
myray = wkb.Ray(x0,y0,z0,p0,q0,r0)

We might then ask the solve to integrate the ode up to a end characteristic time (s_end) with 100 linearly spaced sampling points between s0 and s_end

s_end = 1.0
ns = 100
test.solve(s_end,ns)

This can be then plotted. We note that this returns the solution at 100 sample points. The ode solver automatically chooses step sizes within tolerances. This can give the impression of larger 'jumps' in regions of high alfven speed than in regions of slower procession. In reality, it is just that the linear sampling in time means there are less points to construct the line. The ode solution itself usually has used an appropriate step size within adaptive tolerances. Beware of this.

fieldlines.py

Handles fieldline tracing and datastructures in an analagous way to the wkb ray tracing. Useful for cases where plotting fieldlines as contours of a known flux function (magnetic vector potential) is impractical (mainly, 3D!)

Examples

You can get a quick idea of how it works in practice by running and reading the scripts in the "examples" directory.

JOThurgood / mhd-ray-tracing