hu-macsy/ipdps2020-mpi-betweenness

ipdps2020-mpi-betweenness

This repository contains the necessary code + helper scripts to recreate the experiments used in "Scaling Betweenness Approximation to Billions of Edges by MPI-based Adaptive Sampling" (A. van der Grinten, H. Meyerhenke).

Requirements:

• C++ compiler, g++ 9.0 or higher recommended
• MPI-runtime, code was tested with OpenMPI 4 and MPICH
• simexpal (see Notes for more details about usage)

Notes:

• To build the necessary code and run the tests use simexpal: https://github.com/hu-macsy/simexpal/. The configuration of the experiments can be found inside experiments.yml.
• Instances are referenced by their names and path. Some of the smaller (both generated and real networks) instances can be downloaded here: https://box.hu-berlin.de/d/f2b95b8f887e4847a114/
• If other instances should be tested, they need to be converted to NetworKitBinary format. See here for more information: https://networkit.github.io/dev-docs/notebooks/IONotebook.html#NetworkitBinaryGraph-file-format You can also get in touch with macsy@informatik.hu-berlin.de if there need for a particular instance.

Results:

• If you are just interest in the results, the subfolder results contains an archive (data.tar.gz) with output data from all experimental runs and several Jupyter notebooks. In the notebooks you can find the evaluation scripts, used for creating plots and table data for the paper.