Python 3.4 compatibility with numpy, scipy, and matplotlib

3
votes

I am very new to Python and am wondering what versions of numpy/scipy/matplotlib are compatible with different versions of Python--or to what degree it matters that they match up.

Right now I am using Python 3.4.3, numpy 1.8.2, scipy 0.13.3 and matplotlib 1.3.1. I saw on the scipy FAQ that "the first release of NumPy to support Python 3 was NumPy 1.5.0. Python 3 support in SciPy starts with version 0.9.0." But how should one know what versions of these modules are appropriate for python 3.4? Eg. I know that there is a scipy0.18, but will it potentially cause problems if attempt to use with python 3.4 and not 3.5? Sorry if this is a silly question, I just want to avoid having to uninstall/update different versions multiple times, which I already have done.

3
pythonnumpymatplotlibscipy
when you try to install these packages they have the requirements check in place so they update if the versions are not compatible. – percusse

3 Answers

3
votes

I would recommend you to take a closer look at the Anaconda installation.

You will install all needed packages in one step and beside that all the packages will be mutually compatible and tested.

If you need packages that aren't included, you can simply install them using internal packaging tool - conda:

conda install package_name
1
votes

Current versions, numpy 0.11.0 and scipy 0.17.1, should be perfectly fine with python 3.4 and 3.5.

To a first approximation, any non-ancient version of numpy or scipy is going to be OK with python 3.4.

It's good idea anyway to run self-tests after installation: >>> from numpy import test; test() and >>> from scipy import test; test().

1
votes

From the very same page:

NumPy and SciPy support the Python 2.x series, (versions 2.6 and 2.7), as well as Python 3.2 and newer

Python versions in the 3.x line should all be backwards compatible anyway.