Structural analysis on Facebook.

Three graph algorithms from a UT Dallas big data assignment, ported to TypeScript so they run live in your browser. The arc moves from descriptive statistics on a real social graph to structural decomposition to reliability under random edge failure.

PageRank and friends on Facebook

The original was a PySpark and GraphFrames pipeline that ran five algorithms over the Facebook musae edges list. The TypeScript port implements PageRank, triangle count, connected components, in degree, and out degree. The 22,000 node graph ships as a bundled CSV. Upload your own edges or use the default.

K core peeling

A classical structural decomposition. Repeatedly remove vertices with degree less than k. The remaining subgraph is the k core. Three sample graphs to try (Petersen, K5, and a small example). Animated peeling shows each round of removals.

Network reliability polynomial

Exact two terminal reliability computed by exhaustively enumerating every edge failure subset. For a graph with E edges that means two to the power of E subsets. For each subset, remove the failed edges, BFS check connectivity between source and target, weight by the probability of that subset. Sum across all subsets to get the reliability at a given edge failure probability. Sweep p from zero to one and plot the polynomial. Edit the graph to change the result live.