/ParMBE

Primary LanguageC++

Introduction

This the re-implementation of the paper, Shared-Memory Parallel Maximal Biclique Enumeration, published at 2019 IEEE International Conferenece on High Performance Computing, Data, and Analysis (HiPC) written by Apurba Das et.al

Architecture

  • include:
    • All the header files
  • src:
    • All the cpp files
  • test
    • Implementations including ParLMBC and ParMBE

Usage

The ParLMBC and ParMBE (the proposed method in this paper) can be found in the test folder. We take ParMBE for example.

cd test/parmbe_test
make clean
make
./parmbe_main <edge_file_path> <thread_num>

The default thread_num is the maximum thread number of the underlying computer.

The format of the input graph file follows the format of KORNET Network shown below. Notably, since the graph is bipartite, the source_ndoe and destination_node should be in two disjoint set. The IDs of these two set start from 1. Our graph reader in src/CSR.cpp will remove the replicated edges. The two disjoint set's IDs can be overlaped.

% Some description of the file
source_node destination_node <other edge information>
... ... ...
... ... ...