Add difub for computing the diameter of directed strongly connected networks #791
Conversation
Right now difub takes the largest scc of a graph since getting directed strongly connected graphs is difficult.
networkit/cpp/distance/Diameter.cpp
Outdated
| count size_f = i < distancesF.size() ? distancesF[i].size() : 0; | ||
| count size_b = i < distancesB.size() ? distancesB[i].size() : 0; | ||
| #pragma omp parallel for | ||
| for (count j = 0; j < size_f + size_b; ++j) { |
There was a problem hiding this comment.
This needs to be omp_index which is a signed integer when using MSVC as - at lest with the parameters NetworKit uses - MSVC doesn't support OpenMP loops with unsigned integers.
There was a problem hiding this comment.
Thanks, I fixed it. Changing j to omp_index should be enough? In cpp/distance/APSP.cpp the loop has a count in its condition.
|
Master branch seems to pass CI now, merged it and now this branch should pass it as well. |
| */ | ||
| template <class InputIt, typename L> | ||
| void BFSfrom(const Graph &G, InputIt first, InputIt last, L handle) { | ||
| void BFSfrom(const Graph &G, InputIt first, InputIt last, L handle, bool reverse = false) { |
There was a problem hiding this comment.
Thank you for the PR. I see that you need to follow in-edges but I am not sure that adding this reverse parameter to BFSfrom is the best way to do it.
What about making the implementation of BFSfrom private (e.g., BFSfromImpl) and exposing two functions BFSfrom (follows out-edges) and ReverseBFSfrom (follows in-edges)? The two functions would both use BFSfromImpl, which can take an additional (template?) boolean parameter to decide whether to follow in- our out-edges (as you are already doing here). This holds for Eccentricity as well.
There was a problem hiding this comment.
Thank you for the feedback. This sounds like a good way to do it. I will implement it like this.
Compute the diameter of directed graphs with one strongly connected component faster than with the current naive approach.
Implemented as the practical part of @michitux lecture on networks.
Difub: http://link.springer.com/chapter/10.1007/978-3-642-30850-5_10