README
¶
Example 09 — Leiden community detection
What it demonstrates
Modularity-optimising community detection with community.Leiden: build
an undirected graph, freeze it into an immutable CSR snapshot, run
Leiden, and read the resulting Partition back — including how to
handle the ghost-slot sentinel (-1) that sharded packing leaves in the
NodeID-indexed Community slice on small graphs.
Domain / scenario
A textbook two-community graph: two K4 cliques (nodes 0–3 and 4–7),
each a complete subgraph, joined by a single bridge edge 3–4.
0───1 4───5
│╲ ╱│ │╲ ╱│
│ ╳ │ │ ╳ │
│╱ ╲│ │╱ ╲│
2───3────────4 (bridge 3–4) 6───7
Each clique is densely connected internally and the two halves touch only through the one bridge, so a modularity-optimising method recovers exactly the two cliques as separate communities.
How to run
go run ./examples/09_leiden
Expected output
Found 2 communities across 8 live nodes
node 0 -> community 0
node 1 -> community 0
node 2 -> community 0
node 3 -> community 0
node 4 -> community 1
node 5 -> community 1
node 6 -> community 1
node 7 -> community 1
Key APIs
graph/adjlist.New/AdjList.AddEdge— build the mutable undirected graph (two cliques plus a bridge).graph/adjlist.AdjList.Mapper— translate the user node values (int) into compactNodeIDs viaLookup.graph/csr.BuildFromAdjList— freeze the builder into an immutable CSR snapshot for analytics.search/community.Leiden/DefaultLeidenOptions— run Leiden community detection with the default parameters.search/community.Partition— the result:NumCommunitiesand a NodeID-indexedCommunityslice whose ghost slots carry the sentinel-1.
Further reading
search/community— community-detection package documentationgraph/csr— the immutable CSR snapshot used as the query surfacegraph/adjlist— the mutable adjacency-list builder- Example 08 — PageRank — another analytics algorithm over a CSR snapshot
- docs/examples-standard.md — the standard every example follows
Documentation
¶
Overview ¶
Example 09_leiden — runs Leiden community detection on two K4 cliques joined by a single bridge edge and prints the discovered communities.
The graph has a textbook community structure: each K4 clique is densely connected internally and the two cliques touch only through one bridge edge, so a modularity-optimising method recovers exactly two communities. The example also shows how to read a Partition back: the Community slice is NodeID-indexed and ghost slots created by sharded packing carry the sentinel -1, so the report iterates the eight live nodes explicitly rather than walking the raw slice.
Sample output: run `go run ./examples/09_leiden` 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