[Networkit] NetworKit 4.2 Release
kolja.esders at student.kit.edu
Tue Dec 13 23:29:06 CET 2016
we just released NetworKit 4.2.
These are the release notes:
NetworKit 4.2 Release
Today we announce the next version of NetworKit, the open-source toolkit
for large-scale network analysis. NetworKit is a Python package, with
performance-critical algorithms implemented in C++/OpenMP.
- New graph drawing algorithm for the Maxent-stress model; the algorithm
can layout even large graphs quickly. It follows the paper by Gansner et
al. with some modifications; the biggest deviation is the use of the LAMG
solver for the Laplacian linear systems
- Parallel implementation for the approximation of the neighborhood
function; class has been refactored from ApproxNeighborhoodFunction to
- New heuristic algorithm for the neighborhood function. It is based on
sampling and the breadth-first search and offers more flexibility with
regards to the tradeoff between running time and accuracy as the number of
samples can be specified by the user. It is also much faster than the
approximation algorithm for networks with a high diameter (e.g. road
- Iterative implementation of components.StronglyConnectedComponents,
which is now the new default. For graphs where edges have been deleted, it
is recommended to use the recursive implementation, which is still
- Removed heuristic for vertex diameter estimation from
centrality.ApproxBetweenness (now the vertex diameter is estimated as
suggested in Riondato, Kornaropoulos: *Fast approximation of betweenness
centrality through sampling*)
- Refactoring of the approximation algorithms in the distance group.
ApproxNAME -> NAMEApproximation.
- Simplified installation procedure: Install required dependencies
Please let us know in case there are any issues.
Yani and Kolja
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the NetworKit