IPython3 (Notes)

last update: 2017-06-08

IPython3 provides a series of advantages over the standard Python shell - in particular with regard to scientific computing. A good start is too look into the IPython documentation. Start by making sure it is properly installed:

$ sudo apt install ipython3 python3-ipdb ipython3-qtconsole ipython3-notebook

IPython can be started via eg.

$ ipython3
$ ipython3 qtconsole --pylab=inline
$ ipython3 notebook --pylab=inline

However, I prefer adding an alias to start it with less typing. This can, for example, be achieved by

$ echo "alias i='ipython3'" >> ~/.bashrc

Shortcuts

_, __
last, second last output
_i[X] / _[X]
input / output number X
var = !command
execute operating system shell command and store output in var
!command $var
use content of Python variable var as arguments to operating system command

Magic Functions

One of IPython's key features are so-called magic functions. They are helpful when Python would not be natural / straight-forward in solving a problem. Examples include %cpaste, %paste and . Refer to the documentation (%quickref) for more. It is possible to obtain help for the magic functions by appending a ? directly after their name.

%cpaste
Open a special prompt for manually pasting Python code to be executed
%debug
Invoke post mortem debugger
%hist
Print command input history for copying to an editor
%logstart
Create log of entire console session
%paste
Execute pre-formatted Python code from clipboard
%pdb
Automatically enter debugger after any exception
%quickref
Display the IPython Quick Reference Card
%run [-d] script.py
Run passed script inside IPython (-d enables the debugger); -p -s cumulative runs the script with profiling enabled.
%timeit statement
Time execution of a Python statement or expression. It has a heuristic to run the statement multiple times to produce a fairly reliable average runtime.
%prun statement
Run a statement through the python code profiler

ipdb

IPython's integrated debugger. Allows inspecting the status of variable by simply typing their name. Should their name overlap with a debugger command, the latter takes precedence. To be able to still access the variable, type

!varname

Inside the debugger, simply type h for help. If you have trouble getting out of the debugger, the following should help

import os
os._exit(0)

For debugging a function available inside ipython, the following snipped (kudos to Wes McKinney) will be helpful

def debug(f, *args, **kwargs):
    from IPython.core.debugger import Pdb
    pdb = Pdb(color_scheme='Linux')
    return pdb.runcall(f, *args, **kwargs)