Go implementation of Markov Clustering. It isn't currently particularly well optimised or anything (although it mostly uses mat64 methods for matrix operations, so I imagine it does OK). But there wasn't another one for golang, so it's better than nothing.
I've tested it on graphs of a few thousand nodes, where it completes in a couple of seconds.
If you've already got a graph in adjacency matrix form, then you can just throw it directly into the MCL method
m := NewMCL(power, inflation, maxLoops)
m.GenerateClusters(adjacencyMatrix)If you just have a bunch of data, I've included a small graph implementation.
g := graph.NewGraph()
n1 := graph.NewNode("1")
n2 := graph.NewNode("2")
n3 := graph.NewNode("3")
n4 := graph.NewNode("4")
g.AddNode(&n1)
g.AddNode(&n2)
g.AddNode(&n3)
g.AddNode(&n4)
g.AddEdge(&n1, &n2, 2)
g.AddEdge(&n2, &n3, 2)
g.AddEdge(&n1, &n3, 2)
c, err := g.GetClusters(2, 2, 100)The implementation won't output clusters smaller than 2 elements, and won't repeat clusters.