Graph Theory & Networks

3 min read#networks

Nodes joined by edges are the universal skeleton of connection — friendships, roads, neurons, and the web all reduce to the same simple picture.

Contents

Graph Theory & Networks

Draw some dots. Join a few with lines. You have just built a graph — and with it a model flexible enough to describe friendships, road maps, food webs, the wiring of the brain, the citation trails of science, and the hyperlinks of the web. A graph keeps only what matters about a system of relationships: who is connected to whom. Strip away everything else and a startling amount of structure survives.

This section is a tour of that structure. We start with the Graph itself — vertices and edges, directed or weighted — and the two ways to walk one: Breadth-First Search in patient layers and Depth-First Search plunging deep. We add weights and ask for the cheapest route with Dijkstra's Algorithm, or the cheapest way to connect everything at once with a Minimum Spanning Tree. Then we step back and ask which nodes matter: Centrality measures importance many ways, and PageRank answers it with a beautiful trick borrowed from linear algebra.

Why networks belong here

Networks are Complex Systems in their purest combinatorial form. There are no forces and no differential equations — just nodes and edges — yet the same lessons return: local connections produce global behavior no single node contains. Two structural patterns dominate the real world. Small-World Networks are tightly clustered yet astonishingly shallow ("six degrees of separation"), while Scale-Free Networks grow lopsided hubs because the rich get richer. Both arise from rules you can state in a sentence.

The linear-algebra connection

A graph can be written as a matrix: row i, column j is 1 when an edge runs from i to j. Once a network is a matrix, the whole machinery of linear algebra applies. PageRank is the headline example — a page's importance is its share of the dominant eigenvector of the link matrix, exactly the Eigenvalues and Eigenvectors idea from the Linear Algebra section. In fact, this very site ranks its related-topic suggestions with a PageRank-style computation over the wiki-link graph you are browsing right now.

Explore

FoundationsGraph · Breadth-First Search · Depth-First Search

Weighted problemsDijkstra's Algorithm · Minimum Spanning Tree

ImportancePageRank · Centrality

Network shapesSmall-World Network · Scale-Free Network

See also