matplotlib

Matplotlib

TODO install currently broken!

Scientific plotting library.

Install

Although it is available in PyPi, there are tons of dependencies which pip may not install, and it even fails silently!

Therefore don't use the pip install: but use the distro's package manager. On Ubuntu:

sudo aptitude install python-matplotlib

Developing

Clone besides of being able to hack Matplotlib, you also get of tons of example files with this!

If you are on Ubuntu get the built dependencies:

sudo aptitude build-dep python-matplotlib

Then build C extensions and install with:

sudo python setup.py install

Test hacks

You cannot put the Python files in your python path simply without installing, because the compiled dependencies won't go to the correct place without an install (crashes on _path not found).

So, after you made changes do again:

sudo python setup.py install

After you build the files C files, which is what takes the longest, you don't have to build them again, so after you hack just:

The problem with this is that you cannot keep the distro default installed also.

TODO how not to install after every change.

Architecture

State machine

Not object based, but state machine based.

This means that you often have a current something, and you modify the current something.

This methods like gca() which get you the current something.

Rationale: easier to type on interactive sessions

Objects

• figure: everything
• axes: each subplot, including several axis
• axis (!= axes): the line with the ticks and numbers

show

Plot to screen:

plt.plot([0,1])
plt.show()

On window close, clears plot, so if you have to replot if you want to reuse the old plot:

plt.plot([1,0])
plt.show()

savefig

Plot to file

Recommended formats are:

• SGV: vector. Very precise, but needs to be transformed into bits before being put in a PDF.
• PNG: lossless compression. Simpler to put in PDF because it represents bits directly.

examples:

plt.plot([0,1])
plt.savefig( 'svg.png', format='png', bbox_inches='tight' )
plt.savefig( 'png.png', format='png', bbox_inches='tight' )

format options

Many format options can be given on either:

• in a single format string at once
• in separate kwargs

use only separate kwargs in non-interactive programs since this is more manageable

string

plt.plot([0,1], 'r--')
plt.show()

kwargs

plt.plot([0,1,2,3], [0,1,4,9], color='r', linestyle='--'  )
save_svg_clear('r--kwargs')

pylab

Part of matplotlib: http://stackoverflow.com/questions/12987624/confusion-between-numpy-scipy-matplotlib-and-pylab