Non-contiguous Input Ids Supported

Features

The fix from the upstream merged for the non-contiguous input ids.

Known Issues

The Fast Consensus (starting from the original version) is not always converges (e.g., SNAP amazon, dblp, etc. networks) when the number of workers is ~ <= p / 2.

Fast Consensus for Weighted Networks

Features

Weighted networks processing is implemented

Known Issues

The Fast Consensus (starting from the original version) is not always converges (e.g., SNAP amazon, dblp, etc. networks) when the number of workers is ~ <= p / 2.

FastConcensus clustering algorithms extended for the Clubmark benchmarking

The FastConcensus clustering algorithms extended for the Clubmark benchmarking:

• arbitrary ranges of ids are supported in the input edgefile
• output directory and the number of worker processes are parameterized
• description and code formatting refined a bit

Note: in spite of the edgefiles with weights are supported as a valid input, the weight are actually ignored (both in this and original implementations).