Here are some notes I've taken about using Jupyter Lab and its integrations with matplotlib.
There are two mechanisms for having matplotlib render directly inside a Jupyterlab window:
% matplotlib inline
% matplotlib widget
inline backend generally just works, and what you see is what you will get
widget backend is interactive, but is extremely
resource hungry, slow, and prone to all manner of problems that I don't like to
deal with. Sadly, the
inline backend has a cryptic configuration interface.
In general you can change parameters in
$PWD/matplotlibrc. However, there are
a set of seemingly arbitrary matplotlib configuration parameters which are
For example, changing font size requires
$ ipython profile create $ vim ~/.ipython/profile_default/ipython_config.py
and then explicitly addding
There may not be a commented-out version of this in the template created by
ipython profile create.