Vincent Traag

I'll try to get this implemented before 0.10. I was just thinking to extend the main function as ```diff igraph_error_t igraph_community_leiden(const igraph_t *graph, const igraph_vector_t *edge_weights, const igraph_vector_t *node_weights, const...

We can have some sort of hierarchy, by using bit fields for errors (although that is also limiting the number of errors we can have) and then testing like `err...

Closely related issues: #2065, #2066.

I've been looking at it a bit, it seems that setting ``` options.ncv = DIM; ``` instead of ``` options.ncv = 9; ``` also alleviates the problem (and so keeping...

I have looked into this a bit further, and I can't figure out what the problem could be. The incorrect eigenvectors only appear when using `ncv = 9`, not for...

The Leiden algorithm from `igraph` only implements two quality functions: CPM and modularity. Both indeed use resolution parameters, and are the equivalent of `CPMVertexPartition` and `RBConfigurationVertexPartition`. The reason the `igraph`...

I haven't performed an in-depth benchmark comparison. But results from a single run of modularity detection on an example (a [Facebook graph](http://konect.uni-koblenz.de/networks/facebook-wosn-links)) is sufficiently revealing I think: ``` Running Leiden...

Good to see the large speed gains! Do note that `g.community_leiden(objective_function='modularity', resolution_parameter=0.15)` won't use any edge weights by default. Possibly, `sc.tl.leiden()`, might use edge weights? This might explain some of...

> I accidentally ran the `igraph_community_leiden` in a for loop, 4 times with the same settings and noticed the leiden output differed slightly each time. Is there a parameter to...

Getting identical output from two runs without using the same seed is unlikely in many cases indeed. In some cases, you really do want to get identical output, so then...