NetworkX Graph Analysis Library

NetworkX is a Python package for the creation, manipulation, and study of the structure, dynamics, and functions of complex networks. Originally developed by Aric Hagberg, Dan Schult, and Pieter Swart at Los Alamos National Laboratory starting in 2004, the project has over 15,000 stars as of 2026 and is the most widely used graph analysis library in Python. Key features include: graph types (Graph for undirected networks, DiGraph for directed networks, MultiGraph and MultiDiGraph for networks with parallel edges), node and edge data (arbitrary Python objects as nodes, with edge attributes supporting weights, labels, colors, timestamps, and custom properties), graph generators (over 100 network generators including classic graphs like complete, path, cycle, star, and tree, random graph models like Erdos-Renyi, Watts-Strogatz, Barabasi-Albert, configuration models, and real-world network datasets), algorithms (shortest paths via Dijkstra, Bellman-Ford, and A-star, maximum flow, minimum spanning tree, centrality measures including betweenness, closeness, eigenvector, and PageRank, community detection via modularity, label propagation, and greedy modularity, graph isomorphism checking via VF2, spectral graph theory with Laplacian and adjacency eigenvalues, and traversal via BFS, DFS, and edge DFS), drawing (Matplotlib, Graphviz, and PyDot integration for 2D visualization), I/O (read and write graphs in GraphML, GEXF, GML, JSON, edge lists, adjacency lists, and Pandas DataFrames), and integration (compatible with SciPy, NumPy, Pandas, and PyTorch).