Weighted graph methods built for dynamic community detection #89
Conversation
antepusic
added
lang: cpp
cpp/mg_utility/mg_graph.hpp
Outdated
| @@ -279,6 +333,9 @@ class Graph : public GraphView<TSize> { | |||
|
|
|||
| std::vector<std::vector<TSize>> adj_list_; | |||
| std::vector<std::vector<TNeighbour>> neighbours_; | |||
There was a problem hiding this comment.
Do neighbours_ contains both in and out neighbours or just out? If so make it explicit
There was a problem hiding this comment.
Due to the way neighbours_ is initialized, with directed graphs it only contains the out-neighbors.
Keeping in mind that in- and out-neighbors are identical for undirected graphs, I’m reluctant to rename the vector, but I’ll clarify this with a comment. Does this sound good to you?
There was a problem hiding this comment.
Yes, that sound ok for now. But I would definitely think about separating different versions of Graph, since now you have directed/undirected/weighted in the same class, and that causes undirected to have properties it doesn't use. What do you think about this @jmatak ?
There was a problem hiding this comment.
If we decide to separate them, undirected graphs would still have to implement certain interfaces because of generality. For example, if an algorithm requires the out-neighbors, we should be able to call OutNeighbours without having to check whether the graph is directed. Of course, for undirected graphs, OutNeighbours would just default to calling Neighbours.
Does the above make sense? I believe we should definitely take it into account if it’s a valid suggestion.
There was a problem hiding this comment.
aha, so we are speaking in the context of algorithms running on both undirected and undirected graphs (sorry for the late response)?
Co-authored-by: Jure Bajic <jure.bajic94@gmail.com>
Co-authored-by: Jure Bajic <jure.bajic94@gmail.com>
Co-authored-by: Jure Bajic <jure.bajic94@gmail.com>
Weighted graph methods for MAGE’s dynamic community detection query module (#66), transcluded into separate PR for easier review per Matak’s suggestion.