Instructions for installing python with ipython
Three options for using iPython are Enthought, Anaconda, and Wakari
1. Enthought Distribution
Once installed:
- you can install symbolic links in /usr/local/bin to the python and ipython binaries in the installed source (on my 64bit mac, that's /Library/Frameworks/EPD64.framework/Versions/7.3/bin/ ), or you can just define an alias
- terminal based interactive python is then started with "ipython"
- terminal based interactive python with access to plotting (via x-windows) is started with "ipython --pylab"
- for the ipython notebook demoed in class, use "ipython notebook --pylab inline --script"
- Any command-line python scripts just start with a path to the python binary, e.g.,
#!/usr/local/bin/python (if you have a symbolic link installed)
Important: The default installation needs to be updated to the latest version of python, matplotlib, etc. In the same bin/ directory as above there's a program enpkg that will determine what needs to be updated. Run
> sudo bin/enpkg --whats-new
it will give a list of packages that need to updated.
(If it complains
Enstaller is out of date. Update? ([y]/n)
type y, and it'll first update enpkg, then rerun enpkg --whats-new )
Using
sudo bin/enpkg --update-all
is supposed to update everything, but if it complains about dependencies,
can also be updated by giving a list
sudo bin/enpkg sympy scipy matplotlib pyzmq ...
Eventually it does update everything, as verified by a final check:
> sudo bin/enpkg --whats-new
Name installed available
============================================================
no new version of any installed package is available
2. Anaconda
Anaconda is another easy-to-install bundle of Python and most of these libraries.
This page (from a course given elsewhere)
has instructions to set up the Anaconda python distribution.
3. Wakari
As an alternative to installing on your own computer, it's possible to sign up for free ipython cloud notebooks at www.wakari.io (.5Gb of RAM plus 10Gb of disk space for free accounts). This will get you up and running instantly, though is probably not the most reliable option over the course of the semester.
One Note:
Last year there were questions about functions like 'plot' not working:
NameError: name 'plot' is not defined
If you start iPython from the command-line with
> ipython notebook --pylab inline --script
then the --pylab inline
calls in all the necessary matplotlib
routines and the 'inline' renders the results of plots in the notebook
(rather than external viewer). The wakari notebooks also have this
set up by default.
At least one person using Anaconda python under windows was starting
up from gui without that option, in which case the command
within the ipython window itself:
%pylab inline
will set up the equivalent environment, with matplotlib and numpy
modules called in, and with graphics done within the notebook window.