Butterfly Algorithm

This is my attempt to make the Butterfly Algorithm more palatable by using pictures, text, math and code.

The notebook can be viewed on nbviewer.

Running the Code

Before running from the command line, check out this repository and then generate the module (from the IPython notebook)

$ git clone https://github.com/dhermes/butterfly-algorithm
$ make generated-module

after doing this, the simulation can be run

$ ./butterfly_on_whale.py --M=50 --L=10

to specify the number of terms in the Taylor series (M=50) and the number of refinements (L=10).

Profiling

The code has been profiled using RunSnakeRun to create an image which shows where the code spends its time. The image was made with M = 50 and L = 10:

$ apt-get install python-wxgtk2.8
$ pip install SquareMap RunSnakeRun
$ make generated-module
$ python -m cProfile -o M50_L10.profile.info butterfly_on_whale.py --M=50 --L=10
$ runsnake M50_L10.profile.info

Trivia

To avoid using STIX fonts (don't look great for LaTeX slides), you can manually edit your IPython install.

First locate it

$ python -c 'import IPython; print IPython.__file__'
/foo/IPython/__init__.pyc

then edit. For example, if it is installed in

/foo/IPython/...

as above, edit the files

/foo/IPython/html/static/notebook/js/mathjaxutils.js
/foo/IPython/nbconvert/templates/html/mathjax.tpl

The key change is in the "HTML-CSS" key passed in to MathJax.Hub.Config:

    "HTML-CSS": {
        availableFonts: ["TeX"],  // Line added
        preferredFont: "TeX",  // Line added
        styles: {'.MathJax_Display': {"margin": 0}},
        linebreaks: { automatic: true }
    }

The blue whale image and the butterfly image are Creative Commons licensed via Attribution 2.0 Generic.