README
¶
Example 08 — PageRank
What it demonstrates
Runs search/centrality.PageRank over a small directed graph and reads
back a per-node stationary-rank vector, then prints the pages ordered
from most to least important. It also shows how to filter the live
nodes out of the rank slice through the adjacency-list mapper.
Domain / scenario
A tiny web of six pages that link to one another:
B -> A H -> B
C -> A H -> C
D -> A
E -> A A -> H
Four peripheral pages (B, C, D, E) all link to a single Authority page A, giving A a high in-degree. A's only out-link feeds a Hub page H, and H endorses just two of the four peripheral pages (B and C).
This topology is deliberately not a symmetric cycle. In a directed cycle every PageRank score is identical, so the ranks tell you nothing. Here the asymmetry produces a clear gradient:
- A wins — every other page endorses it.
- H is second — it receives the authority's entire single out-link.
- B and C outrank D and E — they collect a share from the hub on top of the uniform teleport mass, while D and E receive only the teleport (random-jump) share.
B and C are symmetric, so they tie; likewise D and E. The example sorts descending by rank and breaks ties by page name, so the printed order is fully deterministic.
How to run
go run ./examples/08_pagerank
Expected output
Converged in 86 iterations (6 live pages)
1. page A: 0.331875
2. page H: 0.307095
3. page B: 0.155515
4. page C: 0.155515
5. page D: 0.025000
6. page E: 0.025000
Key APIs
graph/adjlist.New/AdjList.AddEdge— build the mutable directed graph.graph/adjlist.AdjList.Mapper— resolve page names to compactNodeIDs (Lookup), used to read only the live entries from the rank slice.graph/csr.BuildFromAdjList— freeze the builder into an immutable CSR snapshot for the analytics pass.search/centrality.PageRank— power-iteration PageRank returning the per-NodeIDrank slice and the iteration count to convergence.search/centrality.DefaultPageRankOptions— the classic Brin-Page parameters (damping 0.85, max 100 iterations, tolerance 1e-6).
Further reading
search/centrality— centrality algorithms package documentationgraph/csr— the immutable CSR snapshot used as the query surfacegraph/adjlist— the mutable adjacency-list builder- Example 01 — basic shortest paths — the minimal build-snapshot-query flow this example extends
- docs/examples-standard.md — the standard every example follows
Documentation
¶
Overview ¶
Example 08_pagerank — runs PageRank on a small directed "authority" graph and prints each page's rank, sorted from most to least important.
The graph models a tiny web of pages that link to one another. Four peripheral pages (B, C, D, E) all link to a single Authority page A, giving A a high in-degree. A is the only node every other page endorses, so the random-surfer model concentrates the most stationary mass on it. A links back to a Hub page H, which in turn links to two of the peripheral pages — this asymmetry is what makes the four peripheral ranks differ from one another instead of being identical.
Unlike a symmetric cycle (where every PageRank score is the same), this topology produces clearly distinct ranks: the authority A wins, the hub H comes next on the strength of A's single outgoing link, the two pages H endorses (B, C) outrank the two it does not (D, E).
Sample output: run `go run ./examples/08_pagerank` and capture the stdout — the output is deterministic for the inputs hard-coded above and serves as the regression baseline a future change should preserve.
Source Files