NetworkX: Difference between revisions
No edit summary |
|||
Line 9: | Line 9: | ||
NetworkX requires a relatively large number of Python libraries to be installed; however, the vast majority can be obtained through yum: | NetworkX requires a relatively large number of Python libraries to be installed; however, the vast majority can be obtained through yum: | ||
<code>sudo yum install atlas blas numpy python-dateutil pytz graphviz graphviz-python graphviz-doc graphviz-graphs graphviz-devel pydot scipy PyYAML pyparsing</code> | <code>sudo yum install atlas blas numpy python-dateutil pytz graphviz graphviz-python graphviz-doc graphviz-graphs graphviz-devel pydot scipy PyYAML pyparsing python-matplotlib</code> | ||
== Using == | == Using == |
Latest revision as of 12:50, 2 July 2008
Introduction
This document describes a minimal process for quick installation of the entire set of libraries and applications needed for use of Python graph processing packages NetworkX and Pygraphviz on standard Fedora system configurations. The prerequisite system packages must be installed by an account with root or sudo access; however, NetworkX and Pygraphviz themselves can be installed to and used from user or project space quite easily. Example code can be found at the projects' respective websites.
Installing Dependencies
NetworkX requires a relatively large number of Python libraries to be installed; however, the vast majority can be obtained through yum:
sudo yum install atlas blas numpy python-dateutil pytz graphviz graphviz-python graphviz-doc graphviz-graphs graphviz-devel pydot scipy PyYAML pyparsing python-matplotlib
Using
Once the above packages are installed, NetworkX and Pygraphviz can be downloaded from http://pypi.python.org/pypi/networkx/ and http://pypi.python.org/pypi/pygraphviz/ respectively. They can be installed system-wide by running the included setup.py scripts with root access, or simply built using a non-privileged account and added to the environment variable PYTHONPATH
; for example, KUSP users may wish to add the following to .bashrc or .bash_profile:
export PYTHONPATH="/projects/kurt/python-lib/:$PYTHONPATH"