Q: Compute complete Betweenness Centrality about cuhornet

[delanyinspiron6400]

Thank you very much, I got it to run with this description. 😄

[delanyinspiron6400]

And unrelated to this question something else, I was redirected to this repository with my question, as the original Hornet is seemingly not maintained anymore, but it seems the current state of the repository does not build.
In the original, I had to fix small stuff in xlib (some includes missing so it works with gcc on ArchLinux), but in this repository, it seems more is missing.
Looking at the repository structure, there was a module rmm, which was removed, but which is referenced throughout the code, even manually pulling it into externals/rmm did not resolve this issue.

Should this repository be useable at the moment as a public repository? Or should I stick with the original?

[ogreen]

I will start off with the 2nd comment\question by saying that the "getting started" documentation for this repo needs to be updated. To build this repo, please visit: https://rapids.ai/start.html. You will need to use conda packages to get rmm working.

The three so called variants of BC are actually two variants: exact and approximate.
There is one class BCCentrality that does the actual traversal according to Brandes algorithm.
Exact BC does the traversal for each root\vertex in the graph.
Approximate BC does the traversal for a subset of vertices. How the subset of vertices is selected is "open" to debate. Two popular approaches are:

1. Pre-select the roots (using some random distribution).
2. Choose one root at a time and after each traversal, choose the next root based on the BC values.