skip to content
 

Scientific computing package for python.

Instructions for users: 

On Ubuntu workstations this is available by default.


On compute clusters the best way to use scipy is to load one of the anaconda modules which will give you access to a recent version of Python, scipy, and large set of other Python modules. You can also access an identical anaconda environment from managed Linux workstations by loading the same anaconda module there too.

Licence Details: 

SciPy has a open-source licence which is BSD-like.

Documentation: 

There is some documentation on the website https://www.scipy.org/.

Source: 

The homepage is https://www.scipy.org/.

Admin notes: 

On Ubuntu this is part of the distro.


You have to build it on SuSE 11.1. This is a fiddly build (although it has improved greatly since I did this for the 10.2 and 10.3 images) and still does not pass its own internal test suite.


First install numpy and nose. I used the blas and lapack that come with SuSE as there is no point trying to build ATLAS for the NFS server as it won't be optimised right for anyone but me.


On the 64-bit image:



 export LAPACK=/usr/lib64/liblapack.a  export BLAS=/usr/lib64/libblas.a  python setup.py install --prefix=/usr/local/shared/suse-10.3/x86_64/python 


then test it with



 python >>> import scipy >>> scipy.test() 


I got one 'error' for the 64-bit build (which I think is this issue) and four for the 32-bit build- the same one as with the 64-bit and three others which I think are this issue. It is NOT the issue mentioned in the Scipy 0.7.0 release notes for weave and gcc 4.3 because I have tried that workaround and it makes no difference.


The Ubuntu 64-bit packaged version also fails the test suite with the same error as above, which is a known bug fixed in 0.80.

System status 

System monitoring page

Can't find what you're looking for?

Then you might find our A-Z site index useful. Or, you can search the site using the box at the top of the page, or by clicking here.