How to run

Either install the dependencies yourself (OpenMPI and HDF5) or use the provided Dockerfile that will provide you with a container with all the dependencies installed.

Then you can just call "make run" to compile and run the program.

Data

To prepare data for this program, it needs to be in a specific format. It needs to be a file of 3 integers per row, where the first two are "source" and "destination" of an edge and the third is irrelevant. Then convert the this edge-set into an edge-list using the provided script. Maybe change up the file path in the script to the data you want to convert.

cd data
python convert_to_list_format.py

Then you can convert it into an HDF5 file using the provided script. Maybe change up the file path in the script to the data you want to convert.

python convert_to_hdf5.py

The file path to the HDF5 file is hardcoded in the C++ program, so you need to change it to the file you want to use. Place the data set "coauth-DBLP-full-proj-graph.txt" in the data folder and run the two scripts in case you don't want to change the hardcoded paths.

Do all of this in docker container

Best, use VSCode with the Remote-Containers extension. It will automatically build the container and mount the project folder into the container. Press ctrl+shift+p and choose option "Dev Container: Reopen in Container".

You can also call "make run MPI_RANKS=<some_number>" to run the program with a different number of MPI ranks. The default is 2. It needs to be a power of 2.