NetworkX is a Python-based package for the creation, manipulation, and study of the structure, dynamics, and function of complex networks.
The structure of a graph or network is encoded in the edges (connections, links, ties, arcs, bonds) between nodes (vertices, sites, actors). If unqualified, by graph we mean an undirected graph, i.e. no multiple edges are allowed. By a network we usually mean a graph with weights (fields, properties) on nodes and/or edges.
The potential audience for NetworkX includes mathematicians, physicists, biologists, computer scientists, and social scientists. The current state of the art of the science of complex networks is presented in Albert and Barabási [BA02], Newman [Newman03], and Dorogovtsev and Mendes [DM03]. See also the classic texts [Bollobas01], [Diestel97] and [West01] for graph theoretic results and terminology. For basic graph algorithms, we recommend the texts of Sedgewick, e.g. [Sedgewick01] and [Sedgewick02] and the survey of Brandes and Erlebach [BE05].
Why Python? Past experience showed this approach to maximize productivity, power, multi-disciplinary scope (applications include large communication, social, data and biological networks), and platform independence. This philosophy does not exclude using whatever other language is appropriate for a specific subtask, since Python is also an excellent “glue” language [Langtangen04]. Equally important, Python is free, well-supported and a joy to use. Among the many guides to Python, we recommend the documentation at http://www.python.org and the text by Alex Martelli [Martelli03].
NetworkX is free software; you can redistribute it and/or modify it under the terms of the NetworkX License. We welcome contributions from the community. Information on NetworkX development is found at the NetworkX Developer Zone https://networkx.lanl.gov/trac.
NetworkX is intended to: